article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
a two - dimensional numerical model of the device was constructed to allow for a greater understanding of the underlying physics of the cell @xcite .
the model solves the nernst - planck equations with advection in the imposed flow for the concentrations and electrostatic potential , assuming a quasi - neutral bulk electrolyte .
fully developed poiseuille flow was assumed in the channel , with reactions occurring at the top and bottom of the channel along thin electrodes ( fig .
[ fig : cell_design ] ) . because the channel width far exceeds the height ,
edge effects can be ignored , validating the two - dimensional assumption .
equilibrium potentials along the cathode and anode were determined by the nernst equation assuming dilute solution theory , and activation losses were estimated using the symmetric butler - volmer equation , consistent with existing kinetics data @xcite . electrolytic conductivity was assumed to depend on the local ( spatially evolving ) hydrobromic acid concentration , and was calculated using empirical data @xcite . for a peclet number of 10,000 and initial concentrations of bromine and hydrobromic acid of 1 m
is overlaid ( a ) .
assembled cell prior to testing ( b).,width=9 ] the cell was operated galvanostatically at room temperature and atmospheric pressure over a range of flow rates and reactant concentrations .
the cell was observed to reach steady state in less than ten seconds , so each data point was collected after sixty seconds of steady state operation to eliminate transient artifacts .
polarization data was collected as a function of peclet number for the hblfb using 1 m hbr and 1 m br@xmath2 , and compared with numerical model results ( fig .
[ fig : pol_curve ] ) .
the peclet numbers of 5,000 , 10,000 , and 15,000 correspond to reynolds numbers of 5.75 , 11.5 , and 17.25 , oxidant flow rates of 0.22 , 0.44 , and 0.66 ml min@xmath1 @xmath0 , and mean velocities of 6.3 , 12 , and 19 mm s@xmath1 , respectively .
the oxidant flow rates correspond to stoichiometric currents of 0.7 , 1.4 , and 2.1 a @xmath0 respectively .
hydrogen was flowed in excess at a rate of 25 sccm .
the slightly enhanced maximum current density of the observed results compared to the predicted results may be attributed to the roughness of the electrode surface producing chaotic mixing that slightly enhances reactant transport . below limiting current ,
the agreement between model and experiment is very good . at low current densities ,
the voltage differences between the different peclet numbers are small , but in each case , the voltage drops rapidly as the cell approaches limiting current , corresponding to the predicted mass transfer limitations . and 1 m hbr . for low concentration reactants ,
mass transport is the dominant source of loss , and limiting current density @xmath3 can be predicted analytically as a function of peclet number @xmath4 using the lvque approximation ( inset ) ( a ) . predicted ( dashed ) and observed ( symbols ) cell voltage and power density as a function of current density and @xmath5 concentration for @xmath6 and 3 m hbr .
the higher reactant concentrations require concentrated solution theory to accurately model , but allow for much higher current and power density ( b).,width=9 ] these limitations can be estimated analytically for fast reactions by assuming a bromine concentration of zero at the cathode and applying the lvque approximation to the bromine depletion layer @xcite . for a channel of length @xmath7 and height @xmath8 , fully developed poiseiulle flow , and initial bromine concentration @xmath9 , the resulting partial differential equation and boundary conditions can be expressed in terms of dimensionless bromine concentration @xmath10 , position @xmath11 and @xmath12 , channel aspect ratio @xmath13 , and peclet number @xmath14 .
@xmath15 a similarity technique can be applied to convert this to an ordinary differential equation . @xmath16{\beta\tilde{x}/2\textrm{pe}}}\nonumber\end{aligned}\ ] ] this equation can be solved exactly in terms of the incomplete gamma function , @xmath17 .
@xmath18 limiting current can be calculated using faraday s law to determine the distribution of current along the length of the electrode and integrating to obtain limiting current @xmath19 as a function of peclet number @xmath20 , reactant concentration @xmath9 and diffusivity @xmath21 , channel height @xmath8 and aspect ratio @xmath22 , faraday s constant @xmath23 , and the number of moles of electrons transferred per mole of reactant @xmath24 .
@xmath25{\frac{\textrm{pe}}{\beta}}\end{aligned}\ ] ] this result has considerable bearing on how laminar flow systems should be designed and operated .
the presence of aspect ratio in the denominator means that shortening the channel results in greater power density , as observed in experiment @xcite .
in addition , the weak 1/3 dependence on peclet number means that increasing the flow rate beyond a certain point yields minimal benefits .
there was excellent agreement between maximum observed current density , maximum numerically predicted current density , and the analytically predicted limiting current density as a function of peclet number ( fig .
[ fig : pol_curve]a ) .
higher bromine concentrations were also investigated by using a more concentrated electrolyte ( 3 m hbr ) to enhance bromine solubility and move beyond the mass transfer limitations of 1 m br@xmath2 .
the performance of this system was investigated at a peclet number of 10,000 by varying the bromine concentration ( fig .
[ fig : pol_curve]b ) .
using 5 m br@xmath2 and 3 m hbr as the oxidant and electrolyte respectively , a peak power density of 0.795 w @xmath0 was observed when operated near limiting current density .
this corresponds to power density per catalyst loading of 1.59 w mg@xmath1 platinum .
the open circuit potential of the cell dropped more than might be predicted simply using the nernst equation at the cathode .
the drop in open circuit potential is consistent with data on the activity coefficient of concentrated hydrobromic acid available in the literature , as well as previous studies that employed concentrated hydrobromic acid @xcite . to account for this effect
, empirical data for the activity coefficient of hydrobromic acid as a function of local concentration was incorporated into the boundary conditions .
the activity coefficient was assumed to vary slowly enough within the electrolyte that gradients in activity coefficient were neglected in the governing equations . taking into account the activity coefficient of hydrobromic acid , there is good agreement between the model and experiment . as in the low concentration data ,
the observed maximum current density is slightly higher than predicted .
otherwise the model captures the main features of the data , including the transition from transport limited behavior at low bromine concentrations , evidenced by a sharp drop in cell voltage , to ohmically limited behavior at high bromine concentrations . in this ohmically limited regime ,
mass transport limitations are less important , and the limiting current solution applied to the low concentration results does not apply .
charging behavior was also investigated by flowing only hbr and applying a voltage to electrolyze hbr back into br@xmath2 and h@xmath2 .
the voltage versus current density behavior of the cell was investigated during charging as a function of hbr concentration at a peclet number of 10,000 ( fig .
[ fig : charging_data ] ) .
experimental conditions were kept identical to those of the discharge experiments , with the exception that no bromine was externally delivered to the cell .
side reactions , in particular the formation of hypobromous acid and the evolution of oxygen become dominant before potentials get sufficiently high to observe limiting current , the numerical model can not accurately describe this behavior at potentials above 1.3 volts , and was not applied to the charging data . at lower voltages , both the electrolyte conductivity and the limiting current density were increased by increasing the hbr concentration , resulting in increased performance ( fig .
[ fig : charging_data ] ) .
roundtrip voltage efficiency was then calculated by dividing the discharge voltage by the charging voltage for a number of reactant concentrations ( fig .
[ fig : power_efficiency ] ) .
voltage efficiencies slightly greater than 100% were observed for low power densities due to differences in the open circuit potential that are generated by the variation in bromine concentration between the charging and discharging experiments , but this anomaly becomes unimportant at higher current densities , where the reactant concentration varies spatially much more strongly .
using high concentration reactants , roundtrip efficiency of 90% was observed when using high concentration reactants at 25% of peak power ( 200 mw/@xmath26 during discharge ) .
this appears to be the first publication of roundtrip charging and discharging of a membrane - less laminar flow battery , and compares very favorably to existing flow battery technologies .
vanadium redox batteries , for example , have demonstrated voltage efficiencies as high as 91% , but only at a discharge power density nearly an order of magnitude lower than the hblfb @xcite .
concentration of 1 m. increasing the hbr concentration increases both the conductivity and the limiting current , resulting in superior performance.,width=9 ] although closed - loop operation was not demonstrated in this study , some insight into coulombic efficiency can be gathered by considering the effect of reactant mixing within the channel . at 25% of peak power ,
the single - pass coulombic efficiency of the cell is only about 1% .
if no attempt were made to separate the reactants at the outlet of the cell , the resulting energy efficiency of the cell would also be very low .
however , if the channel outlet were split to conserve the volume of stored electrolyte and oxidant , only the br@xmath2 that had diffused into the electrolyte layer would be lost . assuming fully developed poiseuille flow , this corresponds to a coulombic efficiency of 72% , or a roundtrip energy efficiency of 66% .
several opportunities for improvement on this design remain to be investigated .
first , the porous hydrogen anode was selected because of its commercial availability , despite the fact that it was intended for use in a proton exchange membrane fuel cell .
the catalyst composition and structure , wettability , and media porosity have not been optimized .
previous work has shown that these parameters can impact the power density of laminar flow cells by nearly an order of magnitude @xcite .
recent work on thin hydrogen oxidation electrodes has demonstrated that excellent catalytic activity can be achieved with platinum loadings as low as 3 @xmath27 g @xmath0 , more than two orders of magnitude lower than the electrodes used in this study @xcite . assuming equivalent performance , an hblfb employing such a hydrogen electrode would have a specific power density of roughly 250 w mg@xmath1 platinum , virtually eliminating platinum as a cost - limiting component of the system .
second , the channel geometry used in these experiments was relatively long in order to achieve high oxidant utilization .
shortening the channel would decrease the average thickness of the depletion layer that develops along the cathode , enabling higher current densities @xcite .
third , reducing the distance between electrodes would greatly reduce ohmic losses , which are dominant when high concentration reactants are fed into the cell .
incorporating further refinements , such as micro patterned chaotic mixing patterns coupled with a non - specific convection barrier or non noble metal catalysts could also improve performance . @xcite . the initial data presented here for the hblfb suggests that high power density and high efficiency energy storage is achievable using a membrane - less electrochemical cell operating at room temperature and pressure .
the hblfb requires no special procedures or facilities to fabricate , and uses kinetically favorable reactions between abundant , low cost reactants .
recent work has shown that a membrane - based hydrogen - bromine flow battery at room temperature can generate 850 mw @xmath0 , or 7% more power than these experiments with the hblfb at room temperature @xcite . however , this was achieved using a stoichiometric oxidant flow rate over 8 times larger than that used in this work , as well as an acid - treated porous bromine electrode with substantially greater active area than the bare graphite bromine electrode of the hblfb .
this work represents a major advance of the state of the art in flow batteries . to the best of the authors knowledge ,
the data presented here represent the highest power density ever observed in a laminar flow electrochemical cell by a factor of three , as well as some of the first recharging data for a membrane - less laminar flow electrochemical cell @xcite .
although previous work has identified the appropriate scaling laws @xcite , the result presented here represents the first exact analytical solution for limiting current density applied to a laminar flow electrochemical cell , and serves as a guide for future designs . the hblfb rivals the performance of the best membrane - based systems available today without the need for costly ion exchange membranes , high pressure reactants , or high temperature operation .
this system has the potential to play a key role in addressing the rapidly growing need for low - cost , large - scale energy storage and high efficiency portable power systems .
a proof of concept electrochemical cell was assembled using a graphite cathode and a commercial carbon cloth gas diffusion anode with 0.5 mg @xmath0 of platinum ( 60% supported on carbon ) obtained from the fuel cell store ( san diego , ca ) .
the cell was housed between graphite current collectors and polyvinylidene fluoride ( pvdf ) porting plates ( fig .
[ fig : cell_design]b ) .
all components were fabricated using traditional cnc machining or die cutting .
no additional catalyst was applied to the cathode .
the hydrogen flow rate through the porous anode was 25 sccm , and an oxidant stream of bromine mixed with aqueous hydrobromic acid passed through the channel in parallel with an electrolyte stream of aqueous hydrobromic acid .
an 800 @xmath27 m thick viton gasket was used to separate the two electrodes and create the channel , which was 1.4 cm long from oxidant inlet to outlet with an active area of 25 mm@xmath28 .
a fixed ratio of ten to one was maintained between the electrolyte and oxidant flow rates , and the net flow rate was adjusted to study the performance of the cell as a function of the peclet number , @xmath20 , defined by the average flow velocity @xmath29 , channel height @xmath8 , and bromine diffusion coefficient @xmath21 such that @xmath30 .
a scaled , dimensionless model was constructed in comsol multiphysics , and results were calculated over a range of flow rates and reactant concentrations .
a complete description of the model has been presented previously @xcite .
bromine concentration varied along the length of the channel , resulting in strong spatial variations in the current density ( fig .
[ fig : cell_design]a ) .
current - voltage data was obtained for a range of reactant concentrations and flow rates for comparison to experimental data by averaging the current density along the length of the cell , and calculating solutions over a range of specified cell voltages .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 _ et al .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , & _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _ _ * * , ( ) . , , & . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , &
. _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
( ) . , , , & _ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) .
_ _ * * , ( ) .
et al . _
_ _ * * , ( ) .
_ * * , ( ) . , & .
_ _ * * , ( ) .
_ _ ( , ) .
_ _ ( , ) .
_ _ ( , ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . , , &
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
_ et al . _ . _
_ ( ) . , &
_ _ * * , ( ) .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
the authors acknowledge financial support from the department of defense ( dod ) through the national defense science & engineering graduate fellowship ( ndseg ) program , as well as the mit energy initiative seed fund .
w.a.b . performed the experiments and simulations ; m.z.b . developed the theoretical plan , and c.r.b . the experimental plan .
all authors wrote and edited the manuscript .
the authors have no competing financial interests . | in order for the widely discussed benefits of flow batteries for electrochemical energy storage to be applied at large scale , the cost of the electrochemical stack must come down substantially .
one promising avenue for reducing stack cost is to increase the system power density while maintaining efficiency , enabling smaller stacks .
here we report on a membrane - less , hydrogen bromine laminar flow battery as a potential high power density solution .
the membrane - less design enables power densities of 0.795 w @xmath0 at room temperature and atmospheric pressure , with a round - trip voltage efficiency of 92% at 25% of peak power .
theoretical solutions are also presented to guide the design of future laminar flow batteries .
the high power density achieved by the hydrogen bromine laminar flow battery , along with the potential for rechargeable operation , will translate into smaller , inexpensive systems that could revolutionize the fields of large - scale energy storage and portable power systems .
low - cost energy storage remains a critical unmet need for a wide range of applications , include grid scale frequency regulation , load following , contingency reserves , and peak shaving , as well as portable power systems @xcite . for applications that require the storage of large quantities of energy economically and efficiently ,
flow batteries have received renewed attention @xcite .
a wide variety of solutions have been proposed , including zinc - bromine and vanadium redox cells @xcite .
this includes recent efforts to incorporate novel concepts such as organic electrolytes for greater voltage stability and semisolid reactants for higher reactant energy density @xcite or chemistries to reduce reactant cost @xcite .
one such permutation is the hydrogen bromine flow battery @xcite .
the rapid and reversible reaction kinetics of both the bromine reduction reaction and the hydrogen oxidation reaction minimize activation losses @xcite , while the low cost ( $ 1.39 kg@xmath1 ) and abundance ( 243,000 metric tons produced per year in the united states alone ) of bromine distinguishes it from many other battery chemistries @xcite .
however , theoretical investigations of such systems have revealed that the perfluorosulfonic acid membranes typically used suffer from low conductivity in the absence of sufficient hydration @xcite . in the presence of hydrobromic acid ,
this membrane behavior is the dominant limitation on overall performance @xcite .
laminar flow electrochemical cells have been proposed to address many of the challenges that face traditional membrane - based systems @xcite .
laminar flow cells eliminate the need for an ion exchange membrane by relying on diffusion to separate reactants . eliminating the membrane decreases cost , relaxes hydration requirements , and opens up the possibility for a much wider range of chemistries to be investigated .
this flexibility has been exploited in the literature ; examples include vanadium redox flow batteries @xcite , as well as methanol @xcite , formic acid @xcite , and hydrogen fuel cells @xcite .
however , none of these systems have achieved power densities as high as their membrane - based counterparts .
this is largely because the proposed chemistries already work well with existing membrane technologies that have been refined and optimized over several decades .
more recently , a laminar flow fuel cell based on borohydride and cerium ammonium nitrate employed a porous separator , chaotic mixing , and consumption of acid and base to achieve a power density of 0.25 w @xmath0 @xcite .
this appears to be the highest previously published power density for a membrane - less laminar flow fuel cell . in this work ,
we present a membrane - less hydrogen bromine laminar flow battery ( hblfb ) with reversible reactions and a peak power density of 0.795 w @xmath0 at room temperature and atmospheric pressure .
the cell uses a membrane - less design similar to previous work @xcite , but with several critical differences that allow it to triple the highest previously reported power density for a membrane - less electrochemical cell and also enable recharging .
first , where many previous laminar flow electrochemical cell designs were limited to low current operation by the low oxygen concentration at the cathode , the hblfb uses gaseous hydrogen fuel and aqueous bromine oxidant @xcite .
this allows for high concentrations of both reactants at their respective electrodes , greatly expanding the mass transfer capacity of the system .
next , both reactions have fast , reversible kinetics , with no phase change at the liquid electrode , eliminating bubble formation as a design limitation .
these two characteristics of the hblfb enable high power density storage and discharge of energy at high efficiency , while avoiding the cost and reliability issues associated with membrane - based systems . |
the main goal of the _ kepler mission _ is to characterize extra - solar planetary systems , particularly earth - like planets in the habitable zone ( e.g. , * ? ? ?
the mission detects the presence of planets through the minute reduction of the light from a star as a planet crosses the line of sight .
several observations of such reductions at fixed time intervals for a given star , and extensive follow - up observations , are used to verify that the effect results from planet transits and to characterize the planet . to ensure a reasonable chance of detection _
kepler _ observes more than 100,000 stars simultaneously , in a fixed field in the cygnus - lyra region .
most stars are observed at a cadence of 29.4 min , but a subset of up to 512 stars can be observed at a short cadence ( sc ) of 58.85s .
_ kepler _ was launched on 6 march 2009 and data from the commissioning period and the first month of regular observations are now available .
the very high photometric accuracy required to detect planet transits @xcite also makes the _ kepler _ observations of great interest for asteroseismic studies of stellar interiors .
in particular , the sc data allow investigations of solar - like oscillations in main - sequence stars . apart from the great astrophysical interest of such investigations they also provide powerful tools to characterize stars that host planetary systems @xcite . in stars with effective temperature @xmath0
we expect to see oscillations similar to those observed in the sun ( e.g. , * ? ? ?
* ) , excited stochastically by the near - surface convection .
these are acoustic modes of high radial order ; in main - sequence stars such modes approximately satisfy the asymptotic relation @xmath1 @xcite . here
@xmath2 is the cyclic frequency , @xmath3 is the radial order of the mode and @xmath4 is the degree , @xmath5 corresponding to radial ( i.e. , spherically symmetric ) oscillations .
also , @xmath6 is essentially the inverse sound travel time across the stellar diameter ; this is closely related to the mean stellar density @xmath7 : @xmath8 .
@xmath9 depends sensitively on conditions near the center of the star ; for stars during the central hydrogen burning phase this provides a measure of stellar age .
finally , @xmath10 is determined by conditions near the stellar surface .
this regular form of the frequency spectrum simplifies the analysis of the observations , and the close relation between the stellar properties and the parameters characterizing the frequencies make them efficient diagnostics of the properties of the star
. this has been demonstrated in the last few years through observations of solar - like oscillations from the ground and from space ( for reviews , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
cccccccccccc name & kic no & @xmath11 ( k ) & [ fe / h ] & @xmath12 & @xmath13 ( cgs ) & @xmath14 & source + & & & & & & @xmath15 & + hat - p-7 & 10666592 & @xmath16 & @xmath17 & @xmath18 & @xmath19 & @xmath20 & ( a ) + & & @xmath21 & @xmath22 & & @xmath23 & & ( b ) + hat - p-11 & 10748390 & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & ( c ) + tres-2 & 11446443 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & @xmath33 & ( d ) + & & @xmath34 & @xmath35 & & @xmath36 & & ( b ) + even observations allowing a determination of @xmath37 provide useful constraints on @xmath7 . with a reliable determination of individual frequencies @xmath7
is tightly constrained and an estimate of the stellar age can be obtained .
this can greatly aid the interpretation of observations of planetary transits ( e.g. , * ? ? ?
* ; * ? ? ?
we note that photometric observations such as those carried out by _ kepler _ are predominantly sensitive to modes of degree @xmath38 . as indicated by eq .
( [ eq : asymp ] ) these are sufficient to obtain information about the core properties of the star .
ground - based transit observations have identified three planetary systems in the _ kepler _ field : tres-2 @xcite , hat - p-7 @xcite , and hat - p-11 @xcite .
these systems have been observed by _
kepler _ in sc mode .
their properties ( cf .
table [ tbl-0 ] ) indicate that they should display solar - like oscillations at observable amplitudes , and hence they are obvious targets for _ kepler _ asteroseismology .
here we report the results of a preliminary asteroseismic characterization of the central stars in the systems , based on the early _ kepler _ data .
we have analyzed data from _ kepler _ for three planet - hosting stars using a pipeline developed for fast and robust analysis of all _ kepler _ p - mode data @xcite .
each time series contains 63324 data points .
sc data characteristics and minor post - pipeline processing are discussed in @xcite .
in addition a limb - darkened transit light curve model fit has been removed and 5-@xmath39 clipping applied to remove outlying data points from each of the time series .
the frequency analysis contains four main steps : 1 .
we calculate an oversampled ( factor of four ) version of the power spectrum by using a least - squares fitting .
we smoothed the spectrum to @xmath40 resolution to remove the fine structure caused by the finite mode lifetime .
we correlated the smoothed power spectrum with an equally spaced comb of delta functions , separated by @xmath41 , and confined to a gaussian - shaped band with a full width at half maximum of @xmath42 .
we adopted the maximum of this convolution over lags between 0 and 0.5 @xmath6 as the filter output for each @xmath6 .
3 . after identifying the peak correlation for the best matched model filter and extracting the large separation corresponding to this peak
we calculate the _ folded spectrum _ ( see fig . [
fig : obs]b ) , i.e. , the sum of the power as a function of frequency modulo the optimum large separation ( the one corresponding to the peak correlation ) .
the summed power is used to locate the p - mode structure and identify the ridges corresponding to the different mode degrees ( based on the asymptotic relation ) .
4 . from the asymptotic relation and the identification of mode degrees we finally identify the position of the individual p - mode frequencies in the smoothed version of the power spectrum ;
when more than one mode is seen near the expected frequency we use the power - weighted average of the two peaks .
those extracted frequencies and the mode identifications are used in the modeling . for observations with low signal - to - noise ratio it may not be possible to identify the individual frequencies . in such cases
the analysis is carried through step 2 , to determine the maximum response and hence an estimate of the large separation .
results on the three individual cases are presented in [ sec : results ] .
stellar evolution models and adiabatic oscillation frequencies were computed using the aarhus codes @xcite , with the opal equation of state @xcite and opacity @xcite and the nacre nuclear reaction parameters @xcite . in some cases
( see below ) diffusion and settling of helium were included , using the simplified formulation of @xcite .
convection was treated with the @xcite mixing - length formulation , with a mixing length @xmath43 in units of the pressure scale height roughly corresponding to a solar calibration . in some models with convective cores ,
overshoot was included over a distance of @xmath44 pressure scale heights .
evolution started from chemically homogeneous zero - age models .
the initial abundances by mass @xmath45 and @xmath46 of hydrogen and heavy elements were characterized by the assumed value of [ fe / h ] , using as reference a present solar surface composition with @xmath47 @xcite and assuming , from galactic chemical evolution , that @xmath48 . from the observed @xmath6 , effective temperature and composition an initial estimate of the stellar parameters was obtained using the grid - based seek pipeline ( quirion et al .
, in preparation ) .
smaller grids were then computed in the vicinity of these initial parameters , to obtain tighter constraints on stellar properties . for hat - p-7
the analysis of the observations yielded frequencies of individually identified modes ; here the analysis was based on @xmath49 where @xmath50 and @xmath51 are the observed and model frequencies , @xmath52 is the standard error in the observed frequencies ( assumed to be constant ) and @xmath53 is the number of observed frequencies .
in addition , we considered @xmath54 , where @xmath55 is the corresponding normalized square difference between the observed and model effective temperature .
when @xmath56 was available we minimized it along each evolution track and considered the resulting minimum values , and the corresponding value of @xmath57 , as a function of the parameters characterizing the models ( see * ? ? ?
* for details ) .
when only the large separation @xmath6 could be determined from the observations , we identified the model along each track which matched @xmath6 and considered the resulting @xmath55 as a function of the model parameters .
the observed power spectrum for hat - p-7 is shown in fig .
[ fig : obs]a .
the presence of solar - like p - mode peaks , with a maximum power around 1.1mhz , is evident . at high frequency
the noise level in the amplitude spectrum is 1.1 parts per million ( ppm ) , with some increase at lower frequency , likely due to the effects of stellar granulation . carrying out the correlation analysis described in [ sec : obs ] we determined the large separation as @xmath58 .
figure [ fig : obs]b shows the resulting folded spectrum .
this clearly shows two closely spaced peaks , identified as corresponding to modes of degree @xmath5 and 2 , and single peak separated from these two by approximately @xmath41 , corresponding to @xmath59 . on this basis
we finally determined the individual frequencies , identifying the modes from the asymptotic relation ; the final set includes 33 p - mode frequencies , determined with a standard error @xmath60 .
these frequencies , corresponding to radial orders between 11 and 24 , are illustrated in fig .
[ fig : obs]c in an chelle diagram ( see below ) .
cccccccccccc @xmath61 & @xmath62 & age & @xmath46 & @xmath45 & @xmath44 & @xmath63 & @xmath7 & @xmath11 & @xmath64 & @xmath56 & @xmath57 + & & ( gyr ) & & & & & @xmath65 & ( k ) & & & + 1 & 1.53 & 1.758 & 0.0270 & 0.6870 & 0.0 & 1.994 & 0.2718 & 6379 & 5.91 & 1.08 & 1.21 + 2 & 1.52 & 1.875 & 0.0290 & 0.6809 & 0.1 & 1.992 & 0.2708 & 6355 & 5.81 & 1.04 & 1.04 + 3 & 1.50 & 2.009 & 0.0270 & 0.6870 & 0.2 & 1.981 & 0.2718 & 6389 & 5.87 & 1.00 & 1.24 + a grid of models was computed for masses between @xmath66 and @xmath67 , [ fe / h ] between 0.17 and 0.38 , and @xmath68 and @xmath69 , extending well beyond the end of central hydrogen burning . the modeling did not include diffusion and settling . at the mass of this star
the outer convection zone is quite thin , and as a result the settling timescale is much shorter than the age of the star . including settling , without compensating effects such as partial mixing in the radiative region or mass loss , leads to a rapid change in the surface composition which is inconsistent with the observed [ fe / h ] ; for simplicity we therefore neglected these effects for hat - p-7 .
the computed frequencies were corrected according to the procedure of @xcite for errors in the modeling of the near - surface layers , by adding @xmath70 where @xmath71 , @xmath72 and @xmath73 . as discussed in [ sec : model ] , for each evolution track , characterized by a set of model parameters , we minimized the departure @xmath56 of the model frequencies from the observations , defining the best model for this set .
we first consider @xmath56 as a function of the effective temperature of the models ( fig .
[ fig : chisq]a ) .
it is evident that there is a clear minimum in @xmath56 ; this is consistent with the determination of @xmath11 by @xcite but not with the somewhat higher temperature obtained by @xcite ( see also table [ tbl-0 ] ) .
thus in the following we use the observed quantities from @xcite . since the frequencies to leading order are determined by the mean stellar density @xmath7 , fig .
[ fig : chisq]b , c show @xmath56 and @xmath57 as functions of @xmath7 .
it is evident that the best - fitting models occupy a narrow range of @xmath7 , with a well - defined minimum .
fitting a parabola to @xmath57 in panel ( c ) we obtain the estimate @xmath74 . in fig .
[ fig : chisq]d @xmath57 is shown against model age . here
the variation with model parameters is substantially stronger , resulting in a greater spread in the inferred age ; in particular , it is evident , not surprisingly , that the results depend on the extent of convective overshoot . from the figure we estimate that the age of hat - p-7 is between 1.4 and 2.3gyr .
examples of evolution tracks are shown in fig .
[ fig : hr ] ; parameters for these models are provided in table [ tbl-2 ] . they were chosen to give the smallest @xmath56 for each of the three values of @xmath44 considered . also shown
are the locations of the models minimizing @xmath56 along each of the computed tracks ; these evidently fall close to a line in the hr diagram , corresponding to the small range in @xmath7 .
the range of luminosities , from @xcite , is based on modeling and hence has not been used in our fit ; even so , it is gratifying that the present models are essentially consistent with these values . also , as indicated by fig .
[ fig : chisq]a and table [ tbl-2 ] , the best - fitting models are close to the value of @xmath11 obtained by @xcite .
the match of the best - fitting model ( model 3 of table [ tbl-2 ] ) to the observed frequencies is illustrated in a so - called _ chelle diagram _
@xcite in fig .
[ fig : obs]c . in accordance with eq .
( [ eq : asymp ] ) the frequency spectrum is divided into slices of length @xmath75 , starting at a frequency of @xmath76 ; the figure shows the location of the observed ( filled symbols ) and computed ( open symbols ) frequencies within each slice , against the starting frequency of the slice ; the model results extend to the acoustical cut - off frequency , @xmath77 , of the model .
there is clearly a very good overall agreement between model and observations , including the detailed variation with frequency which reflects the frequency dependence of the large separation , as a possible diagnostics of the outer layers of the star ( e.g. , * ? ? ?
we have finally made a fit of the inferred @xmath7 , as well as @xmath11 and [ fe / h ] from @xcite , to computed evolutionary tracks from the yonsei - yale compilation @xcite . this was based on a markov chain monte carlo analysis to obtain the statistical properties of the inferred quantities ( see * ? ? ? * for details ) .
this resulted in @xmath78 , @xmath79 and an age of @xmath80gyr .
we note that the age estimate reflects the specific assumptions in the yonsei - yale evolution calculations ; as indicated by fig .
[ fig : chisq]d the true uncertainty in the age determination is likely somewhat larger . for hat - p-11 the oscillation amplitudes were much smaller than in hat - p-7 , as expected from the general scaling of amplitudes with stellar mass and luminosity ( e.g. , * ? ? ?
thus with the present short run of data it has only been possible to determine the large separation @xmath81 from the maximum in the correlation analysis . we have matched this to a grid of models , including diffusion and settling of helium , with masses between 0.7 and @xmath82 and [ fe / h ] between 0.21 and 0.41 .
these models provide a good fit to the observed @xmath11 and @xmath12 ; note that in the present case the luminosity is based on a reasonably well - determined parallax .
we have determined an estimate of @xmath7 by averaging the results of those models which match the observed @xmath6 and lie within 2 standard deviations ( @xmath83k ) from the value of @xmath11 provided by @xcite ; the result is @xmath84 .
although the formal error is extremely small , owing to a tight relation between the large separation and the mean density for stars in this region in the hr diagram , the true error is undoubtedly substantially larger . in particular
, we neglected the error in the determination of @xmath6 and these data have not allowed a correction for the systematic errors in the modeling of the near - surface layers of the star . here also we were unable to determine individual frequencies from the present set of data .
the expected amplitudes are smaller than for hat - p-7 , and the noise level higher due to the fainter magnitude of tres-2 .
the correlation analysis yielded two possible values of @xmath6 : @xmath85 and @xmath86 . for this star @xmath7
has been determined from the analysis of the transit light curve . @xcite
obtained @xmath87 , while @xcite found @xmath88 . from the scaling with
@xmath89 the smaller of the two possible values of @xmath6 is clearly inconsistent with these values of @xmath7 , while @xmath90 yields models that are consistent with the observed @xmath11 and @xmath13 of @xcite as well as with these values of the mean density .
here we considered a grid of models with helium diffusion and settling , masses between 0.85 and @xmath91 and [ fe / h ] between @xmath92 and @xmath93 .
determining again the mean value of @xmath7 for those models that matched @xmath6 and had @xmath11 within two standard deviations of the value of @xcite we obtained @xmath94 .
as in the case of hat - p-11 the true error is likely substantially higher .
the present preliminary analysis provides a striking demonstration of the potential of _ kepler _ asteroseismology and its supporting role in the analysis of planet hosts .
these stars will undoubtedly be observed throughout the mission and hence the quality of the data will increase substantially . for hat - p-7
the detected frequencies are already close to what will be required for a detailed analysis of the stellar interior , beyond the determination of the basic parameters of the star .
thus here we can look forward to a test of the assumptions of the stellar modeling ; the resulting improvements will further constrain the overall properties of the star , in particular its age .
also , given the observed @xmath14 we expect a rotational splitting comparable to that observed in the sun , and hence likely detectable with a few months of observations . for the other two stars there is strong evidence for the presence of solar - like oscillations ; thus continued observations will very likely result in the determination of individual frequencies and hence further constraints on the properties of the stars .
funding for this discovery mission is provided by nasa s science mission directorate .
we are very grateful to the entire _ kepler _ team , whose efforts have led to this exceptional mission . the present work was supported by the danish natural science research council .
_ facilities : _ ammler - von eif , m. , santos , n. c. , sousa , s. g. , fernandes , j. , guillot , t. , israelian , g. , mayor , m. , & melo , c. 2009 , , 507 , 523 angulo , c. , et al . 1999 , , 656 , 3 bakos , g. . , et al .
2010 , , in the press ( arxiv:0901.0282v2 ) bhm - vitense , e. 1958 , , 46 , 108 borucki , w. , et al . 2009 , in proc .
iau symp .
253 , transiting planets , eds f. pont , d. sasselov & m. holman , iau and cambridge university press , 289 borucki , w. , et al .
2010 , science , submitted christensen - dalsgaard , j. 2002 , rev .
phys . , 74
, 1073 christensen - dalsgaard , j. 2008a , , 316 , 13 christensen - dalsgaard , j. 2008b , , 316 , 113 christensen - dalsgaard , j. , arentoft , t. , brown , t. m. , gilliland , r. l. , kjeldsen , h. , borucki , w. j. , & koch , d. 2008 , in proc .
helas ii international conference : helioseismology , asteroseismology and the mhd connections , eds l. gizon & m. roth , j. phys .
ser . , 118 , 012039 grec , g. , fossat , e. , & pomerantz , m. 1983 , , 82 , 55 grevesse , n. , & noels , a. 1993 , in origin and evolution of the elements , eds n. prantzos , e. vangioni - flam & m. cass ( cambridge : cambridge univ . press ) , 15 houdek , g. , & gough , d. o. 2007 , , 375 , 861 huber , d. , stello , d. , bedding , t. r. , chaplin , w. j. , arentoft , t. , quirion , p .- o . , & kjeldsen , h. 2009 , comm . in asteroseismology , 160 , 74 iglesias , c. a. , & rogers , f. j. 1996 , , 464 , 943 kjeldsen , h. , & bedding , t. r. 1995 , , 293 , 87 kjeldsen , h. , bedding , t. r. , & christensen - dalsgaard , j. 2008 , , 683 , l175 kjeldsen , h. , bedding , t. r. , & christensen - dalsgaard , j. 2009 , in proc .
iau symp .
253 , transiting planets , eds f. pont , d. sasselov & m. holman , iau and cambridge university press , 309 koch , d. , et al .
2010 , apjl , submitted michaud , g. , & proffitt , c. r. 1993 , in proc .
iau colloq . 137 : inside the stars , eds a. baglin , & w. w. weiss , asp conf . ser .
, san francisco , 40 , 246 nutzman , p. , gilliland , r. l. , mccullough , p. r. , charbonneau , d. , christensen - dalsgaard , j.,kjeldsen , h. , nelan , e. p. , brown , t. m. , & holman , m. j. 2010 , , submitted odonovan , f. t. , et al . 2006 , , 651 , l61 pl , a. , et al . 2008 , , 680 , 1450 rogers , f. j. , swenson , f. j. , & iglesias , c. a. 1996 , , 456 , 902 southworth , j. 2009 , , 394 , 272 sozzetti , a. , torris , g. , charbonneau , d. , latham , d. w. , holman , m. j. , winn , j. n. , laird , j. b. , & odonovan , f. t. 2007 , , 664 , 1190 tassoul , m. 1980 , , 43 , 469 yi , s. , demarque , p. , kim , y .- c . , lee , y .- w . ,
ree , c. h. , lejeune , t. , & barnes , s. 2001 , , 136 , 417 vandakurov , yu . v. 1967 , , 44 , 786 ( english translation : , 11 , 630 ) | in addition to its great potential for characterizing extra - solar planetary systems the _ kepler mission _ is providing unique data on stellar oscillations . a key aspect of _ kepler _ asteroseismology is the application to solar - like oscillations of main - sequence stars . as an example we here consider an initial analysis of data for three stars in the _ kepler _ field for which planetary transits were known from ground - based observations .
for one of these , hat - p-7 , we obtain a detailed frequency spectrum and hence strong constraints on the stellar properties .
the remaining two stars show definite evidence for solar - like oscillations , yielding a preliminary estimate of their mean densities . |
nonlinear models in ( 1 + 1)-dimensional space - time play important hole both in field theory and in quantum mechanics .
some of such models possess defect solutions representing domain walls that appear , for example , in high energy physics @xcite and condensed matter @xcite . in this work
we extend the method introduced in ref .
@xcite to present new potentials bearing topological ( kinklike ) or non topological ( lumplike ) solutions .
for example , we use the extended deformation procedure to build a semi - vacuumless model , and the corresponding domain wall which serves as seed for generation of dimension bubbles , " as proposed in refs .
we also show how to deform models having lumplike solutions , relevant to the discussion of tachyonic excitations , to generate new ones presenting topological solutions . to begin , consider a theory of a single scalar real field in a ( 1 + 1)-dimensional space - time , described by the lagrangian density @xmath0 we use the metric @xmath1 , and we work with dimensionless fields and coordinates , fields being defined in the whole space .
the equation of motion for static fields is @xmath2 where the prime stands for the derivative with respect to the argument .
mathematically , this corresponds to consider two - point boundary value problems ( with conditions imposed at @xmath3 and @xmath4 ) for second - order ordinary differential equations @xcite .
consider the broad class of potentials having at least one critical point @xmath5 ( that is , @xmath6 ) , for which @xmath7 . in this case , solutions satisfying the conditions @xmath8 obey the first order equation ( a first integral of ( [ eqm2 ] ) ) @xmath9 thus , such solutions equally share their energy densities between gradient and potential parts .
many important examples can be presented : the @xmath10-model , with @xmath11 , is the prototype of theories having topological solitons ( kinklike solutions ) connecting two minima , @xmath12 in this case ; a situation where non topological ( lumplike ) solutions exist is the inverted
@xmath10-model " , with potential given by @xmath13 , lumplike defects being @xmath14 .
one notice that the potential need not be nonnegative for all values of @xmath15 but the solution must be such that @xmath16 for the whole range @xmath17 .
both topological and non topological solutions can be deformed , according to the prescription introduced in @xcite , to generate infinitely many new soluble problems .
this method can be stated in general form as the following theorem : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath18 be a bijective function having continuous non - vanishing derivative . for each potential @xmath19 , bearing solutions satisfying conditions ( [ cond ] ) ( or equivalently eq . ( [ eqfo ] ) ) , the @xmath20-deformed model , defined by @xmath21}{[f^{\prime}(\phi)]^2 } \ , , \label{vtil}\ ] ] possesses solution given by @xmath22 where @xmath23 is a solution of the static equation of motion for the original potential @xmath19 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to prove this assertion , notice that the static equation of motion of the new theory is written in terms of the old potential as @xmath24 - 2v[f(\phi)]\frac{f^{\prime \prime}(\phi)}{[f^{\prime}(\phi)]^3 } \ , .
\label{dd1}\ ] ] on the other hand , taking the second derivative with respect to @xmath25 of eq .
( [ phitil ] ) , one finds @xmath26 ^ 3 } \left ( \frac{d\phi}{dx } \right)^2 \ , .
\label{dd2}\ ] ] it follows from ( [ eqm2 ] ) , ( [ eqfo ] ) and ( [ phitil ] ) that @xmath27 $ ] and @xmath28 $ ] so that @xmath29 satisfies ( [ dd1 ] ) , as stated .
the energy density of the solution ( [ phitil ] ) of the @xmath20-deformed potential ( [ vtil ] ) is given by @xmath30 naturally , the deformation procedure heavily depends on the deformation function @xmath31 .
assume that @xmath32 is bijective and has no critical points .
in this case , the @xmath20-deformation ( and the deformation implemented by its inverse @xmath33 ) can be applied successively and one can define equivalence classes of potentials related to each other by repeated applications of the @xmath20- ( or the @xmath33- ) deformation .
each of such classes possesses an enumerable number of elements which correspond to smooth deformations of a representative one , all having the same topological characteristics .
the generation sequence of new theories is depicted in the diagram below .
@xmath34 & & \widehat{v } \ar[d ] \ar[ll]_{f^{-1 } } & & v \ar[d ] \ar[ll]_{f^{-1 } } \ar[rr]^{f } & & { \widetilde}{v } \ar[d ] \ar[rr]^{f } & & { \widetilde}{{\widetilde}{v } } \ar[d ] & \cdots\\ \cdots & \widehat{\!\widehat{\phi}}_d & & \widehat{\phi}_d \ar[ll]^{f } & & \phi_d \ar[ll]^{f } \ar[rr]_{f^{-1 } } & & { \widetilde}{\phi}_d \ar[rr]_{f^{-1 } } & & { \widetilde}{\!{\widetilde}{\phi}}_d & \cdots}\ ] ] as an example not considered before , take the @xmath35-model .
this model , for which the potential @xmath36 has three degenerated minima at @xmath37 and @xmath38 , is important since it allows the discussion of first - order transitions .
it possesses kinklike solutions , @xmath39/2}$ ] , @xmath40/2}$ ] , connecting the central vacuum with the lateral ones .
take @xmath41 as the deforming function .
the sinh - deformed @xmath35-potential is @xmath42 ^ 2 \label{defpfi6}\ ] ] and the sinh - deformed defects are @xmath43 } \right ) , \\ { { \widetilde}\phi}^-_{\pm}(x)&=&\pm{\rm arcsinh } \left ( \sqrt{\frac{1}{2}\left [ 1-\tanh ( x)\right ] } \right).\label{solphi6-}\end{aligned}\ ] ] notice that , since @xmath44 for the sinh - deformation , the energy of the deformed solutions is diminished with respect to the undeformed kinks .
the reverse situation emerges if one takes the inverse deformation implemented with @xmath45 .
interesting situations arise if one takes polynomial functions implementing the deformations .
consider @xmath46 with @xmath47 for all @xmath48 .
these are bijective functions from @xmath49 into @xmath49 possessing positive derivatives .
fixing @xmath50 corresponds to a trivial rescaling of the field . for @xmath51 , taking @xmath52 , one has @xmath53 with inverse given by @xmath54 $ ] .
thus , the @xmath55-deformed @xmath10 model , for which the potential has the form @xmath56 supports topological solitons given by @xmath57\ , .
\label{solp3}\ ] ] naturally , the inverse deformation can be implemented leading to another new soluble problem .
but if one takes @xmath58 , the inverse of @xmath59 can not be in general expressed analytically in terms of known functions .
this leads to circumstances where one knows analytically solutions of potentials which can not be expressed in term of known functions and , conversely , one has well - established potentials for which solitonic solutions exist but are not expressible in terms of known functions .
for example , take @xmath60 .
the @xmath61-deformed @xmath10 model has potential given by @xmath62 but its solutions @xmath63 are not known analytically . on the other hand , @xmath64 \label{solp5}\ ] ] are topological solutions of the potential @xmath65 ^ 2)^2/(2[p_5^{-1\ , \prime}(\phi)]^2)$ ] which can not be written in terms of known functions .
the procedure can also be applied to potentials presenting non topological , lumplike , solutions which are of direct interest to tachyions in string theory @xcite .
take , for example , the lorentzian lump @xmath66 which solves eq .
( [ eqm2 ] ) for the potential @xmath67 and satisfies conditions ( [ cond ] ) .
distinctly of the topological solitons , this kind of solution is not stable .
in fact , the ` secondary potential ' , that appears in the linearized schrdiger - like equation satisfied by the small perturbations around @xmath68 @xcite is given by @xmath69 this potential , a symmetric volcano - shaped potential , has zero mode given by @xmath70 , which does not correspond to the lowest energy state since it has a node .
deforming the potential ( [ lpot ] ) with @xmath71 leads to the potential @xmath72 $ ] which possesses the lumplike solution @xmath73}$ ] .
the deformation prescription is powerful .
the conditions under which our theorem holds are maintained if we consider a function for which the contra - domain is an interval of @xmath49 , that is , if we take @xmath74 . in this case , however , the inverse transformation ( engendered by @xmath75 ) can only be applied for models where the values of @xmath15 are restricted to @xmath76 .
we illustrate this possibility by asking for a deformation that leads to a model of the form needed in ref .
@xcite , described by a `` semi - vacuumless '' potential , in contrast with the vacuumless potential studied in ref .
consider the new deformation function @xmath77 , acting on the potential @xmath11 .
the deformed potential is @xmath78 which is depicted in fig . 1 .
the kinklike solution is @xmath79 \ , .\ ] ] the deformed potential ( [ dpm ] ) engenders the required profile : it has a minimum at @xmath80 $ ] and another one at @xmath81 .
it is similar to the potential required in ref .
@xcite for the existence of dimension bubbles .
the bubble can be generated from the above ( deformed ) model , after removing the degeneracy between @xmath5 and @xmath81 , in a way similar to the standard situation , which is usually implemented with the @xmath10 potential , the undeformed potential that we have used to generate ( [ dpm ] ) .
a key issue is that such bubble is unstable against collapse , unless a mechanism to balance the inward pressure due to the surface tension in the bubble is found . in ref .
@xcite , the mechanism used to stabilize the bubble requires another scalar field , in a way similar to the case of non topological solitons previously proposed in ref . @xcite .
this naturally leads to another scenario , which involves at least two real scalar fields .
the deformation procedure can be extended even further , by relaxing the requirement of @xmath20 being a bijective function , under certain conditions .
suppose that @xmath20 is not bijective but it is such that its inverse @xmath33 ( which exists in the context of binary relations ) is a multi - valued function with all branches defined in the same interval @xmath82 . if the domain of definition of @xmath33 contains the interval where the values of the solutions @xmath23 of the original potential vary , then @xmath83 are solutions of the new model obtained by implementing the deformation with @xmath20 .
however , one has to check out whether the deformed potential @xmath84/(f^{\prime}(\phi))^2 $ ] is well defined on the critical points of @xmath20 .
in fact , this does not happen in general but occurs for some interesting cases .
consider , for example , the function @xmath85 ; it is defined for all values of @xmath15 and its inverse is the double valued real function @xmath86 , defined in the interval @xmath87 .
if we deform the @xmath10 model with this function we end up with the potential @xmath88 .
the deformed kink solutions are given by @xmath89 with @xmath23 replaced by the solutions ( @xmath90 ) of the @xmath10 model , which reproduce the known solutions of the @xmath35 theory .
the important aspect , in the present case , is that the @xmath91-kink corresponds to field values restricted to the interval @xmath92 which is contained within the domain of definition of the two branches of @xmath93 .
the fact that the @xmath35 model can be obtained from the @xmath10 potential in this way is interesting , since these models have distinct characteristics .
notice that the critical point of @xmath20 at @xmath94 does not disturb the deformation in this case ; this always occur for potentials having a factor @xmath95 , since the denominator of @xmath96 ( eq . [ vtil ] ) is cancelled out .
one can go on and apply this deformation to the @xmath35 model ; now , one finds the deformed potential @xmath97 with solutions given by @xmath98 } \right ) } \ , , \ ] ] corresponding to kinks connecting neighbouring minima ( located at @xmath99 , @xmath100 , @xmath37 , @xmath101 and @xmath102 ) of the potential ( [ phi10 ] ) .
this potential is illustrated in fig .
2 . repeating the procedure for the potential ( [ phi10 ] ) , the deformed potential is a polynomial function of degree @xmath103 , having sixteen kink solutions connecting adjacent minima of the set @xmath104 , and so on and so forth .
the deformation implemented by the function @xmath105 can also be applied to a potential possessing lumplike solutions .
consider the inverted @xmath10 potential @xmath106 , which has the lump solutions @xmath107 .
the deformed potential , in this case , is given by @xmath108 this potential , which is also unbounded from below , vanishes for @xmath109 , has an absolute maximum at @xmath94 and local minima and maxima for @xmath110 and @xmath111 , respectively .
figure 3 shows a plot of this potential .
if we take @xmath112 as the original lump , then the deformed solutions obtained are given by @xmath113 } \
, , \label{slump2}\ ] ] corresponding to lumps which start in the local minima ( for @xmath114 ) , go to the lateral zeros ( when @xmath115 ) and come back to the same minima ( for @xmath116 ) of the potential ( [ vlump2 ] ) . on the other hand
, if the undeformed lump is @xmath117 , taking naively @xmath118 leads to functions that do not have derivative at @xmath115 and , therefore , are not acceptable solutions of the equations of motion .
in fact , the deformation of the @xmath119 lump results in deformed kinks given by @xmath120 } \,\,\,\,\,\,\ , x\leq 0 \\ + \sqrt{\frac{1}{2}\left[1 - { \rm sech}(x)\right ] } \,\,\,\,\,\,\ , x\geq 0 \end{array } \right . \,\ , .
\label{slumpkink}\ ] ] each of such solutions ( e.g. , the @xmath121 one ) starts at a minimum ( @xmath122 ) when @xmath114 , goes to the absolute maximum ( @xmath94 ) at @xmath115 running along one of the branchs of @xmath33 ( the lower branch ) and , passing continuously to the other branch ( the upper one ) , reaches the other local minimum ( @xmath123 ) when @xmath116 .
one notice that , again , the number of solutions duplicates using such a deformation , whose inverse is a double - valued function .
but , in this case , novel topological solutions emerge as deformations of a non topological one .
potentials which have a factor @xmath124 can also be deformed using the function @xmath125 , producing many interesting situations .
in fact , suppose the potential can be written in the form @xmath126 this is always possible for all well - behaved potentials that vanish at both values @xmath127 , as shown by taylor expansion .
then , the sin - deformation leads to the potential @xmath128 \ , , \label{vud}\ ] ] which is a periodic potential , the critical points of @xmath129 not causing any problem to the deformation process .
the inverse of the sine function is the infinitely valued function @xmath130 , with @xmath131 and @xmath132 being the first determination of @xmath133 ( which varies from @xmath134 , for @xmath135 , to @xmath136 , when @xmath137 ) , defined in the interval @xmath138 $ ] .
so to each solution of the original potential , whose field values range in the interval @xmath138 $ ] , one finds infinitely many solutions of the deformed , periodic , potential .
consider firstly the @xmath10 model . applying the sin - deformation to it
, one gets @xmath139 which is one of the forms of the sine - gordon potential .
the deformed solutions thus obtained is given by @xmath140+k\pi$ ] , which correspond to all the kink solutions ( connecting neighbouring minima ) of this sine - gordon model .
for example , the kink solutions @xmath141 , which connect the minima @xmath127 of the @xmath10 model in both directions , are deformed into the kinks @xmath142=2{\rm arctan}(e^{\pm x } ) - \pi /2 $ ] ( which runs between @xmath143 and @xmath144 ) if one takes @xmath145 while , for @xmath146 , the resulting solutions connect the minima @xmath147 and @xmath148 of the deformed potential .
this example can be readily extended to other polynomial potentials , leading to a large class of sine - gordon type of potentials .
for instance , the @xmath35 model , @xmath36 , deformed by the sine function , becomes the potential @xmath149 having kink solutions given by @xmath150}\right ) + k\pi \ , .
\label{s6sin}\ ] ] if , on the other hand , one considers the potential @xmath151 , which is unbounded below and supports kinklike solutions connecting the two inflection points at @xmath38 , one gets the potential @xmath152 with solutions given by @xmath153 a particularly interesting situation appears if one consider the inverted @xmath10 model , which presents lumplike solutions .
the sin - deformation of the potential @xmath154 leads to the potential @xmath155 . in this case
, the lump solutions of @xmath19 , namely @xmath156 , are deformed into @xmath157+k\pi$ ] .
consider the @xmath121-solution and take initially @xmath145 .
as @xmath25 varies from @xmath158 to @xmath37 , @xmath159 goes from @xmath37 to @xmath102 , and @xmath160=2{\rm arctan}(e^x)$ ] changes from @xmath37 to @xmath144 .
if one continuously makes @xmath25 goes from @xmath37 to @xmath161 , then the deformed solution passes to the @xmath146 branch of @xmath133 , @xmath162+\pi$ ] ( @xmath163 for @xmath164 ) , which varies from @xmath144 to @xmath165 as @xmath25 goes from @xmath37 to @xmath161 .
thus , in this case , the lump solution @xmath166 of the inverted @xmath10 model is deformed in the kink of the sine - gordon model connecting the minima @xmath94 and @xmath167 . under reversed conditions ( taking the @xmath146 branch before the @xmath145 one ) , the lump solution @xmath119 leads to the anti - kink solution of the sine - gordon model running from the minimum @xmath167 to @xmath37 .
the other topological solutions of the sine - gordon model are obtained considering the other adjacent branches of @xmath133 .
this is another remarkable example since one has non topological solutions being deformed in topological ones .
finally , we mention that many other soluble models can be construct following the procedure presented in this paper .
also , the discussion raised here might be extended to consider situations where the fields are constrained to intervals of @xmath168 , thus representing two - point boundary value problems defined in finite or semi - infinite intervals of the @xmath25-axis .
such study is left for future work .
+ + * acknowledgments * + the authors would like to thank w. freire for comments , and capes , cnpq , procad and pronex for financial support .
caa thanks funcap for a fellowship .
99 vilenkin a and shellard e p s 1994 _ cosmic strings and other topological defects _ ( cambridge up : cambridge ) walgraef d 1997 _ spatio - temporal pattern formation _ ( springer - verlag : new york ) bazeia d , losano l and malbouisson j m c 2002 _ phys .
_ d * 66 * 101701(r ) blau s k and guendelman e i 1989 _ phys .
_ d * 40 * 1909 morris j r 2003 _ phys .
_ d * 67 * 025005 peixoto m m and da silva a r 1999 _ j. london math .
soc . _ * 60 * 517 zwiebach b 2000 _ j. high energy phys . _ * 09 * 028 minahan j a and zwiebach b 2000 _ j. high energy phys . _ * 09 * 029 jatkar d p and vathsan r 2001 _ j. high energy phys . _ * 06 * 039 jackiw r 1977 _ rev .
phys . _ * 49 * 681 cho i and vilenkin a 1999 _ phys .
_ d * 59 * 021701(r ) ; 1999 * 59 * 063510 bazeia d 1999 _ phys .
_ d * 60 * 067705 frieman j a , gelmini g b , gleiser m and kolb e w 1988 _ phys .
lett . _ * 60 * 2101 | we extend a deformation prescription recently introduced and present some new soluble nonlinear problems for kinks and lumps . in particular , we show how to generate models which present the basic ingredients needed to give rise to dimension bubbles , " having different macroscopic space dimensions on the interior and the exterior of the bubble surface .
also , we show how to deform a model possessing lumplike solutions , relevant to the discussion of tachyonic excitations , to get a new one having topological solutions . |
according to the theory that best reproduces the observations , galaxy formation proceeds in a hierarchical fashion , driven by gravity , in a universe dominated by dark matter of unknown nature ( e.g. * ? ? ?
the hierarchy of merging is accompanied by changes in galaxy structure and morphology . in particular early - type galaxies ( es and s0s )
are thought to form by the gas - rich merging of spiral galaxies or by gas starvation of spirals , followed by subsequent collisionless mergers ( e.g. * ? ? ?
three key global parameters can be used to characterize galaxies structure while studying this sequence of merging of galaxies and dark matter : ( i ) the angular momentum , which varies during mergers and increases with the amount of gas dissipation , ( ii ) the stellar population , which records the history of star formation events during the gas - rich mergers , and ( iii ) the mass - to - light ratio , which is affected by both the population and by the dark - matter fraction .
the large majority of the galaxies in the universe are to first order axisymmetric ( except for bars ) and posses disks .
this includes fast - rotator early - type galaxies @xcite and spiral galaxies .
both the fast - rotator early - types @xcite and the spiral galaxies @xcite appear dominated by the stellar matter inside one half - light radius .
observations suggest that they have a dynamical structure characterized by a flattening of the velocity ellipsoid in the @xmath7 direction parallel to the galaxy symmetry axis @xcite .
the goal of this paper is to include the knowledge of the structure of the fast - rotator and spiral galaxies , into a realistic but simple dynamical modeling method , which can be applied to the measurement of both the mass - to - light ratio and the amount of rotation ( for which the inclination is needed ) in the central regions of these galaxies , while also allowing for the inclusion of dark matter and the study of multiple kinematical components .
the success of the adopted model s assumptions in describing the kinematics of real galaxies is verified against integral - field observations of the stellar kinematics obtained with ` sauron ` @xcite .
in section [ sec : solving_jeans ] we briefly review the theory and past applications of the jeans equations and of the shape of the velocity ellipsoid in galaxies . in section [ sec : anisotropic_jeans_solutions ] we describe our new anisotropic jeans formalism , which we apply and test in section [ sec : tests ] .
finally our results are summarized in section [ sec : summary ] .
the positions @xmath8 and velocities @xmath9 of a large system of stars can be described by the distribution function ( df ) @xmath10 . when the system is in a
_ steady state _ under the gravitational influence of a smooth potential @xmath11 , the df must satisfy the fundamental equation of stellar dynamics , the steady - state collisionless boltzmann equation ( @xcite , hereafter bt ; equation [ 4 - 13b ] ) @xmath12 given that @xmath13 is a function of six variables , an infinite family of solutions satisfies equation ( [ eq : boltzmann ] ) .
additional assumptions and simplifications are required for a practical application of the equation .
one classic way of constraining the problem consists of drastically reducing it , from that of recovering the df , to that of studying only the velocity moments of the df .
this approach leads to the jeans equations , which are discussed in the next section .
if we rewrite equation ( [ eq : boltzmann ] ) in standard cylindrical coordinates @xmath14 and we make the important assumption of _ axial symmetry _ ( @xmath15 ) we obtain ( cf
. bt equation [ 4 - 17 ] ) @xmath16 multiplying equation ( [ eq : boltz_cyl ] ) respectively by @xmath17 and by @xmath18 , and integrating over all velocities , we obtain the two , are zero in axisymmetry . ]
jeans equations ( @xcite ; bt equation [ 4 - 29a , c ] ) @xmath19 where we use the notation @xmath20 these equations are still quite general , as they derive from the steady - state boltzmann equation ( [ eq : boltzmann ] ) with the only assumption of axisymmetry .
they do _ not _ require self consistency ( a potential @xmath11 generated by the luminosity density @xmath21 ) and they make no assumption on the df .
however , even if one assumes @xmath11 to be known ( it may be derived from the observed @xmath21 via the poisson equation ) , the two equations ( [ eq : jeans_cyl_r ] ) and ( [ eq : jeans_cyl_z ] ) are still a function of the four unknown @xmath22 , @xmath23 , @xmath24 and @xmath25 and do not uniquely specify a solution .
given that the axisymmetric jeans equations relate four functions of the meridional plane @xmath1 coordinates , one needs to specify at least two functions of @xmath1 for a unique solution . a natural way to provide a unique solution for the jeans equations consists of specifying the shape and orientation of the intersection of the velocity ellipsoid everywhere in the meridional plane .
in fact the three components @xmath22 , @xmath23 and @xmath25 , of the velocity dispersion tensor , uniquely describe the equation of the velocity dispersion ellipse , whose shape can be derived by diagonalizing the tensor .
the velocity dispersion ellipse will have the major axis at an angle @xmath26 with respect to the equatorial plane @xmath27 and an axial ratio @xmath28 given by @xmath29 the specification of the ellipse geometry ( @xmath26 and @xmath30 ) is then sufficient to write two of the three variables as a function of the remaining one .
the most common choice which is made consists of assuming a circular velocity - ellipsoid in the meridional plane ( sometimes called semi - isotropy condition ) .
this assumption implies @xmath31 and @xmath32 , and is sufficient to ` close ' the set of equations to provide a unique solution for the remaining variables @xmath23 and @xmath24 .
other common options are discussed in section [ sec : choice ] .
a well studied special case of semi - isotropic system is one in which the 3-integral df which generally characterizes a stationary system , is assumed to depend only on the two classical integrals of motion @xmath33 , where @xmath34 is the potential energy , and @xmath35 is the angular momentum with respect to the symmetry @xmath7-axis .
for this reason the semi - isotropic jeans models are often called two - integral jeans models . in this paper
we are investigating whether it is possible to make an alternative assumption on the shape of the velocity ellipsoid , which retains the simplicity of the widely used semi - isotropic assumption , while also providing a better description of real galaxies and still leading to an efficient solution of the jeans equations .
due to its simplicity , the semi - isotropy assumption has proven remarkably useful in a large variety of applications of the jeans equations .
it was used to quantify the amount of rotation in galaxies @xcite , to search for hidden disks in elliptical galaxies @xcite , to measure their mass - to - light ratio @xcite and dark matter profiles @xcite , to study the connection between line - strength in galaxies and the local escape velocities @xcite , to study the scatter in the fundamental plane @xcite , to measure the deviations of the gas kinematics from circular velocities @xcite , and to estimate the masses of supermassive black holes @xcite .
although these models have been largely superseded by @xcite orbit - superposition method , when good kinematic data are available and the maximum generality is required , they are becoming useful to study the mass - to - light ratios and rotation of galaxies at high - redshift , where the data quality still does not justify more sophisticated approaches @xcite . to explore the possibility of making a simple but sufficiently realistic assumption on the shape and orientation of the velocity ellipsoid in the meridional plane of axisymmetric galaxies , we need to understand what that shape is expected to be
. a qualitative insight into the orientation of the velocity ellipse in real galaxies can start from the analysis of the special case of separable potentials @xcite . in an axisymmetric oblate
separable potential the equations of motion for the stellar obits can be separated in a prolate spheroidal coordinates system @xmath36 which also defines the boundaries of the orbits . in other words the orbital motions can be written as the linear combination of two independent oscillations in @xmath37 and @xmath21 , plus a non - uniform rotation around the symmetry axis .
for this reason the velocity ellipse is aligned with the coordinate system and the cross term @xmath38 vanishes @xcite .
the prolate spheroidal coordinates are characterized by the fact that they tend to be aligned with the cylindrical coordinates @xmath14 at small radii and with the polar ones @xmath39 at large radii .
separable potentials are characterized by a central constant - density region .
the cylindrical alignment happens in that region , as the stars there tend to move like an harmonic oscillator .
for these reasons one expects the velocity dispersion ellipse to be cylindrically aligned at small radii and radially aligned at large radii .
the gravitational potential of real galaxies is _ not _ of the separable form .
separable potentials are in fact necessarily characterized by smooth analytic centers , while real galaxies possess central singularities due to cusped density profiles ( e.g. * ? ? ?
* ; * ? ? ?
* ) and supermassive black holes @xcite .
still , numerical integrations of orbits in non - separable axisymmetric potentials show that most of them are still bounded by curves , which qualitatively resemble the spheroidal coordinates ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for this reason one can expect the velocity dispersion ellipse to behave in real galaxies in a way that is qualitatively similar to the separable case .
numerical calculations confirm this fact in non - separable triaxial and axisymmetric potentials @xcite . in the limit of a point mass
, the velocity ellipsoid has to be spherically aligned for symmetry .
this suggests that at small radii , where the supermassive black hole or the stellar cusp dominates , the velocity ellipsoid in real galaxies should be more spherically oriented than in the separable case .
the df of a stationary system is a function of the three separable integrals of motion @xcite . in general it
can not be recovered from real galaxies without at least another three - dimensional observable quantity .
this quantity is now being provided by integral - field observations of the stellar kinematics ( e.g. * ? ? ?
* ) , which allow the stellar line - of - sight velocity - distribution to be measured at every position of the galaxy image on the sky .
we used these observations , in combination with orbit - based three - integral axisymmetric models , to measure the shape and orientation of the velocity ellipsoid at every position within the meridional plane ( within one half - light radius ) , for a sample of 25 early - type galaxies @xcite .
[ fig : coordinates ] qualitatively summarizes the main findings of that paper regarding the shape of the velocity dispersion ellipsoid for the fast - rotator galaxies : ( i ) the ellipsoids qualitatively behave like in separable potentials and already near 1 they are essentially spherically oriented ; ( ii ) the axial ratio of the ellipsoids varies gradually as a function of the polar angle , in such a way that the ellipsoid has nearly the same shape on both the equatorial plane ( where the density is the highest ) and the symmetry axis .
the net effect is that to first order the global anisotropy of the galaxies can be described as a simple flattening of the velocity ellipsoid in the @xmath7-direction ( @xmath40 ) . of the fast - rotator early - type galaxies ( derived from @xcite ) .
the ellipses show the intersection of the velocity ellipsoid with the @xmath41 plane .
the solid lines show a representative prolate spheroidal system of coordinates .
the dashed line is a representative iso - density contour .
the ellipsoids are aligned in spheroidal coordinates , but the axial ratio of the ellipses varies with the polar angle in such a way that the shape and orientation of the ellipsoids are similar on both the equatorial plane and the symmetry axis.,title="fig : " ] one can think of three physically motivated choices of the coordinates system in which to align the velocity ellipsoid , for a unique solution of the axisymmetric jeans equations . in this section
we examine advantages and problems of each one in turn .
the observed behavior of the ellipsoid could be approximated by solving the jeans equations in prolate spheroidal coordinates in a generic axisymmetric potential @xcite . to qualitatively reproduce the behavior of fig .
[ fig : coordinates ] , the simplest realistic model would need at least four free parameters .
two parameters are required to define the coordinate system .
a third parameter could define the shape of the velocity ellipse on the equatorial plane and another parameter would describe the angular variation .
a problem of this approach is that the general solution , to derive the observables for realistic galaxy potentials , requires at least a computationally - expensive triple numerical quadrature .
the derived equations are also rather cumbersome and not justified by the fact that these models can at best provide a qualitative description of real galaxies . for these reasons
those solutions have currently only been applied to a handful of analytic potentials , and no application to real galaxies exist .
given that the velocity ellipsoids in fig .
[ fig : coordinates ] are nearly spherically aligned , a much simpler choice would consist of solving the anisotropic jeans equation in spherical coordinates . a solution for this case
was presented by @xcite and applied to real galaxies by @xcite and @xcite .
however even under this spherical assumption the solution still has the same form as in the spheroidal coordinates .
the computation still involves a triple numerical quadrature .
this is in fact a generic feature of the axisymmetric jeans solution : even in the simple case in which the density is assumed to be stratified on similar oblate spheroids , one quadrature is needed to obtain the potential , a second quadrature provides the intrinsic velocity s second moments and a third quadrature finally gives the projected observables . in the spherically - oriented case
the coordinates system has no free parameters and the simplest realistic model to reproduce fig .
[ fig : coordinates ] could have two parameters ( anisotropy and its angular variation ) . a more radical alternative is to assume the velocity ellipse is oriented with the cylindrical coordinate system .
this option was tested by e.g. @xcite , however until now , for the reasons described in section [ sec : spheroidal ] , it was not considered a sensible choice to describe the shape of the velocity ellipsoid in real galaxies .
our new results , based on the ` sauron ` integral - field observations @xcite , make this choice worth exploring again .
this option can not provide a formally accurate representation of fig .
[ fig : coordinates ] , however it is accurate near the equatorial plane , where the density is at its maximum , and near the minor axis . models with cylindrically - oriented velocity ellipsoid provide a good qualitative description of the empirical observation that the global anisotropy in fast - rotator galaxies is best characterized as a flattening of the velocity ellipsoid in the vertical @xmath7-direction ( * ? ? ? * their fig . 2 ) .
the near - cylindrical orientation of the velocity ellipsoid may be due to the presence of disks , where this orientation appears natural .
for this reason the cylindrical orientation is certainly appropriate to describe the dynamics of spiral galaxies @xcite . in the next section
we show that the assumption of a cylindrically - aligned velocity - dispersion ellipsoid , combined with the powerful multi - gaussian expansion ( mge ) method of @xcite , can generate simple solutions that well reproduce the integral - field kinematics of real galaxies , and also allow for variable mass - to - light ratios ( e.g. dark matter ) , spatially varying anisotropy and multiple kinematical components .
all this while still requiring a _ single _ numerical quadrature to predict the observables on the sky plane ( or any other projection ) .
we start from the general axisymmetric jeans equations ( [ eq : jeans_cyl_r ] ) and ( [ eq : jeans_cyl_z ] ) and we make the following two assumptions : ( i ) the velocity ellipsoid is aligned with the cylindrical coordinate system @xmath14 and ( ii ) the anisotropy is constant and quantified by @xmath42 . in this case
the jeans equations reduce to @xmath43 which corresponds to the semi - isotropic case ( two - integral ) when @xmath44 . with the boundary condition @xmath45 as @xmath46 the solution reads @xmath47 + r \nu\frac{\partial\phi}{\partial r } \label{eq : jeans_sol_r}.\end{aligned}\ ] ]
a general caveat regarding the jeans equations is that the existence of a solution does not guarantee the existence of a corresponding physical positive df .
this can only be verified using different techniques . to derive solutions for the jeans equations we make an explicit choice for the parametrization of the stellar density and the total density ( which can include dark matter and a central black hole ) .
we adopt for _ both _ the mge parametrization of @xcite due to its flexibility in accurately reproducing the surface - brightness of real galaxies and the availability of robust routinesmxc / idl/ ] to fit the galaxy photometry @xcite .
if the @xmath48-axis is aligned with the photometric major axis , the surface brightness @xmath49 at the location @xmath50 on the plane of the sky can be written as @xmath51},\ ] ] where @xmath52 is the number of the adopted gaussian components , having total luminosity @xmath53 , observed axial ratio @xmath54 and dispersion @xmath55 along the major axis .
the deprojection of the surface brightness to obtain the intrinsic luminosity density is not unique unless the axisymmetric galaxy is seen edge - on ( @xmath56 ) @xcite , and the degeneracy becomes especially severe when the galaxy is seen at low inclinations @xcite .
the mge method provides a simple possible choice for the deprojection @xcite .
one of the advantages of the mge method is that one can easily enforce the roundness of the model @xcite , thus producing realistic densities , which look like real galaxies when projected at any angle .
the method can not eliminate the intrinsic degeneracy of the deprojection and this has to be considered when interpreting results of galaxies that are close to face - on . the deprojected mge oblate axisymmetric luminous density @xmath21
can be written as @xmath57,\ ] ] where the individual components have the same luminosity @xmath53 and dispersion @xmath55 as in the projected case ( [ eq : surf ] ) , and the intrinsic axial ratio of each gaussian becomes @xmath58 where @xmath59 is the galaxy inclination ( @xmath56 being edge - on ) .
the total density @xmath60 can be generally described by a different set of m gaussian components @xmath61.\ ] ] in the self - consistent case the gaussians are the same as in equation ( [ eq : dens ] ) and one has @xmath62 , @xmath63 , @xmath64 and @xmath65 , where @xmath66 is the mass - to - light ratio , which can be different for different components . in the non - self - consistent case the density can be described with the sum of two sets of gaussians : the first derived by deprojecting the surface brightness with equation ( [ eq : dens ] ) , and the second e.g. obtained by fitting a ( one - dimensional ) mge model to some adopted analytic parametrization for the dark matter ( e.g. * ? ? ? * ) .
the gravitational potential generated by the density of equation ( [ eq : mass ] ) is given by @xcite @xmath67 where @xmath68 is the gravitational constant and with @xmath69 \right\}}}{{\sqrt { 1 - ( 1 - q_j^2 ) u^2 } } } .
\label{eq : integral}\ ] ] a supermassive black hole can be modeled by adding a keplerian potential @xmath70 to equation ( [ eq : mge_potential ] ) . however , as pointed out by @xcite , an even simpler approach consists of modeling the black hole as a small gaussian in equation ( [ eq : mge_potential ] ) , having mass @xmath71 , @xmath72 and @xmath73 , where @xmath74 is the smallest distance from the black hole that one needs to accurately model ( e.g. one could choose @xmath75 ) .
now we apply the mge formalism to the solution of the axisymmetric anisotropic jeans equations of section [ sec : jeans_solution ] .
our derivation is an extension of what was done in the semi - isotropic self - consistent case ( @xmath76 and @xmath77 ) in section 3.4 of @xcite .
as already pointed out by @xcite , his equations can be used to model the kinematics of different dynamical tracers , as long as they all move in the same potential ( e.g. * ? ? ?
* ; * ? ? ?
* ) . to maintain generality we will then write the solution for the individual @xmath52 luminous gaussian components , which can then be assumed to have different anisotropy .
this fact can be used e.g. to model anisotropy gradients , or to study the anisotropy of kinematical subcomponents in galaxies ( e.g. bulge and disk ) . substituting equations ( [ eq : dens ] ) and ( [ eq : mge_potential ] ) into equations ( [ eq : jeans_sol_z ] ) and ( [ eq : jeans_sol_r ] ) , the integral in @xmath7 can be performed analytically and we obtain @xmath78_k = b_k[\nu\overline{v_z^2}]_k}\\ \label{eq : sigma_z } \lefteqn{[\nu\overline{v_z^2}]_k = 4\pi g\!\ ! \int_0 ^ 1\!\ ! \sum_{j=1}^m \frac{\sigma_k^2 q_k^2 \nu_k q_j \rho_{0j } \mathcal{h}_j(u)\ , u^2 } { 1-\mathcal{c } u^2}\ , { \mathrm{d}}u}\\ \label{eq : vel2_phi } \lefteqn{[\nu\overline{v_{\phi}^2}]_k = b_k[\nu\overline{v_z^2}]_k } \nonumber\\ & & { } + 4\pi g \int_0 ^ 1 \sum_{j=1}^m \frac{\nu_k q_j \rho_{0j } \mathcal{h}_j(u)\ , u^2 } { 1-\mathcal{c } u^2}\mathcal{d}\,r^2 { \mathrm{d}}u\\ & & { } = 4\pi g \int_0 ^ 1 \sum_{j=1}^m \frac{\nu_k q_j \rho_{0j } \mathcal{h}_j(u)\ , u^2 } { 1-\mathcal{c }
u^2}\left(\mathcal{d}\,r^2 + b_k \sigma_k^2 q_k^2\right ) { \mathrm{d}}u\nonumber,\end{aligned}\ ] ] where we defined @xmath79 , @xmath80 and @xmath81 @xmath82 u^2.\ ] ] in all the equations of this paper the index @xmath83 refers to the parameters , or the anisotropy , of the gaussians describing the galaxy s luminosity density ( equation [ [ eq : dens ] ] ) , while the index @xmath84 refers to the parameters of the gaussians describing the total mass ( equation [ [ eq : mass ] ] ) , from which the potential is obtained .
when @xmath85 is not the same for the individual luminous gaussians , the total luminosity - weighted anisotropy at a certain spatial location @xmath1 of an mge model is given by the standard definition @xcite , combined with equation ( [ eq : sigma_r ] ) : @xmath86_k } { \sum_k b_k [ \nu\overline{v_z^2}]_k}\approx 1 - \frac{\sum_k\nu_k } { \sum_k b_k\nu_k}.\ ] ] the last approximation comes from the fact that @xmath87_k$ ] , being mostly a function of the total mge potential , varies relatively little for the different gaussians , while @xmath88 can be completely different and varies by many orders of magnitude for the various luminous mge components .
this allows the global anisotropy of an mge model , at a certain spatial location in the meridional plane , to be approximately estimated from a simple luminosity - weighted sum of @xmath85 .
a similar reasoning applies to the estimation of the total @xmath89 parameter in in section [ sec : satoh ] and the total @xmath90 parameter in section [ sec : mge_spherical ] .
the intrinsic quantities have to be integrated along the line - of - sight ( los ) to generate the observables which can be compared with the galaxy kinematics .
for this we define a system of sky coordinates with the @xmath91 axis along the los and the @xmath92 axis aligned with the galaxy projected major axis ( see equation [ [ eq : surf ] ] ) .
the galaxy coordinates @xmath93 are related to the ones in the sky plane by @xmath94 where the @xmath7-axis coincides with the galaxy symmetry axis and the cylindrical radius is defied by @xmath95 .
the projected second velocity moment along the line - of - sight @xmath96 , for one luminous gaussian component , is then given by @xmath97_k = \int_{-\infty}^{\infty } \left\ { [ \nu\overline{v_z^2}]_k \cos^2 i \right . }
\nonumber\\ & & + \left.\left([\nu\overline{v_r^2}]_k\sin^2\phi + [ \nu\overline{v_\phi^2}]_k\cos^2\phi\right)\sin^2 i \right\ } { \mathrm{d}}z',\end{aligned}\ ] ] where @xmath98 , while the total observed second moment , for the whole mge model , will be @xmath99_k.\ ] ] after substitution of equations ( [ eq : sigma_r])([eq : vel2_phi ] ) , the @xmath91 integral can be written explicitly .
summing over all the @xmath52 luminous gaussian components we obtain the final expression @xmath100 } } } \nonumber \\ & & \times\exp\left\{-\mathcal{a}\left[x'^2+\frac{(\mathcal{a}+\mathcal{b } ) y'^2}{\mathcal{a } + \mathcal{b}\cos^2 i}\right]\right\}{\mathrm{d}}u,\end{aligned}\ ] ] where we defined @xmath101 and @xmath102 @xmath103}\right\}\ ] ] as expected equation ( [ eq : second_moment ] ) reduces in equation ( 63 ) of @xcite .
] to equation ( 61 ) of @xcite when @xmath76 and @xmath77 . as in the semi - isotropic case
this formula is very quick to evaluate as it still requires a single numerical quadrature and involves no special functions .
all this starting directly from a fit to the galaxy surface - brightness , without the need for numerical deprojection or psf deconvolution , as required in other approaches .
the formula has various possible applications , as it can be used to model realistic galaxies with variable anisotropy or multiple kinematic components , dark - matter , variable stellar @xmath0 and supermassive black holes ( representing the point mass with a small gaussian as described in the last paragraph of section [ sec : mge_formalism ] ) .
simple expressions , involving a single quadrature , can be derived for the second velocity moments also for the case of the keplerian potential ( [ eq : kepler ] ) of a black hole , as done in the semi - isotropic case in appendix a of @xcite .
the black hole moments can then be quadratically co - added to the ones in equation ( [ eq : second_moment ] ) to obtain the observed velocity moments for the galaxy .
we found no advantage in speed or accuracy when performing a separate calculation for the black hole and the galaxy potential .
for this reason we will then not give separate expressions for the keplerian case . using equation ( [ eq : second_moment ] ) with a black hole it is however important to use a quadrature routine which samples the integrand function at a sufficiently high number of initial points , to properly recognize the sharp peak in the integrand near @xmath104 .
here we used the vectorized adaptive quadrature algorithm of @xcite .
the second moments @xmath105 provided by equation ( [ eq : second_moment ] ) are a good approximation for the observed quantity @xmath106 , where @xmath4 is the stellar mean velocity and @xmath107 is the velocity dispersion . in @xcite we used realistic semianalytic dynamical models of galaxies and found that , to extract @xmath105 from the simulated data , one should use a single gaussian losvd and adopt as @xmath4 and @xmath107 the mean velocity and dispersion of that gaussian . due to the sensitivity of the second moments to the uncertain wings of the losvd ,
this approach is preferable than trying to extract a more complex losvd , e.g. by fitting the gauss - hermite parametrization @xcite or a fully non - parametric losvd , and numerically integrate @xmath105 from that . the projected first velocity moments @xmath108 are given by @xmath109 in this case , the two assumptions we made in section [ sec : jeans_solution ] are not sufficient any more to provide a unique prediction and therefore additional assumptions are needed .
the jeans equations ( [ eq : jeans_beta_r ] ) and ( [ eq : jeans_beta_z ] ) in fact only give a prediction for @xmath110 and one has to decide how the second moments separate into the contribution of ordered and random motion , as defined by @xmath111 this need for extra assumptions on the tangential anisotropy is a fundamental limitation of the first - moments equations , and it is the reason why one should fit the more general equation ( [ eq : second_moment ] ) to @xmath112 , and only subsequently fit any extra parameter of the first moment solution to @xmath4 , instead of simultaneously fitting the jeans solutions to both @xmath4 and @xmath107 .
the first moment equations however are very useful to quantify the amount of rotation in galaxies and for this reason have been used in the past ( section [ sec : review ] ) .
one can think of two natural options to parameterize the separation of random and ordered streaming rotation around the symmetry axis .
the first option consists of assuming a constant anisotropy , for each gaussian component , in the @xmath113 coordinates , analogously to equation ( [ eq : sigma_r ] ) @xmath114_k = c_k[\nu\overline{v_r^2}]_k,\ ] ] which would imply @xmath115_k = \left([\overline{v_{\phi}^2}]_k - c_k[\overline{v_r^2}]_k\right)^{1/2}.\ ] ] a practical limitation of this approach is that the mean velocity would depend in a non - linear way on the @xmath116 parameter . for this reason nearly all previous authors adopted , in the semi - isotropic case , a second approach , introduced by @xcite .
this consists of defining a constant @xmath83 which quantifies how much the model velocity field is scaled with respect to that of the isotropic rotator ( @xmath117 ) . in this way the first moment can be computed only once , for the isotropic rotator , and is scaled linearly with @xmath83 .
substituting equation ( [ eq : satoh2 ] ) into equation ( [ eq : first_moments_projection ] ) , and using equations ( [ eq : sigma_r])([eq : vel2_phi ] ) , we obtain the projected first velocity moment of the whole mge model @xmath130^{1/2}\!\!\!\ !
{ \mathrm{d}}z'.\nonumber\end{aligned}\ ] ] when both @xmath131 and @xmath77 , this equation reduces to a self - consistent isotropic rotator as in equation ( 59 ) of @xcite .
a double quadrature seems unavoidable here , but when @xmath89 is assumed to be constant for the whole mge , this integral has to be evaluated only once with @xmath132 , at the best fitting @xmath133 parameters previously determined from a fit to the more general second moment equation ( [ eq : second_moment ] ) , and then @xmath134 can be linearly scaled by @xmath89 to fit the data . )
is used instead of equation ( [ eq : satoh ] ) , the term @xmath135 in equation ( [ eq : first_moments_projection_final ] ) becomes @xmath136 $ ] . ] the cylindrically oriented assumption for the shape of the velocity ellipsoid discussed in section [ sec : cylindrical_jeans ] is likely to be unrealistic for slow - rotator elliptical galaxies . as a class these objects
are weakly triaxial , and indeed some of them show clear twists in their kinematical axes @xcite .
the most general and realistic models for these objects are triaxial and orbit- or particle - based @xcite .
however also in the triaxial approximation a degeneracy in the recovery of the galaxy shape is still generally present so that no unique solution can be obtained .
when one is only interested in global galaxy quantities , or to test the results of more general models , it is still useful to construct simpler and approximate models .
the isophotes of the slow rotators are in projection close to circular , especially in their central parts where the kinematics is generally obtained , and this implies they must be intrinsically not far from spherical .
this suggests that one can use simple spherical models as a first order approximation to the dynamics of at least some of these objects .
the spherical solution of the anisotropic jeans equations , in the mge formalism , will be discussed in the following sections . when the boltzmann equation ( [ eq : boltzmann ] ) is written in spherical coordinates @xmath39 , by analogy with our derivation of the axisymmetric jeans equations one can obtain the jeans equations in spherical symmetry ( @xcite ; equation [ 4 - 54 ] of bt ) @xmath137 where @xmath138 for symmetry and we defined @xmath139 .
the solution of this linear first - order differential equation with constant anisotropy @xmath90 and the boundary condition @xmath140 as @xmath141 is given by ( e.g. * ? ? ?
* ) @xmath142 considering that @xmath143 .
after projection along the line - of - sight @xmath91 , we obtain the observed second velocity moment ( see equation [ 4 - 60 ] of bt ) @xmath144 \ , { \mathrm{d}}r\nonumber,\end{aligned}\ ] ] where @xmath145 is the projected radius , measured from the galaxy center . integrating by parts the double integral
can be reduced to a single quadrature , involving special functions @xmath146 where @xmath147,\end{aligned}\ ] ] with @xmath148 , @xmath149 is the gamma function and @xmath150 is the incomplete beta function ( equation [ 6.6.1 ] of @xcite ) . in the isotropic limit @xmath151 and equation ( [ eq : sph_jeans ] ) reduces to the spherical isotropic form of equation ( 29 ) of @xcite . some numerical implementations provide the incomplete beta function only for positive arguments ( e.g. * ? ? ? * ) , for negative values ( @xmath152 ) one can use its analytic continuation via gauss s hypergeometric function @xmath153 ( equation [ 6.6.8 ] of @xcite ) @xmath154 for which efficient routines exist @xcite . for @xmath155 ,
the @xmath150 function is divergent , but for all practical purposes we found it is sufficient to perturb the @xmath90 value by a negligible amount to avoid the singularity .
we proceed as in section [ sec : mge_solution ] to derive an explicit solution for the spherical jeans equation using the mge parametrization for both the luminosity density and the total density .
the expressions for the surface brightness , luminosity density and total density are as in equation ( [ eq : surf ] ) , ( [ eq : dens ] ) and ( [ eq : mass ] ) , with @xmath156 @xcite . for a single gaussian component they read @xmath157
@xmath158 @xmath159 the mass of a gaussian contained within the spherical radius @xmath160 is @xmath161,\ ] ] with @xmath162 the error function ( equation [ 7.1.1 ] of @xcite ) .
using equation ( [ eq : sph_jeans ] ) and summing over all gaussian components with equation ( [ eq : sum_vz2 ] ) , we obtain the projected second velocity moment for the whole mge model @xmath163 { \mathrm{d}}r,\ ] ] where @xmath164 and @xmath165 are given by equation ( [ eq : dens_sph ] ) and ( [ eq : massr ] ) respectively , and @xmath166 is obtained by replacing the @xmath90 parameter in equation ( [ eq : spherical_anisotropic_final ] ) with the anisotropy @xmath167 of each luminous gaussian component .
we explicitly included the mass @xmath168 of a central supermassive black hole .
as in the axisymmetric case , this formula involves a single numerical quadrature .
it can be used to model nearly spherical objects with a variable anisotropy profile , variable stellar @xmath0 , a supermassive black hole and dark matter . an approximate way to construct a model with a certain smooth radial profile of anisotropy @xmath169 ,
consists of defining @xmath170 for each mge gaussian with dispersion @xmath55 .
this simple method works because in a spherical mge each @xmath83-th gaussian produces a significant contribution to the total luminosity density only near its @xmath55 ( @xcite ; see also equation [ [ eq : var_beta ] ] ) .
the jeans anisotropic mge ( jam ) package of idl procedures , providing a reference implementation for the equations described in this section , together with other routines to evaluate auxiliary quantities , like the circular velocity , from mge models , is available online from http://www - astro.physics.ox.ac.uk/@xmath171mxc / idl/.
the formalism derived in sections [ sec : los_mu2 ] and [ sec : satoh ] generalizes to anisotropy the widely used semi - isotropic formalism for the solution of the jeans equations while maintaining its simplicity .
it was motivated by our modeling results on the orbital distribution of a sample of real galaxies @xcite .
however the method would be of limited usefulness if it did not describe the dynamics of fast - rotator early - type galaxies well . in this section
we show that the simple assumptions we made provide a remarkable good description of the main features of the kinematics @xcite of the fast rotator early - type galaxies , as observed with the ` sauron ` integral - field spectrograph @xcite as part of the ` sauron ` survey @xcite . in this section
we compare the anisotropy derived from the jeans models , with the results for the global anisotropy obtained with the more general @xcite method on the same galaxies and from the same data in @xcite .
we use equation ( [ eq : second_moment ] ) , with constant anisotropy @xmath172 and constant mass - to - light ratio @xmath173 for the whole mge model , to predict the velocity second moments . following standard practice we parameterize the anisotropy with the variable @xmath174 . the problem becomes a function of two nonlinear variables @xmath175 and of the scaling factor @xmath173 . to further limit the number of free variables in the model , and to eliminate the degeneracy in the deprojection of the surface brightness , we select for the comparison the galaxies for which the photometry already constrains the inclination to be close to edge - on ( @xmath56 ) . for this
we select from the mge models in table b1-b2 of @xcite , the fast - rotator galaxies for which the flattest gaussian has @xmath176 , which forces a strict limit on the inclination @xmath177 .
we further exclude the galaxy ngc 4550 , due to the presence of two counter - rotating stellar disks , which complicate the interpretation of the models , and ngc 4526 , which has a strong dust disk affecting the observed stellar kinematics .
this selection leads to the four galaxies ngc 821 , ngc 3377 , ngc 4621 and ngc 5845 .
we evaluated the mge model predictions given by equation ( [ eq : second_moment ] ) , adopting @xmath178 as in @xcite , for a direct comparison with the anisotropy determinations .
the model predictions were convolved with the psf and integrated over the pixels , before comparison with the observables , as described in appendix [ sec : psf ] . for each galaxy we computed the model predictions for @xmath179 inside each voronoi bin on the sky , at different anisotropy values @xmath3 . at every anisotropy
the best fitting mass - to - light ratio @xmath173 is obtained from the simple scaling relation @xmath180 as a linear least - squares fit .
the best - fitting scaling factor for the model velocities is @xmath181 , where @xmath182 is the cosine of the angle between the data @xmath183 and model @xmath184 vectors , which implies : @xmath185 where the vectors @xmath183 and @xmath184 have elements @xmath186_n$ ] and @xmath187_n$ ] , with @xmath188_n\equiv\sqrt{v_n^2+\sigma_n^2}$ ] the measured values for the @xmath189 voronoi bins from @xcite and @xmath190_n$ ] the corresponding errors , while @xmath191_n$ ] are the model predictions for @xmath192 . a qualitative comparison between the observed velocity second moments and the model ones , for different values of the anisotropy parameter @xmath3 is shown in fig . [ fig : models_beta ] . , describing the agreement between the data and the models ,
is plotted against the anisotropy parameter @xmath3 .
the four panels show the galaxies ngc 821 , ngc 3377 , ngc 4621 and ngc 5845 .
the solid line with open circles represent the @xmath193 of the jeans models .
the dashed horizontal line indicates the level @xmath194 , which corresponds to the @xmath195 confidence level for one degree - of - freedom .
the filled square with error bars shows the value of the global anisotropy measured using schwarzschild models in @xcite.,title="fig : " ] to ease the visual comparison , given that the models are bi - symmetric by construction , the observations have been bi - symmetrized with respect to the kinematical major axis pa@xmath196 given in @xcite . the actual fits and @xmath197 calculations are always performed using the original data and errors .
it is clear that the isotropic models ( @xmath198 ) do not provide a good description for the observed second velocity moments of any of the four galaxies , the disagreement being strongest for ngc 3377 and weakest for ngc 4621 . using the models presented in this paper we can now move away from the isotropic assumption and quantify that the model of ngc 4621 requires a small amount of anisotropy to appear qualitatively like the data : any @xmath199 produces in fact two vertical ` lobes ' along the minor axis , which are not observed .
both ngc 821 and ngc 5845 require a more significant anisotropy @xmath200 to reduce the amount of rotation ( high @xmath201^{1/2}$ ] ) along the major axis to the level of the observations .
a larger @xmath202 is strongly excluded by the data , as the model immediately produces a significant elongation along the minor axis , which is not observed .
finally , in the case of ngc 3377 , only the most extreme anisotropy shown ( @xmath202 ) is able to reduce the gradient of @xmath179 along the minor axis to the shallow level of the observations . to quantify more precisely the findings of the previous paragraph , in fig .
[ fig : chi2_beta ] we show the plots of @xmath203 , describing the agreement between the data and the models , as a function of @xmath3 .
we also show , as a filled square with error bars , the value of the global anisotropy we measured using more general axisymmetric schwarzschild s models in @xcite .
the @xmath193 plots confirm the visual impression from the maps of fig .
[ fig : models_beta ] and the fact that the jeans models give a tight constrain to the anisotropy
. the small size of the jeans formal errors is of course illusory , as it is due to the restrictive model assumptions .
however the comparison with the more general results show that the simple anisotropic jeans models are able to recover the anisotropy within the errors .
there seems to be a systematic difference of @xmath204 between the anisotropic jeans and the schwarzschild determinations , with the jeans results being lower .
this difference is at the level of the errors in the schwarzschild models ( see section 4.3 in @xcite ) , and not particularly surprising given the radically different modeling approaches adopted . if real , it may be due to the fact that the jeans models force a constant anisotropy everywhere in the galaxy model , while the schwarzschild ones allow for a more realistic and more spherical velocity ellipsoid at intermediate latitudes ( @xmath205 ; see fig .
[ fig : coordinates ] ) . the jeans models may need a lower anisotropy to compensate the mismatch with the data at these intermediate latitudes .
when studying the stellar population or the dark - matter content of galaxies , one is interested more in the @xmath0 determination than in the anisotropy .
it is important to test whether an error in the anisotropy can lead to a significant error in @xmath0 . for this in fig .
[ fig : ml_variation ] we show the variation in the dynamical @xmath0 ( @xmath206-band ) for each of the models shown in fig .
[ fig : models_beta ] . also shown for reference is the @xmath207 determined from the same ` sauron ` data in @xcite , using three integral schwarzschild models .
the comparison shows that the @xmath0 varies very little for the ranges of anisotropy observed in fast rotators early - type galaxies . in all cases the jeans @xmath0 at the best - fitting @xmath3
represents an improvement over the @xmath0 determined from semi - isotropic models ( already given in table 1 of @xcite ) . with the exception of ngc 5845
, the improvement is generally at the level of the measurement errors .
the latter galaxy is an interesting case on its own : it has an small size with @xmath208 @xcite and constitute a unique example of an elliptical galaxy with integral - field stellar kinematics out to @xmath209 .
our standard self - consistent model still well reproduces the ` sauron ` kinematics without the need to invoke a dark matter halo .
the small dependence of @xmath0 on the anisotropy explains the fact that , in a comparison between the semi - isotropic @xmath210 and the three - integral @xmath207 for a sample of 25 early - type galaxies ( including the ones in fig .
[ fig : ml_variation ] ) , we found an excellent agreement between the two determinations .
we did not detect any significant bias in the jeans values and the differences could be explained by random errors at the 6 per cent rms level ( * ? ? ?
* their fig . 7 ) . versus anisotropy
the different symbols , connected by lines , show the variation with anisotropy @xmath3 in the dynamical @xmath0 ( @xmath206-band ) for each of the jeans models shown in fig .
[ fig : models_beta ] . also shown for reference , with horizontal lines using the same style and color , is the @xmath207 , determined from the same data in @xcite , using three - integral schwarzschild models .
the symbols represent ngc 821 ( green circles ) , ngc 3377 ( blue upward triangles ) , ngc 4621 ( black downward triangles ) and ngc 5845 ( red squares ) .
the filled symbols correspond to the best - fitting @xmath3 from fig .
[ fig : chi2_beta ] .
the large cross shows the characteristic error in @xmath3 and @xmath207.,title="fig : " ] in the previous section we showed that the anisotropic jeans formalism introduced in this paper provides a good qualitative description of the integral - field second velocity moments observed with ` sauron ` .
we also showed that for edge - on galaxies the recovered anisotropy agrees well with the one derived from more general models .
here we present additional examples of model fits to the second moments and we show that , using the simple @xcite assumption for the splitting of the random and ordered rotation ( section [ sec : satoh ] ) , a remarkably good prediction of the galaxies mean velocity can be obtained as well , in many cases . in fig .
[ fig : models_velocity ] the data - model comparison is presented for both the first and second velocity moments .
the model fits were determined with the following procedure : 1 . for each pair of the nonlinear model parameters @xmath175 we evaluated the predicted second moments @xmath179 using equation ( [ eq : second_moment ] ) , with @xmath211 and @xmath212 .
the model predictions were scaled by the best - fitting mass - to - light ratio @xmath173 via equation ( [ eq : ml_fit ] ) and the @xmath197 , describing he agreement of data and models was determined ; 3 .
step ( i)(ii ) were repeated for a grid of @xmath175 parameters and the best fitting value was found . in the grid ,
the inclination was sampled at equally spaced intervals in the intrinsic axial ratio of the flattest mge component , while @xmath3 was sampled linearly ; 4 .
the gaussians describing the mass of the mge ( equation [ [ eq : mass ] ] ) were scaled by the best fitting @xmath173 and a prediction for the first velocity moment @xmath134 was computed with equation ( [ eq : first_moments_projection_final ] ) , with @xmath132 ; 5 .
the model velocity field @xmath134 was scaled by the factor @xmath213_n|},\ ] ] where @xmath189 is the number of voronoi bins in the observed ` sauron ` data and @xmath214 is the corresponding flux . @xmath215 and @xmath216_n$ ] are the observed and model velocities respectively , and @xmath217 are the bin coordinates ( the @xmath92 axis being along the projected major axis ) .
the last step ensures that the model has the same projected angular momentum as the observed galaxy , within the observed region .
doing a normal least - squares fit would heavily underestimate the amount of rotation of the model with respect to the data , in cases where the galaxy contains counter - rotating stellar components , which are clearly excluded by the simple assumption of a constant @xmath89 factor for the whole mge model .
the angular momentum of the model could be computed by numerically integrating @xmath218 over the galaxy volume using equation ( [ eq : vel2_phi ] ) .
more useful may be to use the fitted galaxy inclination to deprojected the _ observed _ galaxy stellar angular momentum per unit mass @xcite .
to model kinematically distinct stellar components one could allow for different @xmath121 components in equation ( [ eq : first_moments_projection_final ] ) , and perform a standard least - squares fit as in equation ( [ eq : ml_fit ] ) . for the examples of fig .
[ fig : models_velocity ] we selected three ellipticals and three lenticulars fast - rotator showing a range of kinematical properties .
for all the galaxies the mge models were fitted to the observed photometry using the public software@xmath219 of @xcite .
although some differences between the data and models are visible , the @xmath197 per degree - of - freedom in the fit presented , computed from original non - symmetrized data , is close to one for all the fits .
this indicates that the models are generally consistent with the data , within the measurements errors , which is remarkable for what is essentially a one parameter model !
a more general statistical investigation of the kinematic parameters for a larger sample of galaxies goes beyond the scope of the present paper and will be presented elsewhere .
here we note that for the galaxies presented in fig .
[ fig : models_velocity ] and [ fig : models_beta_inc ] the parameter @xmath220 within a few percent accuracy ( except for ngc 821 ) .
this appears a general characteristics also of the other fast - rotators we modeled .
it implies @xmath221 and confirms the results of more general models .
it indicates that the velocity ellipsoid of fast - rotator early - type galaxies tends to be oblate ( * ? ? ?
* their fig . 2 ) .
interestingly fig .
[ fig : models_velocity ] shows that even the complex double - hump structure observed in the velocity fields of ngc 2549 , ngc 4660 or ngc 5845 , which seems to imply a complex dynamical structure in the galaxy , can be well reproduced by these one - parameter models , once an accurate mge parametrization for the surface brightness is given .
the structures are explained by the presence of thin stellar disks , already visible in the photometry as strongly disky isophotes .
there is no evidence for a significantly different anisotropy between the bulge and the disk of these galaxies , as a constant anisotropy well reproduces the observations ( within 1 ) .
these examples show that a great deal of information on the kinematical structure of the fast - rotators is contained in the photometry alone !
we have shown in the previous sections that the anisotropic jeans axisymmetric models presented in this paper , varying essentially only one free parameter @xmath3 , provide a good qualitative description of the observed shape of the first and second velocity moments of fast - rotator early - type galaxies , when they are already constrained by the photometry to be close to edge - on .
we also confirmed that the fast - rotators are characterized by a velocity ellipsoid which is flattened in the @xmath7 direction so that the anisotropy @xmath6 @xcite . in this section
we study the variations of the models when the inclination is allowed to vary .
in @xcite we showed , using real data and analytic tests , that there appears to be a degeneracy in the recovery of the galaxy inclination ( or its corresponding shape ) using general three - integral axisymmetric models . this is true even in the case of state - of - the - art integral - field data , and assuming the galaxy potential is accurately known at every inclination and can be uniquely recovered from the surface brightness .
this result was confirmed for a larger sample of galaxies in @xcite , and in the oblate limit , but using a triaxial modeling code , by @xcite . in a realistic situation both the galaxy potential and the stellar luminosity density
are only approximately known , due to the important intrinsic degeneracy in the deprojection of the surface brightness , due to the possible presence of bars , and due to the likely contribution of a ( small ) fraction of dark - matter .
moreover the kinematical data are always affected by low - level systematics , which are difficult to control and quantify .
the dynamical models are also affected by numerical approximations and discretisation effects .
all this makes it unlikely that any inclination derived from dynamical models of the stellar kinematics can be trusted , unless further assumptions are made . in this section
we explore whether the anisotropic jeans models , combined with realistic assumptions on the galaxy anisotropy , can be used to recover the galaxy inclination . for our experiments on the recovery of the inclination we selected the same four galaxies with ` sauron ` integral - field kinematics that we presented for the same reason in appendix a of @xcite .
these galaxies have a photometry which , under the axisymmetric assumption , can be deprojected for a wide range of inclinations .
most importantly they possess disks of gas or dust , which allow the inclination to be determined independently from the stellar dynamics . in that paper we found that the inclination derived using semi - isotropic models appeared in agreement with the one from the dust disks . here
we try to understand the reason for the good agreement .
our prototypical test case is the fast - rotator e4 galaxy ngc 2974 .
the inclination of this galaxy was studied extensively by a number of authors using the kinematics of the gas ( e.g. * ? ? ?
* ) . from accurate models of the ` sauron ` gas kinematics in the regions where we have our stellar kinematics @xcite
find a best fitting inclination @xmath222 .
a sequence of model predictions for the velocity second moments @xmath179 as a function of anisotropy @xmath3 and inclination @xmath59 is shown in fig . [ fig : models_beta_inc ] .
a strong similarity is evident between the effect on the models of the anisotropy and of the inclination .
another general feature one can see from the figure is that the second moments are more weakly sensitive to an anisotropy variation when the galaxy is intrinsically flat ( @xmath223 ) than when it is rounder ( @xmath224 ) .
a thin disk can be more strongly anisotropic without producing the vertical elongation of the second moments ( see @xmath225 ) , which is never observed in real galaxies .
the observed behavior can be qualitatively understood as follows : a more face - on view makes the model intrinsically flatter and disk - like , causing the second moments to approach the @xmath226 behavior along the galaxy isophotes as one would expect for a thin rotating disk , with a peak on the projected major axis and a strong minimum on the minor axis .
a similar increase of the second moments on the projected major axis , with respect to the minor axis , can be produced by reducing the fraction of stars on radial orbits ( lowering @xmath3 ) , thus correspondingly increasing the stars on tangential orbits .
due to projection these in fact produce zero motion along the minor axis and show a peak on the major axis . for example in fig .
[ fig : models_beta_inc ] the model predictions with @xmath227 are very similar to the ones for @xmath228 .
the contours of the @xmath197 , describing the agreement between the jeans models and the observed ` sauron ` @xmath112 are shown in fig .
[ fig : chi2_beta_inc ] . within a 3@xmath107 confidence level
any inclination between @xmath229 and @xmath230 , and correspondingly any anisotropy between @xmath231 and 0.1 is equally consistent with the data . , describing the agreement between the data and the anisotropic jeans models ,
are plotted as a function of the anisotropy parameter @xmath3 and the galaxy inclination @xmath59 in degrees .
the lowest levels correspond to the formal 1 , 2 and 3@xmath107 ( thick red line with @xmath232 ) confidence levels for two degrees of freedom .
additional contours are characterized by a factor of 2 increase in @xmath193 .
the two horizontal blue solid lines enclose the region of acceptable inclinations as determined from the gas kinematics or dust geometry ( appendix a of @xcite ) .
the vertical solid green line indicates isotropy ( @xmath198 ) .
the different panels correspond to ngc 4150 , ngc 2974 , ngc 4459 and ngc 524 respectively .
there is a clear degeneracy between inclination and @xmath3 anisotropy .
however if one enforces the observationally - motivated constraint @xmath233 , the recovered inclination always lies within the error bars of the gas / dust determinations.,title="fig : " ] in ( * ? ? ?
* their fig .
2 ) we found that all the fast - rotators have anisotropy @xmath234 .
for the nearly edge - on galaxies we confirmed this determination using jeans models in section [ sec : comp_schw ] and section [ sec : velocity_examples ] . forcing this observationally - motivated constraint , the inclination of ngc 2974 becomes tightly constrained to the range @xmath229 to @xmath235 .
the qualitative behavior seen in ngc 2974 is representative of what we observed for the 48 galaxies we modeled using this method from the ` sauron ` sample ( scott et al . in preparation ) . in fig .
[ fig : chi2_beta_inc ] we show the @xmath197 contours for the three additional galaxies . in the case of ngc 4150
the inclination is well constrained by the data independently of anisotropy , while in the case of ngc 4459 and ngc 524 the best fitting location has @xmath236 and the inclination is heavily dependent on the assumed anisotropy .
for these two galaxies the un - physical location of the @xmath197 minimum is likely due to the degeneracy in the deprojection of the surface brightness ( section [ sec : mge_formalism ] ) .
these two galaxies have a quite low inclination and the deprojected mass distribution is probably significantly in error even at the true inclination , which is then formally excluded by the data .
like all the fast - rotators , these galaxies likely possess stellar disks , which can not be detected at these low inclinations @xcite .
this would explain the general fact that we tend to better reproduce the kinematics of the fast - rotators which are close to edge - on than the kinematics of the more face - on ones .
nonetheless , in all the four cases , if one enforces the constrain @xmath234 , the recovered inclination lies within the error bars of the independent gas / dust determinations .
this indicates that one can use these anisotropic jeans models to recover the inclinations of the fast - rotators by enforcing the requirement @xmath233 , and this explains the good agreement with the determinations from isotropic models we found in @xcite .
the jeans equations are useful when one needs a predictions for the spatial variation of the kinematics in galaxies .
however the stars in galaxies also satisfy more general global relations between kinematic observables , the most important of which are the virial equations @xcite .
these are a tensor generalization of the well known virial theorem @xmath237 , which relates the total kinetic energy @xmath238 and the potential energy @xmath239 of an isolated stellar system in equilibrium .
the small dependence of the measured @xmath0 on the anisotropy ( section [ sec : ml_comparison ] ) can be qualitatively understood as a consequence of the tensor virial equations . for an axisymmetric galaxy in equilibrium
the total kinetic energy is proportional to the galaxy mass .
the ratio of the kinetic energy @xmath240 along the direction of the symmetry @xmath7-axis and the kinetic energy @xmath241 along a direction orthogonal to it , is only a function of the galaxy shape ( equation [ 4 - 91 ] of @xcite ) .
the effect of increasing the anisotropy @xmath3 is simply that of redistributing the same kinetic energy @xmath241 from the tangential to the radial direction .
this reduces the velocity second moments along the galaxy projected major axis , while correspondingly increasing them along the minor axis .
the projected luminosity - weighted second moments , from which the @xmath0 is measured , remain nearly unchanged when averaged over the full galaxy image .
the virial equations also provide a qualitative understanding of why the observations can constrain the inclination of an axisymmetric galaxy ( section [ sec : inc_recovery ] ) .
in fact they state that the ratio @xmath242 increases with galaxy flattening .
in other words , for a given projected surface brightness , a flat system needs more kinetic energy in a direction parallel to its the equatorial plane , to be in hydrostatic equilibrium . the fast rotator ngc 2974 of fig .
[ fig : models_beta_inc ] shows high velocity second moments along the projected major axis .
although the galaxy does not appear very flat in projection , its kinematics is already an indication that the system is likely to be intrinsically flat .
a flat model has more kinetic energy in the equatorial plane and will be able to easily reproduce the high @xmath112 along the major axis , without invoking extreme tangential anisotropy . a quantitative example , using both the first and second velocity moments , of how the tensor virial equations can be used to constrain the galaxy inclination is presented in fig .
11 of @xcite .
there we showed that the diagram of @xcite can be used to provide a qualitative estimate of the inclination of fast - rotators galaxies , given some assumptions on the anisotropy . although the virial equations can be used to estimate the @xmath0 and the inclination of galaxies , they have some obvious limitations : ( i ) they do not allow one to make full use of the spatial information contained in the integral - field observations .
the whole two - dimensional kinematics is reduced to one number ; ( ii ) they do not allow the limited spatial extent of the kinematics to be properly taken into account ; ( iii ) they do not easily deal with non - similar isophotes and multiple photometric components in galaxies .
we present a generalization of the widely used semi - isotropic ( two - integral ) axisymmetric jeans modeling method to describe the stellar dynamics of galaxies .
our method uses the powerful multi - gaussian expansion ( mge ) technique to accurately parameterize the galaxies photometry .
it represents an anisotropic extension of what was presented in the semi - isotropic case by @xcite , and it maintains its simplicity and computational efficiency .
we assume ( i ) a constant mass - to - light ratio and ( ii ) a velocity ellipsoid which is aligned with the cylindrical @xmath1 coordinates and has a flattening quantified by the classical @xmath7-anisotropy parameter @xmath2 , where @xmath7 is the galaxy symmetry axis .
we test the technique using ` sauron ` integral - field observations of the stellar kinematics @xcite for a small set of fast - rotator galaxies with a variety of kinematical properties .
for galaxies that are constrained by the photometry to be close to edge - on we find that , although in the semi - isotropic limit ( @xmath198 ) the models do not provide a good description of the data , the variation of the _ single _ global anisotropy @xmath3 is generally sufficient to accurately predict the shape of both the first ( @xmath4 ) and second ( @xmath243 ) velocity moments , once an detailed mge parametrization of the photometry is given . an accurate description of the photometry , including ellipticity variations and disky isophotes , appears crucial to reproduce in detail the features of the kinematics .
in all cases we find that @xmath234 , while generally @xmath221 with good accuracy .
this confirms previous findings on the dynamical structure of these galaxies , showing that their velocity ellipsoid tends to be oblate @xcite .
the anisotropy derived with our anisotropic jeans dynamical modeling method agrees within the errors with the one previously measured using a more general schwarzschild approach . for fast - rotator galaxies that are _
not _ constrained by the photometry to be close to edge - on , we find that in general the inclination @xmath59 ( or the corresponding galaxy shape ) and the anisotropy @xmath3 are highly correlated and can not be independently determined .
however , if we introduce the observationally - motivated constraint @xmath234 , the inclination becomes constrained to a narrow range of values and it agrees with independent determinations when those are available .
we are applying this method to determine the inclination , the mass - to - light ratio and the amount of rotation of a large sample of galaxies for which integral field kinematics are available . for galaxies close to edge - on , the global anisotropy or the dynamical structure of different galaxy subcomponents ( e.g. bulge and disk ) can also be investigated .
we are using this method to test independent determinations of the masses of supermassive black holes in the nuclei of fast - rotator galaxies .
this technique is ideal to study the dark matter content and the anisotropy of disks of spiral galaxies .
i acknowledge fruitful discussions and feedback from eric emsellem , davor krajnovi and glenn van de ven .
i am grateful to roger davies , tim de zeeuw and remco van den bosch for constructive criticism on the draft .
i thank nicholas scott for providing the mge model of three of the galaxies presented in fig .
[ fig : models_velocity ] , before publication .
i am grateful to the anonymous referee for useful comments .
i acknowledge support from a stfc advanced fellowship ( pp / d005574/1 ) .
the ` sauron ` observations were obtained at the william herschel telescope , operated by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
m. , stegun i. a. , 1965 , handbook of mathematical functions with formulas , graphs , and mathematical tables .
dover books on advanced mathematics , new york : dover , 1965 , corrected edition , edited by abramowitz , milton ; stegun , irene a. r. , 1995 , mnras , 276 , 293 r. , 1985 , a&a , 143 , 84 r. , et al .
, 2001 , mnras , 326 , 23 r. , simien f. , monnet g. , 1983 , a&a , 128 , 405 e. f. , de jong r. s. , 2001 , apj , 550 , 212 o. , 1991 , apj , 366 , 599 f. , cinzano p. , corsini e. m. , rix h .- w .
, zeilinger w. w. , 1995 , apj , 448 , l13 j. , 2005 , mnras , 363 , 937 j. , mamon g. a. , 1982 , mnras , 200 , 361 binney j. , tremaine s. d. , 1987 , galactic dynamics .
princeton univ . press , princeton , nj j. j. , davies r. l. , illingworth g. d. , 1990 , apj , 361 , 78 m. , 2002 , mnras , 333 , 400 m. , et al . , 2006 , mnras , 366 , 1126 m. , et al . , 2007 , mnras , 379 , 418 c. m. , danziger i. j. , 1994 , mnras , 270 , 523 c. m. , de zeeuw p. t. , van der marel r. p. , danziger i. j. , qian e. e. , 1995 , apj , 441 , l25 p. , rix h .- w . , sarzi m. , corsini e. m. , zeilinger w. w. , bertola f. , 1999 , mnras , 307 , 433 p. , van der marel r. p. , 1994 ,
mnras , 270 , 325 e. m. , et al . , 1999 , a&a , 342 , 671 corts j. r. , kenney j. d. p. , hardy e. , 2008 , apj , 683 , 78 n. , de zeeuw p. t. , van der marel r. p. , rix h .- w . , 1999 ,
apjs , 124 , 383 n. , van den bosch f. c. , 1999 , apj , 514 , 704 r. l. , sadler e. m. , peletier r. f. , 1993 , mnras , 262 , 650 f. , debattista v. p. , gerhard o. , sambhus n. , 2007 , mnras , 376 , 71 p. t. , et al . , 2002 , mnras , 329 , 513 p. t. , hunter c. , 1990 , apj , 356 , 365 t. , 1985 , mnras , 216 , 273 w. , gerhard o. e. , 1993 , mnras , 261 , 311 h. , de zeeuw t. , 1988 , apj , 333 , 90 a. s. , 1915 , mnras , 76 , 37 e. , et al . , 2007 , mnras , 379 , 401 e. , et al . , 2004 , mnras , 352 , 721 e. , monnet g. , bacon r. , 1994 , a&a , 285 , 723 n. w. , de zeeuw p. t. , 1994 , mnras , 271 , 202 n. w. , lynden - bell d. , 1991 , mnras , 251 , 213 s. m. , et al . , 2007 , apj , 665 , 265 l. , merritt d. , 2000 , apj , 539 , l9 l. , van den bosch f. c. , ford h. c. , jaffe w. , oconnell r. w. , 1994 , aj , 108 , 1598 j. a. , 1986 , aj , 91 , 1096 k. , et al . , 2000 , apj , 539 , l13 o. , kronawitter a. , saglia r. p. , bender r. , 2001 , aj , 121 , 1936 o. e. , 1993 , mnras , 265 , 213 o. e. , binney j. j. , 1996 , mnras , 279 , 993 j. , kuijken k. , merrifield m. r. , 1997 , mnras , 288 , 618 j. , kuijken k. , merrifield m. r. , 2000 , mnras , 317 , 545 j. h. , 1915 , mnras , 76 , 70 j. h. , 1922 , mnras , 82 , 122 c. l. , et al . , 2001 , apj , 550 , 668 s. a. , de jong r. s. , weiner b. j. , 2006 , apj , 643 , 804 c. s. , rybicki g. b. , 1996 , mnras , 280 , 1257 l. v. e. , treu t. , bolton a. s. , burles s. , moustakas l. a. , 2006 , apj , 649 , 599 j. , bender r. , 1996 , apj , 464 , l119 + d. , cappellari m. , emsellem e. , mcdermid r. m. , de zeeuw p. t. , 2005 , mnras , 357 , 1113 krajnovi d. , et al . , 2008 , mnras , 390 , 93 b. , ciotti l. , 2003 , a&a , 404 , 819 t. r. , et al .
, 1995 , aj , 110 , 2622 j. , 1999 , mnras , 302 , 530 j. , et al . , 1998 , aj , 115 , 2285 d. , 1980 , apjs , 43 , 435 g. , bacon r. , emsellem e. , 1992 , a&a , 253 , 366 r. , miyamoto m. , 1976 , pasj , 28 , 1 j. f. , frenk c. s. , white s. d. m. , 1996 , apj , 462 , 563 noordermeer e. , merrifield m. r. , aragn - salamanca a. , 2008 , mnras , 388 , 1381 p. , williams t. b. , 2000 , aj , 120 , 2884 m. , salucci p. , stel f. , 1996 , mnras , 281 , 27 w. h. , teukolsky s. a. , vetterling w. t. , flannery b. p. , 1992
, numerical recipes in fortran . the art of scientific computing .
cambridge : university press , 1992 , 2nd ed .
e. e. , de zeeuw p. t. , van der marel r. p. , hunter c. , 1995 , mnras , 274 , 602 d. o. , 1982 , apj , 252 , 496 a. , lanzoni b. , bonoli s. , ciotti l. , 2005 , a&a , 443 , 133 h .- w .
, carollo c. m. , freeman k. , 1999 , apj , 513 , l25 h .- w . , white s. d. m. , 1990 , apj , 362 , 52 h .- w . , white s. d. m. , 1992 , mnras , 254 , 389 a. j. , kochanek c. s. , 1997 , mnras , 287 , 35 d. , kochanek c. s. , 2005 , apj , 623 , 666 g. b. , 1987 , in de zeeuw p. t. , ed . , structure and dynamics of elliptical galaxies
127 of iau symposium , deprojection of galaxies - how much can be learned .
p. 397
c. , 1980 , pasj , 32 , 41 m. , 1979 , apj , 232 , 236 l. f. , 2008 , j. comp .
, 211 , 131 z. , jianming j. , 1996 , computation of special functions .
new york , wiley k. l. , gerssen j. , van der marel r. p. , 2003
, aj , 126 , 2707 v. , et al . , 2005 , nature , 435 , 629 t. s. , dejonghe h. , smecker - hane t. , 1999 , aj , 117 , 126 j. , saglia r. p. , bender r. , thomas d. , gebhardt k. , magorrian j. , corsini e. m. , wegner g. , 2007 , mnras , 382 , 657 s. , richstone d. o. , byun y .-
i . , dressler a. , faber s. m. , grillmair c. , kormendy j. , lauer t. r. , 1994 , aj , 107 , 634 t. s. , bertin g. , stiavelli m. , 1995 , mnras , 276 , 1255 g. , hunter c. , verolme e. k. , de zeeuw p. t. , 2003 , mnras , 342 , 1056 f. c. , 1997 , mnras , 287 , 543 r. c. e. , van de ven g. , verolme e. k. , cappellari m. , de zeeuw p. t. , 2008 , mnras , 385 , 647 r. c. e. , van de ven g. , 2008 , mnras , submitted ( arxiv:0811.3474 ) r. p. , 1991 , mnras , 253 , 710 r. p. , 1994 ,
mnras , 270 , 271 r. p. , binney j. , davies r. l. , 1990 , mnras , 245 , 582 r. p. , cretton n. , de zeeuw p. t. , rix h .- w . , 1998 , apj , 493 , 613 r. p. , franx m. , 1993 , apj , 407 , 525 r. p. , van dokkum p. g. , 2007a , apj , 668 , 738 r. p. , van dokkum p. g. , 2007b , apj , 668 , 756 van der wel a. , van der marel r. p. , 2008 ,
apj , 684 , 260 a .- m . , krajnovi d. , van de ven g. , oosterloo t. a. , morganti r. , de zeeuw p. t. , 2008 , mnras , 383 , 1343 l. m. , bureau m. , cappellari m. , 2008 , apj , 676 , 317
the first and second velocity moments have to be convolved with the instrumental point - spread - function ( psf ) and integrated over the aperture used in the observations , before making comparisons with the observed quantities .
the psf effects are particularly important when studying the nuclear regions of galaxies ( e.g. to estimate the mass of supermassive black holes ) .
the psf - convolved velocity moments are given by the general formulas ( e.g. equations [ 51]-[53 ] of @xcite ) @xmath244 @xmath245_{\rm obs } = \frac{(\sigma \overline{v_{\rm los } } ) \otimes { \rm psf}}{\sigma_{\rm obs}}\ ] ] @xmath246_{\rm obs } = \frac{(\sigma \overline{v_{\rm los}^2 } ) \otimes { \rm psf}}{\sigma_{\rm obs}},\ ] ] with @xmath247 representing convolution and where the normalized psf is generally described as the sum of @xmath248 gaussians @xmath249 with @xmath250 .
to efficiently evaluate numerically the above convolutions up to large radii , and even for very flat models , one can evaluate the models predictions @xmath251 and @xmath252 on a grid linear in the logarithm of the elliptical radius and in the eccentric anomaly .
this is done by defining a logarithmically - spaced radial grid @xmath253 and then computing the moments at the coordinate positions @xmath254 , for linearly spaced @xmath255 values , with @xmath256 a characteristic ( e.g. the median ) observed axial ratio of the mge model .
the model is then re interpolated onto a fine cartesian grid before the convolution with the psf using fast fourier methods .
finally the model is rebinned into the observed apertures . alternatively ,
especially when the apertures sparsely sample the plane of the sky , instead of performing the convolutions of equations [ eq : conv_surf][eq : conv_mu2 ] on a regular grid , one can evaluate the same convolution integrals only for the observed apertures , while also including the integration onto the apertures ( appendix d of * ? ? ?
given a rectangular aperture of sides @xmath257 and @xmath258 , aligned with the coordinate axes at position @xmath50 on the sky , and assuming a psf given by equation ( [ eq : psf ] ) , the psf - convolved observable @xmath259 inside the aperture is @xmath260 where @xmath261 has to be replaced by @xmath49 , @xmath252 , or @xmath251 respectively , and correspondingly @xmath262 becomes @xmath263 , @xmath264_{\rm obs}$ ] , or @xmath265_{\rm obs}$ ] respectively .
the kernel is given by @xmath266+{\textrm{erf}}\left[\frac{(l_x/2)+x}{\sqrt{2 } \sigma_i } \right]\right\}\nonumber\\ & \times & \left\{{\textrm{erf}}\left[\frac{(l_y/2)-y}{\sqrt{2 } \sigma_i } \right]+{\textrm{erf}}\left[\frac{(l_y/2)+y}{\sqrt{2 } \sigma_i } \right]\right\}.\end{aligned}\ ] ] in practice the integral of equation ( [ eq : convol ] ) can be limited to the region where @xmath267 is significantly nonzero : @xmath268 | we present a simple and efficient anisotropic generalization of the semi - isotropic ( two - integral ) axisymmetric jeans formalism which is used to model the stellar kinematics of galaxies .
the following is assumed : ( i ) a constant mass - to - light ratio @xmath0 and ( ii ) a velocity ellipsoid that is aligned with cylindrical coordinates @xmath1 and characterized by the classic anisotropy parameter @xmath2 .
our simple models are fit to ` sauron ` integral - field observations of the stellar kinematics for a set of fast - rotator early - type galaxies . with only two free parameters ( @xmath3 and the inclination )
the models generally provide remarkably good descriptions of the shape of the first ( @xmath4 ) and second ( @xmath5 ) velocity moments , once a detailed description of the surface brightness is given .
this is consistent with previous findings on the dynamical structure of these objects . with the observationally - motivated assumption that @xmath6 , the method is able to recover the inclination .
the technique can be used to determine the dynamical mass - to - light ratios and angular momenta of early - type fast - rotators and spiral galaxies , especially when the quality of the data does not justify more sophisticated modeling approaches .
this formalism allows for the inclusion of dark matter , supermassive black holes , spatially varying anisotropy , and multiple kinematic components .
galaxies : elliptical and lenticular , cd galaxies : evolution galaxies : formation galaxies : kinematics and dynamics galaxies : structure |
recently @xcite we introduced a hubbard - like model which describes spin-@xmath2 fermions on a chain of @xmath3 sites , hopping with long - range amplitude , @xmath4^{-1}$ ] . here ,
@xmath5 $ ] is the chord distance between sites @xmath6 and @xmath7 on the chain closed into a ring ( the lattice spacing @xmath8 is set to unity ) . in the thermodynamic limit , @xmath9 , and for a fixed distance , @xmath10 , the purely imaginary hopping @xmath11 becomes @xmath12 , leading to the hubbard hamiltonian with `` @xmath0-hopping '' , @xmath13 here @xmath14 is the strength of the usual local hubbard @xcite interaction . for even @xmath3
we choose antiperiodic boundary conditions , so that the dispersion relation is linear in wave vector , namely , @xmath15 for @xmath16 ( @xmath17 , @xmath18 . for @xmath19
the fermi sea is the ground state with all @xmath20-states from @xmath21 to @xmath22 filled @xcite , where @xmath23 is the total particle density ( the `` filling '' ) .
note that the kinetic and potential energy operators are respectively odd and even under partity , and thus parity is , in general , not a good quantum number in our model . in the large @xmath14 limit
the hamiltonian ( [ hamilt ] ) becomes the @xmath0-tj model including pair - hopping " terms ( @xmath24 projects onto the subspace with no double occupancies ) @xcite : @xmath25 \nonumber \\ % % & & \phantom { \hat{p}_{d=0 } \biggl\ { } - \sum_{l \neq n \neq m \neq l } \frac{t_{l , n}t_{n , m}}{u } \sum_{\sigma\sigma ' } \left ( \sigma\sigma ' \right ) \hat{c}_{l\sigma}^{+}\hat{c}_{n-\sigma}^{+ } \hat{c}_{n-\sigma'}^{\mbox{}}\hat{c}_{m\sigma'}^{\mbox { } } \biggr\ } \hat{p}_{d=0 } \quad .
\label{tj}\end{aligned}\ ] ] as usual , @xmath26 , and @xmath27 is the spin-1/2 vector operator ( @xmath28 , @xmath29 , @xmath30 ) . at half filling , where the first and third terms in ( [ tj ] ) vanish , the model ( [ tj ] ) reduces to the ( parity symmetric ) @xmath31-heisenberg model introduced by haldane and shastry @xcite , with @xmath32 . in ref .
we conjectured the full excitation spectrum and associated degeneracies , and were thus able to calculate the free energy .
the exact solution allowed us to identify two zero temperature ( @xmath33 ) phase transitions : the first was a mott - hubbard @xcite metal - to - charge - insulator transition ( mcit ) in the half - filled @xmath0-hubbard model .
this was signaled by the opening of a charge gap for @xmath34 where @xmath35 is the electron bandwidth .
the second , a metal - to - spin insulator transition ( msit ) was associated with the opening of a spin gap in the @xmath0-tj model ( [ tj ] ) for @xmath36 $ ] . in this paper
we clarify the nature of these transitions and attempt to separate the features due to the special form of the dispersion in ( [ hamilt ] ) from more generic properties which may be expected to survive for other dispersions and/or in higher dimensions .
the plan of the paper is as follows : we start in section [ sec2 ] by recalling the form of our exact solution for the spectrum and ground state energies , from which we identify the location of the mcit and msit . in section [ sec3 ]
we discuss the thermodynamic properties of our two models , namely the specific heat , compressibility , and magnetic susceptibility . in sec .
[ cft ] we use conformal field theory techniques to extract the long - range behavior of ground state correlation functions .
[ g - ology ] contains a discussion of the connection with the g - ology " approach to one - dimensional systems , which we exemplify by calculating the drude weight in the frequency dependent electrical conductivity for the @xmath0-hubbard model .
finally , a summary and conclusions are presented in sec .
[ summary ] .
we recall the effective hamiltonian , introduced in ref . , @xmath37 \right\ } \label{heff}\ ] ] which describes spin ( @xmath38 ; @xmath39 , @xmath40 ) and charge ( @xmath41 ; @xmath42 , @xmath43 ) degrees of freedom in an occupation number representation with a hard core constraint , @xmath44 for each @xmath45 .
furthermore , @xmath46 , @xmath47 , @xmath48 , and @xmath49/2 \geq 0 $ ] .
the chemical potential in the presence of an external magnetic field is given by @xmath50 . in the following
we will set @xmath51 , and @xmath52 . to be precise
, we also restrict ourselves to @xmath53 and @xmath54 in which case we were allowed to identify @xmath55 because @xmath56 vanishes for @xmath57 .
note that , formally , the entire spectrum _ appears _ to display spin - charge separation _ at all energies _ in the sense that the hamiltonian splits up into independent spin and charge contributions ( `` strong '' spin - charge separation ) . in reality , however , spin and charge excitations are coupled by the constraint , @xmath44 ; and spin - charge separation only occurs at sufficiently low energies / temperatures , where spin and charge excitations contribute independently to various physical properties .
as already discussed above , we will concentrate on the physics of two special limits : ( i ) the @xmath0-hubbard model in the vicinity of half filling , @xmath58 , and ( ii ) the @xmath0-tj model for @xmath59 .
the latter is obtained by taking the limit @xmath60 of ( [ heff ] ) which projects out all double occupancies ( @xmath61 ) . with
the help of the completeness constraint one arrives at the tj effective hamiltonian , @xmath62 \label{tjeff},\ ] ] where , for @xmath63 , the exchange coupling , @xmath64 , reduces to @xmath65 $ ] .
below , we will adopt the usual standpoint for the tj model @xcite , and treat @xmath66 as an _ independent _ parameter . from eq .
( [ heff ] ) we can immediately extract the form of eigenstates in an occupation number representation in terms of the effective spin ( @xmath67 ) and charge ( @xmath68 ) degrees of freedom .
for example , the ground state is expressed symbolically as @xmath69 \ldots \left [ \uparrow \downarrow \right ] \biggr|_{{\cal k}_{f}=\pi ( 2n-1 ) } \circ \ldots \circ\ ] ] where @xmath70 , @xmath71 , @xmath72 , and @xmath73 represent @xmath74 , @xmath75 , @xmath76 , @xmath77 , respectively .
we brace those pairs at @xmath78 and @xmath45 which contribute an interaction @xmath56 @xcite .
the ground state can be regarded as a short - range rvb state in @xmath45-space obtained by filling @xmath45 states with @xmath79$]-pairs from @xmath80 to @xmath81 , where @xmath82 . although the real - space structure of the ground state wave function is not immediately apparent , it should be clear , however , that it also consists of long - range ( overlapping ) singlet pairs .
in fact , the recently constructed wave functions for the @xmath83 limit ( the `` @xmath0-t - model '' ) @xcite are genuine rvb states @xcite of the gutzwiller - jastrow type
. the corresponding ground state energy density , @xmath84 , can easily be obtained by applying the effective hamiltonian ( [ heff ] ) to the symbolic ground state wave function given above : @xmath85 + ( 1/2\pi)\int_{{\cal k}_f}^{\pi } d{\cal k } ( -t{\cal k})$ ] , where the factor @xmath2 in the first term takes into account the fact that only every second @xmath45 value contributes to the first integral .
a simple calculation then gives @xmath86 , \label{groundstate}\ ] ] where @xmath87 is the bandwidth .
using particle - hole symmetry @xcite we obtain @xmath88 , and correspondingly , for the chemical potential at zero temperature , @xmath89 , @xmath90 . from ( [ groundstate ] )
one can easily see that @xmath91 becomes finite for @xmath92 : @xmath93 .
equivalently , the charge compressibility , @xmath94 at @xmath95 vanishes for @xmath14 above @xmath96 , as expected at a mott metal - to - charge - insulator transition ( mcit ) .
we are now in the position to identify the low - lying spin and charge excitations which we discuss separately for the half - filled and less than half - filled cases : we begin with the _ spin excitations_. the four degenerate @xmath97 , @xmath98 lowest - lying spin excitations are represented by @xmath99 \ldots \left [ \uparrow \downarrow \right ] \sigma\biggr|_{{\cal k}_1 } \left [ \uparrow \downarrow \right ] \ldots \left [ \uparrow \downarrow \right ] \sigma'\biggr|_{{\cal k}_{2 } } \left [ \uparrow \downarrow \right ] \ldots \left [ \uparrow \downarrow \right ] \quad .\ ] ] for @xmath100 there is a triplet excitation only . in the thermodynamic limit
the excitation energy is given by @xmath101/2 $ ] , and we may thus identify the lowest - lying spin excitation with two _
spinons_. they are characterized as spin-1/2 objects which always come in pairs , and are separated in @xmath45-space by an _ even multiple _ of @xmath102 .
we thus rescale @xmath103 to retain the proper spacing @xmath104 of @xmath105 values .
the spinons then have the dispersion relation , @xmath106 which we depict in fig . [ fig1 ] .
the spin excitations are always gapless at @xmath107 , and at @xmath108 for @xmath109 .
their corresponding velocities at @xmath110 can now be calculated from @xmath111 as [ spinonvelocities]@xmath112 v_s^l & \equiv & v_s ( { \cal k } = \pi/2 ) = - \frac{v_f}{u / w-1 } \quad \text{for } \quad u / w > 1 \quad .
\label{vsl}\end{aligned}\ ] ] here , @xmath113 is the fermi velocity of the bare particles .
the density of states for _ spin excitations _ is then calculated as @xmath114 . with the help of eqs .
( [ spinonvelocities ] ) , for low energies we obtain @xmath115 1/\left ( 2\pi v_{s}^{r}\right ) + 1/\left ( 2\pi |v_{s}^{l}|\right ) & \text{for } & u / w > 1 \end{array } \right .
\quad . \label{dsdos}\ ] ] we next turn to the _ charge excitations_. we restrict ourselves to a fixed particle number ( @xmath116 ) and we only obtain two of the four @xmath117 , @xmath118 lowest - lying states .
these charge excitations are represented by @xmath119 \ldots \left [ \uparrow \downarrow \right ] \circ\biggr|_{{\cal k}_{1 } } \left [ \uparrow \downarrow \right ] \ldots \left [ \uparrow \downarrow \right ] \bullet\biggr|_{{\cal k}_{2 } } \left [ \uparrow \downarrow \right ] \ldots \left [ \uparrow \downarrow \right ] \quad .\ ] ] for @xmath100 there is a low lying triplet excitation ( @xmath118 , @xmath120 ) , with the singlet excitation ( @xmath117 , @xmath120 ) always at high energy ( anti - bound state ) .
again , we may identify the lowest - lying charge excitation with two _ chargeons_. in the half - filled case we always get a pair of a _ holon _ and a _ doublon _ which are separated by an even multiple of @xmath104 .
rescaling @xmath121 leads to the chargeon dispersion relation , @xmath122 as depicted in fig .
the charge excitations are gapless at @xmath108 for @xmath123 , with a velocity given by @xmath124 for @xmath125 a gap , @xmath126 , opens in the charge spectrum .
correspondingly , the density of states for _ charge excitations _ at low energies takes the form @xmath127 { \cal o}\left [ \exp \left(-e/\delta\mu_c\right)\right ] & \text{for } & u / w > 1 \end{array } \right .
\quad . \label{dcdos}\ ] ] at this point a peculiarity of our model becomes evident : the charge velocity _ increases _ as a function of @xmath128 and eventually diverges at @xmath129 .
the effective charge mass @xmath130 which is connected to the density of states or the velocities by @xmath131 _ decreases _ as a function of @xmath128 .
this is in contrast to the brinkman - rice scenario for the mott - hubbard transition in that case @xmath132 _ increases _ and charge excitations tend to localize close to the transition @xcite .
indeed , for the hubbard model with nearest - neighbor hopping ( cosine dispersion ) @xmath133 diverges at the mcit for @xmath134 when the transition is approached from below half - filling ( @xmath135 ) @xcite . in this case
there are only right - moving spinons with velocity @xmath137 , eq .
( [ vsr ] ) .
the spinons are further restricted to @xmath138 with @xmath139 .
the lowest - lying charge excitations are now given by _
alone while the doublons are always gaped : @xmath119 \ldots \left [ \uparrow \downarrow \right ] \circ\biggr|_{\cal k } \left [ \uparrow \downarrow \right ] \ldots [ \uparrow \biggr|_{{\cal k}_{f } } \downarrow ] \circ \ldots \circ \ ; .\ ] ]
although holons need not come in pairs , they are still separated by an even multiple of @xmath102 in @xmath45-space .
rescaling again we find for the charge excitation @xmath140 the holons are gapless at @xmath141 , and the corresponding velocity , @xmath142 , leads to @xmath143 in turn , this implies that the density of states for charge excitations at low energies still remains @xmath144 the results for the ground state energy density @xmath145 , eq .
( [ groundstate ] ) , allow us to calculate the chemical potential @xmath146 and the @xmath33 compressibility @xmath147 . by turning on a small external magnetic field @xmath148
we can also obtain @xmath149 ( see appendix [ appa ] , eq .
( [ e0 m ] ) ) , and the magnetic susceptibility , @xmath150 , where the magnetization density , @xmath7 , is related to @xmath148 by @xmath151 .
we summarize our results as follows : \(i ) @xmath136 or ( @xmath95 , @xmath152 ) : @xmath153 \kappa & = & \frac{1}{\pi |v_c^l| } = \frac{2}{w\left ( 1 + u/\sqrt{(w+u)^2 - 4wun}\right ) } \label{kappa}\\[3pt ] m & = & \chi { \cal h}_{0 } \\[3pt ] \chi & = & \frac{1}{\pi v_s^r } = \frac{2}{w } \left ( 1 + \frac{u}{w } \right ) \quad .
\label{chismaller}\end{aligned}\ ] ] \(ii ) @xmath95 , @xmath154 : @xmath155 \mu ( n=1^+ ) & = & u-\frac{w}{2}\\[3pt ] \delta\mu_c & = & u - w\\ \kappa & = & 0 \label{compressn1}\\ m & = & \chi { \cal h}_0\\ \chi & = & \frac{1}{\pi v_s^r } + \frac{1}{\pi |v_s^l| } = \frac{4u}{w^2 } \quad . \label{chihalffilling}\end{aligned}\ ] ] the system is an incompressible charge insulator at half - filling and @xmath125 .
note that in the limit @xmath156 the compressibility stays _
finite _ for @xmath125 ( for @xmath157 , @xmath158 for @xmath159 ) .
this is , nevertheless , consistent with eq .
( [ compressn1 ] ) since , in contrast with the situation of the usual hubbard model @xcite , in our case the function @xmath160 is not differentiable at @xmath161 and @xmath162 .
we further note that eqs .
( [ kappa ] ) , ( [ chismaller ] ) , and ( [ chihalffilling ] ) are the generic behavior expected of a `` luttinger - liquid '' @xcite , a point which we exploit in our discussion of the ground - state correlation functions ( sec .
[ cft ] ) and the drude conductivity ( sec . [ g - ology ] ) .
since the case of half - filling has already been considered by haldane @xcite , here we will be mainly interested in the less that half - filled situation . from the effective tj hamiltonian ( [ tjeff ] )
it is easy to see that the hole kinetic energy favors all particles to be as close to @xmath163 as possible while the exchange interaction @xmath65 $ ] tries to distribute the particles symmetrically around @xmath164 .
we thus expect a transition at some critical @xmath165 in this case . in general ,
the ground state can be represented as @xmath166 \ldots \left [ \uparrow \downarrow \right ] \biggr|_{{\cal k}_{2 } } \circ \ldots \circ\ ] ] where @xmath167 has to be determined from the minimization of the ground state energy .
we find ( see appendix [ appa ] ) [ e0tj]@xmath168 & = & - \frac{j_c\pi^2 n}{12}\left [ \left ( \frac{j}{j_c } -1 \right ) n(3 - 2n ) + 3 - 3n+n^2\right]\end{aligned}\ ] ] with @xmath169 and @xmath170 & = & - \frac{j_c^2\pi^2 n}{24j } \left [ \left(3 - n^2\right ) \left ( \frac{j^2}{j_c^2 } -1 \right ) + 2 \left ( 3- 3n + n^2\right ) \right]\end{aligned}\ ] ] with @xmath171 . here , @xmath172 is the critical coupling .
note that @xmath165 is proportional to @xmath173 , i.e. , it decreases with increasing ( hole ) doping .
below we discuss the two cases , @xmath174 and @xmath175 , separately : for @xmath176 we can make use of the @xmath0-hubbard model results ( sec .
[ hmsec ] ) .
in particular , @xmath177 for @xmath178 , and the spinon velocity at @xmath179 is @xmath180 it is amusing to note that this is precisely the result for the heisenberg - chain with _ nearest - neighbor _ interactions @xcite . this is consistent with the fact that the gutzwiller projected fermi sea @xmath181@xcite , the ground state wave function of the @xmath0-tj model , is also an excellent trial state for the heisenberg chain with nearest neighbor interaction @xcite , as well as for the nearest neighbor supersymmetric tj model ( @xmath182 ) @xcite . as in sec .
[ hmsec ] the corresponding low energy density of states is given by @xmath183 for @xmath184 , the holon velocity is calculated as @xmath185 . \label{vcltjsmaller}\ ] ] note that the limit @xmath186 is peculiar : precisely _ at _
@xmath187 we have a free gas of holons with @xmath188 for @xmath189 , but with allowed @xmath45-values spaced by @xmath190 . on the other hand , the limit @xmath186 only gives _
half _ of the excitations since the @xmath45 values are now spaced by @xmath102 rather than @xmath190 .
this is because , for @xmath191 , half of the @xmath187 excitations develop a gap , corresponding to the energy required to break a spinon pair , @xmath192 .
consequently , the low energy density of states for charge excitations , @xmath193 is only half as big for @xmath194 as for @xmath187 .
for @xmath175 the lowest - lying spin excitation can be represented as @xmath195 \ldots \left [ \uparrow \downarrow \right ] \circ \circ \biggr|_{{\cal k}_{2 } } \circ \ldots \circ \quad .\ ] ] in the present case nothing can be said about the ground state wave function in terms of the original fermions , since the link to the @xmath0-hubbard model can no longer be made .
again , the spinons come in pairs but they can now have _
arbitrary _ separation in the region outside of @xmath196 .
a single spinon has the excitation energy @xmath197 and thus , a _
proportional to @xmath198 opens in spinon spectrum : @xmath199 .
\label{deltamus}\end{aligned}\ ] ] @xmath200 then corresponds to the onset of a metal - to - spin - insulator transition ( msit ) . for @xmath175 a charge excitation is represented by @xmath201 \ldots \left [ \uparrow \downarrow \right ] \circ \biggr|_{{\cal k } } \left [ \uparrow \downarrow \right ] \ldots \left [ \uparrow \downarrow \right ] \biggr|_{{\cal k}_{2 } } \circ \ldots \circ\ ] ] and , after rescaling @xmath45 , the excitation energy for @xmath202 is @xmath203 .
the corresponding velocities are @xmath204 at @xmath205 and @xmath206 , respectively .
we see that , for low energies , the gapless excitations are _ parity - symmetric _ although the underlying hamiltonian itself does _ not _ conserve parity . the low energy density of states for charge excitations , @xmath207 doubles at the msit . the @xmath33 chemical potential , compressibility , magnetization , and magnetic susceptibility can be calculated from eqs .
( [ e0tj ] ) , ( [ e0mtjsmaller ] ) , ( [ e0mtjbigger ] ) , and ( [ criticalh ] ) of appendix [ appa ] .
as in the case of the @xmath0-hubbard model we summarize our results ( @xmath172 ) : \(i ) @xmath174 : @xmath208 & = & - \frac{\pi^2 ( 1-n)j_c}{4 } \left [ 2n \left ( \frac{j}{j_c } -1 \right ) + 1 \right ] \label{mutjsmaller } \\[6pt ] \kappa & = & \frac{1}{\pi |v_c^l| } = \frac{2}{\pi^2 \left [ \left(1-n\right ) \left(j_c - j\right ) + nj\right ] } \label{kappatjsmaller}\\[3pt ] m & = & \chi { \cal h}_{0 } \\[3pt ] \chi & = & \frac{1}{\pi v_s^r } = \frac{2}{\pi^2 j } \quad .
\label{chitjsmaller}\end{aligned}\ ] ] \(ii ) @xmath175 : @xmath209 & = & - \frac{\pi^2 j_c^2 ( 1-n)}{8j } \left [ \left(1+n\right ) \left ( \frac{j^2}{j_c^2 } -1 \right ) + 2\right ] \label{mutjlarger } \\[6pt ] \kappa & = & \frac{1}{\pi v_c^r } + \frac{1}{\pi |v_c^l| } = \frac{4}{j \pi^2 n}\\[6pt ] \delta\mu_s & = & \frac{\pi^2 ( 1-n)}{4 } \left ( 1-\frac{j_c}{j } \right ) \left [ j(1+n)- j_c ( 1-n ) \right]\\[3pt ] m & = & 0\quad \text{for}\quad { \cal h}_0 < { \cal h}_0^c = \delta\mu_s/2 \\[6pt ] \chi & = & 0 \quad . \label{chibigger}\end{aligned}\ ] ] the system is a spin insulator for @xmath210 . note that the compressibility @xmath211 doubles at the transition because another low - energy charge excitation replaces the spin mode which becomes gaped at @xmath200 .
as noted in ref . , the effective hamiltonian , ( [ heff ] ) , is equivalent to a _ classical _ ashkin - teller type model in the presence of an inhomogeneous `` magnetic field '' @xcite , with periodic boundary conditions .
the thermodynamic properties of our system can then be calculated by transfer - matrix techniques @xcite .
we define the abbreviations @xmath212 , @xmath213 , @xmath214 , and @xmath215 where @xmath216 ( @xmath217 ) is the inverse temperature .
the transfer matrix between sites @xmath218 and @xmath45 , @xmath219 has two vanishing eigenvalues ; the remaining two eigenvalues are given by @xmath220 } \
, \right ] \quad , \ ] ] where we introduced the abbreviation @xmath221 .
the partition function for all chemical potentials @xmath222 , magnetic fields @xmath223 , interaction strengths @xmath128 , and temperatures @xmath224 , @xmath225 leads , in the thermodynamic limit ( @xmath226 , @xmath227 ) , to the free energy density @xmath228 $ ] , @xmath229 here we made use of the fact that @xmath230 @xcite .
this general form ( [ freenergy ] ) remains valid for the @xmath0-tj model ( [ tj ] ) if we set @xmath231 . from eq .
( [ tjeff ] ) it is easy to see that at half - filling the problem is equivalent to an ising model on a ring with nearest - neighbor coupling , @xmath232 , and one thus recovers all of haldane s results for the spin - chain @xcite .
first note that at half - filling ( @xmath95 , i.e. @xcite , @xmath233 ) without external field the spectrum is completely specified in terms of _ independent _ spin and charge excitations .
in fact , the free - energy in that case has the simple spin - charge separated form @xmath234 & & - \frac{2}{2\pi\beta } \int_{-\pi/2}^{\pi/2 } d{\cal k } \left\ { \ln \left [ 1 + \exp \left ( - \beta \epsilon_{s}({\cal k } ) \right ) \right ] + \ln \left [ 1 + \exp \left ( - \beta \epsilon_{c}({\cal k } ) \right ) \right ] \right\ } \label{freenergyn=1}\end{aligned}\ ] ] where the dispersion relations for the spin ( up " and `` down '' spinons ) and charge ( holons and doublons ) excitations were already given in eqs .
( [ spinondispersion ] ) , ( [ chargeondispersion ] ) ( see figure [ fig1 ] ) . note that the rescaling of @xmath45-values gives an additional factor of two in front of the integral such that ( [ freenergyn=1 ] ) corresponds to the result of two independent free fermion systems for charge and spin .
the spinon and chargeon densities of states , @xmath235 , are shown in figure [ fig2]a and [ fig2]b , respectively . already before the mcit
the density of states for the spinons develops a van - hove singularity at @xmath236 .
this could be interpretated as the formation of the lower ( spinon ) and upper ( chargeon ) hubbard band which are , however , not yet separated .
given the densities of states , the internal energy density @xmath237 can be obtained from @xmath238 e f(e)$ ] , where @xmath239^{-1}$ ] is the fermi - dirac distribution function .
the specific heat @xmath240 is then given by @xmath241 at low temperatures this reduces to @xmath242 with @xmath243 \quad .
\label{problem}\ ] ] this luttinger liquid relation @xcite , which remains valid for all fillings , will allow us to identify the conformal charge of the conformal field theory that governs the low energy behavior of our model ( see sec .
[ cft ] ) .
the behavior of the sommerfeld coefficient @xmath244 as a function of the interaction parameter , @xmath128 , can be extracted from eqs .
( [ spinonvelocities ] ) , ( [ dsdos ] ) , ( [ vcl ] ) , and ( [ dcdos ] ) : @xmath245 where @xmath246 is the non - interacting density of states . note
that @xmath247 is _ unrenormalized _ below the mcit .
the reason is that the _ increase _ in the density of states for spin excitations is exactly compensated by the _ decrease _ in the density of states for the charge excitations .
this precise cancellation of the two contributions is a peculiarity of our model and does not occur in the conventional hubbard model . in a realistic scenario for the mott - hubbard transition
we expect a growing effective charge mass ( or , equivalently , @xmath248 ) because the transport of charge becomes more difficult due to the coulomb repulsion @xcite .
also , the spin transport should become less effective because the spin exchange energy smoothly reduces from @xmath249 to @xmath250 and we thus also expect an increasing effective spin mass ( or @xmath251 ) .
consequently , there should be an increase of @xmath244 below the mcit in any realistic hubbard - type model .
the specific heat as a function of temperature is shown in fig .
the general structure of @xmath252 reflects the behavior of the density of states ( see fig .
[ fig2 ] ) . even before the mcit
, at @xmath236 , the specific heat develops a two peak structure which reflects the van - hove singularity in the density of states for spin excitations . as already seen in eq .
( [ gamma ] ) , the sommerfeld factor is continuous at the transition .
well above the transition ( @xmath253 ) the lower and upper hubbard bands are well separated , which is reflected in a narrow low - temperature peak in @xmath252 and a broad maximum around @xmath254 . in spite of the peculiarities of our model the overall shape of @xmath255
is expected to be generic for any hubbard model with a smooth dispersion relation , in the absence of perfect nesting .
next we discuss the isothermal compressibility @xmath256 which is related to the particle number fluctuation by @xmath257 . here ,
a direct calculation gives @xmath259^{-1 } \quad .
\label{kappat}\ ] ] note that the compressibility probes the system slightly away from half - filling .
it is thus seen that not only the chargeon dispersion @xmath260 but also the spinon dispersion @xmath261 enters the expression for the compressibility .
therefore , strong charge - spin separation , in the sense of a completely decoupled response of charge and spin excitations to an external force , does not exist even at half - filling .
it is only for @xmath262 that the spinons do not contribute to the compressibility ( or holons to the magnetic susceptibility ) , as can be seen from eqs .
( [ kappa ] ) , ( [ chismaller ] ) , and ( [ chihalffilling ] ) . the fluctuation of the particle density at half - filling is shown in figure [ fig4 ] . at low temperatures and below the mcit the fluctuations are linear in temperature , i.e.
, the compressibility is constant and given by ( [ kappa ] ) . above the transition the charge gap opens , and there are only exponentially small particle number fluctuations at low temperatures . at high temperatures the fluctuations saturate at @xmath263 .
this is the classical value for spin-1/2 electrons on a lattice where on average half of the sites are doubly occupied or empty .
the magnetic susceptibility at zero external magnetic field is given by @xmath264 .
it can be directly evaluated at half - filling in terms of both spinon and chargeon degrees of freedom as @xmath265^{-1 } \quad .
\label{chit}\ ] ] the result is plotted in figure [ fig5 ] . at low temperatures
it shows pauli behavior which is strongly enhanced by the interaction , especially above the mcit ( see eqs .
( [ chismaller ] ) , ( [ chihalffilling ] ) ) . with increasing @xmath128 , due to the enhanced density of spin excitations at low energies ,
the susceptibility develops a strong peak at low temperatures . at high temperatures the susceptibility shows curie behavior , @xmath266 , the classical value for spin-1/2 electrons on a lattice where on average half of the sites
are singly occupied .
this behavior of @xmath267 is familiar from the hubbard model with cosine dispersion at half - filling @xcite , and is thus a generic feature of hubbard - type models .
away from half - filling there is no compact representation of the free energy in terms of the spinon and chargeon dispersion and one must use the general form in eq .
( [ freenergy ] ) , with @xmath268 , and @xmath269 .
furthermore , the chemical potential now depends on both temperature and density .
we will concentrate on a typical filling factor , @xmath270 , which corresponds to a critical value of the coupling @xmath271 .
the chemical potential as a function of @xmath224 for fixed @xmath270 and various values of @xmath272 is shown in figure [ fig6 ] .
it is seen that @xmath273 depends smoothly on @xmath224 for @xmath274 . both above and below the msit @xmath222
is only weakly temperature dependent at low temperatures , a behavior which is especially pronounced for larger values of @xmath272 where it remains at its @xmath33 value , eq .
( [ mutjlarger ] ) , for all @xmath275 .
although @xmath276 develops stronger @xmath224-dependence at low temperatures in the vicinity of the msit , it remains continuous as a function of @xmath66 and shows no anomalies at @xmath200 .
the specific heat as a function of @xmath224 for fixed @xmath270 and various values of @xmath272 is shown in figure [ fig7 ] .
there is just one maximum which gradually becomes broader and shifts to higher temperatures with increasing coupling strength @xmath66 .
more revealing features appear in the temperature dependence of the sommerfeld coefficient @xmath277 , shown in fig .
[ fig8 ] , which is nothing but @xmath278 , the temperature derivative of the entropy .
first of all , it is seen that the luttinger liquid relation @xmath279 remains valid ( compare eqs .
( [ dstjsmaller ] ) , ( [ dctjsmaller ] ) , ( [ dctjlarger ] ) , and ( [ problem ] ) ) . for low temperatures and just above the msit , @xmath280 shows a prominent peak , reflecting the large density of states for spin excitations just above the spin gap .
as @xmath66 becomes larger , this features broadens and shifts to higher temperatures and is completely washed out for @xmath281 .
[ fig9 ] shows the isothermal compressibility @xmath282 for several values of @xmath272 , ( a ) below and ( b ) above the msit . for @xmath174
the compressibility decreases with increasing @xmath66 , especially through the suppression of the peak at @xmath283 . at the transition ,
the compressibility at @xmath33 doubles because of the appearance of an extra gapless holon excitation in the spectrum ( see sec . [ tjcomp ] ) . just below the transition
this additional density of states for charge excitations is already present at low but finite energies and causes a sharp increase in the slope of @xmath284 for low temperatures such that the compressibility for @xmath285 is actually higher than for @xmath286 in a temperatures region around @xmath287 . above the transition the additional density of states for gapless charge excitations results in a new low temperature peak , which broadens and shifts to higher temperatures ( and
is eventually completely suppressed ) as @xmath66 is further increased above the msit .
finally , fig .
[ fig10 ] shows the magnetic susceptibility @xmath288 for the @xmath0-tj model for various values of @xmath272 below and above the msit .
the susceptibility can be cast into the form @xmath289 + 4 \exp \left [ \beta \left ( t{\cal k}-\mu\right ) \right ] + 4 \exp \left [ \beta j \left ( \pi^2 -{\cal k}^2\right)/4 \right ] \right\}^{-1/2 } \ , .
\label{chitjtemperature}\ ] ] as expected , for @xmath174 we find a finite ( pauli ) susceptibility at @xmath33 , eq .
( [ chitjsmaller ] ) . evaluating the linear term in @xmath224 from eq .
( [ chitjtemperature ] ) one finds @xmath290 t + { \cal o}(t^2)$ ] , independent of @xmath136 .
close to the transition , however , the temperature region over which this expansion is valid shrinks to zero , and the susceptibility thus seems to have a negative temperature gradient at low temperatures close to the msit ( @xmath291 ) . just below the transition we have a strong increase in the density of states for charge excitations at low energies which already showed up in the compressibility .
consequently , the density of states for low - energy spin excitations is considerably reduced .
this results in a low - temperature dip of the magnetic susceptibility close to the msit . for @xmath175
the magnetic susceptibility shows activated behavior reflecting the opening of the spin gap , @xmath292 ( eq . ( [ deltamus ] ) ) .
it is seen that the curves for @xmath285 and @xmath293 qualitatively differ from each other only in a region of very low temperatures ( @xmath294 ) .
since we do not know how to express the original electron operators in terms of the eigenstates of our effective hamiltonian , we can not directly calculate any correlation functions for our model .
fortunately , away from any phase transitions , our model belongs to the class of luttinger liquids @xcite for which the low temperature / energy behavior is dominated by two gapless excitations for charge and spin with linear spectrum and _ different _ velocities @xmath295 , @xmath296 ( charge - spin separation ) .
it is then natural to attempt to calculate the low energy behavior of various correlation functions of our model by recasting our results within the framework of conformal field theory @xcite and @xmath1-ology @xcite techniques , both of which have proved extremely powerful in extracting the physics of the luttinger liquid fixed point . in this section
we focus on the former approach and leave for the next section the discussion of @xmath1-ology .
it is well known that @xmath33 can often be viewed as the `` critical point '' in two - dimensional classical or one - dimensional quantum field theories : correlation functions decay algebraically instead of exponentially for long times and/or distances , i.e. , there is no intrinsic length scale .
the _ assumption _ that , in addition to linearizing the fermionic spectrum , conformal invariance also holds at low energies / temperatures restricts the behavior of the lowest order @xmath297-corrections to the ground state energy density ( @xmath298 ) , and the energies ( @xmath299 ) and momenta ( @xmath300 ) of low - lying states of system of finite size , @xmath3 . in turn , this information is sufficient to determine the long range , long time behavior of correlation functions .
[ an equivalent approach which is based on a landau expansion around the ground state for bethe - ansatz solvable problems gives the same results @xcite ] .
in particular , for a one component fermi gas with a linear spectrum conformal invariance implies : [ corrections ] @xmath301 % e_{h^{\pm}}^{l } - e_0^l & = & \frac{2\pi}{l } v \left ( n^{+ } + n^{- } + h^{+ } + h^{-}\right ) \label{correctionsexc}\\[6pt ] % p_{h^{\pm}}^{l } - p_0^l & = & \frac{2\pi}{l } \left ( h^{+ } - h^{-}+ n^{+ } - n^{-}\right ) + p_h^{\infty } \label{correctionsp } \quad .\end{aligned}\ ] ] here , @xmath302 is the conformal charge , @xmath303 is the velocity of the the right- and left - moving elementary excitations , @xmath304 are their conformal dimensions , @xmath305 is the momentum of the sound excitations in the thermodynamic limit , and @xmath306 are integers @xcite .
a particularly simple way of determining the conformal charge from the sommerfeld coefficient of the specific heat is due to affleck @xcite . for a one - component system with right- and left - moving elementary excitations @xmath307 where @xmath308 is the density of states for low energies . the information of eqs .
( [ corrections ] ) and ( [ affleckcv ] ) determines the large distance @xmath309 , long - time @xmath310 behavior of correlation functions of the ( primary ) fields @xmath311 as follows @xcite @xmath312 we are interested in the long - range behavior of the spin - spin correlation function @xmath313 , the density - density correlation function @xmath314 , and the one - particle green s function @xmath315 , @xmath316 c^{\rm nn } ( r , t ) & = & \left [ \frac{1}{l } \sum_{l=1}^{l } \langle \left ( \hat{n}_{l+r,\uparrow}(t ) + \hat{n}_{l+r,\downarrow}(t ) \right ) \left ( \hat{n}_{l,\uparrow } + \hat{n}_{l,\downarrow } \right ) \rangle \right ] - \left ( \frac{n}{2}\right)^2 \\[3pt ] g_{\sigma } ( r , t ) & = & \frac{-i}{2\pi } \frac{1}{l } \sum_{l=1}^{l } \langle { \cal t } \hat{c}_{l,\sigma}^{\phantom{+ } } \hat{c}_{l+r,\sigma}^{+ } ( t ) \rangle\end{aligned}\ ] ] where @xmath317 is the time - ordering operator .
the only remaining task in computing the asymptotic behavior of these correlation functions is identifying the appropriate ( primary ) fields associated with the physical operators of interest . in what follows we use our exact results to carry out the procedure outlined above
. the obvious generalization of ( [ affleckcv ] ) to a two - component system with charge and spin excitations is to replace @xmath318 by @xmath319 . as could have been expected , comparing with eq .
( [ problem ] ) gives @xmath320 @xcite .
we note that , since this identification requires finite spin and charge velocities , it breaks down at the mit .
this could have been inferred from the excitation spectra for charge and spin ( eqs .
( [ spinondispersion ] ) , ( [ chargeondispersion ] ) ) which are no longer linear near @xmath321 but behave as @xmath322 with @xmath323 .
this is reflected in the finite size corrections to the ground state energy which behave as @xmath324 instead of @xmath297 .
more precisely , from eq .
( [ appe0 ] ) , one obtains for half - filling @xmath325 a similar breakdown of conformal invariance occurs in the context of superintegrable chiral @xmath326-state potts models @xcite for @xmath327 .
it is also worth pointing out that , as shown in appendix [ appb ] , even away from the transition , there are some _ additive _ corrections to the usual formula ( [ corrections0 ] ) in our case . more precisely , we find @xmath328 \label{surprise}\ ] ] ( for @xmath95 and @xmath125 one has to read @xmath329 instead of @xmath330 ) .
it is easy to see that the corrections guarantee that there are _ no _ @xmath297 corrections present at @xmath19 where @xmath331 is the exact result for all @xmath3 .
this requirement follows from the fact that in our case the entire spectrum of the kinetic energy operator @xmath332 can be obtained from @xmath333 independent su(2 ) algebras in each sector with fixed total momentum @xmath334 : @xmath335 .
thus , in this case the algebraic structure is much simpler than the usual virasoro algebra @xcite ; this invalidates the conventional arguments based on conformal invariance and justifies the absence of finite size correction in the strict @xmath19 limit . according to eqs .
( [ correctionsexc ] ) , ( [ correctionsp ] ) , in order to identify the conformal dimensions we need to consider those lowest - lying states with the quantum numbers of the physical excitation of interest .
we will consider two- and one - particle excitations which determine the long - range behavior of the spin - spin and density - density correlation functions , and one - particle green s function , respectively .
[ [ n1 ] ] @xmath136 : + + + + + + + + + + + + + + + + + + away from the transition the lowest - lying two - spinon excitation is obtained from the ground state by breaking a @xmath79$]-pair at @xmath336 and @xmath337 , and placing the electrons into the @xmath98 , @xmath338 state at the same @xmath45 points .
the corresponding excitation energy is @xmath339 corresponding to @xmath340 ( only right - moving spinons ) .
furthermore , @xmath341 = \frac{2\pi}{l } h_s^+\ ] ] where we took into account @xcite that each bound spin - pair contributes a momentum of @xmath342 , and each unbound spin at @xmath45 adds @xmath343 . as a result ,
@xmath344 .
the charge excitation can be treated analogously and one finds ( again for @xmath136 ) @xmath345 & = & \frac{2\pi}{l } |v_c^l| + { \cal o}\left ( 1/l^2 \right ) \\[6pt ] p_{h_c^{-}}^{l } - p_0^l & = & \frac{1}{2 } \left [ - \left ( { \cal k}_f - 3\delta/2 \right ) + \left ( { \cal k}_f + \delta/2 \right ) \right ] = \frac{2\pi}{l},\end{aligned}\ ] ] implying @xmath346 , @xmath347 .
finally , the one - particle excitation from the @xmath326-particle ground state is given by the @xmath348-particle ground state : @xmath349 \ldots [ \uparrow \biggr|_{{\cal k}_f } \downarrow ] \circ \ldots \circ\ ] ] we then find , @xmath350 & = & \mu ( n ) + \frac{2\pi}{l } \left [ \frac{v_s^r}{4 } + \frac{|v_c^l|}{4}\right ] + { \cal o}\left ( 1/l^2 \right ) \\[9pt ] p_{n+1,0}^{l } - p_{n,0}^l & = & \frac{1}{2 } \left [ \left(-\pi + \delta/2 \right ) + \left ( { \cal k}_f + \delta/2 \right ) \right ] = \frac{\pi}{l } + \pi(n-1 ) \quad , \end{aligned}\ ] ] and correspondingly , @xmath351 , @xmath352 ( @xmath353 is the electronic fermi momentum ) .
[ [ n1 - 1 ] ] @xmath95 : + + + + + + + + + + + + + + + + + + below the transition ( @xmath354 ) the same results as for @xmath136 apply . above the transition ( @xmath355 ) we have no gapless charge excitations .
this implies that all correlation function involving charge excitations ( density - density correlation function , one - particle green s function ) decay exponentially because their excitation energies always involve the charge gap @xmath356 . on the other hand
, we have additional spin excitations . besides the excitation with two right - moving spinons ( @xmath357 , @xmath358 ) we also find the corresponding excitation with two left - moving spinons ( @xmath359 , @xmath360 ) .
furthermore , we may split the two spinons to form the excited state @xmath361 \ldots [ \uparrow \downarrow ] \sigma'\ ] ] leading to @xmath362 & = & \frac{2\pi}{l } \left [ \frac{v_s^r}{4 } + \frac{|v_c^l|}{4}\right ] + { \cal o}\left ( 1/l^2 \right ) \\[9pt ] p_{h_s^{\pm}}^{l } - p_0^l & = & \pi + \frac{1}{2 } \left [ \left(-\pi + \delta/2 \right ) + \left ( \pi - \delta/2 \right ) \right ] = \pi \quad .\end{aligned}\ ] ] this gives @xmath363 , and @xmath364 .
[ [ n1-or - n1-and - uw1 ] ] @xmath136 or ( @xmath95 and @xmath365 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + according to the results in the last subsection and the general formulae from conformal field theory ( [ corrfunctions ] ) we are now in position to deduce the long - range behavior of correlation functions : [ ccg ] @xmath366 c^{\rm nn } ( x , t ) & \sim & b \left ( \frac{1}{x + |v_c^l|t } \right)^2 \\[3pt ] g_{\sigma } ( x , t ) & \sim & \frac{e^{-ik_f^e x}}{2\pi } \frac{1}{\sqrt{\left ( x - v_s^r t \right ) + i/\lambda_t } } \frac{1}{\sqrt{\left ( x+|v_c^l| t \right ) + i/\lambda_t}}\end{aligned}\ ] ] where @xmath367 is a cut - off parameter ( @xmath368 ) .
note that the two - particle correlation functions are of the fermi liquid form , with renormalized velocities . on the other hand
, the one - particle green s function displays luttinger liquid behavior involving square - root singularities rather than the conventional quasi - particle asymptotic form , @xmath369 .
consequently , the fourier transformed one - particle green s function shows _ no _ quasiparticle peak , i.e. , there is no contribution proportional to @xmath370 in the one - particle spectral function . in fact
, the form of @xmath371 is very interesting but rather complicated .
a detailed analysis of its properties was given recently in refs . .
nevertheless , we can already see by dimensional analysis that there is a step - discontinuity in the momentum distribution @xmath372 at @xmath373 .
our model provides a novel example of a system which displays a discontinuity in @xmath374 in the absence of single - electron like quasi - particle excitations .
such systems have been termed `` free luttinger liquids '' @xcite or `` gutzwiller liquids '' @xcite .
this unusual behavior would reflect itself , for example , in the dependence of the kondo - temperature on the kondo coupling impurity embedded in a gutzwiller liquid " @xcite .
[ [ n1-and - uw1 ] ] @xmath95 and @xmath355 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the charge - charge correlation function and the one - particle green s function decay exponentially , while the spin - spin correlation function has the asymptotic behavior @xmath375 due to the mixture of right- and left - moving spinons , an additional structure ( `` @xmath376-oscillations '' ) appears in @xmath377 , indicating strong antiferromagnetic correlations beyond the mcit .
conformal field theory only allows us to calculate the long - range ( large @xmath309 _ and _
@xmath310 ) behavior of correlation functions . on the other hand ,
the large @xmath309 behavior at @xmath378 , involves contributions from _ all _ frequencies . from eqs .
( [ ccg ] ) it is already clear that the low - frequency modes lead to a @xmath379 decay for the equal - time spin - spin or density - density correlation function . what can be said about the high - frequency contributions ? [ [ n1-or - n1-and - uw1 - 1 ] ] @xmath136 or @xmath95 and @xmath365 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we know that there is a sharp cut - off wave vector for both charge- and spin - excitations
. in particular , no spin - excitations are possible above @xmath380 .
the highest momentum spin excitation of the ground state is @xmath381 \ldots [ \uparrow\downarrow ] \sigma ' \biggr|_{{\cal k}_f } \circ\ldots\circ\ ] ] with @xmath382/2 = n \pi$ ] .
this corresponds to a sharp edge in the correlation function as a function of momentum @xmath383 at @xmath384 ( for @xmath136 or @xmath95 and @xmath365 the correlation functions do not diverge at @xmath384 ) . as in ordinary fermi liquids , this reflects itself in long - range oscillations in real space : @xmath385 .
the same argument applies to the @xmath378 density - density correlation function which also shows long - range @xmath376-oscillations @xcite .
finally , the single - particle momentum distribution shows step discontinuities at both ends of the @xmath19 fermi surface " .
[ [ n1-and - uw1 - 1 ] ] @xmath95 and @xmath355 : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this regime , the entire large distance behavior is already contained in the large momentum - transfer modes which are also gapless . more explicitly , for large @xmath386 we find @xmath387 which shows strong antiferromagnetic correlations above the mcit . the slow oscillating decay of @xmath388 corresponds to a logarithmic divergence of the spin - spin correlation function in momentum space near @xmath389 . as expected , eq .
( [ problem ] ) implies @xmath320 @xcite .
note that the formulae ( [ corrections ] ) can not be applied exactly at the msit , @xmath200 . at the transition we have _ both _ gapless right - moving spin - excitations _ and _ gapless right- and left - moving charge - excitations with _ finite _ velocities @xmath390 , @xmath391 . nevertheless , the correct form of the @xmath297 correction at @xmath200 , @xmath392(j = j_c ) = -\left[\pi/(6l)\right ] \pi
n j_c$ ] , does not contain a contribution from all these three gapless excitations .
rather , since a finite density of spin - excitations produces an effective gap for the charge excitations and vice versa , only one of the two excitations , spin or charge , should be taken into account at @xmath200 .
more precisely , the factor , @xmath393 in the @xmath297 correction to the energy can be written in either of two forms , @xmath394 ( for the origin of the term @xmath395 see the discussion for the @xmath0-hubbard model , eq .
( [ surprise ] ) ) , or @xmath396 .
in other words , the correct value of the @xmath297 corrections is obtained by taking the limit from either above or below the transition .
[ [ j - j_c ] ] @xmath176 : + + + + + + + + + + + + + + + + + + + + + we can again use all results from the @xmath0-hubbard model in the limit of large @xmath128 . for two - particle excitations we then obtain @xmath397 , @xmath358 ; @xmath398 , @xmath347 . note that for half - filling we have additional excitations which give @xmath399 , @xmath400 .
for the one - particle excitations all excitations involving charges at gaped for @xmath95 , while for @xmath136 we have @xmath351 , @xmath352 .
[ [ jj_c-2 ] ] @xmath175 : + + + + + + + + + + + + + + + + + + + + since above the transition we have no gapless spin excitations , all correlation functions involving spin excitations ( spin - spin correlation function , one - particle green s function ) decay exponentially .
on the other hand , we have additional charge excitations : besides the right - moving holon near @xmath401 ( @xmath402 , @xmath403 ) we also find a left - moving holon near @xmath404 ( @xmath405 , @xmath347 ) .
furthermore , we can have a left - moving holon near @xmath401 and a right - moving holon near @xmath404 .
these excitations obey @xmath406 , @xmath407 . for @xmath176
we obtain the same results as for the hubbard model away from the transition , while @xmath175 both the spin - spin correlation function and the one - particle green s function decay exponentially . on the other hand
, the charge - charge correlation function behaves asymptotically like ( @xmath408 ) @xmath409 + b_2 \left [ \frac{\exp \left ( i2\pi n x\right)}{\left ( x -v_ct\right ) } + \frac{\exp \left ( -i2\pi n x\right)}{\left ( x + v_ct\right ) } \right ] \quad .
\label{cnnaftertransition}\ ] ] note that since in this case large momentum - transfer excitations are also gapless , @xmath410 displays `` @xmath411-oscillations '' @xcite . for @xmath174 we know that the exact wave function is the gutzwiller - projected fermi - sea , in which case all ground state correlation functions are explicitly known @xcite for all densities @xmath412 and distances @xmath386 .
in particular @xcite , @xmath413\ ] ] where @xmath414 is the sine - integral .
at half - filling , the asymptotic behavior @xcite , @xmath415 , reflects the strong antiferromagnetic correlations .
the formulae for the density - density correlation function are rather involved but they show the expected large - distance behavior @xmath416 the momentum distribution is found to have a jump discontinuity @xcite of size @xmath417 .
it is comforting that all general considerations from conformal field theory are verified in a limit where a complete description of the ground state properties is available .
for @xmath175 , both spin - spin correlation function and the one - particle green s function decay exponentially . the large momentum transfer processes for the density - density correlation function are no longer gaped and the large distance behavior even at @xmath378 can be deduced from conformal field theory .
we find @xmath418 the momentum transform of the density - density correlation function shows a logarithmic divergence at @xmath419 .
at low energies / temperatures normal electronic systems in one dimension can be described by a continuum field theory @xcite .
the essential idea is to linearize the electron excitation spectrum near the two fermi points , resulting in left- and right - moving fermions ( momentum and/or energy transfer cutoffs @xmath420 , @xmath421 , must then be introduced to regularize integrals ) .
these two species of fermions interact via several scattering channels characterized by coupling constants @xmath422 , @xmath423 which can also be spin - dependent hamiltonian ) .
the two large momentum transfer processes are described by @xmath424 , which parametrizes the scattering process which interchanges right- and left - moving particles ( `` backscattering '' ) , and @xmath425 , which represents the scattering of two left - moving particles into two right - moving particles and vice versa ( `` umklapp - scattering '' ) .
the remaining processes , described by @xmath426 and @xmath427 , involve a small momentum transfer , between a left - moving and a right - moving electron ( @xmath426 ) , and between electrons on the same branch ( @xmath427 ) . the model with @xmath426 and @xmath427
only is the luttinger model @xcite which has been solved exactly @xcite . in general
, there is no recipe to link the @xmath1-ology coupling constants to the parameters of a given lattice hamiltonian without solving the lattice model exactly @xcite .
moreover , this identification is only valid for interactions strengths smaller than the cutoff . in the case of the hubbard model with cosine dispersion
the mott - hubbard transition happens at half - filling for @xmath428 @xcite , so that the entire low - energy physics , including the physics of the transition , can be described within @xmath1-ology @xcite .
more generally , this approach is applicable to a wide class of one - dimensional electron systems , the so - called luttinger liquids , the low energy behavior of which is controlled by a weak coupling fixed point @xcite . below we will restrict ourselves to the discussion of the metallic phase ( @xmath136 , or @xmath95 and @xmath365 ) of the @xmath0-hubbard model , although a similar treatment can be also given for the metallic phase of the @xmath429 model .
the @xmath0-hubbard model is particularly simple as it describes only right - moving electrons which are , however , hopping _ on a lattice_. it is thus immediately evident that , in the context of @xmath1-ology , the corresponding low - energy physics is described by a _
pure @xmath427-model _ ( chiral luttinger model " ) , with a hamiltonian @xmath430 here , @xmath431 , the system volume ( length of the ring ) is @xmath432 , @xmath433 , and all terms of the hamiltonian are understood to be normal ordered with respect to the ground state of the non - interacting hamiltonian , @xmath434 . the coupling matrix in ( [ ghamilt ] )
can be decomposed as @xmath435 . for sufficiently small values of @xmath14 ( @xmath436 )
the lattice plays no role , and we may obviously identify @xmath437 , @xmath438 .
after bosonization @xcite the hamiltonian ( [ ghamilt ] ) becomes _ diagonal _ in the new bosonic operators for charge ( @xmath439 ) and spin ( @xmath440 ) and thus , the pure @xmath427-model is a `` non - interacting '' luttinger liquid or a gutzwiller liquid " @xcite .
as discussed above , such a model does have a jump discontinuity in the momentum distribution although it is not a fermi liquid .
the bosonic version of the @xmath427-hamiltonian now reads @xmath441 with @xmath442 to leading order in @xmath128 , @xmath443 , and thus , @xmath444 , @xmath445 @xcite .
note that the lattice provides the natural cut - off parameters , @xmath446 and @xmath447 .
it is amusing that , for our model , there is a way of extending the @xmath1-ology solution beyond the perturbative regime . in particular , by identifying the velocities , @xmath448 , with the velocities of the charge and spin excitations found in our exact solution , eqs .
( [ vhl ] ) , ( [ vsr ] ) , we can extract parameters @xmath449 for all values of @xmath128 ( @xmath450 for @xmath95 ) .
the resulting expressions read ( @xmath451 ) : [ gident ] @xmath452 g_4^s & = & - u \frac{w}{u+w } < 0 \quad .\end{aligned}\ ] ] for @xmath453 ( half - filling ) @xmath454 for @xmath455 , reflecting the mcit in the exact solution .
the identification ( [ gident ] ) , for which we give a different argument in appendix [ appc ] ( for @xmath95 ) , allows us to follow the solution all the way to the mcit by using the @xmath1-ology parametrization .
this clarifies the fact that , in our model , the mcit is not caused by the the relevance of @xmath425-processes at half - filling as in the hubbard model with cosine dispersion ( perfect nesting property ) , but rather , it arises as a pure renormalization of @xmath456 due to the presence of the lattice . in this section
we make use of the @xmath1-ology approach to calculate the zero temperature drude weight , @xmath457 , of the zero - frequency peak of the real part of the conductivity , @xmath458 = { \cal d}_c \delta ( \omega ) \quad \text{for } \quad \omega \to 0.\ ] ] in the non - interacting limit , @xmath459 . to obtain the electrical conductivity
we start with the charge - charge green s function ( @xmath460 is the density operator in momentum space ) , @xmath461 the retarded part of which is easily obtained from the @xmath1-ology analysis @xcite : @xmath462 this form becomes exact for our model for small values of @xmath383 and @xmath463 , and implies that the low - energy charge transport is entirely dominated by holons , eq .
( [ vcl ] ) . as a check
one may calculate the zero temperature compressibility as @xmath464 in agreement with our direct calculation , eq .
( [ kappa ] ) . up to constant prefactors which we will put back in the end , the electrical conductivity follows from the generalized einstein relation , @xmath465 which can be derived using the linear response formula for @xmath466 and the continuity equation , @xmath467 ( see also appendix [ appd ] ) .
the real part of the conductivity can then be calculated from @xmath468 = |v_c^l| \delta ( \omega - v_c^l q)$ ] , @xmath469 = gives then the final expression for the drude weight @xmath470 \quad.\ ] ] note that in our model the drude weight _ increases _ with interaction .
in fact , at half - filling , @xmath471 is seen to _ diverge _ as @xmath472 for small @xmath463 and @xmath365 .
at the same time , the compressibility goes to zero , in such a way that the product , @xmath473 , remains constant .
this behavior is in contrast to the mcit observed in the hubbard model with nearest - neighbor hopping , where , at finite @xmath14 the mcit can be approached from the metallic state by increasing the filling , @xmath135 , @xmath134 @xcite . in that case , the compressibility _ diverges _ and the conductivity tends to zero .
in fact , it is clear that for any realistic dispersion the conductivity is bounded from above ( `` f - sum rule '' @xcite ) and can not diverge : @xmath474 .
as shown in appendix [ appd ] , @xmath475 is finite whenever the bare dispersion @xmath476 has a finite second derivative with respect to @xmath20 everywhere .
it is thus seen that it is our discontinuous dispersion relation @xmath477 which allows for a diverging conductivity .
finally , we note that since we do not know how a physical vector potential couples to the effective particles of our exact solution one can not reliably determine the drude weight by twisting the boundary conditions @xcite , corresponding to adding a flux , @xmath478 , through the ring .
in fact , if we naively replace all values @xmath45 by @xmath479 to determine the ground state energy @xmath480 , and try to evaluate @xmath481 we would obtain the _ incorrect _ result for the drude weight , @xmath482 \nonumber\\[3pt ] & = & { \cal d}_{s } + { \cal d}_c \quad , \end{aligned}\ ] ] which involves contributions from both charge _ and _ spin excitations .
above we discussed various aspects of the physics of two exactly solvable models of one dimensional lattice fermions : the @xmath0-hubbard model and a related @xmath0-tj model .
the special feature of these models is that , due to absence of perfect nesting , they display non - trivial metal - to - charge - insulating and spin - insulating states , respectively , at a _ finite _ value of the coupling constants .
the @xmath0-hubbard model thus provides for an explicit realization of mott s and hubbard s ideas @xcite albeit in a rather pathological model in which the dispersion has a jump discontinuity at the zone boundaries .
this peculiarity allows for a diverging drude weight when the mcit is approach from the metallic side , associated with the _ vanishing _ of the charge mass at the transition .
while the behavior in the charge sector in the @xmath0-hubbard model is rather non - generic , the spin sector reflects the expected physics of correlated fermi systems : for example , the magnetic susceptibility strongly increases with interaction and develops a strong low - temperature peak at the spin - exchange energy scale @xmath483 . after the mcit the static spin - spin correlation function displays strong antiferromagnetic correlations with a logarithmic divergence in momentum space at @xmath389 .
the @xmath0-tj model with pair - hopping terms displays a metal - to - spin - insulator transition at @xmath484 .
the spin liquid state is described by the gutzwiller projected fermi - sea and provides an exactly solvable example of a genuine rvb - state . at the msit , simultaneously with the opening of a spin gap , the zero - frequency `` spin - conductivity '' ( drude weight for spin transport ) drops to zero , after having increased with @xmath485 ( to a finite value ) on the spin liquid side .
the msit appears as a pure level - crossing effect , and thus should not be interpreted as a mott - type transition @xcite .
we also found that , with a few changes related to the high symmetry of the spectrum in the non - interacting ( @xmath19 and @xmath187 ) models , conformal field theory is applicable in the metallic and spin liquid states ( i.e. , away from transitions ) . using conformal field theory and @xmath1-ology techniques allowed us to calculate long - distance / time properties of ground state correlation functions .
the two - particle sector shows fermi liquid behavior .
however , the single - particle green s function does not lead to a quasi - particle peak in the spectral function , even though the jump discontinuity in the momentum distribution survives .
such systems are referred to as `` free luttinger liquids '' or `` gutzwiller liquids '' . within the @xmath1-ology approach
the @xmath0-hubbard model is identifyed as a `` pure @xmath427-model '' or `` chiral luttinger model '' . in our model
the mcit is driven by short distance lattice effects which lead to a divergence of the interaction parameter @xmath456 at the mcit . despite their conceptual shortcomings the two models provide instructive examples for strongly correlated electron systems .
although a rigorous proof is still missing we think that our models and results can be used to check the capability of numerical techniques and the applicability of various approximations @xcite .
f.g . would like to thank his colleagues at rutgers university for their hospitality during a visit , and the deutsche forschungsgemeinschaft ( dfg ) for traveling funds .
this research was supported in part by a grant from the u.s .
department of energy , office of basic energy sciences , division of materials research ( d.l.c . ) , and by onr grant # n00014 - 92-j-1378 and a sloan foundation fellowship ( a.e.r . ) .
in this appendix we calculate some ground state properties in the presence of a weak magnetic field @xmath148 , and will eventually let @xmath486 . in the presence of a magnetic field we will have a finite magnetization .
spin pairs will be broken and turned into the direction of the magnetic field .
the positions of these unpaired spins is determined by the spinon dispersion ( [ spinondispersion ] ) . from the discussion above and
figure [ fig1 ] it is clear that they will always be located around @xmath163 .
for @xmath95 and @xmath154 there are also some upturned spins around @xmath487 .
we exemplify the calculation for the latter case . for @xmath95 , @xmath154 , and @xmath488 the ground state
is represented by @xmath489 \ldots \left [ \uparrow \downarrow \right ] \biggr|_{{\cal k}_{2 } } \uparrow\ldots\uparrow\ ] ] where the magnetization @xmath7 is given by @xmath490 .
the ground state energy density is calculated as @xmath491 $ ] where the additional factor @xmath2 took into account that only every second @xmath45 value contributes to the integral .
we are only interested in small fields and expand @xmath492 with @xmath493 . to second order
we then obtain with the help of eqs .
( [ groundstate ] ) , ( [ vsr ] ) , and ( [ vsl ] ) @xmath494 using @xmath495 we minimize @xmath496 with respect to @xmath497 , and obtain @xmath498 .
the magnetization becomes @xmath499 and the ground state energy reads @xmath500 a similar calculation gives the ground state energy density for @xmath136 or @xmath95 , @xmath152 as @xmath501}{4 } \nonumber \\ & - & \frac{1}{24wu } \left [ \left ( \left ( w+u \right)^{2 } - 4wum \right)^{3/2 } - \left ( \left ( w+u \right)^{2 } - 4wun \right)^{3/2 } \right ] \label{e0m}\end{aligned}\ ] ] which is valid for all @xmath7 . note that the limits @xmath502 and @xmath503 do not commute beyond the metal - to - insulator transition , just as in the usual hubbard model @xcite .
the ground state representation in the presence of a ( not too big ) magnetic field is given by @xmath504 \ldots \left [ \uparrow \downarrow \right ] \biggr|_{{\cal k}_{2 } } \circ\ldots\circ\biggr|_{\pi}\ ] ] with @xmath505 and @xmath506 .
a straightforward integration gives @xmath507 - ( j/(16\pi ) ) \left [ \pi^2({\cal k}_2-{\cal k}_1 ) - ( { \cal k}_2 ^ 3 -{\cal k}_1 ^ 3)/3\right]$ ] .
we have to minimize @xmath508 with respect to @xmath509 , where we have to distinguish the cases ( i ) @xmath510 , and ( ii ) @xmath511 to allow for a transition at some @xmath200 .
( i ) @xmath510 : we are below the transition , and we see that @xmath512 , @xmath506 . in this case we can find the ground state energy density even without the minimization procedure . for small external fields it has the simple form @xmath513 \quad .
\label{e0mtjsmaller}\ ] ] by differentiating this expression with respect to @xmath7 gives the connection between the external magnetic field @xmath148 and the ( small ) magnetization density @xmath7 @xmath514 ( ii ) @xmath511 : from the minimization equations one easily finds that @xmath515 .
we thus get a solution @xmath516 note that the presence of a magnetic field _
stabilizes _ the spin - liquid phase where spin - excitations are gapless : @xmath517 .
the ground state energy density then reads @xmath518 \quad .
\label{e0mtjbigger}\ ] ] by differentiating this expression with respect to @xmath7 gives the connection between the external magnetic field @xmath148 and the ( small ) magnetization density @xmath7 [ criticalh ] @xmath519 \\[3pt ] { \cal h}_0^c & = & \frac{\pi^2(1-n)}{8 } \left ( 1- \frac{j_c}{j } \right ) \left [ j ( 1+n ) - j_c(1-n ) \right ] \quad .\end{aligned}\ ] ] it now takes a _
finite _ magnetic field @xmath520 to magnetize the system .
the spin gap is @xmath521 ( see ( [ deltamus ] ) ) because we break a pair of spins ( @xmath97 ) to form an @xmath522 state .
the ground state energy for finite system sizes @xmath3 ( even ) and a finite particle number @xmath326 ( even ) is calculated for @xmath523 as @xmath524 .
the prime indicates that only every second of the @xmath525 @xmath18 has to be taken ( @xmath526 ) .
after rescaling @xmath527 the sums give @xmath528 - \frac{1}{2 } \sum_{r=0}^{n/2 - 1 } \sqrt{w^2 + u^2 - 4wu(2r+1)/l}\quad .\ ] ] to calculate the sum we use the poisson sum formula ( see , e.g. , ref . )
@xmath529 do the first integral and a partial integration in the second , and arrive at @xmath530 away from @xmath157 and @xmath95 we can do further and further partial integrations in the integral on the right hand side of this equation , and generate an expansion in @xmath297 .
keeping only the first term gives @xmath531 + { \cal o } \left(1/l^{2}\right ) \quad .\ ] ] doing the remaining sum and rearranging terms ( @xmath532 is the fermi velocity ) gives @xmath533 \quad .\ ] ] with the help of eqs .
( [ spinonvelocities ] ) and ( [ vcl ] ) this can be cast into the form of eq .
( [ surprise ] ) .
there is a rather simple recipe to obtain @xmath449 for all @xmath534 without referring to the exact solution . at half - filling we know that @xmath535 lies symmetrically around the upper and lower band edge ( @xmath536 at @xmath537 ) . for a pure @xmath427-model one can exactly calculate the _ screened _ electron - electron interaction with the help of ward - identities @xcite .
one finds @xmath538 the effective interaction integrated over all times is given by the @xmath539 contribution which gives @xmath540 if we now _ demand _ that the time - integrated part of the screened interaction has to stay unrenormalized at its value for small @xmath128 , @xmath541 and @xmath542 , we find that @xmath543 g_4^s = - u \frac{1}{1+u / w}\end{aligned}\ ] ] which is indeed the correct result .
so far , we have not been able to obtain a general form for @xmath449 away from half - filling , without appealing to the exact solution .
for a hamiltonian @xmath544 we assume that the interaction only contains density operators @xmath545 .
we omit spin indices in the following . the particle density operator @xmath546 then commutes with @xmath547 .
we expand the exponential in the particle density operator @xmath548 for small momenta @xmath383 ( this step could cause problems in the case of long - range hopping ) .
the continuity equation then reads @xmath549 the sum on the right - hand side is identified with the particle current operator @xmath550 using the heisenberg equations of motion we can calculate @xmath551 as @xmath552 if the dispersion relation is differentiable , this reduces to the standard expression for the current operator involving the group velocity @xmath553 .
our linear dispersion relation is not differentiable at @xmath554 .
instead we may think that the linear dispersion relation is the result of a limiting process where we start with a differentiable dispersion @xmath555 which becomes ever steeper near @xmath556 .
this corresponds to ever faster left - moving electrons near @xmath557 , in addition to the right - moving electrons for @xmath558 . from this viewpoint
it becomes clear that we have to expect a _
contribution from these `` left - moving '' electrons .
indeed , we find for @xmath559 @xmath560 and @xmath561 .
now let @xmath562 and assume that we may approximately replace @xmath563 .
for the minimum @xmath564 one has @xmath565 $ ] . as `` the current at @xmath566 '' one may _
\quad .\end{aligned}\ ] ] we obviously get an extensive ( i.e. , singular ) contribution from the states near the brillouin zone boundary .
we are interested in the real part of the transverse conductivity for small @xmath383 at temperature @xmath33 @xmath568 = \frac { e^2}{l\omega } \lim_{q \to 0 } { \rm re } \left\ { \int_{-\infty}^{t } dt'e^{i \omega ( t - t ' ) } \langle \left [ \hat{j}^{+ } ( q , t ) , \hat{j } ( q , t ' ) \right ] \rangle \right\ } \quad .\ ] ] using the continuity equation ( [ continuity ] ) , the equation of motion ( in @xmath310 ) ( eq .
( [ eqmotion ] ) ) , and integrating by parts ( in @xmath569 ) one finds @xmath570 & = & - \frac { e^2}{l \omega q^{2 } } { \rm re } \biggl\ { ( -i ) \left . \langle
\left [ \left [ \hat{\rho}^{+ } ( q , t ) , \hat{t } \right ] , \hat{\rho } ( q , t ' ) \right ] \rangle e^{i\omega ( t - t ' ) } \right|_{- \infty}^{t } \nonumber \\ & & - ( -i)(-i\omega ) \int_{-\infty}^{t } dt ' e^{i \omega ( t - t ' ) } \langle \left [ \left [ \hat{\rho}^{+ } ( q , t ) , \hat{t } \right ] , \hat{\rho } ( q , t ' ) \right ] \rangle \biggr\}\ ; . \label{monster}\end{aligned}\ ] ] the first term in @xmath571 is purely imaginary and drops out . to obtain the f - sum rule we integrate over all @xmath463
this gives @xmath572 \nonumber \\ & = & - \frac { \pi e^2}{l q^{2 } } \langle \left [ \left [ \hat{\rho}^{+ } ( q ) , \hat{t } \right ] , \hat{\rho } ( q ) \right ] \rangle\end{aligned}\ ] ] because the expression ( [ monster ] ) only depended on @xmath573 .
the double commutator gives @xmath574 , \hat{\rho } ( q ) \right ] = \sum_k \hat{c}_k^{+ } \hat{c}_k^{\phantom{+ } } \biggl [ \epsilon ( k ) - \epsilon ( k+q)+ \epsilon(k ) - \epsilon ( k - q)\biggr]\ ] ] thus , @xmath575 \right\ } \langle \hat{n}_k \rangle\ ] ] and @xmath576 .
for twice differentiable dispersion relations this reduces to @xmath577 $ ] . in particular , for a cosine dispersion , @xmath578 which is the standard result @xcite . for the linear dispersion
we get @xmath579 the minimum value of @xmath383 is @xmath580 .
thus @xmath581 \quad .\ ] ] the f - sum rule is thus _ extensive _ and leaves room for a diverging drude weight .
this can only happen , if the dispersion relation @xmath476 is not twice differentiable with respect to @xmath20 .
f. gebhard and a.e .
ruckenstein , phys .
lett . * 68 * , 244 ( 1992 ) .
j. hubbard , proc .
london , ser . a * 276 * , 238 ( 1963 ) .
the diameter of the fermi body is @xmath582 as usual .
this is the wave vector that enters in the long distance behavior of correlation functions , and is identified with `` @xmath583 '' .
anderson in _ solid state physics _
, ed . by f.
seitz and d. turnbull , vol .
* 14 * , p. 99
( 1963 ) ; j.e .
schrieffer and p.a .
wolff , phys .
rev . * 149 * , 491 ( 1966 ) ; a.b .
harris and r.v .
lange , phys .
* 157 * , 295 ( 1967 ) .
haldane , phys .
lett . * 60 * , 635 ( 1992 ) ; b.s .
shastry , phys .
lett . * 60 * , 639 ( 1992 ) .
mott , rev .
* 40 * , 677 ( 1968 ) ; _ metal - insulator transitions _ , ( taylor & francis , london , 1974 ) .
anderson , g. baskaran , z. zou , and t. hsu , phys .
lett . * 58 * , 2790 ( 1987 ) ; f.c .
zhang and t.m .
rice , phys .
* b 37 * , 3759 ( 1988 ) ; f.c .
zhang , c. gros , t.m .
rice , and h. shiba , supercond .
sci . and techn .
* 1 * , 36 ( 1988 ) .
the configurations @xmath584 $ ] and @xmath585 $ ] represent charge ( @xmath586 ) and spin singlets ( @xmath587 ) .
wang , q.f .
zhang , and p. coleman , ( unpublished ) ( 1993 ) .
anderson , science * 235 * , 1196 ( 1987 ) . in a grand canonical ensemble
we use @xmath588 .
the kinetic energy operator is invariant , and @xmath589 . at half - filling this
must be an operator identity so that we have to fix @xmath590 .
w.f . brinkman and t.m .
rice , phys .
* b 2 * , 4302 ( 1970 ) .
d. vollhardt , rev .
phys . * 56 * , 99 ( 1984 ) .
schulz , phys .
lett . * 64 * , 2831 ( 1990 ) ; int .
j. mod .
phys . * b 5 * , 57 ( 1991 ) ; _ proceedings of the adriatico research conference and miniworkshop on strongly correlated electron systems ii _
, ed . by g.
baskaran et al .
( progress in high temperature superconductivity vol .
29 , world scientific , singapore , 1991 ) ; c.f .
coll , phys .
* b 9 * , 2150 ( 1974 ) .
h. shiba and p.a .
pincus , phys .
* b 5 * , 1966 ( 1972 ) ; h. shiba , prog .
phys . * 48 * , 2171 ( 1972 ) ; h. shiba , phys .
* b 6 * , 930 ( 1972 ) .
haldane , j. phys .
c * 14 * , 2585 ( 1981 ) ; phys .
* 45 * , 1358 ( 1980 ) ; * 47 * , 1840 ( 1981 ) .
haldane , phys .
lett . * 66 * , 1529 ( 1991 ) .
j. des cloizeux and j.j .
pearson , phys .
rev . * 128 * , 2131 ( 1962 ) .
the unique ground state at @xmath19 ( fermi sea ) and for @xmath591 contain one and the same configuration with no double occupancy .
it is the only one to survive for @xmath60 , hence the gutzwiller projected fermi sea is the exact ground state for the @xmath0-tj model at all fillings .
h. yokoyama and h. shiba , j. phys .
* 56 * , 1490 ( 1987 ) ; _ ibid . _ 3570 ( 1987 ) ; _ ibid . _ 3582 ( 1987 ) ; _ ibid . _ * 57 * , 2482 ( 1988 ) ; c. gros , r. joynt , and t.m .
rice , phys .
rev . * b 36 * , 381 ( 1987 ) ; c. gros , phys . rev .
* b 38 * , 931 ( 1988 ) . f. gebhard and d. vollhardt , phys .
* 59 * , 1472 ( 1987 ) ; phys .
* b 38 * , 6911 ( 1988 ) . h. yokoyama and m. ogata , phys .
* 67 * , 3610 ( 1991 ) .
j. ashkin and e. teller , phys .
* 64 * , 178 ( 1943 ) .
l. onsager , phys .
rev . * 65 * , 117 ( 1944 ) ; see also r.j .
baxter , _ exactly solved models in statistical mechanics _ , ( academic press , new york , 1982 ) .
there is always a true gap between @xmath592 and @xmath593 so that the contribution of the @xmath592 to the free energy density does not show up in a @xmath297-expansion .
a. a. belavin , a.m. polyakov , and a.b .
zamolodchikov , nucl .
* b 241 * , 333 ( 1984 ) .
cardy , nucl .
* b 270 * [ fs16 ] , 186 ( 1986 ) ; h.w .
blte , j.l .
cardy , and m.p .
nightingale , phys .
lett . * 56 * , 742 ( 1986 ) ; a.d .
mirnov and a.v .
zabrodin , phys .
66 * , 534 ( 1991 ) . h. frahm and v.e .
korepin , phys .
* b 42 * , 10553 ( 1990 ) ; * b 43 * , 5653 ( 1991 ) .
n. kawakami and s .- k .
yang , phys .
* 148a * , 359 ( 1990 ) ; phys .
* 65 * , 2309 ( 1990 ) ; prog . theor .
suppl . * 107 * , 59 ( 1992 ) .
j. slyom , adv .
* 28 * , 201 ( 1979 ) .
j. voit , phys .
* b 45 * , 4027 ( 1992 ) .
w. metzner and c. di castro , phys .
b 47 * , 16107 ( 1993 ) .
j. carmelo and a.a .
ovchinnikov , j. phys .
* c 3 * , 757 ( 1991 ) ; j.m.p .
carmelo , p. horsch , and a.a .
ovchinnikov , phys .
* b 45 * , 7899 ( 1992 ) ; j.m.p .
carmelo and p. horsch , phys .
68 * , 871 ( 1992 ) .
i. affleck , phys .
lett . * 56 * , 746 ( 1986 ) .
note that our model is chiral , i.e. , the density of states is only half as big as in the usual case .
g. albertini , b.m .
mccoy , j.h.h .
perk , s. tang , nucl .
* b 314 * , 741 ( 1989 ) ; r.j .
baxter , j. stat .
* 57 * , p. 1
v. meden and k. schnhammer , phys .
rev . * b 46 * , 15753 ( 1992 ) , and unpublished ( 1993 ) .
j. voit , phys .
* b 47 * , 6740 ( 1993 ) , and unpublished ( 1993 ) .
lee and j. toner , phys . rev .
* 69 * , 3378 ( 1992 ) , and private communication . w. metzner and d. vollhardt , phys
. rev .
lett . * 59 * , 121 ( 1987 ) ; phys .
b 37 * , 7382 ( 1988 ) .
luttinger , j. math .
* 4 * , 1154 ( 1963 ) . d.c .
mattis and e.h .
lieb , j. math .
* 6 * , 304 ( 1965 ) .
lieb and f.y .
wu , phys .
lett . * 20 * , 1445 ( 1968 ) .
rezayi , j. sak , and j. slyom , phys .
* b 20 * , 1129 ( 1979 ) ; _ ibid .
_ * b 23 * , 1342 ( 1981 ) ; k. penc and j. slyom , phys .
* b 44 * , 12690 ( 1991 ) .
see , e.g. , d. pines and p. nozires , _ theory of quantum liquids _
( benjamin , new york , 1966 ) ; p.f .
maldague , phys .
* b 16 * , 2437 ( 1977 ) .
shastry and b. sutherland , phys .
lett . * 65 * , 243 ( 1990 ) .
w. kohn , phys .
rev . * 133 * , a 171 ( 1964 ) ; for recent applications , see c.a .
stafford , a.j .
millis , and b.s .
shastry , phys .
rev . * b 43 * , 13660 ( 1991 ) ; r.m .
fye , m.j .
martins , d.j .
scalapino , j. wagner , and w. hanke , phys .
* b 45 * , 7311 ( 1992 ) .
f. gebhard and a. girndt , `` comparison of variational approaches for the exactly solvable @xmath0-hubbard model '' , available as preprint cond - mat/9307036 from the preprint library cond - mat`@`babbage.sissa.it ( accepted for publication in z. phys .
b ) . l.d .
landau and e.m .
lifshits , _ statistical physics , part 1 _ , course on theoretical physics vol .
v , 60 ( 3rd edition , pergamon press , new york , 1980 ) . | we discuss the transition from a metal to charge or spin insulating phases characterized by the opening of a gap in the charge or spin excitation spectra , respectively .
these transitions are addressed within the context of two exactly solvable hubbard and tj chains with long range , @xmath0 hopping .
we discuss the specific heat , compressibility , and magnetic susceptibility of these models as a function of temperature , band filling , and interaction strength .
we then use conformal field theory techniques to extract ground state correlation functions .
finally , by employing the @xmath1-ology analysis we show that the charge insulator transition is accompanied by an infinite discontinuity in the drude weight of the electrical conductivity .
while the magnetic properties of these models reflect the genuine features of strongly correlated electron systems , the charge transport properties , especially near the mott - hubbard transition , display a non - generic behavior . |
in line with information theory , we treat a literary text as the output of a stationary and ergodic source that takes values in a finite alphabet and we look for information about the source through a statistical analysis of the text @xcite . here
we focus on correlations functions , which are defined after specifying an observable and a product over functions .
in particular , given a symbolic sequence * s * ( the text ) , we denote by @xmath2 the symbol in the @xmath3-th position and by @xmath4 ( @xmath5 ) the substring @xmath6 . as observables , we consider functions @xmath7 that map symbolic sequences * s * into a sequence * x * of numbers ( e.g. , @xmath1 s and @xmath0 s ) .
we restrict to local mappings , namely @xmath8 for any @xmath3 and a finite constant @xmath9 .
its autocorrelation function is defined as : @xmath10 where @xmath11 plays the role of time ( counted in number of symbols ) and @xmath12 denotes an average over sliding windows , see supporting information ( si ) sec .
i for details .
the choice of the observable @xmath7 is crucial in determining whether and which `` memory '' of the source is being quantified . only once a class of observables sharing the same properties is shown to have the same asymptotic autocorrelation , it is possible to think about long - range correlations of the text as a whole . in the past ,
different kinds of observables and encodings ( which also correspond to particular choices of @xmath7 ) were used , from the huffmann code @xcite , to attributing to each symbol an arbitrary binary sequence ( ascii , unicode , 6-bit tables , dividing letters in groups , etc . )
@xcite , to the use of the frequency - rank @xcite or parts of speech @xcite on the level of words .
while the observation of long - range correlations in all cases points towards a fundamental source , it remains unclear which common properties these observables share .
this is essential to determine whether they share a common root ( conjectured in ref .
@xcite ) and to understand the meaning of quantitative changes in the correlations for different encodings ( reported in ref .
@xcite ) . in order to clarify these points we use mappings @xmath7 that avoid the introduction of spurious correlations .
inspired by voss @xcite and ebeling _ et al . _
@xcite we use @xmath13 s that transform the text into binary sequences @xmath14 by assigning @xmath15 if and only if a local matching condition @xmath16 is satisfied at the @xmath3-th symbol , and @xmath17 otherwise ( e.g. , @xmath18 _ k - th symbol is a vowel _ ) .
see si - sec .
ii for specific examples . once equipped with the binary sequence @xmath19 associated with the chosen condition @xmath16 we can investigate the asymptotic trend of its @xmath20 .
we are particularly interested in the long - range correlated case @xmath21 for which @xmath22 diverges . in this case
the associated random walker @xmath23 spreads super - diffusively as @xcite @xmath24 in the following we investigate correlations of the binary sequence @xmath19 using eq .
( [ eq.mu ] ) because integrated indicators lead to more robust numerical estimations of asymptotic quantities @xcite .
we are mostly interested in the distinction between short- @xmath25 and long- @xmath26 range correlations .
we use normal ( anomalous ) diffusion of @xmath27 interchangeably with short- ( long- ) range correlations of * x*. an insightful view on the possible origins of the long - range correlations can be achieved by exploring the relation between the power spectrum @xmath28 at @xmath29 and the statistics of the sequence of inter - event times @xmath30 s ( i.e. , one plus the lengths of the cluster of @xmath1 s between consecutive @xmath0 s ) .
for the short - range correlated case , @xmath31 is finite and given by @xcite : @xmath32 for the long - range correlated case , @xmath33 and eq . ( [ eq.spectrum ] ) identifies two different origins : ( i ) _ burstiness _ measured as the broad tail of the distribution of inter - event times @xmath34 ( divergent @xmath35 ) ; or ( ii ) long - range correlations of the sequence of @xmath30 s ( not summable @xmath36 ) . in the next section we show how these two terms give different contributions at different linguistic levels of the hierarchy . building blocks of the hierarchy depicted in fig .
[ fig.1 ] are binary sequences ( organized in levels ) and links between them .
levels are established from sets of semantically or syntactically similar conditions @xmath16 s ( e.g. , vowels / consonants , different letters , different words , different topics ) .
each binary sequence @xmath14 is obtained by mapping the text using a given @xmath13 , and will be denoted by the relevant condition in @xmath16 .
for instance , * prince * denotes the sequence @xmath14 obtained from the matching condition @xmath37 `` prince '' .
a sequence @xmath38 is linked to @xmath14 if for all @xmath39 s such that @xmath40 we have @xmath41 , for a fixed constant @xmath42 .
if this condition is fulfilled we say that @xmath14 is _ on top of _ @xmath38 and that @xmath14 belongs to a higher level than @xmath38 . by definition , there are no direct links between sequences at the same level .
a sequence at a given level is on top of all the sequences in lower levels to which there is a direct path . for instance , * prince * is on top of * e * which is on top of * vowel*. as will be clear later from our results , the definition of link can be extended to have a probabilistic meaning , suited for generalizations to high levels ( e.g. , `` prince '' is more probable to appear while writing about a topic connected to war ) .
we now show how correlations flow through two linked binary sequences . without loss of generality
we denote @xmath19 a sequence on top of @xmath43 and @xmath44 the unique sequence on top of @xmath38 such that @xmath45 ( sum and other operations are performed on each symbol : @xmath46 for all @xmath47 ) .
the spreading of the walker @xmath48 associated with @xmath38 is given by @xmath49 where @xmath50 is the cross - correlation .
using the cauchy - schwarz inequality @xmath51 we obtain @xmath52 define @xmath53 , as the sequence obtained reverting @xmath54 on each of its elements @xmath55 .
it is easy to see that if @xmath45 then @xmath56 . applying the same arguments above , and using that @xmath57 for any @xmath19
, we obtain @xmath58 and similarly @xmath59 .
suppose now that @xmath60 with @xmath61 .
in order to satisfy simultaneously the three inequalities above , at least two out of the three @xmath62 have to be equal to the largest value @xmath63 . next we discuss the implications of this restriction to the flow of correlations up and down in our hierarchy of levels . * up . *
suppose that at a given level we have a binary sequence @xmath38 with long - range correlations @xmath64 .
from our restriction we know that at least one sequence @xmath19 on top of @xmath43 , has long - range correlations with @xmath65 .
this implies , in particular , that if we observe long - range correlations in the binary sequence associated with a given letter then we can argue that its anomaly originates from the anomaly of at least one word where this letter appears , higher in the hierarchy of a word containing the given letter is on top of the sequence @xmath43 of that letter .
if @xmath43 is long range correlated ( lrc ) then either @xmath19 is lrc or @xmath66 is lrc .
being finite the number of words with a given letter , we can recursively apply the argument to @xmath66 and identify at least one lrc word . ] . * down .
* suppose @xmath19 is long - range correlated @xmath67 . from eq .
( [ eq.sum ] ) we see that a fine tuning cancellation with cross - correlation must appear in order for their lower - level sequence @xmath43 ( down in the hierarchy ) to have @xmath68 . from the restriction derived above
we know that this is possible only if @xmath69 , which is unlikely in the typical case of sequences @xmath38 receiving contributions from different sources ( e.g. , a letter receives contribution from different words ) .
typically , @xmath38 is composed by @xmath70 sequences @xmath71 , with @xmath72 , in which case @xmath73 .
correlations typically flow down in our hierarchy of levels .
* finite - time effects . *
while the results above are valid asymptotically ( infinitely long sequences ) , in the case of any real text we can only have a finite - time estimate @xmath74 of the correlations @xmath75 .
already from eq .
( [ eq.sum ] ) we see that the addition of sequences with different @xmath76 , the mechanism for moving down in the hierarchy , leads to @xmath77 if @xmath78 is computed at a time when the asymptotic regime is still not dominating .
this will play a crucial role in our understanding of long - range correlations in real books . in order to give quantitative estimates
, we consider the case of @xmath43 being the sum of the most long - range correlated sequence @xmath19 ( the one with @xmath79 ) and many other independent non - overlapping and @xmath44 are non - overlapping if for all @xmath47 for which @xmath80 we have @xmath81 . ]
sequences whose combined contribution is written as @xmath82 , with @xmath83 an independent identically distributed binary random variable .
this corresponds to the random addition of @xmath0 s with probability @xmath84 to the @xmath1 s of @xmath19 . in this case
@xmath85 shows a transition from normal @xmath86 to anomalous @xmath87 diffusion .
the asymptotic regime of * z * starts after a time @xmath88 where @xmath89 and @xmath67 are obtained from @xmath90 which asymptotically goes as @xmath91 .
note that the power - law sets at @xmath92 only if @xmath93 .
a similar relation is obtained moving up in the hierarchy , in which case a sequence @xmath19 in a higher level is built by random subtracting @xmath0 s from the lower - level sequence @xmath43 as @xmath94 ( see si - sec .
iii - a for all calculations ) .
* burstiness .
* in contrast to correlations , burstiness due to the tails of the inter - event time distribution @xmath34 is not always preserved when moving up and down in the hierarchy of levels .
consider first going down by adding sequences with different tails of @xmath34 .
the tail of the combined sequence will be constrained to the shortest tail of the individual sequences . in the random addition example , @xmath95 with @xmath19 having a broad tail in @xmath34 , the large @xmath96 asymptotic of @xmath43 has short - tails because the cluster of zeros in @xmath19 is cut randomly by @xmath97 @xcite .
going up in the hierarchy , we take a sequence on top of a given bursty binary sequence , e.g. , using the random subtraction @xmath98 mentioned above .
the probability of finding a large inter - event time @xmath96 in @xmath43 is enhanced by the number of times the random deletion merges two or more clusters of @xmath1 s in @xmath19 , and diminished by the number of times the deletion destroys a previously existent inter - event time @xmath96 .
even accounting for the change in @xmath99 , this moves can not lead to a short - ranged @xmath34 for @xmath19 if @xmath34 of @xmath43 has a long tail ( see si - sec .
iii - b ) .
altogether , we expect burstiness to be preserved moving up , and destroyed moving down in the hierarchy of levels . * summary . * from eq .
( [ eq.spectrum ] ) the origin of long - range correlations @xmath100 can be traced back to two different sources : the tail of @xmath34 ( burstiness ) and the tail of @xmath101 .
the computations above reveal their different role at different levels in the hierarchy : @xmath75 is preserved moving down , but there is a transfer of _ information _ from @xmath34 to @xmath101 .
this is better understood by considering the following simplified set - up : suppose at a given level we observe a sequence @xmath19 coming from a renewal process with broad tails in the inter - event times @xmath102 with @xmath103 leading to @xmath104 @xcite .
let us now consider what is observed in * z * , at a level below , obtained by adding to @xmath19 other independent sequences .
the long @xmath96 s ( a long sequence of 0 s ) in eq .
( [ eq.renewal ] ) will be split in two long sequences introducing at the same time a cut - off @xmath105 in @xmath34 and non - trivial correlations @xmath106 for large @xmath3 . in this case , asymptotically the long - range correlations ( @xmath107 ) is solely due to @xmath106 .
burstiness affects only @xmath74 estimated for times @xmath108 .
a similar picture is expected in the generic case of a starting sequence @xmath19 with broad tails in both @xmath34 and @xmath101 . .
( b , d ) transport @xmath109 defined in eq .
( [ eq.mu ] ) .
the numerical results show : ( a ) exponential decay of @xmath110 with @xmath111 inset : @xmath34 in log - linear scales ; ( b ) @xmath112 ; ( c ) non - exponential decay of @xmath110 with @xmath113 ; and ( d ) @xmath114 .
all panels show results for the the original and @xmath115-shuffled sequences , see legend . ]
is an indicator of the burstiness of the distribution @xmath34 .
@xmath74 is a finite time estimator of the global indicator of long - range correlation @xmath75 .
a poisson process has @xmath116 .
the twenty most frequent symbols ( white circles ) and twenty frequent words ( black circles ) of wrnpc are shown ( see si - tables for all books ) . @xmath117 indicates the case of * vowels * and @xmath118 of * blank space*. the red dashed - line is a lower - bound estimate of @xmath74 due to burstiness ( see si - sec .
vi ) . this diagram is a generalization for long - range correlated sequences of the diagrams in ref .
equipped with previous section s theoretical framework , here we interpret observations in real texts .
we use ten english versions of international novels ( see si - sec .
iv for the list and for the pre - processing applied to the texts ) . for each book @xmath119
binary sequences were analyzed separately : vowel / consonants , @xmath120 at the letter level ( blank space and the @xmath121 most frequent letters ) , and @xmath120 at the word level ( @xmath122 most frequent words , @xmath123 most frequent nouns , and @xmath123 words with frequency matched to the frequency of the nouns ) .
the finite - time estimator of the long - range correlations @xmath74 was computed fitting eq .
( [ eq.mu ] ) in a broad range of large @xmath124 $ ] ( time lag of correlations ) up to @xmath125 of the book size .
this range was obtained using a conservative procedure designed to robustly distinguish between short and long - range correlations ( see si - sec .
we illustrate the results in our longest novel , `` war and peace '' by l. tolstoy ( wrnpc , in short , see si - tables for the results in all books ) .
one of the main goals of our measurements is to distinguish , at different hierarchy levels , between the two possible sources of long - range correlations in eq .
( [ eq.spectrum ] ) burstiness corresponding to @xmath34 with diverging @xmath35 or diverging @xmath126 . to this end
we compare the results with two null - model binary sequences @xmath127 obtained by applying to @xmath19 the following procedures : * shuffle the sequence of @xmath128 s .
destroys all correlations .
* shuffle the sequence of inter - event times @xmath30 s .
destroys correlations due to @xmath101 but preserves those due to @xmath34 . starting from the lowest level of the hierarchy depicted in fig .
[ fig.1 ] , we obtain @xmath129 for the sequence of vowels in wrnpc and @xmath74 between @xmath130 and @xmath131 in the other @xmath132 books ( see si - fig .
the values for @xmath133 and @xmath134 were compatible ( two error bars ) with the expected value @xmath135 in all books .
figures [ fig.2]ab show the computations for the case of the letter `` e '' : while @xmath34 decays exponentially in all cases ( fig .
[ fig.2]a ) , long - range correlations are present in the original sequence @xmath136 but absent from the a2 shuffled version of @xmath136 ( fig . [ fig.2]b ) .
this means that burstiness is absent from @xmath136 and does not contribute to its long - range correlations .
in contrast , for the word `` prince '' fig .
[ fig.2]c shows a non - exponential @xmath34 and fig .
[ fig.2]d shows that the original sequence @xmath137 and the @xmath138 shuffled sequence show similar long - range correlations ( black and red curves , respectively ) .
this means that the origin of the long - range correlations of * prince * are mainly due to burstiness tails of @xmath34 and not to correlations in the sequence of @xmath30 s @xmath101 . in fig .
[ fig.3 ] we plot for different sequences the summary quantities @xmath74 and @xmath139 a measure of the burstiness proportional to the relative width of @xmath34 @xcite .
a poisson process has @xmath140 .
all _ letters _ have @xmath141 , but clear long - range correlations @xmath142 ( left box magnified in fig .
[ fig.3 ] ) .
this means that correlations come from @xmath101 and not from @xmath34 , as shown in fig .
[ fig.2](a , b ) for the letter `` e '' .
the situation is more interesting in the higher - level case of _ words_. the most frequent words and the words selected to match the nouns mostly show @xmath141 so that the same conclusions we drew about letters apply to these words .
in contrast to this group of function words are the most frequent _ nouns _ that have large @xmath139 @xcite and large @xmath74 , appearing as outliers at the upper right corner of fig .
[ fig.3 ] .
the case of `` prince '' shown in fig .
[ fig.2](c , d ) is representative of these words , for which burstiness contributes to the long - range correlations . in order to confirm the generality of fig .
[ fig.3 ] in the @xmath143 books of our database , we performed a pairwise comparison of @xmath74 and @xmath144 between the @xmath123 nouns and their frequency matched words .
overall , the nouns had a larger @xmath74 in @xmath145 and a larger @xmath146 in @xmath147 out of the @xmath148 cases ( p - value @xmath149 , assuming equal probability ) . in every single book at least @xmath150 out of @xmath123 comparisons
show larger values of @xmath74 and @xmath139 for the nouns .
we now explain a striking feature of the data shown in fig .
[ fig.3 ] : the absence of sequences with low @xmath74 and high @xmath139 ( lower - right corner ) .
this is an evidence of correlation between these two indicators and motivates us to estimate a @xmath139-dependent lower bound for @xmath74 , as shown in fig .
[ fig.3 ] .
note that high values of burstiness are responsible for long - range correlations estimate @xmath151 , as discussed after eq .
( [ eq.renewal ] ) .
for instance , the slow decay of @xmath34 for intermediate @xmath96 in * prince * ( fig . [ fig.2]c ) leads to @xmath152 and an estimate @xmath153 at intermediate times .
burstiness contribution to @xmath74 ( which gets also contributions from long - range correlations in the @xmath30 s ) is measured by @xmath154 , which is usually a lower bound for the total long - range correlations : @xmath155 .
more quantitatively , consider an @xmath138-shuffled sequence with power - law @xmath34 as in eq .
( [ eq.renewal ] ) with an exponential cut - off for @xmath156 . by increasing @xmath157 we have
that @xmath139 monotonously increases [ it can be computed directly from @xmath34 ] . in terms of @xmath154 ,
if the fitting interval @xmath124 $ ] used to compute the finite time @xmath154 is all below @xmath157 ( i.e. @xmath159 ) we have @xmath160 ( see eq .
( [ eq.renewal ] ) ) while if the fitting interval is all beyond the cutoff ( i.e. @xmath161 ) we have @xmath162 .
interpolating linearly between these two values and using @xmath163 we obtain the lower bound for @xmath74 in fig .
[ fig.3 ] .
it strongly restricts the range of possible @xmath164 in agreement with the observations and also with @xmath74 obtained for the @xmath138-shuffled sequences ( see si - sec .
vi for further details ) . the pre - asymptotic normal diffusion anticipated in sec .
* finite - time effects * is clearly seen in fig .
[ fig.4 ] .
our theoretical model explains also other specific observations : \1 .
key - words reach higher values of @xmath74 than letters ( @xmath165 ) .
this observation contradicts our expectation for asymptotic long times : * prince * is on top of * e * and the reasoning after eq .
( [ eq.sum ] ) implies @xmath166 .
this seeming contradiction is solved by our estimate ( [ eq.tt ] ) of the transition time @xmath167 needed for the finite - time estimate @xmath74 to reach the asymptotic @xmath75 .
this is done imagining a surrogate sequence with the same frequency of `` e '' composed by * prince * and randomly added @xmath0 s . using the fitting values of @xmath168 for * prince * in eq .
( [ eq.tt ] ) we obtain @xmath169 , which is larger than the maximum time @xmath170 used to obtain @xmath74 .
conversely , for a sequence with the same frequency of `` prince '' built as a random sequence on top of * e * we obtain @xmath171 . these calculations
not only explain @xmath165 , they show that * prince * is a particularly meaningful ( not random ) sequence on top of * e * , and that * e * is necessarily composed by other sequences with @xmath172 that dominate for shorter times .
more generally , the _ observation _ of long - range correlations at low levels is due to widespread correlations on higher levels .
the sharper transition for keywords .
the addition of many sequences with @xmath100 explains the slow increase in @xmath173 for letters because sequences with increasingly larger @xmath75 dominate for increasingly longer times .
the same reasoning explains the positive correlation between @xmath174 and the length of the book ( pearson correlation @xmath175 , similar results for other letters ) .
the sequence @xmath176 also shows slow transition and small @xmath74 , consistent with the interpretation that it is connected to many topics on upper levels .
in contrast , the sharp transition for * prince * indicates the existence of fewer independent contributions on higher levels , consistent with the observation of the onset of burstiness @xmath177 .
altogether , this strongly supports our model of hierarchy of levels with keywords ( but not function words ) strongly connected to specific topics which are the actual correlation carriers .
the sharp transition for the keywords appears systematically roughly at the scale of a paragraph ( @xmath178 symbols ) , in agreement with similar observation in refs .
@xcite .
additional insights on long - range correlations are obtained by investigating whether they are robust under different manipulations of the text @xcite . here
we focus on two non - trivial shuffling methods ( see si - sec .
vii for simpler cases for which our theory leads to analytic results ) .
consider generating new same - length texts by applying to the original texts the following procedures * keep the position of all blank spaces fixed and place each word - token randomly in a gap of the size of the word .
* recode each word - type by an equal length random sequence of letters and replace consistently all its tokens .
note that m1 preserve structures ( e.g. , words and letter frequencies ) destroyed by m2 . in terms of our hierarchy , m1 destroys the links to levels above word level while m2 shuffles the links from word- to letter - levels . since according to our picture correlations
originate from high level structures , we predict that m1 destroys and m2 preserves long - range correlations . indeed simulations unequivocally show that long - range correlations present in the original texts ( average @xmath74 of letters in wrnpc @xmath179 and in all books @xmath180 ) are mostly destroyed by m1 ( @xmath181 and @xmath182 ) and preserved by m2 ( @xmath183 and @xmath184 ( see si - tables for all data ) . at this point
it is interesting to draw a connection to the _ principle of the arbitrariness of the sign _ , according to which the association between a given sign ( e.g. , a word ) and the referent ( e.g. , the object in the real world ) is arbitrary @xcite .
as confirmed by the m2 shuffling , the long - range correlations of literary texts are invariant under this principle because they are connected to the semantic of the text .
our theory is consistent with this principle .
( local derivative of the transport curve in fig .
[ fig.2]bd ) .
results for three sequences in wrnpc are shown ( from top to bottom ) : the noun `` prince '' , the most frequent letter `` e '' , and the word `` so '' ( same frequency of `` prince ' ' ) .
the horizontal lines indicate the @xmath74 , the error bars , and the fitting range .
inset ( from top to bottom ) : the @xmath150 other nouns appearing as outliers in fig .
[ fig.3 ] , the @xmath150 most frequent letters after `` e '' , and the @xmath150 words matching the frequency of the outlier - nouns . ]
from an information theory viewpoint , long - range correlations in a symbolic sequence have two different and concurrent sources : the broad distribution of the distances between successive occurrences of the same symbol ( burstiness ) and the correlations of these distances .
we found that the contribution of these two sources is very different for observables of a literary text at different linguist levels .
in particular , our theoretical framework provides a robust mechanism explaining our extensive observations that on relevant semantic levels the text is high - dimensional and bursty while on lower levels successive projections destroy burstiness while preserving the long - range correlations of the encoded text via a flow of information from burstiness to correlations .
the mechanism explaining how correlations cascade from high- to low - levels is generic and extends to levels higher than word - level in the hierarchy in fig .
[ fig.1 ] .
the construction of such levels could be based , e.g. , on techniques devised to extract information on a `` concept space '' @xcite .
while long - range correlations have been observed at the concept level @xcite , further studies are required to connect to observations made at lower levels and to distinguish between the two sources of correlations .
our results showing that correlation is preserved after random additions / subtractions of @xmath0 s help this connection because they show that words can be linked to concepts even if they are not used every single time the concept appears ( a high probability suffices ) .
for instance , in ref .
@xcite a topic can be associated to an axis of the concept space and be linked to the words used to build it . in this case , when the text is referring to a topic there is a higher probability of using the words linked to it and therefore our results show that correlations will flow from the topic to the word level . in further higher levels , it is insightful to consider as a limit picture the renewal case eq .
( [ eq.renewal ] ) for which long - range correlations originate only due to burstiness . this _
limit case _ is the simplest toy model compatible with our results .
our theory predicts that correlations take form of a bursty sequence of events once we approach the semantically relevant topics of the text .
our observations show that some highly topical words already show long - range correlations mostly due to burstiness , as expected by observing that topical words are connected to less concepts than function words @xcite .
this renewal limit case is the desired outcome of successful analysis of anomalous diffusion in dynamical systems and has been speculated to appear in various fields @xcite . using this limit case as a guideline we can think of an algorithm able to automatically detect the relevant structures in the hierarchy by pushing recursively the long - range correlations into a renewal sequence .
next we discuss how our results improve previous analyses and open new possibilities of applications .
previous methods either worked below the letter level @xcite or combined the correlations of different letters in such a way that asymptotically the most long - range correlated sequence dominates @xcite . only through our results
it is possible to understand that indeed a single asymptotic exponent @xmath75 should be expected in all these cases .
however , and more importantly , @xmath75 is usually beyond observational range and an interesting range of finite - time @xmath74 is obtained depending on the observable or encoding . on the letter level , our analysis ( figs .
[ fig.2 ] and [ fig.3 ] ) revealed that all of them are long - range correlated with no burstiness ( exponentially distributed inter - event times ) .
this lack of burstiness can be wrongly interpreted as an indication that letters @xcite and most parts of speech @xcite are well described by a poisson processes .
our results explain that the non - poissonian ( and thus information rich ) character of the text is preserved in the form of long - range correlations ( @xmath100 ) , which is observed also for all frequent words ( even in the most frequent word `` the ' ' ) .
these observations violate not only the strict assumption of a poisson process , they are incompatible with any finite - state markov chain model .
these models are the basis for numerous applications of automatic semantic information extraction , such as keywords extraction , authorship attribution , plagiarism detection , and automatic summarization @xcite .
all these applications can potentially benefit from our deeper understanding of the mechanisms leading to long - range correlations in texts . apart from these applications
, more fundamental extensions of our results should : ( i ) consider the mutual information and similar entropy - related quantities , which have been widely used to quantify long - range correlations @xcite ( see @xcite for a comparison to correlations ) ; ( ii ) go beyond the simplest case of the two point autocorrelation function and consider multi - point correlations or higher order entropies @xcite , which are necessary for the complete characterization of the correlations of a sequence ; and ( iii ) consider the effect of non - stationarity on higher levels , which could cascade to lower levels and affect correlations properties .
finally , we believe that our approach may help to understand long - range correlations in any complex system for which an hierarchy of levels can be identified , such as human activities @xcite and dna sequences @xcite .
we thank b. lindner for insightful suggestions and s. graffi for the careful reading of the manuscript .
g.c . acknowledges partial support by the firb - project rbfr08uh60 ( miur , italy ) .
m. d. e. acknowledges partial support by the prin project 2008y4w3cy ( miur , italy )
. 99 schenkel a , zhang j , zhang y ( 1993 ) long range correlation in human writings .
_ fractals _ 1:47 - 55 .
alvarez - lacalle e , dorow b , eckmann jp , moses e , ( 2006 ) hierarchical structures induce long - range dynamical correlations in written texts .
_ proc natl acad sci usa _ 103:7956 - 7961 .
voss r , clarke j ( 1975 ) ` 1/f noise ' in music and speech .
_ nature _ 258:317 - 318 .
gilden d , thornton t , mallon m ( 1995 ) 1/f noise in human cognition .
_ science _ 267:1837 - 1839 .
muchnik l , havlin s , bunde a , stanley he ( 2005 ) scaling and memory in volatility return intervals in financial markets .
_ proc natl acad sci usa _ 102:9424 - 9428 .
rybski d , buldyrev sv , havlin s , liljeros f , makse ha ( 2009 ) scaling laws of human interaction activity .
_ proc natl acad sci _ 106:12640 - 12645 .
kello ct , brown gda , ferrer - i - cancho r , holden jg , linkenkaer - hansen k , rhodes t , van orden gc ( 2010 ) scaling laws in cognitive sciences .
_ trends cogn sci _ 14:223 - 232 press wh ( 1978 ) flicker noises in astronomy and elsewhere .
_ comments on astrophysics _
7:103 li w , kaneko k ( 1992 ) long - range correlation and partial @xmath185 spectrum in a noncoding dna sequence .
_ europhys lett _ 17:655 - 660 . peng ck , buldyrev s , goldberger a , havlin s , sciortino f , simons m , and stanley he ( 1992 ) long - range correlations in nucleotide sequences .
_ nature _ 356 : 168 - 171 .
voss rf ( 1992 ) evolution of long - range fractal correlations and @xmath186 noise in dna base sequences .
_ phys rev lett _ 68:3805 - 3808 .
c.d . manning , h. schtze ( 1999 ) _ foundations of statistical natural language processing _ , ( the mit press , cambridge , massachusetts , usa ) .
stamatatos e ( 2009 ) a survey of modern authorship attribution methods .
_ journal of the american society for information science and technology _ 60:538 - 556 .
oberlander j and brew c ( 2000 ) stochastic text generation .
_ phil trans r soc lond a _ 358:1373 - 1387 .
o usatenko , v yampolskii ( 2003 ) binary n - step markov chains and long - range correlated systems .
_ phys rev lett _ 90:110601 .
amit m , shmerler y , eisenberg e , abraham m , shnerb n ( 1994 ) language and codification dependence of long - range correlations in texts . _ fractals _ 2:7 - 13 ebeling w , neiman a ( 1995 ) long - range correlations between letters and sentences in texts . _
physica a _
215:233 - 241 .
ebeling w , pschel t ( 1994 ) entropy and long - range correlations in literary english .
_ europhys lett _ 26:241 - 246 .
allegrini p , grigolini p , palatella l ( 2004 ) intermittency and scale - free networks : a dynamical model for human language complexity .
_ chaos , solitons and fractals _ 20:95 - 105 .
melnyk ss , usatenko ov , and yampolskii va ( 2005 ) competition between two kinds of correlations in literary texts .
_ phys rev e _ 72:026140 .
herrera jp , pury pa ( 2008 ) statistical keyword detection in literary corpora .
_ eur phys j b _ 63:135 - 146 .
montemurro ma , zanette d ( 2010 ) towards the quantification of the semantic information encoded in written language .
_ adv comp syst _ 13:135 - 153 .
cover tm , thomas ja ( 2006 ) _ elements of information theory _
( wiley series in telecommunications and signal processing ) herzel h , groe i ( 1995 ) measuring correlations in symbol sequences .
_ physica a : statistical mechanics and its applications _ , 216:518 - 542 grassberger p ( 1989 ) estimating the information content of symbol sequences and efficient codes .
_ ieee transactions on information theory _ ,
35:669 - 675 .
kokol p , podgorelec v ( 2000 ) complexity and human writings .
_ complexity _ 7:1 - 6 .
kanter i , kessler da ( 1995 ) markov processes : linguistics and zipf s law .
_ phys rev lett _ 74:4559 - 4562 .
montemurro ma , pury pa ( 2002 ) long - range fractal correlations in literary corpora .
_ fractals _ 10:451 - 461 .
trefn g , floriani e , west bj and grigolini p , ( 1994 ) dynamical approach to anomalous diffusion : response of levy processes to a perturbation .
_ phys rev e _
50:2564 - 2579 .
cox dr , lewis paw ( 1978 ) _ the statistical analysis of series of events _ ( chapman and hall , london ) .
b. lindner ( 2006 ) superposition of many independent spike trains is generally not a poisson process .
_ phys rev e _ 73:022901 .
allegrini p , menicucci d , bedini r , gemignani a , paradisi p ( 2010 ) complex intermittency blurred by noise : theory and application to neural dynamics .
_ phys rev e _ 82:015103 .
goh k - i , barabasi a - l , burstiness and memory in complex systems .
_ europhys lett _ 81 : 48002 .
ortuno m , carpena p , bernaola - galvan p , munoz e , somoza am ( 2002 ) keyword detection in natural languages and dna .
_ europhys lett _ 57:759 - 764 . altmann eg , pierrehumbert jb , motter ae ( 2009 ) beyond word frequency : bursts , lulls , and scaling in the temporal distributions of words .
_ plos one _ 4:e7678 .
doxas i , dennis s , oliver wl ( 2009 ) the dimensionality of discourse .
_ proc natl acad science usa _ 107:4866 - 4871 .
saussure f de ( 1983 ) course in general linguistics , eds .
charles bally and albert sechehaye .
roy harris .
la salle , illinois ) badalamenti af ( 2001 ) speech parts as poisson processes .
_ journal of psycholinguistic research _ 30:31 .
schmitt ao , ebeling w , herzel h ( 1996 ) the modular structure of informational sequences .
_ biosystems _ 37:199210 . supporting information
given an ergodic and stationary stochastic process , correlation functions are defined as @xmath187 where @xmath188 denotes an average over different realizations @xmath19 of the process . stationarity guarantees that @xmath189 depends on the time lag @xmath11 only . in practice
, one typically has no access to different realizations of the process but only to a single finite sequence . in our case , any binary sequence
* x * is obtained from a single text of length @xmath190 through a given mapping . in such cases
it is possible to use the assumption of ergodicity to approximate the correlation function ( [ eq.corr ] ) by @xmath191 where @xmath12 means averaging , for each fixed @xmath11 , over all pairs @xmath192 and @xmath193 for @xmath194 as @xmath195
consider the sentence `` this paper is a paper of mine '' . by choosing the condition @xmath16 to be _ the @xmath3-th symbol is a vowel _
the projection @xmath196 maps the sentence into the sequence @xmath197 .
if @xmath16 is _ the k - th symbol is equal to ` e ' _ than we get : @xmath198 .
generally , we can treat any n - gram of letters in the same way , as for example by choosing the condition @xmath16 to be _
the @xmath199-gram starting at the k - th symbol is equal to ` er ' _ , that projects using a sliding window the sentence to:@xmath200 .
words are encoded using their corresponding n - gram , for example @xmath16 could be _ the 7-gram starting at the k - th symbol is equal to ` paper ' ( blank spaces included ) _ that gives : @xmath201 .
it is possible to generalize these procedures to more _ semantic conditions _ @xmath16 that associate @xmath0 to either all or part of the symbols that appears in a sentence that is attached to a specified topic .
these topics can be quantitatively constructed from the frequency of words using methods such as latent semantic analysis @xcite or the procedures to determine the so - called _ concept space _ @xcite .
we describe two simple procedures to construct two binary sequences * x * and * z * such that * x * is _ on top of _ * z*. these procedures will be based either on the addition " of @xmath0 s to * x * or on the subtraction " of @xmath0 s of * z*. in the simplest cases of _ random _ addition and subtraction , we explicitly compute how long - range correlations flow from * x * to * z * ( corresponding to a flow from upper to lower levels of the hierarchy ) and how burstiness is preserved when extracting * x * from z ( moving from lower to upper levels in the hierarchy ) .
recall that a sequence * x * is _ on top of _ * z * if for all @xmath39 such that @xmath202 we have @xmath203 , for a fixed constant @xmath204 . without loss of generality in the following calculations we fix for simplicity @xmath205 .
we now define simple operations that map two binary sequences into a third binary sequence : * given two generic binary sequences * z * and * @xmath97 * we define their multiplication * y*=*@xmath97 * * z * as @xmath206 . by construction
* y * is on top of * z*. * given two non - overlapping sequences * x * and * y * we define their sum * z * = * x*+*y * as @xmath207 . by construction
* x * and * y * are on top of * z*. we say that sequences @xmath14 and @xmath44 are non - overlapping if for all @xmath47 for which @xmath80 we have @xmath81 . in general , two independent binary sequences * x * and
* @xmath97 * will overlap . a sequence @xmath44 which is non
- overlapping with @xmath14 can be constructed from * @xmath97 * as * y*= * @xmath97*(*1 * -*x * ) , where * 1 * denotes the trivial sequence with all @xmath0 s . in this case
, we say that * z * = * x*+*y * , with * y * = * @xmath97*(*1 * -*x * ) is a sequence lower than * x * in the hierarchy that is constructed by a _ random addition _
( of 1 s ) to * x*. similarly , if * @xmath208 * is independent of * z * , the sequence * @xmath208**z * is a _ random subtraction _ ( of 1 s ) of * z * consider a sequence * z * constructed as a random addition of @xmath0 s to a given long - range correlated sequence @xmath19 : @xmath209 , with * y * = * @xmath97*(*1 * -*x * ) and @xmath97 a sequence of _ i.i.d . _
binary random variables .
the associated random walker @xmath48 spreads anomalously with the same exponent of @xmath27 .
this asymptotic regime is masked at short times by a pre - asymptotic normal behavior .
here we first compute explicitly the spreading of @xmath48 in terms of that of @xmath27 and @xmath210 and then we compute a bound for the transition time @xmath167 to the asymptotic anomalous diffusion of @xmath48 . as written in eq .
( 5 ) of the main text we have @xmath211 for our particular case we obtain @xmath212 and @xmath213 from eqs .
( [ eq.y ] ) and ( [ squared - app ] ) and noting that @xmath214 and @xmath215 we obtain @xmath216 the correlation term in eq .
( [ eq.sum ] ) can also be obtained through direct calculations : @xmath217\\ & = & -\left < \xi \right > \sigma^2_x(t).\label{eq.correl}\end{aligned}\ ] ] finally , inserting eqs . ( [ eq.sigma2 ] ) and ( [ eq.correl ] ) into eq . ( [ eq.sum ] )
we have @xmath218 as @xmath27 superdiffuses so it will @xmath48 and they both have the same asymptotic behavior . on the other hand the asymptotic regime
is masked at short times by a pre - asymptotic normal behavior , given by the linear term in @xmath11 .
we stress that , even if the non - overlapping condition for * y * forces both @xmath219 and @xmath220 to have the same asymptotic behavior of @xmath109 , their cumulative contributions does not cancel out unless we trivially have @xmath221 .
we now give a bound on the transition time @xmath167 to the asymptotic anomalous diffusion of eq .
( [ eq.sumsimple ] ) . without loss of generality
consider the case in which even the asymptotic anomalous behavior of @xmath27 is masked by generic pre -
asymptotic @xmath222 such that @xmath223\ ] ] with @xmath224 and @xmath222 increasing and such that @xmath225 for @xmath226 ( to guarantee that the asymptotic behavior is dominated by @xmath227 ) and @xmath228 ( as @xmath229 ) .
the asymptotic behavior @xmath230 in eq.([eq.sumsimple ] ) dominates only after a time @xmath167 such that : @xmath231 using the fact that the term @xmath232 is positive and that @xmath233 is monotonically increasing we finally have @xmath234 which corresponds to eq .
( 7 ) of the main text . in practice ,
any finite - time estimate @xmath235 is close to the asymptotic @xmath236 only if the estimate is performed for @xmath237 , otherwise @xmath238 ( @xmath239 if @xmath240 ) .
as noted in the main text , if * z * = * x*+*y * then @xmath241 = @xmath242 + * y*. applying to this relation the same arguments above , similar pre - asymptotic normal diffusion and transition time appear in the case of random subtraction , moving up in the hierarchy .
more specifically , starting from a sequence * z * such that asymptotically @xmath243 and constructing @xmath244 , with * @xmath208 * independent of * z * , we obtain a transition time @xmath167 for @xmath19 given by : @xmath245 which corresponds to eq .
( [ eq.tt ] ) above after properly replacing @xmath246 , @xmath247 and @xmath248 .
we consider the case of sequences as in eq .
( 8) of the main text : @xmath38 is a sequence emerging from a renewal process with algebraically decaying inter - event times , i.e. @xmath249 and @xmath250 .
given now a fixed @xmath251 , we consider the random subtraction @xmath98 where each @xmath252 is eventually set to @xmath253 with probability @xmath254 .
it is easy to see that the inter - event times of the new process will be distributed as : @xmath255 asymptotically @xmath256 is dominated by the long tails of @xmath257 : given a large @xmath96 , fix @xmath258 eventually diverging with @xmath259 and split accordingly the sum over @xmath3 in the second term of the right hand side .
the term corresponding to the sum @xmath260 is exponentially dominated by @xmath261 and arbitrary small , while the remaining finite sum over @xmath262 is controlled again by the tail of @xmath34 .
in our investigations we considered the english version of the @xmath143 popular novels listed in si - tab . books .
the texts were obtained through the gutenberg project ( http://www.gutenberg.org ) .
we implement a very mild pre - processing of the text that reduces the number of different symbols and simplifies our analysis : we consider as valid symbols the letters `` a - z '' , numbers `` 0 - 9 '' , the apostrophe `` '' and the blank space `` '' .
capitalization , punctuations and other markers were removed .
a string of symbols between two consecutive blank spaces is considered to be a word .
no lemmatization was applied to them so that plurals and singular forms are considered to be different words .
as described in the main text , the distinction between long - range and short - range correlation requires a finite - time estimate @xmath74 of the asymptotic diffusion exponent @xmath75 of the random - walkers associated to a binary sequence . in practice
, this corresponds to estimate the tails of the @xmath263 relation and it is therefore essential to estimate the upper limit in @xmath11 , denoted as @xmath170 , for which we have enough accuracy to provide a reasonable estimate @xmath74 .
we adopt the following procedure to estimate @xmath170 .
we consider a surrogate binary sequence with the same length @xmath190 and fraction of symbols ( @xmath0 s ) , but with the symbols randomly placed in the sequence .
for this sequence we know that @xmath264 .
we then consider instants of time @xmath265 equally spaced in a logarithmic scale of @xmath11 ( in practice we consider @xmath266 , with @xmath47 integer and @xmath267 ) .
we then estimate the local exponent as @xmath268/\log_{10}(1.2)$ ] . for small @xmath11 , @xmath269 but for larger @xmath11 statistical fluctuations
arise due to the finiteness of @xmath190 , as illustrated in fig .
[ fig.a1](a ) .
we choose @xmath170 as the smallest @xmath265 for which @xmath270 are all outside @xmath271 $ ] ( see fig . [ fig.a1]a ) .
we recall that our primary interest in the distinction between @xmath264 and @xmath272 .
the procedure described above is particularly suited for this distinction and an exponent @xmath142 obtained for large @xmath273 can be confidently regarded as a signature of super diffusion ( long - range correlation ) . in fig .
[ fig.a1 ] we verify that @xmath170 show no strong dependence on the fraction of @xmath0 s in the binary sequence ( inset ) and that it scales linearly with @xmath190 . based on these results , a good estimate of @xmath170 is @xmath274 , i.e. the safe interval for determining long - range correlation ends two decades before the size of the text .
this phenomenological rule was adopted in the estimate of @xmath74 for all cases .
the @xmath170 is only the upper limit and the estimate @xmath74 is performed through a least - squared fit in the time interval @xmath275 , where @xmath276 . in practice ,
we select @xmath143 different values of @xmath265 around @xmath277 and report the mean and variance over the different fittings as @xmath74 and its uncertainty , respectively .
we start clarifying the validity of the inequality @xmath278 where @xmath74 is the finite - time estimate of the total long - range correlation @xmath75 of a binary sequence * x * and @xmath279 is the estimate for the correlation due to the burstiness ( which can be quantified by shuffling * x * using the procedure @xmath138 of the main text ) . equation ( 4 ) of the main text shows that both burstiness @xmath280 and long - range correlations in the sequence of @xmath30 s contribute to the long - range correlations of a binary sequence @xmath19 .
while the @xmath139 contribution is always positive , the contribution from the correlation in @xmath30 s can be positive or negative . in principle , a negative contribution could precisely cancel the contribution of @xmath139 and violates the inequality ( [ eq.inequality ] ) .
conversely , this inequality is guaranteed to hold if the asymptotic contribution of the correlation in @xmath30 s of @xmath19 to @xmath281 is positive .
we now show that this is the case for the sequences we have argued to provide a good account of our observations .
consider high in the hierarchy a renewal sequence * x * with a given @xmath282 and broad tail in @xmath34 ( diverging @xmath139 ) .
adding many independent non - overlapping sequences , we construct a lower level sequence that still has long range correlation , with the same exponent @xmath75 ( see sec . [ sec.operations ] above ) . for this sequence we know that the broad tail in @xmath34 has a cutoff @xmath157 and thus burstiness gives no contribution to @xmath75 . instead , @xmath100 results solely from the correlations in the @xmath96 s , which are therefore necessarily positive .
it is natural to expect that this positiveness of the asymptotic correlation extends to finite times , in which case the ( finite time ) inequality ( [ eq.inequality ] ) holds .
indeed , for small @xmath283 , the distribution @xmath34 is not strongly affected by the independent additions and thus for @xmath108 a finite time estimate @xmath74 will receive contributions from both burstiness and @xmath284 correlations .
finally , we have directly tested the validity of eq .
( [ eq.inequality ] ) by comparing @xmath74 of different sequences * x * to the @xmath154 obtained from the corresponding @xmath285 ( a2-shuffled sequences of * x * , see main text ) . the inequality ( [ eq.inequality ] ) was confirmed for every single sequence we have analyzed , as shown by the fact that @xmath154 ( red symbols ) in fig .
[ fig.3si ] are systematically below their corresponding @xmath235 ( black circles ) .
we now obtain a quantitative lower bound for @xmath74 using eq .
( [ eq.inequality ] ) .
we consider a renewal sequence ( in which case @xmath286 ) with an inter - event time distribution given by @xmath287 where @xmath157 is the cut - off time , @xmath279 is the anomalous diffusion exponent for a renewal sequence with no cutoff @xmath288 , @xmath289 is a lower cut - off ( we fixed it at @xmath290 ) , and @xmath291 is a normalization constant .
we obtain the lower bound for @xmath74 as a function of @xmath139 by considering how @xmath154 and @xmath139 change with @xmath157 in the model above . for short times ( @xmath292 )
the corresponding walkers have not seen the cutoff and their diffusion will be anomalous with exponent @xmath293 . at longer time
( @xmath294 @xcite ) the diffusion becomes normal @xmath162 .
correspondingly , if the fitting interval @xmath295 $ ] used to compute the finite time @xmath154 ( see sec . [ ssec.confidence ] ) is all below @xmath157 ( i.e. @xmath296 ) we have @xmath293 while if the fitting interval is all beyond the cutoff ( i.e. @xmath297 ) we have @xmath162 . when @xmath157 is inside the fitting interval we approximate @xmath154 by linearly interpolating between @xmath279 and @xmath0
finally , we can compute @xmath139 by directly calculating the first and second moments of the distribution ( [ cutoffpt ] ) .
particularly important are the values @xmath298 and @xmath299 obtained evaluating @xmath139 at the critical values of the cutoff @xmath300 and @xmath301 , respectively . using the fact that @xmath139 is a monotonic increasing function of @xmath157 we can obtain explicitely the @xmath74 dependency on @xmath139 .
the @xmath154 for the case of a binary sequence with distribution ( [ cutoffpt ] ) is given by @xmath302,\\ \hat{\gamma}_{a2}&=\gamma_{a2 } \qquad \qquad \qquad \qquad \qquad & \textrm{if } \qquad \sm > s_2.\end{aligned}\ ] ] the red dashed line in fig .
[ fig.3si ] ( fig . 3 of the main text ) was computed using the fitting range corresponding to the book wrnpc @xmath303 ( see sec . [ ssec.confidence ] ) , and @xmath304 ( compatible with @xmath74 observed for words with large @xmath139 ) .
in addition to the shuffling methods presented in the main text , we discuss here briefly two cases : * * shuffle words * + mixing words order kills correlations for scales larger than the maximum word length @xcite .
even the blank space sequence @xmath305 becomes uncorrelated because its original correlations originate ( as in the case of all letters ) from the correlation in @xmath30 and not from tails in @xmath34 .
* * keep all blank spaces * in their original positions and fill the empty space between them with : * * two letters @xmath306 , placed randomly with probabilities @xmath307 and @xmath308 . * * the same letters of the book , placed in random positions .
+ by construction , correlation for blank space is trivially preserved .
what do we expect for the other letters ?
the following simple reasoning indicates that long - range correlation should be expected asymptotically in both cases : any letter sequence @xmath19 is on top of the reverted blank space sequence @xmath309 ; the results in sec .
[ sec.operations ] above show that either the selected sequence @xmath19 or its complement @xmath66 ( such that @xmath310 ) has @xmath311 ; and eq .
( [ eq.sumsimple ] ) above shows that any randomly chosen @xmath19 on top of @xmath309 has @xmath311 . in practice
these exponents are relevant only if the subsequence is dense enough in order for @xmath167 in eq .
( [ eq.ttsub ] ) above to be inside the observation range . for the first shuffling method and for our longest book ( wrnpc )
, we obtain that only if @xmath312 one finds @xmath313 book size . since the most frequent letter in a book has much smaller frequency ( around @xmath314 ) , we conclude that in practice all sequences obtained using the second shuffling mehthod have @xmath315 for all books of size smaller than @xmath316 symbols ( @xmath317 pages ) .
+ these simple calculations show that @xmath318 does not explain the correlations observed in the letters of the original text , as has been speculated in ref .
their origin are the long - range correlations on higher levels .
& & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 41414 & 0.440 & 0.020 & 1.18 & 0.05 & & & & + _ & 26666 & 0.379 & 0.011 & 1.13 & 0.06 & 1.13 & 0.06 & 1.13 & 0.06 + e & 13545 & 0.812 & 0.003 & 1.20 & 0.04 & 1.11 & 0.04 & 1.01 & 0.04 + t & 10667 & 0.858 & 0.003 & 1.17 & 0.05 & 1.05 & 0.03 & 1.05 & 0.03 + a & 8772 & 0.838 & 0.003 & 1.14 & 0.05 & 1.07 & 0.03 & 0.98 & 0.04 + o & 8128 & 0.920 & 0.002 & 1.25 & 0.05 & 1.13 & 0.04 & 0.99 & 0.04 + i & 7500 & 0.887 & 0.002 & 1.20 & 0.04 & 1.10 & 0.03 & 1.03 & 0.03 + h & 7379 & 0.848 & 0.003 & 1.15 & 0.04 & 1.11 & 0.04 & 1.04 & 0.03 + n & 7001 & 0.895 & 0.002 & 1.09 & 0.03 & 1.13 & 0.04 & 1.02 & 0.03 + s & 6497 & 0.925 & 0.002 & 1.11 & 0.04 & 1.09 & 0.03 & 1.07 & 0.03 + r & 5418 & 0.905 & 0.002 & 1.15 & 0.04 & 1.15 & 0.04 & 1.04 & 0.03 + d & 4928 & 0.878 & 0.003 & 1.06 & 0.03 & 1.10 & 0.04 & 0.97 & 0.04
+ l & 4704 & 1.081 & 0.003 & 1.20 & 0.06 & 1.12 & 0.04 & 1.00 & 0.03 + u & 3469 & 0.901 & 0.004 & 1.15 & 0.04 & 1.15 & 0.04 & 1.07 & 0.03 + w & 2681 & 0.966 & 0.003 & 1.11 & 0.04 & 1.23 & 0.05 & 0.99 & 0.04 + g & 2529 & 0.986 & 0.003 & 1.13 & 0.04 & 1.16 & 0.05 & 0.97 & 0.05 + c & 2397 & 0.980 & 0.005 & 1.15 & 0.05 & 1.11 & 0.04 & 1.00 & 0.03 + y & 2259 & 1.070 & 0.004 & 1.23 & 0.04 & 1.05 & 0.03 & 1.00 & 0.04 + m & 2103 & 1.030 & 0.005 & 1.16 & 0.04 & 1.24 & 0.05 & 0.98 & 0.03 + f & 1988 & 1.089 & 0.006 & 1.17 & 0.05 & 1.14 & 0.04 & 1.02 & 0.04 + p & 1514 & 1.143 & 0.006 & 1.17 & 0.04 & 1.13 & 0.04 & 1.05 & 0.03 + _ the _ & 1635 & 0.971 & 0.005 & 1.29 & 0.07 & & & & + _ and _ & 868 & 0.973 & 0.008 & 1.08 & 0.04 & & & & + _ to _ & 734 & 1.013 & 0.006 & 1.08 & 0.04 & & & & + _ a _ & 624 & 1.057 & 0.007 & 1.08 & 0.03 & & & & + _ she _ & 542 & 1.548 & 0.012 & 1.34 & 0.06 & & & & + _ it _ & 530 & 1.172 & 0.008 & 1.22 & 0.04 & & & & + _ alice _ & 386 & 0.885 & 0.008 & 1.01 & 0.05 & & & & + _ in _ & 367 & 0.959 & 0.008 & 1.04 & 0.03 & & & & + _ way _ & 57 & 1.098 & 0.021 & 1.21 & 0.04 & & & & + _ turtle _ & 57 & 4.066 & 0.039 & 1.47 & 0.12 & & & & + _ hatter _ & 55 & 4.978 & 0.050 & 1.46 & 0.12 & & & & + _ gryphon _ & 55 & 3.541 & 0.038 & 1.43 & 0.10 & & & & + _ quite _ & 55 & 1.290 & 0.025 & 1.16 & 0.04 & & & & + _ mock _ & 55 & 3.919 & 0.045 & 1.45 & 0.12 & & & & + _ are _ & 54 & 1.250 & 0.024 & 1.10 & 0.03 & & & & + _ think _ & 52 & 1.268 & 0.039 & 1.09 & 0.04 & & & & + _ more _ & 49 & 1.040 & 0.023 & 1.11 & 0.04 & & & & + _ head _ & 49 & 1.230 & 0.024 & 1.07 & 0.03 & & & & + _ never _ & 48 & 1.083 & 0.049 & 1.09 & 0.05 & & & & + _ voice _ & 47 & 1.359 & 0.054 & 1.07 & 0.03 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 358397 & 0.454 & 0.020 & 1.25 & 0.04 & & & & + _ & 208375 & 0.434 & 0.009 & 1.34 & 0.04 & 1.34 & 0.04 & 1.34 & 0.04 + e & 123056 & 0.817 & 0.002 & 1.18 & 0.04 & 1.22 & 0.05 & 0.96 & 0.05 + t & 86576 & 0.836 & 0.002 & 1.18 & 0.04 & 1.15 & 0.04 & 0.98 & 0.04 + a & 78914 & 0.847 & 0.002 & 1.11 & 0.04 & 1.11 & 0.04 & 1.03 & 0.03 + o & 67896 & 0.885 & 0.001 & 1.20 & 0.04 & 1.24 & 0.04 & 0.96 & 0.05
+ n & 64597 & 0.865 & 0.002 & 1.13 & 0.04 & 1.18 & 0.04 & 1.02 & 0.04 + i & 63755 & 0.889 & 0.001 & 1.22 & 0.04 & 1.15 & 0.04 & 1.07 & 0.03 + s & 62383 & 0.909 & 0.001 & 1.21 & 0.04 & 1.20 & 0.04 & 1.05 & 0.03 + r & 59027 & 0.870 & 0.001 & 1.17 & 0.04 & 1.09 & 0.04 & 1.05 & 0.03 + h & 54880 & 0.861 & 0.002 & 1.23 & 0.04 & 1.07 & 0.05 & 1.09 & 0.03
+ l & 38467 & 1.033 & 0.001 & 1.28 & 0.05 & 1.10 & 0.04 & 1.03 & 0.03 + d & 37051 & 0.923 & 0.001 & 1.24 & 0.04 & 1.19 & 0.04 & 1.01 & 0.03 + c & 27687 & 0.978 & 0.001 & 1.28 & 0.04 & 1.17 & 0.04 & 1.11 & 0.03 + u & 24776 & 0.957 & 0.001 & 1.15 & 0.04 & 1.12 & 0.04 & 1.04 & 0.03 + f & 24052 & 0.955 & 0.001 & 1.18 & 0.04 & 1.12 & 0.04 & 1.08 & 0.03 + m & 21509 & 0.987 & 0.001 & 1.20 & 0.04 & 1.20 & 0.04 & 1.03 & 0.04 + w & 19172 & 1.010 & 0.001 & 1.36 & 0.05 & 1.21 & 0.04 & 1.07 & 0.03 + g & 18284 & 0.962 & 0.001 & 1.20 & 0.04 & 1.09 & 0.05 & 0.99 & 0.03 + p & 16742 & 1.040 & 0.001 & 1.24 & 0.04 & 1.10 & 0.03 & 1.04 & 0.03 + y & 15700 & 0.993 & 0.002 & 1.26 & 0.04 & 1.15 & 0.04 & 1.03 & 0.04 + _ the _ & 16882 & 0.924 & 0.002 & 1.21 & 0.04 & & & & + _ of _ & 9414 & 0.970 & 0.002 & 1.25 & 0.04 & & & & + _ and _ & 5765 & 0.897 & 0.003 & 1.10 & 0.04 & & & & + _ a _ & 5326 & 1.097 & 0.003 & 1.19 & 0.04 & & & & + _ in _ & 4287 & 1.022 & 0.003 & 1.14 & 0.04 & & & & + _ to _ & 4080 & 1.051 & 0.003 & 1.20 & 0.04 & & & & + _ water _ & 417 & 1.509 & 0.011 & 1.26 & 0.04 & & & & + _
little _ & 412 & 1.117 & 0.011 & 1.08 & 0.03 & & & & + _ where _ & 349 & 1.086 & 0.011 & 1.06 & 0.03 & & & & + _ sea _ & 348 & 1.534 & 0.015 & 1.29 & 0.04 & & & & + _ much _ & 338 & 1.112 & 0.010 & 1.08 & 0.04 & & & & + _ country _ & 337 & 1.519 & 0.011 & 1.28 & 0.05 & & & & + _ land _ & 318 & 1.387 & 0.012 & 1.25 & 0.04 & & & & + _ must _ & 317 & 1.290 & 0.009 & 1.12 & 0.04 & & & & + _ feet _ & 312 & 1.391 & 0.013 & 1.23 & 0.04 & & & & + _ may _ & 311 & 1.118 & 0.010 & 1.09 & 0.03 & & & & + _ species _ & 303 & 2.459 & 0.022 & 1.55 & 0.05 & & & & + _ found _ & 303 & 1.217 & 0.010 & 1.16 & 0.04 & & & & + _ me _ & 301 & 1.206 & 0.012 & 1.07 & 0.04 & & & & + _ day _ & 301 & 1.375 & 0.007 & 1.12 & 0.03 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 237050 & 0.420 & 0.020 & 1.26 & 0.05 & & & & + _ & 151300 & 0.404 & 0.010 & 1.53 & 0.05 & 1.53 & 0.05 & 1.53 & 0.05 + e & 78161 & 0.843 & 0.002 & 1.16 & 0.04 & 1.09 & 0.04 & 1.04 & 0.04 + t & 58475 & 0.873 & 0.002 & 1.32 & 0.05 & 1.19 & 0.04 & 1.04 & 0.03 + a & 53663 & 0.854 & 0.002 & 1.21 & 0.04 & 1.17 & 0.04 & 1.08 & 0.03 + o & 47726 & 0.896 & 0.001 & 1.18 & 0.04 & 1.24 & 0.04 & 0.97 & 0.04 +
n & 44497 & 0.874 & 0.002 & 1.14 & 0.04 & 1.18 & 0.04 & 1.12 & 0.03 + h & 44473 & 0.832 & 0.002 & 1.37 & 0.05 & 1.27 & 0.04 & 1.17 & 0.05 + i & 40025 & 0.906 & 0.001 & 1.34 & 0.05 & 1.23 & 0.05 & 1.13 & 0.04 + s & 37500 & 0.941 & 0.001 & 1.32 & 0.05 & 1.37 & 0.04 & 1.07 & 0.03 + r & 34514 & 0.888 & 0.002 & 1.19 & 0.04 & 1.19 & 0.04 & 1.09 & 0.04 + d & 30491 & 0.929 & 0.001 & 1.31 & 0.06 & 1.22 & 0.04 & 1.00 & 0.04 + l & 24876 & 1.059 & 0.001 & 1.20 & 0.04 & 1.10 & 0.04 & 1.07 & 0.03 + u & 17475 & 0.943 & 0.001 & 1.16 & 0.04 & 1.22 & 0.04 & 0.99 & 0.05 + w & 17213 & 0.948 & 0.002 & 1.36 & 0.06 & 1.30 & 0.05 & 1.06 & 0.03 + m & 14754 & 0.977 & 0.001 & 1.18 & 0.04 & 1.24 & 0.04 & 1.04 & 0.03 + c & 14148 & 1.006 & 0.001 & 1.37 & 0.05 & 1.18 & 0.04 & 1.10 & 0.04 + g & 14069 & 0.994 & 0.002 & 1.28 & 0.06 & 1.27 & 0.04 & 1.06 & 0.03 + f & 13862 & 1.016 & 0.002 & 1.25 & 0.04 & 1.21 & 0.04 & 1.09 & 0.04 + y & 10868 & 1.068 & 0.002 & 1.29 & 0.05 & 1.16 & 0.04 & 0.95 & 0.05 + p & 9940 & 1.074 & 0.002 & 1.24 & 0.05 & 1.24 & 0.04 & 1.06 & 0.04 + _ the _ & 8930 & 1.018 & 0.003 & 1.34 & 0.04 & & & & + _ and _ & 7280 & 0.958 & 0.002 & 1.27 & 0.04 & & & & + _ of _ & 4365 & 1.113 & 0.003 & 1.42 & 0.07 & & & & + _ to _ & 4190 & 1.077 & 0.003 & 1.20 & 0.04 & & & & + _ a _ & 4158 & 1.152 & 0.004 & 1.22 & 0.04 & & & & + _ he _ & 3311 & 2.158 & 0.011 & 1.60 & 0.05 & & & & + _ him _
& 1184 & 2.009 & 0.013 & 1.42 & 0.05 & & & & + _ jurgis _ & 1098 & 2.077 & 0.010 & 1.48 & 0.07 & & & & + _ i _ & 485 & 6.141 & 0.275 & 1.54 & 0.06 & & & & + _ man _ & 463 & 1.301 & 0.013 & 1.27 & 0.04 & & & & + _ said _ & 367 & 1.975 & 0.019 & 1.38 & 0.04 & & & & + _ time _ & 356 & 1.209 & 0.013 & 1.15 & 0.04 & & & & + _ men _ & 329 & 1.768 & 0.011 & 1.33 & 0.05 & & & & + _ now _ & 325 & 1.077 & 0.009 & 1.11 & 0.03 & & & & + _ day _ & 280 & 1.378 & 0.021 & 1.15 & 0.04 & & & & + _ other _ & 279 & 1.244 & 0.014 & 1.16 & 0.04 & & & & + _ place _ & 263 & 1.227 & 0.013 & 1.17 & 0.04 & & & & + _ only _ & 261 & 1.042 & 0.010 & 1.03 & 0.04 & & & & + _ before _ & 235 & 1.117 & 0.010 & 1.09 & 0.03 & & & & + _ home _ & 229 & 1.759 & 0.012 & 1.23 & 0.04 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 235370 & 0.445 & 0.020 & 1.48 & 0.05 & & & & + _ & 146786 & 0.429 & 0.009 & 1.65 & 0.05 & 1.65 & 0.05 & 1.65 & 0.05 + e & 76483 & 0.850 & 0.002 & 1.40 & 0.05 & 1.11 & 0.04 & 1.10 & 0.03 + t & 59660 & 0.858 & 0.002 & 1.24 & 0.04 & 1.18 & 0.04 & 1.02 & 0.04 + a & 51642 & 0.859 & 0.002 & 1.23 & 0.04 & 1.18 & 0.04 & 1.11 & 0.04 + o & 47123 & 0.890 & 0.002 & 1.23 & 0.04 & 1.22 & 0.04 & 1.05 & 0.04 + n & 44064 & 0.869 & 0.002 & 1.24 & 0.04 & 1.24 & 0.04 & 1.08 & 0.03 + i & 42750 & 0.920 & 0.001 & 1.30 & 0.04 & 1.19 & 0.04 & 1.11 & 0.03 + s & 38995 & 0.940 & 0.001 & 1.34 & 0.06 & 1.23 & 0.04 & 1.20 & 0.04 + h & 36904 & 0.859 & 0.002 & 1.40 & 0.05 & 1.13 & 0.04 & 1.20 & 0.04 + r & 35465 & 0.912 & 0.001 & 1.34 & 0.05 & 1.19 & 0.04 & 1.16 & 0.04 + d & 27682 & 0.974 & 0.001 & 1.40 & 0.06 & 1.24 & 0.04 & 1.05 & 0.03
+ l & 24910 & 1.055 & 0.001 & 1.20 & 0.04 & 1.12 & 0.04 & 1.06 & 0.03 + u & 17372 & 0.947 & 0.002 & 1.20 & 0.04 & 1.17 & 0.04 & 1.07 & 0.03 + w & 15554 & 0.996 & 0.002 & 1.30 & 0.04 & 1.25 & 0.04 & 1.10 & 0.03 + m & 14940 & 1.006 & 0.002 & 1.29 & 0.04 & 1.27 & 0.04 & 1.09 & 0.04 + c & 14884 & 1.042 & 0.001 & 1.35 & 0.05 & 1.21 & 0.04 & 1.21 & 0.06 + f & 14234 & 1.006 & 0.001 & 1.24 & 0.05 & 1.14 & 0.04 & 1.04 & 0.03 + g & 12890 & 1.044 & 0.001 & 1.26 & 0.04 & 1.17 & 0.04 & 1.09 & 0.03 + y & 11994 & 1.022 & 0.002 & 1.34 & 0.04 & 1.14 & 0.04 & 1.02 & 0.03 + p & 11087 & 1.093 & 0.002 & 1.30 & 0.05 & 1.17 & 0.04 & 1.16 & 0.05 + _ the _ & 9091 & 1.043 & 0.003 & 1.38 & 0.04 & & & & + _ and _ & 5898 & 0.995 & 0.003 & 1.34 & 0.05 & & & & + _ of _ & 4380 & 1.033 & 0.003 & 1.32 & 0.05 & & & & + _ a _ & 4057 & 1.098 & 0.003 & 1.22 & 0.04 & & & & + _ to _ & 3545 & 1.095 & 0.004 & 1.24 & 0.04 & & & & + _ in _ & 2555 & 1.031 & 0.004 & 1.14 & 0.04 & & & & + _ would _
& 480 & 1.552 & 0.012 & 1.26 & 0.04 & & & & + _ river _ & 478 & 2.176 & 0.014 & 1.43 & 0.06 & & & & + _ water _ & 242 & 1.899 & 0.015 & 1.38 & 0.05 & & & & + _ she _ & 239 & 2.055 & 0.022 & 1.44 & 0.06 & & & & + _ boat _ & 212 & 1.921 & 0.028 & 1.32 & 0.05 & & & & + _ here _ & 210 & 1.508 & 0.015 & 1.24 & 0.04 & & & & + _ night _ & 177 & 1.609 & 0.012 & 1.30 & 0.05 & & & & + _ can _ & 177 & 1.392 & 0.015 & 1.13 & 0.04 & & & & + _ go _ & 176 & 1.275 & 0.010 & 1.16 & 0.04 & & & & + _ head _ & 175 & 1.612 & 0.017 & 1.41 & 0.06 & & & & + _ pilot _ & 172 & 2.652 & 0.047 & 1.40 & 0.05 & & & & + _ long _ & 172 & 1.246 & 0.013 & 1.06 & 0.03 & & & & + _ first _ & 164 & 1.132 & 0.018 & 1.11 & 0.04 & & & & + _ miles _ & 162 & 1.816 & 0.030 & 1.49 & 0.05 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 356037 & 0.441 & 0.020 & 1.45 & 0.05 & & & & + _ & 215939 & 0.424 & 0.009 & 1.54 & 0.05 & 1.54 & 0.05 & 1.54 & 0.05 + e & 116938 & 0.859 & 0.002 & 1.29 & 0.04 & 1.12 & 0.04 & 1.05 & 0.03 + t & 87882 & 0.860 & 0.002 & 1.23 & 0.04 & 1.25 & 0.04 & 1.00 & 0.03 + a & 77820 & 0.851 & 0.002 & 1.24 & 0.04 & 1.22 & 0.05 & 1.05 & 0.03 + o & 69258 & 0.900 & 0.001 & 1.27 & 0.04 & 1.16 & 0.04 & 1.09 & 0.03 +
n & 65552 & 0.886 & 0.001 & 1.20 & 0.04 & 1.20 & 0.04 & 1.07 & 0.03 + i & 65349 & 0.905 & 0.001 & 1.28 & 0.04 & 1.11 & 0.04 & 1.09 & 0.03 + s & 64148 & 0.917 & 0.001 & 1.34 & 0.05 & 1.31 & 0.04 & 1.15 & 0.04 + h & 62824 & 0.856 & 0.002 & 1.32 & 0.04 & 1.38 & 0.06 & 1.21 & 0.04 + r & 52073 & 0.900 & 0.002 & 1.32 & 0.04 & 1.19 & 0.04 & 1.14 & 0.04 + l & 42733 & 1.051 & 0.001 & 1.22 & 0.04 & 1.20 & 0.04 & 0.99 & 0.03 + d & 38192 & 0.969 & 0.001 & 1.42 & 0.05 & 1.19 & 0.04 & 1.05 & 0.03 + u & 26672 & 0.968 & 0.001 & 1.23 & 0.04 & 1.09 & 0.03 & 1.02 & 0.03 + m & 23243 & 0.998 & 0.001 & 1.22 & 0.04 & 1.16 & 0.04 & 0.96 & 0.04 + c & 22482 & 1.031 & 0.001 & 1.32 & 0.04 & 1.30 & 0.04 & 1.15 & 0.04 + w & 22193 & 0.957 & 0.001 & 1.24 & 0.04 & 1.23 & 0.04 & 1.06 & 0.03 + f & 20812 & 0.997 & 0.001 & 1.33 & 0.05 & 1.20 & 0.04 & 1.01 & 0.04 + g & 20801 & 1.009 & 0.001 & 1.32 & 0.04 & 1.11 & 0.03 & 1.07 & 0.04 + p & 17233 & 1.057 & 0.001 & 1.23 & 0.04 & 1.13 & 0.04 & 1.12 & 0.03 + y & 16852 & 1.037 & 0.001 & 1.25 & 0.05 & 1.22 & 0.04 & 0.98 & 0.04 + _ the _ & 14404 & 1.033 & 0.002 & 1.40 & 0.04 & & & & + _ of _ & 6600 & 1.073 & 0.003 & 1.47 & 0.06 & & & & + _ and _ & 6428 & 0.962 & 0.002 & 1.23 & 0.04 & & & & + _ a _ & 4722 & 1.137 & 0.003 & 1.34 & 0.05 & & & & + _ to _ & 4619 & 1.023 & 0.003 & 1.15 & 0.04 & & & & + _ in _ & 4166 & 1.021 & 0.003 & 1.30 & 0.05 & & & & + _ whale _ & 1096 & 2.162 & 0.018 & 1.57 & 0.07 & & & & + _ from _ & 1085 & 1.143 & 0.006 & 1.15 & 0.04 & & & & + _ man _ & 476 & 1.252 & 0.007 & 1.21 & 0.04 & & & & + _ them _ & 474 & 1.214 & 0.012 & 1.16 & 0.04 & & & & + _ sea _ & 453 & 1.311 & 0.009 & 1.24 & 0.04 & & & & + _ old _ & 450 & 1.507 & 0.012 & 1.33 & 0.04 & & & & + _ we _ & 445 & 1.646 & 0.011 & 1.28 & 0.05 & & & & + _ ship _ & 438 & 1.522 & 0.012 & 1.31 & 0.04 & & & & + _ ahab _ & 436 & 3.056 & 0.021 & 1.53 & 0.06 & & & & + _ ye _ & 431 & 2.680 & 0.018 & 1.43 & 0.04 & & & & + _ who _ & 344 & 1.136 & 0.012 & 1.22 & 0.04 & & & & + _ head _ & 342 & 1.346 & 0.012 & 1.35 & 0.05 & & & & + _ time _ & 333 & 1.086 & 0.014 & 1.08 & 0.04 & & & & + _ long _ & 333 & 1.092 & 0.009 & 1.07 & 0.03 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 203916 & 0.437 & 0.020 & 1.20 & 0.04 & & & & + _ & 122194 & 0.450 & 0.008 & 1.41 & 0.05 & 1.41 & 0.05 & 1.41 & 0.05 + e & 69370 & 0.828 & 0.002 & 1.19 & 0.04 & 1.12 & 0.04 & 1.08 & 0.04 + t & 46645 & 0.872 & 0.002 & 1.10 & 0.05 & 1.09 & 0.04 & 1.02 & 0.03 + a & 41688 & 0.849 & 0.002 & 1.11 & 0.03 & 1.18 & 0.04 & 1.04 & 0.04 + o & 40041 & 0.891 & 0.001 & 1.18 & 0.04 & 1.09 & 0.03 & 1.00 & 0.04 + i & 37830 & 0.870 & 0.002 & 1.16 & 0.04 & 1.31 & 0.04 & 1.09 & 0.04 +
n & 37689 & 0.884 & 0.001 & 1.13 & 0.04 & 1.16 & 0.04 & 1.09 & 0.03 + h & 34067 & 0.869 & 0.002 & 1.31 & 0.04 & 1.04 & 0.04 & 1.10 & 0.03 + s & 33114 & 0.956 & 0.001 & 1.06 & 0.03 & 1.11 & 0.03 & 1.06 & 0.03 + r & 32299 & 0.882 & 0.001 & 1.18 & 0.04 & 1.09 & 0.04 & 1.06 & 0.04 + d & 22303 & 0.917 & 0.002 & 1.15 & 0.04 & 1.11 & 0.03 & 1.05 & 0.03 + l & 21594 & 1.036 & 0.001 & 1.19 & 0.05 & 1.10 & 0.04 & 1.03 & 0.04 + u & 14987 & 0.971 & 0.002 & 1.26 & 0.04 & 1.17 & 0.04 & 1.03 & 0.03 + m & 14764 & 0.963 & 0.002 & 1.17 & 0.04 & 1.17 & 0.04 & 1.02 & 0.03 + c & 13461 & 1.005 & 0.002 & 1.27 & 0.06 & 1.20 & 0.04 & 1.04 & 0.04 + y & 12706 & 0.992 & 0.002 & 1.37 & 0.05 & 1.20 & 0.05 & 1.04 & 0.03 + w & 12305 & 0.949 & 0.002 & 1.23 & 0.04 & 1.14 & 0.04 & 1.06 & 0.04 + f & 11998 & 0.988 & 0.002 & 1.23 & 0.04 & 1.05 & 0.03 & 1.05 & 0.03 + g & 10031 & 0.949 & 0.002 & 1.06 & 0.04 & 1.17 & 0.04 & 1.03 & 0.03 + b & 9088 & 0.943 & 0.002 & 1.19 & 0.05 & 1.08 & 0.03 & 0.99 & 0.04 + _ the _ & 4331 & 1.083 & 0.003 & 1.24 & 0.04 & & & & + _ to _ & 4163 & 0.945 & 0.003 & 1.11 & 0.03 & & & & + _ of _ & 3609 & 0.974 & 0.003 & 1.21 & 0.04 & & & & + _ and _ & 3585 & 0.859 & 0.003 & 1.18 & 0.04 & & & & + _ her _ & 2225 & 1.592 & 0.015 & 1.31 & 0.04 & & & & + _ i _ & 2068 & 2.915 & 0.014 & 1.46 & 0.05 & & & & + _ at _ & 788 & 1.071 & 0.006 & 1.10 & 0.04 & & & & + _ mr _ & 786 & 1.218 & 0.007 & 1.32 & 0.06 & & & & + _ they _ & 601 & 1.459 & 0.010 & 1.26 & 0.04 & & & & + _ elizabeth _ & 597 & 1.192 & 0.027 & 1.17 & 0.06 & & & & + _ or _ & 300 & 1.026 & 0.010 & 1.00 & 0.03 & & & & + _ bennet _ & 294 & 2.047 & 0.034 & 1.37 & 0.07 & & & & + _ who _ & 284 & 1.148 & 0.010 & 1.06 & 0.03 & & & & + _ miss _ & 283 & 1.536 & 0.015 & 1.35 & 0.07 & & & & + _ one _ & 268 & 1.066 & 0.009 & 1.06 & 0.04 & & & & + _ jane _ & 264 & 1.741 & 0.016 & 1.29 & 0.06 & & & & + _ bingley _ & 257 & 3.166 & 0.019 & 1.45 & 0.08 & & & & + _ we _ & 253 & 1.546 & 0.013 & 1.26 & 0.04 & & & & + _ own _ & 183 & 1.078 & 0.015 & 1.06 & 0.04 & & & & + _ lady _ & 183 & 1.924 & 0.023 & 1.38 & 0.06 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 638882 & 0.430 & 0.020 & 1.26 & 0.04 & & & & + _ & 402964 & 0.415 & 0.009 & 1.34 & 0.05 & 1.34 & 0.05 & 1.34 & 0.05 + e & 204300 & 0.840 & 0.002 & 1.25 & 0.04 & 1.25 & 0.04 & 1.03 & 0.03 + t & 157193 & 0.867 & 0.002 & 1.26 & 0.04 & 1.25 & 0.04 & 1.03 & 0.03 + a & 138706 & 0.841 & 0.002 & 1.25 & 0.04 & 1.28 & 0.04 & 1.09 & 0.03 + o & 136541 & 0.881 & 0.001 & 1.24 & 0.04 & 1.29 & 0.04 & 1.03 & 0.03 + h & 117821 & 0.852 & 0.002 & 1.23 & 0.04 & 1.10 & 0.04 & 1.06 & 0.03 +
n & 115898 & 0.866 & 0.002 & 1.23 & 0.04 & 1.17 & 0.04 & 1.05 & 0.04 + i & 112746 & 0.881 & 0.001 & 1.26 & 0.04 & 1.19 & 0.04 & 1.04 & 0.03 + s & 106979 & 0.935 & 0.001 & 1.28 & 0.05 & 1.27 & 0.04 & 1.06 & 0.03 + r & 92501 & 0.910 & 0.001 & 1.27 & 0.04 & 1.28 & 0.04 & 1.06 & 0.03 + d & 76655 & 0.929 & 0.001 & 1.34 & 0.04 & 1.22 & 0.04 & 1.03 & 0.03 + l & 62107 & 1.108 & 0.002 & 1.24 & 0.04 & 1.28 & 0.05 & 1.01 & 0.03 + u & 46589 & 0.949 & 0.001 & 1.23 & 0.04 & 1.17 & 0.04 & 1.00 & 0.04 + m & 42945 & 0.992 & 0.001 & 1.21 & 0.04 & 1.25 & 0.04 & 1.04 & 0.05 + f & 38552 & 0.977 & 0.001 & 1.24 & 0.04 & 1.24 & 0.04 & 1.05 & 0.03 + w & 38209 & 0.986 & 0.001 & 1.29 & 0.04 & 1.21 & 0.04 & 1.05 & 0.03 + c & 37602 & 0.984 & 0.001 & 1.21 & 0.04 & 1.26 & 0.04 & 1.08 & 0.03 + g & 31927 & 0.988 & 0.001 & 1.22 & 0.04 & 1.23 & 0.04 & 1.03 & 0.03 + y & 31053 & 1.048 & 0.001 & 1.25 & 0.04 & 1.26 & 0.04 & 1.05 & 0.04 + p & 23880 & 1.069 & 0.001 & 1.23 & 0.04 & 1.26 & 0.04 & 1.11 & 0.03 + _ the _ & 20652 & 1.050 & 0.002 & 1.36 & 0.04 & & & & + _ and _ & 16835 & 0.908 & 0.002 & 1.22 & 0.05 & & & & + _ to _ & 13184 & 1.031 & 0.002 & 1.26 & 0.04 & & & & + _ of _ & 12173 & 1.033 & 0.002 & 1.26 & 0.04 & & & & + _ that _ & 7515 & 1.023 & 0.002 & 1.21 & 0.04 & & & & + _ in _ & 6716 & 1.023 & 0.002 & 1.11 & 0.03 & & & & + _ by _ & 2069 & 1.042 & 0.004 & 1.09 & 0.03 & & & & + _ sancho _ & 2063 & 3.762 & 0.025 & 1.63 & 0.05 & & & & + _ or _ & 2048 & 1.154 & 0.004 & 1.15 & 0.04 & & & & + _ quixote _ & 2002 & 3.214 & 0.016 & 1.55 & 0.06 & & & & + _ other _ & 609 & 1.072 & 0.008 & 1.11 & 0.04 & & & & + _ knight _ & 606 & 2.175 & 0.016 & 1.43 & 0.05 & & & & + _ take _ & 546 & 1.195 & 0.008 & 1.14 & 0.04 & & & & + _ master _ & 545 & 1.720 & 0.013 & 1.38 & 0.04 & & & & + _ thy _ & 510 & 2.252 & 0.017 & 1.35 & 0.05 & & & & + _ senor _ & 509 & 1.632 & 0.009 & 1.28 & 0.04 & & & & + _ worship _ & 470 & 2.337 & 0.012 & 1.37 & 0.04 & & & & + _ here _ & 467 & 1.237 & 0.007 & 1.14 & 0.04 & & & & + _ god _ & 467 & 1.169 & 0.011 & 1.12 & 0.04 & & & & + _ way _ & 466 & 1.056 & 0.006 & 1.07 & 0.04 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 110026 & 0.432 & 0.020 & 1.23 & 0.04 & & & & + _ & 71180 & 0.402 & 0.010 & 1.50 & 0.05 & 1.50 & 0.05 & 1.50 & 0.05 + e & 35603 & 0.864 & 0.002 & 1.30 & 0.04 & 1.05 & 0.05 & 0.99 & 0.04 + t & 28825 & 0.858 & 0.002 & 1.16 & 0.04 & 1.23 & 0.04 & 0.99 & 0.04 + a & 23478 & 0.858 & 0.002 & 1.17 & 0.04 & 1.03 & 0.03 & 1.13 & 0.04 + o & 23192 & 0.898 & 0.001 & 1.22 & 0.04 & 1.06 & 0.03 & 0.98 & 0.04 +
n & 20146 & 0.866 & 0.002 & 1.12 & 0.04 & 1.23 & 0.04 & 1.07 & 0.03 + h & 19565 & 0.861 & 0.002 & 1.18 & 0.04 & 1.11 & 0.04 & 1.07 & 0.04 + i & 18811 & 0.910 & 0.002 & 1.16 & 0.05 & 1.23 & 0.04 & 1.05 & 0.03 + s & 17716 & 0.951 & 0.001 & 1.19 & 0.04 & 1.13 & 0.04 & 1.08 & 0.03 + r & 15247 & 0.917 & 0.002 & 1.30 & 0.05 & 1.13 & 0.04 & 1.10 & 0.05 + d & 14850 & 0.950 & 0.002 & 1.20 & 0.04 & 1.20 & 0.04 & 1.01 & 0.03 + l & 12136 & 1.086 & 0.002 & 1.17 & 0.04 & 1.14 & 0.04 & 1.06 & 0.03 + u & 8942 & 0.949 & 0.002 & 1.18 & 0.04 & 1.07 & 0.04 & 1.06 & 0.03 + w & 8042 & 0.949 & 0.002 & 1.13 & 0.03 & 1.18 & 0.04 & 1.12 & 0.04 + m & 7135 & 0.977 & 0.002 & 1.22 & 0.04 & 1.18 & 0.04 & 1.02 & 0.03 + y & 6725 & 1.043 & 0.002 & 1.36 & 0.04 & 1.04 & 0.03 & 1.00 & 0.04 + g & 6606 & 1.041 & 0.002 & 1.16 & 0.04 & 1.15 & 0.05 & 1.06 & 0.03 + c & 6497 & 1.030 & 0.003 & 1.23 & 0.05 & 1.09 & 0.05 & 1.16 & 0.04 + f & 6004 & 1.047 & 0.003 & 1.22 & 0.04 & 1.11 & 0.03 & 1.02 & 0.03 + b & 4958 & 0.959 & 0.003 & 1.10 & 0.04 & 1.25 & 0.04 & 1.02 & 0.03 + _ the _ & 3703 & 1.154 & 0.004 & 1.35 & 0.04 & & & & + _ and _ & 3105 & 1.008 & 0.003 & 1.21 & 0.04 & & & & + _ a _ & 1863 & 1.085 & 0.005 & 1.20 & 0.04 & & & & + _ to _ & 1727 & 1.054 & 0.004 & 1.14 & 0.03 & & & & + _ of _ & 1436 & 1.127 & 0.005 & 1.21 & 0.04 & & & & + _ he _ & 1197 & 1.770 & 0.015 & 1.40 & 0.04 & & & & + _ tom _ & 689 & 1.740 & 0.014 & 1.39 & 0.06 & & & & + _ with _ & 647 & 1.068 & 0.008 & 1.15 & 0.04 & & & & + _ if _ & 237 & 1.404 & 0.011 & 1.23 & 0.04 & & & & + _ huck _ & 223 & 3.228 & 0.024 & 1.46 & 0.07 & & & & + _ boys _ & 155 & 1.767 & 0.019 & 1.24 & 0.06 & & & & + _ did _ & 150 & 1.336 & 0.018 & 1.22 & 0.04 & & & & + _ joe _ & 133 & 2.248 & 0.051 & 1.38 & 0.06 & & & & + _ never _ & 131 & 1.185 & 0.017 & 1.14 & 0.04 & & & & + _ boy _ & 122 & 1.788 & 0.054 & 1.29 & 0.06 & & & & + _ back _ & 121 & 0.968 & 0.015 & 1.04 & 0.03 & & & & + _ off _ & 99 & 1.335 & 0.019 & 1.12 & 0.04 & & & & + _ night _ & 98 & 2.025 & 0.057 & 1.29 & 0.04 & & & & + _ other _ & 96 & 1.145 & 0.019 & 1.12 & 0.03 & & & & + _ becky _ & 96 & 2.701 & 0.036 & 1.55 & 0.10 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 440676 & 0.456 & 0.020 & 1.61 & 0.06 & & & & + _ & 265304 & 0.436 & 0.009 & 1.78 & 0.06 & 1.78 & 0.06 & 1.78 & 0.06 + e & 141465 & 0.855 & 0.002 & 1.28 & 0.04 & 1.30 & 0.06 & 1.11 & 0.03 + t & 100183 & 0.904 & 0.001 & 1.54 & 0.07 & 1.37 & 0.06 & 1.05 & 0.03 + a & 93129 & 0.877 & 0.001 & 1.32 & 0.05 & 1.25 & 0.06 & 1.06 & 0.03 + o & 91403 & 0.930 & 0.001 & 1.19 & 0.04 & 1.35 & 0.05 & 1.10 & 0.04 + i & 81407 & 0.914 & 0.001 & 1.44 & 0.07 & 1.33 & 0.06 & 1.21 & 0.05 +
n & 80138 & 0.897 & 0.001 & 1.34 & 0.05 & 1.27 & 0.04 & 1.26 & 0.06 + s & 76915 & 0.950 & 0.001 & 1.40 & 0.06 & 1.31 & 0.06 & 1.23 & 0.06 + h & 72550 & 0.906 & 0.002 & 1.61 & 0.06 & 1.23 & 0.05 & 1.44 & 0.08
+ r & 69852 & 0.918 & 0.001 & 1.49 & 0.06 & 1.22 & 0.04 & 1.30 & 0.06 + l & 55052 & 1.074 & 0.001 & 1.41 & 0.06 & 1.39 & 0.06 & 1.19 & 0.05 + d & 49093 & 0.980 & 0.001 & 1.44 & 0.05 & 1.17 & 0.04 & 1.04 & 0.04 + u & 33272 & 0.982 & 0.001 & 1.25 & 0.04 & 1.36 & 0.06 & 1.16 & 0.04 + m & 31535 & 1.025 & 0.001 & 1.29 & 0.05 & 1.35 & 0.05 & 1.05 & 0.03 + c & 29894 & 1.072 & 0.001 & 1.62 & 0.07 & 1.31 & 0.04 & 1.36 & 0.07 + g & 27791 & 1.031 & 0.001 & 1.36 & 0.06 & 1.24 & 0.05 & 1.19 & 0.04 + f & 26638 & 1.025 & 0.001 & 1.30 & 0.05 & 1.22 & 0.04 & 1.12 & 0.03 + w & 26164 & 1.056 & 0.001 & 1.53 & 0.07 & 1.32 & 0.05 & 1.31 & 0.06 + y & 24251 & 1.032 & 0.001 & 1.36 & 0.05 & 1.15 & 0.03 & 1.01 & 0.03 + p & 22440 & 1.124 & 0.002 & 1.46 & 0.06 & 1.27 & 0.05 & 1.25 & 0.05 + _ the _ & 14952 & 1.071 & 0.002 & 1.44 & 0.06 & & & & + _ of _ & 8141 & 1.121 & 0.003 & 1.63 & 0.07 & & & & + _ and _ & 7217 & 1.167 & 0.003 & 1.53 & 0.05 & & & & + _ a _ & 6518 & 1.144 & 0.003 & 1.24 & 0.04 & & & & + _ to _ & 4963 & 1.157 & 0.003 & 1.38 & 0.07 & & & & + _ in _ & 4946 & 1.002 & 0.002 & 1.16 & 0.04 & & & & + _ were _ & 510 & 1.461 & 0.013 & 1.27 & 0.04 & & & & + _ stephen _ & 505 & 4.955 & 0.099 & 1.64 & 0.06 & & & & + _ we _ & 425 & 2.427 & 0.085 & 1.25 & 0.04 & & & & + _ man _ & 415 & 1.388 & 0.019 & 1.16 & 0.04 & & & & + _ into _ & 330 & 1.179 & 0.011 & 1.14 & 0.04 & & & & + _ eyes _ & 329 & 1.921 & 0.013 & 1.21 & 0.04 & & & & + _ where _ & 310 & 1.214 & 0.014 & 1.11 & 0.03 & & & & + _ hand _ & 308 & 1.295 & 0.017 & 1.18 & 0.04 & & & & + _ street _ & 293 & 1.394 & 0.013 & 1.21 & 0.04 & & & & + _ our _ & 291 & 1.556 & 0.018 & 1.23 & 0.04 & & & & + _ first _ & 278 & 1.306 & 0.011 & 1.19 & 0.04 & & & & + _ father _ & 277 & 1.631 & 0.013 & 1.62 & 0.05 & & & & + _ day _ & 250 & 1.131 & 0.012 & 1.10 & 0.03 & & & & + _ just _ & 249 & 2.014 & 0.012 & 1.20 & 0.04 & & & & + & & & + sequence & @xmath319 & @xmath144 & error & @xmath74 & error & @xmath74 & error & @xmath74 & error + vowels & 945097 & 0.430 & 0.020 & 1.55 & 0.05 & & & & + _ & 565161 & 0.426 & 0.009 & 1.50 & 0.05 & 1.50 & 0.05 & 1.50 & 0.05 + e & 312626 & 0.834 & 0.002 & 1.39 & 0.05 & 1.35 & 0.04 & 1.05 & 0.03 + t & 224180 & 0.886 & 0.001 & 1.40 & 0.04 & 1.27 & 0.04 & 1.09 & 0.03 + a & 204154 & 0.869 & 0.002 & 1.42 & 0.05 & 1.18 & 0.04 & 1.05 & 0.03 + o & 191126 & 0.904 & 0.001 & 1.45 & 0.05 & 1.40 & 0.04 & 1.04 & 0.04 +
n & 182910 & 0.860 & 0.002 & 1.28 & 0.04 & 1.43 & 0.05 & 1.17 & 0.04 + i & 172403 & 0.894 & 0.001 & 1.48 & 0.05 & 1.46 & 0.05 & 1.21 & 0.04 + h & 166290 & 0.852 & 0.002 & 1.50 & 0.05 & 1.26 & 0.04 & 1.28 & 0.05 + s & 161889 & 0.955 & 0.001 & 1.32 & 0.04 & 1.47 & 0.05 & 1.04 & 0.03 + r & 146667 & 0.919 & 0.001 & 1.38 & 0.04 & 1.34 & 0.04 & 1.10 & 0.03 + d & 117632 & 0.923 & 0.001 & 1.48 & 0.05 & 1.33 & 0.04 & 1.05 & 0.03
+ l & 95888 & 1.064 & 0.001 & 1.26 & 0.04 & 1.30 & 0.04 & 1.04 & 0.04 + u & 64788 & 0.971 & 0.001 & 1.26 & 0.04 & 1.26 & 0.04 & 1.04 & 0.03 + m & 61162 & 1.018 & 0.001 & 1.30 & 0.04 & 1.26 & 0.04 & 0.98 & 0.04 + c & 60576 & 1.009 & 0.001 & 1.54 & 0.05 & 1.39 & 0.04 & 1.17 & 0.04 + w & 58852 & 0.978 & 0.001 & 1.28 & 0.04 & 1.29 & 0.04 & 1.24 & 0.05 + f & 54419 & 1.064 & 0.001 & 1.49 & 0.05 & 1.23 & 0.04 & 1.05 & 0.03 + g & 50819 & 1.014 & 0.001 & 1.49 & 0.05 & 1.37 & 0.04 & 1.08 & 0.04 + y & 45847 & 1.035 & 0.001 & 1.36 & 0.04 & 1.34 & 0.04 & 1.01 & 0.04 + p & 44680 & 1.080 & 0.001 & 1.48 & 0.05 & 1.34 & 0.04 & 1.12 & 0.03 + _ the _ & 34495 & 1.128 & 0.002 & 1.59 & 0.05 & & & & + _ and _ & 22217 & 0.874 & 0.002 & 1.22 & 0.04 & & & & + _ to _ & 16640 & 1.056 & 0.001 & 1.24 & 0.04 & & & & + _ of _ & 14864 & 1.168 & 0.002 & 1.59 & 0.05 & & & & + _ a _ & 10525 & 1.119 & 0.002 & 1.21 & 0.04 & & & & + _ he _ & 9860 & 1.893 & 0.006 & 1.44 & 0.06 & & & & + _ so _ & 1900 & 1.180 & 0.005 & 1.14 & 0.03 & & & & + _ prince _ & 1890 & 3.862 & 0.026 & 1.68 & 0.06 & & & & + _ pierre _ & 1796 & 6.563 & 0.042 & 1.78 & 0.06 & & & & + _ an _ & 1625 & 1.131 & 0.004 & 1.17 & 0.04 & & & & + _ could _
& 1115 & 1.285 & 0.005 & 1.15 & 0.05 & & & & + _ natasha _ & 1098 & 6.334 & 0.036 & 1.74 & 0.06 & & & & + _ man _ & 1081 & 1.407 & 0.007 & 1.33 & 0.04 & & & & + _ will _ & 1066 & 1.530 & 0.011 & 1.36 & 0.04 & & & & + _ andrew _ & 1047 & 4.321 & 0.021 & 1.71 & 0.06 & & & & + _ do _ & 1037 & 1.273 & 0.010 & 1.16 & 0.04 & & & & + _ time _ & 926 & 1.100 & 0.008 & 1.13 & 0.03 & & & & + _ princess _ & 915 & 5.431 & 0.033 & 1.72 & 0.06 & & & & + _ face _ & 893 & 1.445 & 0.007 & 1.25 & 0.04 & & & & + _ french _ & 879 & 2.287 & 0.029 & 1.53 & 0.05 & & & & + | the complexity of human interactions with social and natural phenomena is mirrored in the way we describe our experiences through natural language . in order to retain and convey such a high dimensional information ,
the statistical properties of our linguistic output has to be highly correlated in time .
an example are the robust observations , still largely not understood , of correlations on arbitrary long scales in literary texts . in this paper
we explain how long - range correlations flow from highly structured linguistic levels down to the building blocks of a text ( words , letters , etc .. ) . by combining calculations and data analysis
we show that correlations take form of a bursty sequence of events once we approach the semantically relevant topics of the text .
the mechanisms we identify are fairly general and can be equally applied to other hierarchical settings . + published as : link : dx.doi.org/10.1073/pnas.1117723109[proc .
nat .
acad .
sci .
usa ( 2012 ) doi : 10.1073/pnas.1117723109 ] literary texts are an expression of the natural language ability to project complex and high - dimensional phenomena into a one - dimensional , semantically meaningful sequence of symbols . for this projection to be successful , such sequences have to encode the information in form of structured patterns , such as correlations on arbitrarily long scales @xcite .
understanding how language processes long - range correlations , an ubiquitous signature of complexity present in human activities @xcite and in the natural world @xcite , is an important task towards comprehending how natural language works and evolves .
this understanding is also crucial to improve the increasingly important applications of information theory and statistical natural language processing , which are mostly based on short - range - correlations methods @xcite .
take your favorite novel and consider the binary sequence obtained by mapping each vowel into a @xmath0 and all other symbols into a @xmath1 .
one can easily detect structures on neighboring bits , and we certainly expect some repetition patterns on the size of words .
but one should certainly be surprised and intrigued when discovering that there are structures ( or memory ) after several pages or even on arbitrary large scales of this binary sequence . in the last twenty years
, similar observations of long - range correlations in texts have been related to large scales characteristics of the novels such as the story being told , the style of the book , the author , and the language @xcite . however , the mechanisms explaining these connections are still missing ( see ref .
@xcite for a recent proposal ) . without such mechanisms , many fundamental questions can not be answered .
for instance , why all previous investigations observed long - range correlations despite their radically different approaches ?
how and which correlations can flow from the high - level semantic structures down to the crude symbolic sequence in the presence of so many arbitrary influences ?
what information is gained on the large structures by looking at smaller ones ? finally , what is the origin of the long - range correlations ? in this paper we provide answers to these questions by approaching the problem through a novel theoretical framework .
this framework uses the hierarchical organization of natural language to identify a mechanism that links the correlations at different linguistic levels . as schematically depicted in fig .
[ fig.1 ] , a topic is linked to several words that are used to describe it in the novel . at the lower level
, words are connected to the letters they are formed , and so on .
we calculate how correlations are transported through these different levels and compare the results with a detailed statistical analysis in ten different novels .
our results reveal that while approaching semantically relevant high - level structures , correlations unfold in form of a bursty signal . moving down in levels ,
we show that correlations ( but not burstiness ) are preserved , explaining the ubiquitous appearance of long - range correlations in texts . ) , letters ( a - z ) , words , and topics . ] |
the muon spin rotation / relaxation / resonance ( @xmath0sr ) technique is a versatile local probe technique to investigate the physical properties of superconductors , magnetic systems , semiconductors and organic materials @xcite . because of its unique capabilities of a local probe technique it gives complementary information to other methods .
the advantages of the technique include extreme sensitivity to small internal magnetic fields , the ability to detect magnetic fluctuations over a large range ( 10@xmath5/s ) and applications in a wide variety of sample environments .
the muons are usually produced by an intensive , energetic ( 0.5 - 3 gev ) proton beam bombarding a graphite target , where the protons generate pions in nuclear reactions , which subsequently decay into muons and neutrinos .
these muons have kinetic energies of the order of mev and above and penetrate deeply into a sample ( mm to cm ) .
therefore , @xmath0sr experiments using these mev muons can only study bulk materials . to overcome these limitations and to extend @xmath0sr to the investigation of thin films , psi developed and operates the low energy muons beam facility ( lem ) where
a cryogenic moderation method @xcite is used to generate fully polarized positive muons with tunable energies in the range of ev to several kev . up to
now these characteristics have allowed the lem facility at psi to play a leading role in low energy muon experiments , extending the @xmath0sr technique to the investigation of nano - materials , layer interfaces , thin films and near - surface regions @xcite .
the lem facility is located at the @xmath0e4 beam line , which is a hybrid - type large acceptance channel to generate an intense beam of so - called surface muons ( positive muons , @xmath6 , originating from pions decaying at rest close to the surface of the graphite target , with a kinetic energy of @xmath7 4 mev ) @xcite .
the intensity of the surface muon beam at the exit of the @xmath0e4 beam line is about 4.6@xmath810@xmath9/s at a proton current of 2.2 ma .
this represents at the moment the highest continuous surface muon flux in the world .
about 40@xmath10 of the beam can be focussed onto the cryogenic moderator target . using a wide - band - gap van der waals solid gas ( s - n@xmath11 , s - ar ) a moderation efficiency ( @xmath12 ) of 10@xmath13 or 10@xmath14
is achieved @xcite .
the moderation of the muons from @xmath15 mev to @xmath16 ev is achieved within @xmath17 ps , such that their initial high polarization is conserved @xcite .
the moderator consists of a 200 - 300 nm thick ar layer ( capped by a 10 nm thin n@xmath11 layer ) deposited on a thin ag foil ( @xmath7125 @xmath0 m ) @xcite which is held by a cryostat at a temperature below 20 k. the mean energy of the moderated muons is about 15 ev with a width of about 20 ev ( full width at half maximum , fwhm ) @xcite .
these moderated muons can be re - accelerated by applying a high positive potential of up to 20 kv to the moderator @xcite .
after acceleration they are transported by electro- and magneto - static beam elements to the sample position .
the rate of moderated muons at the sample is up to 4@xmath810@xmath18/s . by tuning the high voltage of the moderator and the acceleration / deceleration high voltage at the sample , low - energy positive muons ( le-@xmath6 ) with tunable kinetic energies between 0.5 and 30 kev
are obtained , corresponding to mean implantation depths ranging from a few nm to a few hundred nm in solid materials @xcite . in transverse and longitudinal field
@xmath0sr measurements an external magnetic field is applied to the sample .
the direction of the external field can be parallel or perpendicular to the muon momentum . in lem experiments the sample
is usually biased by application of an electric potential to adjust the final implantation energy .
the corresponding electric field is superimposed onto the magnetic field . since both fields may influence the beam spot size , we investigated the effect using the musrsim simulation package @xcite which is based on geant4 @xcite . using these simulations
we optimize the settings of the last electrostatic steering and focusing element for various magnetic fields - either perpendicular or parallel to the muon momentum - and varying implantation energies ( electric fields ) .
these parameters allow to run the experiment with optimized beam transport onto the sample under different magnetic and electric field configurations .
the results of this analysis are presented in section [ sec2 ] .
the present le-@xmath6 beam size at the sample is about 14 mm ( fwhm ) in horizontal and vertical directions ( @xmath1 and @xmath2 ) , and the corresponding root - mean - square values ( @xmath19 , @xmath20 ) are 6.0 mm @xcite .
since typical thin - film samples are deposited on @xmath21 mm@xmath22 or @xmath23 mm@xmath22 substrates , the current beam size is larger than such samples .
the beam size is significantly affected by the interaction of the muons with the carbon foil of the start ( or trigger ) detector , which provides the implantation time ( time zero ) of the muons and start the clock of the @xmath0sr data acquisition . the foil causes energy loss and broadening of the energy distribution of the muons , and increases the angular divergence of the beam ( transverse phase space ) @xcite . in order to understand and quantify the effect of the start detector on the beam size
, we simulated the muon beam with and without the start detector .
the corresponding @xmath19 , @xmath20 values are calculated and then compared to experimental measurements on the beam spot . an additional route for optimizing the beam spot size using collimators
was also simulated and investigated .
the results of this study are presented and discussed in section [ sec3 ] .
sr experiments at psi .
the red and blue arrows in the figure represent the muon s momentum and direction of spin polarization , respectively .
@xmath24 represents the spin angle with respect to the initial direction of the 4-mev muon beam , corresponding to @xmath25 direction in the simulation.,width=321 ]
a schematic of the lem apparatus is shown in figure [ fig1 ] .
the source of the low - energy muon beam , the cryogenic moderator , has a quadratic area of @xmath26 mm@xmath22 .
adjustable positive high voltages applied to the moderator and to a set of grids generate an accelerating electric field to reaccelerate the moderated muons from ev energy to higher energies , typically between 10 - 20 kev ( low energy muons ) . since the moderation efficiency is of the order of 10@xmath13 to 10@xmath14 most of the surface muons stop in the moderator target or leave the moderator with kev to hundreds of kev energies ( fast muons ) . an electrostatic mirror with an angle of 45@xmath27 with respect to the muon momentum
is used to separate low energy muons from fast muons . after being focussed by einzel lens 1 ( l1 )
the low energy muons are deflected by 90@xmath27 with respect to the initial muon direction , while fast muons continue to move towards the mcp detector .
after the deflection the spin polarization is perpendicular to the muon momentum . a spin rotator with crossed static magnetic and electric fields @xmath28 can be used to rotate the muon spin parallel or anti - parallel to its momentum @xcite .
this enables to carry out longitudinal field measurements ( lf-@xmath0sr , where the muon spin is parallel or anti - parallel to the applied field at the sample ) .
the le-@xmath6 spin direction ( @xmath24 ) can be changed to any angle from -90@xmath27 to + 90@xmath27 with respect to the initial muon spin direction by tuning the magnetic and electric fields in the spin rotator , such that the ratio e / b is chosen to match the velocity of the muons .
= 15 kv , @xmath24=-10@xmath27 , l1=9.0 kv , l2=10.5 kv , l3=11.5 kv and ra=11.9 kv . the initial number of muons at the moderator ( @xmath29)is @xmath30.,width=264 ] = 15 kv.,title="fig:",width=7 ] + = 15 kv.,title="fig:",width=7 ] downstream of the spin rotator the einzel lens 2 ( l2 ) focuses the muon beam onto the retractable start detector .
this detector provides the fast start timing signal which is necessary for @xmath0sr experiments at a continuous muon beam .
it consists of a 10-nm - thin carbon foil transverse to the muon momentum , a microchannel plate ( mcp ) transverse to the carbon foil and a grid system used to direct secondary electrons , ejected by muons passing through the foil , to a mcp detector @xcite .
the muons lose 0.5 - 1.0 kev energy in the foil , depending on energy , with an energy width of same order ( fwhm ) .
multiple scattering in the foil leads to an increase of transverse phase space , which , in combination with the larger width of the energy distribution , causes an increase of the attainable beam spot size at the sample position .
additionally , neutralization processes in the foil can lead to the formation of neutral muonium atoms ( a hydrogen - like bound state consisting of a positive muon and an electron ) .
the final beam size at the sample position is determined by einzel lens 3 ( l3 ) and the strongly focusing conical lens ( also called ring anode , ra ) . in the lem beam geant4 simulation ,
the initial muon beam is started at the moderator , at the edge of the high - voltage acceleration region .
the moderator is located at a distance of 499 mm upstream to the center of the electrostatic mirror .
we assume homogeneous electric fields in the acceleration volume .
the high voltage @xmath31 applied at the moderator varies between 8.5 kv and 15 kv .
the initial beam is homogeneously distributed on an area of 30@xmath830 mm@xmath22 which is the active area of the moderator , and the initial mean kinetic energy is 15 ev with a fwhm of 20 ev . due to the homogeneous angular distribution of slow muons inside the moderator @xcite a @xmath32 angular divergence
is generated in the simulation for muons escaping from the moderator layer ( @xmath33 is the angle of the muon momentum with respect to the initial beam momentum ) .
the momentum distribution is generated in the simulation as follows , @xmath34 where @xmath35 is the muon momentum , and @xmath36 and @xmath37 are random numbers between 0 and 1 .
field maps of the beam elements ( spin rotator , l1 , l2 , l3 and ra ) have been either calculated or measured @xcite , and are scaled according to the experimental settings . in the electrostatic mirror a homogeneous electric field is assumed . a vertical magnetic field up to 300 gauss , parallel to the sample surface ,
is supplied by the aew magnet for transverse field @xmath0sr measurements ( tf-@xmath0sr , muon spin transverse to the applied field ) , and a magnetic field of up to 3400 gauss can be applied perpendicular to the sample surface and along the muon beam direction ( wew magnet , lf- or tf-@xmath0sr , depending on the initial muons spin polarization ) . an acceleration / deceleration high voltage of up to @xmath3812.5 kv can be applied at the sample plate in order to tune the implantation energy of le-@xmath6 between 0.5 and 30 kev @xcite .
+ 20 mm@xmath22 at the sample for different wew magnetic fields for a moderator potential of 12 kv .
ra@xmath39 and ra@xmath40 are the best values in the experiments and simulations , respectively.,width=264 ] and @xmath20 of the muon beam spot at the sample at different magnetic fields along the muon momentum before and after tuning the ra with v@xmath41=15 kv .
data labelled `` before '' and `` after '' are for ra=11.9 kv for all magnetic fields before optimizing ra in the simulation and after setting the ra according to the values of table [ table8 ] , respectively.,width=264 ] = @xmath42mm}}{n^{out}_{total}}$ ] at the sample position with different magnetic fields parallel to the muon momentum before and after tuning the ra with v@xmath41=15kv .
data labelled `` before '' and `` after '' are for ra=11.9 kv for all magnetic fields ( before optimizing ra in the simulation ) and after tuning the ra according to table [ table8 ] respectively . , width=264 ] in this section we investigate the effects of a vertical magnetic field ( field transverse to the beam direction , aew magnet ) on the beam spot for different fields and implantation energies .
in addition , we simulate the steering and focusing effects of the segmented conical lens ( ra ) in front of the sample position .
this allows us to optimize the ra settings for the different magnetic field configurations , which can be used as reference values for the experiment .
figure [ fig2 ] shows the simulated muon beam spot at the sample position with start detector in - beam for a muon extraction voltage of v@xmath41 = 15 kv and zero applied magnetic and electric fields on the sample .
the beam spot is well centered in this case .
figure [ fig3 ] shows beam spots viewed from downstream for different aew magnetic fields : 50 , 100 , 150 , 200 , 250 and 300 gauss .
the beam spot shifts almost linearly to the left with increasing magnetic field while the shape of the beam spot is only marginally affected .
the mean horizontal and vertical positions ( meanx and meany ) for different electric and vertical magnetic fields are shown in figure [ fig4 ] .
the absolute value of meanx increases nearly linearly with increasing magnetic field , and the beam spot shift is larger in a decelerating electric field .
the meany remains almost constant for the magnetic and electric fields usable in the experiment .
the conical lens ra at the end of the lem beam line consists of four segments : rat , rab , ral and rar ( figure [ fig5 ] ) .
when the electrostatic potentials of the four sections are not equal , the beam spot at the sample is shifted from the center .
we compare the experimental and simulated beam spots at the sample for different ra parameters . in the experiment ,
the beam spot at the sample position is measured using a 40 mm diameter roentdek dld40 delay line detector with three resistance matched hamamatsu mcps in a z configuration .
beam spots for muons accelerated with 10 kv at the moderator are shown in figure [ fig6 ] .
the top two are for ral - rar = -0.64 kv ( ral = 7.18 kv , rar = 7.82 kv , rat = rab = 7.50 kv ) , where the beam spot center shifts to the left side with meanx= -9.5 mm ( in experiment ) and -8.4 mm ( in simulation ) .
the results for ral - rar = 0.56 kv ( ral = 7.78 kv , rar = 7.22 kv , rat = rab = 7.50 kv ) are shown in the bottom two figures of figure [ fig6 ] . in this case
the beam spot shifts to the right side with meanx= 8.8 mm ( in experiment ) and 7.6 mm ( in simulation ) .
the agreement between experiment and simulation is fairly good .
the small deviations can be partially explained by the uncertainty in the applied potentials in the experiment , where it is known that the used high voltage power supplies have an absolute uncertainty between 10 and 30 v. the beam spot is deformed to a trapezoidal shape , which is different to the effect of the magnetic field which does nt cause a deformation .
this deformation is caused by lens abberations in the ra due to proximity of the beam to the electrodes and the broken four fold symmetry of the electric field in case of a potential difference applied between two opposite segments . in the simulation ,
the transmission from the moderator to the sample at 10 kv is about 55% , compared to about 70% at 15 kv . here
, the transmission is defined as @xmath43 @xmath8100% , where @xmath44 is the number of muons in the sample plane , and @xmath29 is the initial number of muons at the moderator .
the reduction of the transmission is due to i ) a longer time - of - flight at 10 kv , which increases the fraction of muons decaying in flight by about 5% at 10 kv compared to 15 kv , ii ) higher muonium ( mu ) formation probability in the carbon foil , by 10% , and iii ) larger beam divergence at 10 kv after the start detector . in the experiment
we observe a corresponding drop of event rate when changing the beam transport settings from 15 kv to 10 kv .
.simulated optimum values of ( ral - rar ) ( unit : kv ) to steer the beam spot to the center at different electric potentials at the sample between -10 kv and 10 kv , and different aew magnetic fields from 0 to 300 gauss , v@xmath41=15 kv .
ral+rar=23.8 kv . [ cols="^,^,^,^,^,^",options="header " , ] [ table1 ] in the simulation we place a collimator with radius @xmath45 mm in front of the start detector to study the effect of beam collimation on the muon beam spot size . table [ table1 ] summarizes the rms values of the beam spot at the sample with and without collimation with the corresponding beam transmission @xmath46 .
after collimation the beam area shrinks to about 30% of its initial value , while at the same time the transmission efficiency significantly drops to less than 10% for the present moderator size of 30@xmath830 mm@xmath22 . for a smaller moderator of 10@xmath810 mm@xmath22 area the beam spot at the sample without collimation at the start detector can be already clearly reduced , with about 2 mm smaller rms values .
additional collimation in front of the start detector can further lower the beam spot size , which is now slightly smaller compared to the collimated beam for the 30@xmath830 mm@xmath22 moderator .
the disadvantage of using collimators is the large reduction of beam transmission of up to about 90% , causing a corresponding reduction of low - energy muon rate . a better focusing of the surface muon beam onto the moderator target
can - at least partially - counteract this loss , as is demonstrated by the simulation result with a 10@xmath810 mm@xmath22 moderator area . to reduce the present lem beam spot size at the sample position two options have been investigated : the removal of the start detector , which is possible when using time integral measurements such as , for example , continuous wave radio frequency @xmath0sr ( rf-@xmath0sr ) , and the use of collimators .
the muon beam spot size at the sample position can be significantly reduced to 3.5 - 4.0 mm rms values without the detector and for the present size of muon moderator of 30@xmath830 mm@xmath22 . to overcome the problem of generating a timing signal for le-@xmath0sr measurements , it is conceivable to use a 2@xmath47 moderator to re - moderate the low energy muons , as it is routinely done in slow positron beam setups @xcite .
secondary electrons , created by the muons impinging on the moderator , could be directed onto a mcp detector to generate the required start signal for the @xmath0sr measurement instead of the start detector currently used .
if successful , such a 2@xmath47 moderator could potentially be useful to further reduce the muon beam spot size to rms values @xmath48 mm .
another possibility to significantly reduce the beam spot size at the sample to @xmath49 3.5 mm rms is the use of a smaller moderator target with a size of 10@xmath810 mm@xmath22 .
additional collimation at the start detector can further lower the rms values to @xmath50 mm . in order to partially compensate for the corresponding loss in beam transmission we are considering an improved beam focus of the @xmath0e4 beamline by using a solenoid as the last focusing element , similar to the existing wsx61/wsx62 solenoids at the beginning of @xmath0e4 to replace the last quadrupole triplet qsm610 - 612 @xcite at the end of the beam line .
the possibility of improving the extraction system after the lem moderator by using a soa lens will be also considered in the future .
such a soa lens has been successfully used at the j - parc / muse facility @xcite and consists of several cylindrical electrodes with asymmetric high voltages .
the advantage of this lens is that it produces a smaller beam cross section compared to the current system of planar grids .
this can reduce the effect of lens abberations on the muon beam spot at the sample position .
in this paper we presented a detailed geant4 investigation to optimize the experimental conditions for the present lem setup , where we studied the beam transport in magnetic and electric fields in the sample region . using these simulations we optimized the electric potential settings of the last segmented electrostatic element ( ra ) in the case of a vertical magnetic field at the sample position .
this field is transverse to the muon momentum and requires steering by ra to center the beam . in the case of a magnetic field along the beam direction the increasing focusing power of the magnet has to be compensated by a reduction of the ra focusing power .
these settings are tabulated in this paper and can be used in future le-@xmath0sr experiments as reference . in order to extend the le-@xmath0sr technique to smaller sample sizes
( @xmath49 10@xmath810 mm@xmath22 ) we studied the influence of the start detector and of collimators on the beam spot size .
this investigation demonstrates the feasibility of achieving a beam spot with @xmath19 , @xmath51 mm .
this requires a smaller muon moderator size ( @xmath49 30@xmath830 mm@xmath22 ) which can be realized by a better focusing of the @xmath0e4 beam onto the moderator by using a solenoid as the last element of the @xmath0e4 beam line
. additionally , an optimized extraction at the moderator and a collimator at the start detector are necessary to achieve this goal , or the replacement of the existing start detector by a 2@xmath47 moderator .
finally , we emphasize that geant4 simulations are very powerful to describe and optimize experimental setups , and to help pushing experimental capabilities to the limit .
the authors thank paolo crivelli for stimulating discussion about muon re - moderation .
ran xiao acknowledges a scholarship from the china scholarship council ( csc ) and financial support from psi for her stay at psi .
10 url # 1`#1`urlprefixhref # 1#2#2 # 1#1 e. morenzoni , physics and applications of low energy muons , muon science : muons in physics , chemistry and materials ( bristol and philadelphia , 1999 ) , sl lee , sh kilcoyne , and r. cywinski , eds 51 ( 1998 ) 343404 .
e. morenzoni , h. glckler , t. prokscha , h. weber , e. forgan , t. jackson , h. luetkens , c. niedermayer , m. pleines , m. birke , et al . , low - energy @xmath0sr at psi : present and future , physica b : condensed matter 289 ( 2000 ) 653657 .
d. harshman , a. mills jr , j. beveridge , k. kendall , g. morris , m. senba , j. warren , a. rupaal , j. turner , generation of slow positive muons from solid rare - gas moderators , physical review b 36 ( 16 ) ( 1987 ) 8850 .
t. prokscha , e. morenzoni , k. deiters , f. foroughi , d. george , r. kobler , a. suter , v. vrankovic , the new @xmath0e4 beam at psi : a hybrid - type large acceptance channel for the generation of a high intensity surface - muon beam , nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment 595 ( 2 ) ( 2008 ) 317331 .
e. morenzoni , r. khasanov , h. luetkens , t. prokscha , a. suter , n. garifianov , h. glckler , m. birke , e. forgan , h. keller , et al .
, low energy muons as probes of thin films and near surface regions , physica b : condensed matter 326 ( 1 ) ( 2003 ) 196204 .
t. prokscha , k. chow , e. stilp , a. suter , h. luetkens , e. morenzoni , g. nieuwenhuys , z. salman , r. scheuermann , photo - induced persistent inversion of germanium in a 200-nm - deep surface region , scientific reports 3 .
e. morenzoni , t. prokscha , h. saadaoui , z. salman , a. suter , b. m. wojek , j. baglo , i. boovi , m. hossain , r. f. kiefl , et al .
, low - energy muons at psi : examples of investigations of superconducting properties in near - surface regions and heterostuctures , in : proceedings of the international symposium on science explored by ultra slow muon ( usm2013 ) , jps conference proceedings , vol .
2 , i d . 010201 , 10 pp .
, vol . 2 , 2014 , p. 0201 .
l. schulz , l. nuccio , m. willis , p. desai , p. shakya , t. kreouzis , v. k. malik , c. bernhard , f. pratt , n. morley , et al . ,
engineering spin propagation across a hybrid organic / inorganic interface using a polar layer , nature materials 10 ( 1 ) ( 2011 ) 3944 .
a. suter , e. morenzoni , t. prokscha , b. wojek , h. luetkens , g. nieuwenhuys , a. gozar , g. logvenov , i. boovi , two - dimensional magnetic and superconducting phases in metal - insulator la 2- x sr x cuo 4 superlattices measured by muon - spin rotation , physical review letters 106 ( 23 ) ( 2011 ) 237003 .
a. boris , y. matiks , e. benckiser , a. frano , p. popovich , v. hinkov , p. wochner , m. castro - colin , e. detemple , v. k. malik , et al .
, dimensionality control of electronic phase transitions in nickel - oxide superlattices , science 332 ( 6032 ) ( 2011 ) 937940 .
a. hofmann , z. salman , m. mannini , a. amato , l. malavolti , e. morenzoni , t. prokscha , r. sessoli , a. suter , depth - dependent spin dynamics in thin films of tbpc2 nanomagnets explored by low - energy implanted muons , acs nano 6 ( 9 ) ( 2012 ) 83908396 .
e. stilp , a. suter , t. prokscha , z. salman , e. morenzoni , h. keller , p. pahlke , r. hhne , c. bernhard , r. liang , et al .
, controlling the near - surface superfluid density in underdoped yba2cu3o6 + x by photo - illumination , scientific reports 4 .
h. saadaoui , z. salman , h. luetkens , t. prokscha , a. suter , w. macfarlane , y. jiang , k. jin , r. greene , e. morenzoni , et al . , the phase diagram of electron - doped
la2- xcexcuo4- @xmath52 , nature communications 6 .
f. al mamari , t. moorsom , g. teobaldi , w. deacon , t. prokscha , h. luetkens , s. lee , g. e. sterbinsky , d. a. arena , d. a. maclaren , et al . ,
beating the stoner criterion using molecular interfaces , nature 524 ( 7563 ) ( 2015 ) 6973 .
l. anghinolfi , h. luetkens , j. perron , m. flokstra , o. sendetskyi , a. suter , t. prokscha , p. derlet , s. lee , l. heyderman , thermodynamic phase transitions in a frustrated magnetic metamaterial , nature communications 6 .
m. flokstra , n. satchell , j. kim , g. burnell , p. curran , s. bending , j. cooper , c. kinane , s. langridge , a. isidori , et al .
, remotely induced magnetism in a normal metal using a superconducting spin - valve , arxiv preprint arxiv:1505.03565 .
e. morenzoni , h. glckler , t. prokscha , r. khasanov , h. luetkens , m. birke , e. forgan , c. niedermayer , m. pleines , implantation studies of kev positive muons in thin metallic layers , nuclear instruments and methods in physics research section b : beam interactions with materials and atoms 192 ( 3 ) ( 2002 ) 254266 .
s. agostinelli , j. allison , k. a. amako , j. apostolakis , h. araujo , p. arce , m. asai , d. axen , s. banerjee , g. barrand , et al .
, geant4a simulation toolkit , nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment 506 ( 3 ) ( 2003 ) 250303 .
j. allison , k. amako , j. apostolakis , h. araujo , p. a. dubois , m. asai , g. barrand , r. capra , s. chauvie , r. chytracek , et al .
, geant4 developments and applications , nuclear science , ieee transactions on 53 ( 1 ) ( 2006 ) 270278 .
z. salman , t. prokscha , p. keller , e. morenzoni , h. saadaoui , k. sedlak , t. shiroka , s. sidorov , a. suter , v. vrankovic , et al . ,
design and simulation of a spin rotator for longitudinal field measurements in the low energy muons spectrometer , physics procedia 30 ( 2012 ) 5560 .
k. s. khaw , a. antognini , p. crivelli , k. kirch , e. morenzoni , z. salman , a. suter , t. prokscha , geant4 simulation of the psi lem beam line : energy loss and muonium formation in thin foils and the impact of unmoderated muons on the @xmath0 sr spectrometer , journal of instrumentation 10 ( 10 ) ( 2015 ) p10025 . c. piochacz , g. kgel , w. egger , c. hugenschmidt , j. mayer , k. schreckenbach , p. sperr , m. stadlbauer , g. dollinger , a positron remoderator for the high intensity positron source nepomuc , applied surface science 255 ( 1 ) ( 2008 ) 98100 .
y. miyake , y. ikedo , k. shimomura , p. strasser , n. kawamura , k. nishiyama , a. koda , h. fujimori , s. makimura , j. nakamura , et al .
, ultra slow muon microscopy by laser resonant ionization at j - parc , muse , hyperfine interactions 216 ( 1 - 3 ) ( 2013 ) 7983 .
t. nagatomo , y. ikedo , p. strasser , s. makimura , j. nakamura , k. nishiyama , k. shimomura , n. kawamura , a. koda , h. fujimori , et al . ,
construction of ultra slow muon beam line at j - parc , in : proceedings of the international symposium on science explored by ultra slow muon ( usm2013 ) , jps conference proceedings , vol .
2 , i d . 010102 , 5 pp .
, vol . 2 , 2014 , p. 0102 .
j. nakamura , t. nagatomo , y. oishi , y. ikedo , p. strasser , n. saito , k. miyazaki , k. yokoyama , k. okamura , y. miyake , et al .
, ultra slow muon microscope at muse / j - parc , in : journal of physics : conference series , vol .
502 , iop publishing , 2014 , p. 012042 . | the low - energy muon ( lem ) facility at psi provides slow fully polarized positive muons with tunable energies in the kev range to carry out muon spin rotation and relaxation ( @xmath0sr ) experiments with nanometer depth resolution on thin films , heterostructures and near - surface regions .
the beam is focussed and transported to the sample by electrostatic lenses .
a spin rotator upstream of the sample can be used to rotate the muon spin .
an electric field along the muon momentum can also be applied to accelerate / decelerate the muons to different energies ( 0.5 - 30 kev ) . moreover , external magnetic fields at the sample are necessary for transverse and longitudinal field @xmath0sr experiments .
all these fields affect the beam spot at the sample position , so that presently the lem beam has a beam spot at the sample position with root - mean - square ( rms ) values in @xmath1 and @xmath2 of about 6 mm .
this practically limits the application of lem to sample sizes of @xmath3 1 @xmath4 . here , we present beam transport simulation results using geant4 , to optimize the beam transport onto the sample for various magnetic and electric fields applied at the sample , and to investigate several options to generate a smaller beam spot .
we study in detail the effect of the muon start detector on the beam spot .
we find that without the detector the rms values of the beam spot can be reduced to 3.5 - 4.0 mm .
we also demonstrate that further reduction of the beam spot to 2.5 mm rms can be achieved by collimation before the start detector and at the source of the low - energy muon beam . muon beam , muon spin rotation , low energy , beam size , geant4 |
random graphs often appear and have been studied in various fields of natural science and engineering .
a recent interest in their applications is generating expander graphs @xcite , regular graphs that shows high connectivity and homogeneity .
such graphs have important applications in designing networks of computers @xcite , infrastructures @xcite , and real and artificial neurons @xcite .
a way to define and generate expanders is maximization of the spectral gap @xmath1 , which is defined as the difference between the largest eigenvalue and the second largest eigenvalue of the adjacency matrix of a graph .
et al . _
@xcite numerically maximized the spectral gap by simulated annealing and generated examples of these graphs .
however , in designing networks , we are also interested in quantitative properties ; specifically how the probability of large spectral gap graphs behaves when the size @xmath2 of graphs increases . in this paper
we apply a method based on multicanonical monte carlo @xcite to the calculation of large deviations in the spectral gap of random graphs .
the method can be regarded as an extension of the method introduced in @xcite ; in @xcite , large deviations in the largest eigenvalue of random matrices are computed by a similar method .
multicanonical monte carlo enables us to estimate tails of the distribution whose probability is very small and can not be computed by naive random sampling . by using the proposed method
, we estimate the distribution of the spectral gap of random 3-regular graphs and quantify the probability @xmath3 that the spectral gap is larger than a given @xmath4 ; when @xmath5 , it is shown that @xmath6 for large @xmath2 .
an undirected graph is described by the corresponding adjacency matrix @xmath7 , whose entries are defined by @xmath8 here we denote eigenvalues of the adjacency matrix by @xmath9 . in the case of @xmath10-regular graphs
, @xmath7 has the trivial largest eigenvalue @xmath11 , therefore we assume @xmath12 the difference @xmath13 between the largest eigenvalue and the largest non - trivial eigenvalue of @xmath7 is called `` spectral gap '' and takes a non - zero value if the corresponding graph is connected .
we are interested in tails of the distribution of the spectral gap . in the case of regular graphs , alon and boppana ( see @xcite ) proved an asymptotic lower bound of the largest non - trivial eigenvalue @xmath14 as @xmath15 assuming that the limit exists , this gives an asymptotic upper bound for the spectral gap as @xmath16 .
graphs with @xmath17 are called `` ramanujan graphs '' @xcite . on the other hand ,
friedman @xcite proved that for @xmath18 and for any constant @xmath19 `` most '' random @xmath10-regular graphs have @xmath14 that satisfies @xmath20 as @xmath21 .
these results indicate that the peak of the distribution of @xmath22 is located near @xmath23 for large @xmath2 .
et al . _
@xcite studied the distribution around this peak using naive random sampling .
none of these studies , however , discusses extreme tails of the distribution of @xmath24 or @xmath1 , which is the main subject of this paper .
here , we will give a brief explanation of multicanonical monte carlo .
the aim of this method is to estimate the density of states @xmath25 defined by @xmath26 where @xmath27 is the dirac @xmath27-function and @xmath28 denotes a multiple integral in the space of matrix @xmath29 .
we define the weight by @xmath30 as a function of @xmath1 , a key quantity of this method .
when a weight function @xmath30 is given , we can generate samples from the distribution defined by the weight using the metropolis algorithm @xcite .
the essential idea is to tune the weight function @xmath30 so as to produce a flat histogram of @xmath1 .
after we find a appropriate weight function @xmath31 that gives a flat histogram , an approximate value of the density of states is estimated by @xmath32 . to obtain this @xmath31 , we modify the weight function step - by - step through the metropolis simulation using the current guess of the weight function @xmath30 .
there are several ways to modify the function @xmath30 . among them , a method proposed by wang and landau @xcite is most successful and used in this study . for a more accurate estimate of the density of states ,
we calculate a histogram @xmath33 by performing a long simulation with fixed @xmath31 .
then @xmath25 is estimated by @xmath34 and the probability distribution @xmath35 of @xmath1 is obtained by @xmath36 .
practically , we calculate the density @xmath35 only in a prescribed interval @xmath37 .
details of the implementation of the metropolis algorithm are as follows .
the simulation starts from an arbitrary @xmath10-regular graph with the desired number of vertices @xmath2 . in each step , a candidate @xmath38 is generated by rewiring edges in the way used in @xcite , where the degree @xmath10 of each vertex is not changed ; a pair of links @xmath39 and @xmath40 that satisfy @xmath41 is selected and rewired as @xmath42 .
then the spectral gap @xmath1 of a candidate is calculated by the householder method and accept / reject decision of the transition from the current state @xmath43 to @xmath38 is made by comparing the metropolis ratio @xmath44 with a random number uniformly distributed in @xmath45 $ ] . a candidate with @xmath46 , indicating a disconnected graph , is always rejected , hence an ensemble of connected random @xmath10-regular graphs is sampled .
of the spectral gap of random 3-regular graphs .
the density at @xmath1 plotted in the vertical axis is obtained from the probability in a small bin around @xmath1 of a fixed width ; the values of them depend on the bin width in the right tail of the distribution , because discreteness of the spectra becomes relevant there .
the arrows indicate @xmath47 ; graphs whose @xmath1 is above this value are ramanujan . ] using the proposed method , we estimate the distribution of the spectral gap of random 3-regular graphs with the number of vertices @xmath48 .
figure [ fig:1 ] shows graphs with the largest spectral gap for @xmath49 and @xmath50 found in the simulations . in figure [ fig:2 ] , we show @xmath35 of random 3-regular graphs with the number of vertices @xmath51 ; the computational time is 7 hours for @xmath52 and 251 hours for @xmath53 using a core of intel xeon x5365 . as @xmath2 increases
, the probability density becomes sharper and the peak becomes closer to around @xmath54 , which is consistent with a theoretical estimate @xcite and a numerical experiment @xcite .
specifically , the probability of graphs with large @xmath1 decreases drastically for large @xmath2 .
s are shown as functions of @xmath2 .
each curve corresponds to different values of @xmath4 .
data are well fitted by quadratic functions when @xmath5 . ] to quantify decreasing rate of the tails of the distribution , we define the probability @xmath55 that @xmath1 is larger than @xmath4 by @xmath56 here , we assume the probability @xmath57 is negligibly smaller than @xmath58 . in figure
[ fig:3 ] , estimated @xmath55s are shown as functions of @xmath2 .
for each @xmath4 , @xmath55 is well fitted by a quadratic functions of @xmath2 , when @xmath5 .
this result indicates that @xmath55 decreases for large @xmath2 as @xmath59 where the rate function @xmath60 is shown in figure [ fig:4 ] . in the region
@xmath61 , @xmath62 is no longer a monotonic decreasing function of @xmath2 and asymptotically approaches to unity for large @xmath2 .
is plotted .
the inset shows semi
log plot of @xmath60 . in the region
@xmath63 , @xmath60 exponentially increases as @xmath4 increases .
@xmath64 is the point that @xmath60 becomes negative . ]
a method based on multicanonical monte carlo is introduced to the estimation of large deviations in the spectral gap of random graphs . by using this method
, we calculate the distribution of the spectral gap @xmath1 and the probability @xmath62 for random 3-regular graphs .
while naive random sampling provides reasonable estimates of @xmath62 only when @xmath4 is around the peak of the distribution @xmath35 , the proposed method enables us to estimate @xmath62 in a wide region of @xmath4 including extreme tails of the distribution .
we find that @xmath62 behaves as @xmath65 for large @xmath2 , when @xmath5 .
our preliminary results indicate that a similar behavior is also seen in the case of random @xmath66- and @xmath67-regular graphs , suggesting that it is a general feature of random @xmath10-regular graphs .
the proposed method can be applied to calculations of large deviations in any statics of any ensemble of random graphs . in the case of non - regular graphs ,
the spectral gap are defined as the smallest non - trivial eigenvalue of the laplacian matrix @xmath68 @xcite ; hence we sample matrices @xmath68 instead of @xmath29 by using multicanonical monte carlo .
recent studies on gaussian or wishart random matrices @xcite showed that the probability of all eigenvalues being negative decreases as @xmath69 when the size @xmath2 of the matrices is large .
our results are regarded as an extension of these results to the spectral gap of random graphs .
we thank prof . m. kikuchi for his support and encouragement . this work is supported in part by global coe program ( core research and engineering of advanced materials - interdisciplinary education center for materials science ) ,
mext , japan .
all simulations were performed on a pc cluster at cybermedia center , osaka university . | graphs with large spectral gap are important in various fields such as biology , sociology and computer science . in designing such graphs ,
an important question is how the probability of graphs with large spectral gap behaves .
a method based on multicanonical monte carlo is introduced to quantify the behavior of this probability , which enables us to calculate extreme tails of the distribution .
the proposed method is successfully applied to random @xmath0-regular graphs and large deviation probability is estimated .
random graph ; spectral gap ; ramanujan graph ; multicanonical monte carlo ; large deviation |
lepton flavor violating ( lfv ) interactions deserve to analyze since , theoretically , they are sensitive the physics beyond the standard model ( sm ) and they are rich in the sense that they exist at loop level and make it possible to predict the free parameters of the underlying theory .
the radiative lfv decays are among the well known candidates of the lfv processes and there are various experimental and theoretical works in the literature .
the current limits for the branching ratios ( @xmath2 s ) of the @xmath0 and @xmath1 decays are @xmath3 @xcite and @xmath4 @xcite , respectively .
the extensive theoretical work has been done on these decays in the supersymmetric models @xcite , in the framework of the two higgs doublet model ( 2hdm ) @xcite and also in a model independent way @xcite . in the general 2hdm , so called the model iii , the possibility of the flavor changing neutral currents ( fcnc s ) at tree level allows the existence of the radiative lfv decays , theoretically .
these decays are induced by the internal neutral higgs bosons @xmath5 and @xmath6 and the yukawa couplings carrying information about the strength of the lepton - lepton - neutral higgs interactions are free parameters of the theory .
these free parameters should be restricted by using the experimental measurements .
this work is devoted to the lfv processes @xmath0 and @xmath1 in the framework of the model iii , with the inclusion of one ( two ) non - universal extra spatial dimension .
the extra dimension scenario is based on the string theories as a possible solution to the hierarchy problem of the standard model ( sm ) .
the effects of extra dimensions on various phenomena have been studied in the literature @xcite-@xcite . for the extra dimension scenarios
the starting point is a fundamental theory , lying in higher dimensions and the assumption that the four dimensional sm is its low energy effective theory .
the procedure to pass from higher dimensions to the four dimension is the compactificitation of each extra dimension to a circle @xmath7 with radius @xmath8 , which is a typical size of corresponding extra dimension .
finally , it leads to appear new particles , namely kaluza - klein ( kk ) modes of the particles in the theory .
if all the fields feel the extra dimensions , so called universal extra dimensions ( ued ) , the extra dimensional momentum is conserved , in other words , the kk number at each vertex is not violated . in the case of a single ued , the compactification size
@xmath8 has been predicted as large as few hundereds of gev @xcite , in the range @xmath9 , using electroweak precision measurements @xcite , the @xmath10 -mixing @xcite and the flavor changing process @xmath11 @xcite .
if the extra dimensions are accessible to some fields but not all in the theory , they are called non - universal extra dimensions . in this case
, the kk number at each vertex is not conserved and tree level interaction of kk modes with the ordinary particles can exist . in @xcite , it was observed that , a very light left handed neutrino can be obtained in the non - universal extra dimensions where only the right handed neutrino feels the extra dimension . the study in @xcite is due to the effect of brane kinetic terms for bulk scalars , fermions and gauge fields in higher dimensional theories .
there are also some phenomenological studies on the non - universal extra dimensions in the literature @xcite . in the present work
, we study the branching ratios ( br s ) of the lfv processes @xmath0 and @xmath12 in the model iii and we assume that the extra dimensions are felt by the new higgs doublet and the gauge sector of the theory .
the compactification of the single ( double ) extra dimension to the circle ( torus ) @xmath7 ( @xmath13 ) brings new particles , kk modes of neutral higgs bosons @xmath5 and @xmath6 , and the additional vertices , inducing kk mode of neutral higgs boson - lepton - lepton interaction , appear . in this case , the kk number in the vertices is not conserved , in contrast to the universal extra dimension ( ued ) case , where all fields experience the extra dimensions .
the non - zero kk modes of neutral higgs fields @xmath14 have masses @xmath15 ( @xmath16 ) with @xmath17 ( @xmath17 , @xmath18 ) . here
@xmath19 ( @xmath20 ) is the mass ( mass@xmath21)of @xmath22th ( @xmath23th ) level kk particle for a single ( double ) non - universal extra dimension .
since the leptons have not kk modes and the additional effects due to the extra dimensions are the functions of the ratio @xmath24 , there is a suppression in the numerical values of the expressions compared with the ones existing in the case of ued .
however , we observe that , the br s of the processes under consideration enhances considerably large for two non - universal extra dimensions , due to the crowd of neutral higgs kk modes .
the paper is organized as follows : in section 2 , we present the br s of lfv interactions @xmath0 and @xmath1 in the model iii version of the 2hdm with the inclusion of non - universal extra dimensions .
section 3 is devoted to discussion and our conclusions .
the radiative lfv decays exist at least in the one loop level and the theoretical values of their br s enhance with the extension of the higgs sector in the models beyond the sm .
the model iii version of the 2hdm accepts the flavor changing neutral currents ( fcnc ) at tree level and it is the candidate model to bring enhancement to the br s of the radiative lfv decays . with the inclusion of possible non - universal extra dimensions which are felt by gauge bosons and the new higgs particles , there appears an additional contribution to the br of the decays under consideration .
the part of lagrangian which carries the interaction responsible for the radiative lfv proceses in 5 ( 6 ) dimension is @xmath25 here @xmath26 , with lepton family indices @xmath27 , are @xmath28-dimensional dimensionful yukawa couplings which can be rescaled to the ones in 4-dimension as @xmath29 , @xmath30 for @xmath31 , are the two scalar doublets , @xmath32 and @xmath33 are lepton doublets and singlets , respectively .
the scalar and lepton doublets are the functions of @xmath34 and @xmath35 ( @xmath35 , @xmath36 ) , where @xmath35 ( @xmath35 , @xmath36 ) is the coordinate represents the @xmath28th dimension . with the choice of the doublets @xmath37 and @xmath38 @xmath39\ , ; \phi_{2}=\frac{1}{\sqrt{2}}\left(\begin{array}{c c } \sqrt{2 } h^{+}\\ h_1+i h_2 \end{array}\right ) \,\
, , \label{choice}\end{aligned}\ ] ] and the vacuum expectation values @xmath40 it is possible to switch off the mixing between neutral scalar higgs bosons and to separate the particle spectrum so that the sm particles are collected in the first doublet and the new particles in the second one .
( @xmath41 ) is the well known mass eigenstate @xmath5 ( @xmath6 ) . ]
notice that the yukawa matrices @xmath26 are responsible for producing the fcnc at tree level and their entries are complex in general .
notice that , in the following , we replace @xmath42 with @xmath43 where `` n '' denotes the word `` neutral '' . in the case of the non - universal two extra dimensions where only the new higgs field @xmath38 is accessible to extra dimension in the higgs sector , the compactification on a torus @xmath13 , causes to appear kk modes @xmath44 of @xmath38 in two spatial extra dimension as @xmath45 where @xmath46 , the indices @xmath22 and @xmath47 are positive integers including zero but both are not zero at the same time . here
@xmath48 the 4-dimensional higgs doublet which includes the charged higgs boson @xmath49 , the neutral cp even ( odd ) @xmath5 ( @xmath6 ) higgs bosons .
the kk mode of the charged higgs boson has the mass @xmath50 .
similarly , the neutral cp even ( odd ) higgs @xmath5 ( @xmath6 ) has the mass @xmath51 ( @xmath52 ) , where @xmath17 and @xmath18 .
furthermore , we assume that the compactification radius @xmath8 is the same for both dimensions .
the expansion for a single extra dimension can be obtained by setting @xmath53 , taking @xmath54 , and dropping the summation over @xmath47 .
notice that the gauge fields feel also the extra dimensions which results in kk modes after the compactification .
however , they are irrelevant in our calculation since they do not play any role in the processes under consideration . at this stage , we consider the flavor violating radiative decays of leptons , @xmath0 and @xmath12 , by introducing a single non - universal extra spatial dimension .
these decays exist at least at the one loop level with the help of the intermediate neutral higgs bosons @xmath5 and @xmath6 . with the inclusion of a single non - universal extra dimension ,
the kk modes of neutral higgs bosons , namely @xmath55 and @xmath56 , contribute after the compactification ( fig .
[ fig1 ] ) . here
the vertices involve two zero modes and one kk mode , since the kk mode conservation does not exist in the case of non - universal extra dimension , in contrary to the universal one . in the loop calculations ,
the logarithmic divergences are eliminated by using the on - shell renormalization scheme . in this scheme ,
the self energy diagrams for on - shell leptons vanish since they can be written as @xmath57 the divergences coming from the vertex diagrams can be eliminated by introducing a counter term @xmath58 so that we have the relation @xmath59 with the renormalized ( bare ) vertex @xmath60 ( @xmath61 ) and using the gauge invariance : @xmath62 where @xmath63 is the photon four momentum vector . taking the vertex diagrams @xmath64 and @xmath65 in fig .
[ fig1 ] and introducing only @xmath66 lepton for the internal line , the decay width @xmath67 reads as @xmath68 where @xmath69 with @xmath70 , left ( right ) chiral amplitudes @xmath71 ( @xmath72 ) .
similarly , the decay width of the lfv process @xmath1 can be obtained as @xmath73 [ dwtaumugam ] where @xmath74 with @xmath70 , left ( right ) chiral amplitudes @xmath75 ( @xmath76 ) .
the functions @xmath77 and @xmath78 in eqs .
( [ a1a2 ] ) and ( [ b1b2 ] ) read @xmath79 and @xmath80 , @xmath81 , @xmath82 is the charge of @xmath66 lepton , the dimensionfull yukawa couplings @xmath83 appearing in the expressions are defined as @xmath84 .
furthermore , the theoretical results is due to the internal @xmath66-lepton contribution since , we respect the sher scenerio @xcite , results in the couplings @xmath85 ( @xmath86 ) , are small compared to @xmath87 @xmath88 , due to the possible proportionality of them to the masses of leptons under consideration in the vertices .
furthermore , we take the yukawa couplings for the interactions lepton - lepton - kk mode of higgs bosons ( @xmath5 and @xmath6 ) as the same as the ones for the interactions of zero mode fields . in the case
two extra spatial dimensions which are felt by the second higgs doublet in the higgs sector , @xmath38 can be expanded into its kk modes as in eq .
( [ sechiggsfield ] ) after the compactification of the extra dimensions on a torus @xmath13 . in the two non - universal spatial extra dimensions
the forms of the expressions ( [ a1a2 ] ) and ( [ b1b2 ] ) are the same , except that , the parameters @xmath89 and @xmath15 are replaced by , @xmath90 and @xmath16 . furthermore , the number @xmath91 in front of the summation should be replaced by @xmath92 and the summation should be done for @xmath93 except @xmath94 .
the radiative lfv decays @xmath0 and @xmath1 are controlled by the yukawa couplings @xmath95 in the model iii version of the 2hdm .
these couplings should be restricted by using the experimental results , since they are free parameters of the theory .
the dominant couplings are @xmath96 , ( @xmath97 ) and @xmath96 ( @xmath98 ) for the @xmath0 and @xmath1 decays , respectively . here
, we assume that , the strength of these couplings are related with the masses of leptons denoted by the indices of them , similar to the cheng - sher scenerio @xcite .
this assumption forces us to consider only the @xmath66 lepton at the internal line for both processes .
we also assume that the yukawa couplings @xmath83 is symmetric with respect to the indices @xmath99 and @xmath100 .
for the numerical values of these couplings we use the following restrictions : * the upper limit of @xmath101 is predicted as @xmath102 ( see @xcite and references therein ) by using the experimental uncertainty , @xmath103 , in the measurement of the muon anomalous magnetic moment and assuming that the new physics effects can not exceed this uncertainty .
* for the yukawa coupling @xmath104 , we use broad range of the prediction , @xmath105 , which is obtained by using the experimental upper limit of br of @xmath106 decay , @xmath107 and predicted upper limit of @xmath108 ( see @xcite ) . * for the yukawa couplings @xmath109 , we use the numerical values which is greater than the upper limit of @xmath101 .
with the addition of the extra dimensions , the new contributions to the physical parameters arise for these decays . in the case of non - universal extra dimensions ,
the kk number at each vertex is not conserved in contrast to the universal one case and lepton - lepton - kk neutral higgs vertices appear . due to the compactification of the a single ( double ) extra dimension on a circle ( torus ) @xmath7 ( @xmath13 ) , the kk modes of internal neutral higgs bosons @xmath5 and @xmath6 appear and they bring additional contributions to the physical quantities of the decays under consideration .
our work is devoted to the prediction of the non - universal extra dimensions on the br of the lfv processes @xmath110 and @xmath111 , in the framework of the type iii 2hdm .
we make this anaysis in one and two extra dimensions . in the case of two extra dimensions
the crowd of kk modes cause to be more enhancement in the @xmath2 of the lfv decays compared to the case of a single extra dimension . on the other hand , the abundance of kk modes in the summation can lead to the divergence in the calculation of the @xmath2 .
however the the ratio @xmath112 appearing in the expressions converges to zero sharply with the increasing values of the integers @xmath22 and @xmath47 and it leads to the convergence of the kk sum for the compactification scale we study , @xmath113 . for the ued case the above ratio has the form @xmath114
since kk number at each vertex is conserved and kk lepton - lepton - kk neutral higgs interactions are switched on . increasing values of @xmath22 and @xmath47 forces the ratio to reach one and the convergence problem of the kk sum should be studied carefully .
[ brmuegamr1 ] ( [ brmuegamr2 ] ) is devoted to the compactification scale @xmath115 dependence of the br of the lfv decay @xmath106 , for @xmath116 , @xmath117 , @xmath118 , and four different values of the coupling @xmath119 , in the case of one ( two ) non - universal extra dimension(s ) . in fig .
[ brmuegamr1 ] , the solid - dashed - small dashed straight lines ( curves ) represent the 2hdm ( the extra dimension ) contribution to the br for @xmath120 .
it is shown that the br is not sensitive to the extra dimension effects for the single extra dimension . in the case of two non - universal extra dimensions ( see fig .
[ brmuegamr2 ] ) , the sensitivity increases and the effects of extra dimensions can reach even to the main contribution for the small values of compactification scale .
this enhancement in the case of two extra dimensions is due to the abundance of kk modes of neutral higgs bosons . in fig .
[ brtaumugamr1 ] ( [ brtaumugamr2 ] ) we present the compactification scale @xmath115 dependence of the br of the lfv decay @xmath121 for @xmath116 , @xmath117 , @xmath118 , and five different values of the coupling @xmath122 , in the case of one ( two ) non - universal extra dimension(s ) . in fig .
[ brtaumugamr1 ] the solid - dashed - small dashed - dotted - dot dashed straight lines ( curves ) represent the 2hdm ( the extra dimension ) contribution to the br for @xmath123 .
similar to the @xmath106 decay the br is not sensitive to the extra dimension effects for the single extra dimension .
however fig .
[ brtaumugamr2 ] shows that the br is much more sensitive to the contribution coming from the kk modes in the case of two exta dimensions , in comparison with the single extra one .
this sensitivity becomes weak for @xmath124 .
now we would like to analyze the ratio @xmath125 where @xmath126 ( @xmath127 ) is the experimental upper limit ( theoretical value ) of the @xmath128 and @xmath129 is the new contributions coming from the extra dimensions .
[ ratiotaumugam1 ] ( [ ratiotaumugam2 ] ) represents the compactification scale @xmath115 dependence of the @xmath130 for the lfv decay @xmath121 for @xmath116 , @xmath117 , @xmath118 , and three different values of the coupling @xmath122 in the case of one ( two ) non - universal extra dimension(s ) . in fig .
[ ratiotaumugam1 ] ( [ ratiotaumugam2 ] ) the solid - dashed - small dashed lines represent the @xmath130 for @xmath131 , for a single ( double ) non - universal extra dimension .
[ ratiotaumugam1 ] shows that the discrepancy between the experimental value and the 2hdm prediction can not be compensated with the addition of extra dimensions . on the other hand
this ratio increases with the increasing values of the coupling constant @xmath132 . for two non - universal extra dimensions it is possible to cover the discrepancy between the experimental and the theoretical values for the small values of the compactification scale @xmath115 .
for @xmath133 the magnitude of extra dimension effect becomes almost @xmath134 of the difference of the experimental and the theoretical values of the br . in fig .
[ ratiotaumugamrr1 ] ( [ ratiotaumugamrr2 ] ) we present the parameter @xmath135 dependence of the @xmath130 for the @xmath121 decay for @xmath117 , @xmath118 , @xmath136 , and three different values of the compactification scale @xmath115 , @xmath137 in the case of one ( two ) non - universal extra dimension(s ) .
these figures show that the @xmath130 is sensitive to the mass ratio of neutral higgs bosons @xmath5 and @xmath6 especially for small @xmath115 .
the increasing values of @xmath47 causes to decrease the ratio and this sensitivity becomes weak for the large values of the scale @xmath115 . in the case that the neutral higgs masses are nearly degenerate the @xmath130 is suppressed almost one order compared to the case that the parameter @xmath138 for the intermediate values of @xmath115 .
now we would like to present the results briefly . *
the single non - universal extra dimension effects on the br s of @xmath106 and @xmath121 decays are considerably weak , however , the addition of one more spatial dimension causes relatively strong sensitivity on the br s . *
the contribution coming from the single non - universal extra dimension is not enough large to compensate the difference between the experimental and the theoretical values of the br . in the case of two non - universal extra dimensions
this discrepancy can be covered by kk modes of neutral higgs bosons for the small values of the compactification scale @xmath115 . *
the extra dimension contributions are sensitive to the mass ratio of neutral higgs bosons @xmath5 and @xmath6 and it becomes weaker with the increasing values of the compactification scale @xmath115 . as a final comment , the effect of the single non - universal extra dimension on the br s of lfv decays @xmath0 and @xmath1 is weak
. however , their contributions enhance in the case of two extra dimensions and the more accurate future experimental results of these decays , hopefully , will be helpful to analyze the possible signals coming from the extra dimensions .
this work has been supported by the turkish academy of sciences in the framework of the young scientist award program .
( eoi - tuba - gebip/2001 - 1 - 8 ) 1 m. l. brooks et .
mega collaboration , _ phys .
lett . _ * 83 * , 1521 ( 1999 ) .
s. ahmed et.al .
, cleo collaboration , _ phys .
rev . _ * d61 * , 071101 ( 2000 ) .
r. barbieri and l. j. hall , _ phys .
lett . _ * b338 * , 212 ( 1994 ) ; r. barbieri , l. j. hall and a. strumia , _ nucl .
phys . _ * b445 * , 219 ( 1995 ) ; r. barbieri , l. j. hall and a. strumia , _ nucl .
phys . _ * b449 * , 437 ( 1995 ) ; p. ciafaloni , a. romanino and a. strumia , _ nucl . phys . _
* b458 * , 3 ( 1996 ) ; t. v. duong , b. dutta and e. keith , _ phys .
* b378 * , 128 ( 1996 ) ; g. couture , et .
al . , _ eur .
phys . j. _ * c7 * , 135 ( 1999 ) ; y. okada , k. okumara and y. shimizu , _ phys .
rev . _ * d61 * , 094001 ( 2000 ) .
e. iltan , _ phys .
rev . _ * d64 * , 115005 , ( 2001 ) ; _ phys .
rev . _ * d64 * 013013 , ( 2001 ) .
r. diaz , r. martinez and j - alexis rodriguez , phys.rev .
d63 ( 2001 ) 095007 .
d. chang , w. s. hou and w. y. keung , _ phys .
rev . _ * d48 * , 217 ( 1993 ) .
n. arkani - hamed , s. dimopoulos , and g. r. dvali , _ phys .
_ * d59 * 086004 ( 1999 ) ; n. arkani - hamed , s. dimopoulos , and g. r. dvali , _ phys .
lett . _ * b429 * , 263 ( 1998 ) ; .
i. antoniadis,_phys .
lett . _ * b246 * , 377 ( 1990 ) ; i.antoniadis and k. benakli , _ phys .
lett . _ * b326 * , 69 ( 1994 ) .
i. antoniadis , k. benakli , and m. quiros , _ phys . lett . _ * b331 * , 313 ( 1994 ) ; a. pomarol and m. quiros , _ phys . lett . _ * b438 * , 255 ( 1998 ) ; i. antoniadis , k. benakli , and m. quiros , _ phys .
* b460 * , 176 ( 1999 ) ; p. nath , y. yamada , and m. yamaguchi , _ phys .
_ * b466 * , 100 ( 1999 ) ; m. masip and a. pomarol , _ phys .
rev . _ * d60 * , 096005 ( 1999 ) a. delgado , a. pomarol , and m. quiros , _ jhep _ * 01 * , 030 ( 2000 ) ; t. g. rizzo and j. d. wells , _ phys .
rev . _ * d61 * , 016007 ( 2000 ) ; p. nath and m. yamaguchi , _ phys .
rev . _ * d60 * , 116004 ( 1999 ) ; a. muck , a. pilaftsis , and r. ruckl , _ phys .
rev . _ * d65 * , 085037 ( 2002 ) ; a. muck , a. pilaftsis , and r. ruckl , hep - ph/0203032 ; c. d. carone , _ phys .
rev . _ * d61 * , 015008 ( 2000 ) .
i. antoniadis , c. munoz , m. quiros , _ nucl . phys _ * b397 * 515 ( 1993 ) ; t. appelquist , h .- c .
cheng , b. a. dobrescu , _ phys .
_ * d64 * 035002 ( 2001 ) . j. papavassiliou and a. santamaria , _ phys .
rev . _ * d63 * , 016002 ( 2001 ) .
d. chakraverty , k. huitu , and a. kundu , phys.lett .
b558 ( 2003 ) 173 - 181 ; a. j. buras , m. spranger , and a. weiler , nucl.phys .
b660 ( 2003 ) 225 . k. agashe , n.g .
deshpande , g .- h .
wu , _ phys .
lett . _ * 514 * 309 ( 2001 ) .
k. r. dienes , e. dudas , t. gherghetta 9811428 , q.
h. shrihari , s. gopalakrishna , c. p. yuan , 0312339 f. agulia , m. p. victoria , j. santiago , _ jhep _ * 0302 * 051 , ( 2003 ) ; _ acta phys.polon._ * b34 * , 5511 ( 2003 ) .
e. o. iltan , hep - ph/0401229 ; e. o. iltan , _ jhep _ * 0404 * 018 , ( 2004 ) . c. s. lam , hep - ph/0302227 , 2003 ; c. a. scrucca , m. serona , l. silvestrini , _ nucl.phys .
_ * b669 * 128 , ( 2003 ) ; m. gozdz , w. a. kaminsk , _ phys .
rev . _ * d68 * 057901 , ( 2003 ) ; c. biggio , et.al , _ nucl.phys . _ * b677 * 451 , ( 2004 ) ; m. carena , et.al , _ phys .
_ * d68 * 035010 , ( 2003 ) ; a. j. buras , et .
al . , hep - ph/0306158 , 2003 ; t. g. rizzo , _ jhep _ * 0308 * 051 ( 2003 ) , a. j. buras , hep - ph/0307202 , 2003 ; s. matsuda , s. seki , hep - ph/0307361 , 2003 ; r. n. mohapatra , _ phys .
* d68 * 116001 , ( 2003 ) ; b. lillie , _ jhep _ * 0312 * 030 ( 2003 ) ; a.a arkhipov , hep - ph/0309327 2003 , e. o. iltan , _ jhep _ * 0402 * 065 , ( 2004 ) , f.feruglia , 0401033 . t. p. cheng and m. sher , _ phy .
rev . _ * d35 * , 3383 ( 1987 ) .
e. iltan , h. sundu , _
acta phys.slov._ * 53 * 17 , ( 2003 ) . | we study the effect of non - universal extra dimensions on the branching ratios of the lepton flavor violating processes @xmath0 and @xmath1 in the general two higgs doublet model .
we observe that these effects are small for a single extra dimension , however , in the case of two extra dimensions there is a considerable enhancement in the additional contributions . |
this note is devoted to answer a question in kirby s problem list @xcite asking whether every simply connected smooth 4-manifold with @xmath1 contains a gompf nucleus @xmath0 ( problem 4.98 in @xcite ) .
we prove that by doing logarithmic transformations on three linearly independent tori in the _
k3_-surface we get a 4-manifold not contaning @xmath0 . to make our statements more precise we need a few definitions .
the hypersurface @xmath2\in { { \bbb { cp}}}^3\ \vert \ \sum _ { i=0 } ^3 z_i ^4=0\rbrace \subset { { \bbb { cp}}}^3 $ ] is a simply connected , smooth 4-manifold with @xmath3 , hence it is a _
k3-surface_. it is known that all simply connected complex surfaces with vanishing first chern class are diffeomorphic , consequently from the differential topological point of view @xmath4 is _ the k3_-surface .
the complex surface @xmath4 admits a holomorphic fibration @xmath5 such that the generic fiber is a smooth elliptic curve a 2-dimensional torus .
such fibrations are called _
elliptic fibrations_. it can be assumed that @xmath6 has a singular fiber homeomorphic to the 2-dimensional sphere @xmath7 such fibers are called _
cusp fibers_. the fibration also has a section @xmath8 , the image of which is a sphere @xmath9 with square @xmath10 ^ 2=-2 $ ] .
the _ gompf nucleus _
@xmath0 is by definition the tubular neighborhood of the union of a cusp fiber and a section in @xmath4 .
the manifold @xmath0 admits a handle decomposition with one 0-handle and two 2-handles , where these two 2-handles are attached to @xmath11 according to the kirby diagram shown in figure [ egyes ] . it can be shown that the _ k3_-surface @xmath4 contains three disjoint copies of @xmath0 .
if a 4-manifold @xmath12 contains a 2-dimensional torus @xmath13 with square 0 , then we can perform a _ logarithmic transformation _ on @xmath13 : deleting the tubular neighborhood of @xmath13 ( which is diffeomorphic to @xmath14 ) and regluing it via a diffeomorphism @xmath15 we get a new manifold @xmath16 .
it turnes out that if @xmath13 is the fiber in a gompf nucleus @xmath17 , then the diffeomorphism type of @xmath16 will depend only on one nonnegative number @xmath18 associated to @xmath19 .
this number is called the _ multiplicity _ of the logarithmic transformation . for more about elliptic surfaces , nuclei and logarithmic transformations
see @xcite , @xcite and @xcite .
now perform logarithmic transformations of multiplicity 2 on the three tori contained by the three disjoint nuclei in the _ k3_-surface @xmath4 .
the resulting manifold will be denoted by @xmath20 .
the 4-manifold @xmath20 does not contain a gompf nucleus @xmath0 .
[ elsotetel ] one of the most interesting questions in 4-manifold theory is whether all 4-manifolds are of simple type or not .
( for the definition of simple type see section [ masodik ] . )
it is known that if @xmath12 contains a homologically essential torus with square 0 , then @xmath21 is of simple type .
there is no example of a 4-manifold with @xmath1 and not containing a torus with square 0 .
one can ask whether every 4-manifold contains a cusp neeighborhood ( a tubular neighborhood of a cusp fiber ) or even a gompf nucleus @xmath0 .
the above theorem shows an example of a manifold which contains no @xmath0 ; @xmath20 is still of simple type , however .
one can modify the question by trying to find more general nuclei @xmath22 in 4-manifolds .
the manifold @xmath22 is described by the kirby diagram shown in figure [ kettes ] it can be defined alternatively as the tubular neighborhhod of the union of a cups fiber and a section in a simply connected elliptic surface ( admitting a section ) with euler characteristic @xmath23 ( see @xcite ) . for every @xmath24
there is a manifold @xmath25 such that @xmath25 does not contain @xmath22 .
[ masodiktetel ] a related question would be to find a 4-manifold @xmath21 with the property that @xmath21 does not contain @xmath26 ; here @xmath26 denotes the 4-manifold with boundary we get by performing two logarithmic transformations ( of multiplicity @xmath18 and @xmath27 ) along the fiber of the gompf nucleus @xmath0 .
using a recent construction of fintushel and stern @xcite such @xmath21 can be found , see @xcite .
_ acknowledgement _ : the author would like to thank msri for their hospitality and support , and zoltn szab for many helpful conversations .
in studying differential topological properties of smooth 4-manifolds , _ seiberg - witten basic classes _ turn out to be very powerful tools . for the definition of these objects see @xcite , @xcite or @xcite , here we restrict ourselves to a very short outline .
assume that @xmath12 is a smooth , oriented , simply connected , closed 4-manifold with @xmath28 .
a cohomology class @xmath29 with @xmath30 uniquely determines a spin@xmath31 structure on @xmath21 , and for such a structure a certain pair of partial differential equations ( the so - called seiberg - witten equations ) can be described involving a choice of metric on @xmath21 , a coupled dirac operator and a perturbation 2-form . by a delicate `` counting argument '' of the solutions of these equations
a number @xmath32 can be associated to the cohomology class @xmath33 .
it turnes out that this number ( up to sign ) is a smooth invariant of the manifold @xmath21 , more precisely if @xmath34 is an orientation preserving diffeomorphism then @xmath35 .
moreover , for a fixed 4-manifold @xmath21 there are only finitely many classes @xmath33 with @xmath36 , and @xmath37 . the cohomology class @xmath38 is called a _ seiberg - witten basic class _ if @xmath36 .
the 4-manifold @xmath21 is of simple type if @xmath36 implies that @xmath39 ; here @xmath40 and @xmath41 stand for the signature and euler characteristic of @xmath21 .
the most important relation between the smooth topology of a 4-manifold @xmath21 and its basic classes @xmath42 is shown by the _ generalized adjunction formula _ : ( kronheimer - mrowka ) if @xmath43 is a smooth , connected 2-dimensional surface of genus @xmath44 , @xmath45\neq 0 $ ] and @xmath45 ^ 2\geq 0 $ ] , then for every basic class @xmath33 we have @xmath46 ^ 2+\vert k([\sigma ] ) \vert .\ ] ]
assume that @xmath47 and @xmath48 are three tori in the _
k3_-surface lying in three disjoint gompf nuclei .
perform logarithmic transformations of multiplicity 2 on each @xmath49 .
the basic classes of the resulting manifold @xmath20 are determined by fintushel and stern @xcite .
_ of theorem [ elsotetel ] : _ assume that @xmath51
. the homology class of the fiber and the section in @xmath0 will be denoted by @xmath52 and @xmath53 respectively ; note that @xmath52 and @xmath54 have square 0 and can be represented by tori .
consequently ( by the generalized adjunction formula ) a basic class @xmath33 of @xmath20 evaluates trivially on @xmath52 and @xmath55 .
now @xmath56 + [ { { t_j}\over 2}]+ [ { { t_k}\over 2}])=0 $ ] and @xmath57 + [ { { t_j}\over 2}]+ [ { { t_k}\over 2}])0 $ ] ( @xmath58 ) implies that @xmath59=0 $ ] , similarly @xmath60 = 0 $ ] for @xmath61 . since the complement of the three disjoint nuclei in @xmath4 have negative definite intersection form , the above equalities imply that @xmath62 $ ] and similarly @xmath63 $ ] .
these two latter equations , however , give a contradiction , since @xmath64 but @xmath65)\cdot ( \sum _ { i=1 } ^3 \beta _ j [ t_j])=0 $ ] .
consequently @xmath0 does not embed in @xmath20 .
_ of theorem [ masodiktetel ] : _ perform logarithmic transsformations of multiplicity @xmath66 on @xmath47 and @xmath48 as above ; the resulting 4-manifold @xmath67 will be denoted by @xmath25 .
the basic classes of @xmath25 are determined in @xcite : these are the poincar duals of the homology classes @xmath68+\beta [ { { t_2}\over { 2n}}]+ \gamma [ { { t_3}\over { 2n}}]$ ] where @xmath69 .
assume now that @xmath70 .
as before , @xmath52 and @xmath53 will denote the homology classes of the fiber and the section in @xmath22 respectively .
note that @xmath71 and @xmath72 .
the class @xmath52 can be represented by a torus , while @xmath73 can be represented by a surface of genus @xmath24 ; moreover @xmath74 .
the same argument as in the proof of theorem [ elsotetel ] gives that @xmath62 $ ] . to show that every basic class of @xmath25 evaluates trivially on @xmath55
involves a little trick : take the basic classes @xmath50\pm [ { { t_2}\over 2}]\pm [ { { t_3}\over 2}]$ ] and evaluate them on @xmath55 . if one of them , say @xmath33 , evaluates nontrivially on @xmath55 , then for @xmath75 ( which is also a basic class ) we have @xmath76 , but then @xmath77 and @xmath55 would violate the generalized adjunction formula .
consequently @xmath78\pm [ { { t_2}\over 2}]\pm [ { { t_3}\over 2}])\cdot g=0 $ ] implying that @xmath79=0 $ ] @xmath80 .
these latter equations show that @xmath81 $ ] , and we have the same contradiction as before . | in this note we show that there are 4-manifolds not containing gompf nucleus @xmath0 ; in this way we answer problem 4.98 of kirby s problem list ( see @xcite ) in the negative .
[ section ] [ fact]lemma [ fact]theorem [ fact]definition [ fact]remark [ fact]proposition # 1#2 # 1 _ # 1 .
_ # 1 _ # 1 .
_ = cmr7 = msbm10 at 11pt = msbm7 at 9pt = msbm5 at 7pt = = = # 1#1 o # 1#2 |
the notion of an independent set plays a fundamental role in the study of hypergraphs .
an @xmath0-uniform hypergraph , or @xmath0-_graph _ , @xmath1 is a pair @xmath14 comprising two sets , the vertices @xmath15 and edges @xmath16 of @xmath1 , where each edge @xmath17 is a set of @xmath0 elements of @xmath15 .
hence a 2-graph is an ordinary graph .
a set @xmath18 is _ independent _ if there is no edge @xmath17 with @xmath19 .
whilst there are many theorems in the literature that can be phrased in terms of estimating the number of independent sets in certain hypergraphs ( we shall mention some of these shortly ) , the question _ per se _ of how many independent sets there can be in a graph has attracted attention only relatively lately . the maximum number of independent sets in a graph of given average degree can be determined easily via the kruskal - katona theorem @xcite , but for regular graphs the maximum is harder to find : following a good estimate by alon @xcite , the exact value for bipartite graphs was determined by kahn @xcite via an elegant entropy argument , and his result was extended to all graphs by zhao @xcite .
there are at most @xmath20 independent sets in a @xmath3-regular graph of order@xmath21 ( that is , having @xmath10 vertices ) , and this number is attained by @xmath22 disjoint copies of @xmath23 .
it would be convenient for many purposes if there were at most @xmath24 independent sets in an @xmath0-graph @xmath1 of order @xmath10 and average degree @xmath3 , but examples like those just cited show this hope to be a forlorn one .
nevertheless , for the applications we have in mind , it is enough to find a good collection @xmath2 of _ containers _ for independent sets : this is a family of subsets of @xmath15 such that , for each independent set @xmath25 , there is a set @xmath26 with @xmath27 , and @xmath28 . of course
, we could just take @xmath29 , but this collection would not be helpful : for @xmath2 to be of use , a further condition is needed that each container @xmath26 is not large , in a sense made precise later . another immediate candidate for @xmath2 is the collection of _ maximal _ independent sets , but this too can be large ; for example , if @xmath3 is even , adding a @xmath30-factor into the vertex classes of each @xmath23 of the graph @xmath31 produces a @xmath32-regular graph with at least @xmath33 maximal independent sets .
( the maximum number of maximal independent sets in any graph of order @xmath10 was determined by moon and moser @xcite . ) the main purpose of this paper is to show that an @xmath0-graph @xmath1 of average degree @xmath3 and order @xmath10 does have a small collection @xmath2 of containers .
typically , but not always , @xmath34 .
results of this kind were known previously in special cases .
sapozhenko @xcite treated regular 2-graphs .
containers for @xmath0-graphs were introduced and used in @xcite in the case of simple regular @xmath0-graphs ( a hypergraph is _ simple _ or _ linear _ if every pair of vertices lies in at most one edge ) . however , the most interesting applications require containers for non - regular @xmath0-graphs . finding such containers
presents significant difficulties and the method here is unrelated to that of @xcite .
we describe our main results about containers in [ sec : containers ] .
the fundamental result is theorem [ thm : cover ] stated in [ subsec : mainthm ] .
it is worth mentioning that the statement applies to all @xmath0-graphs @xmath1 but it gives useful information only if @xmath3 is large ( though independently of @xmath10 ) .
following [ sec : containers ] , the containers are constructed in [ sec : online ] and their properties analysed in [ sec : calc ] , where theorem [ thm : cover ] is proved .
some corollaries , more amenable to application , are developed in [ sec : iteration ] and [ sec : uniform ] , and in [ sec : largecont ] we give an example to show that the main result is , in a sense , best possible .
finally , in [ sec : list][sec : sparse ] we give details of some applications . before getting down to details , though , we outline those applications to be discussed .
each of them involves , implicitly , dealing with the independent sets in some hypergraph @xmath1 , and in each case it suffices to find an appropriate set of containers .
once some simple parameter of @xmath1 has been calculated , the existence of the desired containers follows straight away from the main theorem or one of its consequences in [ sec : iteration][sec : uniform ] , and the applications are finished off by routine arguments ( plus , in [ sec : sumfree ] , a non - trivial removal lemma ) .
the results obtained are generally best possible .
balogh , morris and samotij @xcite have independently developed a container theorem akin to corollary [ cor : sparse_container ] together with applications , including a proof of theorem [ thm : klr3 ] for balanced graphs .
we use standard notation .
in particular , for @xmath35 we let @xmath36=\{1,\ldots , n\}$ ] and @xmath37=\{m,\ldots , n\}$ ] . for collections of subsets we write , for example , @xmath37^{(s)}=\{\sigma\subset[m , n]:|\sigma|=s\}$ ] , @xmath37^{(>s)}=\{\sigma\subset[m , n]:|\sigma|>s\}$ ] , and so on .
as usual , @xmath38 $ ] denotes the collection of all subsets of @xmath36 $ ] . if @xmath1 is a hypergraph we write @xmath39 for the number of edges of @xmath1 and @xmath40 for the number of vertices of @xmath1 .
if @xmath41 then @xmath42 $ ] denotes the subhypergraph of @xmath1 induced by @xmath43 , that is , @xmath42=(s , e(g)\cap\mathcal{p } s)$ ] .
a @xmath44-graph @xmath1 is said to be @xmath9-_choosable _ if , whenever for each vertex @xmath45 we assign a list @xmath46 of @xmath9 colours to @xmath47 , then it is possible to choose a colour for @xmath47 from the list @xmath46 , so that no two adjacent vertices receive the same colour . the _ list chromatic number _ @xmath48 ( also called the _ choice number _ ) is the smallest @xmath9 such that @xmath1 is @xmath9-choosable .
if all the lists are the same then a list colouring is just an ordinary @xmath9-colouring and so @xmath48 is at least @xmath49 , the ordinary chromatic number of @xmath1 .
this natural definition was first studied by vizing @xcite and by erds , rubin and taylor @xcite .
one of the main discoveries of @xcite is that @xmath48 can be much larger than @xmath49 , because @xmath50 , whereas @xmath51 .
in fact , unlike @xmath49 , @xmath48 must grow with the minimum degree of the graph @xmath1 .
alon @xcite showed that @xmath52 holds for any graph @xmath1 of minimum degree @xmath3 .
there is a straightforward reason , as pointed out by alon and kostochka @xcite , why the same is not true for @xmath0-graphs if @xmath6 .
let @xmath53 be some graph on @xmath10 vertices , say @xmath54 , and let @xmath1 be some @xmath0-graph each of whose edges contains an edge of @xmath53 . then @xmath55 , so in this example @xmath56 , whereas the average degree of @xmath1 can be large
. however , if we restrict to simple @xmath0-graphs the situation is different . following work of haxell and pei @xcite on steiner systems , haxell and verstrate
@xcite proved that @xmath57 for simple @xmath3-regular @xmath58-graphs @xmath1 .
alon and kostochka @xcite showed @xmath59 for simple @xmath3-regular @xmath0-graphs @xmath1 , and in @xcite this was improved to @xmath60 .
we extend this to all simple @xmath0-graphs , at the same time giving a better constant .
[ thm : chil ] let @xmath61 be fixed .
let @xmath1 be a simple @xmath0-graph with average degree @xmath3 . then , as @xmath62 , @xmath63 holds .
moreover , if @xmath1 is regular then @xmath64 note that , for @xmath5 , this improves alon s bound @xcite by a factor of 2 and is best possible .
we think that the bound given for regular @xmath0-graphs might hold for general @xmath0-graphs and moreover that it too might be best possible ( see [ sec : list ] ) .
theorem [ thm : chil ] is a weaker version of corollary [ cor : chil ] , which gives a bound for some non - simple @xmath0-graphs also .
we do not give a general bound for all @xmath0-graphs because it would be rather complicated to state : however , in any particular instance , a bound can readily be derived from the results in [ sec : list ] .
this would cover , for example the theorem of alon and kostochka @xcite that if at least half the @xmath65-tuples of vertices of @xmath1 lie in at least @xmath66 edges then @xmath67 .
an @xmath7-graph on vertex set @xmath68 $ ] is said to be @xmath8-_free _ if it contains no subgraph isomorphic to the @xmath7-graph @xmath8 .
as far as @xmath8-free graphs are concerned , our main result is this : for any given @xmath7-graph @xmath8 , though there are many @xmath8-free @xmath7-graphs , each of these is contained in one of a very few @xmath7-graphs that are almost @xmath8-free .
more exactly , there is a small collection @xmath2 of @xmath7-graphs , each @xmath8-free @xmath7-graph being a subgraph of an @xmath7-graph in @xmath2 , and no @xmath7-graph in @xmath2 having more than @xmath69 copies of @xmath8 .
the main content of the theorem is that the number of containers is very small . for graphs at least
, szemerdi s regularity lemma gives a set of containers with @xmath70 , but the size of @xmath2 in the theorem is much smaller .
it is expressed in terms of a parameter @xmath13 that appears often in the literature .
[ def : mh ] for an @xmath7-graph @xmath8 with @xmath71 , let @xmath72 sometimes , @xmath8 is called ( strictly ) balanced if the maximum is attained ( uniquely ) when @xmath73 .
however , this restriction is not needed in any of our arguments and it is ignored . we then obtain the following , where @xmath74 . [
thm : ffree_cover ] let @xmath8 be an @xmath7-graph with @xmath71 and let @xmath75 .
for some @xmath76 and for @xmath77 sufficiently large , there exists a collection @xmath2 of @xmath7-graphs on vertex set @xmath68 $ ] such that * for every @xmath8-free @xmath7-graph @xmath25 on vertex set @xmath68 $ ] there exists @xmath26 with @xmath27 , * moreover , for every pair @xmath78 in ( a ) , there exists @xmath79 where @xmath80 , @xmath81 and @xmath82 , such that @xmath83 , * for every @xmath7-graph @xmath84 , the number of copies of @xmath8 in @xmath85 is at most @xmath86 , and @xmath87 , * @xmath88 .
the condition ( b ) just says that @xmath85 is determined by @xmath89 , which is comprised of small subsets of @xmath15 ( actually subsets of @xmath25 this sometimes matters , because it means ( b ) is stronger than ( d ) ) .
the existence of @xmath2 follows straightforwardly from the results in [ sec : containers ] , as shown in [ sec : hfree ] , by considering the @xmath90-graph @xmath91 , whose @xmath92 vertices are the @xmath7-sets in @xmath68 $ ] , and whose edges are subsets of @xmath15 spanning a copy of @xmath8 in @xmath68 $ ] .
the subsets of @xmath15 are then @xmath7-graphs with vertex set @xmath68 $ ] , independent sets in @xmath1 corresponding to @xmath8-free @xmath7-graphs .
one corollary of theorem [ thm : ffree_cover ] is the next one . in the case
@xmath93 , it was proved for complete @xmath8 by erds , kleitman and rothschild @xcite and for general @xmath8 by erds , frankl and rdl @xcite .
nagle , rdl and schacht @xcite proved it for general @xmath7 using hypergraph regularity methods .
the proof here follows quite easily from theorem [ thm : ffree_cover ] using the supersaturation theorem of erds and simonovits @xcite ; details are in [ sec : hfree ] .
[ cor : ffree_count ] let @xmath8 be an @xmath7-graph .
the number of @xmath8-free @xmath7-graphs on vertex set @xmath68 $ ] is @xmath94 . for @xmath7-graphs @xmath8 which satisfy @xmath95 (
when @xmath93 this means @xmath8 is bipartite ) , we have @xmath96 , and corollary [ cor : ffree_count ] is unhelpful .
nevertheless our results can still be useful , provided appropriate information about @xmath97 is available .
the simplest case is @xmath93 and @xmath98 , where it is well known that @xmath99 ( erds , rnyi and ss @xcite ) , implying the trivial upper bound @xmath100 for the number of @xmath101-free graphs .
a direct application of our results readily gives the bound @xmath102 , but we do nt give details because kleitman and winston @xcite obtained a finer bound , namely @xmath103 . more generally , the number of @xmath104-free graphs has been well estimated by balogh and samotij @xcite . alongside the many results about @xmath8-free graphs , there is a corresponding corpus about _ induced _
@xmath8-free graphs , that is , graphs with no induced subgraph isomorphic to @xmath8 .
the number of induced @xmath8-free graphs was closely estimated by prmel and steger @xcite , and there have been many subsequent refinements .
if @xmath25 is an induced @xmath8-free @xmath7-graph , we need to ask what kind of object @xmath85 must be in order that the inclusion @xmath27 is helpful ; if @xmath85 itself is just an @xmath7-graph and @xmath27 means @xmath25 is a subgraph of @xmath85 , as in theorem [ thm : ffree_cover ] , then the induced subgraphs of @xmath25 differ from those of @xmath85 , which is no use .
we borrow the notion of _ 2-coloured multigraph _ from @xcite .
a 2-coloured @xmath7-multigraph @xmath85 on vertex set @xmath68 $ ] is a pair of edge sets @xmath105^{(\ell)}$ ] , which we call the red and the blue edge sets .
let @xmath25 be an @xmath7-graph on @xmath68 $ ] .
then we write @xmath27 if @xmath106 and @xmath68^{(\ell)}\setminus e(i)\subset c_b$ ] .
thus edges in @xmath107 do not affect whether @xmath27 .
[ thm : ind_ffree_cover ] let @xmath8 be an @xmath7-graph and let @xmath75
. for some @xmath76 and for @xmath77 sufficiently large , there exists a collection @xmath2 of 2-coloured @xmath7-multigraphs on vertex set @xmath68 $ ] such that * for every @xmath7-graph @xmath25 on vertex set @xmath68 $ ] with no induced copy of @xmath8 there exists @xmath26 with @xmath27 , * for every @xmath84 , the number of copies of @xmath8 in @xmath85 is at most @xmath86 , * @xmath108 .
this theorem can be used to recover basic results , akin to corollary [ cor : ffree_count ] , about the number of induced @xmath8-free @xmath7-graphs .
in fact we can state a probabilistic version just as readily .
let @xmath109 be a random @xmath7-graph obtained by choosing edges independently from the complete @xmath7-graph @xmath110 with probability @xmath111 .
the next statement involves a function @xmath112 , whose definition is natural enough but which is deferred until [ sec : hfree ] .
[ thm : ind_ffree_measure ] let @xmath113 and let @xmath8 be an @xmath7-graph
. then @xmath114 for graphs , that is , @xmath93 , this theorem was proved for @xmath115 by prmel and steger ( * ? ? ?
* theorem 1.3 ) and for general @xmath111 by bollobs and thomason ( * ? ? ?
* theorem 1.1 ) ( as illuminated by marchant and thomason @xcite ) . for general @xmath7
it was proved for @xmath115 by dotson and nagle @xcite , using hypergraph regularity techniques .
it will be clear that similar arguments to those described in this section can be used to obtain container results about other structures , such as tournaments .
let @xmath53 be either a finite field or the set of integers @xmath68 $ ] .
we consider linear systems of equations @xmath116 , where @xmath117 is a @xmath118 matrix with entries in @xmath53 , @xmath119 and @xmath120 .
we call such a triple @xmath121 a _ @xmath122 linear system_. [ defn : exteqnox ] for a @xmath122 linear system @xmath121 , a subset @xmath123 is _ solution - free _ if there is no @xmath124 with @xmath125 , and @xmath126 is the maximum size of a solution - free subset .
the notion of a solution - free subset is analogous to that of an @xmath8-free hypergraph in the previous section . once again , our contribution to this topic is a container theorem for solution - free sets .
the statement ( which extends to equations over abelian groups ) is given in theorem [ thm : eqn_cover ] ; it requires a few technical definitions so we omit it from the introduction .
nevertheless we mention a consequence for counting solution - free subsets . for an equation @xmath125 ,
how many solution - free subsets of @xmath53 are there ? a well - known instance of this question is to find the number of subsets @xmath127 $ ] containing no solution to @xmath128 ; the asymptotic answer , conjectured by cameron and erds @xcite , was given by green @xcite and by sapozhenko @xcite .
for a general system , every subset of a solution - free set is itself solution - free , so there are at least @xmath129 solution - free sets . for a single equation ( the case @xmath130 ) , it was shown by green @xcite that there are at most @xmath131 solution - free subsets .
the same bound does not always hold for @xmath132 . if some variables are closely tied to other variables
say the equations imply that @xmath133 then there can be significantly more than @xmath129 solution - free sets . however , a natural condition on @xmath117 rules out closely tied variables , and in this case green s bound holds good .
we say that @xmath117 has _ full rank _ if given any @xmath134 there exists @xmath119 with @xmath125 .
we then say that @xmath117 is _ abundant _ if it has full rank and every @xmath135 submatrix obtained by removing a pair of columns from @xmath117 still has full rank . [ thm : eqn_countnox ]
let @xmath121 be a @xmath118 linear system with @xmath117 abundant .
then the number of solution - free subsets of @xmath53 is @xmath131 . here @xmath136 as @xmath137 , with @xmath117 fixed in the case @xmath138 $ ] .
for example , take @xmath139 and @xmath140 .
theorem [ thm : eqn_countnox ] says that the number of sum - free subsets of @xmath68 $ ] is @xmath141 , giving a new proof of the weak form of the cameron and erds conjecture , proved independently by alon @xcite , by calkin @xcite and by erds and granville ( unpublished ) .
similar results hold when @xmath53 is an abelian group . for the proof of theorem [ thm : eqn_countnox ]
we need a result for equations analogous to the supersaturation theorem for graphs : this is the removal lemma of krl , serra and vena @xcite .
more information is given in [ sec : sumfree ] .
for linear systems where @xmath142 , theorem [ thm : eqn_countnox ] is uninformative .
one of the most prominent examples is that of sidon sets . a set @xmath143 $ ] is _ sidon
_ if every sum of two elements is distinct , i.e. , there are no solutions to @xmath144 with @xmath145 .
it is easy to see that a sidon set has size at most @xmath146 , since each of the @xmath147 values @xmath148 , where @xmath149 and @xmath150 , are distinct and lie in @xmath151 .
erds and turn @xcite improved this upper bound to @xmath152 , and there are examples achieving this bound .
it is natural to ask , as cameron and erds did @xcite , how many sidon sets there are , and the answer clearly lies between @xmath153 and @xmath154 .
neither of these bounds , it turns out , is tight .
[ thm : sidon ] there are between @xmath155 and @xmath156 sidon subsets of @xmath36 $ ] . the lower bound gives a negative answer to the open question of whether there are only @xmath153 sidon sets .
the upper bound , also proved by kohayakawa , lee , rdl and samotij @xcite ( in fact with a better constant ) , follows directly by plugging in the appropriate numbers into a container - counting theorem . for details
see [ sec : sidon ] . in recent times , there has been interest in the extent to which theorems holding for dense structures hold also for sparse random substructures .
our results can be applied in this context , and we give some illustrative examples involving the notions of @xmath8-free graphs and solution - free subsets already discussed .
the application of our results always fits a simple paradigm .
typically we want some statement to hold for a random substructure , with high probability ; by considering an appropriate collection of containers , the fact that there is a small number of containers means that the work is reduced , via the union bound , to establishing a ( generally much simpler ) statement for a single container .
for example , consider a random @xmath7-graph @xmath109 , as defined in [ subsec : hfree ] .
evidently there are @xmath8-free subgraphs of @xmath109 with @xmath157 edges , but are there significantly larger @xmath8-free subgraphs ?
kohayakawa , uczak and rdl @xcite conjectured that if @xmath158 then @xmath8-free subgraphs of @xmath109 almost surely have at most @xmath159 edges .
this conjecture was recently proved by conlon and gowers @xcite ( for strictly balanced @xmath8 ) and by schacht @xcite , using different methods .
our methods give an alternative proof .
for each container @xmath26 given by theorem [ thm : ffree_cover ] , it is easily seen that with high probability @xmath109 contains not much more than @xmath160 edges of @xmath85 , and by the union bound this holds for all @xmath26 , and hence also for all @xmath8-free @xmath7-graphs .
[ thm : ffree_sparse ] let @xmath8 be an @xmath7-graph and let @xmath161 . for some @xmath76 , for @xmath77
sufficiently large and for @xmath162 , the following event holds with probability greater than @xmath163 : @xmath164 other related conjectures , including what has become known as the kr conjecture , were made in @xcite , not all of which have previously been proved in full , but they all follow from theorem [ thm : ffree_cover ] in a similar way .
details are in [ sec : sparse ] .
the same arguments can be applied to solution sets of linear equations .
here is a typical consequence .
[ thm : szem_sparse ] let @xmath165 and @xmath75 .
there exists a constant @xmath76 such that for @xmath166 , if @xmath167 $ ] is a random subset chosen with probability @xmath111 , then with probability tending to @xmath30 as @xmath168 , any subset of @xmath169 of size @xmath170 contains an arithmetic progression of length @xmath7 .
further examples and details can be found in [ sec : sparse ] .
a couple of simple notions are needed for the statement of the main theorem , and we define these now .
they are the co - degree function and degree measure .
the present results about containers were originally motivated by the study of the list chromatic number of simple hypergraphs , described in [ sec : list ] .
the main difficulties in the construction of containers are already present in the simple case .
however the method can be adapted efficiently to any hypergraph .
the size and number of the containers depends on the way the edges overlap , but the dependence can be encapsulated by a single parameter , which is usually quite straightforward to compute .
this parameter appears in most of the theorems .
first , we define the degree of a subset of vertices , in the natural way .
the _ degree _ of set of vertices @xmath172 is the number of edges containing @xmath173 ; that is , @xmath174 if @xmath175 , that is @xmath176 where @xmath45 , we generally write @xmath177 instead of @xmath178
. we can now define the co - degree function @xmath171 .
this is a function of the parameter @xmath179 , a parameter used in the construction of containers .
[ defn : delta ] let @xmath1 be an @xmath0-graph of order @xmath10 and average degree @xmath3 .
let @xmath180 . given @xmath45 and @xmath181 ,
let @xmath182 if @xmath183 we define @xmath184 by the equation @xmath185 then the _ co - degree function _
@xmath171 is defined by @xmath186 if @xmath187 we define @xmath188 .
there is nothing significant about the powers of 2 in the definition ; they are just constants needed for lemma [ lem : maxdeg ] .
[ rem : delta ] the parameter @xmath179 appears in the main theorem , theorem [ thm : cover ] , and the smaller that @xmath179 can be made the stronger the result becomes .
the constraint on @xmath179 comes from a lower bound on @xmath171 .
it can be seen that if @xmath179 decreases then the values of the @xmath184 increase , and hence so does @xmath171 ; indeed @xmath189 as @xmath190 .
a typical application will have @xmath179 as small as possible subject to @xmath171 being less than some constant , say @xmath191 . here
are some observations intended to indicate the optimal size of @xmath179 .
observe first that , unless @xmath1 has isolated vertices , @xmath192 holds for all @xmath47 , and so @xmath193 .
the largest of these bounds is @xmath194 ( @xmath179 is invariably less than one ) and so , for fixed @xmath0 and large @xmath3 , it will always be that for @xmath171 to be small we must choose @xmath179 at least as large as @xmath195 . in a simple hypergraph
, @xmath196 holds whenever @xmath197 , and so @xmath198 . in this case the largest of the @xmath184 s is @xmath199 , and we can make @xmath171 small by choosing @xmath179 just a little larger than @xmath195 .
in fact , for any hypergraph whose edges are uniformly distributed , @xmath199 is once again the @xmath184 which dominates , as a simple calculation ( which we omit ) shows , so here again @xmath171 is small if @xmath179 is larger than @xmath195 .
sometimes , though , the dominant @xmath184 is not @xmath199 .
one example of this is in the case of sidon sets : when @xmath200 it is the value of @xmath201 which is the most important .
another example is the hypergraph describing @xmath8-free @xmath7-graphs : here the most important @xmath184 is determined by which subgraph @xmath202 achieves the maximum of @xmath203 , and this is how @xmath13 enters in .
but in each of our examples the values are easily checked . in summary
, we must always choose @xmath204 , and for simple or uniformly distributed hypergraphs the value need not be much larger .
but there are applications which are far from uniformly distributed , where @xmath179 needs to be larger and where the behaviour of @xmath171 will prove crucial .
we mentioned in the introduction that the containers must not be too large . for a substantial number of applications it suffices that @xmath205 for some constant @xmath206
this is achievable for regular hypergraphs but it clearly is unattainable in general ; for example , if @xmath207 ( which , for large @xmath10 , has average degree close to @xmath208 ) then some container must have size at least @xmath209 .
other applications require that the number of edges inside a container , that is , @xmath210)$ ] , is small .
this is attainable in general but there are applications where what matters is @xmath211 rather than @xmath210)$ ] .
in fact we measure the size of containers by what we call _ degree measure_. it turns out that if the degree measure is bounded then it is possible to recover all the properties of containers that are needed .
[ def : mu ] let @xmath1 be an @xmath0-graph of order @xmath10 and average degree @xmath3 .
let @xmath212 .
the _ degree measure _ @xmath213 of @xmath43 is defined by @xmath214 thus @xmath215 is a probability measure on @xmath15 .
note that if @xmath1 is regular then @xmath216 , which is the _ uniform _ measure of @xmath43 .
the essential idea for demonstrating that a collection of containers is small is this : each container is specified by just a small set of vertices , meaning that , given some small set @xmath217 , there is a construction which produces another larger set @xmath83 , and the construction is such that for any independent set @xmath25 there is some @xmath218 which produces a @xmath85 with @xmath27 .
if @xmath89 is small , the number of choices for @xmath89 is not large , and so the number of containers is not large . actually , we shall generate @xmath85 from not just one small set but an @xmath0-tuple @xmath219 of small sets ; the principle is the same .
we already introduced the parameter @xmath179 .
essentially , @xmath179 will be the value of @xmath220 , which is why we want @xmath179 to be as small as possible .
in fact , the theorem guarantees @xmath221 where @xmath222 is some small constant at our disposal .
often we shall take @xmath223 but sometimes it is useful to choose a smaller value .
we use one more piece of shorthand .
let @xmath224 $ ] and let @xmath225 $ ] .
then we define @xmath226\,=\ , ( t_{r-1}\cap [ w],\ldots , t_1\cap [ w],t_0\cap [ w ] ) .
$ ] [ thm : cover ] let @xmath1 be an @xmath0-graph with vertex set @xmath36 $ ] , where @xmath177 decreases with @xmath47 . let @xmath227 satisfy @xmath228 . then there is a function @xmath229\to\mathcal{p}[n]$ ] , such that , for every independent set @xmath230 $ ] there exists @xmath231 with * @xmath232 , * @xmath233 , * @xmath234 , and * @xmath235 .
moreover @xmath85 has the online property , meaning that @xmath236=c(t\cap[w])\cap[w]$ ] for all @xmath237 $ ] and @xmath225 $ ] .
in fact , the above is true for all sets @xmath230 $ ] for which either @xmath238 $ ] is @xmath239-degenerate or @xmath240)\le 2r \tau^re(g)/ \zeta$ ] . [ rem : cover ] it is worth making a few observations at this point . * roughly speaking , the theorem says that for each @xmath25 there exists @xmath218 with @xmath241 , @xmath232 and @xmath242 , provided @xmath179 is large enough to make @xmath171 small . *
the online property is needed only for certain applications , principally theorem [ thm : uniform ] .
more is said about this in [ subsec : mainproof ] . *
ordering the vertices by degrees is generally unnecessary : it is used only to obtain ( c ) and the online property simultaneously , and to accommodate @xmath243-degenerate graphs . *
the container construction method makes essentially no use of the independence of the sets @xmath25 , so we include an extension to two kinds of sparse subset , where either @xmath238 $ ] is @xmath243-degenerate or @xmath240)\le bn$ ] , and @xmath243 is small . as usual , we say @xmath238 $ ] is _ @xmath243-degenerate _ if for every subset @xmath244 there is a vertex in @xmath245 $ ] of degree at most @xmath243 .
this is equivalent to saying that every subgraph @xmath245 $ ] is sparse ; indeed if @xmath238 $ ] is @xmath243-degenerate then @xmath245 $ ] has at most @xmath246 edges for all @xmath244 , and conversely if @xmath245 $ ] has at most @xmath246 edges for all @xmath244 then @xmath238 $ ] is @xmath247-degenerate . *
the theorem is best possible in two senses .
first , the examples in [ sec : largecont ] show that our method can not give @xmath248 .
secondly , the bound on the measure of the generating sets @xmath249 , implicit in the constraint @xmath228 which determines how small @xmath179 can be , can not be improved significantly .
one way to see this is that an improvement here would give an improvement in some of the applications , say theorem [ thm : ffree_sparse ] , but these are known to be best possible .
as mentioned in the introduction , theorem [ thm : cover ] has a variety of consequences and weaker forms which are easier to apply directly .
these are discussed in [ sec : iteration][sec : uniform ] , but we state a couple of them here for illustration . in [ sec :
iteration ] we give a number of consequences of theorem [ thm : cover ] concerning the number of edges @xmath210)$ ] inside the container .
the following is the weakest of these ; it provides , in a handy format , a collection of containers each with few internal edges , the size of the collection being bounded by a simple function of @xmath179 .
[ cor : sparse_container ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] .
let @xmath250 .
suppose that @xmath179 satisfies @xmath251 and @xmath252 .
then there exists a constant @xmath253 and a collection @xmath254 $ ] such that * for every independent set @xmath25 there exists @xmath255 with @xmath256 , @xmath257 and @xmath258 , * @xmath210 ) \le \epsilon e(g)$ ] for all @xmath259 ,
* @xmath260 .
moreover , ( a ) holds for all sets @xmath230 $ ] for which either @xmath238 $ ] is @xmath261-degenerate or @xmath240)\le 24 \epsilon r!r\tau^r e(g)$ ] .
next we give a consequence for applications when the size @xmath211 of the container is of interest . as mentioned earlier ,
examples such as @xmath262 show that it is not possible always to guarantee that @xmath211 is bounded .
however , it is possible to bound @xmath263|$ ] for some initial segment @xmath264 $ ] of the vertex set @xmath36 $ ] , and this can be done so that the number of different sets @xmath263|$ ] is small ( a function of @xmath47 rather than of @xmath10 ) .
more exactly , each container @xmath85 `` nominates '' @xmath265 , so that @xmath263|$ ] is small and not many containers nominate any given @xmath47 .
this consequence of the main result is suitable for dealing with list colourings , for example .
the statement is actually in terms of tuples @xmath266 of containers rather than individual containers , since this is ultimately more efficient .
the theorem is explained more in [ sec : uniform ] .
[ thm : uniform ]
let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] , for which the degree sequence is decreasing .
let @xmath267 .
suppose that @xmath268 , that @xmath269 , and that @xmath270 $ ] satisfies @xmath271)\le \zeta/2r!$ ] .
let @xmath272 .
then there exists a collection @xmath273 $ ] and a map @xmath274 $ ] , with the following properties : * for all independent sets @xmath25 there is some @xmath26 with @xmath27 , * for all @xmath275 $ ] @xmath276,\ldots , c_t\cap[v])\,:\,g(c_1,\ldots , c_t)=v\ , \}\right|\,\le\,\zeta^{-2 } vtr\tau\log(1/\tau),\ ] ] * and for all @xmath277 @xmath278|\,\le\,(1-\frac{1}{r!}+8\zeta)v\,,\ ] ] where @xmath279 .
moreover , ( a ) holds for all sets @xmath230 $ ] for which either @xmath238 $ ] is @xmath280-degenerate or @xmath240)\le 2 r\tau^re(g)/ \zeta$ ] .
in this section we describe the method of building containers and establish the basic facts about them .
sapozhenko @xcite gives a way to build containers in the case of ordinary ( 2-uniform ) graphs that are close to being regular .
however even for graphs of this kind there is more than one approach which works .
it is not obvious how to extend these graph methods to hypergraphs , each method offering a few plausible possibilities , most of which fail but some of which succeed .
indeed we eventually found more than one way to construct containers for regular 3-graphs , and one of these , which extended to regular @xmath0-graphs , was described in @xcite : it is a random - based method .
but we have found it difficult to find a method which fulfils the goals of working for general ( not just regular ) @xmath0-graphs and which also has the necessary online property .
the method of @xcite would be enough for some applications , such as list colouring , if it were true that each @xmath0-graph of average degree @xmath3 contained a subgraph with degrees in the range @xmath281 to @xmath282 , for some constants @xmath283 , @xmath284 and @xmath285 .
but it is possible to construct examples of @xmath0-graphs where this is far from the truth ( these examples bear some similarity to the @xmath44-graphs of pyber , rdl and szemerdi @xcite ) .
the method given below fulfils all our requirements .
the reason for the remarks just made is that , whilst we do attempt to motivate the construction , we can not do so fully because it is hard to explain why one construction should work when another similar one does not .
the devil is sometimes in the detail , and a prolonged discussion would be unjustified .
there are two aspects to the construction : the production of a suitable small set @xmath89 from a set @xmath25 , which we call _ pruning _ , and the production of @xmath85 from @xmath89 , which we call _ building_. ( in fact , as mentioned already , we produce not just one small set but an @xmath0-tuple @xmath219 of small sets . )
these two aspects are very closely intertwined , and it is convenient to describe both in terms of a single algorithm which has two slightly differing modes , a _ build mode _ and a _ prune mode_. given an @xmath0-graph @xmath1 with vertex set @xmath36 $ ] , and a sequence of subsets @xmath286 of @xmath36 $ ] , we produce a sequence of @xmath287-_multigraphs _
@xmath288 for @xmath289 .
this means that @xmath288 is @xmath287-uniform but multiple edges are allowed ; in other words @xmath290 is a _
multiset_. each edge @xmath291 with @xmath292 will come from an edge @xmath293 , where @xmath294 and @xmath295 , @xmath296 .
equivalently , each edge of @xmath288 is an edge of @xmath297 whose first vertex , which is in @xmath298 , has been removed .
the reason @xmath288 is defined as a multigraph , even if @xmath1 itself does not have multiple edges , is so that distinct edges of @xmath1 give rise to distinct edges of @xmath288 .
the multigraph @xmath299 is @xmath30-uniform : its edges are single vertices .
if @xmath25 is an independent set and the sets @xmath300 are chosen within @xmath25 , as they will be , then evidently the members of @xmath301 can not be vertices in @xmath25 , and so the container @xmath85 can be chosen from vertices not in @xmath301 .
our first aim , then , is to ensure that @xmath301 is as large as possible , and to this end we attempt to make @xmath290 large for each @xmath287 .
however this aim has to be balanced against keeping the sets @xmath298 small .
hence we shall choose a parameter @xmath179 , so that , roughly speaking , @xmath298 will comprise a proportion @xmath179 of the vertices ( in degree measure ) , and we might hope the size of @xmath290 to be roughly @xmath179 times the size of @xmath302 .
this means the average degree of @xmath288 will typically be around @xmath303 .
the parameter @xmath179 is the same as that discussed in [ subsec : codeg ] and the constraint @xmath204 described there is precisely what is needed to ensure that @xmath301 contains something worthwhile .
however not every edge of @xmath1 with its first @xmath304 vertices in @xmath305 will be admitted as an edge of @xmath288 , but only a selection of these .
we do not allow edges into @xmath288 if they increase the degree of some vertex , or the degree of some subset @xmath306 $ ] , beyond some agreed threshold .
we define the degree of @xmath173 in the multigraph @xmath288 to be @xmath307 where we note that here we are counting edges with multiplicity in the multiset @xmath290 .
( naturally we may write @xmath308 instead of @xmath309 if @xmath275 $ ] . )
there are several reasons for wanting to bound the degrees in @xmath288 .
one reason is the hope of keeping the vertex degrees near to @xmath310 times the degrees in @xmath1 , so that degree measure in @xmath288 relates to measure in @xmath1 ; in particular , small sets of vertices can not account for most of the edges of @xmath288 unless those sets have large measure in @xmath1 .
a second reason for controlling degrees of subsets is that only by doing so can we restrain the degrees of vertices at later stages : this comes out in the proof of lemma [ lem : maxdeg ] .
so we proceed in the following way .
we begin with @xmath311 , and then apply the straightforward algorithm below to construct @xmath288 from @xmath297 using @xmath298 , with @xmath287 taking the values @xmath312 in turn . during the application of the algorithm , the degrees @xmath313 in @xmath288 will grow , as edges are added .
we denote by @xmath314 the collection of vertices and subsets whose degrees have reached their bound , and we do not permit the addition to @xmath288 of any edge which contains a current member of @xmath314 .
the set @xmath314 will grow too during the construction .
as mentioned before , the algorithm can be run in two modes : prune mode and build mode .
the aim of successive runs of prune mode is to produce sets @xmath315 , and the aim of build mode is to produce sets @xmath316 which will form the basis of a container .
thus in a single run of prune mode , an independent set @xmath25 is input and some set @xmath298 is output , whereas in build mode a set @xmath298 is input and a set @xmath317 is output .
both forms of the algorithm include the construction of @xmath288 from @xmath297 .
the essential difference between the modes is this .
the algorithm inspects the vertices in @xmath36 $ ] one by one . in prune mode , where @xmath25 is part of the input ,
if the vertex is in @xmath25 a decision is made , according to some rule , whether to place the vertex into @xmath298 . in build mode , where @xmath298 is input but @xmath25 is not , the vertex is inspected to see whether it passes the rule ; if it passes , but the vertex is not in @xmath298 , we know it could not have been in @xmath25 , so it is removed from @xmath317 ( with the aim of making a container via @xmath317 ) . in both modes , if the vertex is in @xmath298 , appropriate edges are added to @xmath288 .
two real numbers are included in the input to the algorithm .
the parameter @xmath179 is the more important and has already been discussed .
the parameter @xmath222 is a small constant , often in practice chosen to be @xmath318 but sometimes chosen smaller .
it is used in the rule to decide membership of @xmath298 .
the independence of the set @xmath25 is not actually used by the algorithm , and it is useful to define the algorithm for general subsets @xmath230 $ ] . lgorithm + nput n @xmath0-graph @xmath1 on vertex set @xmath36 $ ] + an @xmath319-multigraph @xmath297 on vertex set @xmath36 $ ] + parameters @xmath320 + = a subset @xmath321 $ ] + a subset @xmath322 $ ] + + an @xmath287-multigraph @xmath288 on vertex set @xmath36 $ ] + a subset @xmath323 $ ] + a subset @xmath324 $ ] + + put @xmath325 and @xmath326 + = put @xmath327 + put @xmath328 $ ] + + for @xmath329 do : + let @xmath330^{(s ) } : \{v\}\cup f\in e(p_{s+1}),\ , \mbox { and } \forall \sigma\in \gamma_s\,\ , \sigma\not\subset f\,\}$ ] + + = if @xmath331 and @xmath332 , add @xmath47 to @xmath298 + if @xmath333 , remove @xmath47 from @xmath317 + if @xmath334 then + = add @xmath53 to @xmath290 + for each @xmath335$],= if @xmath336 , add @xmath337 to @xmath314 + for each @xmath338^{(>1)}$ ] , if @xmath339 , add @xmath173 to @xmath314 the algorithm adds to @xmath288 all @xmath287-edges which , with @xmath340 as first vertex , form an edge of @xmath297 and which do not contain ( at that moment ) any subset in @xmath314 .
the degree threshold for a vertex entering @xmath314 is in terms of its degree @xmath341 in the original graph @xmath1 , whereas for a larger subset @xmath173 it is in terms of its degree in @xmath297 ; this difference is for technical reasons arising in the proof of theorem [ thm : cover ] .
clearly , if in prune mode we construct @xmath298 from @xmath25 , then , when in build mode with the same set @xmath298 as input , the condition @xmath331 can happen only if either @xmath342 or @xmath343 .
evidently , therefore , @xmath344 , so @xmath345 is one option for a container for @xmath25 .
another option for a container , as mentioned earlier , is @xmath36 $ ] minus the vertices in @xmath301 . in particular ,
if @xmath25 is independent then @xmath346-\gamma_1 $ ] .
( strictly speaking , the set @xmath347 is a set of singletons of vertices rather than a set of vertices , but we identify @xmath347 with @xmath348:\{u\}\in\gamma_1\}$ ] .
furthermore @xmath347 is not output by the algorithm , but it is easily recoverable from @xmath299 .
indeed , @xmath301 is just a multiset of vertices , and @xmath349 is the multiplicity of @xmath350 in @xmath301 . by construction , @xmath351:d_1(u)>\tau^{r-1}d(u)\}$ ] . )
so each of @xmath345 , @xmath352 , and @xmath36-\gamma_1 $ ] is a container for @xmath25 ; our aim is to ensure that at least one of these is a good container , meaning that is not close to @xmath36 $ ] . here then is a way of viewing the operation of the algorithm . if @xmath347 is large then @xmath36-\gamma_1 $ ] is a good container for @xmath25 .
if @xmath347 is not large then , since the degrees in @xmath299 are bounded , the average degree of @xmath299 must be small . but
@xmath311 , whose average degree is not small , so there must be some @xmath287 for which @xmath297 has large average degree ( of order @xmath353 ) but @xmath288 has small average degree ( much smaller than @xmath303 ) . since the degrees are bounded , there must have been plenty of vertices of @xmath297 which could have contributed edges to @xmath290 but did not do so .
why did they not do so ?
only because they are not in @xmath25 and so not available for @xmath298 .
these are exactly the vertices which are removed from @xmath317 : hence for this value of @xmath287 , @xmath345 will be a good container for @xmath25 .
we are thus lead to two important definitions .
[ def : ttuple ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] and let @xmath230 $ ] .
let @xmath320 .
let @xmath315 be the sets constructed by repeated applications of the algorithm in prune mode .
let @xmath354 .
then we define @xmath355 the @xmath0-tuple @xmath89 is the fruit of running the algorithm in prune mode , from which the container for @xmath25 will be built . as noted earlier , if @xmath25 is an independent set then @xmath356 .
hence the introduction of @xmath357 is unnecessary if we wish to find containers only for independent sets , but by introducing @xmath357 we can produce containers for non - independent sets too .
it will turn out that @xmath357 is small , as desired , provided @xmath25 is sparse ; see [ subsec : mut ] . now comes the main definition that of containers .
[ def : container ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] and let @xmath224 $ ] .
let @xmath227 .
let @xmath316 be constructed by repeated applications of the algorithm in build mode , using @xmath315 .
let @xmath358-\gamma_1 $ ] .
the _ container _ @xmath359 is then @xmath360 [ lem : container ] if @xmath361 then @xmath362 .
we noted earlier that @xmath344 for @xmath363 .
moreover , @xmath364 by definition , since @xmath358-\gamma_1 $ ] and @xmath354 .
hence @xmath365 . before computing the size of the containers @xmath359 and the number of them , we make note of their online property , namely that @xmath366 $ ] is determined just by @xmath367 $ ] .
[ lem : online ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] and let @xmath237 $ ] .
then , for each @xmath368 $ ] , @xmath369=c(g , t\cap[w],\tau,\zeta)\cap[w]$ ] holds .
a little reflection on the algorithm makes the lemma clear .
suppose that @xmath370 $ ] is given , together with the multiset of edges @xmath371 such that @xmath372\ne\emptyset$ ] .
then the algorithm , run with @xmath373 rather than @xmath374 , will correctly produce the edges @xmath375 such that @xmath372\ne\emptyset$ ] and will find all @xmath376 such that @xmath377\ne\emptyset$ ] ( together with some other members of @xmath290 and @xmath314 that we shall not need ) .
moreover , when running in build mode , this restricted version of the algorithm will correctly determine @xmath378 $ ] .
therefore , by running the restricted algorithm for @xmath379 , we can determine @xmath380,\ldots , c_1\cap[w]$ ] .
finally , since we have determined @xmath381 $ ] we can find @xmath382 $ ] , and because we are given @xmath383 $ ] this means we know @xmath384 $ ] .
in this section we estimate the measure of the tuples @xmath385 and of the containers @xmath359 , thereby proving theorem [ thm : cover ] . before making these estimates
we need information on how large the degrees can be in @xmath288 .
the intention behind the set @xmath314 is to prevent degrees being much larger than the target degrees , namely @xmath386 for the vertex @xmath350 ; after the degree of @xmath350 attains this level , no further edges containing @xmath350 are added to @xmath288 . however ,
when a vertex @xmath350 enters @xmath314 , it does so because some multiset @xmath53 has been added to @xmath290 . since @xmath53 can include many edges that contain @xmath350 , the degree @xmath387 can increase significantly in one step , from an initial value at most the target value @xmath386 to something much larger .
the extent of this problem depends ultimately on the way the edges of @xmath1 overlap each other .
the reason @xmath314 is defined the way it is in the algorithm , is to keep control of the degree problem without increasing @xmath179 more than is necessary .
this can be expressed succinctly in terms of the co - degree function @xmath171 introduced in [ subsec : codeg ] .
first we need a small calculation .
[ lem : asj ] for @xmath388 and @xmath389 , let @xmath390 be given by the equations @xmath391 and @xmath392 for @xmath393 , where @xmath184 was defined in definition [ defn : delta ]
. then @xmath394 holds for @xmath395 . since @xmath396 , it is enough to prove that @xmath397 .
now by dint of the definition it is clear that @xmath390 is a linear combination of the numbers @xmath398 , @xmath399 .
we claim that the coefficient of @xmath398 in @xmath390 is at most @xmath400 .
this is certainly true if @xmath401 , since the only positive coefficient is that of @xmath184 ( i.e. @xmath402 ) .
for @xmath393 we may prove the claim on the assumption that it is true for @xmath403 . if @xmath402 then the coefficient of @xmath398 in @xmath404 is zero , and the claim follows because @xmath405 . if @xmath406 we have @xmath407\le 2^{\binom{r}{2}-\binom{s+\ell}{2}+\ell}\ ] ] and the claim follows in this case too .
hence the claim always holds , and so @xmath408 by definition of @xmath171 .
here is the main lemma about degrees in @xmath288 , explaining the role of the co - degree function @xmath171 .
[ lem : maxdeg ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] with average degree @xmath3 .
let @xmath311 and let @xmath409 be the multigraphs constructed by successive applications of the algorithm .
then @xmath410 holds for all subsets @xmath411 $ ] and for @xmath412 . by analogy with definition [ defn : delta ]
we define @xmath413^{(j)}\,\}\,,\ ] ] for @xmath414 , where here it is the final values of these quantities that are used that is , we measure these quantities in the output multigraph @xmath288 .
when @xmath401 the lemma is true by definition of @xmath415 , so from now on we assume @xmath416 .
suppose that @xmath417^{(j)}$ ] .
if @xmath418 then @xmath419 . if @xmath420 then @xmath173 was added to @xmath314 after some vertex @xmath340 was inspected and @xmath53 was added to @xmath290 . before this took place , @xmath421 held ; since the number of edges of @xmath53 containing @xmath173 was at most @xmath422 , the final value of @xmath313 satisfies @xmath423
. this inequality holds for all @xmath417^{(j)}$ ] .
let @xmath424 $ ] ; then @xmath425 for some @xmath417^{(j)}$ ] , so @xmath426 we claim that @xmath427 } d_s^{(j)}(u)\le a_s^{(j)}\tau^{r - s+j-1}nd\,,\ ] ] where @xmath390 was defined in lemma [ lem : asj ] . indeed , for @xmath401 the claim ( with equality ) is just the definition of @xmath184 , and for @xmath416 it follows immediately by induction ( on @xmath304 ) from inequality ( [ eqn : dsj ] ) and the definition of @xmath390 .
hence , for @xmath428 , we have by lemma [ lem : asj ] @xmath427 } d_{s+1}^{(2)}(u)\le 4^{1-s}\tau^{r - s}nd\,\delta(g,\tau)\ , .
\label{eqn : ds2}\ ] ] now let @xmath429 . if @xmath430 then @xmath431 . if @xmath432 then @xmath350 was added to @xmath314 after some vertex @xmath340 was inspected and @xmath53 was added to @xmath290 . since @xmath53 has at most @xmath433 edges containing @xmath350 , the degree of @xmath350 is at most @xmath434 .
now @xmath435 so , using ( [ eqn : ds2 ] ) , we have @xmath436 which establishes the lemma .
we now estimate the measures of the sets @xmath298 .
[ lem : muts ] let @xmath230 $ ] and @xmath437 .
then @xmath438 for @xmath352 .
the set @xmath298 is output when the algorithm is run in prune mode . during the run of the algorithm , each vertex @xmath47 which enters @xmath298 contributes a set @xmath53 of at least @xmath439 edges to @xmath290 .
therefore , writing @xmath3 for the average degree of @xmath1 , lemma [ lem : maxdeg ] yields @xmath440}d_s(u)\\\le \tau^{r - s}nd(1 + 4^{1-s}\delta(g,\tau))\end{aligned}\ ] ] and this proves the lemma .
the set @xmath357 needs a different argument . as noted before , @xmath441 if @xmath25 is independent .
it turns out that we can allow @xmath25 to be @xmath243-degenerate , for fixed @xmath243 , at essentially no cost compared with @xmath25 being independent , though to handle this case we shall need to make sure the degree sequence of @xmath1 is decreasing . on the other hand , requiring only that @xmath240)\le bn$ ] will incur a small cost , which is that @xmath179 must increase typically from around @xmath195 to around @xmath442 in order to keep @xmath357 small ( @xmath3 being the average degree of @xmath1 ) .
[ lem : mut0 ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] with average degree @xmath3 .
let @xmath230 $ ] and @xmath437 . if @xmath240 ) \le bn$ ] then @xmath443 .
if @xmath238 $ ] is @xmath243-degenerate and @xmath444 for @xmath445 $ ] , then @xmath446)\le ( \tau/\zeta)(1+\delta(g,\tau))$ ] .
recall that @xmath447 , so @xmath448 for each @xmath449 .
let @xmath450 . recall that distinct @xmath30-edges @xmath451 correspond to distinct @xmath0-edges @xmath452 with @xmath453 and @xmath454 , so these edges lie in @xmath245 $ ] .
it follows that @xmath455)$ ] . if @xmath240)\le bn$ ] then we simply have @xmath456 ) \le e(g[i])\le bn$ ] , so @xmath443 .
if @xmath238 $ ] is @xmath243-degenerate then we consider @xmath457 instead , where @xmath458 denotes @xmath459 $ ] for any @xmath460 $ ] .
if @xmath461 then each of the above - mentioned edges @xmath462 lies within @xmath457 , since @xmath463 . therefore @xmath464)\le b|j^*|$ ] , by @xmath243-degeneracy .
then , by definition of @xmath465 $ ] , @xmath466 therefore @xmath467 , and so @xmath468 , from which the result follows via lemma [ lem : muts ] .
we now prove the crucial fact that the measure of the container @xmath359 is bounded above by some constant less than one .
this can be established with a fairly simple argument , but just a little more care yields a bound close to @xmath469 , and this is best possible , as shown in [ sec : largecont ] .
[ lem : mucont ] let @xmath470 $ ] .
then @xmath471 recall from definition [ def : container ] that @xmath358-\gamma_1 $ ] and that @xmath316 are constructed by the algorithm in build mode .
let @xmath472 .
we define @xmath473 \setminus c \\ d_2 & = \{v\in[n ] : \{v\}\in \gamma_2 , v\notin d_1 \}\\
d_3 & = \{v\in[n ] : \{v\}\in \gamma_3 , v\notin ( d_1 \cup d_2)\ } \\ & \ \vdots \\ d_{r-1 } & = \{v\in[n ] : \{v\}\in \gamma_{r-1 } , v\notin ( d_1 \cup \cdots \cup d_{r-2})\ } \\
d_r & = [ n ] \setminus ( d_1 \cup \cdots \cup d_{r-1})\,.\end{aligned}\ ] ] evidently , @xmath474 form a partition of @xmath36 $ ] .
since @xmath475 , it is enough to prove that @xmath476 .
it is convenient to define @xmath477 , @xmath478 and @xmath479\setminus d_{\le s}$ ] .
for @xmath395 we also need certain subsets of the edges of @xmath288 : @xmath480 we further define the numbers @xmath481 by @xmath482 , @xmath483 and @xmath484 , where @xmath3 is the average degree of @xmath1 .
observe that @xmath485 , that is , @xmath486 .
note that @xmath487 and each member of @xmath488 meets @xmath489 .
so @xmath490 where the last line employs lemma [ lem : maxdeg ] .
now @xmath491 for all @xmath492 : for @xmath393 this is because @xmath493 , and for @xmath401 it holds trivially .
hence @xmath494 , so we obtain @xmath495 . adding @xmath496 to each side
gives @xmath497 for each @xmath395 .
multiplying this inequality by @xmath498 and summing over @xmath499 , noting that @xmath500 , @xmath501 and @xmath502 , we obtain @xmath503 where we used @xmath504 .
suppose @xmath395 and @xmath505 .
if @xmath506 and @xmath507 then @xmath175 , say @xmath508 .
but @xmath509 implies @xmath510 by definition of @xmath511 , which contradicts @xmath512 .
thus @xmath513 contains no member of @xmath514 .
let @xmath47 be the first vertex of @xmath513 .
then @xmath515 since @xmath516 , so @xmath517 .
by the construction of @xmath518 , @xmath47 is the first vertex of fewer than @xmath519 edges of @xmath288 that contain no member of @xmath514 , so it is the first vertex of fewer than @xmath519 edges in @xmath520 .
therefore @xmath521 .
hence @xmath522 .
in particular @xmath523 , because @xmath524 . now let @xmath525 and put @xmath526 . by definition of @xmath527 ,
each member of @xmath527 contains a member of @xmath43 .
therefore @xmath528 hence @xmath529 for @xmath525 .
since @xmath530 this means @xmath531 ; by repeating and applying both @xmath502 and @xmath532 , this yields @xmath533 for @xmath525 .
the inequality holds for @xmath534 also .
substituting this inequality into inequality ( [ eqn : xy ] ) we obtain @xmath535 the coeffient of @xmath536 is negative , and so @xmath537 , which is what we needed to prove .
the simple and obvious choices for @xmath89 and @xmath85 in theorem [ thm : cover ] are @xmath361 and @xmath538 .
indeed this choice works perfectly well for parts ( a ) , ( b ) and ( d ) in the ( most important ) case when @xmath25 is independent , and also when @xmath240)$ ] is bounded .
however for the case when @xmath238 $ ] is @xmath243-degenerate a different choice is required .
we need containers determined by small sets of vertices none of which has low degree . the online property offers , as a by - product , a convenient way to achieve this aim , which furthermore has the side benefit of yielding ( c ) .
we label the vertex set so that the degrees are decreasing , and hence the vertices of small degree are specified by some terminal segment of @xmath36 $ ] .
we then run the algorithm just on the initial segment @xmath465 $ ] of vertices of high degree , and add the trailing segment @xmath539 $ ] to the container .
these modified containers , determined by subsets of @xmath465 $ ] , still have the online property .
[ lem : trunc ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] and let @xmath540 $ ] .
let @xmath230 $ ] , let @xmath541 and let @xmath542 $ ] .
put @xmath543\,.\ ] ] then @xmath85 is a container , determined by @xmath89 , with the online property : that is , @xmath27 , and @xmath544 $ ] is determined by @xmath367 $ ] for all @xmath225 $ ] .
the lemma is a simple consequence of lemma [ lem : online ] .
notice that the theorem is trivial if @xmath545 , since in that case the function @xmath546 $ ] works , with @xmath547 representing all @xmath25 .
recall too from [ subsec : codeg ] that @xmath189 as @xmath190 .
hence the condition @xmath228 is satisfiable by making @xmath179 large enough , although if @xmath548 the theorem is again trivial . let @xmath3 be the average degree of @xmath1 . we define @xmath540
$ ] by @xmath465=\{v\in [ n ] : d(v)\ge\zeta d\}$ ] .
let @xmath549 $ ] and write @xmath550 .
we then put @xmath542 $ ] , so @xmath551 where @xmath552 $ ] , @xmath553 .
finally , we define @xmath554 $ ] .
we shall show that the theorem holds with this choice of @xmath89 and @xmath85 .
first , note that ( a ) and the online property both hold by virtue of lemma [ lem : trunc ] . to show that ( b ) holds
, we apply lemmas [ lem : muts ] and [ lem : mut0 ] ; notice that the sets denoted by @xmath298 in those lemmas are here denoted by @xmath555 . for @xmath428
we obtain @xmath556 by lemma [ lem : muts ] because , as remarked , may assume @xmath557 . as for @xmath357 , if @xmath25 is independent then @xmath441 . if @xmath238 $ ] is @xmath243-degenerate then lemma [ lem : mut0 ] shows @xmath558)\le 2\tau/\zeta$ ] , because @xmath559 for @xmath445 $ ] .
if @xmath240)\le bn$ ] then lemma [ lem : mut0 ] shows @xmath560 .
hence ( b ) holds in every case .
since @xmath561 $ ] for each @xmath287 , we have @xmath562 , so ( c ) is a consequence of ( b ) .
now @xmath563)$ ] .
lemma [ lem : mucont ] applied to @xmath359 shows @xmath564 , because ( b ) holds . finally , note that @xmath565)<\zeta$ ] by definition of @xmath465 $ ] , so ( d ) holds and we are done .
the method of theorem [ thm : cover ] will not , in general , give containers of measure less than @xmath469 ( we shall give examples to show this in [ sec : largecont ] ) . on a more positive note ,
it is possible via iteration to obtain smaller , almost optimal , containers , at the cost of sacrificing the online property . some applications , most notably list colouring , depend crucially on the online property , but for many applications it is unnecessary , and in such cases it is generally well worthwhile using smaller , iterated , containers .
let @xmath1 be an @xmath0-graph of order @xmath10 and average degree @xmath3 , and let @xmath25 be an independent set in @xmath1 .
suppose we have a container @xmath85 for @xmath25 with @xmath566 , such as that supplied by theorem [ thm : cover ] where @xmath206 is close to @xmath191 .
it follows that the induced subgraph @xmath567 $ ] can not be dense ; to be precise , @xmath568)\le ( 1/r)\sum_{v\in c}d(v)=(1/r)nd\mu(c ) = \mu(c)e(g)\le ( 1-c)e(g)\ , .
\label{eqn : mutoe}\ ] ] theorem [ thm : cover ] can now be re - applied , this time to the hypergraph @xmath567 $ ] , to get a container @xmath569 , with @xmath570)\le ( 1-c)e(g[c])\le(1-c)^2e(g)$ ] .
repeating this operation often enough will give a container with @xmath571 edges .
the reason that the online property of theorem [ thm : cover ] is lost following iteration is that , in order to compute , say , @xmath572 $ ] , it is necessary to know the whole of @xmath567 $ ] , for which it is necessary to know the whole of @xmath85 , not just @xmath573 $ ] ; hence @xmath572 $ ] can not be computed online .
another drawback of iteration is the increase in the number of sets @xmath249 needed to specify the eventual container , leading to a greater number of possible containers .
usually in practice this turns out to be unimportant . here is a simple lemma to help count the number of containers being generated .
[ lem : entropy ] there are at most @xmath574 @xmath287-tuples of subsets @xmath575 $ ] with @xmath576 , where @xmath577 .
let there be @xmath578 such @xmath287-tuples with @xmath579 .
we wish to bound @xmath580 . plainly , the ordinary generating function @xmath581 for the sequence @xmath582 is equal to @xmath583 .
therefore , since @xmath584 , we have @xmath585 . the next theorem is a version of theorem [ thm : cover ] more suited to iteration .
[ thm : coveroff ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] .
suppose that @xmath586 and that @xmath587 .
then there exists a collection @xmath273 $ ] such that * for every independent set @xmath25 there exists @xmath588 with @xmath589 and @xmath590 , * @xmath591 , and * @xmath210 ) \le ( 1 - 1/2r!)e(g)$ ] for all @xmath26 . moreover , ( a ) holds for all sets @xmath230 $ ] for which either @xmath238 $ ] is @xmath592-degenerate or @xmath240)\le 24r!r\tau^re(g)$ ] .
we apply theorem [ thm : cover ] to @xmath1 with @xmath223 ( we may assume the degree order of @xmath1 is decreasing ) . for each set
@xmath25 we have @xmath551 and a container @xmath593 satisfying properties ( a)(d ) of that theorem .
take @xmath2 to be the collection of all such @xmath85 . since @xmath594 , we have @xmath595 , so @xmath596 .
it follows from inequality ( [ eqn : mutoe ] ) that @xmath210)\le ( 1 - 1/2r!)e(g)$ ] .
hence ( a ) and ( c ) of the present theorem are satisfied and it remains to check ( b ) .
theorem [ thm : cover ] tells us that each container @xmath85 is specified by sets @xmath597 each of size at most @xmath598 , where @xmath599 . by lemma [ lem : entropy ]
we have @xmath600 establishing ( b ) and completing the proof .
theorem [ thm : coveroff ] is easy to apply iteratively .
though a similar theorem can be derived using the proof method of theorem 3.1 of @xcite in place of theorem [ thm : cover ] , which actually gives a better bound in ( c ) , with a polynomial in @xmath0 in place of @xmath601 , the bound in ( b ) is worse , with @xmath602 in place of @xmath179 .
we remark , though , that it is not possible to iterate theorem 3.1 of @xcite directly , because it applies only to regular @xmath0-graphs , or , more precisely , to @xmath0-graphs with no large sparse subset . even if @xmath1 itself is regular , @xmath567 $ ] might be far from regular .
the next theorem is the result of applying theorem [ thm : coveroff ] repeatedly to end up with a collection of sparse containers .
its appearance is rather technical but it is in a form which can be readily applied .
[ thm : iteration ] let @xmath1 be an @xmath0-graph on vertex set @xmath36 $ ] .
let @xmath603 .
suppose that , for each @xmath604 $ ] with @xmath605 ) \ge e_0 $ ] , the function @xmath606 satisfies both @xmath607,\tau ) \le 1/12r!$ ] and @xmath608 .
for @xmath609 define @xmath610,\ , e(g[u ] ) \ge m \ } \\
\tau^ * & = \max\{\ , \tau(u ) : u \subset [ n],\ , e(g[u ] ) \ge e_0 \}\end{aligned}\ ] ] let @xmath611 .
then there exists a collection @xmath254 $ ] such that * for every independent set @xmath25 there exists @xmath612 with @xmath256 , @xmath613 and @xmath614 , * @xmath210 ) \le e_0 $ ] for all @xmath259 , * @xmath615
. moreover , ( a ) holds for all @xmath230 $ ] for which either @xmath238 $ ] is @xmath616)/12rr!|u|\rfloor$]-degenerate or @xmath240)\le 24r!r\tau(u)^re(g[u])$ ] , for all @xmath411 $ ] with @xmath605)\ge e_0 $ ] . we will show that for all @xmath617 with @xmath618 , there exists a collection @xmath619 $ ] satisfying conditions ( a)(c ) , where the constant @xmath620 has been replaced by @xmath617 in ( a)(c ) , and @xmath9 is replaced by @xmath621 .
when @xmath622 , we may take @xmath623\}$ ]
. otherwise , suppose @xmath624 .
it is enough to show that @xmath625 exists provided @xmath626 exists .
each @xmath627 is specified by a tuple @xmath628 with @xmath629 . if @xmath630 ) \le t$ ] , let @xmath631 .
otherwise , apply theorem [ thm : coveroff ] with @xmath632 to the @xmath0-graph @xmath633 $ ] , and let @xmath634 be the collection of containers given by the theorem .
then put @xmath635 .
if @xmath636 then @xmath85 is specified completely by @xmath637 , together with the @xmath0-tuple appearing in condition ( a ) of theorem [ thm : coveroff ] if the theorem was applied .
hence @xmath85 is specified completely by a tuple of size at most @xmath638 , so satisfying condition ( a ) . if @xmath639 then either @xmath630 ) \le t$ ] in which case @xmath640 , or @xmath630 ) > t$ ] in which case @xmath641 hence @xmath642 finally for @xmath643 , note that @xmath210 ) \le t$ ] , since if @xmath630 ) > t$ ] then by condition ( c ) of theorem [ thm : coveroff ] @xmath210 ) \le ( 1 - 1/2r ! ) e(g[d ] ) \le t$ ] . for certain applications
the technical detail of theorem [ thm : iteration ] is not needed ; what is required is a simple statement that a few iterations will produce a container with a negligible proportion of the original edges .
such a statement was presented earlier as corollary [ cor : sparse_container ] .
let @xmath644 .
observe that for @xmath604 $ ] , if @xmath645 then @xmath607,\tau ) \le \delta(g,\tau)/\epsilon \le 1/12r!$ ] .
therefore we may apply theorem [ thm : iteration ] to the graph @xmath1 with @xmath646 and @xmath647 for all @xmath648
. then @xmath649 and @xmath650 .
hence we obtain a collection @xmath2 satisfying conditions ( a ) and ( b ) of the corollary , and @xmath651 giving condition ( c ) .
theorem [ thm : cover ] provides containers of bounded degree measure , and [ sec : iteration ] provides containers with few edges inside . for some purposes ,
though , what is required is containers of bounded size , that is , of bounded uniform measure . for regular hypergraphs the results of [ sec : iteration ] can be used directly . if @xmath1 is @xmath3-regular and @xmath652 $ ] is such that @xmath210)=o(nd)$ ] then @xmath653 .
however for non - regular hypergraphs we get no such information .
indeed , it is perfectly possible to have @xmath654 and @xmath210)$ ] both small whilst @xmath211 is close to @xmath10 .
hence it is not possible to guarantee a container of bounded uniform measure . instead
, what we can show is that there is some initial interval @xmath264\subset[n]$ ] such that @xmath655|$ ] is bounded .
the basic lemma which translates information about @xmath215-measure into information about uniform measure is the following one . in the lemma
, @xmath43 is a multiset , so @xmath213 , @xmath656|$ ] and so on have their natural interpretations counting with multiplicities .
[ lem : lemmoid ] let @xmath657\to\mathbb r$ ] be a measure with @xmath658 , and let @xmath460 $ ] be a multiset
. then @xmath659\ , : \,|s\cap [ v]|\ge\alpha v\,\}\right ) \,\le\,\mu(s)\ ] ] holds for all @xmath660 .
let @xmath661|\ge\alpha v\}$ ] .
we must show @xmath662 .
let @xmath663 where @xmath664 and @xmath665 .
define the numbers @xmath666 by @xmath667|$ ] and @xmath668|$ ] for @xmath669 .
then we have @xmath670)\ge s_i\mu(w_i)$ ] , because @xmath671 . therefore @xmath672 ) + \mu(s\cap[w_1 + 1,w_2 ] ) + \cdots + \mu(s\cap[w_{k-1}+1,w_k ] ) \nonumber \\ & \,\ge\ , & s_1\mu(w_1)+s_2\mu(w_2)+\cdots+s_k\mu(w_k ) \nonumber \\ & \,=\ , & \sum_{i=1}^k \alpha\mu(w_i ) + ( s_1+\cdots+s_i - \alpha i)(\mu(w_i)-\mu(w_{i+1}))\ , , \nonumber \\ & \,=\ , & \alpha\mu(w ) + \sum_{i=1}^k ( s_1+\cdots+s_i - \alpha i)(\mu(w_i)-\mu(w_{i+1}))\ , , \label{eqn : lemmoid}\end{aligned}\ ] ] where @xmath673 is defined to be zero .
now @xmath674|=s_1+\cdots+s_i$ ] holds for @xmath675 , and so @xmath676 , because @xmath677 .
in particular , @xmath678 , since @xmath679 .
moreover @xmath215 is a decreasing function , so each summand in ( [ eqn : lemmoid ] ) is non - negative , and the lemma follows .
in fact we shall need not just that @xmath263|$ ] is bounded for a single container @xmath85 but that the average @xmath680|$ ] is bounded for a collection @xmath681 .
the next lemma prepares the way .
[ lem : g ] let @xmath215 be a probability measure on @xmath36 $ ] with @xmath682 .
let @xmath683 be subsets of @xmath36 $ ] , with @xmath684 for @xmath685 and @xmath686 for @xmath687 , where @xmath688 .
suppose moreover that @xmath270 $ ] and @xmath271)\le \eta c$ ] .
then there exists @xmath689 $ ] with @xmath690| < \frac{\lambda}{\eta}v\mbox{\qquad and \qquad } \frac{1}{t}\sum_{i=1}^t |c_i\cap[v]| < ( 1-c)v\,.\ ] ] let @xmath691| \ge s\lambda v/\eta\}$ ] . writing @xmath43 for the multiset which is the disjoint union of @xmath692 , so that @xmath693 and @xmath656| = \sum_{i=1}^s |t_i\cap[v]|$ ] , we can apply lemma [ lem : lemmoid ] with @xmath694 to obtain @xmath695 .
in like manner , let @xmath696| \ge t(1-c)v\}$ ] .
writing now @xmath43 for the multiset which is the disjoint union of @xmath681 , so that @xmath697 and @xmath656| = \sum_{i=1}^t |c_i\cap[v]|$ ] , we apply lemma [ lem : lemmoid ] with @xmath698 to obtain @xmath699 .
it follows that @xmath700 ) <
\eta + 1 -\eta/(1-c ) + \eta c < 1 $ ] , so there exists @xmath275 $ ] not contained in @xmath701 $ ] .
this @xmath47 satisfies the conditions of the corollary .
we can now prove the main result about containers and uniform measure , which was stated earlier as theorem [ thm : uniform ] .
apply theorem [ thm : cover ] to @xmath1 to obtain a collection @xmath2 of containers @xmath593 for @xmath470 $ ] . since @xmath269
we have @xmath702 so @xmath703 .
let @xmath704 , where @xmath272 .
each @xmath705 is specified by an @xmath0-tuple of sets @xmath300 , so the whole collection @xmath266 is specified by @xmath706 sets which , after re - labelling , we call @xmath707 , with @xmath221 for @xmath708
. let @xmath709 and @xmath710
. then @xmath271)\le\zeta/2r!<\eta c$ ] so the conditions of lemma [ lem : g ] are satisfied with @xmath711 and @xmath712 , and we may choose @xmath713 $ ] with @xmath690| < \frac{\tau}{\zeta^2}v\mbox{\qquad and \qquad } \frac{1}{t}\sum_{i=1}^t |c_i\cap[v]| < ( 1-\frac{1}{r!}+8\zeta)v\,.\ ] ] define @xmath714
. then ( a ) and ( c ) of the theorem are satisfied . to obtain ( b ) ,
we need that the containers have the online property : in other words , the @xmath617-tuple @xmath715,\ldots , c_t\cap[v])$ ] is determined by @xmath716,\ldots , t_s\cap[v]$ ] . this online property is guaranteed by theorem [ thm : cover ] .
hence the size of the set @xmath717,\ldots , c_t\cap[v]):g(c_1,\ldots , c_t)=v \}$ ] is bounded by the number of tuples @xmath718,\ldots , t_s\cap[v])$ ] .
now @xmath719| < s\theta v$ ] , where @xmath720 .
so by lemma [ lem : entropy ] @xmath721 which completes the proof .
theorem [ thm : cover ] provides a small collection of containers for independent sets in an @xmath0-graph , each container having degree measure at most @xmath469 ( plus a term that is usually small ) .
it is conceivable that an algorithm different to the one in [ sec : online ] might yield smaller online containers .
but in this section we describe examples to indicate that @xmath469 is the limit of the present method . indeed , for each @xmath722
, there are examples of simple @xmath0-uniform hypergraphs @xmath1 and independent sets @xmath25 for which @xmath723 can be as large as @xmath724 when @xmath361 .
because a precise detailed definition of the examples would be very long and opaque , we give instead a sketch showing how the examples work , together with a few words about how the details can be filled in .
( 7,1.7 ) ( 0,0 ) in a @xmath726-graph @xmath1.,title="fig : " ] ( 0.15,0.75)@xmath117 ( 1.6,1)@xmath727 ( 3.7,0.75)@xmath728 ( 5.3,1.3)@xmath729 ( 6.4,1.3)@xmath730 ( 3,0.1)@xmath1 ( 4.7,0.4)@xmath731 ( 5.9,0.3)@xmath732 ( 7.05,0.4)@xmath299 suppose @xmath1 is a simple @xmath3-regular @xmath726-graph with the following properties ( see figure [ fig : r4factorial ] ) . let the vertex set be @xmath733 , where @xmath734 , @xmath735 , @xmath736 , @xmath737 , @xmath738 , so @xmath739 and so on , where @xmath740 as @xmath741 . since @xmath1 is regular , the degree ordering used by theorem [ thm : cover ] is not determined by the 4-graph itself , so we may suppose we have an ordering with @xmath742 .
let the induced subgraph of @xmath1 on @xmath743 be approximately @xmath3-regular ( so that only a few edges meeting @xmath744 also meet @xmath117 ) , and suppose every edge has 3 vertices in @xmath727 and 1 vertex in @xmath745 ( the sizes of the sets @xmath746 have been chosen to ensure that this is possible ) .
suppose we run the algorithm with @xmath747 ( which is roughly the typical value of @xmath179 for a @xmath3-regular simple @xmath726-graph ) .
first the algorithm is run with @xmath748 .
after the vertices in @xmath117 have been inspected but before any vertex in @xmath727 has been inspected , the partially constructed subgraph of @xmath731 will contain @xmath58-edges that depend on @xmath25 and on the edges of @xmath1 that contain a vertex of @xmath117 . but suppose for now that we can choose @xmath749 so that at this point , @xmath731 contains a @xmath750-regular graph on @xmath745 , in which every edge has two vertices in @xmath728 and one vertex in @xmath751 ( see figure [ fig : r4factorial ] ) .
then since @xmath752 contains vertices with degree at least @xmath753 , all vertices in @xmath745 will be in @xmath752 .
therefore , as the algorithm for @xmath731 proceeds to inspect the vertices after @xmath117 , all vertices in @xmath727 will be kept in @xmath754 ( since , for each vertex @xmath755 , there are very few edges whose first vertex is @xmath47 that do not meet @xmath752 ) .
furthermore , all the vertices in @xmath756 will be kept in @xmath754 since they are not the first vertex of any @xmath726-edge in @xmath1 .
similarly , if we can arrange things so that when the algorithm is run with @xmath534 , @xmath757 contains a @xmath758-regular bipartite graph , and when the algorithm is run with @xmath759 , @xmath730 contains a @xmath30-regular @xmath30-graph , then the sets @xmath760 will be kept in @xmath761 and @xmath762 .
also @xmath347 contains no singletons from @xmath760 , so @xmath763 , and so @xmath764 as required .
this broad setup generalizes to all @xmath0 by taking sets @xmath765 of size @xmath766 , such that the induced graph of @xmath1 on @xmath767 is approximately @xmath3-regular with every edge having @xmath768 vertices in @xmath727 and @xmath30 vertex in @xmath769 . then provided that the @xmath288 are similarly generalized
, we will have @xmath770 and so @xmath771 .
in the example above , we did not specify how to guarantee that the graphs @xmath288 are as stated .
this is achieved by adding edges to @xmath1 containing vertices from @xmath117 and specifying @xmath772 appropriately ; we indicate how this may be achieved for @xmath725 .
suppose @xmath773 .
choose an equipartition @xmath774 of @xmath728 , and consider the 4 vertex classes @xmath775 .
the last three classes each have size @xmath776 .
add edges to @xmath1 forming a simple @xmath726-partite @xmath726-graph between these classes , such that in this set of edges the vertices in @xmath117 have degree @xmath3 and the vertices in @xmath777 have degree @xmath750 .
now let @xmath778 .
consider what happens when the algorithm is run on this graph .
each vertex @xmath779 will be added to @xmath780 and the corresponding edges will be added to @xmath731 , producing the required @xmath750-regular graph on @xmath745 .
suppose now that @xmath781 , @xmath782 , @xmath783 .
consider the 4 vertex classes @xmath784 .
add edges forming a simple @xmath726-partite @xmath726-graph between these classes , such that in this set of edges the vertices in @xmath785 have degree @xmath3 , the vertices in @xmath786 have degree @xmath750 , and the vertices in @xmath757 have degree @xmath758 .
add also edges between @xmath786 and @xmath787 so that every vertex in @xmath788 has degree @xmath3 in @xmath1 .
let @xmath789 , and suppose the degree ordering on @xmath117 is @xmath790 .
when the algorithm is run , the vertices in @xmath785 will be added to @xmath780 .
let @xmath791 be the last vertex in the degree order in @xmath785 .
after vertex @xmath792 has been inspected by the algorithm , the graph @xmath731 will be a @xmath58-partite graph between classes @xmath793 , where the vertices in @xmath786 have degree @xmath750 and the vertices in @xmath757 have degree @xmath758 .
every vertex @xmath794 will be added to @xmath795 , and the graph @xmath732 will then be the required @xmath758-regular graph on @xmath757 .
the graph @xmath299 can be achieved similarly .
this argument overlooks the point that the graphs @xmath299 , @xmath732 , @xmath731 can not be simultaneously produced as stated ; for example , it would require the degree in @xmath731 of a vertex in @xmath796 to be @xmath797 , whereas the maximum degree , by construction of @xmath731 , is @xmath798 .
nonetheless , we can adjust some of the edge sets slightly so that the example works .
these constructions generalize to all @xmath0 .
in @xcite , a lower bound for the list colouring number of a regular hypergraph was proved .
theorem 2.1 of that paper , based on a simple probabilistic argument , gave a bound of approximately @xmath799 provided there is a collection @xmath2 of containers for the independent sets , with @xmath800 for each @xmath26 and with @xmath801 .
this proof fails to work for a general hypergraph because it is not possible to find containers of bounded size , but theorem [ thm : uniform ] provides conditions under which we can recover the proof .
let @xmath802 $ ] be a set of colours .
we say that a collection of lists @xmath803 : u \in [ n ] \}$ ] is _
@xmath2-compatible _ if there is a colouring function @xmath804\to[t]$ ] and a tuple @xmath704 , such that , for each @xmath424 $ ] , @xmath805 and @xmath806 .
[ thm : uncompatible_lists ] let @xmath807 .
then there exists @xmath808 , such that the following property holds for all @xmath809 .
let @xmath810 and let @xmath811 .
let @xmath812 and let @xmath813 $ ] .
suppose that there is a map @xmath814 $ ] , such that @xmath278|\,\le\,(1-c)v \leqno{\quad(a)}\ ] ] holds for every @xmath277 , where @xmath279 .
suppose moreover that @xmath815,\ldots , c_t\cap[v])\,:\,g(c_1,\ldots , c_t)=v\ , \}\right|\,\le\,e^{vt / k}\leqno{\quad(b)}\ ] ] holds for all @xmath275 $ ] .
then there is a collection of lists @xmath816 \}$ ] , each of size @xmath817 , which is not @xmath2-compatible . for each @xmath818 $ ] , let @xmath819^{(\ell)}$ ]
be a subset of @xmath802 $ ] of size @xmath7 chosen uniformly and independently at random , and let @xmath820 \}$ ] be the collection of lists . given a tuple @xmath821 , we define , for each @xmath424 $ ] , the set of colours @xmath822
: u \in c_i \}\,.\ ] ] we say that @xmath823 _ fits _
@xmath266 if @xmath824 for each @xmath424 $ ] .
this is the same as saying there is a function @xmath804\to[t]$ ] with @xmath825 , or in other words , @xmath805 and @xmath806 .
hence we shall prove the theorem by showing that , with positive probability , @xmath823 fits no tuple @xmath266 , since then @xmath823 is not @xmath2-compatible .
in fact , we claim something stronger : with positive probability , @xmath823 _ rejects _ every tuple @xmath266 , meaning that there is some @xmath826 $ ] with @xmath827 , where @xmath279 .
to establish the claim , fix for the time being some tuple @xmath266 and let @xmath279 .
let @xmath826 $ ] and let @xmath828 be the probability that @xmath829 , or equivalently @xmath830 \setminus b_u$ ] .
then @xmath831 we note here that @xmath832 if @xmath833 is large enough and thus @xmath834 .
write @xmath835 for the average of the values @xmath836 for @xmath826 $ ] ; then by condition ( a ) of the lemma we have @xmath837 } z_u \ge\sum_u t-|b_u| = vt - \sum_u |b_u| = vt - \sum_{i=1}^t |c_i\cap [ v]| \geq vct\,.\ ] ] so , since the function @xmath838 is convex for @xmath839 , we have @xmath840 } p_u = \sum_{u\in[v]}\binom{z_u}{\ell } \binom{t}{\ell}^{-1 } \,&\ge\ , v \binom{z}{\ell}\binom{t}{\ell}^{-1 } \nonumber\\ \,&\geq\ , v \binom{ct}{\ell}\binom{t}{\ell}^{-1 } \,\geq\ , v ( c-(\ell-1)/t)^\ell \,.\nonumber\end{aligned}\ ] ] since @xmath406 we have @xmath841 , and so @xmath842 .
hence the probability that @xmath843 fails to reject @xmath266 is @xmath844 ) & = \
, \prod_{u\in[v ] } ( 1-p_u ) \nonumber\\ & \leq\ , \exp \{-\sum_{u\in[v ] } p_u\ } \leq \exp \{-v c^\ell/2 \}\,.\nonumber\end{aligned}\ ] ] notice that the probability of @xmath266 not being rejected depends only on the tuple @xmath715,\ldots , c_t\cap[v])$ ] which , by condition ( b ) of the lemma , takes at most @xmath845 values as @xmath266 ranges over @xmath846 . hence if we fix @xmath47 and write @xmath847 for the probability that there is some tuple @xmath266 with @xmath279 which is not rejected , then @xmath848 \right\}\\ & \le & \exp \left\ { -\frac{v}{2k } k^{\epsilon/2 } \right\ }
\mbox{\ \ \ if $ k_0 $ is large enough}\\ & \le&\exp \left\{-\frac{1}{2}v^{\epsilon/2}\right\ } \mbox{\ \ \ since $ k\le v$}\\ & \le&v^{-2}\mbox{\ \ \
if $ k_0 $ is large enough.}\end{aligned}\ ] ] finally , if we consider all tuples @xmath277 , the probability that one of them is not rejected is at most @xmath849 } p_v \,\le\,\sum_{v\ge k}v^{-2}\,<\,1\ ] ] if @xmath833 is large .
this establishes our claim and so proves the theorem .
we can now prove a lower bound on the list chromatic number of a hypergraph .
it is awkward to state the most general conditions under which a non - trivial bound can be obtained , so instead we state a couple of typical results .
the first applies to many @xmath0-graphs of average degree @xmath3 , and certainly to simple @xmath0-graphs ; indeed , theorem [ thm : chil ] stated earlier is the special case when @xmath1 is simple .
[ cor : chil ] let @xmath61 be fixed .
let @xmath1 be an @xmath0-graph with average degree @xmath3 .
suppose that @xmath850 for every @xmath45 and for @xmath851 ( recall definition [ defn : delta ] ) , where @xmath136 as @xmath62 .
then @xmath852 moreover , if @xmath1 is regular then @xmath64 we shall apply theorem [ thm : uniform ] to @xmath1 .
choose @xmath853 so that , as @xmath62 , then @xmath854 , @xmath855 and @xmath856 for all @xmath45 and @xmath181 .
let @xmath857 .
then @xmath858 because @xmath854 . also , recalling definition [ defn : delta ] , @xmath859 , so @xmath860 .
moreover @xmath861 because @xmath855 .
let @xmath862 .
then @xmath863 . let @xmath45 .
for each @xmath864 , at most @xmath865 edges contain both @xmath350 and @xmath47 , and so @xmath866 .
it follows that @xmath271)\le ( 1/nd)knd^{(r-2)/(r-1)}\zeta^{-1}\le\zeta^5\le\zeta/2r!$ ] .
so the conditions of theorem [ thm : uniform ] are satisfied .
it follows that there exists a collection @xmath2 of containers for the independent sets of @xmath1 , satisfying properties ( b ) and ( c ) of theorem [ thm : uniform ] , and since @xmath867 it follows that conditions ( a ) and ( b ) of theorem [ thm : uncompatible_lists ] are satisfied , with @xmath868 .
consequently there are lists of size @xmath869 that are not @xmath2-compatible , which is to say lists of size at least @xmath870 .
since @xmath2 is a set of containers for the independent sets of @xmath1 , the first claim of the corollary follows .
the proof for regular graphs is similar , except that we take @xmath871 . to achieve this we make use of corollary [ cor : sparse_container ] instead of theorem [ thm : uniform ] . with @xmath179 , @xmath222 and @xmath9
defined as before , we can take @xmath872 in corollary [ cor : sparse_container ] because @xmath873 .
we obtain a collection @xmath2 of containers such that @xmath210)\le \zeta e(g)=o(e(g))$ ] for all @xmath26 .
because @xmath1 is regular this means that @xmath800 , where @xmath871 and @xmath874 . we can now apply theorem [ thm : uncompatible_lists ] by defining @xmath875 for all @xmath266 ; note that condition ( b ) of the theorem is satisfied because , by corollary [ cor : sparse_container ] , @xmath876 .
the remainder of the proof is the same .
the bound given for @xmath0-graphs of average degree @xmath3 is weaker than that for regular @xmath0-graphs because we only had containers of measure @xmath469 available , rather than @xmath877 .
probably this is an artifact of our algorithm , and @xmath878 holds for @xmath0-graphs of average degree @xmath3 .
the bound for regular graphs is tight .
indeed , let @xmath879 be the complete @xmath0-partite @xmath0-graph with @xmath66 vertices in each class . suppose that lists of size @xmath7 are given to the vertices .
randomly choose , for each colour in the palette , a vertex class on which that colour is forbidden to be used ; then the expected number of vertices with no available colour is @xmath880 which is less than one if @xmath881 , and so @xmath882 ( see haxell and verstrate @xcite ) .
this graph is @xmath3-regular where @xmath883 so @xmath884 .
note that @xmath885 .
it is not hard to construct an @xmath66-regular simple subgraph @xmath1 of @xmath879 , and so ( putting @xmath886 ) we have simple @xmath3-regular @xmath0-graphs with @xmath887 .
quite possibly @xmath884 in this case too , because a subgraph of @xmath1 with @xmath888 vertices in each class is likely to be very sparse , and a random colouring might be repairable if @xmath889 , or @xmath890 . but this argument is far from rigorous . as an illustration of the use of containers for non - independent sets we finish with the next result .
let @xmath1 be a graph with average degree @xmath3 .
then , for each @xmath864 there is a list @xmath891 of @xmath892 colours , such that it is not possible to choose a colour @xmath893 with the vertices of each colour spanning a planar graph .
we follow the proof of corollary [ cor : chil ] with @xmath5 , except we use a set @xmath2 of containers for those subsets @xmath25 for which @xmath238 $ ] is planar .
since a planar graph is @xmath894-degenerate , we can apply theorem [ thm : uniform ] and continue with the proof exactly as before , provided @xmath895 .
but @xmath896 so this condition holds comfortably .
theorem [ thm : ffree_cover ] is obtained by a routine application of corollary [ cor : sparse_container ] to the following hypergraph , whose independent sets correspond to @xmath8-free @xmath7-graphs on vertex set @xmath68 $ ] .
[ def : gnh ] let @xmath8 be an @xmath7-graph .
let @xmath897 .
the @xmath0-graph @xmath97 has vertex set @xmath68^{(\ell)}$ ] , where @xmath898 is an edge whenever @xmath899 , considered as an @xmath7-graph with vertices in @xmath68 $ ] , is isomorphic to @xmath8 .
[ lem : ffree_delta ] let @xmath8 be an @xmath7-graph with @xmath900 and let @xmath901 . for @xmath77 sufficiently large , @xmath902 .
let @xmath903 .
consider @xmath904^{(\ell)}$ ] ( so @xmath173 is both a set of vertices of @xmath1 and an @xmath7-graph on vertex set @xmath68 $ ] ) .
the degree of @xmath173 in @xmath1 is the number of ways of extending @xmath173 to an @xmath7-graph isomorphic to @xmath8 . if @xmath173 as an @xmath7-graph is not isomorphic to any subgraph of @xmath8 , then clearly @xmath905
otherwise , let @xmath906 be the number of vertices in @xmath173 considered as an @xmath7-graph , so there exists @xmath907 $ ] , @xmath908 with @xmath909 .
edges of @xmath1 containing @xmath173 correspond to copies of @xmath8 in @xmath68^{(\ell)}$ ] containing @xmath173 , each such copy given by a choice of @xmath910 vertices in @xmath68-v$ ] and a permutation of the vertices of @xmath8 .
hence @xmath911 for some integer @xmath912 in the range @xmath913 .
thus for @xmath77 sufficiently large , @xmath914 for @xmath45 and @xmath915 , the quantity @xmath916 is the maximum of @xmath917 over all @xmath904^{(\ell)}$ ] with @xmath918 and @xmath919 .
thus @xmath920 let @xmath921 .
since @xmath922 and @xmath901 , for @xmath923 @xmath924 by definition of @xmath925 and @xmath13 , @xmath926 .
hence @xmath927 and so , with @xmath897 , @xmath928 as claimed .
we use a well - known supersaturation theorem to bound the number of edges in containers .
[ prop : supersaturation ] let @xmath8 be an @xmath7-graph and let @xmath929 .
there exists @xmath930 and @xmath931 such that if @xmath85 is an @xmath7-graph on @xmath932 vertices containing at most @xmath933 copies of @xmath8 then @xmath934 .
let @xmath935 be given by proposition [ prop : supersaturation ] , let @xmath936 , let @xmath903 , let @xmath897 and let @xmath937 by lemma [ lem : ffree_delta ] , @xmath938 .
the vertex set of @xmath1 is @xmath68^{(\ell)}$ ] , @xmath939 ; and the edge set of @xmath1 is the set of @xmath7-graphs isomorphic to @xmath8 , of which there are at most @xmath940 .
we claim that the collection @xmath2 given by corollary [ cor : sparse_container ] applied with @xmath941 for @xmath77 sufficiently large satisfies the conditions of the theorem .
write @xmath942 for the constant @xmath253 in corollary [ cor : sparse_container ] .
condition ( a ) : @xmath2 covers the independent sets of @xmath1 , which are precisely the @xmath8-free @xmath7-graphs on vertex set @xmath68 $ ]
. condition ( b ) : for each @xmath25 , the corresponding @xmath26 with @xmath27 is specified by a tuple @xmath943 with @xmath287 bounded by a constant depending on @xmath0 and @xmath944 . furthermore @xmath945 for @xmath206 sufficiently large as a function of @xmath946 .
condition ( c ) : the number of copies of @xmath8 in @xmath85 is at most @xmath947 . note also that since @xmath948 , proposition [ prop : supersaturation ] implies that @xmath949 .
condition ( d ) : this holds as @xmath950 for @xmath206 and @xmath77 sufficiently large .
we may assume that @xmath71 , since otherwise the result is trivial .
let @xmath929 and let @xmath951^{(\ell)})$ ] be the collection given by theorem [ thm : ffree_cover ] ( we may assume that @xmath77 is sufficiently large ) . note that @xmath952 and @xmath953 for each @xmath259 . since every @xmath8-free @xmath7-graph on @xmath77 vertices is a subgraph of some graph @xmath259 , this means there are at most @xmath954 @xmath8-free graphs on @xmath77 vertices .
since @xmath75 was arbitrary this completes the proof .
the proof is very similar to theorem [ thm : ffree_cover ] , and we only sketch the details .
let @xmath955 .
let @xmath1 be the @xmath0-graph whose vertex set is two copies of @xmath68^{(\ell)}$ ] , denoted by @xmath956 and @xmath957 ( vertices in @xmath956 correspond to @xmath7-edges and vertices in @xmath957 correspond to non-@xmath7-edges ) , and whose edges correspond to copies of @xmath8 ; thus @xmath958 is an edge of @xmath1 whenever @xmath959 and @xmath960 are the edges and non - edges , respectively , of an @xmath7-graph isomorphic to @xmath8 with vertices in @xmath68 $ ] .
note that every induced-@xmath8-free @xmath7-graph @xmath961^{(\ell)}$ ] corresponds to an independent set of @xmath1 , namely the set of vertices in @xmath956 corresponding to the edges of @xmath25 together with the set of vertices in @xmath957 corresponding to non - edges of @xmath25 .
@xmath1 has very similar properties to @xmath962 given in definition [ def : gnh ] , where @xmath963 is the complete @xmath7-graph on @xmath964 vertices . in particular , for fixed @xmath179 , the @xmath184 for @xmath1 differ only by a constant factor by those for @xmath965 .
let @xmath966 , so that @xmath967 . as in the proof of theorem [ thm : ffree_cover ] , we may apply corollary [ cor : sparse_container ] with @xmath968 to obtain the required collection @xmath969 , where each @xmath26 is identified with a 2-coloured @xmath7-multigraph in the obvious way .
we now prove theorem [ thm : ind_ffree_measure ] .
first we must define the function @xmath112 . for a 2-coloured @xmath7-multigraph @xmath970 with red and blue edge sets @xmath971 and @xmath972 , define @xmath973 observe that the probability that a random @xmath7-graph on the same vertex set is a subgraph of @xmath970 is @xmath974 .
let @xmath975^{(\ell ) } , h \not\subset j\ , \}\,.\ ] ] then we put @xmath976 .
as always , we need a supersaturation lemma for the proof of theorem [ thm : ind_ffree_measure ] .
the proof of the following is essentially the same as that of proposition [ prop : supersaturation ] , and we do not give details .
[ lem : ind_supersaturation ] let @xmath8 be an @xmath7-graph and let @xmath977 .
there exists @xmath930 and @xmath931 such that if @xmath85 is a 2-coloured @xmath7-multigraph on @xmath932 vertices containing at most @xmath933 copies of @xmath8 then @xmath978 .
the lower bound on the stated probability follows from the definition of @xmath112 .
let @xmath75 and let @xmath935 be given by lemma [ lem : ind_supersaturation ] .
let @xmath2 be the collection of 2-coloured @xmath7-multigraphs given by theorem [ thm : ind_ffree_cover ] satisfying @xmath979 and for every @xmath26 , the number of copies of @xmath8 in @xmath85 is at most @xmath933 .
by lemma [ lem : ind_supersaturation ] , for each @xmath26 we have @xmath980 .
since every induced-@xmath8-free graph on vertex set @xmath68 $ ] is contained in some @xmath26 , @xmath981 since @xmath75 was arbitrary this completes the proof of theorem [ thm : ind_ffree_measure ] .
in this section , @xmath53 denotes a finite field , the set of integers @xmath68 $ ] , or an abelian group .
we consider linear systems of the form @xmath125 , where @xmath117 is a @xmath122 matrix . as in the introduction , when @xmath53 is a finite field or @xmath68 $ ] , then @xmath117 has entries in @xmath53 .
our methods apply to abelian groups as well : in this case , let @xmath117 have integer entries , where integer - group multiplication @xmath982 , @xmath983 , @xmath984 , is @xmath985 copies of @xmath986 , @xmath987 ; or @xmath988 copies of @xmath989 if @xmath985 is negative .
the definitions of _ full rank _ and _ abundant _ given in the introduction extend to abelian groups .
often one wishes to discount solutions to an equation @xmath125 where the vector @xmath986 contains repeated values .
for example , in forbidding a 3-term arithmetic progression , we take @xmath990 and @xmath991 and discount solutions of the form @xmath992 . to accommodate this setup , we let @xmath993 be a set of discounted solutions .
we also call @xmath994 a _ @xmath122 linear system _ , where a solution to the system is a vector @xmath995 such that @xmath125 .
a subset @xmath996 is _ solution - free _ if there is no @xmath997 with @xmath125 . in this setup , we could also define @xmath998 to be the maximum size of a solution - free subset when @xmath999 .
however , this turns out to be unnecessary , since typically @xmath1000 so @xmath1001 by proposition [ prop : eqn_removal ] .
[ def : ma ] let @xmath994 be a @xmath122 linear system with @xmath117 abundant .
when @xmath53 is a finite field or @xmath68 $ ] , define ( as in rdl and ruciksi @xcite ) @xmath1002,\ , |j| \ge 2 } \frac{|j|-1}{|j|-1+\mbox{rank}(a_j)-k},\ ] ] where the matrix @xmath1003 is the @xmath1004 submatrix of @xmath117 obtained by deleting columns indexed by @xmath970 .
otherwise , when @xmath53 is an abelian group , let @xmath617 be the maximum value of @xmath1005 for which @xmath1003 has full rank whenever @xmath1006 , and let @xmath1007 it can readily be checked that if @xmath117 is abundant then the denominators appearing in the definition of @xmath1008 are strictly positive .
the separate definition of @xmath1008 when @xmath53 is an abelian group is necessary since the rank of an integer matrix over an abelian group is not well - defined ; in general , when the pair @xmath1009 could either be considered a finite field or an abelian group with @xmath117 integer valued , the value of the second definition is at least as big as the value of the first definition .
this is since @xmath1010 when @xmath1011 , and is otherwise at least @xmath1012 .
this is our main theorem for linear equations .
the _ determinantal _ of a @xmath122 integer matrix is the greatest common divisor of the determinants of its @xmath1013 submatrices .
[ thm : eqn_cover ] let @xmath75 and let @xmath994 be a @xmath122 linear system with @xmath117 abundant .
there are constants @xmath1014 , depending on @xmath1015 in the case @xmath138 $ ] , and depending only on @xmath1016 otherwise , such that if @xmath1017 and @xmath1018 then there exists @xmath1019 satisfying * for every solution - free subset @xmath996 there exists @xmath26 with @xmath27 , * moreover , for each pair @xmath78 in ( a ) , there exists @xmath79 where @xmath80 , @xmath81 and @xmath1020 , such that @xmath83 , * for every @xmath259 , the number of solutions to @xmath1021 with @xmath1022 is at most @xmath1023 , * if @xmath53 is a finite field or @xmath68 $ ] , or @xmath53 is an abelian group and the determinantal of @xmath117 is coprime to @xmath1024 , then @xmath1025 , * @xmath1026 .
[ thm : eqn_count ] let @xmath994 be a @xmath118 linear system with @xmath1000 and @xmath117 abundant , and additionally with the determinantal of @xmath117 coprime to @xmath1024 in the case that @xmath53 is an abelian group
. then the number of solution - free subsets of @xmath53 is @xmath131 . here @xmath136 as @xmath137 , with @xmath117 fixed in the case @xmath138 $ ] .
note that theorem [ thm : eqn_count ] generalizes theorem [ thm : eqn_countnox ] by including abelian groups and allowing @xmath999 . if @xmath117 is not abundant , then the conclusion of theorem [ thm : eqn_count ] need not hold .
for example , let @xmath1027 , @xmath140 , and consider the cyclic group @xmath1028 for @xmath10 odd .
observe that the pairs @xmath1029 such that @xmath1030 and @xmath1031 partition @xmath1032 .
therefore @xmath1033 .
however , one can construct a solution - free set by including either @xmath986 or @xmath1034 or neither for each pair @xmath1029 , so there are at least @xmath1035 solution - free sets .
there are similar examples with larger values of @xmath9 and @xmath1036 .
additionally , when @xmath138 $ ] , the condition that @xmath117 is fixed as @xmath1037 is necessary .
for example , for the equation @xmath1038 , the maximum size of a solution - free subset of @xmath68 $ ] is @xmath1039 ( since for every pair @xmath1040 $ ] with @xmath1041 , a solution - free set can include at most one of @xmath1042 or @xmath986 ) , but there are at least @xmath1043 solution free sets , since for every @xmath1040 $ ] with @xmath1041 and @xmath1044 , we can include either @xmath1042 or @xmath986 or neither to form a solution - free set .
theorem [ thm : eqn_cover ] follows from an application of corollary [ cor : sparse_container ] to the following hypergraph , whose independent sets correspond to solution - free subsets of @xmath53 .
[ def : gfab ] let @xmath994 be a @xmath122 linear system .
the @xmath0-partite @xmath0-graph @xmath1045 has vertex set @xmath1046 , where each @xmath1047 is a disjoint copy of @xmath53 , and edge set @xmath1048 .
[ fact : eqn_deg_codeg ] let @xmath53 be a finite field or abelian group , let @xmath117 be a @xmath1049 matrix and let @xmath120
. if @xmath117 has full rank then there are @xmath1050 solutions to @xmath125 .
more generally if @xmath53 is a finite field , there are at most @xmath1051 solutions to @xmath125 .
if @xmath117 has full rank , then for every @xmath1052 there exists @xmath1053 with @xmath1054 . thus if @xmath1055 is a solution to @xmath1056 then @xmath1057 , so by symmetry every @xmath120 has @xmath1058 solutions to @xmath116 .
the case when @xmath53 is a finite field is standard .
[ lem : eqn_delta ] let @xmath994 be a @xmath122 linear system where @xmath53 is a finite field or abelian group , @xmath117 is an abundant matrix and @xmath1059 .
let @xmath1045 , @xmath901 and @xmath1060
. then @xmath1061 .
the number of edges in @xmath1 is the number of solutions to @xmath1021 not in @xmath1062 .
the matrix @xmath117 has full rank , so by fact [ fact : eqn_deg_codeg ] the number of edges of @xmath1 is @xmath1063 . for @xmath45 and @xmath414 , @xmath916 is the maximum over all sets @xmath1064 with @xmath918 of the number of edges of @xmath1 containing @xmath173 . if @xmath173 contains two vertices
in the same part @xmath1047 then there are no edges containing @xmath173 .
otherwise suppose @xmath1065 for @xmath1066 , where @xmath1067 .
let @xmath1068 .
the number of edges containing @xmath173 is at most the number of solutions to @xmath125 with @xmath1069 for @xmath1066 , which for some @xmath1070 is the number of solutions to @xmath1071 , @xmath1072 .
we now split the proof into two cases depending on whether @xmath53 is a finite field or an abelian group .
_ when @xmath53 is an abelian group : _ if @xmath1073 ( recall definition [ def : ma ] ) , then @xmath1003 has full rank by assumption , and so fact [ fact : eqn_deg_codeg ] implies the number of solutions is at most @xmath1074 .
when @xmath1075 , we have @xmath1076 , so @xmath1077 when @xmath1078 , the bound @xmath1079 implies that @xmath1080 when @xmath1081 , the bound @xmath1082 implies that @xmath1083 since @xmath1084 and @xmath1085 , this implies that @xmath1086 for all @xmath1005 .
when @xmath53 is a finite field : _ by fact [ fact : eqn_deg_codeg ] the number of solutions to @xmath1071 is at most @xmath1087 .
hence @xmath1088,\ , |j|=j } |f|^{r - j-\rm{rank}(a_j)}.\ ] ] using @xmath1089 and @xmath901 , this implies that @xmath1090,\ , |j|=j } |f|^{1-j+k-\rm{rank}(a_j)+(j-1)/m_f(a ) } \gamma.\ ] ] the exponent is at most @xmath1091 be definition of @xmath1008 so @xmath1086 . in both cases , @xmath1092
the proof of part ( d ) of theorem [ thm : eqn_cover ] requires the following removal lemma of krl , serra and venna .
note that if @xmath117 has determinantal coprime to @xmath1024 then in particular @xmath117 has full rank .
[ prop : eqn_removal ] let @xmath121 be a @xmath118 linear system where @xmath53 is a finite field or abelian group and @xmath117 has full rank .
suppose further that @xmath117 has determinantal coprime to @xmath1024 in the case that @xmath53 is an abelian group .
let @xmath1093 .
then there exists @xmath1094 such that for all @xmath1095 and @xmath1096 , if there are at most @xmath1097 solutions to @xmath1021 with @xmath1098 , then there are sets @xmath1099 with @xmath1100 such that there is no solution to @xmath1021 with @xmath1101 .
we may assume that @xmath53 is a finite field or abelian group .
indeed , @xmath68 $ ] can be embedded into the finite field @xmath1102 for a sufficiently large prime @xmath111 .
taking @xmath111 in the range @xmath1103 , where @xmath1104 is the sum of the absolute values of @xmath117 , guarantees that @xmath117 is still abundant in @xmath1102 and that a solution to @xmath1105 is also a solution to @xmath125 ( provided , say , @xmath1106 ; but we may assume this since otherwise there are no solutions to @xmath125 in @xmath68 $ ] ) .
then the result of this theorem for @xmath1107 implies the result for @xmath1108,a , b , z)$ ] , since @xmath1109 is bounded by a constant depending only on @xmath117 .
let @xmath935 be given by proposition [ prop : eqn_removal ] applied with @xmath1110 .
let @xmath1111 .
note that @xmath1112 .
let @xmath1113 and let latexmath:[$\tau=12r !
} @xmath938 .
note also that @xmath1115 for @xmath1024 sufficiently large , so corollary [ cor : sparse_container ] gives a collection of sets @xmath1116 covering the independent sets of @xmath1 .
for @xmath639 , let @xmath1117 be the part of @xmath1118 in the @xmath1119th copy of @xmath53 and let @xmath1120 we claim that @xmath2 satisfies the conditions of the theorem .
condition ( a ) : consider a solution - free set @xmath123 .
the subset @xmath970 of @xmath15 formed by taking a copy of @xmath25 in each @xmath1047 is an independent set in @xmath1 . in particular ,
it is contained in some @xmath639 , hence @xmath1121 .
condition ( b ) : for @xmath78 in ( a ) , there exists a tuple @xmath79 such that @xmath1122 , @xmath1123 , @xmath1124 and @xmath1125 for @xmath206 sufficiently large as a function of @xmath0 and @xmath941 . thus @xmath85 is specified by the @xmath1126-tuple @xmath1127 .
this verifies condition ( b ) ( with @xmath1126 in place of @xmath287 ) .
condition ( c ) : consider @xmath26 .
each solution to @xmath1021 with @xmath1128 corresponds to an edge of @xmath633 $ ] , of which there are at most @xmath1129 .
condition ( d ) : there are at most @xmath1130 such @xmath1131 with @xmath125 , so by proposition [ prop : eqn_removal ] there exists a set @xmath1132 , @xmath1133 such that there are no @xmath1134 with @xmath125 , implying that @xmath1135 . condition ( e ) : observe that @xmath1136 for @xmath206 sufficiently large as a function of @xmath0 and @xmath941 .
the proof is the same as the proof of corollary [ cor : ffree_count ] , where the collection given by theorem [ thm : eqn_cover ] is used instead of the collection given by theorem [ thm : ffree_cover ] .
in this section we prove theorem [ thm : sidon ] . to prove the upper bound we construct , in the natural way , the hypergraph representing the solutions to @xmath144 in a subset @xmath460
we then apply theorem [ thm : iteration ] in an entirely mechanical way ; all that is needed is to set appropriate values and to check the conditions .
we remark that the codegree function @xmath1137 for this hypergraph exhibits a change in behaviour when @xmath200 , as the dominant contribution then comes from @xmath201 rather than @xmath1138 ( see equation ( [ eqn : sidon_delta ] ) ) ; kohayakawa , lee , rdl and samotij @xcite noticed an interesting behavioural change at the same point , for a closely related problem .
_ lower bound construction .
_ suppose @xmath1139 for some prime @xmath111 .
ruzsa @xcite shows that there is a set @xmath1140 $ ] of size @xmath1141 such that every sum of two elements of @xmath43 is distinct modulo @xmath1142 .
thus for any @xmath1143 satisfying @xmath1144 for @xmath1145 , the set @xmath1146 is a sidon subset of @xmath1147 $ ] , where @xmath1148 .
this gives @xmath1149 sidon subsets of @xmath36=[4p(p-1)]$ ] .
the general case follows by embedding @xmath1147 $ ] into @xmath36 $ ] , where @xmath111 is the largest prime such that @xmath1150 , and using the fact that the ratio of successive primes tends to 1 .
( we note that any construction for large modular sidon sets could have been used here ; this includes the classical constructions of singer @xcite and of bose @xcite . )
_ upper bound .
_ let @xmath1 be the @xmath726-graph on vertex set @xmath36 $ ] , where @xmath1151^{(4)}$ ] is an edge whenever @xmath1152 .
sidon sets correspond to independent sets in @xmath1 ( although the converse is not always true , since solutions to @xmath144 where @xmath1153 or @xmath1154 do not correspond to edges of @xmath1 )
. we shall apply theorem [ thm : iteration ] to the graph @xmath1 ; our task is to bound @xmath1155 . to this end , let @xmath1156 , let @xmath1157 , and consider @xmath604 $ ] where @xmath1158 . for @xmath1159 $ ] , let @xmath1160 .
note that @xmath1161 .
each pair of sets @xmath1162 with @xmath1163 corresponds to an edge with @xmath144 , and each such edge corresponds to the two pairs @xmath1162 and @xmath1164 .
hence the number of edges in @xmath1165 $ ] satisfies @xmath1166 ) = \frac{1}{2 } \sum_{i=1}^{n-1 } { t_i \choose 2 } \ge \frac{n-1}{2 } { \frac{1}{n-1 } \sum_i t_i \choose 2 } \ge u^4 / 20n,\end{aligned}\ ] ] where the last inequality holds for @xmath1167 .
let @xmath1168 , as used for theorem [ thm : iteration ] .
thus @xmath1169 .
let @xmath1170 .
for the application of theorem [ thm : iteration ] , put @xmath1171 we must check that the conditions of theorem [ thm : iteration ] hold .
recall the definition of @xmath1172 . in @xmath1165
$ ] , observe that @xmath1173 , since for @xmath1174 , the number of solutions of the form @xmath1152 is at most @xmath1175 and the number of solutions of the form @xmath1176 is at most @xmath350 ; similarly @xmath1177 and @xmath1178 . hence @xmath1179 and ( since @xmath1180 ) @xmath1181 then both terms on the right hand side of ( [ eqn : sidon_delta ] ) are less than @xmath1182 , so @xmath1183 is satisfied . to apply theorem [ thm : iteration ]
, we also require @xmath1184 . if @xmath1185 , then the constraint @xmath1186 is automatically satisfied ( since @xmath1187 and @xmath1188 ) , and @xmath1189 where the first inequality holds since @xmath1190 is an increasing function of @xmath179 when @xmath1191 , and the second inequality holds since @xmath1192 is an increasing function of @xmath350 when @xmath1193 , and @xmath1194 which is less than @xmath1195 because @xmath1169 . alternatively , if @xmath1196 then the constraint @xmath1186 is automatically satisfied , and @xmath1197 where the second inequality holds since @xmath1198 is an increasing function of @xmath350 when @xmath1199 and @xmath1194 , together with the bound @xmath1200 .
let @xmath1201 for @xmath1202 .
let @xmath1203 and @xmath1204 .
the conditions of theorem [ thm : iteration ] hold , so let @xmath2 be the collection of containers given by theorem [ thm : iteration ] for the graph @xmath1 , where each @xmath259 satisfies @xmath1205 ( and hence @xmath1206 ) , and @xmath1207 . since @xmath1208 and @xmath1209 are non - increasing functions of @xmath66 , @xmath1210 for @xmath1211 .
note that @xmath1212 and @xmath1213 , so @xmath1214 observe that @xmath1215 when @xmath1216 ( and hence @xmath1217 ) , so @xmath1218 each sidon set in @xmath36 $ ] is a subset of size at most @xmath1219 of some @xmath259 , @xmath1206 . by lemma [ lem : entropy ]
the number of such subsets is at most @xmath1220 , where @xmath1221 is the maximum size of a sidon set , so @xmath1222 .
letting @xmath1223 be the collection of sidon subsets of @xmath36 $ ] , @xmath1224 which completes the verification .
in this section we prove theorem [ thm : ffree_sparse ] and related theorems . note that the condition @xmath162 in theorem [ thm : ffree_sparse ] is tight up to the value of @xmath206 .
indeed , if @xmath1225 , it is readily checked that for some subgraph @xmath1226 with @xmath1227 , the expected number of copies of @xmath1228 is much less than the number of edges , and removing very few edges will result in an @xmath8-free subgraph . as a further illustration of the paradigm described in [ subsec : sparse ] ,
we prove two other conjectures of kohayakawa , uczak and rdl @xcite .
the first of these has already been proved , by conlon and gowers @xcite for strictly balanced graphs and by samotij @xcite , following schacht @xcite , for all graphs .
it states that , for non - bipartite @xmath8 , not only does every @xmath8-free subgraph @xmath25 of a random graph have at most @xmath1229 edges , but in the case that @xmath25 has close to @xmath1230 edges , it can be made @xmath1231-partite by removing a small number of edges .
the dense ( @xmath1232 ) version of this theorem is the stability theorem of erds and simonovits @xcite .
[ thm : klr_conj ] let @xmath8 be a @xmath44-graph with @xmath1233 and let @xmath161 . there exist constants @xmath1234 such that for @xmath77 sufficiently large and for @xmath162 , the following is true
let @xmath1235 be the event that there exists an @xmath8-free subgraph @xmath1236 with @xmath1237 which can not be made @xmath1231-partite by removing at most @xmath1238 edges .
then @xmath1239 the other conjecture from @xcite , sometimes known as the kr conjecture , has a more technical statement .
let @xmath1 be a graph .
for @xmath1240 , write @xmath1241 for the set of edges of @xmath1 with one vertex in @xmath648 and one vertex in @xmath1242 .
let @xmath1243 and write @xmath1244 for the _ edge density_. for @xmath1245 , say that the pair @xmath1246 is
_ @xmath1247-regular _
if for every @xmath1248 with @xmath1249 and @xmath1250 with @xmath1251 , the edge density satisfies @xmath1252 this extends the notion of regularity to sparse graphs of density @xmath111 .
let @xmath8 be a graph on vertex set @xmath1253 $ ] .
in what follows , @xmath1254 is a partition of @xmath68=[hn]$ ] , where each part has size @xmath1255 .
if @xmath1 is a graph on vertex set @xmath68 $ ] , say that @xmath1 is @xmath1256-regular if for every pair @xmath1257 with @xmath1258 , the bipartite subgraph of @xmath1 between @xmath1259 and @xmath1260 is @xmath1247-regular .
let @xmath1261 denote a graph chosen uniformly at random from all @xmath1262-partite graphs with parts @xmath1263 , having @xmath1264 if @xmath1258 and @xmath1265 otherwise .
we say that @xmath1266 is @xmath8-free if there is no set of vertices @xmath1267 with @xmath1268 such that @xmath1269 is an edge of @xmath1 whenever @xmath1258 . a similar result to theorem [ thm : ffree_sparse ] applies to other structures , such as solutions to linear equations as in [ sec : sumfree ] .
so we can say , for example , that if @xmath970 is a solution - free subset of @xmath53 of maximum size , and if a subset @xmath1274 is chosen with probability @xmath111 , then almost certainly the maximum size of a solution - free subset of @xmath169 is @xmath1275 , provided @xmath111 is not too small .
[ thm : eqn_sparse ] let @xmath161 and let @xmath994 be a @xmath118 linear system with @xmath117 abundant , and additionally with the determinantal of @xmath117 coprime to @xmath53 in the case that @xmath53 is an abelian group .
then there exist constants @xmath1234 , depending on @xmath1276 when @xmath138 $ ] , and depending only on @xmath1277 otherwise , such that for @xmath1017 , @xmath1278 and @xmath1279 , if @xmath1280 is a random subset with each element included independently with probability @xmath111 , then the following event holds with probability greater than @xmath1281 when @xmath53 is a finite field , the condition @xmath1282 in theorem [ thm : eqn_sparse ] is known to be tight up to the value of the constant @xmath206 appearing , at least under some slightly more restrictive assumptions on @xmath994 .
see rdl and ruciski @xcite .
[ lem : sparse ] given @xmath1283 and @xmath428 , there is a constant @xmath1284 such that the following holds .
let @xmath1285 be a set , @xmath1286 , and let @xmath1287 .
let @xmath1288 , let @xmath1289 and let @xmath1290 .
suppose for each @xmath1291 there exists both @xmath1292 and @xmath1293 , where @xmath1294 , @xmath1295 and @xmath1296 .
let @xmath1297 be a random subset where each element is chosen independently with probability @xmath111
. then @xmath1298 consider @xmath1291 and @xmath1299 .
let @xmath1300 .
let @xmath1301 be the event that @xmath1302 the event @xmath1301 is contained in @xmath1303 , where @xmath1304 is the event that @xmath1305 and @xmath1306 is the event that @xmath1307 . since @xmath1304 and @xmath1306 are independent , @xmath1308 .
now @xmath1309 if @xmath1310 is large .
so by standard estimates ( e.g. , chernoff s bound ) , @xmath1311 where @xmath1312 is the binomial random variable .
note that @xmath1313 .
let @xmath1314 , so @xmath1315 .
if @xmath1310 is large we may assume @xmath1316 , so @xmath1317 holds if @xmath1310 , and therefore @xmath986 , is large .
if there exists @xmath1318 , @xmath1291 with @xmath1319 , then the event @xmath1320 holds .
hence the probability in ( [ eqn : sparse ] ) is bounded by @xmath1321 as claimed .
let @xmath1322 be the set of @xmath8-free @xmath7-graphs on vertex set @xmath68 $ ] .
let @xmath1323 and @xmath1324^{(\ell)}$ ] .
for @xmath1291 , let @xmath1325 , @xmath83 and @xmath1326 be given by theorem [ thm : ffree_cover ] .
our aim is to apply lemma [ lem : sparse ] with @xmath1327 and @xmath1328 the conditions of lemma [ lem : sparse ] then hold with @xmath1329 , noting that @xmath1330 and that @xmath1331 if @xmath206 is large enough .
finally , note that in ( @xmath1332 ) , each @xmath8-free @xmath7-graph @xmath1291 is contained in @xmath1333 and @xmath1334 , so the probability in the statement of the theorem is bounded by @xmath1335 completing the proof .
notice that @xmath1336 and @xmath1337 .
it is a standard exercise , using either the stability arguments of erds and simonovits or using szemerdi s regularity lemma , that there exists @xmath75 such that if @xmath85 is a @xmath44-graph on vertex set @xmath68 $ ] for @xmath77 sufficiently large with @xmath1338 and such that @xmath85 contains at most @xmath1339 copies of @xmath8 , then there exists a subgraph @xmath1340 of size @xmath1341 such that @xmath1342 is @xmath1231-partite .
we may and shall assume that @xmath1343 and @xmath1344 .
let @xmath1322 be the set of @xmath8-free graphs on vertex set @xmath68 $ ] .
for @xmath1291 let @xmath1325 , @xmath83 and @xmath1326 be given by theorem [ thm : ffree_cover ] with @xmath944 as above .
let @xmath1345 for @xmath1346 let @xmath1347 be as above , so that @xmath1348 is @xmath1231-partite .
let @xmath1349 .
let @xmath1350 be the event that there exists @xmath1318 , @xmath1346 such that @xmath1351 .
let @xmath1352 be the event that there exists @xmath1318 , @xmath1353 such that @xmath1354 .
observe that @xmath1355 .
the probability of @xmath1352 is bounded by applying lemma [ lem : sparse ] to the collection @xmath1356 , with @xmath1324^{(2)}$ ] , @xmath1357 , @xmath1358 , @xmath1359 , @xmath1360 , @xmath1361 and @xmath1362 ; provided @xmath162 and @xmath1363 are sufficiently large , @xmath1364 the probability of @xmath1350 is bounded by applying lemma [ lem : sparse ] to the collection @xmath1365 , with @xmath1366 , @xmath1367 , @xmath1368 , @xmath1361 and @xmath1362 ; provided @xmath162 and @xmath1363 are sufficiently large , @xmath1369 since @xmath1370 and @xmath1371 is large , this completes the proof of theorem [ thm : klr_conj ] .
let @xmath1372 , let @xmath1373 and let @xmath1374 .
for ease of notation we shall often identify graphs with their edge sets .
let @xmath1375 , @xmath1376 and @xmath1377 be sufficiently small constants to be chosen later .
let @xmath2 be given by theorem [ thm : ffree_cover ] applied with @xmath8 and @xmath944 .
for @xmath79 , let @xmath1378 , and define the following probabilistic events : @xmath1379 theorem [ thm : ffree_cover ] tells us that if @xmath169 is @xmath8-free then there exists @xmath79 with @xmath1380 , and if in addition @xmath169 is @xmath1256-regular then @xmath1301 holds . by the union bound ,
it is hence sufficient to show that @xmath1381 .
theorem [ thm : ffree_cover ] also states that @xmath1382 and @xmath1383 , where @xmath1384 , so it is easily checked that if @xmath169 is @xmath1256-regular then @xmath1385 is @xmath1386-regular provided that @xmath1387 .
this holds provided @xmath1388 for @xmath206 sufficiently large , which we shall assume .
therefore @xmath1389 and so @xmath1390 .
in order to bound @xmath1391 , let @xmath89 be fixed , let @xmath1392 and let @xmath1393 .
consider an @xmath935-regular szemerdi partition of @xmath85 refining the partition @xmath1263 , i.e. , for @xmath0 sufficiently large depending only on @xmath935 , and for each @xmath1394 $ ] an equitable partition @xmath1395 so that all but at most @xmath1396 of the bipartite graphs between pairs @xmath1397 with @xmath1258 are @xmath935-regular . provided @xmath935 and @xmath944 are sufficiently small then , by standard arguments , for every choice of @xmath1398 $ ] there exists some pair @xmath1258 such that the pair @xmath1399 is either irregular or has density @xmath1400 ( since otherwise there would be more than @xmath1401 copies of @xmath8 in @xmath85 , contradicting the conditions of theorem [ thm : ffree_cover ] ) .
in particular , by averaging , there exists @xmath1402 such that at least @xmath1403 of the pairs @xmath1404 with @xmath1405 $ ] are either irregular or of density less than @xmath1406 .
let @xmath1407 $ and $ d_c(v_{i_0,x},v_{j_0,y})<\nu$ } \ } , \\
s & = \{\{u , w\}:\mbox{$u\in u$ , $ w\in w$ , for some $ ( u , w)\in r$}\}.\end{aligned}\ ] ] since there are at most @xmath1408 irregular pairs @xmath1404 with @xmath1405 $ ] , this implies that @xmath1409 for @xmath935 sufficiently small . allowing for the sets @xmath1410
not being all exactly the same size , this implies that @xmath1411 .
note that , by the definition of @xmath1412 , @xmath1413 holds .
write @xmath1414 for the set of pairs of vertices @xmath1415 . if @xmath1306 holds , then @xmath1416 is @xmath1386-regular and so @xmath1417 however @xmath1418 and if @xmath1306 holds then @xmath1419 , so the probability of ( [ eq : klr3_gt ] ) is small . specifically , in generating the random graph @xmath1420 when conditioned on @xmath1421 , we are selecting a set of @xmath1422 edges uniformly from at least @xmath1423 possible edges , and for ( [ eq : klr3_gt ] ) to hold , we must select at least @xmath1424 edges from a set of @xmath1425 possibilities .
this probability is at most @xmath1426 provided @xmath1375 is small enough . since @xmath79 where @xmath1382 , lemma [ lem : entropy ] tells us that @xmath1430 where @xmath1431 is the average size of the @xmath249 .
now @xmath1432 and @xmath1433 .
we can make @xmath1434 as small as we wish to by making @xmath206 large , thus ensuring @xmath1435 , as desired .
let @xmath1108,a , b , z)$ ] be the @xmath1437 linear system corresponding to forbidding an @xmath7-term arithmetic progression in @xmath68 $ ] .
for example if @xmath1438 then @xmath1439 , @xmath991 and @xmath1062 is the set of solutions of the form @xmath1440 that are discounted , so @xmath1441 .
it can readily be checked that @xmath1442}(a)=\ell-1 $ ] . since @xmath1443 , the result immediately follows by applying theorem [ thm : eqn_sparse ] .
n. alon , restricted colorings of graphs , in _ surveys in combinatorics 1993 _ , london mathematical society lecture notes series , * 187 * ( k. walker , ed . ) , cambridge university press , cambridge , ( 1993 ) , 133 .
erds , some recent results on extremal problems in graph theory ( results ) , in : p. rosenstiehl ( ed . ) , theory of graphs ( international symposium , rome , 1966 ) , gordon and breach , newyork , dunod , paris , 1967 , pp .
118123 .
m. simonovits , a method for solving extremal problems in graph theory , stability problems , in : p. erds , g. katona ( eds . ) , theory of graphs ( proceedings of the colloquium , tihany , 1966 ) , academic press , new york , 1968 , pp . 279319 . | we develop a notion of containment for independent sets in hypergraphs . for every @xmath0-uniform hypergraph @xmath1 , we find a relatively small collection @xmath2 of vertex subsets , such that every independent set of @xmath1 is contained within a member of @xmath2 , and no member of @xmath2 is large ; the collection , which is in various respects optimal , reveals an underlying structure to the independent sets .
the containers offer a straightforward and unified approach to many combinatorial questions concerned ( usually implicitly ) with independence . with regard to colouring
, it follows that many ( including all simple ) @xmath0-uniform hypergraphs of average degree @xmath3 have list chromatic number at least @xmath4 . for @xmath5
this improves alon s bound and is tight . for @xmath6 ,
previous bounds were weak but the present inequality is close to optimal . in the context of extremal graph theory
, it follows that , for each @xmath7-uniform hypergraph @xmath8 of order @xmath9 , there is a collection @xmath2 of @xmath7-uniform hypergraphs of order @xmath10 each with @xmath11 copies of @xmath8 , such that every @xmath8-free @xmath7-uniform hypergraph of order @xmath10 is a subgraph of a hypergraph in @xmath2 , and @xmath12 where @xmath13 is a standard parameter ( there is a similar statement for induced subgraphs ) .
this yields simple proofs of many hitherto difficult results : these include the number of @xmath8-free hypergraphs , sparsity theorems of conlon - gowers and schacht , and the full kr conjecture . likewise , for systems of linear equations the containers supply , for example
, bounds on the number of solution - free sets ( including sidon sets , for which we give both lower and upper bounds ) and the existence of solutions in sparse random subsets .
balogh , morris and samotij have independently obtained related results . |
let @xmath0 be the frequency of the @xmath1th species in a random sample of size @xmath2 from a multinomial population with a perhaps countably infinite number of species and let @xmath3 be probability measures under which the @xmath1th species has probability @xmath4 of being sampled , where @xmath5 with .
let @xmath6 and @xmath7 denote the sum of the probabilities of the unobserved species , and the total number of species represented @xmath8 times in the sample , respectively , that is , @xmath9 where @xmath10 .
then @xmath11 is called the sample coverage which is the sum of the probabilities of the observed species .
@xcite proposed the estimator @xmath12 for @xmath13 .
the good estimator @xmath14 has many applications such as shakespeare s general vocabulary and authorship of a poem [ @xcite , @xcite ] , genom [ @xcite ] , the probability of discovering new species in a population [ @xcite , @xcite ] , network species and data confidentiality [ @xcite ] . @xcite
considered the problem of predicting @xmath6 .
they studied prediction and prediction intervals , and gave a real - data example . on the theoretical aspects , many authors studied the asymptotic properties [ cf .
esty ( @xcite ) , @xcite , and @xcite and references therein ] .
@xcite proved the following asymptotic normality : @xmath15 under the condition @xmath16 where @xmath17 recently , @xcite found a necessary and sufficient condition for the asymptotic normality ( [ clt - thm - esty ] ) under the condition @xmath18 that is , under condition ( [ moment - condition-1 ] ) , ( [ clt - thm - esty ] ) holds if and only if both @xmath19 and for any @xmath20 , @xmath21 where for any @xmath22 , @xmath23 in this paper , we consider the moderate deviation problem for the good estimator .
it is known that the moderate deviation principle is a basic problem .
it provides us with rates of convergence and a useful method for constructing asymptotic confidence intervals .
the moderate deviations can be applied to the following nonparameter hypothesis testing problem : @xmath24 where @xmath25 and @xmath26 are two probability measures under which the @xmath1th species has , respectively , probability @xmath27 and @xmath28 of being sampled , where @xmath29 with @xmath30 , @xmath31 .
we can define a rejection region of the hypothesis testing by the moderate deviation principle such that the probabilities of type i and type ii errors tend to @xmath32 with an exponential speed .
the asymptotic normality provides @xmath33 as the asymptotic variance and approximate confidence statements , but it does not prove that the probabilities of type i and type ii errors tend to @xmath32 with an exponential speed .
the moderate deviations can be applied to a hypothesis testing problem for the expected coverage of the sample .
@xcite have established a general delta method on the moderate deviations for estimators .
but the method can not be applied to the good estimator . in order to study the moderate deviation problem for the good estimator
, we need refined asymptotic analysis techniques and tail probability estimates . the exponential moments inequalities , the truncation method , asymptotic analysis techniques and the poisson approximation in @xcite play important roles .
our main results are a moderate deviation principle and a self - normalized moderate deviation principle for the good estimator .
the rest of this paper is organized as follows .
the main results are stated in section [ sec2 ] . some examples and applications to the hypothesis testing problem and the confidence interval are also given in section [ sec2 ] .
the proofs of the main results are given in section [ sec3 ] .
some basic concepts for large deviations and the proofs of several technique lemmas are given in the .
in this section , we state the main results and give some examples and applications .
let @xmath34 , @xmath35 , be a function taking values in @xmath36 such that @xmath37 we introduce the following lindeberg - type condition : for any positive sequence @xmath38 with @xmath39 and any @xmath20 , @xmath40 [ rmk-2 - 1 ] for any @xmath41 , @xmath42 in particular , take @xmath43 . if @xmath44 , then ( [ mdp - lindeberg - condition ] ) holds .
[ main - thm - mdp ] suppose that the conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold .
then @xmath45 satisfies a large deviation principle with speed @xmath46 and with rate function @xmath47 .
in particular , for any @xmath48 , @xmath49 [ main - thm - mdp - self ] suppose that conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold
. then @xmath50 satisfies a large deviation principle with speed @xmath46 and with rate function @xmath47 .
let @xmath51 , @xmath52 be a sequence of positive numbers such that @xmath53 then theorems [ main - thm - mdp ] and [ main - thm - mdp - self ] give the following estimates which are much easier to understand and apply : @xmath54 and @xmath55 set @xmath56 .
then @xmath57 is called the expected coverage of the sample in the literature . by theorems [ main - thm - mdp ] and [ main - thm - mdp - self ] , and lemma [ main - thm - self - lem ] , @xmath58 as an estimator of @xmath59 also satisfies moderate deviation principles .
[ main - thm - mdp - expected ] suppose that conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold .
then @xmath60 and @xmath61 satisfy the large deviation principle with speed @xmath46 and with rate function .
@xcite considered the problem of predicting @xmath6 , and obtained conditionally unbiased predictors and exact prediction intervals based on a poissonization argument .
the moderate deviations for the predictors are also interesting problems . in this subsection , we apply the moderate deviations to hypothesis testing problems and confidence interval .
let @xmath6 be the unknown total probability unobserved species , and let @xmath62 be the estimator defined by ( [ def - estimator ] ) .
first , let us consider a nonparametric hypothesis testing problem .
let @xmath25 and @xmath26 be two probability measures under which the @xmath1th species has , respectively , probability @xmath27 and @xmath28 of being sampled , where @xmath63 with @xmath30 , @xmath31 .
denote by @xmath64 and @xmath65 suppose that the conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold for @xmath66 , @xmath31 , and that @xmath67 consider the nonparameter hypothesis testing @xmath68 we take the statistic @xmath69 as test statistic .
suppose that the rejection region for testing the null hypothesis @xmath70 against @xmath71 is , where @xmath72 is a positive constant .
the probability @xmath73 of type i error and the probability @xmath74 of type ii error are @xmath75 respectively .
it follows @xmath76 therefore , corollary [ main - thm - mdp - expected ] implies that @xmath77 the above result tells us that if the rejection region for the test is@xmath78 , then the probability of type i error tends to @xmath32 with exponential decay speed @xmath79 , and the probability of type ii error tends to @xmath32 with exponential decay speed @xmath80 for all @xmath48 .
but the asymptotic normality does not prove that the probabilities of type i and type ii errors tend to @xmath32 with an exponential speed .
we also consider a hypothesis testing problem for the expected coverage of the sample .
we denote by @xmath81 the probability measures under which the expected coverage of the sample is @xmath57 and set @xmath82 suppose that the conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold for @xmath81 for each @xmath83 .
let @xmath84 be two real numbers preassigned .
consider the hypothesis testing @xmath85 we also take the rejection region @xmath86 , where @xmath72 is a positive constant .
when @xmath87 , @xmath88 and when @xmath89 , @xmath90 next , we apply the moderate estimates to confidence intervals . for given confidence level @xmath91 ,
set @xmath92 .
then by theorem [ main - thm - mdp ] , the @xmath91 confidence interval for @xmath6 is @xmath93 , that is , @xmath94 but the confidence interval contains unknown @xmath95 .
we use theorem [ main - thm - mdp - self ] to obtain another confidence interval with confidence level @xmath91 for @xmath6 which does not contain unknown @xmath95 , @xmath96 let us check that some examples in @xcite also satisfy moderate deviation principles if @xmath97 , where @xmath98 . for a given decreasing density function @xmath99 on @xmath100 .
define @xmath101 , where @xmath102 .
two concrete examples are as follows : let @xmath103 , where @xmath104 and @xmath105 . by example 1 in @xcite , @xmath106 and @xmath107 , where @xmath108 thus ( [ moment - condition-1 ] ) and ( [ moment - condition-2 ] ) hold . by remark [ rmk-2 - 1 ] , ( [ mdp - lindeberg - condition ] ) also holds .
therefore , theorems [ main - thm - mdp ] and [ main - thm - mdp - self ] hold .
let @xmath109 , where @xmath110 for some constant @xmath111 .
then by example 2 in @xcite , @xmath112 and @xmath113 when @xmath39 .
thus , ( [ mdp - lindeberg - condition ] ) is equivalent to @xmath114 which holds if and only if @xmath115 .
therefore , ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold if and only if @xmath115 .
in this section we give proofs of the main results .
let us explain the idea of the proof of theorem [ main - thm - mdp ] .
first , we divide the proof into two cases : case i and case ii , according to the limit @xmath116 and @xmath32 . for case i , by the truncation method and the exponential equivalent method
, we simplify our problems to the case which @xmath117 is uniformly bounded . for case ii , by the poisson approximation and the exponential equivalent method ,
we simplify our problems to the case of independent sums satisfying an analogous lindeberg condition .
for the two cases simplified , we establish moderate deviation principles by the method of the laplace asymptotic integral ( lemmas [ laplace - int - lem ] and [ mdp - poisson - approx ] ) . the exponential moment estimate ( lemma [ exp - moment - inq - lem-1 ] ) plays an important role in the proofs of some exponential equivalence ( lemmas [ exp - moment - estimates - lem-1 ] and [ exp - eqv - lemma-1 ] ) . the main technique in the estimate of the laplace asymptotic integral lemma [ laplace - int - lem ] is asymptotic analysis .
in particular , we emphasis a transformation defined below ( [ laplace - int - lem - eq-4 ] ) which plays a crucial role in the proof of lemma [ laplace - int - lem ] .
we can assume that the population is sampled sequentially , so that @xmath118 , @xmath119 , are i.i.d .
@xmath120 under @xmath121 , where @xmath122 ; @xmath123 can be viewed as a multinomial @xmath124 vector under @xmath121 , that is , for all integers @xmath119 , @xmath125 it is obvious that @xmath126 .
since for any @xmath127 , @xmath128 we have that @xmath129 and if @xmath130 , then @xmath131 . without loss of generality , we can assume that @xmath132.\ ] ] otherwise , we consider subsequence
. the proof of theorem [ main - thm - mdp ] will be divided into two cases , @xmath133 now let us introduce the structrue of the proofs of main results . in section [ sec3.1 ] ,
we give several moment estimates and exponential moment inequalities which are basic for studying the moderate deviations for the good estimator . a truncation method and some related estimates are also presented in the subsection .
the proofs of cases i and ii of theorem [ main - thm - mdp ] are given , respectively , in sections [ sec3.2 ] and [ sec3.3 ] . in section [ sec3.4 ] , we prove theorem [ main - thm - mdp - self ] .
the proofs of several technique lemmas are postponed to the . for any @xmath134 and @xmath135 , set @xmath136 and @xmath137 [ truncation - lem - l ] if @xmath138 , then for any positive sequence @xmath139 with @xmath39 , @xmath140 in particular , condition ( [ mdp - lindeberg - condition ] ) is valid .
similarly to remark [ rmk-2 - 1 ] , for any @xmath41 , @xmath141 therefore , ( [ truncation - lem - l - eq-1 ] ) holds . from lemma 1 in @xcite ,
under conditions ( [ moment - condition-1 ] ) and ( [ moment - condition-2 ] ) , @xmath142 and if @xmath138 , then @xmath143 . [ sn - comparison - lem-1 ] assume that ( [ mdp - lindeberg - condition ] ) holds . if @xmath144 and @xmath145 then @xmath146 set @xmath147 .
then for any @xmath148 , for @xmath2 large enough , @xmath149 therefore , by ( [ mdp - lindeberg - condition ] ) , the above inequality implies that @xmath150 as @xmath151 . on the other hand ,
it is clear that for any @xmath148 , when @xmath2 is large enough , @xmath152 which yields that @xmath153 .
thus ( [ sn - comparison - lem-1-eq-1 ] ) is valid .
[ truncation - lem - rho ] for any @xmath135 , @xmath154 since @xmath155 holds uniformly on @xmath156 for @xmath157 , we obtain that @xmath158 that is , ( [ truncation - lem - rho - eq-1 ] ) holds . in order to obtain the exponential moment inequalities ,
we need some concepts of negative dependence ; cf .
@xcite , @xcite .
let @xmath159 be real random variables .
@xmath159 are said to be negatively associated if for every two disjoint index finite sets @xmath160 , @xmath161 for all nonnegative functions @xmath162 and @xmath163 that are both nondecreasing or both nonincreasing .
[ negative - dep - lem
] @xmath164 is a sequences of negatively associated random variables , and for each @xmath165 @xmath166 is also negatively associated .
let @xmath167 denote the frequency of the @xmath1th species in the @xmath168th sampling , that is , @xmath169 then @xmath170 are zero - one random variables such that @xmath171 . by lemma 8 in @xcite , @xmath170 , are negative associated . since @xmath172 , @xmath173 , are i.i.d . under @xmath121 , @xmath174 , @xmath175 , are negative associated . noting that @xmath176 and @xmath177 where @xmath178}(x)$ ] is a decreasing function , we obtain that @xmath179 and @xmath180 are two sequences of negatively associated random variables .
[ exp - moment - inq - lem-1 ] let @xmath181 be a subset of the set @xmath182 of positive integers .
then for any @xmath183 , @xmath184 and for any @xmath185 , @xmath186\\[-8pt ] & & \qquad\leq\prod_{k\in m } \biggl ( \bigl(e^r-1 \bigr)\sum_{l=0}^j \frac { n!}{(n - l)!l!}p_{kn}^l ( 1-p_{kn})^{n - l}+1 \biggr ) . \nonumber\end{aligned}\ ] ] for any @xmath183 given , set @xmath187 , @xmath188 .
then , when , all @xmath189 , @xmath190 are nonnegative and increasing ; when @xmath191 , all @xmath189 , @xmath190 are nonnegative and decreasing
. therefore , by lemma [ negative - dep - lem ] , @xmath192 similarly , we can obtain ( [ exp - moment - inq - lem - eq-2 ] ) . as applications of lemma [ exp - moment - inq - lem-1 ] , we have the following exponential moment estimates .
its proof is given in appendix [ sec5 ] .
[ exp - moment - estimates - lem-1 ] ( 1 ) for any @xmath193 and @xmath183 , @xmath194\\[-10pt ] & & \qquad{}\times\log e_n \biggl(\exp \biggl(\frac{ra^2(b(n ) ) } { b^2(n)}\sum _ { k\in m_{n\varrho}^{c } } \sum_{l=0}^j \bigl(\delta_{kl}(n)-e_n\bigl(\delta_{kl}(n ) \bigr)\bigr ) \biggr ) \biggr ) \leq0.\nonumber\end{aligned}\ ] ] \(2 ) if @xmath138 , then for any @xmath193 and @xmath183 , @xmath195\\[-10pt ] & & \qquad{}\times\log e_n \biggl(\exp \biggl(\frac{ra(b(n ) ) } { b(n)}\sum_{k\in m_n^{lc } } \sum _ { l=0}^j\bigl(\delta_{kl}(n)-e_n \bigl(\delta_{kl}(n)\bigr)\bigr ) \biggr ) \biggr)\leq0\nonumber\end{aligned}\ ] ] and @xmath196\\[-10pt ] & & \qquad{}\times\log e_n \biggl(\exp \biggl(\frac{r na(b(n ) ) } { b(n)}\sum_{k\in m_n^{lc } } p_{kn } \bigl(\delta_{k0}(n)-e_n\bigl(\delta_{k0}(n ) \bigr)\bigr ) \biggr ) \biggr ) \leq0.\nonumber\end{aligned}\ ] ] in this subsection , we use the grtner ellis theorem to show theorem [ main - thm - mdp ] under @xmath138 .
the laplace asymptotic integral plays a very important role . by lemma [ truncation - lem - l ] , if @xmath197 , when @xmath198 is large enough , @xmath199 and @xmath200 where @xmath201 . in this subsection , we assume that @xmath198 is large enough such that @xmath202 and @xmath203 .
the following laplace asymptotic integral is a key lemma .
it will be proved in appendix [ sec5 ] .
[ laplace - int - lem]suppose that conditions ( [ moment - condition-1 ] ) and ( [ moment - condition-2 ] ) hold . if @xmath204 , then for any @xmath205 , @xmath206\\[-10pt ] & & \qquad{}\times\log e_n \biggl(\exp \biggl\{\frac{\alpha a(b^l(n ) ) } { b^l(n ) } \sum _ { k\in m_n^l}\bigl(\delta_{k1}(n)-np_{kn}\delta _ { k0}(n)\bigr ) \biggr\ } \biggr)= \frac{\alpha^2}{2}.\nonumber\end{aligned}\ ] ] proof of theorem [ main - thm - mdp ] under @xmath138 by the grtner ellis theorem [ cf .
theorem 2.3.6 in @xcite ] and lemma [ laplace - int - lem ] , @xmath207 satisfies a large deviation principle with speed @xmath208 and with rate function @xmath209 . by lemma [ exp - eqv - lemma-1 ]
, we only need to check @xmath210 it is obvious that @xmath211 from ( [ truncation - bn - eq-2 ] ) and @xmath212 satisfies the large deviation principle , we obtain that for any @xmath148 , @xmath213 by lemma [ exp - moment - estimates - lem-1 ] and chebyshev s inequality , we have that for any @xmath148 , @xmath214 and for @xmath215 , @xmath216 which implies that for any @xmath148 , @xmath217 now , ( [ main - thm - case - i - eq-0 ] ) follows from ( [ main - thm - case - i - eq-2 ] ) and ( [ main - thm - case - i - eq-3 ] ) . therefore , the conclusion of theorem [ main - thm - mdp ] holds under @xmath138 .
in this subsection , we show theorem [ main - thm - mdp ] under @xmath218 . in this case , since @xmath219 can not be truncated as a uniformly bounded sequence , the asymptotic analysis techniques in the first case can not be used .
the proof of this case is based on the poisson approximation [ cf .
@xcite ] and the truncation method .
let first us introduce the poissonization defined by @xcite .
define @xmath220 let @xmath221 be a poisson process independent of @xmath222 with @xmath223 . define the poissonization @xmath224 of @xmath225 as follows : @xmath226 under probability @xmath121 , @xmath227 are independent poisson variables with means @xmath228 , so that @xmath229 are independent zero - mean variables with variance @xmath230 .
then the poissonization @xmath231 satisfies the following moderate deviation principle .
[ mdp - poisson - approx ] let conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold
. then @xmath232 satisfies a large deviation principle with speed @xmath233 and with rate function . for any @xmath205 , @xmath234 for any @xmath235 $ ] such that @xmath236 , for @xmath2 large enough , we can write @xmath237 by ( [ mdp - lindeberg - condition ] ) , @xmath238 and @xmath239 therefore , by @xmath240 as @xmath241 , we have that @xmath242 which implies the conclusion of the lemma by the grtner ellis theorem ; cf .
theorem 2.3.6 in @xcite . by lemmas [ mdp - poisson - approx ] and [ exp - eqv - lemma - basic ]
, we need the following exponential approximation : for any @xmath20 , @xmath243 let us first give a maximal exponential estimate . its proof is postponed to appendix [ sec5 ] .
[ exp - eqv - lemma-1 ] let conditions ( [ moment - condition-1 ] ) , ( [ moment - condition-2 ] ) and ( [ mdp - lindeberg - condition ] ) hold , and let @xmath218 .
for any @xmath244 fixed , set @xmath245 , @xmath246 .
then for any @xmath20 , @xmath247}|\zeta_{\lambda_n n}- \zeta_{t n}|\geq\varepsilon a\bigl(b(n)\bigr ) \bigr)=-\infty.\ ] ] proof of theorem [ main - thm - mdp ] under @xmath218 by lemmas [ mdp - poisson - approx ] and [ exp - eqv - lemma - basic ] , we only need to prove ( [ exp - eqv - lemma-2-eq-1 ] ) .
set @xmath248 .
then @xmath51 has gamma@xmath249 distribution and @xmath250 .
therefore , for any @xmath20 and any @xmath244 , @xmath251}|\zeta _
n- \zeta _ { t n}|\geq\frac{\varepsilon a(b(n))}{2 } \biggr).\end{aligned}\ ] ] by lemma [ exp - eqv - lemma-1 ] , @xmath252}|\zeta_n- \zeta_{t n}|\geq \frac{\varepsilon a(b(n))}{2 } \biggr)=-\infty.\hspace*{-34pt}\ ] ] by chebyshev s inequality , it is easy to get that @xmath253 therefore , we only need to prove that @xmath254 it is sufficient that for any @xmath48 , @xmath255 in fact , by lemma [ exp - moment - inq - lem-1 ] , we can get that for any @xmath48 , @xmath256 where @xmath257 , which implies that ( [ exp - eqv - lemma-2-eq-1 ] ) holds . by the comparison method in large deviations [ cf . theorem 4.2.13 in @xcite ] , in order to obtain theorem [ main - thm - mdp - self ] , we need the following lemma .
[ main - thm - self - lem ] for any @xmath20 , for @xmath157 , @xmath258 by ( [ exp - moment - estimates - lem-1-eq-1 ] ) , for @xmath157 , for any @xmath135 and @xmath20 , @xmath259 therefore , by lemma [ truncation - lem - rho ] , it suffices to show that @xmath260 & & \quad{}\times\log p_n \biggl ( \frac{1}{b(n ) } \biggl|\sum _ { k\in m_{n\varrho } } \biggl(\delta_{kj}(n)- \frac{1}{j!}(np_{kn})^j e^{-np_{kn } } \biggr ) \biggr|\geq
\varepsilon \biggr)\\[-2pt ] & & \qquad=-\infty .
\nonumber\end{aligned}\ ] ] now , let us show ( [ main - thm - self - lem - eq-3 ] ) . using the partial inversion formula for characteristic function due to @xcite
[ see also @xcite , @xcite ] , for any @xmath183 , @xmath261
\hspace*{-4pt}&&\qquad= \frac{n!}{2\pi n^ne^{-n}}\int_{-\pi}^\pi\prod _ { k\in m_{n\varrho}^c } e_n \bigl(\exp \bigl\{iu \bigl(y_k(n)-np_{kn}\bigr ) \bigr\ } \bigr ) \\[-2pt ] \hspace*{-4pt}&&\hspace*{60pt}\qquad\quad{}\times \prod_{k\in m_{n\varrho } } e_n \biggl(\exp \biggl\{iu \bigl(y_k(n)-np_{kn}\bigr)\\[-2pt ] \hspace*{-4pt}&&\qquad\hspace*{147pt } { } + r \biggl(i_{\{y_k(n ) = j\}}-\frac{1}{j!}(np_{kn})^j e^{-np_{kn } } \biggr ) \biggr\ } \biggr)\,du,\end{aligned}\ ] ] where @xmath262 are independent random variables and @xmath263 is poisson distributed with mean @xmath264 .
let @xmath265 be defined as in the proof of lemma [ laplace - int - lem ] , that is , @xmath266 set @xmath267 then for any @xmath205 , @xmath268 set @xmath269 .
then @xmath270 , and noting that @xmath271 , @xmath272 we obtain that for @xmath273 small enough , @xmath274 } \biggl{\vert}e^{n(e^{i u}-1-iu ) } \prod_{k\in m_{n\varrho } } \vartheta _ k(u,\alpha)\biggr{\vert}\biggr ) \\ & & \qquad\leq - \frac { b(n)n \tau^2(n)}{a^2(b(n ) ) } \biggl(1+o \biggl(\frac { \log n } { n\tau^2(n ) } \biggr)+o ( \varrho ) \biggr ) \to-\infty.\end{aligned}\ ] ] since @xmath275}\sup_{k\in m_{n\varrho } } |np_{kn}(1-\cos u)|\leq\varrho$ ] , on @xmath276 $ ] , @xmath277 thus @xmath278 and so @xmath279 this yields that ( [ main - thm - self - lem - eq-3 ] ) holds .
proof of theorem [ main - thm - mdp - self ] by lemma [ main - thm - self - lem ] , for any @xmath20 , @xmath280 now , by @xmath281 and the elementary inequality @xmath282 for all @xmath283 , we obtain that @xmath284 therefore , the conclusion of the theorem follows from lemma [ exp - eqv - lemma - basic ] or theorem 4.2.13 in @xcite .
for the sake convenience , let us introduce some notions in large deviations [ @xcite ] .
let @xmath285 be a metric space .
let @xmath286 , @xmath287 be a sequence of probability spaces and let @xmath288 be a sequence of measurable maps from @xmath289 to @xmath290 .
let @xmath38 be a sequence of positive numbers tending to @xmath291 , and let @xmath292 $ ] be inf - compact ; that is , @xmath293 $ ] is compact for any @xmath294 .
then @xmath295 is said to satisfy a large deviation principle ( ldp ) with speed @xmath296 and with rate function @xmath297 , if for any open measurable subset @xmath298 of @xmath290 , @xmath299 and for any closed measurable subset @xmath300 of @xmath290 , @xmath301 assume that @xmath295 satisfies @xmath302 in law and a fluctuation theorem such as central limit theorem , that is , there exists a sequence @xmath303 such that @xmath304 in law , where @xmath305 is a constant and @xmath306 is a nontrivial random variable .
usually , @xmath307 is said to satisfy a moderate deviation principle ( mdp ) if @xmath308 satisfies a large deviation principle , where @xmath309 is an intermediate scale between @xmath310 and @xmath311 , that is , @xmath115 and @xmath312 . in this paper ,
the following exponential approximation lemma is required .
it is slightly different from theorem 4.2.16 in @xcite .
[ exp - eqv - lemma - basic ] let @xmath288 and @xmath313 , @xmath134 be sequences of measurable maps from @xmath289 to @xmath290 .
assume that for each @xmath134 , @xmath314 satisfies a ldp with speed @xmath315 and with rate function @xmath297 .
if @xmath316 and for any @xmath148 , @xmath317 the @xmath288 satisfies a ldp with speed @xmath296 and with rate function @xmath297 .
set @xmath318 .
for any closed subset @xmath300 , @xmath319 where @xmath320 . by ( [ exp - eqv - lemma - basic - eq-2 ] ) , @xmath321 for large @xmath2 and @xmath198 .
therefore , for large @xmath2 and @xmath198 @xmath322 and so @xmath323 the argument for open sets is similar and is omitted .
in this appendix , we give the proofs of several technique lemmas .
the proofs of lemmas [ exp - moment - estimates - lem-1 ] and [ exp - eqv - lemma-1 ] are based some exponential moment inequalities for negatively associated random variables and martingales .
the refined asymptotic analysis techniques play a basic role in the proof of lemma [ laplace - int - lem ] .
proof of lemma [ exp - moment - estimates - lem-1 ] ( 1 ) by lemma [ exp - moment - inq - lem-1 ] , we have that for any @xmath183 , and @xmath193 , @xmath324 \hspace*{-2pt}&&\qquad\leq\frac{a(b(n))}{b(n)}\frac{r^2a^2(b(n))}{b(n ) } \\[-4pt ] \hspace*{-2pt}&&\qquad\quad{}\times\frac{1}{b(n ) } \sum _ { k\in m_{n\varrho}^{c } } \biggl ( \frac{2}{\varrho } np_{kn}e^{-n p_{kn}}+ \sum_{l=1}^j\frac{n!}{(n - l)!l ! } p_{kn}^l e^{-(n - l)p_{kn } } \biggr).\end{aligned}\ ] ] therefore , ( [ exp - moment - estimates - lem-1-eq-1 ] ) holds .
\(2 ) similarly to the proof of ( [ exp - moment - estimates - lem-1-eq-1 ] ) , we also have that @xmath325 & & \qquad\leq\frac{r^2a^2(b(n))}{b(n ) } \frac{1}{b(n)}\sum_{k\in m_n^{lc } } \biggl ( \frac{2}{l } np_{kn}e^{-n p_{kn}}+ \sum _ { l=1}^j\frac{n!}{(n - l)!l ! }
p_{kn}^l e^{-(n - l)p_{kn } } \biggr)\\[-4pt ] & & \qquad\to0.\end{aligned}\ ] ] finally , let us prove ( [ exp - moment - estimates - lem-1-eq-3 ] ) .
by lemma [ exp - moment - inq - lem-1 ] , for any @xmath326 , @xmath327 & & \qquad\leq\sum_{k\in m_n^{lc } } \biggl ( \log \biggl ( \biggl(\exp \biggl\ { \frac{r na(b(n))}{b(n ) } p_{kn } \biggr\}-1 \biggr ) ( 1- p_{kn})^n+1 \biggr ) \\[-4pt ] & & \hspace*{126.5pt}\qquad\quad{}-\frac{r na(b(n ) ) } { b(n)}p_{kn}(1- p_{kn})^n \biggr).\end{aligned}\ ] ] therefore @xmath327 & & \qquad\leq 4\sum_{k\in m_n^{lc } } \biggl(\frac { r na(b(n))}{b(n ) } p_{kn } \biggr)^2e^{-n p_{kn}}i_{\ { { |r| na(b(n))}p_{kn}/{b(n)}\leq1 \ } } \\[-4pt ] & & \qquad\quad { } + 12\sum_{k\in m_n^{lc } } \exp \biggl\{\frac{2|r| na(b(n))}{b(n ) } p_{kn } \biggr\}e^{-n p_{kn}}i_{\ { { |r| na(b(n))}p_{kn}/{b(n)}\geq1 \ } } \\[-4pt ] & & \qquad\leq \frac{4 r^2 a(b(n))}{b(n ) } a_{nl}+ 24 |r| a_{n},\vadjust{\goodbreak}\end{aligned}\ ] ] where @xmath328 , @xmath329 and @xmath330 by the proof of lemma [ truncation - lem - l ] , @xmath331 . by ( [ truncation - lem - l - eq-1 ] ) , @xmath332 therefore , ( [ exp - moment - estimates - lem-1-eq-3 ] ) holds .
proof of lemma [ laplace - int - lem ] it is known that @xmath333 where @xmath262 are independent random variables , and @xmath263 is poisson distributed with mean @xmath264 .
then , using the partial inversion formula for characteristic function due to @xcite [ see also @xcite , @xcite ] , for any @xmath205 , @xmath334 where @xmath335 and @xmath336 it is obvious that @xmath337 . by stirling s formula , @xmath338 it suffices to show that for any @xmath205 , @xmath339 since @xmath340 uniformly in @xmath341 , we can write that for @xmath2 large enough , @xmath342 where @xmath343 . choose a positive function @xmath344 such that @xmath345 and @xmath346 , and define @xmath347 , @xmath348 and then @xmath349,@xmath350 noting that for @xmath2 large enough , @xmath351}(1-\cos u)\geq\tau^2(n)/4 $ ] , we have that @xmath352}\bigl|h_n(u , \alpha)\bigr| \bigr ) \\ & & \qquad\leq \frac { b^l(n)\log n } { 2 a^2 ( b^l(n))}- \frac { b^l(n)n \tau ^2(n ) } { 4 a^2 ( b^l(n))}+\frac { b^l(n ) } { a^2 ( b^l(n))}\sum _ { k\in m_n^l}\log\sup _ { u \in(\tau(n),\pi]}\bigl|h_k(u , \alpha)\bigr| \\ & & \qquad= - \frac { b^l(n)n \tau^2(n)}{a^2(b^l(n ) ) } \biggl(1+o \biggl(\frac { \log n}{n\tau^2(n ) } \biggr)+o \biggl(\frac{a(n)}{n\tau^2(n ) } \biggr ) \biggr ) \to -\infty,\end{aligned}\ ] ] which implies that @xmath353 } n ^{1/2}h_n(u , \alpha ) \,du\biggr{\vert}=-\infty.\ ] ] therefore , it suffices to show that @xmath354 in order to show ( [ laplace - int - lem - eq-4 ] ) , let us define a transformation as follows . for
@xmath205 given , set @xmath355 , and define @xmath356 where @xmath357 denotes the complex plane .
the transformation plays an important role .
let @xmath358 denote the closed path formed by the ordered points @xmath359 , @xmath360 , @xmath361 , @xmath362 , @xmath359 on the complex plane .
then by cauchy s formula , @xmath363 noting that @xmath364 , by @xmath365 and @xmath366 , similarly to the proof of ( [ laplace - int - lem - eq-3 ] ) , we have that @xmath367 similarly , @xmath368 .
therefore , it suffices to prove that @xmath369 let @xmath370 and @xmath371 denote the real part and the imaginary part of a complex number @xmath372 , respectively .
then @xmath373 and @xmath374 for convenience , let @xmath375 , @xmath185 , denote uniformly bounded real functions such that @xmath376 for all @xmath377 , and @xmath378 .
then for @xmath2 large enough , for all @xmath379 $ ] , @xmath380 and @xmath381 therefore @xmath382 and so @xmath383 now , by @xmath384 we obtain ( [ laplace - int - lem - eq-5 ] ) .
the proof of lemma [ laplace - int - lem ] is complete .
proof of lemma [ exp - eqv - lemma-1 ] for any @xmath385 , we can write [ cf .
( a.1 ) in @xcite ] @xmath386 therefore , it suffices to prove that @xmath387 and @xmath388\\[-8pt ] & & \qquad\hspace*{242pt}\geq \varepsilon a\bigl(b(n)\bigr ) \biggr)\nonumber\\ & & \qquad=-\infty .
\nonumber\end{aligned}\ ] ] let us first prove ( [ exp - eqv - lemma-1-eq-3 ] ) .
set @xmath389 and @xmath390 .
since @xmath391 are independent variables with mean zero and independent of @xmath392 , @xmath393 is a martingale , and by the maximal inequality for supermartingales , we have that for any @xmath20 , for any @xmath48 , @xmath394 and @xmath395 for any @xmath396 , take @xmath397 . then for @xmath2 large enough , @xmath398 therefore , for @xmath2 large enough , @xmath399 where @xmath400 @xmath401 and @xmath402 by ( [ mdp - lindeberg - condition ] ) , @xmath403 then , by @xmath404 under @xmath218 , @xmath405 and @xmath406 , @xmath407 and @xmath408 thus @xmath409 this yields ( [ exp - eqv - lemma-1-eq-3 ] ) by chebyshev s inequality .
next , we show ( [ exp - eqv - lemma-1-eq-4 ] ) . noting that @xmath410 it suffices to show that for any @xmath20 , @xmath411\\[-8pt ] & & \qquad=-\infty\nonumber\end{aligned}\ ] ] and @xmath412 since @xmath413 a similar argument to the proof of ( [ exp - eqv - lemma-1-eq-5 ] ) gives @xmath414 which implies ( [ exp - eqv - lemma-1-eq-6 ] ) .
similarly , we can obtain ( [ exp - eqv - lemma-1-eq-7 ] ) .
the author is very grateful to the editor , professor t. cai , the associate editor and an anonymous referee for their helpful comments and suggestions .
the author is also thankful to the referee for recommending the reference @xcite . | in this paper , we consider moderate deviations for good s coverage estimator . the moderate deviation principle and the self - normalized moderate deviation principle for good s coverage estimator
are established .
the results are also applied to the hypothesis testing problem and the confidence interval for the coverage . |
the redshift evolution of the stellar mass function is a fundamental observable of galaxy evolution , as it directly measures the stellar mass buildup of galaxies .
recently , it has been measured with unprecedented precision by a number of wide , deep near - infrared surveys ( muzzin et al .
2013 ; ilbert et al .
2013 ; moustakas et al . 2013 ; davidzon et al . 2013
; tomczak et al .
these surveys are revealing new frontiers in galaxy evolution ; for example , the fourstar galaxy evolution survey ( zfourge ) is mass - complete to @xmath13 at @xmath11 , for both star - forming and quiescent galaxies ( tomczak et al .
this impressive depth has revealed that simple schechter fits are not a good representation of the mass function in the high - redshift universe . the observed relationship between the star formation rate ( sfr ) and stellar mass of star - forming galaxies ( the star - forming sequence " ) , measuring the derivative of the mass buildup of galaxies
, has also been of great interest in the literature ( brinchmann et al .
2004 ; noeske et al . 2007 ;
daddi et al .
2007 ; peng et al . 2010 ; karim et al .
2011 ; whitaker et al . 2012
; guo , zheng , & fu 2013a ; speagle et al .
recently , this has been measured robustly out to high redshift and over a wide variety of sfr indicators ( oliver et al .
2010 ; wuyts et al .
2011 ; karim et al . 2011
; whitaker et al . 2012 ) , and further extended down to low stellar masses and star formation rates in the local universe ( huang et al .
much work has been done to bring these different sfr indicators into agreement ( wuyts et al . 2011 ) , and there has been some success in putting these many studies into a consistent framework ( speagle et al .
2014 ) .
while there has been great progress in reducing errors due to limited depth and field - to - field variations , the interpretation of these data is still subject to systematic uncertainties .
stellar mass measurements , particularly at higher redshifts , are uncertain by at least a factor of two , due to unknowns such as emission line contributions , star formation histories , dust content , and metallicities ( marchesini et al .
2009 ; conroy , gunn , & white 2009 ; behroozi , conroy , & wechsler 2010 ; mitchell et al .
2013 ) . while in principle
these stellar mass measurements can be calibrated with dynamical masses , only a handful of galaxies at @xmath14 have reliable stellar velocity dispersions so far ( van de sande et al .
2013 ; belli et al .
star formation rates also suffer from uncertainties , particularly in low - mass galaxies at high redshift , due to flux limits , systematic differences between sfr indicators , and selection effects ( mitchell et al .
2014 ; speagle et al .
2014 ) . given these systematic uncertainties , it is important to test whether the observed star - forming sequence and the observed evolution of the mass function are consistent with one another .
a version of this phenomenological test was applied by bell et al .
( 2007 ) , and has been implicitly performed by several studies since ( drory & alvarez 2008 ; peng et al .
2010 ; behroozi , wechsler , & conroy 2013c ) .
most recently , weinmann et al .
( 2012 ) has showed that in order for the star - forming sequence to be consistent with the number density evolution of low - mass galaxies at @xmath15 , either ( 1 ) for sfr @xmath0 m@xmath16 , @xmath3 must be greater than 0.9 , or ( 2 ) the rate of destruction by mergers must be substantial .
this analysis was limited to low redshift , however , thus missing the peak of the cosmic star formation rate density at @xmath17@xmath122 .
extention of this analysis to higher redshift requires accurate measurements of the number density of low - mass quiescent galaxies , which have only recently been made possible by the ultra - deep zfourge mass functions ( tomczak et al . 2014 ) .
we use the zfourge mass functions to take a fresh look at the consistency between the stellar mass function and the star - forming sequence between @xmath18 .
we compare this to the star - forming sequence from whitaker et al .
( 2012 ) , which has been mapped with deep medium - band nir imaging and consistent uv+ir sfr indicators between @xmath19 .
the layout of the paper is as follows . in section 2 ,
we construct a smooth analytical description of the redshift evolution of the stellar mass function . in section 3 ,
we build a model to compare the growth of the mass function implied by different low - mass extrapolations of the star - forming sequence to the observed growth of the mass function , and based on this comparison , postulate a new functional form for the star - forming sequence .
section 4 discusses the implications of our results and the remaining uncertainties , and the conclusion is in section 5 .
we use a standard @xmath20cdm cosmology , with @xmath21 = 0.3 , @xmath22 = 0.7 , and @xmath23 , and use a chabrier initial mass function ( imf ) .
we adopt mass functions from the zfourge survey , measured between @xmath2 ( tomczak et al .
zfourge is the deepest measurement of the stellar mass function to date , and makes use of ground - based near - infrared medium - bandwidth filters which improve the accuracy of photometric redshifts ( van dokkum et al .
the survey also incorporates _ hst _ imaging from the cosmic assembly near - infrared deep extragalactic legacy survey ( candels ) ( grogin et al .
2011 ; koekemoer et al .
zfourge use the candels @xmath24 filter as their detection criteria .
zfourge imaging covers an area of 316 arcmin@xmath25 , and the zfourge catalogs include data from the wider but shallower newfirm medium - band survey ( nmbs ; whitaker et al .
2011 ) to tighten constraints at the massive end .
star - forming and quiescent galaxies are separated by their rest - frame uvj colors as described in williams et al .
( 2009 ) .
we supplement the zfourge mass functions with low - redshift mass functions measured from sdss-@xmath26 data ( moustakas et al . 2013 ) .
these cover the redshift range @xmath27 , and have @xmath28 photometry and spectroscopic redshifts from the sdss data release 7 ( abazajian et al .
they include @xmath29 photometry from the 2mass extended source catalog ( jarrett et al .
2000 ) and photometry at 3.4 and 4.6 @xmath30 m from the wise all - sky data release ( wright et al .
star - forming and quiescent galaxies are separated via uv luminosity as measured by the _ galaxy evolution explorer _ ( @xmath26 ; martin et al .
2005 ) .
we use these data to construct an analytical description of the evolution of the stellar mass function with time .
we aim to smoothly parameterize the redshift evolution of the stellar mass function as @xmath31 .
historically , the stellar mass function has been fit with a schechter function ( schechter 1976 ) .
recently , however , deeper measurements of the stellar mass function have shown that a double schechter function is necessary to describe the steepening of the slope at masses below @xmath32 m@xmath33 ( baldry , glazebrook , & driver 2008 ; li & white 2009 ; drory et al .
2009 ; moustakas et al . 2013 ; muzzin et al . 2013
; ilbert et al . 2013 ; tomczak et al .
the logarithmic form of the double schechter function is : @xmath34 dm { \addtocounter{equation}{1}\tag{\theequation}}\end{aligned}\ ] ] where , in this equation , m = log(m / m@xmath33 ) and m@xmath35 = log(m@xmath35/m@xmath33 ) . while in principle
each component of the double schechter could have a separate m@xmath35 , the data are consistent with a single value of m@xmath35 for each component of the double schechter function .
it is difficult to simply use the best - fit schechter parameters to construct a smooth model for the evolution of the mass function , however , due to degeneracy between parameters in a double schechter function : an unconstrained chi - squared minimization leads to schechter parameters which do not evolve smoothly with redshift .
we have found that simply interpolating in redshift between the best - fit schechter parameters will introduce spurious increases and decreases in number density between observed redshift windows . to avoid these numerical artifacts , we re - fit constrained double schechter functions separately to the observed star - forming , quiescent , and total mass functions in each redshift window .
we limit degenerate solutions by fixing the faint - end slopes to the best - fit values at @xmath36 from tomczak et al .
( 2014 ) , which are : @xmath37 we follow drory & alvarez ( 2008 ) in fitting second - order polynomials to the redshift evolution of the remaining best - fit schechter parameters , namely @xmath38 , @xmath39 , and @xmath40 .
we then re - fit @xmath41 and @xmath40 , constraining them to be within 40% of the best - fit second - order polynomials .
we perform this fit iteratively , constraining the best - fit @xmath38 , @xmath39 , and @xmath40 to be increasingly close to the second - order polynomials during each fit .
the goals of this procedure are to : ( 1 ) reproduce the evolution of the observed mass function in the observed redshift windows , and ( 2 ) to enforce smooth , monotonic evolution of the mass function between observed redshift windows .
this iterative approach maximizes the effectiveness of the polynomial fits in building a smooth model for redshift evolution of the stellar mass function , though we recognize that it does not necessarily guarantee a unique solution .
the resulting redshift evolution of the schechter parameters is : @xmath42 as we find no significant redshift evolution in the quiescent @xmath40 , we fix it to its average value . the parameterized mass function growth , @xmath31 , is compared directly with the measured zfourge mass functions in figure [ schechparms ] .
there is overall good agreement with the observed mass function .
noticeably , the observed number density of low - mass quiescent galaxies is underpredicted by @xmath43 dex at @xmath36 and overpredicted by @xmath44 dex at @xmath45 .
the data show negligible evolution in the number density of low - mass quiescent galaxies between these two redshift windows
. however , there is still significant evolution from @xmath6 to @xmath46 , and from @xmath47 to @xmath48 , and the smooth evolution in the model reflects the broader trend of declining number density in low - mass quiescent galaxies with increasing redshift . in practice
, this offset has a negligible effect on our conclusions : the quiescent mass function is only used to calculate the fraction of star - forming galaxies as a function of mass , and star - forming galaxies dominate at low masses regardless of the normalization offset . in the following analysis , we extrapolate the stellar mass function to below the nominal stellar mass completeness when necessary .
the assumed shape of the stellar mass function below the observed completeness limit does not have a significant effect on our results , however .
we now compare the evolution of the mass function , as implied by the star - forming sequence , to the observed evolution of the mass function , as parameterized in section [ massmod ] .
we begin by describing the model which connects the star - forming sequence to the growth of the mass function .
we also include a simple model for the effect of galaxy mergers on the evolution of the mass function .
we then examine the effects of applying different star - forming sequences to the mass function .
we implement a model to describe the time evolution of the mass function due to the observed star - forming sequence .
a schematic of this model is shown in figure [ schematic ] .
we also include a simple model for galaxy mergers . at each redshift step
, mass is added to the mass function by star formation .
star formation will cause the mass function to shift to the right " with time . at fixed mass ,
galaxies have a range of star formation rates ( van dokkum et al .
however , when describing the evolution of the mass function , the complex distribution of star formation rates can be reduced to the average star formation rate as a function of mass ( discussed further in section [ scatter ] ) .
in general , each bin of the mass function contains a mix of star - forming and quiescent galaxies . adopting this division ,
the rate of mass addition due to star formation in each bin of the mass function is given by : @xmath49 with @xmath50 the average star formation rate of star - forming galaxies and @xmath51 the average star formation rate of quiescent galaxies .
we assume that quiescent galaxies have a negligible rate of star formation and set @xmath51 to zero .
this assumption is justified by observations indicating that uvj - selected quiescent galaxies have average star formation rates that are at least 20 - 40 times lower than uvj - selected star - forming galaxies , and thus can be safely neglected ( fumagalli et al . 2013 ) .
we define @xmath52 , the fraction of star - forming galaxies , to be the following : @xmath53 we calculate @xmath54 directly from our smoothed model for the evolution of the mass function , and fit it with : @xmath55\ ] ] the fit is restricted to @xmath56 , where the mass function is mass - complete and the effects of cosmic variance are minimized at the high - mass end ( tomczak et al .
2014 ) . by construction ,
the adopted star - forming fraction asymptotes at high masses to @xmath57 , consistent with studies of bcgs in the local universe ( bauer et al .
2013 ; oliva - altamirano et al .
2014 ) . at low masses , the star - forming fraction asymptotes to @xmath58 ;
the typical best - fit value of @xmath58 is @xmath59 .
we take @xmath50 to be : @xmath60 @xmath61 is the star formation rate implied by the star - forming sequence .
@xmath62 is the fraction of mass ejected from a stellar population during the course of passive stellar evolution , primarily due to winds and outflows . in general ,
@xmath63 , starting at @xmath64 when the stellar population is formed and ending at @xmath65 after @xmath66 gyr ( assuming a chabrier imf )
. however , as most of the mass loss occurs within the first hundred million years , we approximate mass loss as instantaneous , fixing @xmath62 to @xmath67 ( see section [ mloss ] for further discussion of this approximation ) . the final equation for the average growth rate of the mass function due to
star formation is thus : @xmath68 the framework described so far models the growth of the mass function solely due to star formation . the other physical process that can change the number density of galaxies is galaxy mergers .
specifically , mergers affect the mass function in two ways : ( 1 ) they directly decrease the number density of galaxies , and ( 2 ) they contribute to the stellar mass growth of galaxies .
we model both of these effects .
we measure the merger rate measured directly from the guo et al .
( 2013b ) semi - analytical model ( sam ) based on the millenium - ii simulation ( boylan - kolchin et al .
2009 ) . at each snapshot in redshift
, we measure the rate at which galaxies merge with more massive galaxies than themselves , which is the rate of destruction " by mergers .
this is shown as a function of mass in figure [ samdirect ] in appendix a. for low - mass galaxies , this rate varies by @xmath69 per snapshot in redshift , but broadly declines from @xmath127% per gyr at @xmath14 to @xmath123% per gyr in the local universe .
this is consistent with observed destruction rates of @xmath1210% per gyr ( bridge , carlberg , & sullivan 2010 ; lotz et al .
2011 ) , though other studies which determine merger rates based on galaxy morphology find destruction rates that are 3 - 5 times higher ( conselice 2014 ) .
we show in section [ otherexp ] that our conclusions are robust to changing the destruction rate .
we interpolate in redshift and stellar mass to apply this measured destruction rate directly to the evolution of the mass function . in order to enforce stellar mass conservation ,
we assume that all mergers have a mass ratio of 1:10 , and correspondingly increase the stellar mass of more massive galaxies . as we show below , including mergers changes the evolution of the mass function at low masses by only @xmath70 dex in number density from @xmath71 to @xmath72 . in appendix a , we show that instead using the measured mass ratios from the guo et al .
( 2013b ) sam has little effect on our results .
we note that using measured mass ratios from the guo et al .
( 2013b ) sam no longer enforces mass conservation in our model , as the stellar mass functions within the semi - analytical model do not match those from observations . in this section ,
we compare the effects of using three different star - forming sequences to grow the mass function .
we start the simulation at @xmath71 with the observed mass function . in each time step
, we then evolve the mass function according to the model described in section [ mgrowthmod ] , and compare it to the observed mass function .
the resulting growth of the mass function , with and without mergers , is illustrated in figure [ mainplot ] , along with the corresponding star - forming sequences .
we describe the adopted star - forming sequences and their corresponding effect on the evolution of the mass function below . .
mass loss is taken to be a constant factor @xmath65 for consistency with the rest of the study . for flat low - mass slopes ,
low - mass galaxies grow at an extremely rapid rate at @xmath11 , inconsistent with the observed growth of the mass function . ]
we first use the observed star - forming sequence from whitaker et al .
( 2012 ) to model the growth of the stellar mass function .
whitaker et al .
( 2012 ) measure stellar masses and photometric redshifts from nmbs , which combines photometry in 5 near - infrared medium - band filters with publicly - available imaging from 0.15 - 8@xmath30 m ( whitaker et al . 2011 ) .
they use a combination of @xmath73-mips and uv imaging to determine sfrs in a uniform manner between @xmath19 .
star - forming galaxies are defined by their rest - frame uvj colors .
we note that the zfourge stellar mass function uses the same color - color cut and also uses the nmbs imaging in derivation of their stellar masses .
thus , the mass function and the star formation rates are measured , to a large extent , from the same galaxies , which will minimize the effect of systematic differences in stellar masses between these two data sets .
whitaker et al .
( 2012 ) provide the following single power law fit to the median dex higher than the median .
the observed star formation rates used in our model may be slightly too low , if the measured average is closer to the true average than the measured median . ]
sfr of star - forming galaxies : @xmath74 with @xmath75 and @xmath76 . due to incompleteness at low masses and low star formation rates ,
the slope of the star - forming sequence at low stellar mass is not well - constrained in this study . for reference ,
the mass - complete limit for nmbs is @xmath77 at @xmath71 and @xmath78 at @xmath79 , and the sfr - complete limit for whitaker et al .
( 2012 ) is @xmath80 m@xmath33/yr at @xmath71 and @xmath81 m@xmath33/yr at @xmath79 .
we first assume that the slope does not change as a function of mass . in the following subsection
, we explore the effects of altering this assumption .
this extrapolation of the observed slope results in a dramatic rate of growth for low - mass galaxies , visible in figure [ mainplot ] .
by @xmath11 , only @xmath82 myr after the simulation has started , the number density of galaxies with @xmath83 = 9 is overpredicted by a factor of 100 .
this offset decreases as stellar mass increases : at @xmath83 = 10.5 , the offset is only a factor of 2 .
the dramatic growth of these galaxies is a result of the relatively flat slope of the star - forming sequence . at @xmath11 ,
the observed slope of the star - forming sequence is 0.44 , meaning that the naive mass - doubling time scales as @xmath84 . with the observed normalization of the star - forming sequence ,
a galaxy at @xmath83= 8 has a doubling time of just 21 million years .
the very rapid assembly of the mass function is thus a direct result of the implied high star formation rates at low masses .
when mergers are turned on , the rapid growth of low - mass galaxies causes a corresponding strong growth in high - mass galaxies .
the merger growth rate is very rapid , such that by @xmath6 , there is actually a deficit of high - mass galaxies , as they have all grown to @xmath85 m@xmath33 .
this smooths out by @xmath72 , as the lower - mass galaxies grow due to star formation and mergers .
this implied rapid growth of low - mass galaxies is consistent with previous analyses from the literature ( drory & alvarez 2008 ; weinmann et al .
weinmann et al .
( 2012 ) examines the difference in the evolution of low - mass galaxies between observations and hydrodynamical+semi - analytical models of galaxy evolution . in the process of this comparison , they model the evolution of the mass function between @xmath86 by varying the slope of the star - forming sequence .
they conclude that the number density of galaxies with masses between 9.27 @xmath87 9.77 is consistent with the observed number density if the slope of the star - forming sequence is steeper than @xmath59 , though their model does not include corrections for quiescent galaxies .
they suggest that either high merger rate or a steep star - forming sequence can reconcile the mass function with the star - forming sequence .
drory & alvarez ( 2008 ) use @xmath88-band selected mass functions and average uv star formation rates to estimate the galaxy merger rates , calculated by subtracting the evolution of the mass function due to star formation from the observed evolution of the mass function .
they conclude that the net merger rate of low - mass galaxies must be high to be consistent with the growth of the mass function : 14% per 100 myr at @xmath89 .
they note that this result is sensitive to the slope of the star - forming sequence ( they adopt a slope of 0.6 ) and to completeness estimates at low stellar mass . the merger rate calculated from the guo et al .
( 2013b ) sam for galaxies with @xmath90 is only @xmath12 0.5 - 0.7% per 100 myr , more than order of magnitude lower than the net merger rate estimated in drory & alvarez ( 2008 ) .
application of the sam merger rate to the evolution of the mass function is indicated by the dashed lines in figure [ mainplot ] .
this rate of galaxy mergers can not compensate for the rapid growth of low - mass galaxies implied by the observed star - forming sequence .
we next explore the effect of modifying the slope at low stellar masses .
we postulate a simple broken power law : @xmath91 with the high - mass slope and normalization fixed to the whitaker et al .
( 2012 ) values .
the low - mass slope is taken to be unity . as can be seen in figure [ mainplot ]
, adopting this star - forming sequence effectively removes the rapid growth of low - mass galaxies .
this is due to a steeper low - mass slope , which in turns implies longer mass - doubling timescales ( @xmath92 ) for low - mass galaxies .
this is illustrated in figure [ tdouble ] , which shows the mass - doubling timescale at @xmath11 as a function of low - mass slope in equation [ bpl_eqn ] . extrapolating the observed high - mass slope of 0.44 at @xmath11 to low masses results in a doubling time of less than 10 million years for a galaxy of mass @xmath83 = 7 , compared to @xmath93 myr for a slope of unity .
thus , adopting a slope of unity at low masses results in evolution of the mass function which compares much more favorably with the observed mass function .
however , there remains an offset from the observed mass function of @xmath94 dex at all masses .
this offset originates at @xmath11 and persists to lower redshifts .
we now generalize the result of section [ bpl_sec ] by constructing a star - forming sequence which exactly reproduces the observed evolution of the mass function . in our simple model with no scatter in star formation rates and a slow or nonexistent merger rate , galaxy populations will evolve along lines of constant number density ( leja et al .
we take a range of evenly - spaced slices in number density between @xmath95 and calculate @xmath96 directly from @xmath31 .
we apply the inverse of the corrections for star - forming fraction and mass loss described above to turn this into a star - forming sequence ( see figure [ schematic2 ] ) .
we then perform a least chi - square fit to the implied star formation rates with the following function : @xmath97 we restrict the fit to masses above the zfourge stellar mass completeness limit , and below @xmath83 = 11 , where star formation is expected to be the dominant mode of stellar mass growth ( drory & alvarez 2008 ; conroy & wechsler 2009 ; leitner 2012 ) .
this model has four parameters : the transition mass @xmath98 , the high - mass slope @xmath99 , the low - mass slope @xmath100 , and the normalization @xmath101 .
as the model is not very sensitive to the location of the transition mass , we fix @xmath102 in all fits .
this lack of sensitivity comes from the fact that , in reality , the slope changes smoothly with mass in the transition region .
however , lacking the theoretical basis for a more physically motivated functional form , we prefer a double power law for simplicity .
the best - fit redshift evolution of these parameters is shown in figure [ parms ] .
in general , at higher redshifts , the mass function prefers a steep low - mass slope and a shallow high - mass slope . at low redshifts , the high - mass and low - mass slopes head towards convergence
this may be a sign that the star - forming sequence locally is well - fit by a single power law down to observed stellar mass completeness limits .
the periodic structure visible in @xmath103 is a result of the zfourge mass completeness limits .
we show in figure [ mainplot ] that this model correctly describes the growth of the observed mass function . there
do exist some small discrepancies . at @xmath72 , the number density of galaxies with masses between @xmath104
is overpredicted by @xmath105 dex .
these low - mass galaxies were below the mass - complete limit at the start of the simulation , and so both their original number densities and their expected growth rates came from fits to galaxies at higher masses , extrapolated to low mass . thus , their low - redshift number densities are not expected to match the observed number densities exactly .
the other noticeable difference is around the transition mass of @xmath106 , where the number density is under - predicted below and over - predicted above by @xmath107 dex .
this is a natural result of fitting a bimodal double - power law to an intrinsically curved distribution of star formation rates : the star formation rates directly below the transition mass are slightly over - predicted , and the star formation rates directly above the transition mass are slightly under - predicted .
a more physically motivated functional form will resolve these differences in the future .
below @xmath108 , the high - mass slope becomes considerably steeper than the low - mass slope .
this is a direct result of the number density of high - mass galaxies increasing substantially at low redshift , an increase which is both intrinsic to the zfourge mass functions and also arises naturally when zfourge is combined with the sdss-@xmath26 mass functions .
we caution that while our model interprets this as an increase in the star formation rate of star - forming massive galaxies at low redshift , this is not seen in observations ( e.g. , noeske et al .
an alternate explanation is that mergers , both major and minor , are important in the evolution of massive galaxies and correspondingly in the evolution of the mass function at the high - mass end ( van dokkum 2005 ; bezanson et al .
2009 ; naab , johansson , & ostriker 2009 ; van dokkum et al .
2010 ; behroozi et al .
2013a ; ferreras et al .
our simple prescription for mergers is insufficient to model the evolution of the high - mass end of the mass function in detail , particularly since while the net effect of mergers on the mass function may be small , the effect on individual galaxies may be quite substantial ( drory & alvarez 2008 ) .
we explicitly test this idea in appendix a by deriving sfrs from the mass function _ including _ mass growth and destruction via mergers .
the low - redshift increase in the high - mass slope then disappears .
instead , the high - mass slope drops rapidly at low redshift , becoming negative .
this is because the growth via mergers is sufficient ( or more than sufficient ) to explain the growth at the high - mass end .
we conclude that in regimes where the growth via star formation is sub - dominant ( i.e. , at low redshifts and high masses ) , the derived slope of the star - forming sequence from the stellar mass function is sensitive to the growth rate via mergers .
this is not the case for the low - mass slope , but is the case for the high - mass slope .
we thus suggest that observational studies are a more robust determinant of the high - mass slope than the analytical model , and do not find significant evidence that the high - mass slope as measured in , e.g. , whitaker et al .
( 2012 ) is incorrect . .
in this study , we have demonstrated that a straightforward extrapolation of the observed star - forming sequence to low masses results in dramatic disagreement with the observed evolution of the mass function .
the assumption of a single - slope starforming sequence implies mass - doubling times of 25 million years for a galaxy of mass @xmath109 m@xmath33 at @xmath11 , which is inconsistent with the observed growth of the mass function .
we further demonstrate that this inconsistency is unlikely to be solved by either mergers or undiscovered populations of low - mass quiescent galaxies .
thus , we posit that the most likely explanation is a steeper slope for the star - forming sequence at low masses than is observed at high masses .
this is the first explicit test of the consistency between the star - forming sequence and the stellar mass function since bell et al .
( 2007 ) , made possible in part by recent , accurate , and deep high - redshift measurements of the stellar mass function and the star - forming sequence ( whitaker et al . 2012 ; tomczak et al .
another key advance in this study is demonstrating that neither mergers nor an unknown population of quiescent low - mass galaxies is likely to solve this problem .
the results of this study are consistent with related joint analyses of the star - forming sequence and the stellar mass function ( drory & alvarez 2008 ; weinmann et al .
2012 ) .
a key question is whether a steep low - mass slope , disjoint from a shallower , high - mass slope , is consistent with observations .
the evidence presented in this study for a steep low - mass slope is most compelling at @xmath11 ( see figure [ mainplot ] ) , due to the low mass - doubling timescales implied by a single - slope star - forming sequence ( figure [ tdouble ] ) .
some studies have observed a downturn in the sfr - mass relationship in star - forming galaxies at high redshift , but this lies close to or below their completeness limits , preventing any strong conclusions ( karim et al . 2011 ; whitaker et al .
salim et al .
( 2007 ) finds a mass - dependent slope in their data before removing agn , but the mass - dependent slope disappears after removing agn from their data . in a companion paper ( whitaker et al .
2014 ) , we present new observational evidence for a mass - dependent slope .
whitaker et al .
( 2014 ) measures the slope of the star - forming sequence separately for galaxies with @xmath83 @xmath110 10.2 and those with @xmath83 @xmath110 10.2 , finding that the low - mass slope is approximately unity , and remains stable within observational uncertainties between @xmath111 , while the high - mass slope decreases from roughly 0.8 at @xmath112 to 0.2 at @xmath72 .
these findings are consistent across observational indicators for the star formation rate . even after adopting a steep slope for the star - forming sequence below @xmath113 ,
there remains a @xmath120.3 dex offset at @xmath11 between the growth of the mass function and the normalization of the star - forming sequence .
this offset is present at all stellar masses , and largely disappears by @xmath6 .
this discrepancy may arise from problems in stellar mass estimates , star formation rate estimates or , most likely , both . since the whitaker et al .
( 2012 ) starforming sequence and the zfourge mass functions both use nmbs imaging to derive stellar masses , the difference is very unlikely to stem from systematic differences between these two studies .
this can not be fixed by adjusting the merger rates , as the high - mass end already overproduces stars even when neglecting growth via mergers .
if the star formation rates are solely responsible for the discrepancy , even after fixing the low - mass slope to unity , star formation rates would have to decrease across all masses at @xmath114 .
the difference at @xmath11 can be read off of figure [ mainplot ] : star formation rates would have to decrease by @xmath115 dex at @xmath116 , decreasing to @xmath117 dex at @xmath118 ( again , before accounting for growth via mergers ) .
the star - forming sequence reported in whitaker et al .
( 2012 ) represents the median star formation rate for star - forming galaxies .
notably , the median sfr is 0.1 dex below the average star formation rate for star forming galaxies ( whitaker et al .
2014 ; kelson 2014 ) .
correcting for this offset does not change the results of this study indeed , it would only further increase the discrepancy between star formation rates and the growth of stellar mass .
some of the discrepancy at low masses can be alleviated by mergers decreasing the number density of galaxies .
however , mergers also contribute to the stellar mass growth of galaxies at the high - mass end , which is implicitly reflected in the growth of the mass function .
thus , accounting for mass growth via mergers would increase the discrepancy between star formation rates and stellar mass growth at the high - mass end , and there will still exist a global difference between star formation rates and stellar mass growth .
we first ask whether this disagreement is particular to the star formation rates measured in the whitaker et al .
( 2012 ) star - forming sequence .
we perform a comparison between the integrated star formation rate densities implied by star - forming sequences from the literature , shown in figure [ litcomp ] ( oliver et al . 2010 ; kajisawa et al . 2010 ; peng et al .
2010 ; karim et al . 2011
; whitaker et al .
we use the abundance of star - forming galaxies from zfourge to convert the star - forming sequence into star formation rate densities .
the implied global star formation rates vary by a factor of two , but display similar redshift evolution .
the similar redshift evolution between studies is due to the broad agreement in the redshift evolution of the normalization of the starforming sequence .
the actual offsets are due to a combination of different normalizations and different slopes ; crucially , changing the limits of integration has a significant effect on this comparison , due to the difference in observed slopes .
these differences in slopes and normalization may come about due to different sfr indicators , different adopted conversions from luminosity to sfr , different dust corrections , selection effects , or different definitions of a star - forming galaxy ( speagle et al .
since the redshift evolution differs little between studies , the relative rate of change may be well - constrained at least for galaxies with @xmath83 @xmath110 10 .
notably , the @xmath43 dex study - to - study variation between star formation rates is comparable in size to the systematic decrease in star formation rates necessary to bring the evolution of stellar mass and star formation rates into agreement .
as many of the observable sfr indicators are primarily driven by radiation from massive stars , one way to decrease observed star formation rates is to postulate a top - heavy imf ( e.g. , van dokkum 2008 ) .
it has been claimed in the past that a top - heavy imf is necessary to reproduce the properties of the submillimeter galaxy population ( baugh et al . 2005 ; dav et al .
2010 ) , though recent studies may have resolved this tension ( hayward et al .
there is not enough systematic evidence for a top - heavy imf to be compelling so far ( bastian , covey , & meyer 2010 ) . alternatively
, systematic overestimation of stellar masses at high redshift could solve the tension with star formation rates .
the cumulative redshift - dependent errors in the mass function necessary to be consistent with the modified star - forming sequence can be read directly from the bottom panels of figure [ mainplot ] : the error at high masses would have to be @xmath119 dex , and @xmath120 dex at low masses . at the high - mass end
, errors could potentially come from systematic errors in fitting the light profile ( bernardi et al .
2013 ) . at the low - mass end
, stellar masses may be significantly overestimated due to emission line contributions to the observed flux . fixing the stellar metallicity to
the solar value is likely a poor approximation at high redshift , and may lower stellar mass estimates as well , particularly at the low - mass end ( mitchell et al .
it is also possible that the exponentially - decaying sfhs are poor fits to true sfhs of @xmath11 galaxies , which would decrease stellar mass at all masses ( papovich et al .
2011 ) . in figure [ litcomp ] ,
we compare the integrated stellar mass density from the zfourge survey to other results from the literature ( marchesini et al . 2009 ; muzzin et al .
2013 ; moustakas et al .
these studies agree to within a factor of two from @xmath121 , but they show different evolution with redshift . in particular , for @xmath122 , zfourge shows less evolution with redshift than other surveys , which implies correspondingly lower star formation rates .
if the integration limits are extended to lower stellar masses , the integrated stellar mass density in zfourge shows even less evolution with redshift , due to relatively high abundance of low - mass galaxies at high redshift in zfourge .
this comparison shows that the uncertainties in the absolute value of stellar mass density in the literature are at least a factor of two , and this difference is redshift - dependent .
these systematic redshift - dependent errors in stellar mass estimates may substantially contribute to the difference between the growth of stellar mass and the star formation rate at @xmath11 .
we note that altering the imf for low - mass stars as suggested by conroy & van dokkum ( 2012 ) will affect star formation rates and stellar masses equally , and thus will not solve the tension between them . here
, we examine the effects of different forms of the star - forming sequence on the global star formation rate . the mass - dependent contribution to the global star formation rate implied by the star - forming sequence , calculated via @xmath123 ,
is shown in figure [ sfrd ] . with no modifications to the low - mass slope of the star - forming sequence , low mass galaxies
dominate the cosmic star formation rate at @xmath11 , and continue to contribute substantially to the cosmic star formation rate at lower redshifts .
this is consistent with reddy & steidel ( 2009 ) , which suggests that sub-@xmath124 galaxies constitute up to 93% of the unobscured uv cosmic star formation rate density at @xmath125 .
however , sobral et al .
( 2014 ) use data from a narrowband survey of h@xmath3 emission at @xmath126 to argue that the contribution to the cosmic star formation density peaks at galaxies of @xmath32 m@xmath33 .
this is in agreement with the @xmath127 star formation rate density measured from the roles survey ( gilbank et al .
semi - analytical models predict a peak at a similar location , though at slightly higher stellar masses : @xmath12@xmath128 m@xmath33 , decreasing to 10@xmath129 m@xmath33 at @xmath130 ( lagos et al .
2014 ) . adopting a steep low - mass slope , as suggested by our analysis ,
substantially decreases the implied contribution from low - mass galaxies , in agreement with gilbank et al .
( 2010 ) and sobral et al .
( 2014) though notably , the gilbank et al .
( 2010 ) star formation rate density falls off more sharply towards low stellar mass , due to differences between the roles and zfourge stellar mass functions ( gilbank et al .
the peak of the cosmic star formation rate density depends on the exact location of the transition between high - mass slope and low - mass slope ; in our model , this is fixed to @xmath128 m@xmath33 .
the star formation rates implied by the evolution of the stellar mass function have an almost identical shape to the modified low - mass star - forming sequence , but with a @xmath43 dex normalization offset at @xmath11 .
both of the modified star - forming sequences presented in our study indicate that the contribution to the cosmic star formation peaks in galaxies of @xmath131 solar masses . an interesting parallel can be drawn with the star formation efficiency ( defined as m@xmath132/m@xmath133 ) .
hydrodynamical simulations and abundance matching models indicate that the star formation efficiency peaks at a similar stellar mass @xmath134 for @xmath135 ( behroozi , wechsler , & conroy 2013b ; genel et al .
2014 ) .
the observed star formation rates of galaxies at @xmath136 have long been in tension with theoretical expectations , which include semi - analytical , hydrodynamical , and semi - empirical models of galaxy formation ( bouch et al .
2010 ; firmani , avila - reese , & rodrguez - puebla 2010 ; dav , oppenheimer , & finlator 2011 ; lilly et al .
2013 ; dekel & burkert 2014 ; genel et al .
2014 ; mitchell et al .
2014 ) . in models of galaxy formation
, the specific star formation rate of galaxies roughly tracks the specific dark matter accretion rate ( weinmann et al .
2012 ; lilly et al .
2013 ; mitchell et al . 2014 ; genel et al .
2014 ; peng et al .
2014 ) , with a normalization offset of @xmath137 , required to match low - redshift observations ( genel et al .
2014 ) . however , between @xmath136 , the observed star formation rates are higher than the model star formation rates by a factor of @xmath138 , despite matching the model star formation rates at both higher and lower redshifts .
the discrepancy increases at low stellar masses , where galaxy formation models prefer a star - forming sequence with a slope of approximately unity , while some observations indicate a shallower slope ( genel et al .
one solution is to de - couple galaxy formation and halo assembly at low stellar masses : weinmann et al .
( 2012 ) suggest this , arguing that observations indicate that the buildup of the number density of galaxies with @xmath139 @xmath83@xmath140 occurred much more recently than in the models .
mitchell et al .
( 2014 ) advocate for a similar picture , arguing that the steep star - forming sequence in semi - analytical models must be adjusted at low stellar masses to match observations of a shallower slope . in figure
[ theorycomp ] , we compare the star - forming sequence in semi - analytical models ( guo et al .
2013a ; mitchell et al .
2014 ) , stellar mass - halo mass abundance matching ( sham ) model ( behroozi & silk 2014 ) , and hydrodynamical models ( genel et al .
2014 ) to the star - forming sequence we derive from the zfourge mass functions .
they are strikingly similar in shape , with a slope of @xmath141 at the high - mass end and a slope of @xmath142 at the low - mass end .
considering that semi - analytical , hydrodynamical , and abundance - matching models must necessarily be consistent with the evolution of the stellar mass function , it is perhaps not surprising that they show similar behavior to the star formation rates derived from the observed redshift evolution of the stellar mass function .
we suggest that a mass - dependent slope for the star - forming sequence would ease the discrepancy between models and observations . a steep low - mass slope at @xmath114
is not ruled out by current studies : see the discussion in section [ obs_sfseq ]
. adopting a star - forming sequence with a slope close to unity at low masses means that low - mass galaxies grow in a self - similar fashion , implying that @xmath143 , where @xmath144 is a constant that depends only on redshift .
thus , the relative growth of low - mass galaxies would be independent of mass , but still dependent on redshift .
the star formation rate of low - mass galaxies will then scale similarly to the specific dark matter accretion rate , which , for @xmath145 , has @xmath146 ( neistein & dekel 2008 ) . enforcing a steep slope in the star - forming sequence at low stellar masses
is one way to construct a consistent picture of the buildup of stellar mass .
however , it is important to explore possible alternative explanations as well : namely , the merger rate and the possibility of a hidden population of quiescent galaxies .
higher merger rates will remove low - mass galaxies before they grow rapidly , while large populations of quiescent galaxies will decrease the average star formation rate as a function of mass .
we calculate the merger rate needed to produce a single - slope star - forming sequence .
this is derived by growing the stellar mass function with the single - slope star - forming sequence , then computing the difference between the observed mass function and the mass function grown by star formation .
this difference , divided by the time step , is the required rate of destruction by mergers .
note that while we neglect the galaxy growth due to mergers in this procedure , in practice , the growth rate due to mergers in our model is negligible for galaxies with stellar mass lower than 10@xmath147 m@xmath33 .
the same effect can be caused by undetected quiescent galaxies . to calculate the required number density of undetected quiescent galaxies
, we first calculate the growth of the stellar mass function in a single time step due to the single - slope star - forming sequence from whitaker et al .
then , we compare this growth rate to the growth rate in a single time step required to match the observed stellar mass function .
the increase in the number of quiescent galaxies necessary to reconcile these two growth rates is inferred from equation 6 .
figure [ sffrac_merge ] shows the merger rate necessary for the observed star - forming sequence to be consistent with the evolution of the mass function .
we compare this directly to the merger rate from the guo et al .
( 2013b ) semi - analytical model . at @xmath11
, merger rates would have to be over two orders of magnitude higher than expected in order be consistent with the observed star - forming sequence .
this difference decreases with redshift , but even at @xmath72 , it is necessary to increase the rate of low - mass mergers by an order of magnitude to match the growth of the mass function .
figure [ sffrac_merge ] also shows the factor by which the number density of quiescent galaxies would have to increase to bring star formation rates and the growth of stellar mass into agreement . the number density of quiescent galaxies at @xmath89 would have to increase by almost three orders of magnitude at @xmath11 . at @xmath15
, it would need to increase by two orders of magnitude .
this would indicate that galaxy surveys are missing the overwhelming majority of quiescent galaxies at @xmath6 .
zfourge is estimated to be mass - complete for quiescent galaxies down to @xmath83 = 9 at @xmath71 .
however , the zfourge stellar mass - complete limit is estimated by comparing the magnitude of a single stellar population formed at @xmath148 to the magnitude limit of the survey .
this calculation does not take into account the confounding effects of dust .
if low - mass quiescent galaxies have a significant amount of dust at higher redshift , this would imply that the nominal mass - completeness of the survey is overestimated .
we conclude that while it is possible that a combination of merger rates and missing quiescent galaxies may be a factor in easing the tension between star formation rates and the growth of stellar mass , they are unlikely to be a dominant factor . the most likely solution remains a modification of the low - mass slope of the star - forming sequence .
we model the growth of the stellar mass function by assuming all galaxies grow at the same rate . at fixed stellar mass , however , galaxies are a diverse population and display a variety of star formation rates ( van dokkum et al .
it is thus important to discuss whether the assumption of a single growth rate at fixed mass is a valid one .
it is possible to mathematically prove that scatter in star formation rates does not influence the evolution of the mass function , and that the only relevant parameter is the _ average _ star formation rate as a function of mass and time . here ,
we sketch this proof briefly ; the full proof will be presented in franx et al .
( in preparation ) .
the evolution of the mass function due to star formation is given by the following equation : @xmath149\ ] ] with @xmath150 defined as the average star formation rate at mass @xmath151 .
this continuity equation follows immediately from the conservation of galaxies , and was first presented by drory & alvarez ( 2008 ) .
the equation is identical to the continuity equation in stellar dynamics ( e.g. binney & tremaine ( 2008 ) , equation 4.204 ) . in this case , the mass @xmath151 is the x - position , and the star formation rate is the velocity .
as further discussed by franx et al ( in preparation ) , the consequence is that the evolution of the mass function due to star formation at any @xmath151 is determined by @xmath150 only , and is the same for all models with a given @xmath152 and @xmath150 .
hence the scatter in the star formation at a given @xmath151 does not influence the evolution of the mass function - as long as @xmath150 is given .
it is not important on what timescale galaxies may move below or above the average .
as shown by franx et al ( in preparation ) , another way to see this quickly is by assuming that scatter is caused by two populations of galaxies with two different star formation rates . in appendix b , we show that sum of the two populations will satisfy the continuity equation , just like the two populations separately .
this argument can be extended to an arbitrary number of subpopulations .
thus , any galaxy population for which the mean star formation rate as a function of mass is identical will evolve in the same fashion .
the actual distribution of star formation rates does not affect the evolution of the stellar mass function : as long as the mean star formation rate as a function of mass is identical in any given distribution at _ every time step _ , the mass function will evolve in an identical fashion .
we adopt an instantaneous stellar mass loss model . in reality
, the mass loss rate of galaxies is time - dependent , with the majority of it occurring in the first hundred million years .
the efficiency of the instantaneous approximation will thus scale inversely with the specific star formation rate of galaxies , meaning that low - mass galaxies at high redshift are most affected by this approximation .
stellar mass functions are tabulated using the current mass in stars and stellar remnants , thus implicitly taking this effect into account . to account for this in our model ,
the fraction of mass returned to the ism , @xmath62 , would have to be modeled as sfh - dependent .
this would likely elevate the implied star formation rates at low masses by @xmath153 .
this is not a dominant effect , but would contribute towards reconciling stellar mass growth and star formation rates at high redshift and low masses .
in this study , we have examined the connection between the observed star - forming sequence and the observed redshift evolution of the stellar mass function .
we have constructed a smooth parameterization of the growth of the stellar mass function from zfourge and sdss - galex data , and compared this growth to the growth implied by the observed star - forming sequence .
we find that a simple extrapolation of the observed slope to low stellar masses is inconsistent with the observed evolution of the mass function .
we conclude from this comparison that one or all of the following must be true : ( 1 ) the star - forming sequence steepens at low masses , ( 2 ) the destruction rate of low - mass galaxies by mergers is very high , ( 3 ) there is a dominant population of low - mass quiescent galaxies missing from high - redshift surveys .
we use merger rates from semi - analytical models to show that the merger rate of low - mass galaxies is several orders of magnitude too low to solve the issue .
we also show that there would have to be several orders of magnitude more quiescent galaxies than observed at @xmath11 to solve the issue , which is unlikely to be true in the mass - complete regime of stellar mass surveys .
we thus conclude that , in order to decrease the star formation rates for galaxies with mass @xmath154 m@xmath33 , the star - forming sequence must steepen at low masses .
this conclusion is supported by the observations of whitaker et al .
( 2014 ) .
we show that a mass - dependent slope for the star - forming sequence makes the global star formation rate more consistent with the growth of stellar mass density .
however , even with a steeper star - forming sequence , there is a discrepancy of @xmath43 dex between the growth of stellar mass density and the global star formation rates at high redshift . comparing different studies in the literature
, we find that there is an inter - study difference of @xmath43 dex in both the implied global star formation rates and the growth of the stellar mass density , and the redshift evolution of these quantities can differ substantially between studies .
it is thus unclear whether measurements of star formation rates , stellar masses , or both need to change in order to tell a consistent story .
future efforts to firmly establish the observational uncertainties in star formation rates and stellar masses will be crucial to resolving this discrepancy .
we thank the anonymous referee for their insightful suggestions and attention to details , which have greatly improved the quality of the paper .
support from nasa grant nnx11ab08 g is gratefully acknowledged .
our fiducial model , presented in the main body of the paper , uses the destruction rate of galaxies from the guo et al .
( 2013b ) sam , and assumes that the resulting mass growth occurs in 1:10 ratio mergers .
this approximation has the advantage of enforcing mass conservation within our model .
however , this approximation also introduces several issues .
firstly , by simplifying all mergers into 1:10 mass ratio mergers , the mass growth rate of a galaxy of mass m is determined only by the destruction rate at 0.1*m , modulated by the ratio of number densities .
in particular , this underestimates the mass growth due to mergers in high mass galaxies , which do nt just accrete intermediate - mass galaxies but also accrete a significant fraction of low - mass galaxies .
secondly , galaxies below the mass cutoff of our simulation ( 10@xmath155 m@xmath33 ) also deposit stellar mass into more massive galaxies , but this is not reflected in our prescriptions .
, low - mass galaxies are being rapidly destroyed , but this rate of destruction decreases as redshift decreases .
the opposite trend is seen at higher masses , whereby the rate of destruction for higher - mass galaxies increases with decreasing redshift .
the growth rate due to mergers is a smooth and continuous function of mass . ] to address these issues , we directly measure the mass growth of galaxies due to mergers in the guo et al .
( 2013b ) sam , and apply them to the stellar mass function .
the destruction rates and growth rates as a function of mass are shown in figure [ samdirect ] .
the growth due to mergers is a smooth function of mass , and decreases smoothly and slowly as redshift decreases .
the destruction rate due to mergers increases with redshift at the low - mass end and decreases with redshift at the high - mass end , such that a destruction rate which is relatively flat with mass at @xmath156 becomes a destruction rate which is quite steep with mass at @xmath157 .
we note that the millenium - ii simulation probes a relatively small volume , such that at the highest masses , there are only a few galaxies from which to measure merger growth rates .
the accuracy of the growth and destruction rates at the highest masses are thus strongly limited by both poisson noise and cosmic variance .
we interpolate between the measurements in figure [ samdirect ] and apply both directly to the growth of the mass function .
the resulting growth of the mass function , when combined with the three parameterizations of the star - forming sequence described in section [ dif_sfseq ] , is shown by the dashed lines in figure [ mainplotappend ] .
the main difference as compared to our preferred model is the growth of galaxies with @xmath83 @xmath110 11 , which increases substantially when using the sam growth rates .
lower - mass galaxies experience less growth due to mergers , but as their growth was already dominated by star formation , this has little effect on the evolution of the mass function . , while the purple line is done in the same way , but after applying the sam - measured merger rates .
the slope of the star - forming sequence becomes an even stronger function of mass after including mergers .
] the main result of this study is that the growth of the mass function requires a mass - dependent slope in the star - forming sequence .
this is demonstrated in section [ zfourgesfr ] , where we derive a star - forming sequence which is consistent with the evolution of the stellar mass function , and show that it has two distinct slopes .
we now repeat this process of deriving a consistent star - forming sequence , but now perform this fit @xmath158 including the effects of mergers .
this star - forming sequence is shown in purple in figure [ mainplotappend ] .
note that it is not a functional fit that is shown ( as is true for the star formation rates derived from the mass function without mergers ) , but the actual inferred star formation rates .
including the mergers when fitting for the star - forming sequence slightly increases the star formation rates at low masses , while decreasing them substantially at high masses .
this is due to the fact that the main effect of mergers at low mass is the destruction of galaxies , while the main effect of mergers at high mass is additional mass growth .
in fact , applying the guo et al .
( 2013b ) sam merger growth rates to the high - mass end is more than sufficient to model the growth of high - mass galaxies in the zfourge mass function : presumably , the merger growth rates in the real universe are lower than those measured from the sam .
we also fit the merger - modulated star formation rates with a double power law with the same form as equation [ brokenpowerlaw ] .
the redshift evolution of the normalization and the slopes is shown in figure [ bpl_append ] .
crucially , the evidence for a mass - dependent slope is not altered by including the growth rate due to mergers .
in fact , this conclusion is strengthened , though this highlights an important point : deriving the high - mass slope of the star - forming sequence is very sensitive to the merger rate ( conversely , the low - mass slope is very stable to the merger rate ) . however , using merger growth rates measured directly from the sam violates mass conservation . specifically , the guo et al .
( 2013b ) mass functions have a higher number density of low - mass galaxies than is observed in zfourge , which means that for the same rate of destruction of galaxies , the merger growth rates from the semi - analytical model will be higher than the simple 1:10 merger growth rates assumed in our model .
this results in mass growth at the high - mass end which is more than enough to explain the growth of the mass function .
this extra growth is reflected in the high - mass drop off in the merger - modulated star formation rates from the stellar mass functions , visible in figure [ mainplotappend ] .
in this section , we prove that equation [ galaxycont ] presented in the text is linear , and thus can be generalized to the entire galaxy population .
first , assume that we have two population of galaxies , each with a different dependence of star formation rate with stellar mass .
the continuity equation for each individual population is @xmath159\ ] ] where the subscript i is either 1 or 2 for the two populations .
we sum both sides of the continuity equation , as defined for galaxy populations 1 and 2 . for the sum of the left hand terms
, we obtain : @xmath160}{\partial t}\ ] ] and for sum of the right hand terms , we obtain : @xmath161 - \frac{\partial}{\partial m}\left [ \psi_2(m , t ) \langle
\dot{m}_2 \rangle \right ] = \ ] ] @xmath162\ ] ] now , define @xmath163 to be the average star formation rate of the combined galaxy population , and @xmath164 to be the total mass function : @xmath165 @xmath166 substitute these into the left and right hand sums , and we recover the original equation : @xmath167\ ] ] thus , the continuity equation for the evolution of the galaxy population is linear . | we examine the connection between the observed star - forming sequence ( sfr @xmath0 @xmath1 ) and the observed evolution of the stellar mass function between @xmath2 .
we find the star - forming sequence can not have a slope @xmath3 @xmath4 0.9 at all masses and redshifts , as this would result in a much higher number density at @xmath5 by @xmath6 than is observed .
we show that a transition in the slope of the star - forming sequence , such that @xmath7 at @xmath8 and @xmath9 ( whitaker et al . 2012 ) at @xmath10 , greatly improves agreement with the evolution of the stellar mass function .
we then derive a star - forming sequence which reproduces the evolution of the mass function by design .
this star - forming sequence is also well - described by a broken - power law , with a shallow slope at high masses and a steep slope at low masses . at @xmath11 ,
it is offset by @xmath120.3 dex from the observed star - forming sequence , consistent with the mild disagreement between the cosmic sfr and recent observations of the growth of the stellar mass density .
it is unclear whether this problem stems from errors in stellar mass estimates , errors in sfrs , or other effects .
we show that a mass - dependent slope is also seen in other self - consistent models of galaxy evolution , including semi - analytical , hydrodynamical , and abundance - matching models . as part of the analysis
, we demonstrate that neither mergers nor hidden low - mass quiescent galaxies are likely to reconcile the evolution of the mass function and the star - forming sequence .
these results are supported by observations from whitaker et al .
( 2014 ) . |
long - period moir patterns are characteristic of well - aligned graphene heterostructures with hbn .
they offer an experimentally - viable route @xcite to observe fractal magnetic miniband spectra generic for electrons in two - dimensional superlattices in strong magnetic fields @xcite .
these fractal miniband spectra , also known as hofstadter s butterflies @xcite , reflect the fact that the translational symmetry of a lattice , suppressed by the presence of a magnetic vector potential in the electron s schrdinger equation , is restricted to magnetic field values @xmath0 corresponding to a rational fraction @xmath1 of magnetic flux quantum @xmath2 per unit cell area @xmath3 of the lattice @xcite .
the generic features of fractal magnetic miniband spectra have been observed @xcite and modeled @xcite in monolayer graphene heterostructures with hbn , where the properties of dirac electrons at the conduction / valence band edge allow for a straightforward interpretation of experimental observations , and have also been studied in slightly misaligned pairs of graphene flakes @xcite .
modeling of the fractal magnetic miniband spectra of a bilayer graphene / hbn heterostructure is much more limited @xcite , even though one of the first observations of this phenomena was in this system @xcite . in this article
, we study fractal magnetic minibands in bilayer graphene ( blg ) subject to a moir superlattice perturbation due to an hbn underlay .
the low - energy hofstadter butterfly spectra of a blg - hbn heterostructure is dominated by bands related to the degenerate `` zero - energy '' landau level ( ll ) states @xmath4 and @xmath5 of unperturbed blg @xcite .
we use a symmetry - based approach to model the influence of hbn on electrons in blg , and we show that the spectra in blg - hbn is a superposition of two very different miniband spectra associated with electrons in opposite valleys ( brillouin zone corners ) of blg s band structure : the miniband spectrum in one valley is only weakly broadened by the superlattice potential and incorporates one unperturbed @xmath4 ll , while the miniband spectrum in the other valley is widely broadened .
this valley asymmetry arises from spatial inversion symmetry breaking produced by the fact that the moir perturbation only directly affects one of the two blg layers .
since the zero energy ll states reside on different layers in opposite valleys , in one valley they are strongly influenced by the moir perturbation , but in the other valley they are not .
these qualitative features agree with the results of the tight - binding model of moon and koshino @xcite .
in addition , we characterize different types of moir superlattice perturbations including the possibilities that the perturbation creates potential asymmetry between the carbon atoms in blg or introduces a spatial modulation of the nearest - neighbor carbon - carbon hopping amplitude .
hence , we find that the dirac point at the conduction - valence band edge can be either gapless or gapped , and that a secondary dirac point can appear in the valence band of blg - hbn .
our model hamiltonian is presented in section [ s : ham ] and its numerical diagonalization in the presence of a magnetic field , including methodology , results and discussion , is described in section [ s : numerics ] , where we also derive simple effective hamiltonians to describe low - energy features in the magnetic minibands and , also , an effective hamiltonian to describe the secondary dirac point . in section [
s : map ] , we show how gaps in the fractal energy spectrum are manifest as observable incompressible states under a variation of carrier density and magnetic field .
to describe the sublattice ( @xmath6 ) and bottom / top ( @xmath7 ) layer composition of electron states in bernal - stacked blg on a hbn underlay , we write down their hamiltonian @xcite as a @xmath8 matrix , @xmath9\!. \nonumber\end{aligned}\ ] ] here , we use the basis of bloch functions @xmath10 for valley @xmath11 ( @xmath12 ) , @xmath13 for valley @xmath14 ( @xmath15 ) , and the @xmath16 pauli matrices @xmath17 act in the space of sublattice components .
the term @xmath18 on the diagonal takes into account the electrons dirac spectrum in each layer where @xmath19@xmath20s@xmath21 @xcite is the dirac velocity , @xmath22 , @xmath23_z =- b$ ] , @xmath24 , and @xmath25 , while @xmath26ev @xcite is the inter - layer hopping between @xmath27 and @xmath28 sublattices .
the term @xmath29 in eq .
( [ eqn:4x4hamiltonian ] ) accounts for the moir superlattice perturbation produced by the hbn substrate @xcite , and is applied only to the layer of the bilayer which is closest to the hbn ( layer ` @xmath30 ' ) @xcite . within the term @xmath29
, the functions @xmath31 , with @xmath32 , are written using the shortest non - zero bragg vectors of the moir superlattice @xmath33 ( 0,{\tfrac { 4\pi } { \sqrt{3}l}})$ ] , where @xmath34 describes anti - clockwise rotation by angle @xmath35 , @xmath36 takes into account the relative lattice mismatch between graphene and hbn @xcite , @xmath37 is the misalignment angle between the two hexagonal lattices and @xmath38@xmath39 is the lattice constant of graphene . for @xmath40 , @xmath41 so that the energy scale @xmath42 obtains its minimum value of @xmath43ev at @xmath44 .
the strength of the various terms included in @xmath29 are controlled by separate dimensionless parameters , @xmath45 , which describe in turn an electrostatic potential which does not distinguish between the two carbon sublattices ( @xmath46 ) , a sublattice - asymmetric part of the potential ( @xmath47 ) , and a spatial modulation of the nearest - neighbor carbon - carbon hopping amplitude ( @xmath48 ) .
one of the most interesting features of electrons in blg at strong magnetic fields is the degeneracy of two orbital lls , with @xmath4 and @xmath5 , which appear at @xmath49 , the edge between the valence and conduction band .
the mixing of these degenerate lls by the moir superlattice potential determines the main features of the lower - energy part of the magnetic miniband spectrum , shown in fig .
[ fig : magnetic_minibands ] for several different choices of moir perturbation parameters @xmath50 .
these low - energy electron states in blg can be described @xcite using a simplified two - band hamiltonian , which can be obtained from eq . by a schrieffer - wolf projection onto the basis of bloch states residing on @xmath51 and @xmath52 sublattices . for a blg - hbn heterostructure , such a projection results in @xcite , @xmath53 { \tfrac { b v^2 } { \gamma_1}}{\hat{\boldsymbol{\pi } } } h_+(r ) & { \tfrac { v^3 b } { \gamma_1 ^ 2}}{\hat{\boldsymbol{\pi } } } g_-(r){\hat{\boldsymbol{\pi}}}^{\dagger } \end{array } \right),\nonumber\\ & \boldsymbol{\tilde{m}}_{\text{-}1}=\left ( \begin{array}{cc } { \tfrac { v^3 b } { \gamma_1 ^ 2 } } { \hat{\boldsymbol{\pi}}}^{\dagger } g_-(r){\hat{\boldsymbol{\pi } } } & { \tfrac { b v^2 } { \gamma_1}}{\hat{\boldsymbol{\pi}}}^{\dagger } h_-(r ) \\[0.3em ] { \tfrac { b v^2 } { \gamma_1}}h_-^*(r){\hat{\boldsymbol{\pi } } } & v b g_+(r ) \end{array } \right),\nonumber \\ & g_{\pm } ( r)=\sum _ { m } e^{i b_m.r } ( u_0\pm i u_3(-1)^m),\nonumber\\ & h_{\pm}(r)= \pm
i u_1 \sum _
m(-1)^me^{i b_m.r}\left(b_m^x \pm i b_m^y \right)/b .
\nonumber\end{aligned}\ ] ] hamiltonian @xmath54 is written in the basis of the bloch states @xmath55 for the @xmath11 valley and @xmath56 for the @xmath14 valley using @xmath57 , and @xmath58 examples of the zero magnetic field @xmath11-valley bandstructure of moir perturbed blg with @xmath44 are displayed in the left panels of fig .
[ fig : magnetic_minibands ] .
these spectra were calculated by numerical diagonalization of the heisenberg matrix constructed from hamiltonian in a basis of unperturbed plane wave states .
the corresponding dispersion in the @xmath14-valley is obtained using the relation , @xmath59 , which follows from the time - reversal symmetry of hamiltonians and for @xmath60 .
each panel in fig .
[ fig : magnetic_minibands ] corresponds to a different set of moir perturbation parameters @xmath61 .
the choice of parameters used in the top panel corresponds to the predictions of a pair of microscopic models , one based on hopping between graphene carbon atoms and hbn atoms , the other on scattering of graphene electrons by the quadropole electric moments of nitrogen atoms .
interestingly , these two models predict the same relative values for the moir potential parameters @xcite , @xmath62 where @xmath63 depends upon the specific microscopic parameters used to describe the hbn underlay . here
we take @xmath64 , chosen to make the moir perturbation strong enough that an almost gapless secondary dirac point ( sdp ) is produced between the red and blue minibands in the valence band of the upper left panel of fig .
[ fig : magnetic_minibands ] .
signatures of this feature were observed experimentally in ref .
@xcite .
the other panels in fig .
[ fig : magnetic_minibands ] illustrate the influence of each @xmath50 parameter taken individually and they exemplify three additional scenarios : for @xmath65 , the original dirac point is gapped and there is no sdp ( in the two valence bands closest to zero energy ) ; for @xmath66 , the original dirac point and the sdp are both gapless ( and there is a sdp in the conduction band ) ; for @xmath67 , the original dirac point is gapless and there is no sdp .
here we describe the numerical diagonalization of hamiltonian in the presence of a magnetic field . to simplify this calculation , we take account of the hexagonal symmetry of the moir pattern and use a non - orthogonal coordinate system @xcite @xmath68 , where @xmath69 and @xmath70 are direct moir lattice vectors and @xmath71 . in this basis
the landau gauge has the form @xmath72 which leads to @xmath73,\end{aligned}\ ] ] where @xmath74 , and @xmath75 is the electron charge . for the wave - vector space ,
this determines @xmath76 with @xmath77 and @xmath78 .
hence , we employ the basis set of magnetic oscillator functions , @xmath79 where @xmath80 are the hermite polynomials . for free electrons in blg ,
the ll spectrum contains two degenerate states at @xmath49 , and pairs of conduction / valence band states at @xmath81 , @xmath82 , @xmath83 in general , the magnetic vector potential @xmath84 breaks the symmetry of the hamiltonian with respect to translations of the moir superlattice .
however , for magnetic flux @xmath85 where @xmath86 and @xmath87 are co - prime natural numbers and @xmath3 is the unit cell area of the moir superlattice , translational symmetry is restored .
because of this , we consider a unit cell of the magnetic superlattice that is @xmath87 times larger than the unit cell of the moir superlattice in both directions ( hence its area is @xmath88 times larger ) @xcite .
the magnetic translational group @xmath89 , where @xmath90 describes geometrical translations and @xmath91 and @xmath92 are integers , commutes with hamiltonian and is isomorphic to the group of geometrical translation , so that its eigenstates form a plane wave basis @xmath93 .
bloch functions @xmath94 exist in the magnetic brillouin zone which is @xmath88 times smaller than that of the moir superlattice @xcite and in which magnetic minibands are @xmath87-fold degenerate . for states with momentum @xmath95 within this magnetic brillouin zone , @xmath96 , and @xmath97 indexes the magnetic sub - bands , and @xmath98 indexes the above mentioned @xmath87-fold degeneracy without loss of generality
, we set @xmath99 and omit it , using the notation @xmath100 , from now on . in order to obtain the energy dispersion
, we calculate the matrix elements of the perturbation , @xmath101 , in this basis and diagonalize the resulting matrix @xcite .
the main panels of fig .
[ fig : magnetic_minibands ] show the magnetic spectrum of the blg - hbn heterostructure for the four choices of moir perturbation described above . for small flux
, the magnetic miniband spectra can be traced to the sequence of landau levels for moir perturbed blg . near zero energy ,
the gap at the original dirac point is seen in the magnetic spectra of the top two panels .
the presence or absence of a secondary dirac point ( depending on the particular parameter set ) is also clearly reflected in the magnetic spectra at small flux at the corresponding energy . at a higher flux ,
the landau levels broaden and split , forming a fractal pattern , with the most striking features around zero energy . for all parameter sets , the valley symmetry of the spectrum , preserved in the absence of a magnetic field ,
is lifted .
this is because the moir perturbation only affects the layer of blg that is adjacent to the hbn and , thus , it breaks inversion symmetry . in conjunction with time - inversion symmetry breaking by a magnetic field ,
this allows the energy spectra for electrons in valleys @xmath11 and @xmath14 to be different @xcite .
in particular , note that , in the absence of the perturbation , the distribution of the wave function among the layers is , for a given landau level , exactly inverted in the two valleys , eqs .
( [ eqn : basis function],[eqn : basisfnall ] ) .
the moir potential affects only the wave function component in the layer adjacent to hbn , hence breaking the layer symmetry and leading to valley - asymmetric spectra with a gap .
importantly , the spectrum in the valley for which the wave function sits on the layer further from the substrate contains a zero - energy landau level completely decoupled from the rest of the spectrum ( shown as a red line in the @xmath14-valley in fig .
[ fig : magnetic_minibands ] ) . the spinor structure eq . of the two zero - energy states , @xmath102 , for unperturbed electrons
shows that the states in valley @xmath11 ( @xmath14 ) are localized on the bottom ( top ) graphene layer only @xcite . since the moir perturbation does not directly influence the top layer , it does not couple the @xmath4 state in the @xmath14 valley to any other state and , thus , the @xmath4 state in this valley always persists .
this coexistence of a zero - energy ll and a fractal spectrum of magnetic sub - bands creates a unique opportunity to observe the interplay between electron - electron interaction and hofstadter s quantization @xcite .
each of the panels in fig .
[ fig : magnetic_minibands ] display further features of interest . for the parameter
set @xmath103 ( upper panel ) , the @xmath60 miniband spectrum displays a slightly gapped sdp found in the first valence miniband , which produces a sequence of ll including tilted zeroth lls , at the corresponding energy and weak magnetic fields ( @xmath104 @xmath105 ) .
we shall discuss the sdp further in section [ s : sdp ] ( a similar feature is present in the lower middle panel ) . for potential modulation taken in isolation , @xmath106 ,
the band structure obeys six - fold rotation symmetry in each valley ( left upper middle panel ) , in contrast to all other band structure images which only display symmetry under three - fold rotation .
for the spatial modulation of the carbon - carbon hopping amplitude , @xmath107 ( lower middle panels ) , the spectrum at both zero magnetic field and finite magnetic field obeys electron - hole symmetry , and the zeroth ll and first ll of the original dirac point are completely degenerate in both valleys ( discussed in section [ s : low - e ] ) .
for the sublattice - asymmetric potential , @xmath67 ( lower panel ) , the bands at zero magnetic field are symmetric under the operation which combines electron - hole - reflection and a rotation by @xmath108 , which is reflected in the electron - hole symmetry of the magnetic miniband widths in a magnetic field .
additionally for this perturbation , we find that the magnetic miniband structure around @xmath49 can form gapless linear or quadratic dirac points , which produce sequences of lls in the magnetic miniband spectrum , best visible around @xmath109 in the @xmath14-valley and @xmath110 in the @xmath11-valley ( also see section [ s : low - e ] ) .
the main features of the broadened magnetic miniband spectrum around zero energy at a simple fraction @xmath111 can be described by truncating the basis of bloch functions eq . to the zeroth and first ll only , yielding effective hamiltonians ,
@xmath112 here @xmath113 is used for the @xmath114-valley and written using the basis ( @xmath115,@xmath116 ) .
also @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath123 and @xmath124 .
the factor @xmath125 $ ] leads to rapid broadening for @xmath126 which slows down for @xmath127 . in @xmath128
the symbol @xmath129 denotes a direct sum , where the state @xmath115 is decoupled from the rest of the spectrum .
similarly , for simple fractions @xmath130 , with integer @xmath131 , the spectrum around zero energy may be described by effective hamiltonians @xmath132 where the @xmath133 block matrices are @xmath134 \\
e^{i l_0 k_1 } [ c_1 + i c_{12 } ] & - c_2 \end{array } \right ) \nonumber \ , , \\ & m_2=\left ( \begin{array}{cc } s_2 & -e^{\text{-}i l_0 k_1 } [ s_1 + i s_{12 } ] \\
-e^{i l_0 k_1 } [ s_1 -i s_{12 } ] & -s_2 \end{array } \right ) \nonumber \ , , \\ & m_3= \left(\begin{array}{cc } -s_2 & e^{\text{-}i l_0 k_1}[\tau^*s_1 -i \tau s_{12 } ] \\ e^{i l_0 k_1 } [ \tau^*s_1 + i \tau s_{12 } ] & s_2 \end{array}\right)\nonumber \ , , \\ & m_4=\left(\begin{array}{cc } c_2 & e^{\text{-}i l_0 k_1 } [ \tau^*c_1 + i \tau c_{12 } ] \\
e^{i l_0 k_1 } [ \tau^*c_1 -i \tau c_{12 } ] & -c_2 \end{array}\right ) \ , .\nonumber \end{aligned}\ ] ] here we use the basis @xmath115,@xmath135,@xmath116,@xmath136 and @xmath137 is the @xmath138 zero matrix . figure [ fig : low_energy_magnetic_minibands ] shows the excellent agreement between the result of diagonalizing effective hamiltonians - ( yellow ) , with the fully numerical diagonalization of hamiltonian ( turquoise ) , for the various choices of magnetic field , valley , and superlattice parameters @xmath50 indicated on the figure .
the computed dispersion surfaces display a wide array of possible forms , including the possibility of gapless linear or quadratic dirac points for the superlattice perturbation @xmath139 .
these features are found either in the @xmath14-valley whenever @xmath86 in @xmath140 is even ( fig .
[ fig : low_energy_magnetic_minibands]c ) , or in the @xmath11-valley whenever @xmath86 in @xmath140 is odd ( fig .
[ fig : low_energy_magnetic_minibands]d ) . for the perturbation with @xmath141 two zero - energy lls of hamiltonian
remain unperturbed ( lower middle panel in fig .
[ fig : magnetic_minibands ] ) , which is reflected in the fact that all matrix elements in hamiltonians - vanish . in this case
, hamiltonian obeys the `` electron - hole '' symmetry @xmath142 , which implies that its matrix elements must obey @xmath143 .
consequently , the resulting heisenberg matrix has at least @xmath144 linearly dependent rows , resulting in zero - energy eigenvalues which are @xmath145-fold degenerate .
as the @xmath48 only perturbation can be considered to be a periodic pseudo - magnetic field @xcite , it could alternatively be created using an abrikosov lattice , i.e. , a vortex structure generated from a type - ii superconductor in a magnetic field .
such a system is also expected to possess degenerate zero energy eigenvalues @xcite .
here we give an analytical description of the almost gapless sdp found at the corner superlattice brillouin zone in the valence band in the top row of fig .
[ fig : magnetic_minibands ] . to do this
, we note that zone folding using bragg vectors @xmath146 brings together three degenerate plane wave states @xmath147 , @xmath148 and @xmath149to each of the two inequivalent corners of the moir brillouin zone where @xmath150 , @xmath151 , and , @xmath152 with @xmath35 the polar angle of momentum @xmath153 , and @xmath154 the band index .
using @xmath155 theory , the vicinity of each moir brillouin zone corner can then be described using an effective hamiltonian acting on a three component vector of smoothly varying envelope functions , written in a basis of the standing wave states , @xmath156 using basis the @xmath157 hamiltonian for the @xmath11-valley is diagonal exactly at the brillouin zone corner , @xmath158 where @xmath159 and @xmath160 denote the real and imaginary parts .
the corresponding hamiltonian for the @xmath14-valley is obtained using time reversal symmetry , @xmath161=@xmath162 .
figure [ fig : sdpll ] shows the ll spectra of hamiltonian using blue(red ) dots for the @xmath163(@xmath164 ) corner , where the signature of an almost gapless sdp is evident at @xmath165 for the @xmath163 corner . to calculate this spectra , we diagonalized the heisenberg matrix of hamiltonian in a basis consisting of the products of magnetic oscillator functions with the standing waves states , @xmath166 , where we take @xmath167 , and @xmath168 , where is @xmath169 sufficiently large that the resulting spectrum converges .
any eigenstates with a large weight on high - index magnetic oscillator functions , @xmath170 with @xmath171 , are discarded because they violate the @xmath155 approximation used to construct hamiltonian .
also , to improve the comparison of its ll spectra with that of hamiltonian , we calculate the moir perturbation correction to the band - edge energies , @xmath172 , using a higher order of perturbation theory than is explicit in eq . .
the result of numerical diagonalization of the two band hamiltonian is displayed using black dots in fig .
[ fig : sdpll ] .
the two spectra agree well , confirming that eq .
is a good description of the sdp .
moreover , we can diagonalize hamiltonian for arbitrarily small magnetic fields , where the size of the heisenberg matrix of magnetic bloch functions needed to diagonalize hamiltonian becomes prohibitively large .
the energy of the``zeroth '' ll originating from the sdp in fig .
[ fig : sdpll ] depends on the magnetic field , due to a non - zero magnetic momentum generated from the influence of a third band ( residing mainly on @xmath173 for hamiltonian ) that mixes with the lls of the first two bands . for the parameter set @xmath103 , this third band is separated by a large gap from the sdp , which permits the approximate two band description for the sdp , @xmath174 where the parameters for the energy shift @xmath175 , gap @xmath176 , effective velocity @xmath177 , and magnetic momentums @xmath178 and @xmath179 can be fitted to the result of numerically diagonalizing eq . .
such a fitting is illustrated in fig .
[ fig : sdpll ] using the grey lines , and produces the best accuracy for low indexed lls .
[ fig : filling_factors ] shows the maps of incompressible states for each parameter set in fig .
[ fig : magnetic_minibands ] , created by filling the corresponding magnetic miniband spectrum with electrons .
a larger gap is depicted with a darker color , resulting in lines , the gradient of which corresponds to the filling factor . for the regions shown in yellow ,
the gap is negligibly small as a result of incomplete filling of a magnetic miniband , each of which can accommodate a density @xmath180 of electrons ( including spin degeneracy ) .
in contrast , the unperturbed zero - energy landau level present in the @xmath181 valley accommodates electron density of @xmath182 , creating a large gapless region portrayed in red .
because of its large electron capacity at exactly zero energy , the presence of this zero - energy mode should be clearly visible in capacitance measurements . electron - hole symmetry is displayed in the plots for @xmath183 ( lower left ) , and @xmath184 ( lower right ) , whereas it is absent in the upper two plots .
many of the largest gaps in the magnetic spectrum plots ( fig .
[ fig : magnetic_minibands ] ) occur between the zeroth and first lls of dirac points ( whether original or secondary ) . in the map of incompressible states ,
these large gaps are manifest as a solid blue line , which intersect the @xmath60 axis at a density @xmath185 , where @xmath186 for the main dirac point or @xmath187 for an sdp at the edge of the first conduction / valence miniband .
for example , there is a ( gapped ) sdp in the valence band of the magnetic spectrum for the moir perturbation @xmath103 ( upper panel in fig .
[ fig : magnetic_minibands ] ) . in the upper panel of fig .
[ fig : filling_factors ] , the gap between the two zeroth lls of this sdp is represented as a blue vertical line , which starts from ( @xmath188 , @xmath60 ) , whereas the two gaps between the zeroth and first lls are shown as two tilted blue lines , with gradients of @xmath189 .
in summary , we have shown that the presence of the hbn substrate lifts the valley degeneracy of bilayer graphene , producing different magnetic hofstadter s butterflies in each valley .
the zero - energy landau level located on the layer furthest from the hbn substrate remains unaffected by the moir perturbation , which makes the blg / hbn spectrum unique in comparison to other known magnetic spectra , for which all landau levels split into sub - bands .
we investigated the influence of different possible characteristics of moir perturbation including an electrostatic potential which does not distinguish between the two carbon sublattices , a sublattice - asymmetric part of the potential , and a spatial modulation of the nearest - neighbor carbon - carbon hopping amplitude , and we identified how they give rise to different features in the electronic spectra including gapped or overlapping bands , or bands connected by a secondary dirac point .
in addition to determining the fractal electronic spectra by numerical diagonalization of a model hamiltonian , we also derived simple effective hamiltonians to describe low - energy features in the magnetic minibands and to describe the secondary dirac point .
finally , we showed how gaps in the fractal energy spectrum lead to the formation of incompressible states that may be observed under a variation of carrier density and magnetic field .
we thank a. k. geim , i. aleiner and m. koshino for useful discussions .
this work was funded by the eu flagship project , epsrc science and innovation award , epsrc grant ep / l013010/1 , and erc synergy grant hetero2d .
l. a. ponomarenko , r. v. gorbachev , g. l. yu , d. c. elias , r. jalil , a. a. patel , a. mishchenko , a. s. mayorov , c. r. woods , j. r. wallbank , m. mucha - kruczyski , b. a. piot , m. potemski , i. v. grigorieva , k. s. novoselov , f. guinea , v. falko and a. k. geim , nature * 497 * , 594 ( 2013 ) .
g. l. yu , r. v. gorbachev , j. s. tu , a. v. kretinin , y. cao , r. jalil , f. withers , l. a. ponomarenko , b. a. piot , m. potemski , d. c. elias , x. chen , k. watanabe , t. taniguchi , i. v. grigorieva , k. s. novoselov , v. i. falko , a. k. geim and a. mishchenko , nat .
phys * 10 * , 525 ( 2014 ) m. diez , j. p. dahlhaus , m. wimmer , and c. w. j. beenakker , phys . rev .
lett . * 112 * , 196602 ( 2014 ) .
p. moon and m. koshino , phys .
b * 90 * , 155406 ( 2014 ) .
l. a. chizhova , f. libisch , and j. burgdrfer , phys .
b * 90 * , 165404 ( 2014 ) .
j. r. wallbank , m. mucha - kruczyski , x. chen , and v. i. falko , ann . phys . * 527 * , 359 ( 2015 ) . g. j. slotman , m. m. van wijk , pei - liang zhao , a. fasolino , m. i. katsnelson , and shengjun yuan phys . rev
. lett . * 115 * , 186801 ( 2015 ) .
r. bistritzer and a. h. macdonald , phys .
b , * 84 * , 035440 ( 2011 ) . | we investigate the magnetic minibands of a heterostructure consisting of bilayer graphene ( blg ) and hexagonal boron nitride ( hbn ) by numerically diagonalizing a two - band hamiltonian that describes electrons in blg in the presence of a moir potential . due to
inversion - symmetry breaking characteristic for the moir potential , the valley symmetry of the spectrum is broken , but despite this , the zero - energy landau level in blg survives , albeit with reduced degeneracy .
in addition , we derive effective models for the low - energy features in the magnetic minibands and demonstrate the appearance of secondary dirac points in the valence band , which we confirm by numerical simulations . then , we analyze how single - particle gaps in the fractal energy spectrum produce a sequence of incompressible states observable under a variation of carrier density and magnetic field . |
most galaxies in the local universe follow a metallicity - luminosity ( or metallicity - mass ) relation in the sense that more luminous galaxies also tend to be more metal - rich .
galaxies in the high - metallicity , high - mass regime with current stellar masses greater than @xmath1 m@xmath2 show very little increase in metallicity towards higher masses though @xcite , while there is a pronounced decline in metallicity towards lower luminosities ( or stellar masses ) and an increased scatter in the relation @xcite . taking three parameters into account , namely the stellar mass , the gas - phase metallicity , and the star formation rate , galaxies up to a redshift of @xmath3
follow a fairly tight `` fundamental metallicity relation '' in this three - dimensional space @xcite . at higher redshifts ,
galaxies show lower metallicities leading to increasing deviations from the relation . here
dilution by infall of pristine gas becomes a dominant factor despite active star formation as long as the dynamical time scales remain shorter than the chemical enrichment time scales @xcite . at lower redshift outflows of enriched material
become more important and a balance between infall and outflows develops @xcite , whereas interacting and merging galaxies deviate from the stellar mass - metallicity trend @xcite . in this review
, we consider metal - poor galaxies in the _ local _ universe , i.e. , galaxies at the low - luminosity , low - mass end of the metallicity - luminosity relation thus , dwarf galaxies .
exactly where to draw the line between `` dwarf '' and `` giant '' galaxies is more a matter of definition than of physical difference . often a simple luminosity criterion is used ( e.g. , a luminosity @xmath4 or @xmath5 mag ; see , e.g. , @xcite ) , which also excludes the massive galaxies along the classical hubble sequence .
dwarf galaxies cover a variety of morphological types , luminosities , present - day star formation activity , gas content , as well as environments @xcite .
late - type , gas - rich , star - forming dwarfs ( such as dwarf spirals , dwarf irregulars , or blue compact dwarfs ) are usually found in low - density environments like the outer regions of galaxy clusters , groups , or in isolation in the field . early - type , gas - deficient , quiescent dwarfs ( such as dwarf ellipticals , dwarf spheroidals , or ultra - compact dwarfs ) are primarily found in high - density environments , e.g. , in the vicinity of massive galaxies in groups and clusters .
this morphology - distance or gas fraction - distance relation @xcite of the dwarf galaxies perpetuates the morphology - density relation observed for giant early - type and late - type galaxies ( e.g. , @xcite ) and suggests that environmental effects may play an important role in shaping the evolution of dwarf galaxies ( e.g. , @xcite ) .
the best - studied dwarfs are the dwarf galaxies of the local group , which are sufficiently close be resolved into individual stars and to even permit us detailed studies of their stellar content down to their oldest main - sequence turn - offs ( see @xcite for recent examples ) .
all nearby dwarf galaxies studied in detail so far , regardless of their morphological type , contain very old stars with ages @xmath6 gyr @xcite . though the fractions of these old populations vary and more massive dwarf galaxies usually show more extended episodes of star formation , old populations ( i.e. , early star formation ) appear to be ubiquitous in dwarfs @xcite . in low - mass galaxies
the younger and/or metal - rich populations tend to be more centrally concentrated ( e.g. , @xcite ) .
similarly , more metal - rich , younger populations tend to be dynamically colder than older , more metal - poor populations ( e.g. , @xcite ) , though not all dwarf galaxies show such population gradients ( e.g. , @xcite ) .
taking advantage of the wealth of information available for nearby , low - metallicity dwarf galaxies and their identifiable , metal - poor old populations , they will be in the focus of this review ( although we will also consider galaxies beyond the local group ) .
many of the local group dwarf galaxies are even close enough for detailed spectroscopic abundance analyses of individual member stars , revealing not only the overall metal content , but the star - to - star variations in light and heavy elements of different nucleosynthetic origin ( e.g. , @xcite ) , which in turn allow us to draw inferences about the conditions under which these metal - deficient galaxies and their stars formed .
as pointed out by @xcite , star formation and molecular gas are closely related , and the ability to form molecular gas depends on metallicity ( including the presence of dust to provide shielding ) .
these authors argue that in the absence of metals , molecular gas can not easily form , leading to a long gas consumption time scale , a low star formation efficiency , and slow enrichment . above a critical metallicity threshold ,
however , both star formation rate and enrichment accelerate , resulting in a rapid increase in stellar mass and metallicity @xcite .
thus , without significant pre - enrichment by population iii stars , one would _ not _ expect to observe a continuously occupied metallicity - luminosity relation , but instead to see a bimodality in metallicity @xcite .
instead , the metallicity - luminosity relation for dwarf galaxies smoothly extends the correlation found for giant galaxies all the way down to ultra - faint dwarf spheroidal galaxies with luminosities as low as @xmath7 mag @xcite . particularly for dwarf irregular and dwarf spheroidal galaxies
, there is also an increasing `` mean age '' trend in the sense that the more metal - poor dwarf galaxies tend to be dominated by stellar populations with older ages ( see also @xcite ; in contrast to the behavior seen in , e.g. , giant elliptical galaxies ) . at optical wavelengths , the vigorously star - forming dwarf galaxies discussed in the next paragraph differ from this trend .
generally , the scatter in the metallicity - luminosity relation increases with decreasing luminosity ( @xcite ; see also @xcite ) .
metal - deficient star - forming dwarf galaxies , in particular blue compact dwarfs ( bcds ) , appear to be too luminous for their low metallicities @xcite , thus deviating from the mean locus of dwarf galaxies in the metallicity - luminosity relation .
their apparent overluminosity is likely largely an effect of the pronounced starbursts they are undergoing at the present time .
the analysis of the light and heavy element content of hii regions in such metal - poor bcds and the low dispersion of the resulting element abundance ratios suggests a probable primary origin of these elements in massive stars @xcite . this used to be considered an argument in favor of these galaxies being `` young '' in the sense that they are now undergoing their first burst of star formation ( e.g. , @xcite ) . the most metal - deficient bcd known , i zw 18 , which has a metallicity of only @xmath8 , has often been considered an epitome of a truly young galaxy in the present - day universe . _ hubble space telescope _ ( hst )
photometry of resolved stars in i zw 18 seemed to support the `` young galaxy '' scenario , since those data showed no trace of an older red giant branch ( rgb ) population @xcite .
even deeper hst photometry , however , revealed that i zw 18 is more distant than previously assumed and contains a well - defined rgb , demonstrating that this extremely metal - poor galaxy is not primordial @xcite . more generally , recent studies of the integrated light have confirmed that bcds are old systems currently experiencing starbursts ( e.g. , @xcite ) .
the stellar mass - weighted ages were found to be as old as up to 10 gyr , while the luminosity - weighted ages are of the order of only 10 myr @xcite .
the current star formation rates exceed the averaged past star formation rates by factors of more than two to three @xcite .
the colors of bcds are best reproduced when invoking continuous star formation and a recent burst @xcite , although theoretical chemical evolution models also suggest series of bursts in combination with metal - enhanced winds @xcite .
observational data suggest that the strongly enhanced star formation activity of bcds is often caused by mergers or interactions with a companion , whereas one possible mechanism to halt the burst activity are supernova - driven winds @xcite .
theoretical models support gas - rich dwarf - dwarf mergers , whose old populations turn into diffuse low - surface - brightness components @xcite .
the starbursts in the compact cores of bcds may be fueled by low - metallicity gas from the outer extended gas disks of the progenitor dwarfs , resulting in metal - poor young stars @xcite .
tidal dwarf galaxies are long - lived , gravitationally bound objects free of dark matter that form from `` recycled '' , mainly gaseous material previously torn out during interactions between massive ( spiral ) galaxies ( see @xcite and references therein ) .
the most massive tidal dwarfs are believed to form from matter concentrations at the tips of tidal tails and will experience starburst activity comparable to that of bcds @xcite .
tidal dwarf galaxies also deviate from the metallicity - luminosity relation . at a given luminosity ,
their metallicity _ exceeds _ that of regular dwarf galaxies since they are metal - rich already `` at birth '' , forming from the pre - enriched gas of much more massive galaxies @xcite .
in fact , their average metallicity ( 1/3 solar ) is essentially independent of their absolute luminosity @xcite .
the star formation efficiency in tidal dwarfs resembles that of spiral galaxies @xcite .
this may be attributed to the improved self - shielding due to their higher - than - normal metallicities ( as compared to other dwarf galaxies ) , facilitating the formation of molecular gas and star formation @xcite .
hst imaging suggests that the tidal dwarf candidates in the m81 group contain young stars formed in situ , but also older rgb stars that may have formed in the massive galaxies of this group @xcite . while too high a metallicity for a given luminosity is a good indicator for young tidal dwarf candidates , it is more difficult to identify _ old _ tidal dwarfs . depending on when they formed , these objects may be as metal - poor as `` normal '' dwarf irregulars @xcite .
in contrast to the previously discussed dwarfs ( sections 2.1 and 2.2 ) , dwarf irregular ( dirr ) , dwarf elliptical ( de ) , and dwarf spheroidal ( dsph ) galaxies typically follow a well - defined metallicity - luminosity relation ( see , e.g. , @xcite ) .
dsph galaxies usually host prominent old populations @xcite , and the mean metallicities of these systems are thus dominated by the metallicities of their ( old and intermediate - age ) rgb populations .
dirrs , on the other hand , have typically experienced continuous star formation with some amplitude fluctuations @xcite .
this mode of extended periods of star formation interrupted by quiescent phases is also called `` gasping '' star formation @xcite . that the ( predominantly old ) dsphs follow the same global metallicity - luminosity ( or metallicity - stellar mass ) relation as the dirrs suggests that the dsphs did not experience substantial tidal stripping despite their proximity to massive galaxies @xcite
. moreover , with the exception of dsphs that are currently being disrupted , their structure and kinematics do not show evidence of their being unbound tidal remnants ( e.g. , @xcite ) .
a closer investigation of the metallicity - luminosity relation reveals that there is an offset between dsphs and dirrs ( e.g. , @xcite ) , which persists even when limiting the comparison to the metallicities of the old populations in the two types of dwarfs @xcite . despite their older mean ages ,
dsph galaxies are more metal - rich than expected from their luminosities , which may imply that they experienced initially more vigorous star formation and enrichment than the slowly evolving dirrs @xcite . at a given stellar mass ,
the metallicities of dsphs are typically higher by a factor of three than those of dirrs @xcite .
this makes it difficult for dirrs to evolve into dsphs if their star formation were to cease since mere passive fading would require more than a hubble time , though such an evolution appears to be possible for low - luminosity transition - type dirr / dsph galaxies @xcite .
dsphs and dirrs may thus be intrinsically different and formed under different conditions @xcite .
alternatively , dsphs may have lost a large amount of their initial mass due to later tidal ( e.g. , @xcite ) and/or ram pressure stripping ( e.g. , @xcite ) while allowing the galaxies to survive as dark - matter - dominated bound entities ( e.g. , @xcite and sect.3 ) .
the low - luminosity end of the metallicity - luminosity relation consists exclusively of the recently discovered faint and ultra - faint ( @xmath9 mag ) dsph galaxies around the milky way and m31 ( e.g. , @xcite ) .
the ultra - faint dsphs are not only of interest for the cosmological substructure or missing satellite problem ( e.g. , @xcite ) , but are also intriguing objects from the point of view of galaxy evolution : they are the least massive , least luminous , most metal - poor , dark - matter - dominated galaxies known . as objects containing only old , metal - deficient stars , they may be potential `` pre - re - ionization fossils '' @xcite .
environment , i.e. , the local galaxy density , the degree of proximity to massive galaxies , the degree of activity in massive galaxies , and immediate galaxy interactions , affect the evolution and properties of low - mass galaxies . the earlier mentioned morphology - density relation is a prominent example of the impact of environment .
one of the intriguing questions in this context , particularly with respect to dwarf galaxies , is the importance of a galaxy s intrinsic properties ( `` nature '' ) vs. the importance of external influences ( `` nurture '' ) .
clearly , both play a role in shaping dwarf galaxies in groups and clusters ( e.g. , @xcite ) . in the local group and other nearby groups , early - type companions with hi masses below @xmath10 m@xmath2
are usually found within @xmath11 kpc around massive galaxies , whereas late - type dwarfs with @xmath12 m@xmath2 are mostly located at larger distances @xcite .
low - mass transition - type dirr / dsph galaxies fill the range in between these properties @xcite .
this seems to suggest that gas removal caused by various processes related to the proximity to massive galaxies may be the cause for the morphological segregation ( see also @xcite ) , in particular ram - pressure stripping , although the present - day galactic halo gas densities appear to be too low @xcite .
cosmological simulations taking into account gas cooling , star formation , supernova ( sn ) feedback , enrichment , and ultraviolet heating lead to satellite galaxies of which 95% are gas - free at the present time @xcite . according to these simulations ,
satellites need total masses of at least @xmath13 m@xmath2 in order to retain their gas .
interestingly , also isolated dwarf galaxies are predicted to be largely gas - deficient at the present time if their total masses are below @xmath14 m@xmath2 @xcite .
these authors conclude that while gas stripping aids in removing gas from dwarf satellites , the total mass of a dwarf galaxy ( thus an intrinsic property ) is the primary factor determining whether a galaxy retains or loses its gas , and the total mass also governs a galaxy s star formation and enrichment .
in contrast , other models advocate a stronger role of galaxian environment . for instance , in simplified models presented by @xcite , feedback - assisted ram pressure stripping or tidal stripping during the period when infalling dwarfs still experienced active star formation may have effectively removed the gas , reproducing the observed morphological segregation .
@xcite investigate the effects of re - ionization on satellite properties .
they find that `` external '' , uniform re - ionization from the cosmic radiation field may have less of an effect than `` internal '' , distance - dependent re - ionization from the most massive milky way progenitor .
assuming that the satellites were already at similar distances as today , the photoevaporation driven by the galactic radiation field first removed star - forming material from the inner satellites , allowing the outer ones to continue star formation for a longer period ( @xcite ; see also @xcite ) .
this `` internal '' scenario can reproduce the observed cumulative radial satellite distribution very well , in contrast to a scenario that only considers the external ionization field @xcite .
@xcite include four epochs of star formation in their simulations : ( 1 ) at @xmath15 star formation occurs in systems large enough for h@xmath0 cooling resulting in very metal - poor stars , ( 2 ) later through hi cooling ( leading to re - ionization ) , ( 3 ) during re - ionization until @xmath16 through further hi cooling in subhalos that are large enough not yet to have lost their gas , and finally ( 4 ) during the last 10 gyr through metal cooling .
ultra - faint dsphs with @xmath17 then formed very early in low - mass halos via h@xmath0 cooling @xcite .
the h@xmath0 cooling threshold of @xmath18 to @xmath10 m@xmath2 implies a luminosity threshold , which excludes the existence of ever fainter satellites .
satellites in the luminosity range of @xmath19 experienced star formation via hi cooling and photo - heating feedback , whereas most of the stars in the brighter satellites formed after re - ionization @xcite .
dwarf galaxies do not represent a simple stellar population of a single age and metallicity .
instead , they usually experience extended episodes of star formation leading to gradual enrichment and , depending on duration of star formation , also to a measurable range of ages ( e.g. , @xcite ) . for nearby dwarfs the spread in metallicity ( and even in individual element abundance ratios ) can be measured directly through absorption line spectroscopy of individual stars , particularly along the rgb . in more distant dwarfs
, the brightest supergiants may still be accessible ( e.g. , @xcite ) , while emission - line spectroscopy of planetary nebulae or hii regions can be conducted over a wide range of distances ( e.g. , @xcite ) .
recent spectroscopic analyses of large numbers of rgb stars in nearby dwarf galaxies resulted in well - sampled metallicity distribution functions ( mdfs ) for many dsphs and dirrs .
often these mdfs show a gradual rise toward higher metallicities and then a steeper fall - off , but exceptions exist ( e.g. , @xcite ) . a g - dwarf problem ( or ,
more accurately , a k - giant problem ) is found in all dwarfs ( e.g. , @xcite ) .
a comparison of the mdfs with simple chemical evolution models generally shows a poor fit for closed - box models or pristine gas models , but better results for pre - enrichment , leaky box , or models with additional gas infall ( e.g. , @xcite ) .
the range in stellar metallicity increases with galaxian luminosity ( or stellar mass ) @xcite .
the interplay between slow gas infall , low - efficiency star formation , and strong galactic winds can reproduce both the observed mdfs and element abundance ratios @xcite .
the pronounced decrease of the metal - rich end of the mdf is then caused by the removal of gas without winds , the resulting mdf would be much more metal - rich than observed @xcite
. other models , however , do not require strong winds but employ external mechanisms for the loss of heated , enriched gas such as ram pressure or tidal interactions @xcite .
shallow gradients of decreasing metallicity with galactocentric radius are commonly found in spectroscopic surveys @xcite , confirming the finding that more metal - rich and/or younger populations are more centrally concentrated and kinematically colder ( see section 1 and , e.g. , @xcite ) , although there are also exceptions ( e.g. , @xcite )
. the lack of a gradient may be due to insufficient radial area coverage in spectroscopic surveys @xcite or be caused by stripping ( see @xcite ) .
tidal stripping would primarily remove the more extended ( and more metal - poor ) stellar populations and homogenize the stellar radial velocities of the components @xcite .
@xcite , on the other hand , point out the danger of overinterpreting small stellar samples in favor of the existence of a gradient .
a number of theoretical models have explored the possible origin of population gradients .
for example , apart from sn - driven blow - outs sne in the outer regions of dwarfs may drive inward - propagating winds that ultimately concentrate cold gas in the inner regions @xcite .
star formation may re - start from this enriched gas @xcite , which would result in a metallicity gradient .
other models suggest that any gradients will be quickly erased since the winds of sne ii tend to homogenize the interstellar medium @xcite .
metallicity gradients are then introduced by the spatially inhomogeneous enrichment through the much rarer sne ia , which occur more frequently in the denser populated central regions @xcite .
@xcite show that the time scales of the change of the dark matter profiles of dwarf galaxies ( from cuspy to cored profiles ) due to feedback and the redistribution of ( baryonic ) matter on the one hand and of the formation of chemical gradients on the other hand may be linked .
the oldest stars in the dwarf move on radial orbits , and inner and outer regions are thus well mixed ( similar to @xcite , but due to a different mechanism ) .
the evolving density profile implies that younger , more metal - rich stars forming in the central regions tend to be on more circular orbits with smaller radial velocity dispersion , resulting in less mixing with the outskirts @xcite .
dwarf galaxies exhibit considerable stellar abundance spreads .
the range of metallicities often spans @xmath20 dex in [ fe / h ] even in galaxies dominated by old populations ( e.g. , @xcite . in several dwarf galaxies studied in detail it turns out that at _ any _ given age there is a spread in metallicity ( e.g. , @xcite )
. moreover , at a given overall metallicity one tends to find scatter in the @xmath21 abundance ratios ( e.g. , @xcite ) .
these galaxies were not well - mixed when these stars formed .
the trends as such differ from galaxy to galaxy , confirming once again that no two dwarfs are alike not even dwarfs of the same luminosity and morphological type @xcite .
there are a number of theoretical models addressing these issues .
it seems that at very early times and very low metallicities star formation was governed by stochasticity and inhomogeneous heavy - element pollution caused by the few early sne ii @xcite .
subsequently , with increasing enrichment , the feedback from these sne led to fairly homogeneous mixing @xcite . at higher metallicities , individual sne ia again led to localized inhomogeneous abundance patterns @xcite , resulting in a substantial element abundance spread even for coeval stars , just as seen in the observations ( see also @xcite ) .
considering trends in [ @xmath21/fe ] vs. [ fe / h ] , stars in dwarf galaxies resemble those in the galactic halo at low metallicity ( [ fe / h ] @xmath22 dex ) ( e.g. , @xcite ) , but tend to show lower [ @xmath21/fe ] ratios at higher metallicities as compared to typical galactic halo stars ( e.g. , @xcite ) .
sne ia thus seem to contribute already at lower metallicities in dwarfs than in our halo .
see @xcite for a nice demonstration of the age dependence of the [ @xmath21/fe ] vs.trend .
the different locations of the `` turn - over '' from the near - constant [ @xmath21/fe ] towards lower ratios appears to roughly depend on a galaxy s stellar mass .
lower ( than galactic ) [ @xmath21/fe ] ratios at a given [ fe / h ] may indicate ( 1 ) low star formation rates ( with few contributions of @xmath21 elements from sne ii ) , ( 2 ) a substantial loss of sn ejecta ( metals ) via galactic winds , or ( 3 ) a larger contribution from sne ia ( enhancing the fe content with respect to @xmath21 elements ) @xcite .
several studies show evidence for early rapid enrichment based on heavy neutron - capture and r - process element abundance ratios ( e.g. , @xcite ) , while low - efficiency , extended star formation with stochastic , inhomogeneous contributions from sne ia ejecta and asymptotic giant branch stars appears to have been important at higher metallicities ( e.g. , @xcite ) .
also , an initial mass function sparsely populated with massive stars due to the low star formation rate may contribute to the slow build - up of metallicity and to the inhomogeneities @xcite .
major efforts were devoted to the search for extremely metal - poor stars in our milky way and in nearby dwarf galaxies .
while no stars as metal - deficient as found in the galactic halo ( @xcite and references therein ) have been detected in galactic satellites yet , the current record holders have [ fe / h ] @xmath23 dex in the dsph sculptor @xcite and @xmath24 dex in the irregular large magellanic cloud ( lmc ) @xcite .
ultra - faint dsphs appear to be a particularly rewarding hunting ground for very metal - deficient stars as suggested by both observational data ( e.g. , @xcite ) and simulations ( e.g. , @xcite ) .
@xcite and @xcite argue that the abundance patterns in extremely metal - poor stars in the ultra - faint dsphs leoiv and hercules are consistent with enrichment through population iii sn explosions . according to simulations , ultra - faint dsphs may have experienced star formation via h@xmath0 cooling of pristine gas at high redshift ( @xcite ; sect .
3 ) , an inefficient process resulting in low star formation rates @xcite . these strongly dark - matter - dominated galaxies themselves may be merger products of very early h@xmath0-cooling minihalos @xcite , which may account for their metallicity spread . while the element abundance ratios in dsphs were once considered an argument against a significant contribution of such galaxies to the build - up of the galactic halo ( see , e.g. , @xcite and references therein ) , this picture has changed recently .
a number of studies point out the good agreement of the abundance ratios in metal - poor stars in the galactic halo and in dsphs .
this chemical similarity makes contributions from dsphs plausible and indicates that the early chemical evolution of galaxies on all scales may be similar ( e.g. , @xcite ) .
some models predict that most of the stars in the stellar halo ( particularly the inner halo ) were formed not in dsphs , but in a small number of massive dirr - like galaxies that then were accreted by the milky way some 10 gyr ago @xcite .
the element abundance ratios in genuinely old , metal - poor field and globular cluster stars in the magellanic clouds turn out to be in good agreement with galactic halo abundances @xcite .
these findings are compatible with the predicted early accretion of dirrs .
ultra - faint dsphs , on the other hand , may be a very important source of extremely metal - deficient stars in the galactic halo .
intriguingly , and contrary to earlier views , these galaxies may even contain _ too many _ such extremely metal - poor stars as compared to the galactic halo @xcite .
however , as emphasized by @xcite , most halo star studies focused on the _ inner _ halo .
carbon - enhanced metal - poor ( cemp ) stars may serve as a tracer for generally very metal - poor stars , and their number increases with decreasing metallicity and with height above the galactic plane @xcite . in the _ outer _ halo
the fraction of cemp stars is approximately twice as much as in the inner halo @xcite .
hence we may expect also a larger ( yet to be detected ) number of extremely metal - poor stars there . combining these findings and arguments ,
a picture emerges where the inner halo may consist largely of stars contributed early on by larger progenitor systems ( which make it difficult to uncover dsph contributions ) , while the outer halo may plausibly contain a significant stellar component accreted from small dwarfs such as dsphs ( @xcite ; see also @xcite ) . clearly , detailed element abundance information for more old stars in the outer halo , in the magellanic clouds , and in ultra - faint and classical dsphs is desirable to explore the importance of early accretion and of the origin of very metal - poor stars .
g. gilmore , and h. m. asiri , h. m. , `` are dsph galaxies galactic building blocks ? '' , in _ assembling the puzzle of the milky way _ , edited by c. reyl , a. robin , and m.schultheis , epj web of conferences * 19 * , edp sciences , les ulis , 2012 , 03001 e. k. grebel , `` evolutionary histories of dwarf galaxies in the local group '' , in _ the stellar content of local group galaxies _ , edited by p. whitelock & r. cannon , iau symp . *
192 * , astronomical society of the pacific , san francisco , 1999 , pp .
1738 e. k. grebel , `` the star formation history of the local group '' , in _
33rd eslab symposium on star formation from the small to the large scale _ , edited by f. favata , a. kaas , and a. wilson , esa sp * 445 * , european space agency , noordwijk , 2000 , pp .
8798 e. k. grebel , `` stellar populations in the local group of galaxies '' , in _ stellar astrophysics with the world s largest telescopes _ , edited by j. mikoaewska and a. olech , aip conf .
* 752 * , american institute of physics , melville , 2005 , pp .
161174 a. kayser , e. k. grebel , d. r. harbeck , et al .
, `` the age - metallicity relation of the smc '' , in _
globular clusters guides to galaxies _ , edited by t. richtler and s. larsen , eso astrophysics symposia , springer verlag , heidelberg , 2009 , pp . | a galaxy s mean metallicity is usually closely correlated with its luminosity and mass . consequently the most metal - poor galaxies in the local universe are dwarf galaxies . blue compact dwarfs and tidal dwarfs tend to deviate from the metallicity - luminosity relation by being too metal - poor or too metal - rich for their luminosity , respectively .
a less pronounced offset separates dwarf spheroidal ( dsph ) and dwarf irregular galaxies , making the former too metal - rich for their luminosity , which indicates different formation conditions for these two types of dwarfs . while environment ( photo - evaporation through local re - ionization by massive galaxies , tidal and ram pressure stripping ) govern the observed morphology - distance relation , intrinsic properties ( in particular total mass ) play a decisive role in dwarf galaxy evolution with respect to the time and duration of star formation and the amount of enrichment .
the metallicity distribution functions of nearby dwarfs can be understood taking pre - enrichment , gas infall , and winds into account .
many dwarfs show evidence for inhomogeneous , localized enrichment .
ultra - faint dsphs , which may have formed their metal - poor stars at high redshift via h@xmath0 cooling , show an overabundance of metal - deficient stars as compared to the ( inner ) galactic halo , but may , along with classical dsphs , have contributed significantly to the build - up of the outer halo .
the abundance ratios measured in the irregular large magellanic cloud are consistent with the postulated early accretion of irregulars to form the inner galactic halo .
address = astronomisches rechen - institut , zentrum fr astronomie der universitt heidelberg , mnchhofstr . 1214 , 69120 heidelberg , germany |
the chermnykh s problem is new kind of restricted three body problem which was first time studied by @xcite . given numerical exploration of chermnykh s problem , in which the equilibrium points and zero velocity curves studied numerically also the non - linear stability for the triangular lagrangian points are computed numerically for the earth - moon and sun - jupiter mass distribution when the angular velocity varies .
the mass reduction factor @xmath0 is expressed in terms of the particle radius @xmath1 , density @xmath2 radiation pressure efficiency factor @xmath3 ( in c.g.s .
system):@xmath4 . where @xmath5 is solar radiation pressure force which is exactly apposite to the gravitational attraction force @xmath6 and change with the distance by the same law it is possible to consider that the result of action of this force will lead to reducing the effective mass of the sun or particle .
@xcite examined the linear stability of triangular equilibrium points in the generalized photogravitational restricted three body problem with poynting - robertson drag , @xmath7 and @xmath8 points became unstable due to p - r drag which is very remarkable and important , where as they are linearly stable in classical problem when @xmath9 .
further the normalizations of hamiltonian and nonlinear stability of @xmath10 in the present of p - r drag has been studied by normal forms are a standard tool in hamiltonian mechanics to study the dynamics in a neighbourhood of invariant objects .
usually , these normal forms are obtained as divergent series , but their asymptotic character is what makes them useful . from theoretical point of view , they provide nonlinear approximations to the dynamics in a neighbourhood of the invariant object , that allows to obtain information about the real solutions of the system by taking the normal form upto a suitable finite order . in this case , it is well known that under certain(generic ) non - resonance conditions , the remainder of this finite normal form turns out to be exponentially small with respect to some parameters .
those series are usually divergent on open sets , it is still possible in some cases to prove convergence on certain sets with empty interior(cantor - like sets ) by replacing the standard linear normal form scheme by a quadratic one . from a more practical point of view , normal forms can be used as a computational method to obtain very accurate approximations to the dynamics in a neighbourhood of the selected invariant object , by neglecting the remainder .
they have been applied , for example , to compute invariant manifolds or invariant tori . to do that
, it is necessary to compute the explicit expression of the normal form and of the ( canonical ) transformation that put the hamiltonian into this reduced form .
a context where this computational formulation has special interest is in some celestial mechanics models , that can be used to approximate the dynamics of some real world problems .
the linearization of hamiltonian , have been studied numerically and analytically .
we have examined the effect of gravitational potential from the belt , oblateness effect and radiation effect .
let us consider the model proposed by @xcite , according to this model the potential of belt is given by : @xmath11\left ( \mathbf{a}+n\right)^2}{n^3\left[r^2+\left(\mathbf{a}+n\right)^2\right]^{5/2}}\end{aligned}\ ] ] where @xmath12 is the total mass of the belt and @xmath13 , @xmath14 are parameters which determine the density profile of the belt , if @xmath15 then the potential equals to the one by a point mass .
the parameter @xmath1 flatness parameterand @xmath16 core parameter , where @xmath17 , @xmath18 , @xmath19 .
then we obtained @xmath20 and @xmath21 .
as in , we consider the barycentric rotating co - ordinate system @xmath22 relative to inertial system with angular velocity @xmath23 and common @xmath24axis .
we have taken line joining the primaries as @xmath25axis .
let @xmath26 be the masses of bigger primary(sun ) and smaller primary(earth ) respectively .
let @xmath27 , @xmath28 in the equatorial plane of smaller primary and @xmath29 coinciding with the polar axis of @xmath30 . let @xmath31 , @xmath32 be the equatorial and polar radii of @xmath30 respectively , @xmath33 be the distance between primaries .
let infinitesimal mass @xmath34 be placed at the point @xmath35 .
we take units such that sum of the masses and distance between primaries as unity , the unit of time taken such that the gaussian constant of gravitational @xmath36 . then perturbed mean motion @xmath37 of the primaries
is given by @xmath38 , where @xmath39 , @xmath40 is oblateness coefficient of @xmath30 . where @xmath41 is mass parameter , @xmath42 with @xmath43
. then coordinates of @xmath44 and @xmath45 are @xmath46 and @xmath47 respectively . the mass of sun @xmath48 @xmath49(the mass of earth ) , hence mass parameter for this system is @xmath50 . in the above mentioned reference system , we determine the equations of motion of the infinitesimal mass particle in @xmath51-plane . now using @xcite profile and ,
the equations of motion are given by : @xmath52 where @xmath53 the energy integral of the problem is given by @xmath54 , where the quantity @xmath55 is the jacobi s constant .
the zero velocity curves @xmath56 are presented in figure ( [ fig : plotzvc ] ) for the entire range of parameters@xmath57 and @xmath58 .
we have seen that there are closed curves around the @xmath10 so they are stable but the stability range reduced ( [ fig : jvc&q0 ] ) ] due radiation effect .
the closed curves around @xmath10 disappeared when @xmath59 .
, for all values of @xmath60,title="fig : " ] , for all values of @xmath60,title="fig : " ] , in frame one @xmath61 , in second frame show the curves in classical case.,title="fig : " ] + , in frame one @xmath61 , in second frame show the curves in classical case.,title="fig : " ] the position of equilibrium points are given by putting @xmath62 i.e. , @xmath63 from equations ( [ eq : eq1pts ] , [ eq : eq2pts ] ) we obtained : @xmath64\label{eq : r1},\\ r_2&&=1+\frac{\mu(1 - 2r_c)m_b}{3\left(r_c^2+t^2\right)^{3/2 } } \label{eq : r2}\end{aligned}\ ] ] from above , we obtained : @xmath65^{-2/3}-y^2\right]^{1/2 } \label{eq : x1lag}\\ & & x=1-\mu\nonumber\\&&\pm\bigl[\left[1-\frac{\mu(1 - 2r_0)m_b}{\left(r_c^2+t^2\right)^{3/2}}\right]^{-2/3}-y^2\bigr]^{1/2}\label{eq : x2lag}\end{aligned}\ ] ] the triangular equilibrium points are given by putting @xmath62 , @xmath66 , then from equations ( [ eq : omegax ] ) and ( [ eq : omegay ] ) we obtained the triangular equilibrium points as:@xmath67}{3\left(r_c^2+t^2\right)^{3/2}}\label{eq : xl4}\end{aligned}\ ] ] @xmath68}{3\left(r_c^2+t^2\right)^{3/2}}\right]^{1/2}\label{eq : yl4 } \end{aligned}\ ] ] all these results are similar with @xcite , @xcite , and others .
we have to expand the lagrangian function in power series of @xmath25 and @xmath69 , where @xmath70 are the coordinates of the triangular equilibrium points
. we will examine the stability of the triangular equilibrium points .
for this we will utilize the method of @xcite . by taking @xmath71
, we will consider linear equations this we have established the relations between perturbed basic frequencies .
the lagrangian function of the problem can be written as @xmath72 and the hamiltonian @xmath73 , where @xmath74 are the momenta coordinates given by @xmath75 let us suppose @xmath76 , with @xmath77 and rewriting the coordinates of triangular equilibrium points as : @xmath78 where @xmath79 .
we shift the origin to @xmath7 , change @xmath80 , @xmath81 . and
@xmath82,then @xmath83 we have to expand @xmath84 in power series of @xmath85 and @xmath69 , for this we use taylor s expansion i.e. @xmath86+\frac{1}{2}[x^2f_{xx}(0,0)+2xyf_{xy}(0,0)+y^2f_{yy}(0,0 ) ] \nonumber\\&&+\frac{1}{3!}[x^3f_{xxx}(0,0)+3 x^2yf_{xxy}(0,0)+3xy^2f_{xyy}(0,0)+y^3f_{yyy}(0,0)]+\dots \label{eq : texp}\end{aligned}\ ] ] using these values in ( [ eq : texp ] ) and with the help of ( [ eq : a ] ) and ( [ eq : b ] ) we get @xmath87 where @xmath88 are @xmath89n^2+\frac{(1-\mu ) q_1 } { \sqrt{a^2+b^2}}\nonumber\\&&+\frac{\mu } { \sqrt{(a-1)^2+b^2}}\left\{1+\frac{a_2}{2(a-1)^2+b^2}\right\}+\frac{m_b}{\left((a-\mu)^2+b^2+t^2\right)^{3/2}}\end{aligned}\ ] ] @xmath90\nonumber\\ & & y\left[2b n^2-\frac{b ( 1-\mu)q_1}{\left(a^2+b^2\right)^{3/2}}-\frac{3bm_b}{\left((a-\mu)^2+b^2+t^2\right)^{5/2}}\right.\nonumber\\&&\left.-\frac{b\mu } { \left(a-1)^2+b^2\right)^{3/2}}\left\{1+\frac{3 a_2}{2(a-1)^2+b^2}\right\}\right]\end{aligned}\ ] ] @xmath91 @xmath92\label{eq : e}\end{aligned}\ ] ] @xmath93\label{eq : f } \end{aligned}\ ] ] @xmath94\label{eq : g } \end{aligned}\ ] ] using the method in @xcite , to find the canonical transformation from the phase space @xmath95 into the phase space product of the angle co - ordinates @xmath96 and the action momenta co - ordinates @xmath97 and of the first order in @xmath98 .
we consider the following linear equations in the variables @xmath99 : @xmath100
@xmath101 \quad { and } \quad a=\left[\begin{array}{c c c c } 2e & g&\lambda & -n\\ g&2f&n&\lambda\\ -\lambda & n & 1 & 0\\ -n & -\lambda & 0 & 1\end{array}\right]\ ] ] the characteristic equation of the hamiltonian @xmath102 is given by @xmath103 i.e. @xmath104 stability is assured only when the discriminant @xmath105 , where @xmath106 from stability condition we obtained : @xmath107\end{aligned}\ ] ] where @xmath108 .
when @xmath105 the roots @xmath109 and @xmath110 ( @xmath111 being the long / short -periodic frequencies ) are related to each other as @xmath112\right.\nonumber\\&&\left.+\frac{81m_b}{2}(12 + 30\epsilon+30\mu+95\mu\epsilon)+135m_ba_2(18 + 58\epsilon+45\mu+188\mu\epsilon)\right.\nonumber\\&&\left.+\frac{1}{2\left(r_c^2+t^2\right)^{3/2}}\bigl\{m_b\left(180 + 2(2r_c-1)(44 + 21)\mu\epsilon\right.\right.\nonumber\\&&+\left.\left.72r_c+4r_ca_2[78 + 180\mu+(253 + 873\mu)\epsilon]\right)\bigr\}\right]\label{eq : w1+w2}\end{aligned}\ ] ] @xmath113 + 18[31\epsilon+m_b(144 + 647\epsilon)]a_2\right.\nonumber\\&&+\left.\frac{m_b\left\{4 [ 36 - 47\epsilon+4r_c(9
+ 28\epsilon)]+3[74 - 297\epsilon+4r_c(44 + 175\epsilon)]a_2\right\}}{\left(r_c^2+t^2\right)^{3/2}}\right]\nonumber\\&&+\mu\left[\frac{27}{4}+\frac{99\epsilon}{8}+\frac{117}{4}+73a_2\epsilon + m_b\left\{\frac{45}{4}+81a_2+\frac{273\epsilon}{8}+\frac{2357a_2\epsilon } { 8}\right.\right.\nonumber\\&&\left.\left.+\frac{396(2r_c-1)+4(772r_c-395)\epsilon+12(326r_c-181)a_2+(18452r_c-9667)a_2\epsilon}{72\left(r_c^2+t^2\right)^{3/2 } } \right\}\right]\nonumber\\&&+\mu^2\left[\frac{-27}{4}+\frac{111\epsilon}{8}+\frac{117}{4}+\frac{161a_2\epsilon}{2 } + m_b\left\{\frac{405}{4}+\frac{495a_2}{2}+\frac{4185\epsilon}{16}+\frac{25275a_2\epsilon } { 16}\right.\right.\nonumber\\&&\left.\left.+\frac{(2r_c-1)[198 - 838\epsilon-1014a_2 - 5117a_2\epsilon]}{36\left(r_c^2+t^2\right)^{3/2 } } \right\}\right]\label{eq : w1w2}\end{aligned}\ ] ] where @xmath114 satisfy the property @xmath115 . and
@xmath116 , @xmath117 the mass parameter is shown in figure @xmath118 , @xmath119 , surface ( 1)@xmath120 ( 2 ) @xmath121 , ( 3)@xmath122,(4)@xmath123 .
for expressing @xmath71 in a simpler form , we consider the set of linear equations ( [ eq : ax ] ) the solution of which can be obtained as @xmath124 substituting @xmath125 and @xmath126 , we obtain the solution sets as @xmath127 @xmath128 @xmath129 @xmath130 where @xmath131 and @xmath132 are constants of proportionality . following the method for reducing @xmath71 to the normal form , as in @xcite , use the transformation @xmath133,\quad t=\left[\begin{array}{c } q_1\\q_2\\p_1\\p_2\end{array}\right]\label{eq : xjt}\ ] ] and @xmath134_{1\leq i , j \leq 4}$ ] i.e. @xmath135\ ] ] @xmath136 under normality conditions : @xmath137 equivalently , @xmath138 @xmath139 being arbitrary , we follow the approach of @xcite and choose @xmath140 , implying that @xmath141 i.e. @xmath142 using equations ( [ eq:4ik1 ] ) , ( [ eq:4ik2 ] ) , ( [ eq : k1k3h1 ] ) and ( [ eq : k2k4h2 ] ) we observe that @xmath143 where @xmath144 it is now verified that @xmath71 takes the form : @xmath145 we observe that @xmath146_{4\times4}=\left[\begin{array}{c c c c } 0 & 0 & -\frac{m_1}{\omega_1\overline{m_1 } } & \frac{i m_2}{\omega_2\overline{m_2}}\\ -\frac{2n\omega_1}{m_1\overline{m_1 } } & \frac{2n\omega_2}{m_2\overline{m_2 } } & \frac{g } { \omega_1m_1\overline{m_1 } } & \frac{i g } { \omega_2 m_2\overline{m_2}}\\ -\frac{\omega_1(m_1 ^ 2 - 2n^2)}{m_1\overline{m_1 } } & \frac{i\omega_2(m_2 ^ 2 - 2n^2)}{m_2\overline{m_2 } } & \frac{-ng}{\omega_1 m_1\overline{m_1 } } & \frac{n ig } { \omega_2 m_2\overline{m_2}}\\ \frac{-\omega_1 g } { m_1\overline{m_1 } } & \frac { i\omega_1 g } { \omega_2 m_2\overline{m_2 } } & -\frac{n(\omega_1 ^ 2-m_1 ^ 2)}{\omega_1 m_1\overline{m_1 } } & -\frac{ni(2\omega_2 ^ 2-m_2 ^ 2)}{\omega_2 m_2\overline{m_2 } } \end{array}\right]\ ] ] with @xmath147 and @xmath148 . applying a contact transformation from @xmath149 to @xmath150 defined by @xcite @xmath151 and @xmath152\ ] ] i.e. @xmath153 the hamiltonian @xmath71 is transformed into the form @xmath154 denoting the angular variables @xmath155 and @xmath156 by @xmath157 and @xmath158 and the actions
@xmath159 and @xmath160 by @xmath161 respectively , the second order part of the hamiltonian transformed into the normal form @xmath162 the general solution of the corresponding equations of motion are @xmath163 if the oscillations about @xmath7 are exactly linear , the equation ( [ eq : intmt ] ) represent the integrals of motion and the corresponding orbits are given by @xmath164
we have seen there are closed curves around the @xmath10 so they are stable but the stability range reduced due to radiation effect . the effect of oblateness and mass of the belt is presented .
we have found the normal form of the second order part of the hamiltonian .
for this we have solved the aforesaid set of equations . under the normality conditions ,
we have applied a transformation defined by @xcite .
we have also utilized the approach of @xcite for reducing the second order part of the hamiltonian into the normal form .
we have found that the second order part @xmath71 of the hamiltonian is transformed into the normal form @xmath165 i would like to acknowledge and extend my heartfelt gratitude to dr .
uday dolas for her vital encouragement and support .
especially , i would like to give my special thanks to my wife mrs .
snehlata kushwah(pinky ) whose patient love enabled me to complete this research article . | linearization of the hamiltonian is being performed around the triangular equilibrium points in the generalized photogravitational chermnykh s problem .
the bigger primary is being considered as a source of radiation and small primary as an oblate spheroid .
we have found the normal form of the second order part of the hamiltonian .
for this we have solved the aforesaid set of equations . .
the effect of radiation pressure , gravitational potential from the belt on the linear stability have been examined analytically and numerically .
_ keywords _ : linearization , chermnykh s problem , radiation pressure , generalized photogravitational , rtbp . |
in this section , we will first explain the basic principle of the photonic faraday rotation . and two degenerate ground states @xmath13 and @xmath14 .
it is also shown in refs.@xcite.,width=151 ] from fig .
1 , a three - level atom is trapped in the cavity .
it has an excited state @xmath15 and two degenerate ground states @xmath13 and @xmath14 .
the transitions between the @xmath15 and @xmath13 , @xmath15 and @xmath14 are assisted with a left - circularly ( @xmath16 ) and right - circularly ( @xmath17 ) polarized photon , respectively .
we can write the hamiltonian of the system as @xcite @xmath18\nonumber\\ & + & \hbar\lambda\sum_{j = l , r}(a_{j}^{\dagger}\sigma_{j-}+a_{j}\sigma_{j+})+h_{r},\end{aligned}\ ] ] with @xmath19\nonumber\\ & + & \int_{-\infty}^{\infty}d\omega\sum_{j = l , r}\overline{\alpha}(\omega)(c_{j}^{\dagger}\sigma_{j-}+c_{j}\sigma_{j+}).\end{aligned}\ ] ] here @xmath20 and @xmath21 are the creation and annihilation operators of the cavity field and the frequency of the cavity field is @xmath22 .
the @xmath23 , @xmath24 , @xmath25 and @xmath26 are the lowering and raising operators of the transition r(l ) , respectively .
the @xmath27 is the atomic frequency and @xmath28 means the hamiltonian of the free reservoirs .
briefly speaking , we consider a single photon pulse with frequency @xmath29 enters the optical cavity .
using the adiabatic approximation , we can obtain the input - output relation of the cavity field in the form of @xcite @xmath30[i(\omega_{0}-\omega_{p})+\frac{\gamma}{2 } ] + \lambda^{2}}{[i(\omega_{c}-\omega_{p})+\frac{\kappa}{2}][i(\omega_{0}-\omega_{p})+\frac{\gamma}{2}]+\lambda^{2}}.\label{reflect1}\end{aligned}\ ] ] from eq .
( [ reflect1 ] ) , if @xmath31 , above equation for an empty cavity can be written as @xmath32 here @xmath33 is the frequency of the photon . @xmath34 and @xmath35 are the cavity damping rate and atomic decay rate .
the @xmath36 is the atom - cavity coupling strength .
the total physical picture of the process can be described as follows : if a left circular polarization photon enters the cavity , and couples with the atom in the state @xmath13 , the photon will be reflected to the output mode but induce a phase shift , which can be described as @xmath37 . here @xmath38 .
interestingly , if the right - circular photon enters the cavity and couples with the atom in the state @xmath13 , it only senses the empty cavity . after the photon reflecting from the cavity
, the whole process can be written as @xmath39 . here
therefore , if a single photon state @xmath41 enters the cavity and couples with the @xmath13 atom , the whole system can be described as @xmath42 on the other hand , if @xmath41 enters the cavity and couples with the @xmath14 atom , the whole system can be described as @xmath43 the @xmath44 is called the faraday rotation .
if we consider the cases that @xmath45 , @xmath46 , and @xmath47 , we can obtain @xmath48 and @xmath49 .
we can rewrite the relationship between the photon and the atom as @xmath50 polarization and reflect the photon in the @xmath51 polarization . here
d1 , d2 , d3 and d4 are four single - photon detectors.,width=264 ]
now we start to explain our protocol . as shown in fig . 1
, two pairs of atomic entanglement are shared by alice and bob , respectively .
suppose that they are in the same arbitrary two - qubit states of the form @xmath53 here @xmath54 and @xmath55 .
first , they prepare two single photons with the same polarization @xmath56 .
they let the two photons in the intput1 and input2 pass through the cavities , respectively .
the four atoms can be written as @xmath57 therefore , after the two photons passing through the four cavities , if both photons do not change and transmit the polarization beam splitters ( pbss ) and finally are detected by single photon detectors d@xmath58 and d@xmath59 , the atomic state will become @xmath60 with the total success probability of @xmath61 .
after they obtain the state @xmath62 in eq .
( [ step1 ] ) , they perform the hadamard operations on the atoms a1 and a2 , which makes the @xmath63 become @xmath64 from eq .
( [ step2 ] ) , after performing the hadamard operation , the atoms a1 and a2 are entangled .
finally , they still prepare a single photon in the input1 mode of the form @xmath41 , and let it pass through the two cavities a1 and a2 . if the photon s polarization does not change , it will project the atoms a1 and a2 into the singlet state @xmath65 therefore , if the photon passes through the cavities and finally be detected by the single - photon detector d@xmath58 in a second time , the state @xmath66 will collapse to @xmath67 the success probability for obtaining the state @xmath68 is @xmath69 .
the total success probability for obtaining the state @xmath68 is @xmath70 .
therefore , we can get @xmath71 . we let @xmath72.,width=302 ] interestingly , from eq .
( [ c ] ) , we have established the relationship between the concurrence of the atomic entanglement and the success probability for measuring the photons . in a practical experiment , we should repeat this protocol for many times to obtain the success probability @xmath73 by calculating the ratio between the detected photon number and the initial total photon number .
so far , we have briefly described our protocol with a simple example .
it is shown that we can complete the detection of the concurrence of arbitrary two - qubit atomic pure state by input - output processing of photons . in the pioneer work of romero _ et al .
_ , they also discussed the measurement for concurrence of atomic two - qubit pure states .
they require the atomic cnot gate to complete the task .
however , the cnot gate can not be well performed in current experimental conditions . in this protocol
, we do not require the cnot gate , and we also do not require the atoms to interact with each other . in our protocol ,
@xmath74rb atom trapped in a fiber - based fabry - perot cavity is a good candidate system @xcite . in ref.@xcite
, the two ground states @xmath13 and @xmath14 can be the states of @xmath75 of level @xmath76 .
the transition frequency between the excited state and the ground state is @xmath77 at @xmath78 .
the cavity length @xmath79 .
the cavity decay rate @xmath80 mhz , and the finesse @xmath81 .
certainly , in a practical experiment , we also should consider the experimental imperfection .
for example , in this protocol , we require @xmath45 , @xmath46 , and @xmath47 . actually , in practical experiment , we can not ensure to control @xmath48 to let @xmath47 . on the other hand
, the photon loss and the imperfect detection will also affect the quality of the whole protocol .
if @xmath82 , we should rewrite the relationship in eq .
( [ relation ] ) as @xmath83 from above description , it is shown that if the atomic state is @xmath84 or @xmath85 , the polarization state is not changed .
that is to say , the imperfect operation does not affect the fidelity of the state .
if the atomic state is @xmath86 or @xmath87 , it will make the polarization of the photon change , which will has some probability to be @xmath41 , and finally be detected by d@xmath58 or d@xmath59 . in this way
, it will contribute a successful case with the probability of @xmath88 in fig .
3 , we calculate the @xmath89 altered with @xmath90 . from eq .
( [ odd1 ] ) , in an ideal case , the item @xmath91 has no contribution to the successful case , while it will has the probability of @xmath92 to make d@xmath58 and d@xmath59 both fire .
the second item @xmath93 also has the probability of @xmath94 to make d@xmath58 and d@xmath59 both fire .
therefore , if @xmath95 , we can estimate the total success probability as @xmath96 here we omit the contribution of @xmath97 for @xmath98 .
therefore , the concurrence will be increased because of the imperfection . on the other hand
, we should also consider the photon loss .
the photon loss is the main obstacle in realistic experiment .
the minor misalignment , cavity mirror absorption and scattering , and even the tiny piece of dust , can induce the photon loss .
the photon loss will decrease the total success probability .
suppose the efficiency of a single - photon detector is @xmath99 , the efficiency of coupling and transmission of the photon from fiber to cavity is @xmath100 , and the transmission of each photon through the other optical components is @xmath101 , we can rewrite the total success probability in a detection round as @xmath102 in practical condition , the efficiency of the single - photon detector is @xmath103 .
suppose that @xmath104 , and @xmath105 .
we can obtain that @xmath106 .
if the generation rate of the single - photon is @xmath107 , this protocol can be completed in 0.01s . in conclusion
, we have described the detection of the nonlocal atomic entanglement assisted with single photons . through the auxiliary single photons input - output process in cavity
quantum electrodynamics ( qed ) , the concurrence of the atomic entanglement can be obtained according to the success probability of picking up the singlet states of the atoms .
we also discussed the practical imperfection of this protocol .
it is shown that the success probability will be increased if @xmath108 .
on the other hand , the photon loss will decrease the total success probability .
this protocol has several advantages : first , we do not require the sophisticated cnot gates .
second , the distributed atoms are not required to intact with each other . third
, the atomic entanglement can be distributed nonlocally , which provides its important applications in distributed quantum computation .
this work is supported by the national natural science foundation of china under grant nos .
11104159 and 11347110 . and
the project funded by the priority academic program development of jiangsu higher education institutions . | we present an efficient way for measuring the entanglement of the atoms . through the auxiliary single photons input - output process in cavity
quantum electrodynamics ( qed ) , the concurrence of the atomic entanglement can be obtained according to the success probability of picking up the singlet states of the atoms .
this protocol has three advantages : first , we do not require the sophisticated controlled - not ( cnot ) gates .
second , the distributed atoms are not required to intact with each other .
third , the atomic entanglement can be distributed nonlocally , which provides its important applications in distributed quantum computation .
entanglement plays an important role in current quantum information processing ( qip)@xcite .
quantum teleportation@xcite , quantum key distribution @xcite , quantum dense coding @xcite , quantum secure direct communication @xcite and other quantum information protocols @xcite , all rely on the entanglement between distant parties .
however , entanglement is difficult to characterize experimentally .
bell inequalities @xcite , entanglement witnesses @xcite can not provide satisfactory results of the entanglement because they disclose the entanglement of some states while fail for other states .
another indirect method for measuring the entanglement is the quantum state tomographic reconstruction @xcite . by reading out the 15 parameters , they can reconstruct the density matrix of a two - qubit state , but it is quite complicated . in the early work of bennett _
et al . _
, they proposed the way for measuring the quantify entanglement named entanglement of formation @xcite . in their protocol ,
an arbitrary two - qubit state with the density matrix @xmath0 can be described in terms of exactly calculable quantity , i. e. the concurrence ( c ) .
the concurrence of the two - qubit state can be defined as @xcite @xmath1 with the @xmath2 .
@xmath3 are the non - negative eigenvalues of the hermitian matrix @xmath4 in decreasing order . here
@xmath5 , and @xmath6 is the pauli operator , and @xmath7 is the complex conjugate of @xmath0 . if we consider an arbitrary two - qubit pure state of the form @xmath8 , with @xmath9 , the concurrence can be described as @xmath10 . in 2006 , walborn _ et al .
_ reported their experiment about the determination of entanglement with a single measurement @xcite . in their experiment
, they demonstrated the measuring of the concurrence for a two polarized state of the form @xmath11 and the concurrence can be described as @xmath12 .
they require the hyperentanglement to complete the task .
recently , the group of cao also discussed the measurement of the concurrence for two - photon polarization entangled pure state with the help of cross - kerr nonlinearity @xcite . on the other hand ,
the cavity quantum electrodynamics ( qed ) has become an important platform to realize the qip @xcite . over past decades
, both excellent theatrical and experiments have focused on qed for qip , such as the generation the single photons @xcite , the performance of the logic gate @xcite , and so on @xcite . in the early work of qed
, they should require the high - quality cavity and the strong coupling between the solid qubit and the cavity .
recently , the qips based on the low - q cavity have been widely discussed .
for example , an _ et al .
_ showed that with the help of faraday rotation , the different polarized photons can obtain the different phase shift after they interact with the atoms which are trapped in a low - q cavity @xcite .
subsequently , a lot of excellent works for atoms were discussed , including the quantum teleportation @xcite , controlled teleportation @xcite , swapping @xcite , entanglement concentration @xcite , and so on @xcite .
interestingly , inspired by the previous works of atoms and the measurement of the concurrence , we find that the faraday rotation can also be used to perform the measurement of the atomic entanglement .
actually , in 2007 , romero _ et al .
_ first discussed the direct measurement for concurrence of atomic two - qubit pure state @xcite . in their protocol , they illustrate the protocol for rydberg atoms crossing three - dimensional microwave cavities and confined ions in a linear paul trap .
they require the controlled - not ( cnot ) gate or the controlled - phase gate between two atoms to complete the task . in 2008 , lee _ et al .
_ presented the concurrence of a two - qubit cavity system with the help of flying atoms @xcite .
the measurement for concurrence based on trapped ions were also proposed @xcite . in this paper
, we will propose the detection of the remote atomic entanglement with the help of single photons .
this protocol is quite different from others .
first , it does not require the sophisticated controlled - not ( cnot ) gates and the high - q cavities , which will greatly release the experimental complexity .
second , the distributed atoms are not required to intact with each other .
third , the atomic entanglement can be distributed nonlocally , which provides its important applications in distributed quantum computation .
this paper is organized as follows : in sec .
2 , we first briefly describe the key element of this protocol , i. e. the basic principle of the faraday rotation . in sec . 3 , we will explain our protocol with a simple example . in sec . 4
, we will make a discussion about the practical imperfection in experiment . in sec . 5
, we will make a conclusion . |
tightness is a fundamental notion in @xmath1-dimensional contact topology .
it is characterized by the thurston - bennequin inequality ( see 1.1 ) .
roughly , for a seifert ( hyper)surface @xmath0 in a contact manifold , this inequality compares the contact structure along @xmath0 with the tangent bundle of @xmath0 by means of relative euler number .
an overtwisted disk is a @xmath2-disk equipped with a certain germ of @xmath1-dimensional contact structure , for which the inequality fails .
a contact manifold is said to be tight if it contains no embedded overtwisted disks .
then the inequality automatically holds for any @xmath0 ( eliashberg@xcite ) .
the @xmath3-jet space @xmath4 for a function @xmath5 is tight ( bennequin@xcite ) .
thus all contact @xmath1-manifolds , which are modelled on @xmath4 , are locally tight . giroux@xcite smoothly approximated a given compact surface in a contact @xmath1-manifold by a surface with certain transverse monotonicity , i.e. , a convex surface ( see 1.2 for the precise definition ) .
thus , in @xmath1-dimensional case , we may easily consider that the above inequality is only for convex surfaces .
contrastingly , in higher dimension , we show the following theorem . in the case where @xmath6 , any open subset of the @xmath3-jet space @xmath7 , which is the model space for contact @xmath8-manifolds , contains a hypersurface @xmath0 such that 1 .
@xmath0 violates the thurston - bennequin inequality , and 2 .
no convex hypersurface smoothly approximates @xmath0 .
this leads us to seriously restrict the inequality to convex hypersurfaces ( see @xcite for a sequel ) .
let @xmath0 be a compact oriented hypersurface embedded in a positive contact manifold @xmath9 @xmath10 which tangents to the contact structure @xmath11 at finite number of interior points .
let @xmath12 ( resp .
@xmath13 ) denote the set of positive ( resp .
negative ) tangent points . with respect to the symplectic structure @xmath14
, the symplectic orthogonal of @xmath15 defines a singular line field @xmath16 .
the integral foliation @xmath17 of @xmath18 on @xmath0 is called the
_ characteristic foliation_. the singularity of @xmath18 coincides with @xmath19 .
the restriction @xmath20 defines a holonomy invariant transverse contact structure of @xmath21 and determines the orientation of @xmath18 ( i.e. , @xmath22 ) .
we define the index @xmath23 of each tangent point @xmath24 by regarding it as a singular point of @xmath18 .
assume that the boundary of each connected component of @xmath0 is non - empty and @xmath18 is outwards transverse to @xmath25 .
then the boundary @xmath26 is said to be _ contact - type_. the unit @xmath2-disk @xmath27 equipped with the germ of contact structure @xmath28 is called an ( the ) overtwisted disk , where @xmath29 is the cylindrical coordinates of @xmath30 . slightly extending @xmath27 , we obtain a disk with contact - type boundary such that the singularity of the characteristic foliation is a single negative sink point .
a contact @xmath1-manifold is said to be _ overtwisted _ or _ tight _ depending on whether it contains an embedded overtwisted disk ( with the same germ as above ) or not .
let @xmath0 be _ any _ surface with contact - type boundary embedded in the @xmath3-jet space @xmath31 equipped with the canonical contact form .
then bennequin@xcite proved the following inequality which immediately implies the tightness of @xmath4 : @xmath32 eliashberg proved the same inequality for symplectically fillable contact @xmath1-manifolds ( @xcite ) , and finally for all tight contact @xmath1-manifolds ( @xcite ) .
the inequality can be written in terms of relative euler number .
let @xmath33 be the above vector field on a hypersurface @xmath34 with contact - type boundary . then , since @xmath35 , we can regard @xmath33 as a section of @xmath36 which is canonical near the boundary @xmath25 .
thus we can define the relative eular number of @xmath36 by @xmath37\rangle = \sum_{p\in s_+(\sigma)}\mathrm{ind}\,p-\sum_{p\in s_-(\sigma)}\mathrm{ind}\,p.\ ] ] then the thurston - bennequin inequality may be expressed as @xmath38\rangle \leq -\chi(\sigma).\ ] ] there is also an absolute version of the thurston - bennequin inequality which is expressed as @xmath39\rangle|\leq -\chi(\sigma)$ ] , or equivalently @xmath40 for any closed hypersurface @xmath0 with @xmath41 .
this holds if the euler class @xmath42 is a torsion , especially if @xmath43 .
note that the inequality and its absolute version can be defined for any oriented plane field on @xmath44 ( see eliashberg - thurston@xcite ) .
they are originally proved for codimension @xmath3 foliations on @xmath44 without reeb components by thurston ( see @xcite ) .
a vector field @xmath33 on @xmath9 is said to be _ contact _ if the lie derivative @xmath45 vanishes on @xmath11 .
let @xmath46 denote the space of contact vector fields on @xmath9 .
we can see that the linear map @xmath47 is an isomorphism .
the function @xmath48 is called the contact hamiltonian function of @xmath33 .
a closed oriented hypersurface @xmath49 is said to be _ convex _ if there exists a contact vector field @xmath50 on a neighbourhood @xmath51 of @xmath52 with @xmath53 ( @xmath54 ) , i.e. , @xmath50 is positively transverse to @xmath0 ( giroux@xcite ) . by perturbing the contact hamiltonian function if necessary
, we may assume that @xmath55 is a hypersurface transverse to @xmath0 .
then @xmath56 separates @xmath0 into the _ positive region _
@xmath57 and the _ negative region _
@xmath58 so that @xmath59 .
each interior @xmath60 has the positive exact symplectic form @xmath61 .
we can modify the function @xmath62 near @xmath56 to obtain a function @xmath63 such that @xmath64 are symplectic and @xmath65 is @xmath50-invariant .
( this is the `` transverse monotonicity '' of @xmath0 . )
note that the dividing set @xmath66 is then the convex ends of the exact symplectic manifolds @xmath60 .
a _ convex hypersurface with contact - type boundary _ is a connected hypersurface @xmath0 which admits a transverse contact vector field @xmath33 such that , for the associated decomposition @xmath59 , @xmath60 are also convex exact symplectic manifolds , and the contact - type boundary @xmath67 is non - empty .
( changing @xmath33 if necessary , we can assume moreover that the dividing set @xmath66 contains @xmath25 . )
then the thurston - bennequin inequality can be expressed as follows .
@xmath68 slightly extending the overtwisted disk @xmath27 , we obtain a convex disk @xmath0 which is the union @xmath69 of a disk @xmath70 and an annulus @xmath71 surrounding @xmath70 .
then @xmath0 violates the thurston - bennequin inequality ( @xmath72 ) , and is called a convex overtwisted disk .
a possible higher dimensional overtwisted convex hypersurface @xmath0 would also satisfy @xmath73 and @xmath74 .
particularly @xmath75 would have to be disconnected ( see @xcite ) .
we show the following proposition .
let @xmath76 be an overtwisted contact @xmath1-manifold and @xmath77 the @xmath78-ball in @xmath79 @xmath80 . then there exist a closed hypersurface @xmath81 and a hypersurface @xmath0 with contact - type boundary in the product contact @xmath8-manifold @xmath82 such that @xmath81 and @xmath0 are not convex , @xmath83 , and @xmath0 violates the thurston - bennequin inequality , where @xmath84 denotes the natural projection to @xmath44 .
let @xmath29 be the cylindrical coordinates of @xmath85 , and consider the functions @xmath86 then we see that the contact structure on @xmath85 defined by the contact form @xmath87 is overtwisted .
an overtwisted disk in @xmath76 has a neighbourhood which is contactomorphic to @xmath88 .
thus , by using the formula @xmath89 we can replace @xmath90 in proposition with @xmath91 .
then we take the hypersurface @xmath92 and its subset @xmath93 we orient @xmath81 so that the characteristic foliation @xmath94 is presented by the vector field @xmath95 indeed the following calculations shows that the vector field @xmath33 satisfies @xmath96 , @xmath97 , and @xmath98 .
@xmath99 @xmath100 @xmath101 @xmath102 @xmath103 we see that the vector field @xmath33 also satisfies @xmath104 with attention to @xmath105 .
thus @xmath26 is contact - type .
the singularity of @xmath106 is the union of @xmath107 and @xmath108 they are respectively a source point and a sink point ( see figure 1 ) . precisely , figure 1 depicts ( the fourfold covering of ) the well - defined push - forward @xmath110 of @xmath33 under the natural projection @xmath111 from @xmath81 to the quarter - sphere @xmath112 the vector field @xmath110 defines the singular foliation @xmath113 .
the singularity consists of the following five points ; two ( quarter-)elliptic points @xmath114 whose preimages under @xmath111 are the above singular points ; other two ( half-)elliptic points @xmath115 whose preimages are the periodic orbits @xmath116 of @xmath33 @xmath117 ; and a hyperbolic point @xmath118 , which is the self - intersection of the leaf corresponding to @xmath119 . slightly changing @xmath78 if necessary , we may assume that the preimage @xmath120 of the hyperbolic singular point of @xmath110 is a union of periodic orbits of @xmath33 .
now we assume that @xmath81 is ( approximately ) convex in order to prove proposition by contradiction .
the assumption implies @xmath121 where @xmath122 are the @xmath123 regions of the convex surface @xmath81 devided by the transverse intersection with the level set @xmath56 of the contact hamiltonian function described in 1.2
. then we can see that the intersection @xmath124 contain a spherical component .
however the eliashberg - floer - mcduff theorem implies that @xmath125(other components ) can not be realized as the boundary of a connected convex symplectic manifold ( see mcduff@xcite ) .
this contradiction proves proposition .
here we omit a similar proof of the non - convexity of the seifert hypersurface @xmath0 .
aaa d. bennequin : _ entrelacements et quations de pfaff _ , astrisque , * 107 - 108 * ( 1983 ) , 83161 .
y. eliashberg : _ filling by holomorphic discs and its applications _ , geometry of low - dimensional manifolds 2 , london math .
note ser . *
151*(1990 ) , 4572 .
y. eliashberg : _
contact 3-manifolds twenty years since j. martinet s work _
fourier , grenoble , * 42 * ( 1991 ) , 165192 .
y. eliashberg and w. thurston : _ confoliations _ ,
a.m.s . university lecture series , * 13 * ( 1998 ) . e.
giroux : _ convexit en topologie de contact _ , comm .
* 66 * ( 1991 ) , 637677 .
d. mcduff : _ symplectic manifolds with contact type boundaries _ , invent . math . * 103*(1991 ) , 651671 .
a. mori : _ reeb foliations on @xmath128 and contact @xmath129-manifolds violating the thurston - bennequin inequality _
, preprint ( 2009 ) , arxiv:0906.3237 .
a. mori : _ the reeb foliation arises as a family of legendrian submanifolds at the end of a deformation of the standard @xmath126 in @xmath128 _ , preprint ( 2011 ) . w. thurston : _ norm on the homology of @xmath1-manifolds _ , memoirs of the ams , * 339 * ( 1986 ) , 99130 . | we show that any open subset of a contact manifold of dimension greater than three contains a certain hypersurface @xmath0 which violates the thurston - bennequin inequality .
we also show that no convex hypersurface smoothly approximates @xmath0 .
these results contrasts with the @xmath1-dimensional case , where any surface in a small ball satisfies the inequality ( bennequin@xcite ) and is smoothly approximated by a convex one ( giroux@xcite ) . |
the cross section being the most important quantity for description and interpretation of physics underlying interactions of particles connects theory and experiment .
accurate knowledge on it is therefore very important for deeper understanding already well established theories like the standard model as well as for search of new physics .
the effective lagrangian of the leptoquark model proposed by buchmller , rckl and wyler in 1986 @xcite obeys the symmetries of the standard model gauge groups @xmath2 .
therefore one may expect that dynamics of some processes in these models can formally coincide so that the calculated cross sections turn out also to be the same up to constant factors associated with different couplings .
this letter shows that exclusive production of @xmath0 bosons in neutrino
photon scattering and production of vector leptoquarks in electron
gluon collisions represent such a situation in the leading order of perturbation theory provided the leptoquarks minimally couple to the gauge fields of the standard model .
the cross sections of both reactions are calculated analytically . a significant difference between the result of this letter and previous analysis of the leptoquark production
following @xcite let us also postulate the lagrangian with minimal couplings between vector leptoquarks and the neutral gauge bosons of the standard model : @xmath3,\label{lagrange}\end{aligned}\ ] ] where @xmath4 is the field strength tensor of vector leptoquarks , @xmath5 are the field strength tensors of the neutral gauge fields of the standard model , @xmath6 are the coupling constants for the interactions between these gauge fields and the leptoquarks .
the covariant derivative @xmath7 is defined by @xmath8 here @xmath9 , @xmath10 and @xmath11 are the photon , @xmath12 boson and gluon fields , respectively ; @xmath13 is the elementary electric charge , @xmath14 denotes the electric charge of a leptoquark , @xmath15 is the strong coupling constant , @xmath16 ( @xmath17 is the third component of the weak isospin , @xmath18 is the weinberg angle ) , @xmath19 are the gell - mann matrices . from ( [ lagrange ] )
it follows that the feynman rules for the @xmath20 and @xmath21 vertices are similar to the @xmath22 @xmath23 ones @xcite .
the couplings of vector leptoquarks to the gluon fields have also analogy with the self - interaction of the gluons .
for example , in fig .
[ fig1 ] the explicit form of the feynman rules for the @xmath23 and @xmath24 vertices are given .
one can see that they differ only by constant factors .
let us consider two exclusive reactions .
the first one is the @xmath0 production in neutrino
photon scattering allowed by the standard model @xcite : @xmath25 where @xmath26e@xmath27 .
the second one is production of vector leptoquarks ( @xmath28 ) in interactions of left / right polarized electrons with gluons appearing in the buchmller
rckl wyler leptoquark model @xcite : @xmath29 the reactions ( [ reac1 ] ) and ( [ reac2 ] ) are closely related to each other from a formal point of view . to illustrate this , it is convenient to represent the considered interactions in the form of the feynman diagrams so that the amplitudes contributing to ( [ reac1 ] ) and ( [ reac2 ] ) will look as shown in fig .
[ fig2 ] and fig .
[ fig3 ] , respectively @xcite .
the leptoquark lagrangian has such symmetry properties that the feynman rules for the vertices @xmath30 respectively coincide in their structure with those of the standard model for the vertices @xmath31 up to constant factors ( see fig .
[ fig44 ] ) . as to the familiar rules corresponding to the @xmath32 and @xmath33 vertices
, they also differ from each other only by the coupling constants and the color coefficients .
this means that in the limit of massless initial state leptons the leading order cross sections of the reactions ( [ reac1 ] ) and ( [ reac2 ] ) will have exactly the same dependence on the mandelstam variables and on the masses of the final state particles differing from each other only by a constant as well .
such relations between processes are well known in quantum field theory .
most simple qcd diagrams are exactly analogous to qed diagrams , and the qcd cross section is obtainable by appropriate replacements of the coupling constants in the qed one corresponding to the replacement of the photon by a gluon @xcite .
suppose that @xmath34 and @xmath35 are the leading order cross sections of the reactions ( [ reac1 ] ) and ( [ reac2 ] ) , where @xmath36 is the center - of - mass ( cm ) energy squared , @xmath37 and @xmath38 are the masses of the final state particles .
then , in accordance with the above discussion , they must satisfy the following condition : @xmath39 note that both cross sections in ( [ ratio1 ] ) are taken with the same values of the masses .
therefore , it is enough to know one of the cross sections to find the other .
the cross section of the reaction ( [ reac2 ] ) is calculated using the diagrams from fig .
[ fig3 ] with the corresponding feynman rules given in figs .
[ fig1](b ) and [ fig44](b ) .
the result reads @xmath40 where @xmath41 is the coupling constant corresponding to the @xmath30 vertex , @xmath42 , @xmath43\right.\nonumber\\\left.+2\frac{\,2 m_1 ^ 2 \left(s^2 - 2m_1 ^ 2s+2m_1 ^ 4\right)+m_2 ^ 2\left(s^2-\left(s-3m_1 ^ 2\right)\left(m_1
^ 2 - 2m_2 ^ 2\right)\right)-m_2 ^ 6}{s^3 } \right.\nonumber\\ \left.\times\log \left[\frac{\left(p^2+m_2 ^ 2\right)^{1/2}+p}{m_2}\right]\right.\nonumber\\\left.-\frac{8 m_1 ^ 4\left(s - m_1 ^ 2\right)-m_2 ^ 2 \left(s^2 + 4m_1 ^ 2s+6m_1 ^ 4\right)-m_2 ^ 4 \left(s-12m_1 ^ 2\right)+2 m_2 ^
6}{s^3}\right.\nonumber\\ \left.\times \log \left[\frac{\left(p^2+m_1 ^ 2\right)^{1/2}+p}{m_1}\right]\right\}.\nonumber\\\label{function}\end{aligned}\ ] ] here @xmath44 is the cm momentum of any of the final state particles .
the similarity between the cross sections of the reactions ( [ reac1 ] ) and ( [ reac2 ] ) discussed in section [ problem ] allows to verify the validity of the cross section ( [ result ] ) .
actually , the problem of calculation of @xmath34 is now reduced to just performing the following obvious replacements of the coupling constants and the masses of the final state particles in ( [ result ] ) : @xmath45 where @xmath46 is the coupling of the weak charged current ( related to the fermi coupling constant @xmath47 by @xmath48 , @xmath49 is the fine structure constant .
note that the coefficient of @xmath50 is the color factor equal to @xmath51 for the process ( [ reac2 ] ) .
so that one finds @xmath52 the cross section for the reaction ( [ reac1 ] ) in the leading order has also been independently calculated in @xcite , however in such a way that the masses of the final state leptons were neglected ( let us denote this cross section by @xmath53 ) .
this means that ( [ result2 ] ) obtained in this letter and the result of @xcite must asymptotically coincide satisfying the following condition : @xmath54 it should be emphasized that ( [ ratio2 ] ) is a criterion to verify the validity of the cross sections obtained in the present letter .
figure [ fig55 ] shows that ( [ ratio2 ] ) is fulfilled as it must be in the case of correct calculations . moreover , there is a perfect coincidence between ( [ result2 ] ) and @xmath53 over the entire range of the energy for the case of the electron in the final state owing to the vanishing electron mass in comparison with @xmath55 .
the cross section for the reaction ( [ reac2 ] ) in the leading order has also been calculated in @xcite whose result ( let us denote it by @xmath56 ) significantly differs from ( [ result ] ) .
this fact is illustrated in fig .
[ fig5 ] , where dependences of both cross sections on the cm energy are shown for production of the @xmath57 and @xmath58 type vector leptoquarks of mass 1000 gev at @xmath59 and @xmath60 .
the difference is more brightly reflected by the ratio of @xmath56 to @xmath35 shown in fig .
one can see that there may be cases in which the cross section ( [ result ] ) exceeds the corresponding quantity from @xcite by about a factor of two .
this is because different choices of the lagrangian responsible for interaction of the leptoquarks with the gluon fields which lead to different feynman rules for the @xmath24 vertex @xcite . in the present letter
the minimal couplings of the leptoquarks to the gauge fields of the standard model are assumed while in @xcite the lagrangian with anomalous interaction terms is used @xcite .
a standard convolution of the cross section ( [ result ] ) with the gluon distribution in the nucleon gives the cross section of inclusive vector leptoquark production measurable in electron
proton collisions : @xmath61 where @xmath62 is the gluon distribution function , @xmath63 . figure [ fig7 ] shows the cross sections for the production of the ( @xmath64 ) and ( @xmath65 ) type vector leptoquarks evaluated using ( [ convolution ] ) with the gluon distribution function adopted from cteq5 @xcite .
the cross section for vector leptoquark production in electron
gluon collisions is calculated analytically using the @xmath2-symmetric lagrangian with the minimal couplings between the leptoquarks and the gauge fields of the standard model .
it is shown that from a formal point of view this process is similar to exclusive production of @xmath0 bosons in neutrino
photon scattering allowed by the standard model whose cross section is also found .
the cross section of the leptoquark production obtained in this letter significantly exceeds the corresponding result of previous calculations presented in the literature .
the cross sections of inclusive production of the ( @xmath64 ) and ( @xmath65 ) vector leptoquarks in electron nucleon interactions observable at @xmath1 colliders are evaluated .
the presented analysis is applicable to studies of the production of vector leptoquarks of the other types as well .
i thank p.m. zerwas for useful comments on production of leptoquarks in several processes .
i am also thankful to m. spira for drawing my attention to their results on production of leptoquarks at @xmath1 colliders as well as for providing useful information .
this work was supported in part by the russian foundation for basic research ( grant 11 - 02 - 12043 ) , by the program for basic research of the presidium of the russian academy of sciences `` neutrino physics and neutrino astrophysics '' and by the federal target program of the ministry of education and science of russian federation `` research and development in top priority spheres of russian scientific and technological complex for 2007 - 2013 '' ( contract no .
16.518.11.7072 ) .
* fig . 6 : * dependence of the cross sections of production of vector leptoquarks of mass 1000 gev on the cm energy .
right panel : the @xmath57 type leptoquark . left panel : the @xmath58 type leptoquark .
the solid curves represent the results of this letter , the dashed curves are the calculations of @xcite .
the couplings @xmath59 and @xmath60 . * fig . 7 :
* dependence of the ratio of the cross section @xmath69 taken from @xcite to that calculated in the present letter on the cm energy for production of the @xmath58 type vector leptoquark of mass 500 gev ( dash - dotted ) , 700 gev ( dashed ) and 1000 gev ( solid ) .
the dotted line corresponds to @xmath70 . *
fig . 8 : * the cross section of the reaction @xmath71 as a function of the leptoquark mass at @xmath72 gev . dashed curve : the @xmath57 type leptoquark .
solid curve : the @xmath58 type leptoquark .
the coupling @xmath41 is divided out , @xmath60 . | the cross section for vector leptoquark production in electron
gluon collisions is calculated analytically using the lagrangian with the minimal couplings between the leptoquarks and the gauge fields of the standard model .
it is found that the cross section significantly exceeds the corresponding quantity previously presented in the literature . the cross section of exclusive @xmath0 boson production in neutrino
photon scattering emerges as a by - product of this letter .
the obtained results can be used for studies at @xmath1 colliders .
leptoquark , electron , gluon , neutrino , photon 14.80.sv , 13.60.-r |
we thank andrey chabanov for useful discussions on cavity design . j.k .
acknowledges support from the national science foundation ( grant no .
dmr-1310138 ) .
this work was performed , in part , at the center for integrated nanotechnologies , a u.s .
department of energy , office of basic energy sciences user facility .
sandia national laboratories is a multiprogram laboratory managed and 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 no .
de - ac04 - 94al85000 .
the work at sandia was supported by the u.s .
department of energy , office of science , materials sciences and engineering division .
growth and characterization completed at purdue by j.d.w . was supported by the department of energy , office of basic energy sciences , division of materials sciences and engineering under award no .
de - sc0006671 .
. acknowledges additional support from the w. m. keck foundation and microsoft research . | nonperturbative coupling of light with condensed matter in an optical cavity is expected to reveal a host of coherent many - body phenomena and states @xcite . in addition
, strong coherent light - matter interaction in a solid - state environment is of great interest to emerging quantum - based technologies @xcite .
however , creating a system that combines a long electronic coherence time , a large dipole moment , and a high cavity quality ( @xmath0 ) factor has been a challenging goal @xcite . here , we report collective ultrastrong light - matter coupling in an ultrahigh - mobility two - dimensional electron gas in a high-@xmath0 terahertz photonic - crystal cavity in a quantizing magnetic field , demonstrating a cooperativity of @xmath1360 .
the splitting of cyclotron resonance ( cr ) into the lower and upper polariton branches exhibited a @xmath2-dependence on the electron density ( @xmath3 ) , a hallmark of collective vacuum rabi splitting .
furthermore , a small but definite blue shift was observed for the polariton frequencies due to the normally negligible @xmath4 term in the light - matter interaction hamiltonian . finally , the high-@xmath0 cavity suppressed the superradiant decay of coherent cr , which resulted in an unprecedentedly narrow intrinsic cr linewidth of 5.6ghz at 2k .
these results open up a variety of new possibilities to combine the traditional disciplines of many - body condensed matter physics and cavity - based quantum optics .
= 1 strong resonant light - matter coupling in a cavity setting is an essential ingredient in fundamental cavity quantum electrodynamics ( qed ) studies @xcite as well as in cavity - qed - based quantum information processing @xcite . in particular , a variety of _ solid - state _ cavity qed systems
have recently been examined @xcite , not only for the purpose of developing scalable quantum technologies , but also for exploring novel many - body effects inherent to condensed matter .
for example , collective @xmath5-fold enhancement of light - matter coupling in an @xmath6-body system @xcite , combined with colossal dipole moments available in solids , compared to traditional atomic systems , is promising for entering uncharted regimes of ultrastrong light - matter coupling .
nonintuitive quantum phenomena can occur in such regimes , including a `` squeezed '' vacuum state @xcite , the dicke superradiant phase transition @xcite , the breakdown of the purcell effect @xcite , and quantum vacuum radiation @xcite induced by the dynamic casimir effect @xcite .
specifically , in a cavity qed system , there are three rates that jointly characterize different light - matter coupling regimes : @xmath7 , @xmath8 , and @xmath9 .
the parameter @xmath7 is the coupling constant , with 2@xmath7 being the vacuum rabi splitting between the two normal modes , the lower polariton ( lp ) and upper polariton ( up ) , of the coupled system .
the parameter @xmath8 is the photon decay rate of the cavity ; @xmath10 @xmath11 @xmath12 is the photon lifetime of the cavity , and the cavity @xmath0 @xmath11
@xmath13 at mode frequency @xmath14 .
the parameter @xmath9 is the nonresonant matter decay rate , which is usually the decoherence rate in the case of solids .
strong coupling is achieved when the splitting , @xmath15 , is much larger than the linewidth , ( @xmath16 , and ultrastrong coupling is achieved when @xmath7 becomes a considerable fraction of @xmath17 .
the two standard figures of merit to measure the coupling strength are @xmath18 @xmath19 @xmath20 and @xmath21 ; here , @xmath18 is called the _ cooperativity
_ parameter @xcite , which is also the determining factor for the onset of optical bistability through resonant absorption saturation @xcite .
in order to maximize @xmath18 and @xmath21 , one should construct a cavity qed setup that combines a large dipole moment ( i.e. , large @xmath7 ) , a small decoherence rate ( i.e. , small @xmath9 ) , a large cavity @xmath0 factor ( i.e. , small @xmath8 ) , and a small resonance frequency @xmath14 .
iii - v semiconductor quantum wells ( qws ) provide one of the cleanest and most tunable solid - state environments with quantum - designable optical properties .
microcavity qw - exciton - polaritons represent a landmark realization of a strongly coupled light - condensed - matter system that exhibits a rich variety of coherent many - body phenomena @xcite
. however , the large values of @xmath14 and relatively small dipole moments for interband transitions make it impractical to achieve large values of @xmath21 using exciton - polaritons .
intraband transitions , such as intersubband transitions ( isbts ) @xcite and cyclotron resonance ( cr ) @xcite , are much better candidates for accomplishing ultrastrong coupling because of their small @xmath14 , typically in the midinfared and terahertz ( thz ) range , and their enormous dipole moments ( 10s of @xmath22- ) .
experimentally , ultrastrong coupling has indeed been achieved in gaas qws using isbts @xcite and cr @xcite . in the latter case ,
a record high value of @xmath21 @xmath11 0.87 has been reported @xcite . in all these previous intraband studies of ultrastrong light - matter coupling
, however , due to ultrafast decoherence ( large @xmath9 ) and/or lossy cavities ( large @xmath8 ) , the value of @xmath18 remained small , i.e. , the standard strong - coupling criterion ( @xmath23 ) was not satisfied . here , we simultaneously achieved small @xmath9 and small @xmath8 in ultrahigh - mobility two - dimensional electron gases ( 2degs ) in gaas qws placed in a high-@xmath0 1d thz photonic - crystal cavity ( pcc ) in a perpendicular magnetic field .
we achieved ultrastrong coupling ( @xmath24 and @xmath25 ) between coherent cr and thz cavity photons , observing vacuum rabi splitting ( rabi oscillations ) in the frequency ( time ) domain .
furthermore , we observed a @xmath2-dependence of 2@xmath7 on the electron density ( @xmath3 ) , signifying the collective nature of light - matter coupling @xcite . a value of @xmath21 @xmath11 0.12 was obtained with just a single qw with a moderate @xmath3 .
finally , the previously identified superradiant decay of cr in high - mobility 2degs @xcite was significantly suppressed by the presence of the high-@xmath0 thz cavity . as a result , we observed ultranarrow polariton lines , yielding an intrinsic cr linewidth as small as 5.6ghz ( or a cr decay time of 57ps ) at 2k .
high - mobility gaas 2deg samples were studied using thz time - domain magnetospectroscopy .
the magnetic field quantized the density of states of the 2deg into landau levels .
as schematically shown in fig.1**a * * , thz cavity photons are coupled with the transition between adjacent landau levels , i.e. , cr .
figure 1**b * * shows our 1d thz pcc design , consisting of two layers of 50-@xmath26m - thick undoped si wafers on each side as a bragg mirror . thanks to the large contrast of refractive index between si ( 3.42 in the thz range ) and vacuum , only a few layers of si were required to achieve sufficient cavity confinement of thz radiation with high @xmath0 values .
a substrate - removed 4.5-@xmath26m - thick gaas 2deg sample was placed on the central `` defect '' layer of the pcc , which was a 100-@xmath26m - thick si ( sapphire ) wafer in cavity 1 ( cavity 2 ) .
calculated electric field distribution inside cavity 1 is shown in fig.1 , * c * , * d * , and * e * , for the first , second , and third cavity modes , respectively .
the spatial overlap of the 2deg and the electric field maximum ensured the strongest light - matter coupling .
figure 1**f * * shows a thz transmission spectrum for cavity 1 , containing a 2deg , at 4k .
three photonic band gaps are seen as transmission stop - bands . at the center of each stop - band , a sharp cavity mode is observed .
as shown in fig.1 , * g * , * h * , and * i * , the full - width - at - half - maximum ( fwhm ) values , or @xmath8 , of these cavity modes were 2.7ghz , 5.0ghz , and 3.8ghz , corresponding to @xmath0 factors of 150 , 243 , and 532 , respectively ; note that these numbers are slightly lower than those for an empty cavity without including the 2deg , which were 183 , 450 , and 810 .
these @xmath0-factors are one to two orders of magnitude higher than those reported for the thz metamaterial resonators employed in previous untrastrong - coupling studies using 2deg cr @xcite . in the following , experimental data recorded with cavity 1 are shown . by varying the magnetic field ( @xmath27 )
, we continuously changed the detuning between the cyclotron frequency ( @xmath28 @xmath11 @xmath29 , where @xmath30 @xmath11 0.07@xmath31 is the electron effective mass of gaas and @xmath31 @xmath11 9.11 @xmath32 10@xmath33 kg ) and the cavity mode frequency ( @xmath14 ) : @xmath34 .
clear anticrossing behavior , expected for strong coupling , is shown in fig.2**a * * for the first cavity mode in cavity 1 .
two polariton branches ( lp and up ) were formed through the hybridization of cr and the thz cavity photons .
the central peak originates from the transmission of the cr - inactive circular - polarization component of the linearly polarized incident thz beam , which does not interact with the 2deg and whose position is independent of @xmath27 .
the fwhm of the central peak is thus given by @xmath8 , while that for the lp and up peaks at @xmath35 @xmath11 0 is given by @xmath36 .
therefore , from the @xmath35 @xmath11 0 spectrum ( @xmath37 @xmath11 1.00 t for this mode ) , we determined @xmath38 @xmath11 ( 74 , 2.6 , 5.6 , 407)ghz , yielding @xmath18 @xmath11 360 and @xmath21 @xmath11 0.09 .
parameter values determined in this manner for all modes in both cavities 1 and 2 are summarized in table [ parameters ] , together with cavity parameters and resonance conditions . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] as in other cavity qed systems based on atoms and microcavity excitons , vacuum rabi splitting in the frequency domain can be directly observed as time - domain oscillations @xcite .
experimentally , for an incident thz beam linearly polarized in the @xmath39 direction , we measured the @xmath40-polarization component , @xmath41 , of the transmitted thz wave , in both positive ( @xmath42 ) and negative ( @xmath43 ) fields , and took the difference @xmath44 @xmath11 @xmath45 , to eliminate any background noise .
the cr inactive mode was numerically filtered out .
as shown in fig.2**b * * , the measured @xmath35@xmath41 signal showed strong beating between the two polariton modes , which can be viewed as coherent repetitive energy exchange between the matter resonance and the cavity photons . at each beating node ( indicated by an arrow ) , energy is stored in the 2deg cr .
the average time separation between two adjacent beating nodes was about 13 - 15ps , matching the @xmath15 splitting in the frequency domain ; see also fig.2**c * * for the fourier transform of the time - domain oscillations .
the beating lasts for dozens of picoseconds , indicating a long intrinsic cr coherence time . in analogy to the physics of many - atom light - matter interactions @xcite ,
one crucial question is whether the rabi splitting observed here is a fully coherent behavior of a large number of individual electrons in the 2deg .
figure 2 , * d * and * e * , show three spectra showing polariton manifestation at @xmath35 @xmath11 0 for different electron densities ( @xmath3 ) , when cr is in resonance with the first cavity mode in cavity 1 .
the vacuum rabi splitting ( @xmath15 ) between the lp and up peaks exhibited a square - root dependence on @xmath3 ( fig.2**e * * ) , which is strong evidence for _ collective _ vacuum rabi splitting , as observed in atomic gases @xcite and spin ensembles @xcite .
this observation validates the notion that billions of 2d electrons are interacting with a common cavity thz photon field in a fully coherent manner . by extrapolation ,
the vacuum rabi splitting for cr of a single electron is estimated to be 0.14mhz .
it is also worth noting that we should be able to increase the vacuum rabi splitting further by using multiple layers of a 2deg with a higher electron density .
the coupled system of landau - quantized 2d electrons and thz cavity photons can be described by the following hamiltonian @xcite : @xmath46 @xmath11 @xmath47 , where @xmath48 @xmath11 @xmath49 , @xmath50 @xmath11 @xmath51 , @xmath52 @xmath11 @xmath53 , and @xmath54 @xmath11 @xmath55 .
the first two terms , @xmath50 and @xmath48 , represent , respectively , the energy of the cavity mode at @xmath14 and the energy of the 2deg in a @xmath27 with frequency @xmath28 .
the operators @xmath56 and @xmath57 ( @xmath58 and @xmath59 ) are the annihilation and creation operators for cavity photons ( collective cr excitations ) , respectively . the light - matter interaction term , @xmath52 , with coupling strength @xmath7 includes counter - rotating terms , @xmath60 , which are usually neglected under the rotating - wave approximation
. also included in the total hamiltonian is the diamagnetic term , @xmath54 , also known as the @xmath4 term because , mathematically , it is proportional to the square of the vector potential @xmath61 of the light field .
the pre - factor @xmath62 of the @xmath4 term suggests that this term is negligible in the weak - coupling regime but can have measurable effects in the ultrastrong - coupling regime where @xmath7 is on the order of @xmath28 .
figure 3**a * * presents the best fits to our data with three different hamiltonians : i ) @xmath47 ( full hamiltonian ) , ii ) @xmath63 , and iii ) @xmath64 ( the jaynes - cummings hamiltonian , @xmath65 ) . at the optimum
fitting , the value of @xmath66 was determined to be 0.09 ( for cavity 1 ) .
the full cr - cavity hamiltonian , i ) , provided the best fit , while the other two failed to show the non - negligible blue shift of the polariton modes . with @xmath66 @xmath11 0.1 , the contribution of the @xmath4 term
is expected to be on the order of 0.01@xmath14 .
this 1% contribution from the @xmath4 term indeed explains the observed blue shift , which is another manifestation of the fact that our system is in the ultrastrong - coupling regime .
furthermore , even the cavity mode is affected by ultrastrong coupling : as shown in fig.3**b * * , the residual cavity mode exhibits a red shift with increasing magnetic field .
this unexpected behavior occurs because the ultrastrong light - matter interaction even enables coupling between the cavity mode and the cr inactive mode at a negative frequency .
the cavity mode is actually part of the up branch for the cr inactive mode , as shown in fig.3**c**. finally , we studied the extracted value of @xmath9 as a function of temperature .
it has previously been shown that the decay of cr in free space is dominated by collective radiative decay , or superradiance , in ultrahigh - mobility 2degs , showing a decay rate that is proportional to @xmath3 @xcite .
this radiative decay mechanism is very strong and dominant at low temperatures , faster than any other phase breaking scattering processes ; thus , cr lines are much broader than expected from the sample mobility . in the present case ,
however , the emitted coherent cr radiation can not readily escape from the high-@xmath0 cavity and thus re - excites coherent cr _
multiple times_. hence , this reversible emission and absorption in a strongly coupled cavity-2deg system strongly suppresses the superradiant decay , revealing the intrinsic cr decoherence rate , @xmath67 ( s@xmath68 ) @xmath11 @xmath69 .
this dramatic suppression of radiative decay is opposite to the purcell effect in a high-@xmath0 cavity expected in the weak - coupling regime and can only be understood within the framework of strong coupling @xcite , where superradiant decay is suppressed by the reversible absorption and emission processes .
figure 4**a * * presents temperature - dependent transmission spectra at @xmath70 for the first cavity mode in cavity 1 .
the lp and up peaks significantly broadens above 20k , becoming unobservable above 80k .
the central peak , on the other hand , remains essentially unchanged as the temperature increases , serving as an excellent linewidth reference . at each temperature , we determined both @xmath8 and @xmath9 , using the procedure described earlier ( fig.2**a * * ) .
as shown in fig.4**b * * , the cr decay time in the cavity is 57 @xmath71ps at 2k , much longer than the superradiance - limited value in free space ( 10ps ) .
therefore , the cr decay time measured in our high-@xmath0 cavity is the intrinsic cr decay time , @xmath72 , due to nonradiative decay mechanisms ( i.e. , scattering ) . above 20k ,
@xmath72 approaches the dc momentum scattering time , @xmath73 , where piezoelectric scattering and polar optical phonon scattering dominate . at 2k
, @xmath72 is still lower than @xmath73 .
how the cr linewidth in a high - mobility 2deg changes with the magnetic field and temperature is a long - standing question @xcite , and a systematic study of ` superradiance - free ' cr widths should provide significant new insight .
we have demonstrated collective and ultrastrong light - matter coupling between the cr of a 2deg and thz cavity photons in a high-@xmath0 1d photonic - crystal cavity , with a cooperativity up to 360 . unlike the near - field coupling of metamaterial resonators , our thz cavity scheme is applicable to both 2d and bulk materials , which will allow us to study various strongly correlated systems with collective many - body excitations in the thz range , e.g. , magnetically ordered systems , high - temperature superconductors , and heavy - fermion systems .
hence , our demonstration of ultrastrong light - matter interaction in high-@xmath0 thz cavities opens a door to a plethora of new possibilities to combine the traditional disciplines of many - body condensed matter physics and quantum optics of cavity qed . |
quantum gravity seeks for the synthesis of general relativity and quantum mechanics @xcite .
the synthesis of special relativity and quantum mechanics still also remains to some extent an open problem @xcite it is anticipated that in general the synthesis of relativity and quantum mechanics can also shed new light on the structure and properties of the subatomic world , an area currently treated by quantum chromodynamics ( qcd ) @xcite . on the other hand the bohr and bohr - sommerfeld rotating electron models of the h atom present striking examples where classical deterministic treatment of a physical problem coupled with the de broglie wavelength expression leads to results in almost quantitative agreement with experiment and with the results of the full quantum mechanical schrdinger treatment . in these very successful classical circular or elliptical orbit models
the centripetal force is the coulombic attraction between the proton and the electron which are both treated as point charges .
it thus becomes interesting to examine the extent to which this deterministic type approach may be useful to provide a satisfactory zeroth order model for other physical problems at the subatomic level such as the condensation of the quark - gluon plasma to form hadrons @xcite .
in such an approach the parton or quark - gluon constituents of hadrons may be viewed as particles with relativistic velocities @xcite .
such an approach appears worthwhile to consider since , as analyzed in several recent papers by t hooft @xcite , attempts to produce physically realistic models for the unification of general or special relativity with quantum mechanics may not be successful unless some level of determinism in the latter is restored @xcite . as shown by t hooft @xcite such models which introduce determinism beneath quantum mechanics and thus restore causality and locality can be constructed so that stochastic behaviour is correctly described by quantum mechanical amplitudes in precise accordance with the copenhagen - bohr - born doctrine @xcite .
a difficulty , however , with such models , which may be essential to restore causality and locality at planckian distance scales @xcite , is to obtain a hamiltonian which is bounded from below and thus leads to a ground state @xcite .
these considerations have provided an incentive for the present work where we present a deterministic model which combines special relativity , newton s gravitational law and the weak equivalence principle to examine the rotational motion of three light neutral or charged particles under the influence of their gravitational attraction with velocities very close to the speed of light @xmath0 .
this leads to an infinity of bound rotational states and , similarly to the bohr model of the h atom , angular momentum quantization is used to select among these states the ones which conform to the de broglie wavelength expression which historically has been the origin of quantum mechanics .
interestingly it is found that when the masses of the light particles are in the mass range of neutrinos @xmath4
@xmath5 @xcite then the masses of the confined composite bound states formed are in the hadrons mass range @xmath6 . both the computed radii , which are in the @xmath1 range , and the estimated magnetic moments are in good agreement with experiment .
the model also provides a good approximation of the mass ratios between the proton and other baryons and also predicts a lower limit for baryon lifetimes @xmath7 which is in agreement with experiment .
this deterministic bohr - type hadronization model involves no unknown parameters and , although the observed very good agreement with experiment could be fortuitous , it appears to be a useful zeroth order model for hadronization .
it is at first quite surprising that this semiclassical deterministic model provides a good fit to experiment by considering gravity between the relativistic particles as the only attractive force .
however , the possibility that gravity may have a significant role at short , femtometer or subfemtometer distances has attracted significant interest for years @xcite and the potential role of special relativity @xcite as well as the feasibility of developing a unified gauge theory of gravitational and strong forces @xcite have been discussed .
of special interest in this area was wheeler s concept of geons ( gravitational - electromagnetic entities ) , i.e. of spherical or toroidal objects made up of electromagnetic waves or neutrinos held together gravitationally @xcite .
geons are classical entities based on general relativity .
small electromagnetic or neutrino geons had been proposed as classical models for elementary particles @xcite . more commonly the formation of baryons and other hadrons via condensation of smaller particles , i.e. of the gluon - quark plasma is analyzed in the qcd theory @xcite .
this condensation occurs at the transition temperature of qcd which is estimated to be around 150 mev in the kt scale , i.e. it corresponds to a particle energy of about 150 mev @xcite .
the gravitational interaction is usually extremely weak and insignificant when discussing interaction between particles such as those forming baryons . in the model discussed here a baryon
is described in terms of the rotational motion of three symmetrically spaced point particles with rest mass @xmath8 being significantly lower than @xmath9 , the rest mass of the baryon .
these three light point particles form a bound rotational state and are moving at a speed close to the speed of light , @xmath0 .
both their kinetic energy and their inertial mass are significantly enhanced by this high relativitic speed .
the rest mass @xmath9 of the baryon is determined by the total rest plus kinetic energy of the three particles , i.e. @xmath10 where @xmath11 is the lorentz factor . when v approaches @xmath0 the lorentz factor @xmath12 and the inertial mass of the particle , @xmath13 @xcite increase very significantly .
thus the first incentive for the present work has been the observation that according to the equivalence principle between inertial and gravitational mass @xcite , such a high value of the inertial mass implies an equally high value of the gravitational mass and thus , according to newton s gravitational law , a much stronger gravitational attraction than that corresponding to non - relativistic velocities .
such a strong attraction can , as shown by the present model , lead to stable highly relativistic rotational states . a second incentive has been the observation depicted in fig . 1 and table a1 in appendix a that the masses of the known uncharmed spin ( 1/2 ) and spin ( 3/2 ) baryons @xcite , i.e. of baryons consisting of u , d and s quarks , follow with good approximation , better than 3% , a @xmath14 law where @xmath15 corresponds to the proton or the neutron mass .
this @xmath14 law is reminiscent of the @xmath16 law obeyed by the energy levels of the h atom . in the model discussed here
the three light particles are assumed to have a spin 1/2 .
they are also allowed to be charged which allows for charged baryons , such as the proton .
however , the coulomb interaction between them is shown to contribute less than 0.1 percent to the composite ( baryon ) mass
. the coulomb interaction seems only to account for small differences between close masses of baryons e.g. between that of the proton and the neutron .
, of the uncharmed baryons , consisting of u , d and s quarks , @xcite with equation , i.e. @xmath17 where @xmath18 is the proton mass.,width=285,height=248 ] we consider a laboratory frame @xmath19 and an instantaneous inertial frame @xmath20 moving with a particle with an instantaneous velocity @xmath21 relative to frame @xmath19 .
it is worth noting that the instantaneous inertial frame @xmath20 is defined by the vector @xmath21 alone and not by the overall type of motion ( e.g. linear or cyclic ) performed by the particle @xcite .
thus one can assign to the frame @xmath20 and corresponding velocity @xmath21 an inertial particle mass , @xmath22 , by considering a test force , @xmath23 , parallel to @xmath21 , acting on the particle . according to the theory of special relativity the case where @xmath23 and @xmath21 are
parallel is the only case where the force is invariant , i.e. the force perceived in @xmath19 and @xmath20 is the same @xcite .
starting from the general relativistic equation of motion , i.e. from @xcite : @xmath24 and using the fact that the test force @xmath23 is taken to be parallel to @xmath21 one obtains after some simple algebra that the measures of the force , @xmath25 , and of the acceleration , dv / dt , are related via : @xmath26 this defines the mass @xmath13 , termed sometimes longitudinal mass @xcite , which is the inertial mass of the particle , @xmath22 , as it equals the ratio of force and acceleration .
thus it is @xmath27 . according to the equivalence principle
, @xmath22 also equals the gravitational mass , @xmath28 , of the particle @xcite , i.e. : @xmath29 it is worth noting that the equivalence principle has been formulated and tested for parallel @xmath23 and @xmath21 vectors , as is the present case .
it is also worth reminding that , as already noted , for given @xmath8 and v the inertial mass @xmath22 and thus the gravitational mass @xmath28 are both uniquely determined by and their value does not depend on the type of motion ( e.g. linear or circular ) performed by the particle .
thus upon considering a second particle of rest mass @xmath8 and instantaneous velocity measure v relative to the observer at @xmath19 and at a distance @xmath30 from the first particle , it follows that the inertial and gravitational mass of the second particle is also given by @xmath13 , as in equation and thus one can use these two @xmath28 values in newton s gravitational law in order to compute the gravitational force , @xmath31 , between the two particles .
one thus obtains : @xmath32 which depends on the @xmath33 power of @xmath12 @xcite .
it is worth remembering that this equation stems directly from special relativity ( eq .
[ eq2n ] ) , the weak equivalence principle ( eq . [ eq3n ] ) and
newton s gravitational law . in our semi - classical model the three interacting particles composing the baryon ,
are assumed to rotate at the same speed around their center of mass , in a circular orbit of radius @xmath34 , as shown in fig .
2 . for the laboratory observer
the tangential velocity v is close to the speed of light , @xmath0 and the center of mass of the three particles is at rest .
the magnitude of the resulting force acting on each particle is : @xmath35 , in a circle of radius @xmath34 around their center of mass .
they are equally spaced .
@xmath36 and @xmath37 are two particle attraction forces and @xmath38 is the resultant , radial , force.,width=263,height=332 ]
one may now consider the relativistic equation of motion ( eq . ) for the case of a circular orbit of radius @xmath34 with @xmath39 . since @xmath40 one obtains : @xmath41 and thus upon combining with eq .
one obtains : @xmath42 i.e. , the gravitational force @xmath31 given by acts as the centripetal force for the rotational motion .
it must be noted on the basis of one might be tempted to assign the value @xmath43 , commonly termed transverse mass @xcite , to the inertial and thus gravitational mass of each particle .
however , as already noted the mass @xmath44 , and thus also @xmath28 is uniquely determined for given @xmath8 and @xmath21 , via the colinear to @xmath21 test force @xmath23 , and does not depend on the type of motion ( e.g. linear or circular ) performed by the particle
. upon utilizing @xmath45 in one obtains : @xmath46 or , equivalently : @xmath47 where @xmath48 is the schwarzschild radius of a particle with rest mass @xmath8 . as shown in figure 3 the function r defined by eq . exhibits a minimum , @xmath49 , at @xmath50 thus v@xmath51
this is the minimum radius for a circular orbit and the corresponding minimum angular momentum is @xmath52 @xmath53 .
this condition is similar to the criteria @xmath54 or @xmath55 found for circular orbits in special relativity @xcite or @xmath56 for the schwarzschild metric in general relativity @xcite with orbits around point masses with @xmath57 potentials .
equation defines two @xmath58 branches ( fig .
3 ) , one corresponding to low @xmath12 values @xmath59 the other corresponding to large @xmath12 values @xmath60 . the first branch corresponds to common gravitational orbits . in this case @xmath12 and thus the velocity @xmath61 decreases with increasing r , e.g. @xmath62 in the non - relativistic case @xmath63 .
the second branch which leads to relativistic velocities defines rotational states where @xmath12 and thus v increase with increasing r. these states with @xmath64 are the states of interest to the present model . for @xmath64 ,
e.g. @xmath65 eq .
reduces to : @xmath66 this equation can be used to eliminate r or @xmath58 in the force expression of eq . .
in the former case ( i.e. elimination of r ) one obtains : @xmath67 and thus , interestingly , for any fixed value of @xmath12 or @xmath34 , the attractive force is uniquely determined by the familiar @xmath68 parameter of the gravitational field equations of general relativity @xcite , i.e. @xmath69 , which relates the einstein tensor @xmath70 with the stress - momentum - energy tensor @xmath71 @xcite . in the latter case
, i.e. elimination of @xmath58 , one obtains : @xmath72 the force equation refers to circular orbits only and thus defines a certain conservative force , since the work done in moving the particles between two points @xmath73 and @xmath74 , corresponding to two rotational states with radii @xmath73 and @xmath74 , is independent of the path taken .
the force vector is always pointing to the center of rotation and thus a conservative vector field is defined which is the gradient of a scalar potential , denoted @xmath75 .
the latter can be termed gravitational potential energy of the three rotating particles and corresponds to the transfer of the particles from the minimum circular orbit radius @xmath76 to an orbit of radius of interest , @xmath34 . the function @xmath75 is obtained via integration of eq . ,
i.e. , denoting by @xmath77 the dummy variable : @xmath78 thus while the magnitude of the gravitational force acting on the rotating particles increases with decreasing radius , r ( eq . ) , the absolute value @xmath79 of the gravitational potential energy increases with increasing @xmath34 .
this behavior is reminiscent of asymptotic freedom @xcite , i.e. the attractive interaction energy is small at short distances and increases significantly with increasing distance @xmath34 .
since it is @xmath80 , the value of @xmath81 can be estimated from newton s gravitational law and is about @xmath82 @xmath83 , which , as shown in section 3.5 , turns out to be of the order of 0.1 ev .
this is quite negligible in comparison with the rest energies of baryons @xmath84 of interest in this work .
thus , in this scale , @xmath85 is quite close to the zero gravitational potential energy of the three free particles with zero velocity at infinite distance from each other .
therefore when this vacuum state is chosen as the reference gravitational potential energy state @xmath86 then for @xmath87 eq . is written with very good accuracy as : @xmath88 or using eq .
: @xmath89 on the other hand
the kinetic energy , @xmath90 , of the three rotating particles is : @xmath91 thus one may now compute the change in total system energy , @xmath92 , upon formation of the rotational bound state from the three originally free particles . denoting by f and i the final and initial states and by ( re ) the rest energy , one obtains from energy conservation : @xmath93-\left[(re)_{i}+t_{i}+v_{g , i } \right]=\\ \nonumber & = & \left[3m_{o}c^{2}+3(\gamma -1)m_{o}c^{2}-5\gamma m_{o}c^{2}\right]-3m_{o }
c^{2}=\\ \nonumber & = & \delta t+\delta v_{g } = -(2\gamma + 3)m_{o } c^{2 } \approx -2\gamma m_{o } c^{2}\end{aligned}\ ] ] where the last equality holds for @xmath64 as is the case of interest here .
the same @xmath92 expression is , of course , obtained regardless of the choice of the reference potential energy state .
thus in view of eqs . , , and one can summarize the dependence of @xmath94 , @xmath95 and @xmath92 on @xmath12 and @xmath34 for @xmath96 as : @xmath97 @xmath98 @xmath99 the negative sign of @xmath92 denotes that the formation of the bound rotational state starting from the three initially free particles happens spontaneously , is exoergic @xmath100 , and the binding energy @xmath101 equals @xmath102 .
the total rest plus kinetic energy of the three rotating particles equals @xmath103 and constitutes at the same time the rest energy of the composite particle formed , i.e. of the rotational bound state . denoting by @xmath104 the rest mass of the composite bound state ,
it is : @xmath105 it is useful to note that in the model the rest mass of the three particles , i.e. @xmath106 , does not change when the bound state is formed .
the transformation of the kinetic energy of the three rotating particles into rest energy of the bound state , is simply due to the change in choice of the boundaries of the system . in the former case ( three individual rotating particles )
the boundaries are geometrically disconnected and encompass each particle individually , in the latter case the system boundary contains all three particles .
thus the formation of the bound rotating state by the three particles provides a simple hadronization mechanism , i.e. generation of rest mass , @xmath104 , starting from an initial rest mass @xmath107 , according to eq .
( [ eq18 ] ) .
it follows from and that : @xmath108 thus the binding energy per light particle is @xmath109 , which for @xmath110 , the proton mass , gives an energy of 208 @xmath111 , not far from the estimated particle energy of 150 @xmath111 at the transition temperature of qcd @xcite .
one may note here that since the potential energy expression does not depend on the number , @xmath112 , of rotating particles but the kinetic energy , t , does , i.e. @xmath113 , it follows from that , according to the model , stable rotational states can not be obtained for @xmath114 since they lead to positive @xmath92 .
the case @xmath115 is interesting , as it may be used to model mesons , and gives results in qualitative agreement with those obtained for the @xmath116 case treated here . for @xmath115 one
finds that the computed composite masses , @xmath104 , are in the @xmath117 range .
we then proceed to identify among the infinity of bound rotational states described by eqs . , and , each corresponding to a different @xmath34 , those rotational states where @xmath34 is an integer multiple of the reduced de broglie wavelength @xmath118 of the light rotating particles .
this can be done by introducing quantization of the angular momentum of the light particles in the form : @xmath119 where the subscript @xmath120 denotes a baryon and indicates that the corresponding @xmath12 , @xmath34 and @xmath121 values satisfy the quantization condition .
it is assumed that @xmath15 corresponds to a proton and thus eq .
yields , for @xmath15 , @xmath122 which equals the reduced de broglie wavelength of the light particle of rest mass @xmath8 moving at a speed @xmath123 and is three times the reduced compton length of the proton , @xmath124 .
the mass @xmath125 is calculated by combining eqs . , and : @xmath126 one notes that @xmath127 is the planck mass , @xmath128 @xcite .
one can then reasonably assume that the ground state @xmath129 corresponds to a proton ( and neutron ) .
this yields quantitative expressions for the proton mass and for @xmath130 : @xmath131 where @xmath18 is the proton mass @xmath132 .
the mass of the heavier baryons follow the series : @xmath133 which is in good agreement with the experimental data as shown in figure 1 and table a1 . from eq .
one obtains the following expression for @xmath136 : @xmath137 substituting the values of the constants one obtains : @xmath138 which , surprisingly , is in the mass range of neutrinos @xcite .
the heaviest neutrinos mass is currently estimated to be between 0.04 and 0.4 ev / c@xmath139 @xcite .
the value of @xmath140 for the proton , @xmath141 , is obtained from eqs .
, i.e. : @xmath142 while for the heavier baryons it increases with n , @xmath143 the rotational radius r for the baryons increases with n , as given by eqs . and : @xmath144 where @xmath145 corresponds to the proton .
( vertical line ) and of eqs .
, , , showing how the bohr quantization condition ( eq . ) determines the rest energy , @xmath146 of the bound state and the corresponding binding energy @xmath147.,width=266,height=359 ] using the definitions of @xmath148 , of @xmath149 and eq .
, i.e. @xmath150 one can rewrite eqs . , and in terms of @xmath151 rather than @xmath152 : @xmath153 @xmath154 @xmath155 plots of these functions are given in fig .
the period of rotation @xmath156 of the light particles within the composite ones , @xmath157 , is , using eq .
, @xmath158 where @xmath159 is the rotation period for the proton .
the time interval @xmath156 provides a rough lower limit for the lifetime of the composite particles , interpreted as baryons , as they can be defined only if the light particles complete at least a revolution ( fig .
2 ) . indeed
all the known lifetimes of the baryons are not much shorter than that estimate .
the lifetime of the @xmath160 baryons , which is the shortest , is @xmath161 @xcite .
so far the model discussed only the mass of the light particles and its effect on the binding forming a neutral composite particle . however
, baryons also possess a spin and many of them also charge .
we therefore allocate to the light particles , in analogy to the quark model @xcite , a charge and spin .
it is assumed that there are two types of light particles , those with charge @xmath162 and those with @xmath163 , where e is the elementary charge .
in addition each light particle is assumed to possess a spin 1/2 . as a result all composite particles have either a total spin 1/2 or 3/2 as is the case for most baryons .
the coulombic forces between charged particles with relativistic velocities have been studied in detail @xcite .
it is well established that coulomb s law correctly gives the force on the test charge for any velocity of the test charge provided the source charge is at rest @xcite . in the simplified geometry of figure 2
the distance between the two particles remains constant , thus in the reference frame of the source charge the test charge is also at rest , thus coulomb s law remains valid without any relativistic corrections .
it is thus possible to estimate the coulomb interaction energy between the rotating particles . in the simplified geometry of figure 2
the total coulomb potential energy for the proton ( charges 2/3 , 2/3 , -1/3 ) vanishes ( since ( 4/9)-(2/9)-(2/9)=0 ) , while for the neutron ( charges -1/3 , -1/3 , 2/3 ) it is negative , i.e. denoting @xmath164 one obtains : @xmath165=-\frac{(e^{2 } /\varepsilon ) } { 3\sqrt{3 } r_{b}}\ ] ] , i.e. there is an overall attractive coulombic interaction . upon substituting @xmath166 from eq . for the case of the proton
@xmath129 one obtains : @xmath167 where @xmath168is the fine structure constant .
thus @xmath169 is about three orders of magnitude smaller than the rest energy of the proton .
this justifies the approximation made by neglecting the coulomb interaction .
interestingly as shown in table a2 ( appendix a ) the mass differences in baryons which differ only in the charge value ( e.g. the n , @xmath170 or @xmath171 baryons ) are generally small ( up to 7 @xmath172 ) and the ratio @xmath173 , @xmath174 or @xmath175 is of the order of @xmath176 , similarly to the value of the ratio @xmath177 obtained from eq . .
thus the coulomb interaction could be the origin of this small difference . if the coulomb interaction is taken into consideration the symmetry of the configuration of fig .
2 is broken as not all three charges are the same .
although the deviation from three - fold symmetry is small , since the coulombic energy is small , and thus one may still use with good accuracy eq . to estimate the attractive interaction between the three particles forming a neutron ,
it is conceivable that this broken symmetry may be related to the relative instability of the neutron ( lifetime 885.7 @xmath178 ) vs the proton ( estimated lifetime @xmath179
it is interesting to compute the magnetic dipole moments , @xmath180 , of these bound rotational states . using the definition of @xmath181 and considering the case @xmath15 ,
corresponding to a proton ( which is a uud baryon ) with charge @xmath162 for u and @xmath163 for d it is : @xmath182=(1/2)erc\ ] ] upon substituting @xmath183 one obtains : @xmath184 where @xmath185 is the nuclear magneton @xmath186 .
this value differs less than 8% from the experimental value of @xmath187 @xcite . in the above computation ( eq . )
one assumes that the spin vectors of the three small particles ( i.e. uud ) are parallel to the vector of rotation of the rotating proton state .
if one considers the neutron which is a udd particle and assumes that the spin of one of the two d quarks is parallel with the rotation vector of the rotating neutron state and the spins of the other two particles are antiparallel to the neutron rotation vector then one obtains : @xmath188=-(1/3)erc\ ] ] and upon substitution from eq . of @xmath189 one
obtains : @xmath190 which is in excellent agreement with the experimental value of @xmath191 .
this good agreement seems to imply that the spin contribution of the light particles to the magnetic moment of the rotating state is small and only the spin vector orientation ( parallel or antiparallel to the baryon rotation vector ) is important .
a simple semi - classical model is presented which explores the use of gravitation as the origin of the strong force by examining the rotational motion of three light particles due to their gravitational attaction under relativistic conditions .
the model uses special relativity and the equivalence principle to express the gravitational force by substituting the inertial mass @xmath192 in newton s gravitational law .
interestingly for any fixed @xmath12 value the gravitational force is found to be uniquely determined by the key parameter , @xmath68 , of the field equations of general relativity .
the model is similar to the original bohr model of the h atom . in the present case
the angular momentum is assumed to be quantized with the condition @xmath193 rather than with @xmath194 as in the bohr model for the h atom .
the model provides surprisingly interesting results concerning the masses that follow a @xmath14 law , the binding energies per light particle , and also the radii and the magnetic moments .
it also provides a reasonable lower limit for the baryon lifetimes .
the mass @xmath195 and spin ( 1/2 ) of the light particles , agrees with that of the neutrino @xcite .
interestingly this is consistent with wheeler s concept of neutrino geons @xcite .
if this is the case then some of these hadron forming neutrinos have to be charged ( at least in charged baryons ) .
one may explain their absence in experiment by a high cross section for absorption , e.g. to form neutral neutrinos or hadrons . as in the standard model ,
the charge allocated to the light particles is @xmath196 and @xmath163 .
the coulomb interaction is typically a factor of @xmath197 smaller than the gravitational one and thus can be as a first step neglected .
it is of the order of the small difference in mass between the proton and the neutron . in the series @xmath198 baryons with n=5 , 7 , 8 and 9 are missing
this raises the possibility that such baryons may exist .
the @xmath199 value for the proton is found to be @xmath200 , which when inserted into eq . enhances the gravitational force by a factor of @xmath201 .
this is consistent with the initial hypothesis that the gravitational energy can be enhanced to the level of @xmath202 .
interestingly it follows from equation that in the case of the proton @xmath129 the inertial and gravitational mass of each rotating particle , @xmath203 , is related to the planck mass , @xmath204 , via a very simple equation , i.e. @xmath205 which provides an interesting direct connection between the planck mass and the present model .
interestingly , in wheeler s geon analysis the minimum mass of a small geon also lies in the planck mass range @xcite .
gravity is generally expected to reach the level of the strong force at energies approaching the planck scale @xmath206 @xcite which is in good agreement with the model results ( eq . ) .
it is interesting to note that when using the inertial or gravitational mass , @xmath203 , in the definition of the compton wavelength , @xmath207 of the particle @xmath208 then one obtains the planck length @xmath209 , but when using the mass corresponding to the total energy of the particles , @xmath210 , then one obtains the proton compton wavelength @xmath211 , which is close to the actual distance between the rotating particles .
another interesting feature of the model is that it is qualitatively consistent with a central experimental observation about the strong force @xcite , i.e. that the normalized angular momentum of practically all hadrons and their excited states is roughly bounded by the square of their mass measured in @xmath212 @xcite .
indeed from and one obtains : @xmath213 which in reasonable qualitative agreement with experiment for small integer @xmath214 values .
it thus appears that deterministic bohr - type models which combine special relativity with the de broglie wavelength equation may provide a useful zeroth order approach to explore some subatomic problems .
the present model suggests that gravity may perhaps provide a mechanism for the binding of light particles , probably neutrinos , to form baryons .
this is qualitatively consistent with wheeler s concept of small neutrino geons @xcite .
while the results are interesting they should be examined in light of the assumptions made , i.e. that ( a ) the force of gravitation for relativistic particles can be expressed by eq .
, obtained by replacing via the equivalence principle the gravitational mass by the inertial mass @xmath13 in newton s gravitational law , and ( b ) that @xmath193 rather than @xmath215 can be used for the quantization of the angular momentum of these rotational states .
cgv acknowledges numerous helpful discussions with professor ilan riess of the physics department at the technion .
26 rovelli , c. : quantum gravity . cambridge university press , ( 2004 ) t hooft , g. : nature * 448 * , 271 ( 2007 ) braun - munzinger , p. & stachel , j. : nature * 448 * , 302 ( 2007 ) nambu , y. : quarks : frontiers in elementary particle physics .
world scientific , singapore , ( 1984 ) schwarz , p.m. & schwarz , j.h .
: special relativity : from einstein to strings .
cambridge university press , ( 2004 ) t hooft , g. : found .
* 10 * , 105 ( 1997 ) t hooft , g. : class .
quantum grav .
* 16 * , 3263 ( 1999 ) t hooft , g. : determinism beneath quantum mechanics , quo vadis quantum mechanics , the frontiers collection , eds .
a. elitzur et al , springer 2005 , isbn 1612 - 3018 , pp .
99 - 111 ; quant - ph/0212095 t hooft , g. : does god play dice ? , physics world * 18*(12 ) , december 2005 griffiths , d. : introduction to elementary particles .
wiley - vch verlag gmbh & co. kgaa , weinheim , ( 2008 ) hehl , f.h .
& ijaki , dj . :
12*(1 ) , 83 ( 1980 ) hoyle , c.d . , et al . :
phys . rev .
lett . * 86 * , 1418 ( 2001 ) long , j.c . , chan , h.w . , churnside , a.b . ,
gulbis , e.a . , varney , m.c.m . ,
price , j.c .
: nature * 421 * , 922 ( 2003 ) oldershaw , r. l. : arxiv:1002.1078 [ physics.gen-ph ] wheeler , j.a . :
phys . rev . * 97*(2 ) , 511(1955 ) misner , c.w . ,
thorne , k.s . &
wheeler , j.a .
: gravitation .
freeman , san fransisco , ( 1973 ) misner , c.w . ,
thorne , k.s . &
zurek , w.h . :
physics today , pp .
40 - 46 , ( 2009 ) gross , d.j . & wilczek , f. : phys .
lett . * 30 * , 1343 ( 1973 ) politzer , h.j .
lett . * 30 * , 1346 ( 1973 ) cabibbo , n. & parisi , g .
b * 59 * , 67 ( 1975 ) french , a.p . : special relativity . w.w .
norton and co. , new york , ( 1968 ) freund , j. : special relativity for beginners .
world scientific publishing , singapore , ( 2008 ) vayenas , c.g .
& souentie , s. : ( 2009 ) arxiv:0904.3850v3 [ physics.gen-ph ] ; ( 2010 ) arxiv:1003.4686 [ physics.gen-ph ] torkelsson , u. : eur . j. phys . * 19 * , 459 ( 1998 ) boyer , t.h . :
am . j. phys . *
72*(8 ) , 992 ( 2004 ) gillies , g.t . :
. phys . * 60 * , 151 ( 1997 ) povh , b. , rith , k. , scholz , ch . &
zetsche , f. : particles and nuclei : an introduction to the physical concepts 5th ed . , springer - verlag berlin heidelberg , ( 2006 )
.experimental @xcite and computed ( eq . ) baryon masses [ cols="<,<,<,<,<,<",options="header " , ] | according to special relativity and the equivalence principle , the newtonian gravitational force between two particles with relativistic velocities increases significantly with velocity and in fact becomes unbound as the latter approaches @xmath0 .
one may thus construct a deterministic model of three fast rotating light particles which uses gravitation as the attractive force , special relativity , the weak equivalence principle and introduces angular momentum quantization as in the bohr model of the h atom .
the model shows the existence of stable rotational states corresponding to highly relativistic particle velocities with radii in the @xmath1 range .
when the rest masses of the three light particles is in the mass range of neutrinos @xmath2 , then surprisingly the rest masses and compton wavelengths of these rotational states are found to correspond to those of baryons and in fact the masses @xmath3 reproduce quite well the light baryon mass spectrum .
interestingly the model also predicts a lower limit for the lifetime of the baryons which is consistent with their measured lifetimes and yields a good approximation for their magnetic moments .
it thus appears that this deterministic bohr - type approach to subatomic phenomena , i.e. a classical special relativistic treatment coupled with the de broglie wavelength expression to seek conformity with quantum mechanics , may be a useful zeroth - order model to explore some quantum gravity problems . +
* keywords * : special relativity , equivalence principle , inertial and gravitational mass , particle confinement , deterministic models , baryon masses , planck mass + * pacs * : 03.30.+p special relativity , 04.60.bc phenomenology of quantum gravity , 04.80.-y experimental studies of gravity , 21.65.qr quark matter constantinos g. vayenas & stamatios souentie _ lcep , caratheodory 1 , st . ,
university of patras , patras gr 26500 , greece _ |
twisted accretion discs around rotating black holes are used for explanations of different observational phenomena .
for example they are invoked to explain jet precession observed in certain active galactic nuclei ( agn ) ( e.g. @xcite and references therein ) , warping of the accretion disc in the maser galaxy ngc 4258 ( e.g. @xcite ) , certain features of emission line profiles coming from agn ( e.g. @xcite ) etc .. @xcite proposed that shear velocities induced by the disc twist ( see below ) provide a mechanism of excitation of high frequency quasi - periodic oscillations observed in many astronomical objects . light curves , forms of emission lines of precessing twisted tori around a kerr black hole as well as their shapes as seen from large distances have recently been investigated numerically by @xcite .
theoretical studies of thin twisted discs started from the seminal paper @xcite , who proposed a simple model of a stationary twisted disc , where a twisted disc was considered as a collection of circular rings interacting with each other due to viscous forces . in frameworks of this model bardeen and petterson came to the conclusion that a stationary twisted disc inclined with respect to the black hole equatorial plane at large distances from the black hole aligns with this plane at smaller radii - the effect called later as `` the bardeen - petterson effect '' .
this alignment was explained as being due to the presence of gravitomagnetic force acting from the side of a rotating black hole on the disc rings .
this force causes precession of orbital planes of free particles around a direction of the black hole spin and breaks the spherical symmetry of the problem .
the bardeen and petterson approach was later developed and generalised on non - stationary twisted discs by e.g. @xcite . in frameworks of hydrodynamical theory of perturbations
@xcite pointed out that the simple model developed in the previous works is , in fact , inconsistent since it does not conserve all components of the angular momentum content of an accretion disc .
they showed that a self - consistent model must , necessarily , contain perturbations of the disc density and velocity fields induced by the disc twist , which determine deviations of the gas particles motion in the disc from the circular one ( see @xcite for a qualitative explanation of this effect ) . in particular , the velocity perturbations should be odd with respect to the disc vertical coordinate thus having the structure of a shear flow .
@xcite also found that when a sufficiently viscous disc is considered in the classical newtonian gravitational field of a point mass a typical alignment scale of a stationary twisted configuration decreases with decrease of the @xcite @xmath0 parameter ( see also e.g. @xcite ) and that non - stationary propagation of twisted disturbances through the disc has a diffusive character with a characteristic time scale being also proportional to @xmath0 ( also e.g. @xcite ) contrary to what was claimed in the previous studies .
these two effects are determined by the well known degeneracy of the keplerian potential , which has orbital and epicyclic frequencies equal to each other .
it was later shown by @xcite that in the opposite case of a low viscosity keplerian disc propagation of non - stationary twisted disturbances through the disc has a wave - like character with a typical speed of the order of speed of sound in the disc .
these studies considered the gravitational field of a black hole as a field of a newtonian point mass with the gravitomagnetic force being treated as a classical force causing precession of the disc rings .
@xcite considered post - newtonian corrections to equations describing velocity perturbations .
they found that when a low viscosity stationary twisted disc is considered and the disc gas is rotating in the same sense as a black hole the bardeen - petterson effect does not take place .
instead , there are radial oscillations of the disc inclination angle with amplitude increasing with decrease of the distance from the black hole .
this effect was later confirmed by time dependent calculations made by @xcite .
@xcite also provided a qualitative explanation of this effect and a general criterion of appearance of these oscillations .
namely , they take place in a low viscosity stationary twisted disc when the signs of the apsidal and nodal precessions are the same . in the opposite case
the disc aligns with the equatorial plane of the black hole for any value of viscosity .
for example , the latter situation is realised when the black hole rotates in the direction opposite to the direction of the orbital motion of the gas in the disc , see @xcite and below or when a newtonian twisted disc around a massive binary star or a binary black hole is considered , e.g. @xcite .
we are going to show below that this criterion is applicable to our fully relativistic problem as well and that whether the disc aligns with the equatorial plane or not depends on the direction of rotation of the black hole relative to the orbital motion .
additionally , @xcite showed that the linear analysis made by @xcite and @xcite can be extended to the case of disc inclination angles , which are larger than the disc opening angle @xmath12 , where @xmath5 is the disc half thickness and @xmath1 is the radial coordinate , and the precise definition of the angle @xmath3 is made in such way that it is a constant independent on @xmath1 , see equations ( [ eq31 ] ) and ( [ eq32 ] ) below .
this was done by employing a formalism based on the so - called `` twisted coordinate '' system introduced by @xcite .
an analogous coordinate system has later been used by @xcite to consider the disc inclination angles of order of unity and , accordingly , to construct a non - linear theory of evolution of sufficiently viscous twisted discs .
as we described above , all previous formalisms of description of the thin twisted discs were based either on the classical or post - newtonian treatments of the problem .
it is , however , important to consider the problem in full general relativity .
this is especially crucial either for the discs having a small value of @xmath13 or in the case of slowly rotating black holes having their rotational parameter @xmath14 , where a stationary accretion disc can be twisted at scales comparable to the black hole gravitational radius , @xmath15 .
a formalism based on general relativity can also be very useful for a quantitative modelling of emission spectra coming from twisted disc as well as for accurate studying of many other physical processes occurring in such discs like e.g. their self - irradiation . in this paper
we construct a fully relativistic theory of thin twisted discs at radii larger than the radius of the last stable orbit around slowly rotating black holes , treating the effect of the black hole rotation in the linear approximation only .
this essentially means that only one term in general equations of motion ( proportional to the rotational parameter @xmath6 and describing the gravitomagnetic force ) can be taken into account and all quantities entering in our final equations describing dynamics of twisted discs can be defined with help of the schwarzschild metric of a non - rotating black hole .
we make the usual assumptions of smallness of the disc opening angle @xmath3 , the disc inclination angle @xmath16 with respect to the equatorial plane and its logarithmic derivative , @xmath17 .
additionally , we neglect self - gravity of the discs . we generalise the twisted coordinate system of @xcite to the relativistic case and
write down our general equations of motion in this generalised coordinate system .
we show that these equations can be naturally split on two different subsets : 1 ) a set of equations describing the standard flat disc accretion and 2 ) a set of equations determining dynamics of variables related to the disc twist . as a set of the standard equations we use equations of the well known @xcite accretion disc model , which is the relativistic generalisation of the @xcite @xmath0-disc model .
equations of the set 2 ) are used to derive a pair of equations ( [ eq50 ] ) and ( [ eq51 ] ) fully describing the dynamics and stationary configurations of twisted discs under our approximations .
similar to the case of newtonian twisted discs ( see e.g. @xcite ) these equations can be formulated for two dependent complex variables - @xmath18 , where @xmath19 is the second euler angle determining longitude of ascending nodes of the disc rings , and @xmath20 determining the form of velocity perturbations in the disc . in the second part of the paper we study in detail shapes of stationary configurations , which are described by equation ( [ eq52 ] ) .
we show that this equation contains only two independent parameters - @xmath0 assumed to be a constant throughout the disc and @xmath21 .
when the disc gas rotates in the same direction as the black hole ( @xmath22 ) , for the gas pressure dominated @xcite models we find that there are three possible qualitatively different shapes of the twisted discs depending on values of @xmath0 .
when @xmath0 is very small the disc exhibits oscillations of the inclination angle , which have a different character of amplitude and radial frequency behaviour at large distances from the black hole and close to the last stable orbit .
the disc behaviour at large radii is the same as described by @xcite . on the other hand oscillations of the inclination angle close to the last stable orbit
proceed in a different regime , with the amplitude and the radial frequency strongly growing towards the last stable orbit .
when @xmath0 has moderately small values the oscillations are suppressed but the disc does not align with the black hole equatorial plane .
the inclination angle is either monotonically growing towards the last stable orbit or exhibits a non - monotonic behaviour decreasing with decrease of @xmath1 at large radii and increasing again towards the last stable orbit at smaller radii . finally , when @xmath0 is sufficiently large the bardeen - petterson effect takes place . in section 2
we present our basic definitions , introduce the twisted coordinate system , describe fundamental equations of motions and the flat disc model used in our study . in section 3
we derive our final equations ( [ eq50 ] ) and ( [ eq51 ] ) describing the dynamics and stationary configuration of the twisted discs .
section 4 is devoted to analytical and numerical studies of the stationary twisted configurations .
note that section 3 and section 4 can be read independently .
we use the natural systems of units throughout the paper setting the gravity constant and the speed of light to unity .
the metric signature is @xmath23 and the usual summation rule over repeating indices is implied .
as was mentioned in introduction we assume that the black hole rotates slowly with the rotational parameter @xmath14 and derive our dynamical equations taking into account only leading terms in @xmath6 .
therefore , for our purposes , it suffices to consider the metric of a kerr black hole in the boyer - lindquist coordinates taking into account only linear in @xmath6 terms @xmath24 it is easy to see that apart from the @xmath25 term the metric coincides with the metric of a non - rotating black hole written in the schwarzschild coordinates , see e.g. @xcite . in the non - relativistic problems the dynamical equations for the twisted discs take simplest form in the so - called `` twisted coordinates '' introduced by @xcite . as we discuss in section 2.2 these coordinates
are obtained from the cylindrical ones by rotation . in order to find a simple relativistic generalisation of the twisted coordinates we would like to bring the metric ( [ eq1 ] ) into another , the so - called spatially isotropic form , where the spatial part of the line element is proportional to the cartesian line element .
this is done by the change of the radial coordinate @xmath26 which brings ( [ eq1 ] ) in the form @xmath27 where the radial and time coordinates are expressed in units of @xmath28 from now on . introducing the cylindrical coordinates @xmath29 we can rewrite ( [ eq3 ] ) in an equivalent form @xmath30 where @xmath31 the metric ( [ eq3 ] ) induces an associated orthonormal tetrad of one - forms @xmath32 we also use cartesian coordinates @xmath33 related to @xmath34 in the usual way .
the twisted coordinates @xmath35 are obtained from the cartesian ones by rotation @xmath36 where the euler angles @xmath37 and @xmath38 are functions to be determined .
the inclination angle @xmath16 is assumed to be small .
it is evident that @xmath39 .
since only @xmath40 is used below we omit the index @xmath41 later on .
it is also convenient to use @xmath42 in our expressions and introduce the variables @xmath43 instead of @xmath16 and @xmath19 and the quantities @xmath44 where partial derivatives over @xmath45 and @xmath1 are denoted by dot and prime , respectively .
the orthonormal one - forms associated with the twisted coordinate system are given by the expressions @xmath46
@xmath47 @xmath48 @xmath49 see appendix a for their relation to the coordinate forms ( [ eq6 ] ) .
when @xmath50 the twisted coordinates are reduced to cylindrical coordinates rotated with respect to the coordinates introduced above by angle @xmath19 and the forms ( [ eq9]-[eq12 ] ) coincide with the expressions ( [ eq6 ] ) .
it is sufficient to have @xmath51 for ( [ eq9]-[eq12 ] ) to coincide with ( [ eq6 ] ) .
clearly , this is due to the spherical symmetry of the schwarzschild space - time . ]
the adjoint basis vectors @xmath52 are obtained from the duality condition @xmath53 .
explicitly , we have @xmath54 @xmath55 @xmath56 @xmath57 in the limit of large @xmath1 the expressions ( [ eq9]-[eq16 ] ) tend to the corresponding expressions calculated for a flat space - time in @xcite .
we project all our dynamical quantities and equations of motion onto the basis ( [ eq9]-[eq16 ] ) . in order to calculate covariant derivatives in this basis the connection coefficients , @xmath58 , should be used .
their explicit forms are given in appendix a. for our purposes we need covariant derivative of a vector and covariant divergence of a tensor projected onto the bases ( [ eq9]-[eq16 ] ) : @xmath59 where @xmath60 stands for covariant derivative .
note that the indices of components of the tensor quantities as well as the ones of the connection coefficients projected onto our orthonormal bases can be raised and lowered with help of the minkowski metric @xmath61 .
our equations of motion follow from the law of mass conservation @xmath62 where @xmath63 is the rest mass density , @xmath64 are components of four velocity , and equality to zero of covariant derivative of the stress energy tensor , @xmath65 we consider the stress energy tensor of a viscous radiative fluid @xmath66 where @xmath67 and @xmath68 are the energy density and pressure , respectively , @xmath69 and @xmath70 are the components of the metric and viscosity tensors , @xmath71 are the components of radiation flux .
we have @xmath72 , where the thermal energy @xmath73 is assumed to be much smaller than the rest energy @xmath63 .
the viscous stress tensor @xmath74 , where @xmath75 is dynamical viscosity , @xmath76 is the shear tensor and @xmath77 is the projection tensor .
note that @xmath78 similar to the analysis of newtonian twisted discs in the fully relativistic case in the linear approximation in angle @xmath16 we can divide our equations of motion ( [ eq18 ] ) and ( [ eq19 ] ) in two parts having different symmetries with respect to the coordinates @xmath79 and @xmath80 : 1 ) a part describing an unperturbed flat disc and 2 ) a part describing the disc s perturbations associated with the disc s twist and warp .
the first ( background ) part follows from standard models of relativistic flat disc accretion , where all dynamical quantities should be projected onto the basis ( [ eq13]-[eq16 ] ) with @xmath81 , @xmath82 and @xmath83 formally set to zero .
in this paper we would like to consider the simplest possible case of a relativistic geometrically thin optically thick stationary @xmath84disc as a background model , and use , accordingly , the results of @xcite , @xcite , hereafter nt , @xcite and @xcite , hereafter rh , to describe it .
the perturbation part is treated in a way similar to what is done in @xcite , hereafter di , it is discussed in section 3 with some technical details relegated to appendix b. for our purposes it is sufficient to consider only a few basic properties of the flat disc models .
additionally , since we consider the case of a slowly rotating black hole in this section we neglect corrections due to a nonzero value of @xmath6 and discuss only accretion discs around a schwarzschild black hole , formally setting @xmath85 .
however , certain basic expressions accounting for linear in @xmath6 corrections are shown in appendix b , for completeness .
equations describing the relativistic flat disc models can themselves be split in two parts : the ones responsible for disc s structure and behaviour determined by effects occurring on a dynamical time scale and the ones determining the disc s structure and behaviour related to a slow viscous time scale .
let us discuss the first group of equations temporarily neglecting viscous interaction between neighbouring rings of the disc . in this approximation
the disc s gas is orbiting in the azimuthal direction with nearly geodesic four - velocity having only two components : @xmath86 and @xmath87 .
we use , accordingly , the standard expressions for geodesic circular motion in the field of a schwarzschild black hole @xmath88 where we set @xmath89 using the fact that we consider the disc situated close to the equatorial plane of a black hole and assume that the disc s gas is orbiting in positive direction with respect to the angle @xmath90 .
an important quantity associated with the circular motion is its frequency with respect to a distant observer , @xmath91 . from equation ( [ eq23 ] ) we get @xmath92 the disc structure in the vertical direction is governed by equation of hydrostatic balance @xmath93 where we neglect @xmath94 .
viscous interactions in the disc result in a slow drift of the disc s gas in the radial direction , and , accordingly , an additional radial component of four velocity , @xmath95 , appears .
equations determining disc s properties related to this effect follow from the laws of conservation of the rest mass , energy and angular momentum , radiation transfer in the vertical direction and properties of viscous interactions in the disc .
they allow one to complete the set of equations of the disc structure and obtain explicit expressions for basic quantities describing the disc .
let us discuss some of these equations relevant for our purposes .
the law of the rest mass conservation may be written in the form @xmath96 where the surface density @xmath97 and the rate of flow of the rest mass through the disc , @xmath98 , is a constant . in the flat disc s models
only @xmath99 and @xmath100 components of the viscous stress tensor are important . as is shown in nt and @xcite
( see also rh ) the laws of energy and angular momentum conservation result in an explicit expression for integrated over the vertical coordinate @xmath99-component of the stress tensor . also ,
integration of this component in these papers is performed over the proper vertical length @xmath101 , see ( [ eq4 ] ) .
therefore , in order to obtain a vertically integrated value of @xmath102 the nt and rh result should be additionally divided by @xmath103 . ]
@xmath104 where the bar stands hereafter for vertically integrated quantities , @xmath105 and @xmath106 .
note that @xmath107 at the marginally stable orbit , when @xmath108 , where the disc truncates .
the value of @xmath109 is obtained from ( [ eqn22 ] ) : @xmath110 . from ( [ eq21 ] ) and ( [ eq23 ] ) it follows another expression for @xmath111 , @xmath112 equating ( [ eq28 ] ) to ( [ eq26nn ] )
we get @xmath113 taking into account that when @xmath114 the expression in the brackets tends to unity we get the well known result that @xmath115 is a constant in the nonrelativistic limit . for our purposes
we need an expression for kinematic viscosity @xmath116 . in this paper
we employ the simplest possible approach assuming that @xmath117 does not depend on the vertical coordinate @xmath79 and use the standard @xcite and @xcite prescription for an estimate of this quantity assuming that it is proportional to the product of a characteristic disc scale height and sound speed , @xmath118 , in the disc @xmath119 where @xmath120 is the proper scale height of the disc related to a coordinate , @xmath5 , of the disc upper boundary in our coordinate system as @xmath121 , and @xmath9 is the usual shakura - sunyaev parameter .
it is assumed to be a constant . from the hydrostatic balance equation ( [ eq25 ] ) it follows that @xmath122 . substituting this relation to ( [ eq30 ] )
we finally define @xmath0 in such a way that @xmath123 using equations ( [ eq29 ] ) and ( [ eq30n ] ) we obtain an important relation @xmath124 finally , we are going to use the ratio @xmath125 in our calculations
. it can be easily obtained from the results of nt provided that one takes into account a correction to the vertical balance equation of nt found in rh .
when the disc is gas pressure dominated and the main source of opacity is determined by the thomson scattering , we have @xmath126 and for the gas pressure dominated disc with the free - free processes giving the main source of opacity we get @xmath127 where @xmath128 is a constant .
note that in both cases @xmath129 at the marginally stable orbit .
it is also important to note that the disc opening angle @xmath130 so defined does not depend on whether we use the boyer - lindquist or the isotropic coordinates .
indeed , taking into account that @xmath131 and @xmath132 we have @xmath133 .
the expressions for the flat nt model listed above are strictly valid only when the radial drift velocity is small : @xmath134 .
this inequality is , however , broken when @xmath135 is small , and , when @xmath136 is very close to the position of the last stable orbit @xmath137 . assuming that the nt expressions can also give order of magnitude estimates even when @xmath138 let us estimate a value of @xmath139 , @xmath140 , where we have @xmath141 .
the expression for @xmath142 follows from equations ( [ eq26 ] ) and ( [ eq30nn ] ) , @xmath143 while the speed of sound may be estimated as @xmath144 , see above . assuming that the new radial variable @xmath139 is small we have from ( [ eq27 ] ) @xmath145 we represent the quantity @xmath146 as @xmath147 , where @xmath148 , and @xmath149 , @xmath150 for the cases of the disc opacity dominated by the thomson scattering and free - free processes , respectively , see ( [ eq31 ] ) and ( [ eq32 ] ) . in the same limit
we can set the values of all variables entering ( [ eqmm1 ] ) and the expression for @xmath151 excepting @xmath146 and @xmath152 equal to their values at @xmath10 .
we get @xmath153 where we use ( [ eqmm2 ] ) . from this equation
we obtain @xmath154 this equation tells that for typical values of @xmath0 and @xmath155 ( see e.g. @xcite , their equation ( 1 ) ) the value of @xmath140 is quite small .
as described above we use equations ( [ eq19 ] ) written in the basis ( [ eq13]-[eq16 ] ) to obtain equations determining dynamics and stationary configurations of a twisted disc and we take into account that our assumption about smallness of the inclination angle @xmath16 allows us to decouple the set of equation ( [ eq19 ] ) onto two subsets : 1 ) the subset of equations describing the standard flat relativistic disc reviewed above ( the `` standard subset '' ) and 2 ) an additional set of equations determining propagation of warped and twisted disturbances through the disc and shapes of non - planar stationary configurations ( we call it later as a subset describing the disc twist ) . all equations of this additional subset depend harmonically on the angle @xmath156 ( or @xmath80 ) and can be obtained from the general set ( [ eq19 ] ) by a procedure employing symmetry of different terms in ( [ eq19 ] ) with respect to the vertical coordinate @xmath79 since in the approximation we use all terms in this subset are either even or odd with respect to the change @xmath157 . namely , let us call the equations of ( [ eq19 ] ) having the index @xmath158 as `` the horizontal part of equations '' and the equation with @xmath159 as `` the vertical equation '' .
then , all terms of the horizontal part having the odd symmetry and all terms of the vertical equation having the even symmetry belong to the subset of equations describing the disc twist while all terms with the opposite symmetries belong to the standard subset .
there are additional simplifications determined by the fact that we consider only the thin discs @xmath160 and slowly rotating black holes with @xmath14 .
the first assumption allows us to consider only terms of order of @xmath161 in the horizontal part and terms of zero and second order in @xmath161 in the vertical equation .
in fact , as we are going to show there is only one zero order term in the vertical equation describing precession of the disc rings due to the lense - thirring effect ( the so - called lense - thirring or gravitomagnetic precession ) .
it is , therefore , proportional to @xmath6 .
thus , this zero order term may also be classified as a `` small '' one due to our second assumption .
a further simplification follows from the fact that the pressure @xmath68 , the thermal density @xmath73 and the dynamic viscosity @xmath162 are small in comparison with @xmath63 : @xmath163 .
thus , these quantities enter in the horizontal part only in expressions containing differentiation over @xmath79 in order to provide terms linear in @xmath161 . in the vertical equation we set @xmath164 due to the same assumption .
this results in absence of @xmath73 in all our expressions used below , see appendix b. the second assumption allows us to suppose that a characteristic evolution time scale of propagation of twisted and warped disturbances through the disc , @xmath165 , is much larger than the disc dynamical time scale @xmath166 . indeed , for a non - rotating black hole with @xmath85
the disc rings inclined with respect to each other can change their mutual orientations only due to interactions determined by either pressure or viscous forces . due to our first assumption
these interactions are small and the corresponding time scales must be proportional to an inverse power of @xmath161 .
for a slowly rotating black hole there is an additional characteristic time scale , @xmath167 , determined by the gravitomagnetic precession of the rings .
it may be estimated as @xmath168 , see e.g. equation ( [ 2a11 ] ) . due to the small ratios @xmath169 and @xmath170 the time derivatives of the euler angles , and , accordingly , the variable @xmath82 defined in ( [ eqn8 ] ) can be neglected in all expressions entering in the horizontal part . as was first noted by @xcite , hereafter pp , for the newtonian problem in order to get a self - consistent set of equations describing the disc twist one must take into account perturbations of velocity , gas density and pressure determined by the twisted disturbances in the disc . in our relativistic generalisation we must use perturbations of four - velocity , and , accordingly , assume that @xmath171 where the quantities with index @xmath172 are determined by the equations of the flat disc model described above . note that in a self - consistent approach we should consider perturbations of the dynamical viscosity , @xmath173 , and radiation flux , @xmath174 , as well .
however , it can be shown that contribution of terms proportional to the unperturbed part of radiation flux , its perturbation and perturbation of the dynamical viscosity , is negligible in our equations at the required order of accuracy .
this is related to the fact that the quantities @xmath175 and @xmath176 are proportional to @xmath177 and @xmath178 .
as we discuss above the terms proportional to @xmath177 give a contribution to the horizontal part of our equations being differentiated over @xmath79 and in the vertical part being multiplied by terms of zero order in @xmath161 .
a direct examination of our set of the perturbed equations shows that this , in fact , does not happen in the linear approximation in angle @xmath179 .
also note that the pressure perturbation @xmath180 does not enter in our set of final equations describing a twisted disc , see the next section .
since both @xmath64 and @xmath181 obey the same normalisation condition @xmath182 the velocity perturbations @xmath183 are orthogonal to @xmath184 : @xmath185 , giving @xmath186 .
the `` horizontal '' velocity perturbations @xmath187 are odd functions of @xmath79 while the `` vertical '' perturbation @xmath188 is an even function of the same coordinate .
the horizontal part of the velocity perturbation is directly determined from our equations for the disc twist , see below .
determination of the vertical part , @xmath188 requires some additional consideration .
namely , as was first shown for the newtonian problem by @xcite , @xmath188 consists of two parts : @xmath189 , the first one is determined by a change of the coordinate @xmath79 with time of a given gas element while the second one is due to the non - coordinate character of our basis vectors ( [ eq13]-[eq16 ] ) .
this results in evolution of projection of the velocity vector @xmath183 onto the vector @xmath190 even when a gas element has a fixed value of @xmath79 provided that either the euler angles are changing with time or there is a non - zero drift velocity component @xmath191 . in order to find @xmath192
let us assume that @xmath193 .
in this case we have @xmath194 , where @xmath195 is given by equation ( [ eq12 ] ) , and @xmath196 is the line element .
using this equation , taking into account that @xmath197 and @xmath198 , see equations ( [ eq9 ] ) and ( [ eq10 ] ) , respectively , we get @xmath199 in what follows we set @xmath193 . this requirement is related to a specific freedom of choice between different twisted coordinate systems corresponding to the same physical situation .
indeed , in order to construct our twisted coordinate system we add two euler angles to the set of the state variables describing our physical system while the number of dynamical equations remains the same .
therefore , in the twisted coordinate system there is a freedom of choice between different sets of dynamical variables satisfying the same equations , which must be fixed by an additional requirement .
as was discussed in @xcite the requirement @xmath193 fixes this freedom and leads to a choice of the most appropriate twisted coordinate system , where perturbations of the state variables are minimal .
since the unperturbed parts of our variables and the perturbations enter in a different way in all equations below , we omit for simplicity the index @xmath172 characterising the unperturbed variables from now on .
the horizontal part of our equations and the vertical equation are shown in appendix b , see equations ( [ 2a1]-[2a3 ] ) and equation ( [ 2a4 ] ) , respectively .
they have a rather complicated structure .
however , they can be further simplified .
indeed , a simple analysis shows that all terms in these equations proportional to the rotational parameter @xmath6 excepting the gravitomagnetic term - the second term in square brackets on the right hand side of ( [ 2a4 ] ) can be neglected . these terms lead to terms in our final equation ( [ eq38 ] ) proportional to a product of two small parameters - @xmath200 .
thus , we can consider an effectively schwarzschild problem with inclusion of only one term determined by rotation of the black hole .
additionally , the time derivatives of @xmath201 and @xmath188 can also be neglected since , similar to the newtonian problem , they result in terms proportional to @xmath202 in the same equation . on the other hand ,
the time derivatives of the perturbed `` horizontal '' velocities @xmath203 and @xmath204 should be retained in a certain combination , see e.g. @xcite , di and below .
the reason for this is the well known degeneracy of the keplerian gravitational potential resulting in closed orbits of particles around a newtonian source of gravity . owing to this degeneracy a certain combination of terms proportional to the time derivatives ( we call it later the `` resonance combination '' ) plays an important role at large distances from the black hole , where the gravitational potential is approximately the newtonian one provided that @xmath205 .
taking into account the simplifications discussed above let us consider equations of appendix b. at first let us deal with the horizontal part of the problem .
it turns out that it is more instructive to use certain linear combinations of equations ( [ 2a1 ] ) and ( [ 2a3 ] ) instead of themselves . neglecting the terms proportional to @xmath6 and @xmath206 in these equations and taking the linear combination @xmath207 we
remove the time derivative of @xmath204 from the result , thus obtaining @xmath208 in the newtonian limit , when @xmath114 , equation ( [ eq35 ] ) reduces to the continuity equation .
another equation reducing in the newtonian limit to the @xmath90 component of perturbed navier - stokes equations can be obtained from ( [ 2a1 ] ) and ( [ 2a3 ] ) taking another linear combination @xmath209 with the result @xmath210 with our simplification being adopted equation ( [ 2a2 ] ) has the form @xmath211 it is easy to show that it reduces to the @xmath1 component of perturbed navier - stokes equations when @xmath114 .
the vertical equation ( [ 2a4 ] ) plays a special role in our analysis .
in fact , as we see later it most important for determination of the dynamics and form of twisted configurations of accretion discs . in the approximation
we use , i.e. setting all terms proportional to @xmath6 excepting the gravitomagnetic term in ( [ 2a4 ] ) , and , accordingly , in ( [ 2a5]-[2a7 ] ) to zero we can bring this equation to the so - called divergent form , which reflects symmetries of the schwarzschild space - time with respect to spacial rotations .
this is analogous to the newtonian problem , see pp and di .
for that we would like to express the density perturbation @xmath212 and the velocity perturbation @xmath204 entering ( [ 2a4 ] ) through @xmath203 using equations ( [ eq35]-[eq37 ] ) . since the resonance combination of terms containing the time derivatives does not enter in the vertical equations we can set @xmath213 when operating with equations ( [ eq35]-[eq37 ] ) .
after this being done we express the term containing @xmath214 in ( [ eq35 ] ) through the velocity perturbations using equations ( [ eq25 ] ) and ( [ eq37 ] ) , and substitute the result in ( [ eq35 ] ) .
we express the velocity perturbation @xmath204 entering explicitly in ( [ eq35 ] ) and ( [ 2a4 ] ) , in terms of @xmath203 with help of ( [ eq36 ] ) , and substitute it in these equations .
then , we use the relation obtained from ( [ eq35 ] ) to express the density perturbation through @xmath215 and the viscous terms and substitute the result in ( [ 2a4 ] ) .
the resulting expression is integrated over @xmath79 , where we note that the term containing the derivative of the perturbed pressure , @xmath180 , vanishes after integration .
this expression also contains derivatives over @xmath79 in the viscous terms ( [ 2a5]-[2a6 ] ) under the integrals .
these are integrated by parts taking into account that the corresponding surface terms are equal to zero .
after a rather tedious but straightforward calculation along the lines outlined above we obtain from ( [ 2a4 ] ) a rather simple equation @xmath216 where we remind that @xmath217 is the surface density and the bar stands for the quantities integrated over @xmath79 . in appendix c
we show that when @xmath85 equation ( [ eq38 ] ) may be brought in a special divergent form determined by rotational symmetries of the schwarzschild space - time . provided a background model of a flat relativistic disc
is specified equations ( [ eq36]-[eq38 ] ) form a complete set . in order to transform the set of equations ( [ eq36]-[eq38 ] ) to a simpler form
we should specify a dependence of density @xmath218 on @xmath79 . since an accurate treatment of the vertical distribution of the density is not important for our purposes we would like to consider the simplest possible case of an isothermal disc @xmath219 where the central density @xmath220 and the disc half - height @xmath5 are functions of @xmath1 .
similar to the newtonian problem it is easy to see that distributions of the velocity perturbations of the form @xmath221 satisfy equations ( [ eq36 ] ) and ( [ eq37 ] ) provided that the kinematic viscosity @xmath117 does not alter with height and the amplitudes @xmath222 , @xmath223 , @xmath224 and @xmath225 are assumed to be functions of the time and radial coordinates . introducing the complex notation @xmath226 and using equations ( [ eq25 ] ) , ( [ eq30n ] ) , ( [ eq40 ] ) , ( [ 2a5 ] ) and ( [ 2a7 ] ) we can rewrite ( [ eq36 ] ) and ( [ eq37 ] ) in the form @xmath227 @xmath228 respectively , where @xmath229 note that @xmath230 is the relativistic epicyclic frequency for a slightly perturbed circular orbit in the schwarzschild space - time , see e.g. @xcite .
as we have mentioned above equations ( [ eq42 ] ) , ( [ eq43 ] ) contain a degeneracy in the newtonian limit @xmath231 , for a formally inviscid disc , due to the well known keplerian resonance between the mean and epicyclic motions .
it is due to this resonance the `` small '' terms @xmath232 and @xmath233 should be retained in these equations . in order to clarify the physical meaning of this resonance let us consider the limit @xmath114 and set @xmath234 in equations ( [ eq42 ] ) and ( [ eq43 ] ) .
we get @xmath235 from equation ( [ eq42 ] ) , and @xmath236 where @xmath86 is assumed to have its keplerian value , @xmath237 . multiplying ( [ eq45 ] ) by @xmath238 and adding the result to ( [ eq46 ] )
, we see that the terms proportional to @xmath239 on the left hand sides of ( [ eq45 ] ) and ( [ eq46 ] ) cancel each other , and we obtain @xmath240 on the other hand we must have @xmath241 , where we remind that @xmath242 is the characteristic dynamical time scale , @xmath165 is the characteristic time scale of evolution of twisted disturbances and @xmath243 .
taking this fact into account we see that the term on the right hand side of ( [ eq47 ] ) is of order of @xmath244 . therefore , it can be neglected in the leading approximation in ( [ eq46 ] ) , and we get either from this equation or from ( [ eq45 ] ) an algebraic relation between @xmath245 and @xmath246 : @xmath247 which is valid only in the leading order in @xmath169 .
thus , in this order the perturbed velocities can be described by only one complex amplitude , either @xmath245 or @xmath248 ) follows from the fact that slightly perturbed circular orbits of free particles in the newtonian potential are ellipses with a small eccentricity . ] .
let us show that the same property approximately holds for the general relativistic problem as well .
for that let us consider a linear combination of ( [ eq42 ] ) and ( [ eq43 ] ) with coefficients chosen in such a way that the term proportional to @xmath249 is eliminated @xmath250 ( i-\alpha)\omega { \bf b } - \left [ ( i+\alpha)u^\varphi \omega - \frac{3i\alpha}{i-\alpha } k_1 ( u^\tau)^2u^\varphi \tilde \omega \right ] { \bf w}^\prime .
\label{eq49}\ ] ] the combination on the left hand side plays a role in dynamics of the disc only when we consider a non - relativistic , low viscosity disc with @xmath251 while in the opposite limits it can be set to zero .
therefore , we can consider all background quantities in this combination equal to their newtonian limits , set @xmath234 there and use equation ( [ eq48 ] ) to express @xmath252 through @xmath248 on the right hand side .
we get @xmath253 ( i-\alpha)\omega { \bf b } - \left [ ( i+\alpha)u^\varphi \omega - \frac{3i\alpha}{i-\alpha } k_1 ( u^\tau)^2u^\varphi \tilde \omega \right ] { \bf w}^\prime \right \rbrace \label{eq50}\ ] ] equation ( [ eq50 ] ) describes the horizontal part of the problem in the leading order in @xmath169 , for all possible values of @xmath0 and @xmath1 .
it is one of two final equations describing a fully relativistic twisted disc around a slowly rotating black hole . in order to get the second final equation describing our problem
let us consider the vertical equation ( [ eq38 ] ) .
at first we use equations ( [ eq39]-[eq41 ] ) and ( [ 2a7 ] ) to rewrite it in terms of the complex variables and perform integration over @xmath79 . after this being done we get expression proportional to @xmath254 and @xmath255 under the derivative over @xmath1 .
we substitute ( [ eq29 ] ) and ( [ eq30n ] ) and take the constant factors out of the derivative .
then , we use equation ( [ eq26 ] ) to express @xmath95 in terms of @xmath98 and the explicit form of @xmath256 given by ( [ eq26nn ] ) . after the resulting expression
is divided over @xmath257 one can see that it contains @xmath98 and @xmath257 only in the combination @xmath258 .
this combination is expressed in terms of @xmath259 with help of ( [ eq30nn ] ) .
we finally obtain @xmath260 equations ( [ eq50 ] ) and ( [ eq51 ] ) form a complete set .
non - stationary solutions of equations ( [ eq50 ] ) and ( [ eq51 ] ) will be discussed in a separate publication . in this paper
we would like to consider a stationary problem setting time derivatives in ( [ eq50 ] ) and ( [ eq51 ] ) to zero .
then , we express @xmath248 through @xmath261 using ( [ eq50 ] ) and substitute the result in ( [ eq51 ] ) to obtain @xmath262 where @xmath263 stands for the complex conjugate , @xmath264 we use the coordinate @xmath136 as an independent variable , and the explicit expression for the gravitomagnetic frequency @xmath265 ( [ 2a11 ] ) is used .
it is easy to see that in the non - relativistic limit @xmath266 and @xmath267 the function @xmath268 tends to the function @xmath269 introduced in paper @xcite , hereafter kp .
a character of solutions to equation depends on only two independent parameters - @xmath0 and @xmath270 a number of authors considered equations describing a shape of a stationary twisted disc far from a rotating black hole .
they used either a purely newtonian analysis with the gravitomagnetic term treated as originating from a newtonian force responsible for precession of the disc rings ( e.g. pp , kp ) or took into account certain post newtonian terms , which may play a significant role in determining of the stationary disc configurations ( e.g. @xcite , hereafter ii ; @xcite ) .
obviously , equation ( [ eq52 ] ) should be reduced to equations exploited by previous authors in the respective limits . in particular , in the paper ii a low viscosity
@xmath271 stationary twisted disc was considered and a simple equation describing the stationary configurations was derived taking into account only the most important post - newtonian correction to the first term on the right hand side of ( [ eq52 ] ) . in the approximation used by ii
we neglect the second term on the right hand side of ( [ eq52 ] ) , which is proportional to @xmath0 and set the values of all quantities there equal to their newtonian limits ( i.e. @xmath272 , @xmath273 , @xmath274 and @xmath275 ) and assume that @xmath234 everywhere excepting the so - called resonance denominator in the first term in ( [ eq53 ] ) , which should contain a term proportional to @xmath9 and a first relativistic correction . in this approximation
we have @xmath276 and obtain from equation ( [ eq52 ] ) @xmath277 where @xmath278 .
this equation coincides with equation ( 33 ) of ii .
pp derived dynamical equations describing a newtonian twisted accretion disc in the limit of relatively large viscosity @xmath279 .
stationary solutions to these equations were examined in the paper kp , for a disc with isothermal distribution of density with height . in order to obtain a fully newtonian limit of ( [ eq52 ] ) and compare it to equation used by kp we set @xmath272 , @xmath273 , and @xmath275 in ( [ eq52 ] ) and ( [ eq53 ] ) .
we also introduce the independent variable @xmath280 , use the form of @xmath281 employed by kp and parametrisation of @xmath146 , @xmath282 , where @xmath283 is a parameter as in kp . in this way we obtain from ( [ eq52 ] ) equation studied by kp @xmath284 - 6\alpha { d\over dx_2}{\bf w } \right ) + 16iax_2 ^ 6{\sqrt r_{ms } \over h_{*}^2}{\bf w}=0.\label{eq55}\ ] ] an important limiting case of ( [ eq52 ] ) is obtained by formally setting @xmath234 in this equation .
when the disc gas rotates in the same sense as the black hole and , accordingly , @xmath285 the behaviour of the disc inclination angle differs drastically from the standard picture of the well known bardeen - petterson effect , where it is expected that the disc aligns with the black hole equatorial plane at sufficiently small radii .
contrary to this picture when considering the shape of low viscosity twisted disc at large distances @xmath286 ii found a phenomenon of spatial oscillations of the disc inclination angle @xmath16 for the case of @xmath7 while in the opposite case the disc does align with the black hole equatorial plane .
our fully relativistic analysis allows us to consider this effect in more details and show that the amplitude of the oscillations can be significantly amplified close to the last stable orbit . setting @xmath234 in ( [ eq52 ] ) we get @xmath287 where @xmath288 note that equation ( [ eqn1 ] ) depends parametrically only on @xmath289 .
since this equation has only real coefficients its solution may be taken to be real .
thus , in this limit only the angle @xmath16 changes with radius while @xmath19 stays unchanged .
it is also important to note that the influence of the black hole gravitomagnetic force on the disc shape is most significant when @xmath8 . also , it is most natural to suppose that the rotation parameter @xmath7 .
the case of @xmath8 and @xmath7 is considered in more details below .
note , however , that the bulk of our expressions below are formally valid for the case @xmath11 as well .
although in the case considered below we may choose @xmath290 we continue to use @xmath291 to keep our notation uniform .
at first let us discuss the form of solutions to ( [ eqn1 ] ) close to the last stable orbit assuming that @xmath292 .
for that we change the independent variable to @xmath139 in ( [ eqn1 ] ) and assume that all quantities entering this equation , which have finite values at @xmath10 are equal to these values .
thus , we take into account only the dependencies of @xmath152 and @xmath146 on @xmath139 in such approximation .
we use equation ( [ eqmm2 ] ) , which tells that @xmath293 close to @xmath10 .
we parametrise the dependency of @xmath146 on @xmath139 as @xmath294 , with @xmath295 being proportional to @xmath3 .
we are going to consider only the values of @xmath296 , which physically corresponds to the condition that the surface density @xmath257 tends formally to zero when @xmath297 .
equations ( [ eq31 ] ) and ( [ eq32 ] ) tell that @xmath149 , @xmath298 and @xmath299 , @xmath300 , for the opacity dominated by the thomson scattering and the free - free processes , respectively . in this way we get from ( [ eqn1 ] ) @xmath301 where @xmath302 is evaluated at @xmath10 . using the parameter @xmath289 given by ( [ eq57 ] ) we can rewrite ( [ eq56a ] ) in the form @xmath303 , where @xmath304 and @xmath305 , the indices @xmath306 and @xmath307 stand for the cases of the thomson and free - free opacities , respectively . when @xmath285 the solution to equation ( [ eq56 ] ) regular at @xmath308 may be expressed through the bessel function , @xmath309 , as @xmath310 where @xmath311 it is easy to see that the solution ( [ eq58 ] ) tends to a non - zero constant when @xmath312 .
this fact may be used to express the constant @xmath313 through the value of @xmath291 at @xmath314 , @xmath315 , as @xmath316 where @xmath317 is the gamma function and we use the well known approximate expression for @xmath309 at small @xmath318 . since the solution contains the bessel functions of a real argument it describes oscillations of the inclination angle @xmath319 the solution contains the bessel function of an imaginary argument , which describes a monotonic decrease of the inclination angle of the disc with decrease of @xmath139 . ] .
let us estimate a change of the amplitude of these oscillations over the range @xmath320 , in the limit of @xmath8 .
as follows from ( [ eq56a ] ) and ( [ eq59 ] ) in this limit the variable @xmath321 takes values @xmath322 even for @xmath323 and , therefore , the region can be divided on two domains : @xmath324 and @xmath325 . in the domain
@xmath324 the value of @xmath291 monotonically decreases with @xmath139 , and , typically , it decreases in six times before oscillations of the inclination angle begin .
thus , we have @xmath326 .
when @xmath327 we use the asymptotic expansion of the bessel function in powers of @xmath328 simplifying ( [ eq58 ] ) to @xmath329 assuming that this expression is approximately valid even when @xmath330 or @xmath331 and that @xmath332 , we can roughly estimate @xmath333 using the quantities @xmath289 and @xmath334 in ( [ eq62 ] ) and remembering that @xmath335 we obtain @xmath336 since @xmath337 for the parameters we use we set this factor to unity in our expressions below . from equation ( [ eq63 ] )
it follows that when @xmath338 and @xmath8 a rise of the amplitude of the oscillation with decrease of @xmath139 near the marginally stable orbit can be quite large .
ii showed that when the limit @xmath8 is considered a low viscosity disc inclined at large distances with respect to the equatorial plane gets twisted at a characteristic scale @xmath340 at such large scales we can use equation ( [ eq54 ] ) considering the limit @xmath341 there . in this limit we get @xmath342 where we remind that @xmath278 and that we consider the case @xmath7 .
note that following ii we neglect a very slow dependence of @xmath146 on @xmath136 for @xmath343 and set @xmath344 in ( [ eq64 ] ) .
the solution to ( [ eq64 ] ) can be expressed through the bessel functions @xmath345 where @xmath346 and @xmath222 and @xmath223 are constants .
( which is equal to our variable @xmath347 ) in front of the square brackets in their equation ( 37 ) should be @xmath348 and not @xmath349 as in the text . ]
the limit of large @xmath350 corresponds to the limit of small @xmath347 . in this limit
the first and second terms in the brackets in ( [ eq65 ] ) being multiplied by @xmath351 tend to a non - zero constant and to zero , respectively .
this fact may be used to express the constant @xmath222 through the asymptotic value of @xmath291 at infinity , @xmath352 , as @xmath353 in the opposite limit of @xmath354 we get a simple approximate expression for @xmath355 from ( [ eq65 ] ) and ( [ eq66 ] ) using the asymptotic expressions for the bessel functions at large values of their arguments @xmath356 . \label{eq67}\ ] ] equation ( [ eq67 ] ) tells that in the region @xmath357 the inclination angle oscillates with the amplitude of oscillations proportional to @xmath358
. the asymptotic expression ( [ eq61 ] ) and ( [ eq67 ] ) can be matched together by a wkbj solution .
indeed , in the limit @xmath359 the ratio of the quantities @xmath360 and @xmath361 entering ( [ eqn1 ] ) , @xmath362 , is large for the range of @xmath136 such as @xmath318 and @xmath347 defined in equations ( [ eq59 ] ) and ( [ eq65 ] ) , respectively , are large . applying the standard wkbj scheme we find the solution in this region in the form @xmath363 where the constants @xmath364 and @xmath365 must be chosen in such a way that the solution ( [ eq61 ] ) is matched in the corresponding asymptotic limit .
it is easy to see that in order to have such a matching we should require that @xmath366 and @xmath367 where it is assumed that @xmath368 and @xmath369 are evaluated at @xmath308 and we use the facts that @xmath370 and that @xmath371 close to @xmath10 . in the limit
@xmath266 we can set the values of @xmath360 and @xmath361 in front of the cosine in ( [ eq68 ] ) equal to their newtonian values .
the integral in ( [ eq68 ] ) can be represented as @xmath372 , where @xmath373 . taking into account that the newtonian value of @xmath374 we have @xmath375 , and , accordingly , @xmath376 the expression ( [ eq70 ] ) can be matched to the asymptotic expression ( [ eq67 ] )
provided that the constant @xmath222 and @xmath223 are appropriately chosen .
a simple calculation gives @xmath377 equations ( [ eq58 ] ) , ( [ eq65 ] ) and ( [ eq68 ] ) provide expressions for an approximate shape of a low viscosity twisted disc in the whole allowed range of @xmath136 , @xmath378 .
note that although we derive these equations assuming that @xmath7 they are formally valid for negative values of @xmath6 as well .
equations ( [ eq60 ] ) , ( [ eq66 ] ) , ( [ eq69 ] ) and ( [ eq71 ] ) allow us to relate the asymptotic value at large radii , @xmath379 , to the value of @xmath291 at the last stable orbit , @xmath380 , as @xmath381 where an explicit form for @xmath382 follows from these equations .
it is important to note that @xmath383 as follows from equations ( [ eq66 ] ) and ( [ eq71 ] ) .
thus , for values of @xmath384 such that @xmath385 , @xmath386 while @xmath387 .
this describes a peculiar `` resonant '' solution to equation ( [ eqn1 ] ) , where a regular at the last stable orbit solution is matched precisely to the solution of ( [ eq64 ] ) proportional to @xmath223 while @xmath388 , see equation ( [ eq65 ] ) . from equation ( [ eqn2 ] ) it follows that the integral @xmath389 in the cosine can be written in the form @xmath390 , where @xmath391 does not depend on @xmath289 , we have @xmath392 and @xmath393 .
this allows us to write the condition for obtaining the resonant solutions as a condition that the parameter @xmath394 has discrete values @xmath395 where @xmath396 is an integer .
thus , a stationary twisted disc described by such a solution has its inclination angle going to zero with radius . note , however , that inclusion of effects determined by the viscosity coefficient @xmath0 leads to a partial suppression of this resonance . as we see later numerical results
show that when the resonant condition ( [ eq72 ] ) is fulfilled and a small but nonzero value of @xmath0 is considered the value of @xmath397 at the resonant values of @xmath289 is much smaller than its neighbouring values instead of being equal precisely to zero . neglecting the possibility of the resonances we can roughly estimate the dependence of @xmath397 on @xmath289 using very simple arguments .
indeed , from equation ( [ eq67 ] ) it follows that the inclination angle scales with @xmath136 as @xmath398 for radii @xmath399 , where @xmath400 is given by equation ( [ eq63a ] ) while when @xmath401 it stays approximately constant .
thus , we can roughly estimate @xmath402 .
this estimate does not take into account the growth of the amplitude of the oscillations close to the last stable orbit , which is described by equation ( [ eq62 ] ) . using this equation and assuming , for simplicity , that the numerical factor @xmath403 there , we estimate @xmath404 . combining both estimates
we get @xmath405 we derive our equations assuming that the inclination angle @xmath319 is small . also , we suppose that the nt model of the flat background disc with a constant value of the parameter @xmath0 is valid .
these assumptions put rather severe limitations on validity of our analysis , especially in the case of the low viscosity twisted disc considered in section 4.2 . since in this case @xmath406 and @xmath407 the condition @xmath408 everywhere in the region @xmath409 leads to the condition @xmath410 .
equation ( [ eq73 ] ) tells that in this case the asymptotic value of the inclination angle at spacial infinity should be quite small @xmath411 another somewhat more stringent constraint stems from the fact that the shear velocities induced in the disc by the disc twist should not be too large .
when considering the oscillations of the disc inclination angle at scales @xmath412 ii supposed that a characteristic amplitude of the shear velocities should not exceed the sound speed , @xmath151 . in the opposite case
the shearing instability could make the disc more turbulent , thus increasing an effective value of @xmath0 .
another possibility is the presence of shocks in the disc ( e.g. @xcite ) , which may heat up the disc thus effectively increasing the relative disc thickness @xmath146 .
ii obtained a condition that the relativistic oscillations do not induce such velocity amplitudes at the scale @xmath412 provided that @xmath413 .
note , however , that our numerical results suggest that viscosity effects suppress the oscillation amplitude , and , accordingly , the velocity amplitude more efficiently than it was assumed in a simplified analysis of ii . therefore , the constraint of ii may to be too restrictive for more realistic models taking into account a small but non - zero value of @xmath9 , see the next section . in order to estimate a characteristic amplitude of the shear velocities induced in the disc
we assume that it is of the order of @xmath414 , where @xmath248 is defined in equation ( [ eq41 ] ) and we remind that it can be considered as a real quantity as far as equation ( [ eqn1 ] ) is concerned . from equation ( [ eq50 ] ) it follows that @xmath415 when @xmath233 and @xmath9 are equal to zero there and @xmath416 , which is appropriate for the radii close to @xmath10 . on the other hand , from the results of section 2.4
it follows that @xmath417 . introducing the variable @xmath418 we have @xmath419 . from equation ( [ eq61 ] ) it follows that when @xmath323 but @xmath325 we have @xmath420 and from equation ( [ eq60 ] ) it follows that @xmath421 , and , therefore , @xmath422 since @xmath423 and the factor @xmath424 we can roughly estimate that @xmath425 in ( [ eq75 ] ) .
we see that the condition @xmath426 is fulfilled when @xmath427 according to our criterion our analysis is valid close to @xmath10 provided that we demand that @xmath428 , at least .
this leads to inequality @xmath429 , which can be reformulated in terms of @xmath352 using equation ( [ eq73 ] ) @xmath430 where @xmath431 is defined in ( [ eq74 ] ) .
we see that this inequality is stronger than the inequality ( [ eq74 ] ) although the difference is not quite significant since the parameter @xmath67 is typically rather small .
we integrate numerically equation ( [ eq52 ] ) starting from the last stable orbit .
similar to the case of almost inviscid twisted disc considered above there is a partial solution regular at the last stable orbit and we use this solution to specify the boundary condition at @xmath10 .
we consider below the case of thomson opacity only .
the case of free - free opacity is quite similar . on the radial coordinate @xmath136
is shown . since the problem is a linear one we set hereafter the value of the inclination angle at large radii , @xmath432 , equal to unity in all figs .
the parameter @xmath433 .
the solid curve is obtained by setting @xmath434 in ( [ eq52 ] ) .
the short dashed curve represents our analytical results discussed above .
we use the expression for @xmath16 given by equation ( [ eq68 ] ) in an intermediate region of @xmath136 , when @xmath435 the expression ( [ eq58 ] ) is used and when @xmath436 or larger we use equation ( [ eq65 ] ) . the dot - dashed , dot - double dashed , long dashed curves correspond to @xmath437 , @xmath438 and @xmath439 , respectively . _
right panel : _ the result of integration of equation ( [ eq52 ] ) presented in parametric form . @xmath440 and @xmath441 , @xmath442 and @xmath443 for this plot . the circle , square , triangle ( pointing up ) , triangle ( pointing down ) and the star correspond to @xmath444 , @xmath445 , @xmath446 , @xmath447 and @xmath448 , respectively.,width=604 ] at first let us discuss the dependency of @xmath16 on @xmath136 for the case of relatively small viscosity parameter @xmath449 and @xmath7 . this case is represented in figs .
[ figg1 ] ( left panel ) and [ figg2 ] , for @xmath433 and @xmath450 , respectively . in these figs , using the fact that our problem is a linear one
. this ratio can be arbitrary large provided that @xmath451 is sufficiently small . ] as seen from these figs .
the oscillations of the inclination angle are quite prominent for the case when @xmath452 in equation ( [ eq52 ] ) as well as for the case of quite small @xmath453 .
also , it is seen that our analytic theory described above gives the curve ( the left panel in fig .
[ figg1 ] ) , which is in a quite good agreement with the numerical one calculated for @xmath234 .
effects of viscosity tend to smooth out the oscillations , which disappear for the curves with @xmath454 and @xmath439 on the left panel in fig .
[ figg1 ] .
the similar tendency is seen in fig .
[ figg2 ] where , additionally , the case of relatively large @xmath455 is shown as well .
it is clear from these figs . that when @xmath9 is small the inclination angle grows with decrease of @xmath136 either on average or monotonically
thus , in these cases the disc does not align with the black hole equatorial plane and the bardeen - petterson effect is absent .
note that the smooth curves do not have sharp gradients of the inclination angle and , accordingly , they correspond to solutions , where the shear velocities induced in the disc are relatively small . therefore , these solutions may be physically realistic even for relatively large values of the inclination angle at large radii , @xmath432 . but for @xmath456 .
the solid , dotted , short dashed , dot - dashed , dot - double dashed and long dashed curves correspond to @xmath457 , @xmath438 , @xmath439 , @xmath458 and @xmath459 , respectively.,width=340 ] instead of showing dependencies of the second euler angle @xmath460 on @xmath136 we plot curves corresponding to our solutions on the plane @xmath461 , which depend parametrically on @xmath136 .
since the angle @xmath19 does not change significantly with @xmath136 for very small values of @xmath462 and the corresponding curves are close to straight lines , the case of small viscosity parameters is represented by only one curve corresponding to the solution with @xmath443 , see the right panel in fig .
[ figg1 ] .
one can see from this fig . that this curve has a form of a tight spiral . in this case @xmath463 and @xmath464 , so the curve starts at @xmath308 when @xmath465 and @xmath466 .
it spirals clockwise with increase of @xmath136 , i.e. in the direction opposite to the orbital motion and black hole rotation , which are counterclockwise .
the dependence of @xmath16 on @xmath136 as well as the solutions in the parametric form for @xmath433 and larger values of @xmath467 , @xmath450 , @xmath468 and @xmath459 are shown in fig .
[ figg3 ] . from the left panel in fig .
[ figg3 ] one can see that in this case the inclination angle monotonically decrease in practically the whole range of @xmath136 while there is a small increase of the angle close to @xmath10 .
in general , the ratio of @xmath469 is quite small and this may be interpreted as manifestation of the bardeen - petterson effect .
contrary to the case of the low viscosity discussed above the initial value of @xmath470 is smaller than one , thus for the curves on the right panel of fig .
[ figg3 ] @xmath471 increases its value when the curves spiral from @xmath10 to larger @xmath136 .
the direction of spiralling with the increase of @xmath136 is also clockwise similar to what is shown on right panel of fig .
[ figg1 ] . however , returning to fig .
[ figg2 ] we see that for @xmath472 the bardeen - petterson effect is absent even when @xmath473 in the sense that the ratio @xmath469 remains to be of order of unity .
thus , in the case of @xmath7 the bardeen - petterson effect occurs only when the viscosity parameter @xmath0 is sufficiently large and @xmath474 is sufficiently small .
but for larger values of @xmath9 .
the solid , dashed , dot - dashed and dot - double dashed curves are plotted for @xmath467 , @xmath450 , @xmath468 and @xmath459 , respectively .
curves of the same style correspond to the same @xmath9 on both panels .
different symbols on the right panel correspond to different values of @xmath136 , @xmath475 , @xmath476 and @xmath477 for the circles , squares and triangles , respectively .
, width=604 ] but for the case of the black hole rotating in the opposite sense with respect to the disc rotation , @xmath11 . the solid , dotted , short - dashed , dot - dashed , dot - double dashed and the long - dashed curves correspond to @xmath234 , @xmath478 , @xmath479 , @xmath450 , @xmath468 and @xmath459 , respectively .
curves of the same style correspond to the same @xmath9 on both panels and positions of different symbols on the right panel correspond to the same @xmath136 as in fig .
[ figg3].,width=604 ] the case of @xmath480 is shown in fig .
[ figg4 ] , for @xmath433 and a range of values of @xmath0 . as is seen from this fig .
contrary to the previous case the disc always aligns with the black hole .
this is valid even in the case of setting @xmath481 in ( [ eq52 ] ) , which is described by the solid curve in fig . [ figg4 ] .
the curves with small values of @xmath0 are close to this limiting curve .
it is interesting to point out that as seen from the right panel in fig .
[ figg4 ] the direction of spiralling when @xmath136 runs from smaller to larger radii is opposite to the case of @xmath482 .
now the curves spirals in the direction of the orbital motion but , again , opposite to the direction of the black hole rotation .
thus , the direction of spiralling is always opposite to the black hole rotation .
this may have some observational consequences . as we discuss above our equation
contains two independent parameters - @xmath0 and @xmath289 .
it is of interest to describe how the most important characteristics of the solutions such as the ratio @xmath469 depend on them . in fig .
[ figg5 ] we show the dependence of this ratio on @xmath289 for various sufficiently small values of @xmath9 .
one can see that the curve corresponding to @xmath481 experiences a quite non - monotonic behaviour with series of peaks at some successive values of @xmath289 .
since the analytical curve based on our wkbj theory developed above is in excellent agreement with the numerical one we can identify these peaks with the resonant solutions of ( [ eq52 ] ) with @xmath234 for which the value of the inclination angle tends to zero when @xmath484 .
the positions of these peaks are , accordingly , given by equation ( [ eq72 ] ) .
note also that our simple estimate ( [ eq73 ] ) of the dependence of @xmath469 on @xmath289 provides a very good approximation to the numerical curve when the resonance peaks are not taken into account . from the analytical theory it follows that when @xmath234 formally the amplitude of peaks in ( [ figg5 ] ) is infinite .
when a small value of @xmath0 is taken into account the peak s amplitudes get finite values decreasing with increase of @xmath0 and/or @xmath289 . for a relatively large value of @xmath485
the peaks practically disappear and the corresponding curve strongly deviates from the theoretical one . in the case the values of @xmath486 in the whole range of considered values of @xmath289 . , on the parameter @xmath487 .
the solid curve is calculated for @xmath9 set to zero in ( [ eq52 ] ) , the short dashed curve represents @xmath488 , where @xmath397 is defined in equation ( [ eq71a ] ) while the dotted curve represents our simple estimate ( [ eq73 ] ) , which does not take into account the effect of the resonances .
the dot - dashed , dot - double dashed and the long dashed curves are for @xmath453 , @xmath438 and @xmath439 , respectively.,width=340 ] on the plane @xmath489 , @xmath490 of the parameters of the problem .
arrows show the values of @xmath469 for each curve .
the dashed curve on the right hand side of the plot separates a region on the parameter plane , where a change of @xmath16 with @xmath136 is larger than 10 per cent of @xmath432 ( to the left of this curve ) from the region , where the black hole influence on the disc twist is relatively small and the change is less than 10 per cent ( to the right of the curve).,width=340 ] on @xmath136 along the curve in fig . [ figg6 ] , where @xmath491 .
the values of @xmath289 are equal to @xmath438 , @xmath439 , @xmath458 and @xmath459 for the solid , dashed , dot - dashed and dot - double dashed curves , respectively.,width=340 ] in fig .
[ figg6 ] we show the curves of constant values of the ratio of @xmath469 on the plane @xmath489 , @xmath490 of the parameters entering equation ( [ eq52 ] ) , for the case of @xmath7 and the value of @xmath289 in the range @xmath492 .
when this ratio is small the disc aligns with the black hole equatorial plane at small radii .
one can see that the area of small values of @xmath469 is situated at the upper left part of the plot .
the opposite case , where @xmath493 corresponds to a situation when the bardeen - petterson effect is absent . for the considered range of @xmath289 the area in the plot , where @xmath493 is larger than the area corresponding to the small values of this ratio .
the curves corresponding to the levels of @xmath494 exhibit oscillatory behaviour .
this effect is most probably determined by the resonant behaviour of @xmath469 at small values of @xmath0 considered above .
the dashed curve on the right hand side of the plot separates the regions on the parameter plane where the quantity @xmath495 is smaller / larger than @xmath450 , for all values of @xmath136 .
the latter condition is satisfied to the left of this curve , where the disc twist induced by the black hole is sufficiently large .
thus , the black hole significantly influences on the disc only when the value of @xmath289 is sufficiently small , @xmath496 depending on value of @xmath9 .
there is a curve in the plot corresponding to the level of @xmath491 .
the dependencies of @xmath497 with parameters @xmath0 and @xmath289 taken along this curve are shown in fig .
[ figg7 ] . as is seen from this fig .
when @xmath289 is sufficiently small the angle @xmath16 behaves in a non - monotonic way , though the curves are smooth and no oscillations of the inclination angle are observed .
such a behaviour looks realistic and it may have serious consequences for the disc structure itself since the inner parts of the disc strongly irradiate the outer part in such cases . also , it may have some interesting outcomes for spectra modelling , etc .
in this paper we derive dynamical equations describing evolution and stationary configurations of a fully relativistic thin twisted disc around a slowly rotating black hole .
we assume that the disc inclination angle @xmath498 , the black hole rotational parameter @xmath6 is small and that the background flat accretion disc is described by the simplest novikov - thorne model with a constant value of the shakura - sunyaev parameter @xmath0 . with these assumptions being
adopted a characteristic evolution time scale of such a disc is much larger than the dynamical time scale .
this helps to simplify our analysis to a great extent .
it is shown that the final dynamical equations may be formulated as equations for two dynamical complex variables @xmath291 describing the disc twist and warp and @xmath248 , which describes shear velocities induced in such a disc .
they may be brought to a form similar to what has been used for description of the classical newtonian twisted discs , see equations ( [ eq50 ] ) and ( [ eq51 ] ) . since all our variables are well defined in the black hole space - time ,
the results of integration of our dynamical equations may be directly applied to the problems connected with observations , e.g. obtaining time - dependant luminosity and spectrum of a twisted accretion disc , its shape as seen from large distances , etc .. we analyse stationary configurations of twisted discs in our model and show that they can be fully described by only two parameters - the viscosity parameter @xmath0 and by the parameter @xmath499 , where @xmath3 is a characteristic opening angle of a background flat accretion disc .
we consider in details the case of a very low viscosity twisted disc formally setting @xmath500 in equation ( [ eq52 ] ) describing the stationary configurations and develop an analytic approach to the problem of finding of these configurations for the most interesting case @xmath8 and @xmath7 , which is in excellent agreement with results of numerical integration of ( [ eq52 ] ) .
we find that the oscillations of the disc inclination angle found by ii in their model of a low viscosity twisted disc far from the black hole can be in a different regime in the relativistic region close to the last stable orbit .
namely , for the background flat disc models having singular profiles with the surface density being formally zero at @xmath10 as in the gas pressure dominated nt models the amplitude of disc oscillations and the radial frequency can grow indefinitely large when @xmath501 . additionally , in the formal limit @xmath234 there are `` resonant '' solutions occurring at some discrete values of @xmath487 , which have a finite value of the disc inclination angle at @xmath308 and the disc inclination tending to zero when @xmath266 .
effects of non - zero viscosity tend to smooth out the relativistic oscillations to an extent more prominent than was obtained in an oversimplified model of ii .
this may explain the results of @xcite who did nt find prominent oscillations of the disc inclination angle in their sph simulations of the bardeen - petterson effect . when viscosity parameter @xmath0 gets sufficiently large ( say , @xmath502 for @xmath433 ) the disc oscillations are not observed . instead of it
there is a growth of the inclination angle towards the last stable orbit , which is opposite to what is expected in the standard picture of the bardeen - petterson effect .
when @xmath0 gets even larger ( say , @xmath458 for @xmath433 ) the inclination angle decreases towards @xmath10 as in the standard picture . in general , there are three possible qualitatively different shapes of the twisted disc depending on values of the parameters of the problem .
\1 ) when @xmath0 is very small the disc inclination angle oscillates , with the oscillation amplitude growing dramatically towards @xmath10 .
when the disc inclination angle at large radii is not very small such a regime is probably unphysical . as we discuss in the text in such a case the shear velocities induced in the disc
can easily exceed the speed of sound .
this may either make the disc more turbulent thus effectively increasing the value of @xmath0 , which may , in its turn , damp the oscillations . on the other hand this can lead to presence of shocks in the disc close to @xmath10 .
the energy dissipated in the shocks may make the disc thicker .
this opens up a possibility of explaining of the presence of more compatible with observations thick discs close to @xmath10 by such a mechanism .
\2 ) when @xmath0 is moderately small the oscillations are absent .
however , the bardeen - petterson effect is absent as well . in such a regime
the disc inclination can either monotonically grow towards the last stable orbit or exhibit a non - monotonic behaviour decreasing with decrease of @xmath136 at large radii and increasing at smaller radii , see figs .
[ figg1 ] and [ figg7 ] .
since the corresponding curves are quite smooth we do not expect large shear velocities , which are proportional to the gradient of @xmath291 .
therefore , such a regime may be physically allowed .
this can lead to interesting possibilities of modifying the disc structure by strong irradiation of outer parts of the disc by inner parts , can be very important for modelling of spectra of the discs , etc ..
\3 ) finally , when @xmath0 is sufficiently large the bardeen - petterson effect is observed .
the region of the parameter plane , where such effect is possible is shown in fig .
[ figg6 ] , for the considered range @xmath503 .
although we consider the rotational parameter of the black hole as a small parameter of the problem we believe that the outlined picture remains qualitatively correct for moderate values of the rotational parameter as well , say when @xmath504 .
we discuss possible generalisations of our results . at first
one can solve time - dependant equations ( [ eq50 ] ) and ( [ eq51 ] ) to clarify dynamics of the twisted discs in the relativistic regime .
this will be done in a separate publication .
secondly , one can try to generalise the results to the case of large values of the rotational parameter , @xmath505 .
we expect that time - dependent equations will be rather cumbersome since in this case the evolution time scale will be of the order of the dynamical one and many terms neglected in our analysis must be taken into account .
the generalisation to the stationary case looks relatively straightforward .
however , it may happen that our twisted coordinate system based on the cylindrical coordinate system is less convenient when @xmath505 than a twisted coordinate system based on the spherical coordinates similar to what was introduced by @xcite for the classical newtonian twisted discs .
one may also try to include ( at least , at some phenomenological level ) the effects of back reaction on the background flat disc model determined by the large shear velocities induced in the twisted disc , and the disc self - irradiation . additionally , as we discuss in section 2.4 equations describing the flat disc models used in this paper are invalid very close to the last stable orbit , where the radial drift velocity gets larger than the sound speed , see equation ( [ eqmm4 ] ) .
one may try to use a slim disc model close to @xmath10 as a background model for our equations to overcome this difficulty .
finally , one may use the shape of the stationary configurations calculated in this paper to model different observed characteristics of particular sources .
we are grateful to n.i .
shakura for many helpful and fruitful discussions .
we also thank m. a. abramowicz and a.f .
illarionov for useful conversations , m. buck and m. h. moore for helpful comments .
vzh was supported in part by research and research / teaching staff of innovative russia for 2009 - 2013 years ( state contract no .
p 2552 on 23 november 2009 ) and in part by the grant rfbr-09 - 02 - 00032 .
pbi was supported in part by the dynasty foundation , in part by research and research / teaching staff of innovative russia for 2009 - 2013 years ( state contract no .
p 1336 on 2 september 2009 ) and in part by the grant rfbr-11 - 02 - 00244-a .
33 aliev a. n. , galtsov d. v. , 1981 , general relativity and gravitation , 13 , 899 bachev r. , 1999 , a&a , 348 , 71 bardeen j. m. , petterson j. a. , 1975 , apj , 195 , l65 cadez a. , brajnik m. , gomboc a. , calvani m. , fanton c. , 2003 , a&a , 403 , 29 caproni a. , mosquera cuesta h. j. , abraham z. , 2004 , apj , 616 , l99 caproni a. , abraham z. , livio m. , mosquera cuesta h. j. , 2007 , mnras , 379 , 135 demianski m. , ivanov p.b . , 1997 , a&a , 324 , 829 dexter j. , fragile , p. c. , 2011 , apj , in press , arxiv:1101.3783 ferreira b. t. , ogilvie g. i. , 2008 , mnras , 386 , 2297 ferreira b. t. , ogilvie g. i. , 2009 , mnras , 392 , 428 fragile , p. c. , blaes , o. m. , 2008 , apj , 687 , 757 hatchett s.p . ,
begelman m.c . , sarazin c.l . ,
apj , 247 , 677 ivanov p.b . ,
illarionov a.f . , 1997 , mnras , 285 , 394 ivanov p. b. , igumenshchev i.v . , novikov i.d . , 1998 ,
apj , 507 , 131 ivanov , p. b. , papaloizou , j. c. b. , 2008 , mnras , 384 , 123 ivanov p.b .
, papaloizou j. c. b. , polnarev , a. g. , 1999 , mnras , 307 , 79 kato s. , 1990 , pasj , 42 , 99 kumar s. , pringle j.e . , 1985 , mnras , 213 , 435 kumar s. , 1990 , mnras , 245 , 670 lubow s.h . , ogilvie g.i .
, pringle j.e . , 2002 , mnras , 337 , 706 martin r. g. , 2008 , mnras , 387 , 830 nelson , r. p. , papaloizou , j. c. b. , 2000 , mnras , 315 , 570 novikov i. d. , thorne k. s. , 1973 , black holes ( les astres occlus ) , 343 ogilvie g. i. , 1999 , mnras , 304 , 557 page d.n . , thorne k.s . , 1974 , apj , 191 , 499 papaloizou j.c.b .
, pringle j.e . , 1983 , mnras , 202 , 1181 papaloizou j.c.b
. , lin d.n.c . , 1995 ,
apj , 438 , 841 papaloizou j. c. b. , terquem c. , lin d. n. c. , 1998 , apj , 497 , 212 petterson j.a . , 1977 , apj , 214 , 550 petterson j.a . , 1978 , apj , 226 , 253 riffert h. , herold h. , 1995 , apj , 450 , 508 shakura n.i . , 1972 , azh , 49 , 921 shakura n.i . , sunyaev r.a . , 1973 , a&a , 24 , 337 wu , s .-
m . , chen , l. , yuan , f. , 2010 , mnras , 402 , 537 zeldovich , ya .
b. , novikov , i. d. , 1971 , _ relativistic astrophysics _
, vol 1 ( _ stars and relativity _ ) , p. 137
, the univ . of chicago press , chicago , london
the transformation law ( [ eq7 ] ) may be considered as a rotation to another cartesian coordinate system @xmath506 with help of the rotational matrix entering ( [ eq7 ] ) .
let us call this matrix as @xmath507 .
the one forms ( [ eq9]-[eq12 ] ) are related to the forms associated with the new coordinates @xmath508 as the forms ( [ eq6 ] ) are related to the forms associated with the old one @xmath509 .
explicitly , the transformation between the forms ( [ eq6 ] ) and the forms associated with the coordinates @xmath510 may be written as @xmath511 where @xmath512 are vectors having the corresponding one forms as their components , e.g. @xmath513 has the components @xmath514 , etc .. @xmath515 is a rotational matrix with components @xmath516 analogously , the transformation between the vector containing the forms ( [ eq9]-[eq12 ] ) , @xmath517 , and the vector containing the forms corresponding to the new cartesian system , @xmath518 , may be written as @xmath519 where the rotational matrix @xmath520 is obtained from @xmath248 by change of the angle : @xmath521 . accordingly , we have @xmath522 the rotational matrix @xmath523 has the components @xmath524 since the transformation ( [ an4 ] ) is given by an orthogonal matrix the transformation between the corresponding sets of adjoint basis vectors is the same .
the non - trivial connection coefficients are given by the following expressions : @xmath525 where @xmath526 .
+ other non - zero connection coefficients are obtained with help of antisymmetry over the first two indices : @xmath527 .
as we discuss in section [ twist ] our subset of equations describing the disc twist follows from equations ( [ eq19 ] ) after the procedure outlined in this section .
the @xmath45-component of ( [ eq19 ] ) gives @xmath528 where @xmath529 and the @xmath1 , @xmath80 and @xmath79-components give , respectively , @xmath530 v^\varphi + { 1\over \rho}f^{r}_{\nu } = \\ w r \frac{\partial_\xi p}{\rho } - a\xi\frac{k_3 ^ 2}{k_1 k_2 } \left ( \frac{k_1}{k_3 } \right ) ^\prime z u^\tau u^\varphi , \label{2a2}\end{gathered}\ ] ] @xmath531 and @xmath532 + { r\over \rho}f^{\xi}_{\nu } = \\ -\left [ \frac{k_2}{k_1 } \partial_\varphi u - 2a\frac{k_3 z}{k_2 } + a\frac{\xi^2}{r } \frac{k_3 ^ 2 z}{k_1 k_2 } \left ( \frac{k_1}{k_3 } \right ) ^\prime \right ]
r u^\tau u^\varphi + ar^2\frac{k_3}{k_1 } ( u^\varphi)^2 \partial_\varphi u. \label{2a4}\end{gathered}\ ] ] where @xmath533 are determined by the presence of viscous interactions in the disc .
since these interactions lead to a non - zero value of the drift component , @xmath534 , we include terms containing @xmath534 in @xmath533 excepting the terms proportional to @xmath95 through the dependence of @xmath188 on @xmath95 , see equation ( [ eq34 ] ) to keep the notation uniform . explicitly ,
we have from ( [ eq19 ] ) @xmath535 and @xmath536 where @xmath537 . @xmath538 are components of the viscous stress tensor given by equations ( [ eq20 ] ) and ( [ eq21 ] ) relevant for our purposes : @xmath539 where we use equations ( [ eq20 ] ) and ( [ eq21 ] ) .
note that we take into account all explicit terms of zero and first order in @xmath6 in equations ( [ 2a1]-[2a7 ] ) . as we discuss in the main text , in fact , the only term linear in @xmath6 , which is important for our purposes , is the gravitomagnetic term - the second term in the square brackets on the right hand side of ( [ 2a4 ] ) , see equation ( [ 2a12 ] ) below .
other terms may be important in other studies and are shown for a reference .
the background quantities @xmath369 , @xmath86 and @xmath540 can also be easily developed up to the linear in @xmath6 order , and , for completeness , we show them below .
we have @xmath541 where @xmath542 and @xmath543 are given by ( [ eq23 ] ) , and @xmath544=\frac{(u^\varphi)^2}{r^2 } \frac{\omega_\perp^2}{\omega^2}\xi , \label{2a9}\ ] ] where @xmath545 we remind that @xmath546 and the frequency of the lense - thirring precession ( see e.g. aliev @xmath547 galtsov 1981 , kato 1990 ) , @xmath265 , is given by the expression @xmath548 it is important to note that from equation ( [ 2a4 ] ) it follows that this characteristic frequency determines the disc s evolution in a certain regime . in order to show it
let us temporarily assume that we consider a pressureless , inviscid disc with gas particles moving on circular orbits around a black hole . in this case we can set @xmath549 in ( [ 2a4 ] ) .
the last two terms on the right hand side can be also neglected since they contain products of two small parameters : @xmath550 and @xmath551 , respectively .
accordingly , we obtain from ( [ 2a4 ] ) @xmath552 remembering that @xmath553 we see that in this case the disc rings precess with the precession frequency equal to @xmath265 .
the space - time of a schwarzschild black hole is symmetric with respect to transformations belonging to the three - dimensional group of rotations , @xmath554 .
accordingly , equations of motion contain three conserved quantities , which may be identified with three components of angular momentum .
conservation of the components perpendicular to the axis @xmath318 requires a special divergent form of equation ( [ eq38 ] ) ( or ( [ eq51 ] ) ) . in order to see that this equation has , indeed , such a form let us represent the term on the right hand side as @xmath555 and the quantity @xmath556 as @xmath557 , where @xmath364 and @xmath558 are functions of @xmath1 and @xmath559 . introducing the complex notation used above we rewrite ( [ eq38 ] ) as @xmath560 where @xmath561 .
we multiply ( [ c1 ] ) by @xmath562 and integrate over @xmath1 from some radius @xmath563 to @xmath564 .
we get @xmath565 where @xmath566 and @xmath567 it can be shown that @xmath568 is proportional to a linear combination of @xmath139 and @xmath569-components of the disc angular momentum within the region @xmath570 , and @xmath571 is the corresponding torque term provided by the gravitomagnetic force . the term @xmath572 can be transformed with help of equation reflecting the law of conservation of angular momentum for a background disc model @xmath573 where @xmath574 is flux of radiation from the disc surface . ) written for a flat disc or can be derived from the results of @xcite and rh .
] we integrate ( [ c4 ] ) by parts and use ( [ c5 ] ) to get @xmath575 equation ( [ c6 ] ) shows that when the black hole is not rotating ( @xmath576 ) a change of the angular momentum content within a region of the disc is determined by fluxes of angular momentum through the region inner and outer boundaries and angular momentum flux carried by radiation emitted from the disc surface .
thus , equation ( [ eq38 ] ) has the required divergent form . | in this paper we derive equations describing dynamics and stationary configurations of a twisted fully relativistic thin accretion disc around a slowly rotating black hole .
we assume that the inclination angle of the disc is small and that the standard relativistic generalisation of the @xmath0 model of accretion discs is valid when the disc is flat .
we find that similarly to the case of non - relativistic twisted discs the disc dynamics and stationary shapes can be determined by a pair of equations formulated for two complex variables describing orientation of the disc rings and velocity perturbations induced by the twist .
we analyse analytically and numerically the shapes of stationary twisted configurations of accretion discs having non - zero inclinations with respect to the black hole equatorial plane at large distances @xmath1 from the black hole .
it is shown that the stationary configurations depend on two parameters - the viscosity parameter @xmath0 and the parameter @xmath2 , where @xmath3 is the opening angle ( @xmath4 , where @xmath5 is the disc half thickness and @xmath1 is large ) of a flat disc and @xmath6 is the black hole rotational parameter . when @xmath7 and @xmath8 the shapes depend drastically on value of @xmath9 .
when @xmath0 is small the disc inclination angle oscillates with radius with amplitude and radial frequency of the oscillations dramatically increasing towards the last stable orbit , @xmath10 .
when @xmath0 has a moderately small value the oscillations do not take place but the disc does not align with the equatorial plane at small radii . the disc inclination angle is either increasing towards @xmath10 or exhibits a non - monotonic dependence on the radial coordinate . finally , when @xmath0 is sufficiently large the disc aligns with the equatorial plane at small radii .
when @xmath11 the disc aligns with the equatorial plane for all values of @xmath0 .
the results reported here may have implications for determining structure and variability of accretion discs close to @xmath10 as well as for modelling of emission spectra coming from different sources , which are supposed to contain black holes .
[ firstpage ] accretion , accretion discs ; hydrodynamics ; black hole physics ; relativity ; binaries : close ; galaxies : nuclei ; celestial mechanics |
with the detection and first characterization of the higgs particle @xcite the standard model of physics has seemingly become complete .
a full description in three families of fundamental material particles behaving as fermions , acting upon each other through three fundamental forces via bosonic force - carrying particles , has been formulated in a consistent framework . however , a number of physical phenomena are not contained in the standard model ( sm ) : dark matter and dark energy are not explained ; a connection of the three fundamental sm - forces to gravity can not be made ; and neither is it understood why gravity is such a feeble force .
deeper questions as to why the universe is built in 3 + 1 dimensions , why there exist only 4 forces , and whether the constants of nature may depend on cosmological history or on local space - time conditions remain unanswered as of yet .
it is the paradigm of the present paper that molecular physics may contribute to finding answers to these questions .
the recent searches for a dipole moment of the electron , setting tight constraints of @xmath9 e@xmath10 cm , have already ruled out some supersymmetric extensions to the standard model .
these studies take advantage of the very strong internal polarizing fields of @xmath11 v / cm in ybf @xcite and tho @xcite molecules as well as of hff@xmath7 ions @xcite . at the most fundamental level laser spectroscopic investigations of molecules
can be applied for an experimental test of the symmetrization postulate of quantum mechanics .
the fact that integer - spin species must exhibit a symmetric wave function upon interchange of identical particles makes that certain rotational quantum states in the @xmath12o@xmath0 @xcite or in the @xmath13c@xmath12o@xmath0 @xcite molecule can not exist .
this provides an experimental platform for the investigation of the symmetrization postulate .
a search for varying constants implies a search for physics beyond the sm .
the fact that the fundamental coupling strengths and other constants such as the mass ratios between particles are inserted as parameters in the fundamental theory suggests that these fundamental constants can be varied at will
. however , a time - variation or a space - time dependence of these constants should be phrased in field theories , and in order for those to remain consistent , even in the simplest approaches , additional fields have to be invoked @xcite .
such fields can also be associated with fifth forces or quintessence @xcite .
the hydrogen molecule is an experimental probe for detecting a possible variation of the proton - electron mass ratio @xmath1 , a dimensionless constant of nature .
the strong dipole - allowed spectrum of the lyman and werner bands in the vacuum ultraviolet spectral range provides a sensitive tool for detecting h@xmath0 .
observation of h@xmath0 absorption spectra , redshifted into the atmospheric transmission window ( @xmath14 nm ) with ground - based telescopes constrain putative variations of @xmath5 over time .
since h@xmath0 is the most abundant molecule in the universe , it is readily observed even up to redshifts as high as @xmath15 @xcite providing look - back times of 12.5 billion years into cosmic history .
alternatively , h@xmath0 absorption spectra as observed from photospheres of white dwarf stars can be employed to probe whether this constant @xmath5 depends on special local conditions , such as e.g. gravitational fields .
another approach toward probing new physics from molecules is to carry out ultra - precise measurements on level energies of molecules under well - controlled laboratory conditions .
a historical inspiration is drawn from the discovery of the lamb shift in atomic hydrogen @xcite , where a measurement of a small shift in the relative energies between @xmath16 and @xmath17 levels led to the birth of quantum electrodynamics ( qed ) , the modern theory describing the interaction between light and matter . in a similar fashion qed - calculations of small molecular systems
can be tested in precision laser spectroscopic experiments to search for phenomena beyond qed .
any deviations from first principles calculations in the framework of qed can be interpreted in terms of physics beyond the standard model , in terms of fifth forces or quintessence , in terms of extra dimensions , or as any other phenomenon beyond the standard model . here
the status of such searches for new physics from spectroscopy of hydrogen molecules and molecular ions will be discussed .
the remainder of the paper is structured as follows . in sections [ muvar ] and [ mu - grav ]
the use of the h@xmath0 spectrum for a search for a varying proton - electron mass ratio is discussed , either on a cosmological time scale or as a local variability with dependence on a gravitational field . in section [ h2metrology ] a number of recent precision measurements on the h@xmath0 molecule , including its dissociation limit , its fundamental ground state vibrational splitting , as well as level energies of a sequence of rotational states and a high vibrational level , are presented . in section [ h2ionmetrology ] some recent precision measurements on the h@xmath8/hd@xmath7 molecular ions are discussed .
these outcomes of experimental precision studies are compared with the most advanced qed - theoretical calculations for these benchmark molecules in section [ qed - theory ] .
the agreement found between the experiments and theory is interpreted , in section [ bsm - physics ] , in terms of setting bounds on a fifth force and on extra dimensions , both concepts beyond the standard model of physics . in the final section [ outlook ]
an outlook is presented on perspectives to perform more sensitive searches for new physics in these areas .
searches for temporal variations of the proton - electron mass ratio @xmath1 can be conducted by comparing absorption spectra of molecules observed astronomically from high redshift objects with the same spectra measured in the laboratory .
such comparisons can be made for many molecules @xcite , but molecular hydrogen is often a species of choice since it is the most abundant molecule in the universe , and readily observed at medium to high redshifts .
the relative variation of @xmath5 can be deduced by imposing the following relation to the obtained redshifted wavelengths @xmath18 and laboratory ( zero redshift ) wavelengths @xmath19 : @xmath20 here the redshift parameter @xmath21 relates to the overall redshift of the absorbing hydrogen cloud , and the parameters @xmath22 represent the sensitivity coefficients expressing the induced wavelength shift upon a varying @xmath5 for each individual line in the absorption spectrum of h@xmath0 @xcite : @xmath23 the laboratory wavelengths @xmath19 are determined in laser - based experiments of the @xmath24 lyman and @xmath25 werner bands of h@xmath0 via extreme ultraviolet laser spectroscopy @xcite achieving accuracies of @xmath26 . even better accuracies were obtained in a study measuring anchor level energies in two - photon doppler - free excitation of the @xmath27 system @xcite , combined with results of a comprehensive fourier - transform emission study delivering relative values for rovibrational levels in @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , and @xmath36 states @xcite .
this two - step process yielded accuracies of @xmath37 @xcite for a number of lyman and werner band lines . for the purpose of comparing with quasar absorption spectra the laboratory values for the absorption wavelengths may be considered exact .
the overall redshift @xmath21 is , within a good approximation , connected to a look - back time @xmath38 into the cosmological history of the universe @xmath39\ ] ] where @xmath40 gyrs is the age of the universe .
a possible small differential effect on the absorption spectrum , due to a varying proton - electron mass ratio and represented by a relative change @xmath41 , may be derived from the spectrum in a comprehensive fit .
in such a fit the molecular physics information on the wavelengths @xmath42 , the oscillator strength @xmath43 , and the natural lifetime broadening parameters @xmath44 for each h@xmath0 line is fixed , the values for the sensitivity coefficients @xmath22 inserted , and the physical parameters describing the absorbing medium , such as the doppler width @xmath45 , the column densities for each @xmath46-level @xmath47 and the redshift @xmath21 are fitted to model the spectrum . from the resulting values of @xmath47
it is found that typical population temperatures of 50 k are found in most cases , although the populations do not exactly represent a boltzmann distribution . in all cases only absorption from the lowest @xmath48 vibrational level is observed for rotational states @xmath49 .
hence the dipole allowed absorption spectrum lies shortward of 1120 and extends to the lyman cutoff of 912 , thus covering the @xmath24 lyman bands from ( 0,0 ) up to ( 17,0 ) and the @xmath25 werner bands up to ( 4,0 ) . in several cases absorption lines of deuterated hydrogen hd
are observed and included in the analysis , by comparing to accurate laboratory wavelengths @xcite and invoking calculated sensitivity coefficients @xcite .
for hd only r(0 ) lines are observed . due to the opacity of the earth s atmosphere ( transparency for @xmath14 nm ) only for systems of redshift @xmath50 a significant number of absorption lines
can be observed with ground based telescopes . in fig .
[ quasarspec ] six different h@xmath0 absorption spectra at redshifts in the range @xmath51 are displayed as observed in the line of sight of quasars .
each system displays a number of distinct velocity features , ranging from a single feature as in b0642 - 5038 to seven distinct features as in q2348 - 011 . in fig .
[ quasarspec ] information about quasar data is compiled from original analyses for quasars j2123 - 0050 at @xmath52 @xcite , for q2348 - 011 at @xmath53 @xcite , for b0642 - 5038 at @xmath54 @xcite , for j1237 + 0647 at @xmath55 @xcite , q0528 - 250 at @xmath56 @xcite , and for j1443 + 2724 at @xmath57 @xcite
. in some panels , blends with typical lyman-@xmath58 forest lines are seen , like for the r(1 ) line in b0642 - 5038 and in j1443 + 2724 .
such features have to be included in the fitting procedure . in a final analysis of each individual spectrum
the parameter @xmath41 is included as a fitting parameter and determined .
the uncertainty in the relative variation of this fundamental constant is obtained from the covariance matrix . after having performed the comprehensive fitting analysis of the six systems presented in fig .
[ quasarspec ] , and the additional works on systems he0027 - 1836 @xcite , q0347 - 383 @xcite , q0405 - 443 @xcite , and q1232 + 082 @xcite an overall average is obtained of @xmath59 in the redshift interval @xmath51 , so for look - back times of 10.5 - 12.4 gyrs .
this finding of a positive value for @xmath41 at the 1.9@xmath4 significance level is too small to be interpreted as evidence for a variation of a fundamental constant , and is interpreted as a null result : a constraint on a variation of the proton - electron mass ratio of @xmath60 ( at 3@xmath4 ) .
the spectrum of molecular hydrogen can also be employed to probe a possible dependence of the proton - electron mass ratio @xmath5 on environmental conditions , connecting to the chameleon scenario for varying constants @xcite . in the photosphere near the surface of collapsed stars ,
also known as white dwarfs , the gravitational field is some 10@xmath61 times stronger than at the earth s surface .
hence the absorption spectra of h@xmath0 , in rare cases observed from such photospheres @xcite , transfer a fingerprint of the value of @xmath5 existing in extreme gravitational field conditions @xcite .
[ gd133 ] displays part of the absorption spectrum of the white dwarf gd133 as observed with the cosmic origins spectrograph aboard the hubble space telescope .
the spectrum consists of a large number of blended absorption lines in the range 1310 - 1410 , which can be assigned to absorptions in lyman bands @xmath24 @xmath62 bearing the largest intensities in bands ( 0,3 ) , ( 0,4 ) , and ( 1,3 ) . the spectrum of this white dwarf is observed in our galaxy , and is therefore not affected by a cosmological redshift , but the gravitational potential induces a gravitational redshift to the spectrum of @xmath63 , where @xmath64 is newton s constant , @xmath65 the speed of light , and @xmath66 and @xmath67 the mass and the radius of the white dwarf star .
the experimental spectrum is compared to a spectral model function @xmath68 for h@xmath0 described as : @xmath69 with @xmath70 a total column density of absorbing h@xmath0 , @xmath71 the partition function representing the quantum state populations as a function of temperature and including a 3:1 ortho - para distribution , and @xmath72 the oscillator strengths for individual transitions , which were calculated with _ ab initio _ methods @xcite .
this spectral function was implemented into a fitting routine , where for each spectral line a lorentzian width @xmath44 corresponding to the excited state lifetime was convolved with a doppler width @xmath45 . in the fit @xmath70 , @xmath45 , the temperature @xmath38 ,
the gravitational redshift @xmath73 and @xmath41 were optimized .
a good reproduction of the gd-133 spectrum was obtained for a temperature @xmath74 k , a doppler shift @xmath75 km / s , a column density @xmath76 , and a redshift @xmath77 @xcite . the high temperature explains why the populated vibrational levels @xmath78 and @xmath79 contribute to the spectrum and why so many h@xmath0 absorption lines lend intensity to the spectrum . in the modeling of the gd-133 spectrum over a thousand h@xmath0 lines are included , represented by the sticks in fig .
[ gd133 ] .
the comprehensive fit of the gd-133 white dwarf spectrum returns a value of @xmath80 delivering a constraint on a dependence of the proton - electron mass ratio on a gravitational potential of @xmath81 , which may be compared with the gravitational potential at the earth s surface of @xmath82 .
note that here the dimensionless gravitational potential is defined as @xmath83 .
an analysis has been performed for another white dwarf gd29 - 38 yielding a similar result @xcite .
in recent years great progress is made in both the experimental and theoretical determination of level energies of rovibrational quantum states in the @xmath84 electronic ground state of the hydrogen molecule and its isotopomers . in a combined effort by the zrich and amsterdam laser precision spectroscopy teams
the ionization potential ( ip ) of the h@xmath0 molecule has been measured following scheme i as illustrated in fig .
[ h2scheme ] .
this work is based on the concept that an accurate value of the dissociation limit can be derived from a measurement of the ip , as was previously shown @xcite .
the dissociation limit is a well - calculable benchmark property of the canonical molecular quantum system .
scheme i in fig .
[ h2scheme ] represents a combination of three separate measurements to determine the ip .
firstly , measurements were performed on the largest energy splitting between of the @xmath27 system , via deep - uv doppler - free two - photon laser spectroscopy @xcite .
secondly , the frequency gap between the @xmath85 state and a @xmath86 rydberg state was measured in single - photon excitation in a molecular beam at 397 nm @xcite .
thirdly , a mm - wave experiment using radiation at wavelengths of 0.23 mm was performed to measure level splittings between rydberg states and extrapolate to the ip using multi - channel quantum defect theory @xcite .
the combination of these three independent measurements yields an experimental value for the ip of h@xmath0 @xcite . from a measurement of the ip - limit e@xmath87(h@xmath0 )
a value of the dissociation limit d@xmath88(h@xmath0 ) can be derived through a thermodynamic cycle @xmath89 as is also illustrated graphically in fig .
[ h2scheme]b .
the dissociation energy of the molecular ion d@xmath88(h@xmath8 ) can be calculated to high accuracy @xcite , while an accurate value for the atomic ionization potential e@xmath87(h ) is known from experiment and theory @xcite . by this three - step sequence of precision measurements
the ionization potentials and dissociation limits were determined for the stable isotopomers of neutral hydrogen , h@xmath0 @xcite , d@xmath0 @xcite , and hd @xcite at an accuracy of @xmath90 @xmath91 .
results of these experimental studies are listed in table [ data ] .
an important benchmark for a precision study in neutral molecules is the measurement of the fundamental vibration of the hydrogen molecule , i.e. the energy splitting between @xmath92 and @xmath93 levels in the ground electronic state .
experimental studies thereof had been performed in the past via the very weak quadrupole absorption spectrum in the infrared range @xcite or via the raman spectrum @xcite .
both approaches suffer from the effect of doppler broadening , while the infrared measurement is also affected by collisional broadening and shifts , since the weak absorptions can only be observed at elevated pressures .
a novel approach was undertaken to measure the vibrational splitting @xmath94 via the combination difference in electronic transitions @xmath27 for ( 0,0 ) and ( 0,1 ) bands .
the excitation schemes , displayed as i and ii in fig .
[ h2scheme]a , sharing a common excited state in the @xmath85 ( @xmath92 ) level , were performed with two - photon excitation at @xmath95 nm ( from the @xmath96 @xmath92 ground state ) and at @xmath97 nm ( from the @xmath96 @xmath93 ground state , prepared in a discharge pulsed beam source ) .
the measurements were carried out in the collision - free conditions of a molecular beam , while the doppler broadening was fully suppressed in a two - beam counter - propagating arrangement @xcite . by this
means linewidths of 36 mhz , much narrower than obtainable in the direct quadrupole transitions , were achieved .
values for the rotationless @xmath98 fundamental vibration for h@xmath0 , d@xmath0 and hd were determined at an accuracy of @xmath99 @xmath91 @xcite and listed in table [ data ] .
these splittings were also measured for @xmath100 and @xmath101 for the three isotopomers at the same accuracy @xcite .
additional experiments have been performed to determine precise level energies of either rotationally or vibrationally excited quantum states in the @xmath84 ground state potential of h@xmath0 , again involving two - photon excitation to the @xmath85 state . for use in various schemes the rovibrational levels of this first electronically excited @xmath85 state , associated with a double - well potential , were calibrated in a preparatory study . as discussed above in section [ muvar ] ,
results from fourier - transform emission @xcite were combined with the absolute frequency calibration of the lowest para and ortho levels in the @xmath85 state from the two - photon uv measurements as in scheme i @xcite .
this combination of methods yields absolute level energies of a large manifold of rovibrational levels of @xmath85 with accuracies in some cases approaching @xmath102 @xmath91 .
these @xmath85 ( @xmath103 ) levels can be used as anchor levels to determine binding energies of @xmath96 ( @xmath103 ) levels as shown in the following two examples . [ cols="<,^,>,^,>,>,>",options="header " , ] + @xmath104value only for @xmath105 , larger uncertainties for @xmath106 and @xmath107 .
+ @xmath108average taken over four measurements for differing rotational states @xmath109 .
+ @xmath110averages taken over the rotational levels involved .
one application is the precise measurement of highly excited rotational levels in the @xmath92 ground state manifold of h@xmath0 @xmath96 @xcite .
the h@xmath0 molecular quantum states of high rotational quantum number were prepared via an ultraviolet photolysis reaction @xcite : @xmath111 and a subsequent chemical reaction : @xmath112 thus giving access to states of up to @xmath113 . by making use of the calibrated @xmath85 ( @xmath114 ) anchor levels and exciting these via two - photon doppler - free spectroscopy , the binding energies of these high rotational states could be experimentally determined at accuracies of 0.005 @xmath91 @xcite . in the most recent experimental study highly - excited vibrational levels in h@xmath0
were investigated lying only some 2000 @xmath91 below the dissociation threshold @xcite .
production of vibrational states in @xmath115 was accomplished via two - photon uv - photolysis of h@xmath0s @xcite .
a separate narrowband uv - laser was employed to perform the spectroscopy , displayed as scheme iii in fig .
[ h2scheme ] .
again two - photon doppler - free spectroscopy with counter - propagating beams was applied , now exciting the @xmath85 @xmath116 levels in h@xmath0 , corresponding to the third level @xmath117 in the outer well of the @xmath85 double - well potential , for which level energies had been calibrated @xcite .
this spectroscopic experiment led to values for the binding energies of levels @xmath118 at an accuracy of @xmath119 @xmath91 @xcite .
this accuracy is somewhat lower than for the vibrational ground tone measurements , due to the rather large ac - stark effects induced in the experiments by the intense laser pulses required when probing the low densities of prepared h@xmath0(@xmath120 ) states .
the data on the experiments discussed are presented in table [ data ] , which also includes the results of direct measurements of quadrupole transitions in the hydrogen isotopomers , in particular on the ( 2,0 ) overtone in h@xmath0 @xcite , the ( 3,0 ) overtone in h@xmath0 @xcite and the ( 2,0 ) overtone in d@xmath0 @xcite .
for convenience of comparison table [ data ] lists values in units of mhz .
the hydrogen molecular ion is an alternative benchmark system for testing first principles qed calculations in experiment @xcite .
being a charged particle it can be subjected to electromagnetic confinement in an ion trap , which has been applied in particular in studies of the hd@xmath7 ion @xcite .
such measurement conditions bear the advantage that extended excitation times can be realized , coping with the weak dipole polarizabilities even in high - order overtone transitions .
moreover , the ions can be sympathetically laser - cooled by adding be@xmath7-ions inside the ion trap , thus reaching mk temperatures at which the doppler broadening is greatly reduced @xcite .
the laser cooling of two - species ionic ensembles can be pushed to produce coulomb crystals in which the lighter hd@xmath7 particles are confined to the axis of the trap where the electromagnetic fields are zero .
the position of the different ions inside a linear paul trap can be monitored on - line by recording the be@xmath7 fluorescence with an imaging camera as is shown in fig .
[ cloud ] , which was produced with our hd@xmath7 ion trap experiment in amsterdam @xcite .
the detection of the hd@xmath7 resonance is accomplished by measuring the loss of hd@xmath7 ions from the trap in a resonance - enhanced multi - photon dissociation ( rempd ) scheme as illustrated in fig [ hd+ ] .
the hd@xmath7 ions , excited to the @xmath121 level by a stabilized spectroscopy laser at 782 nm , are dissociated by a second laser at @xmath122 nm @xcite . for probing the @xmath123 intermediate level
the combination 1442 nm/266 nm can be employed @xcite .
the loss of deuterated hydrogen ions can be quantified by a radio - frequent excitation of the coulomb crystal ( secular excitation ) , therewith liquifying the crystal structure . under these conditions
the intensity of the photons scattered for laser cooling the be@xmath7-ions will increase proportionally to the number of hd@xmath7 ions , thus delivering the fluorescence signal to monitor the spectroscopic transition .
over 50 unresolved hyperfine components contributing to the transition are modeled into a lineshape function , based on the known hyperfine structure for hd@xmath7 @xcite . analyzing a variety of systematic effects , such as the ac and dc stark effects on the transition frequency @xcite , the redistribution of rotational sub - states due to black - body radiation in molecular dynamics simulations , and the analysis of hyperfine line strengths at the magnetic sub - state level
, the experimental result on the r(2 ) line of the ( 8,0 ) band of hd@xmath7 yields a transitions frequency of @xmath124 mhz @xcite .
this result and previous results on transitions in the ( 4,0 ) @xcite and ( 1,0 ) @xcite bands are included in table [ data ] .
attempts to measure transitions in the ( 6,0 ) band have not yet resulted in accurate values @xcite .
work is in progress to measure the spectrum of h@xmath8 in an ion trap @xcite , which is more difficult than probing hd@xmath7 where electric dipole transitions are allowed , unlike for the homo - nuclear species .
alternatively , the spectroscopy of the h@xmath8 ion can be approached through measurements on the rydberg states of the neutral hydrogen molecule , as had been done in the 1990s in measuring the hyperfine structure @xcite .
the zrich group applied laser excitation of rydberg states of neutral h@xmath0 and subsequent millimeter ( mm)-wave radiation to probe the level structure of the rydberg manifolds at principal quantum numbers at @xmath125 @xcite .
accurate values for the binding energies of the neutral states could be determined and by application of multi - channel quantum defect theory an extrapolation could be made for @xmath126 , yielding results on ionic states , for all three isotopomers @xcite .
remarkably , in these studies the complex hyperfine structure of the ionic cores was better resolved than in the direct measurements on the ions , due to the smaller doppler broadening achievable in mm - wave spectroscopy .
recently , this technique was implemented to measure the @xmath127 rotational level splitting in para - h@xmath8 @xcite yielding the accurate value of @xmath128 mhz included in table [ data ] .
simultaneously with the reports on improved measurements of the dissociation limits of the neutral hydrogen isotopomers h@xmath0 , hd , and d@xmath0 , novel _ ab initio _ calculations on the rovibratonal levels of these species were performed in the framework of non - relativistic quantum electrodynamics and an evaluation in orders of the electromagnetic coupling constant @xmath58 : @xmath129 the non - relativistic energy @xmath130 is calculated via an accurate born - oppenheimer potential @xcite , where adiabatic and non - adiabatic corrections are calculated in a perturbative approach yielding non - relativistic binding energies for all rovibrational levels @xcite .
for the first time accurate relativistic corrections of order @xmath131 and qed - contributions of order @xmath132 were calculated , while an estimate of qed - contributions of order @xmath133 was made , thus leading to an accurate value for the dissociation limits for h@xmath0 and d@xmath0 @xcite , and a full set of binding energies of all @xmath134 rovibrational levels in the @xmath84 electronic ground state for h@xmath0 and d@xmath0 @xcite and for hd @xcite .
typical accuracies for the binding energies are @xmath135 @xmath91 with some variation over the @xmath134 set .
these calculations now stand as the benchmark for comparison with experiments as test of qed in molecular systems .
they constitute an order of magnitude improvement over the previous reference values for the binding energies of ( @xmath103 ) rovibrational levels by wolniewicz @xcite .
it is for the first time that qed - phenomena in neutral molecules are explicitly calculated , leading to a precision competitive with experiment . in fig .
[ energy_contrib ] the values for the correction terms to the born - oppenheimer energies , the adiabatic and non - adiabatic as well as the combined relativistic and qed corrections are plotted .
also included in the figure are the uncertainties as currently obtained from first principles calculations . all values in fig .
[ energy_contrib ] are restricted to @xmath98 states .
the latter values provide insight into approaches to test the qed calculations , with the largest uncertainties remaining for vibrational levels @xmath136 .
results from the calculations are also included in table [ data ] .
the uncertainties obtained are listed as theoretical uncertainties @xmath137 and the deviations between experiment and theory are included as @xmath138 .
uncertainties from theory and experiment are added in quadrature to obtain values : @xmath139 which are considered as the 1@xmath4 combined uncertainties for testing theory .
comparison between the combined uncertainties @xmath140 and the deviations @xmath138 provides a test of qed calculations in these neutral molecular systems . for all cases
it is found that @xmath141 , except for the ground tone vibration in h@xmath0 , where the deviation is just slightly over 1@xmath4 @xcite .
for the four measurements on level energies in @xmath115 of h@xmath0 @xcite the deviation @xmath138 is averaged and the result is in agreement with qed .
this leads us to conclude that qed - theory is found in agreement with the precision measurements on h@xmath0 , hd and d@xmath0 , within the uncertainties as listed .
this holds for the @xmath98 levels as represented in table [ data ] , but also for the rotational states up to @xmath113 @xcite .
it is noted here that the hyperfine structure in the neutral molecules is neglected in this treatment . from the measurements on the hyperfine structure performed by ramsey and coworkers in the 1950s
it was established that the hyperfine splittings for the @xmath100 ortho - levels in h@xmath0 is below @xmath142 mhz @xcite , while it is even smaller in the deuterated molecules because of the smaller nuclear @xmath143-factor of the deuteron , with respect to that of the proton .
the para - levels in h@xmath0 do not exhibit hyperfine structure .
it is therefore safe to assume that hyperfine structure does not play a role for qed - tests in the neutral hydrogen molecules , at the present level of accuracy . for the hd@xmath7 molecular ion
the calculations of moss had long stood as the most accurate @xcite .
these calculations did not include hyperfine structure and they were performed in the framework of the born - oppenheimer approximation . in recent years , the hyperfine structure of the molecular hydrogen ions was revisited in a series of theoretical works @xcite .
the highest precision so far has been achieved for h@xmath8 , and excellent agreement with measured hyperfine intervals @xcite was found @xcite .
this gives confidence that the hyperfine structure in hd@xmath7 is understood as well .
alternatively , the hd@xmath7 ion was considered as a three - body coulomb system for which variational calculations were performed , bypassing the born - oppenheimer approximation @xcite .
subsequently leading - order relativistic and radiative corrections were computed @xcite , as well as higher order qed - terms @xmath144 @xcite and @xmath145 @xcite , leading to transition frequencies for hd@xmath7 below the 0.1 ppb - level @xcite .
this makes these calculations more accurate than those for the neutral hydrogen four - particle molecules where electron correlation plays a complicating role .
results from calculations for transitions between rovibrational states in hd@xmath7 are included in table [ data ] and compared to observations , again deducing @xmath140 and @xmath138 values . while the experimental and theoretical findings for the ( 4,0 ) @xcite and ( 8,0 ) @xcite bands produce agreement within @xmath146 , the result of the ( 1,0 ) band shows a deviation at the @xmath147 level .
the discrepancy is considered insufficient to call for a disagreement with qed and is treated as an outlier within the full data set of experimental level energies to be compared with theory .
in the previous section a number of recent precision measurements on level energies and level splittings in the neutral and ionic hydrogen molecules has led to the conclusion that measurements are in agreement with calculations in the framework of qed .
for all entries ( except that for the ( 1,0 ) transition in hd@xmath7 ) in table [ data ] the relation @xmath141 was verified .
the @xmath140 values may serve as input for the search for physics beyond the standard model ( bsm ) , which can be cast in terms of mathematical tests in which expectation values @xmath148 of quantum mechanical operators relating to new physical phenomena can be calculated and compared with the existing constraints on qed .
this concept was recently applied to two possible extensions of the standard model , one relating to the possible existence of a fifth force acting between hadronic particles within molecules @xcite and one related to the possible existence of extra dimensions @xcite . constraints on bsm - physics are then given by the inequality relation : @xmath149 in molecular systems gravity is very weak , as the gravitational field between two protons is weaker than the electromagnetic field by @xmath150 .
the strong interaction is confined to femtometer - scale distances and therefore negligible at the length scale of a chemical bond in molecules .
this nuclear force exhibits an influence on the shape of the electronic wavefunction and hyperfine structure through nuclear finite - size effects .
these contributions are well known and readily taken into account . due to the smallness of these effects , the discrepancy between proton size values obtained from different methods ( the proton size puzzle @xcite ) does not play a significant role in the present work .
the weak interaction is also extremely small , below the hz level , in the hydrogen atom @xcite , and by proxy it is assumed to be similarly small in the lightest molecules h@xmath0 and h@xmath8 .
this implies that experimental tests of transition frequencies and level energies as calculated by qed in fact represent tests of the standard of model of physics ; only the electromagnetic interaction is to be considered when describing the structure of molecules .
the occurrence of a hypothetical fifth force can be expressed in terms of a yukawa potential of the form @xcite : @xmath151 where @xmath152 is the coupling strength of this force , in analogy with the fine structure constant @xmath153 , and @xmath154 is the range of the force . in a quantum field approach
this force is then mediated by the exchange of force - carrying bosonic particles of mass @xmath155 . in the following @xmath156
is adopted and we restrict the treatment to hypothetical fifth forces between hadronic particles for which nucleon numbers @xmath157 and @xmath158 are included , @xmath159 for the proton and @xmath160 for the deuteron .
the effect of the fifth force on the energy of a molecular state @xmath161 can be calculated as an expectation value @xmath162 with @xmath163 in this treatment the range of the force @xmath154 is retained as a parameter , while the expectation value can be numerically evaluated by inserting the well - known wave functions @xmath164 for the quantum states of the h@xmath0 molecules , their isotopomers , and the hd@xmath7 ion , by integrating over the coordinate @xmath165 representing the internuclear distance in the molecule .
this effect of a fifth force on level energies can be extended to transition frequencies by calculating the differences in expectation values for ground states @xmath166 and excited states @xmath167 , represented by @xmath168 .
constraints on a fifth force can be derived by combining eq .
( [ v5expect ] ) and eq .
( [ v5ineq ] ) to @xcite : @xmath169 and computing the fifth force expectation value for each specific precision measurement and for each value of the range parameter @xmath154 .
this then results in the constraints as graphically represented in fig .
[ limit-5th ] .
the figure shows that the result of the measurement on the ( 8,0 ) high overtone band hd@xmath7 yields the strongest constraint .
this is caused by the fact that the wave function densities and average internuclear distance of @xmath121 deviate more strongly from the ground state wave function @xmath92 in comparison to the other wave functions probed , @xmath93 @xcite and @xmath123 @xcite .
the measurements in hd@xmath7 are currently more accurate than for h@xmath0 , but the effect of the smaller equilibrium distance in h@xmath0 with respect to hd@xmath7 is reflected in a steeper decline of the constraint for the ions below 0.3 .
note also that constraints from d@xmath0 are tighter than those from h@xmath0 because of the @xmath170 factor .
note , however , that this is based on the assumption that protons and neutrons deliver an additive contribution to the fifth force acting between hadrons .
the figure also shows how the values for the dissociation energy d@xmath88 are more constraining than those of the fundamental vibration measured at higher accuracy .
this is because the dissociation limit corresponds to a state with @xmath171 .
a similar reasoning can be followed when considering the effect of extra dimensions .
a theory was developed by arkani - hamed and coworkers @xcite , known as add - theory , postulating that the three forces of the standard model act in three dimensions ( on a 3-brane ) , while gravity may act in a higher ( @xmath172)-dimensional space , where @xmath173 refers to the number of extra dimensions .
as it turns out the effects of gravity become larger in higher dimensional space .
the extra dimensions , if existent , should not extend over large distances otherwise it would be in contradiction with observations . as was postulated already in 1926 by klein the extra dimensions might exist nevertheless in compactified form @xcite . in our analysis
we assume further that the entire molecule of size @xmath174 is contained within the compactification length @xmath175 of the extra dimensions , so that it effectively probes the higher - dimensional space .
this assumption can be made , since the bounds on extra dimensions have not yet been experimentally limited to this molecular scale . indeed , as the outcome of the present analysis we find @xmath176 .
salumbides _ et al . _
@xcite applied the concepts of add - theory in a phenomenological fashion to diatomic molecules where the two nuclei are considered to act as test bodies in a cavendish - type gravity experiment . again , as in the case of the fifth force , the expectation value @xmath177 of this potential is calculated in a perturbative manner by integrating over the molecular wave function @xmath178 . while in the previous analysis @xcite geometrical prefactors , related to volumes and surfaces of hyperspheres , had been ignored , these prefactors will be included here , giving rise to a small numerical adaptation of results .
the flux conservation law for the gravitational force @xmath179 or the gravitational field @xmath180 acting on a test mass @xmath181 can be recast in @xmath182 dimensions as @xmath183 where @xmath184 is the gravitational constant and @xmath66 the mass contained within the spherical volume @xmath185 . assuming that @xmath179 is spherically symmetric ( equivalent to the requirement that @xmath186 ) , the choice of an @xmath182-dimensional hypersphere as the enclosing gaussian volume simplifies the integration .
the integral is taken over the hypersurface surrounding the @xmath182-dimensional hypersphere , defined as an @xmath187 dimensional surface @xmath188 , which relates to the volume @xmath189 of the @xmath182-dimensional hypersphere with radius @xmath165 via @xmath190 , yielding : @xmath191 r^{n-1 } , \label{hypersurface}\ ] ] where @xmath192 is the surface area for a unit hypersphere ( @xmath193 ) .
this then yields an equation for newton s law in @xmath182 dimensions as : @xmath194 where the newtonian inverse square law dependence is recovered for 3 dimensions .
note that @xmath184 have different dimensional units when compared to the ( 3-dimensional ) gravitational constant @xmath195 .
following the analysis of @xcite , assuming that the @xmath173 extra spatial dimensions have the same compactification length @xmath175 , the @xmath182-dimensional force @xmath196 can be derived in the limiting case when @xmath197 to be @xmath198 this should be in correspondence with normal newtonian gravity , i.e. @xmath199 as @xmath200 , which implies that the factor inside the braces @xmath201 is equivalent to the 3-dimensional gravitational constant @xmath64 . indeed the definition for the gravitational constant @xmath202 for the case of @xmath182-dimensions : @xmath203 { ( r_{c})^{n_e } } g \label{gn - factor}\ ] ] is the very solution invoked in add theory @xcite to solve the hierarchy problem on the weakness of gravity compared to the other sm interactions . the @xmath182-dimensional gravitational potential can be derived via the relation @xmath204 , yielding @xmath205 { ( r_{c})^{n_e } } g \frac{m m } { r^{(n-2 ) } } , \label{n - pot}\ ] ] where @xmath206 has been expressed in terms of @xmath64 .
note that the geometrical prefactor [ @xmath207 for the potential : @xmath208 = \left [ \frac{4\pi \gamma(\frac{n}{2}+1)}{n(n-2)\pi^{n/2 } } \right ] \label{prefactor}\ ] ] scales differently as a function of @xmath182 , but reduces to [ @xmath209=1 for 3-dimensional space .
however for extra dimensions the prefactor becomes [ @xmath210=@xmath211 , [ @xmath212=@xmath213 , [ @xmath214=@xmath215 , etc .
the extra - dimensional gravitational potential of add - theory ( within the compactification radius @xmath216 ) is expressed in a convenient form for application in molecular theory as @xmath217 \frac{1}{r^{(n_e+1)}},\ ] ] where @xmath218 is set to be the nucleon mass(es ) , @xmath219 represents the very small gravitational coupling constant , and @xmath157 and @xmath158 the number of hadrons ( protons and neutrons ) in the attracting bodies .
outside the compactification radius ( @xmath220 ) gravity is assumed to follow the 3-dimensional newtonian law : @xmath221 which will also be used for scaling purposes .
the perturbative effect on a quantum state is then , again following @xcite : @xmath222 \times \nonumber \\ & & \braket{\psi(r ) | \frac{1}{r^{(n_e+1 ) } } | \psi(r ) } \end{aligned}\ ] ] where the number of extra dimensions @xmath173 and the compactification radius @xmath175 ( taken equal for all extra dimensions ) are retained as parameters .
similarly as in the case of the fifth force treatment a perturbative effect can also be calculated on a transition frequency by taking the differing perturbations on the ground and excited quantum states .
the experimental data are then implemented , as prescribed in eq .
( [ v5ineq ] ) , to constrain the sizes of the extra dimensions for each number of extra dimensions @xmath173 by : @xmath223 } \times \nonumber \\ & & { \left [ \braket{\frac{1}{r^{n_e+1}}}_{\psi_e } - \braket{\frac{1}{r^{n_e+1}}}_{\psi_g } \right]^{-1}}\end{aligned}\ ] ] with calculation of the expectation values for ground @xmath224 and excited @xmath225 states in a transition .
hence constraints on the compactification radius @xmath175 can be deduced for cases of differing @xmath173 extra dimensions .
for the special case of the measurement of the ( 8,0 ) band of hd@xmath7 results are graphically displayed in fig .
[ limit - dim ] . in the figure
the extra - dimensional contribution to gravity as represented by @xmath177 is scaled to normal gravity @xmath226 , thus demonstrating that the effect of the former is much larger . from fig .
[ limit - dim ] a constraint on a compactification radius can be read for each number @xmath173 .
in particular for the special case of the m - theory which lives in 10 spatial dimensions @xcite a constraint follows of @xmath227 m , under the assumption that all extra dimensions have equal extension .
some 10 h@xmath0 absorption systems at high redshift in the range @xmath51 have been analyzed from spectra observed with the best large dish optical telescopes of the 8 - 10 m class ( eso very large telescope at paranal , chile , and the keck telescope at hawaii ) equipped with high resolution spectrometers resulting in a constraint on a varying proton - electron mass ratio of @xmath3 at 3@xmath4 . with current technologies no decisive improvements
are expected , although in a quest for varying constants even incremental improvements are valuable .
a number of 23 additional h@xmath0 absorbers have been identified @xcite , some of which apparently of the same quality as the 10 already analyzed in detail .
another set of 23 systems is reported exhibiting certain or only tentative detection of h@xmath0 @xcite .
improvements on the quality of data can be accomplished with the planned larger dish telescopes ( the european extremely large telescope , the thirty - meter - telescope , and the giant magellan telescope ) and with the use of fiber - fed spectrometers ( like espresso @xcite planned for the vlt ) , as well as the use of frequency - comb assisted frequency calibration @xcite .
this should lead to at least an order - of - magnitude tighter constraint on a varying @xmath5 .
detection of h@xmath8 ions in space , either at high redshift or galactic , might impose similar constraints on varying constants .
attempts to detect these ionic forms of hydrogen molecules have been unsuccessful so far , despite dedicated searches by radio astronomy on the spin - splitting resonance at @xmath228 ghz @xcite .
also searches for vibrationally excited states have been performed @xcite , in view of the fact that the h@xmath8 to h@xmath229 conversion is slower for vibrationally excited states .
the development of new radio observatories of extreme sensitivity as the square kilometer array @xcite bears the prospect of detecting h@xmath8 in outer space .
it is remarkable that none of the electronically excited states of the h@xmath8 ion , calculated accurately already in the early days of quantum mechanics by teller @xcite has ever been spectroscopically observed , neither in space nor in the laboratory .
this is due to the fact that the electronically excited states ( @xmath230 , @xmath231 , @xmath232 , etc . )
have potential wells lying at such large internuclear separation that only franck - condon allowed decay is possible to very highest vibrational levels and the dissociation continuum of the @xmath233 ground state .
such decay would produce radiation of @xmath234 ev , hence in the vacuum ultraviolet part of the spectrum .
an exception is the @xmath235 state which exhibits a strongly repulsive well , but supports nevertheless some weakly bound states at larger internuclear separation , which were observed through microwave spectroscopy in h@xmath8 @xcite and in hd@xmath7 @xcite . as for laboratory measurements
it is noted that , since the advent of quantum mechanics , progress in measurements and theory of the dissociation limit of h@xmath0 has been mutually stimulating , therewith producing steady progress .
this history , both for theoretical and experimental advances , is presented in fig .
[ fig - dis ] in terms of the accuracy of the values determined .
the data on theory starts with the calculation by heitler and london @xcite mainly aiming at the quantum mechanical explanation that the ground state of the h@xmath0 is bound .
thereafter a great number of calculations were performed of ever increasing accuracy . as is common for ab initio calculations ,
uncertainties were not explicitly stated , and in fig .
[ fig - dis ] estimates of uncertainties were derived from the number of digits given in the original papers . however , in view of the fact that qed - effects were not explicitly calculated the accuracies are estimated to be limited to @xmath236 @xmath91 @xcite .
the results of some of those calculations appeared to be accurate , but that is due to the fact that not - addressed relativistic and qed - contributions cancel for a large part .
only for the final insert the ab initio calculation explicitly includes relativistic and qed - effects up to @xmath133 @xcite achieving an accuracy of @xmath237 @xmath91 .
the first experimental determination of the hydrogen dissociation energy was established just before the advent of quantum mechanical calculations in 1926 @xcite .
this measurement was superseded over the years by experimental studies of increasing accuracy with the most accurate measurement achieving a precision of @xmath90 @xmath91 @xcite .
[ fig - dis ] demonstrates how the improvement in the accuracy of this benchmark value of molecular physics has increased by some seven orders of magnitude in less than a century , and how the improvement of experiment and theory have gone hand - in - hand . for the future a number of possible improvements may be identified . through the measurement of the binding energy of the @xmath238 state @xcite a novel route for bridging the gap between the electronic ground state and the ip - limit
is opened , as an alternative to the @xmath85 , @xmath92 state as an intermediate ( see also fig . [ h2scheme]a ) .
this novel route involves only a two - step process , instead of the three - step process via @xmath85 .
it requires however doppler - free 2-photon excitation at @xmath239 nm , which could possibly be generated at sufficient intensity by harmonic upconversion in a kbbf crystal @xcite .
another prospect for improvement is in the remeasurement of the anchor lines , either of the @xmath85 state or the @xmath240 state , using ramsey - comb spectroscopy with ultrafast laser pulses @xcite . in this case
excitation takes place with two ultrafast phase - locked laser pulses derived from a frequency comb laser .
ramsey spectroscopy signal can be obtained by varying the delay between the pulses on two different time scales via adjustments of the frequency comb laser . from a series of ramsey
signals the transition frequencies can be obtained with possibly two orders of magnitude improved accuracy ( @xmath241 khz ) compared to the single - nanosecond pulse excitation technique used so far .
on the theory side improvements have been made recently in accurate calculations of h@xmath0 level energies , in particular for the adiabatic corrections @xcite and the non - adiabatic corrections @xcite . for a comparison with experiment and more stringent test of qed in molecular systems
further improvements on the calculations of relativistic and qed contributions are awaited .
experimental work on h@xmath0 at increasing accuracy may also provide a handle to target @xmath242 symmetry - breaking effects in the homo - nuclear species , calculated to be at the @xmath243 @xmath91 level @xcite for the very highest vibrational states , which might be probed by the photolysis preparation methods explored @xcite .
tests of qed in neutral molecules can be extended to tritium containing nuclei , already included in the calculations , while only few experimental studies have been performed @xcite . for the more distant future
it may be noted that the lifetimes of the ( @xmath103 ) rovibrational levels of the @xmath84 electronic ground state of h@xmath0 are in the range of @xmath244 seconds @xcite . when performing measurements at infrared frequencies , i.e. probing quadrupole transitions , of some @xmath245 hz , the spectral lines exhibit a natural line width of @xmath246 of the transition frequency , which might in principle allow for a 20-digit determination of level energies to be compared with theory .
of course many experimental difficulties need to be overcome to reach such accuracies , where activities in decelerating , cooling and trapping of h@xmath0 @xcite may become useful ingredients . in any case there
exists a vast `` room - at - the - bottom '' for improving precision metrology of neutral hydrogen in search for new physics . for the charged hydrogen molecule
the precision achieved in theory is much higher , with calculations up to order @xmath247 , while experiments on hd@xmath7 are currently the most accurate , and the most constraining for testing new physics ( see e.g. fig .
[ limit-5th ] ) .
prospects are that accuracy can be improved and proposals for molecular clocks based on hydrogen molecular ions are discussed @xcite . a value of the proton - electron mass ratio @xmath5 can ultimately be determined from spectroscopy , as was already postulated long time ago by wing _
et al . _
calculations of level energies have recently reached a precision at which they are limited by the value of @xmath5 rather than the theoretical methods , while the sensitivity on other fundamental constants such as the deuteron - proton mass ratio and the proton size has become significant @xcite .
experiments for the charged species are however much less accurate than theory , even with the trapping and laser - cooling methods applied .
spectroscopy of improved precision in the hd@xmath7-ion trap can be achieved by doppler - free two - photon laser excitation with counter - propagating beams .
transition probabilities were found to be weak using degenerate two - photon excitation @xcite . a scheme based on two - photon excitation with near - degenerate laser beams
was proposed bearing the advantage of increased transition probabilities @xcite . in a scheme where two different wavelengths are imposed ( see fig . [ hd+ ] ) , involving a first the transition @xmath248 at @xmath249 nm and a second transition @xmath250 at @xmath251 nm
, the lamb - dicke regime can be reached .
this makes this measurement measurement scheme essentially doppler - free if counter - propagating laser beams are used .
calculations and numerical simulations have shown that accuracies at the @xmath252 level can be reached @xcite , therewith providing a greatly enhanced accuracy on the transition frequencies in hd@xmath7 .
this will produce much tighter constraints on fifth forces and extra dimensions , in addition to a competitive value of the proton - electron mass ratio @xmath5 .
the authors wish to thank julija bagdonaite , john barrow , martin barstow , denise bailly , jurriaan biesheuvel , mario dapr , gareth dickenson , beatriz gato - rivera , laurent hilico , urs hollenstein , toncho ivanov , christian jungen , lex kaper , jean - philippe karr , julian king , jacek komasa , vladimir korobov , jinjun liu , frederic merkt , michael murphy , mingli niu , krzystof pachucki , simon preval , mourad roudjane , evelyne roueff , bert schellekens , daniel sprecher , lydia tchang - brillet , michel vervloet and john webb for their collaboration and the fruitful discussions .
this work is financially supported by the netherlands foundation for fundamental research of matter ( fom ) through the program `` broken mirrors & drifting constants '' and by the european research council ( erc ) under the european union s horizon 2020 research and innovation programme ( grant agreement no 670168 ) .
w. ubachs , r. buning , k. s. e. eikema , e. reinhold , on a possible variation of the proton - to - electron mass ratio : h@xmath255 spectra in the line of sight of high - redshift quasars and in the laboratory , 241 ( 2007 ) 155179 .
j. philip , j. sprengers , t. pielage , c. a. de lange , w. ubachs , e. reinhold , highly accurate transition frequencies in the h@xmath255 lyman and werner absorption bands , can .
82 ( 2004 ) 713722 .
w. ubachs , e. reinhold , highly accurate h@xmath0 lyman and werner band laboratory measurements and an improved constraint on a cosmological variation of the proton - to - electron mass ratio , 92 ( 2004 ) 101302 .
e. reinhold , r. buning , u. hollenstein , a. ivanchik , p. petitjean , w. ubachs , indication of a cosmological variation of the proton - electron mass ratio based on laboratory measurement and reanalysis of h@xmath255 spectra , 96 ( 2006 ) 151101 .
s. hannemann , e. j. salumbides , s. witte , r. t. zinkstok , e. j. van duijn , k. s. e. eikema , w. ubachs , frequency metrology on the @xmath258 ( 0,0 ) transition in h@xmath255 , hd , and d@xmath255 , phys .
a 74 ( 2006 ) 062514 .
d. bailly , e. salumbides , m. vervloet , w. ubachs , accurate level energies in the @xmath259 , @xmath260 , @xmath261 , @xmath262 , @xmath263 , @xmath264 , @xmath265 , @xmath266 , @xmath267 states of h@xmath255 , 108 ( 2010 ) 827846 .
e. j. salumbides , d. bailly , a. khramov , a. l. wolf , k. s. e. eikema , m. vervloet , w. ubachs , improved laboratory values of the h@xmath255 lyman and werner lines for constraining time variation of the proton - to - electron mass ratio , 101 ( 2008 ) 223001 .
t. i. ivanov , g. d. dickenson , m. roudjane , n. d. oliveira , d. joyeux , l. nahon , w .- u .
l. tchang - brillet , w. ubachs , fourier - transform spectroscopy of hd in the vacuum ultraviolet at @xmath154 = 87 - 112 nm , 108 ( 2010 ) 771786 .
a. l. malec , r. buning , m. t. murphy , n. milutinovic , s. l. ellison , j. x. prochaska , l. kaper , j. tumlinson , r. f. carswell , w. ubachs , keck telescope constraint on cosmological variation of the proton - to - electron mass ratio , 403 ( 2010 ) 15411555 .
h. rahmani , m. wendt , r. srianand , p. noterdaeme , p. petitjean , p. molaro , j. b. whitmore , m. t. murphy , m. centurion , h. fathivavsari , s. dodorico , t. m. evans , s. a. levshakov , s. lopez , c. j. a. p. martins , d. reimers , g. vladilo , the uves large program for testing fundamental physics - ii .
constraints on a change in @xmath5 towards quasar he 0027@xmath2681836 , 435 ( 2013 ) 861878 .
r. i. thompson , j. bechtold , j. h. black , d. eisenstein , x. fan , r. c. kennicutt , c. martins , j. x. prochaska , y. l. shirley , an observational determination of the proton - to - electron mass ratio in the early universe , 703 ( 2009 ) 16481662 .
j. bagdonaite , e. j. salumbides , s. p. preval , m. a. barstow , j. d. barrow , m. t. murphy , w. ubachs , limits on a gravitational field dependence of the proton - electron mass ratio from h@xmath255 in white dwarf stars , 113 ( 2014 ) 123002 .
e. j. salumbides , j. bagdonaite , h. abgrall , e. roueff , w. ubachs , h@xmath0 lyman and werner band lines and their sensitivity for a variation of the protonelectron mass ratio in the gravitational potential of white dwarfs , 450 ( 2015 ) 12371245 .
j. liu , e. j. salumbides , u. hollenstein , j. c. j. koelemeij , k. s. e. eikema , w. ubachs , f. merkt , determination of the ionization and dissociation energies of the hydrogen molecule , 130 ( 2009 ) 174306 .
a. osterwalder , a. west , f. merkt , c. jungen , high - resolution millimeter wave spectroscopy and multichannel quantum defect theory of the hyperfine structure in high rydberg states of molecular hydrogen h@xmath0 , 121 ( 2004 ) 1181011838 .
s. kassi , a. campargue , k. pachucki , j. komasa , the absorption spectrum of d@xmath0 : ultrasensitive cavity ring down spectroscopy of the ( 20 ) band near 1.7 @xmath5 m and accurate ab initio line list up to 24 000 @xmath91 , 136 ( 2012 ) 184309 .
u. bressel , a. borodin , j. shen , m. hansen , i. ernsting , s. schiller , manipulation of individual hyperfine states in cold trapped molecular ions and application to hd@xmath274 frequency metrology , phys .
lett . 108 ( 2012 ) 183003 .
p. m. aker , g. j. germann , j. j. valentini , state - to - state dynamics of h+hx collisions .
i. the h+hx @xmath275 h@xmath255+x ( x = cl , br , i ) abstraction reactions at 1.6 ev collision energy , 90 ( 1989 ) 47954808 .
a. campargue , s. kassi , k. pachucki , j. komasa , the absorption spectrum of h@xmath255 : crds measurements of the ( 2 - 0 ) band , review of the literature data and accurate ab initio line list up to 35 000 @xmath91 , phys .
( 2012 ) 802815 .
hu , h. pan , c .- f .
cheng , y. r. sun , x .- f .
li , j. wang , a. campargue , a .- w .
liu , the @xmath279 s(0)-s(3 ) electric quadrupole transitions of h@xmath255 near 0.8 @xmath5 m , astroph .
j. 749 ( 2012 ) 76 .
z. w. fu , e. a. hessels , s. r. lundeen , determination of the hyperfine structure of h@xmath8 ( @xmath284 ) by microwave spectroscopy of high-@xmath285 rydberg states of h@xmath0 , phys .
a 46 ( 1992 ) r5313r5316 .
d. sprecher , c. jungen , f. merkt , determination of the binding energies of the @xmath286 rydberg states of h@xmath0 , hd , and d@xmath0 from high - resolution spectroscopic data by multichannel quantum - defect theory , 140 ( 2014 ) 104303 .
j. komasa , k. piszczatowski , g. ach , m. przybytek , b. jeziorski , k. pachucki , quantum electrodynamics effects in rovibrational spectra of molecular hydrogen , j. chem .
theory comput . 7
( 2011 ) 31053115 .
r. pohl , a. antognini , f. nez , f. d. amaro , f. biraben , j. m. r. cardoso , d. s. covita , a. dax , s. dhawan , l. m. p. fernandes , a. giesen , t. graf , t. w. hnsch , p. indelicato , l. julien , c. y. kao , p. knowles , e .- o .
l. bigot , y .- w .
liu , j. a. m. lopes , l. ludhova , c. m. b. monteiro , f. mulhauser , t. nebel , p. rabinowitz , j. m. f. dos santos , l. a. schaller , k. schuhmann , c. schwob , d. taqqu , j. f. c. a. veloso , f. kottmann , the size of the proton , nature 466 ( 2010 ) 213216 .
f. a. pepe , s. cristiani , r. rebolo lopez , n. c. santos , a. amorim , g. avila , w. benz , p. bonifacio , a. cabral , p. carvas , r. cirami , j. coelho , m. comari , i. coretti , v. de caprio , h. dekker , b. delabre , p. di marcantonio , v. dodorico , m. fleury , r. garca , j. m. herreros linares , i. hughes , o. iwert , j. lima , j .- l .
lizon , g. lo curto , c. lovis , a. manescau , c. martins , d. mgevand , a. moitinho , p. molaro , m. monteiro , m. monteiro , l. pasquini , c. mordasini , d. queloz , j. l. rasilla , j. m. rebordo , s. santana tschudi , p. santin , d. sosnowska , p. span , f. tenegi , s. udry , e. vanzella , m. viel , m. r. zapatero osorio , f. zerbi , espresso : the echelle spectrograph for rocky exoplanets and stable spectroscopic observations , in : spie conf .
, vol . 7735 , 2010 , p. 9
m. t. murphy , t. udem , r. holzwarth , a. sizmann , l. pasquini , c. araujo - hauck , h. dekker , s. dodorico , m. fischer , t. w. hnsch , a. manescau , high - precision wavelength calibration of astronomical spectrographs with laser frequency combs , 380 ( 2007 ) 839847 .
t. wilken , g. lo curto , r. a. probst , t. steinmetz , a. manescau , l. pasquini , j. i. gonzalez hernandez , r. rebolo , t. w. hnsch , t. udem , r. holzwarth , a spectrograph for exoplanet observations calibrated at the centimetre - per - second level , nature 485 ( 2012 ) 611614 .
h. beutler , the dissociation heats of the hydrogen molecule h@xmath0 , determined at 850 angstrom from the rotation structure on the long - wave border of the absorption continuum , zeitschr . f. phys .
29b ( 1935 ) 315327 . | spectroscopy of hydrogen can be used for a search into physics beyond the standard model .
differences between the absorption spectra of the lyman and werner bands of h@xmath0 as observed at high redshift and those measured in the laboratory can be interpreted in terms of possible variations of the proton - electron mass ratio @xmath1 over cosmological history .
investigation of some ten of such absorbers in the redshift range @xmath2 yields a constraint of @xmath3 at 3@xmath4 .
observation of h@xmath0 from the photospheres of white dwarf stars inside our galaxy delivers a constraint of similar magnitude on a dependence of @xmath5 on a gravitational potential @xmath6 times as strong as on the earth s surface . while such astronomical studies aim at finding quintessence in an indirect manner , laboratory precision measurements target such additional quantum fields in a direct manner .
laser - based precision measurements of dissociation energies , vibrational splittings and rotational level energies in h@xmath0 molecules and their deuterated isotopomers hd and d@xmath0 produce values for the rovibrational binding energies fully consistent with quantum ab initio calculations including relativistic and quantum electrodynamical ( qed ) effects .
similarly , precision measurements of high - overtone vibrational transitions of hd@xmath7 ions , captured in ion traps and sympathetically cooled to mk temperatures , also result in transition frequencies fully consistent with calculations including qed corrections .
precision measurements of inter - rydberg transitions in h@xmath0 can be extrapolated to yield accurate values for level splittings in the h@xmath8-ion .
these comprehensive results of laboratory precision measurements on neutral and ionic hydrogen molecules can be interpreted to set bounds on the existence of possible fifth forces and of higher dimensions , phenomena describing physics beyond the standard model . molecular hydrogen , laser spectroscopy , fundamental forces , varying constants , extra dimensions |
the distribution of rational points on an algebraic variety @xmath1 is very subtle .
for example , it can often happen that there is a proper subvariety @xmath2 of @xmath1 such that the set of rational points of @xmath2 has density one in the set of rational points of @xmath1 , where density is used in the sense of weil heights .
roughly speaking , such subvarieties @xmath2 are called accumulating subvarieties .
thus , if one is interested in the arithmetic of @xmath1 , one must first identify which rational points of @xmath1 lie on @xmath2 , and which do not .
unfortunately , even the purely geometric problem of identifying potential accumulating subvarieties @xmath2 can be very difficult .
it would be helpful to have a local and arithmetic criterion to identify points @xmath0 which lie on accumulating subvarieties . in this paper
, we do not quite manage to construct such a criterion , but we do identify an invariant , called the approximation constant of @xmath0 on @xmath1 with respect to a divisor @xmath3 , which describes how well @xmath0 can be approximated by other rational points of @xmath1 .
( see definition [ bigdef ] for details . )
if @xmath0 lies on an accumulating subvariety , then it can be well approximated by other rational points , and therefore will have a small approximation constant .
in the course of computing this constant in many examples , it became clear that sequences which best approximate a rational point @xmath0 tend to lie along curves . while there are counterexamples to show that this principle can not hold in general , we are able to formulate a conjecture which predicts that it should be true whenever @xmath0 lies on a rational curve defined over the field of coefficients ( see conjecture [ ratcurve ] ) .
we prove this conjecture for a wide range of examples .
the basic technique we use for proving conjecture [ ratcurve ] is inductive .
we start by proving the conjecture for @xmath4 ( see theorem [ ratapprox ] ) .
we then prove a number of inductive results which enable us to take advantage of the structure of the nron - severi group of @xmath1 . using these inductive results , we prove conjecture [ ratcurve ] for a wide range of rational surfaces by means of a careful analysis of the nef cone . in the final section ,
we make several remarks about further directions in which to attack conjecture [ ratcurve ] , and drawbacks of our current techniques .
we describe some cases which are unlikely to be proven using our current techniques , and we further deduce conjecture [ ratcurve ] from vojta s main conjecture ( conjecture 3.4.3 of @xcite ) for a generic variety of non - negative kodaira dimension .
i am grateful to several people for helpful conversations about this material , including doug park , mike roth , and cam stewart .
i would particularly like to thank kevin hare for writing some extremely useful computer programs , without which many of the theorems in this paper would still be conjectures , and the anonymous referee , whose invaluable suggestions improved the manuscript enormously .
in this paper , all heights are absolute and multiplicative .
we fix a number field @xmath5 .
[ ratapprox ] let @xmath6 be any @xmath5-rational point .
let @xmath7 be any sequence of @xmath5-rational points , and let @xmath8 be the set of archimedean places of @xmath5 , with the convention that pairs of complex conjugate embeddings count as one place .
suppose there is a positive real constant @xmath9 such that for all @xmath10 , we have : @xmath11 where the function @xmath12 denotes the local distance function on @xmath13 induced by @xmath14 .
then the set @xmath7 is a subset of the finite union of lines through @xmath0 of plcker height at most @xmath15 , where @xmath16 is a constant which does not depend on @xmath17 . _
proof : _ without loss of generality , we may assume that @xmath18 $ ] , and that none of the @xmath19 lie on the line @xmath20 .
fix a set @xmath21 of representatives of the class group of @xmath5 .
write @xmath22 $ ] , where the @xmath23 and @xmath24 are integers ( that is , elements of @xmath25 ) with @xmath26 .
let @xmath27 be any archimedean place of @xmath5 .
up to multiplication by a bounded function ( which wo nt affect the conclusion of the theorem ) , we may write the @xmath14-distance as : @xmath28 and the height as : @xmath29}\ ] ] note that with our choice of representation , the non - archimedean absolute values do not contribute more than a ( multiplicatively ) bounded function to the height .
now suppose that @xmath30 for all @xmath10 .
then we get : @xmath31 } \prod_{v\in s}\max\{|a_{i1}|_v,\ldots,|a_{in}|_v,|b_i|_v\ } \leq c^{[k:\q]}\ ] ] and hence @xmath32 } \max\{|a_{i1}/b_i|_v,\ldots,|a_{in}/b_i|_v,1\ } \leq c^{[k:\q]}\ ] ] and so _ a fortiori _ , we obtain : @xmath33 in particular , there are only finitely many choices for each @xmath23 , and each choice corresponds to a line of height at most @xmath34}$ ] . in light of this theorem , we make the following definitions : let @xmath35 be an algebraic variety defined over @xmath5 , and let @xmath8 be the set of archimedean absolute values on @xmath5 .
define : @xmath36 for any rational points @xmath0 and @xmath37 on @xmath1 .
note that this is a well defined distance function on the metric space @xmath38 , where @xmath39 is the completion of @xmath5 with respect to @xmath14 . in this paper , all limits will be computed with respect to the topology induced by the distance function @xmath40 .
[ appconst ] let @xmath1 be an algebraic variety defined over @xmath5 , and let @xmath41 be any rational point .
let @xmath3 be any divisor on @xmath1 , with corresponding height function @xmath42 .
assume that there is a positive constant @xmath17 such that @xmath43 for all @xmath37 in some zariski open neighbourhood of @xmath0 .
( this will be satisfied if , for example , @xmath3 is ample , or more generally , if some multiple of @xmath3 is basepoint free . ) for any sequence @xmath44 with @xmath45 , define the _ approximation constant on @xmath1 of @xmath7 with respect to @xmath0 and @xmath3 _ to be the smallest non - negative real number @xmath46 such that : @xmath47 if there is no such smallest non - negative real number , then the sequence does not have an approximation constant . notice that although the definition of @xmath40 depends on the choice of embedding of @xmath1 in projective space , the definition of the approximation constant does not .
this is because two different embeddings of @xmath1 are diffeomorphic , and hence distances change by no more than a multiplicative function bounded away from zero and infinity . in this paper
, this ambiguity will never be significant , so we will refer to @xmath40 as a function independent of the embedding of @xmath1 .
[ noconst ] for example , let @xmath48\in\p^2 $ ] , and let @xmath49 . on any @xmath50-rational line @xmath51 through @xmath0
, we can find an infinite sequence of rational points @xmath7 such that @xmath52 , where @xmath53 is a constant that depends on @xmath51 .
( see for example theorem 2.1 of @xcite . ) therefore , there is a positive constant @xmath54 and an infinite sequence @xmath55 of rational points such that @xmath56 for all @xmath57 .
moreover , we can choose the points @xmath58 such that no two of them are collinear with @xmath0 .
thus , by theorem [ ratapprox ] , the approximation constant of this sequence , if it exists , is greater than one .
but it is clear that for every @xmath59 , there exists a positive constant @xmath60 such that @xmath61 , so any constant of approximation for this sequence is at most one .
we conclude that the sequence @xmath58 does not have a constant of approximation .
[ bigdef ] let @xmath1 , @xmath0 , @xmath3 , and @xmath42 be as in definition [ appconst ] .
define the _ approximation constant of @xmath0 on @xmath1 with respect to @xmath3 _ to be the minimum ( if it is achieved ) of all approximation constants on @xmath1 of sequences @xmath44 with respect to @xmath3 .
a _ sequence of best approximation to @xmath0 with respect to @xmath3 _ is a sequence whose corresponding approximation constant is equal to the approximation constant of @xmath0 .
curve of best approximation to @xmath0 _ is a curve on @xmath1 passing through @xmath0 that contains a sequence of best approximation to @xmath0 .
notice that by this definition , sequences with no approximation constant ( such as the one described in example [ noconst ] ) can not be a sequence of best approximation to a point @xmath0 .
if conjecture [ ratcurve ] is true , then theorem [ curve ] makes it clear that this is a reasonable restriction
. we may therefore improve theorem [ ratapprox ] as follows .
[ projlines ] let @xmath6 be any rational point .
then there exists a sequence of best approximation to @xmath0 .
moreover , any such sequence is a subset of a finite set of lines and has constant of approximation equal to 1 , and any line through @xmath0 is a curve of best approximation to @xmath0 . _
proof : _ we first prove the theorem for @xmath62 . without loss of generality , we may assume that @xmath63 $ ] .
fix a set of representatives @xmath21 of the class group of @xmath5 .
let @xmath64 $ ] be any point in @xmath65 , where @xmath66 and @xmath67 lie in @xmath25 with @xmath68 and @xmath69 .
since @xmath70\}$ ] is clearly a sequence with constant of approximation equal to 1 , it suffices to show that @xmath71 , where @xmath72 is a constant depending only on @xmath5 . since the arithmetic mean is always greater than the geometric mean , for @xmath37 close enough to @xmath0 we compute : @xmath73\prod_v|a / b|_v^{1/[k:\q]}\ ] ] @xmath74(n(a)/n(b))^{1/[k:\q]}\ ] ] where all sums and products are over archimedean places @xmath14 , and @xmath75 denotes the norm @xmath76 .
we also compute : @xmath77}\geq n(b)^{1/[k:\q]}\ ] ] where the product again ranges over archimedean places @xmath14 . note that since @xmath78 , if we change the representation of @xmath37 , then we will change the ideal @xmath79 by multiplication by an element of @xmath80 .
since no two elements of @xmath21 represent the same ideal class , it follows that the representation changes by multiplication by a unit of @xmath25 , and so @xmath81 does not change .
since @xmath82 is nonzero , it follows that @xmath83 , so that : @xmath84\ ] ] and we have proven the result for @xmath62 .
in general , if @xmath85 is any rational point , we can find an embedding @xmath86 such that @xmath87 is a line , and @xmath88 .
since @xmath89 only changes distances and heights by multiplication by a function bounded away from zero and infinity , it follows that a sequence of best approximation to @xmath0 along the line @xmath87 has constant 1 . but by theorem [ ratapprox ] , any sequence with approximation constant at most 1 must lie in a finite set of lines through @xmath0 .
the result follows . in light of theorem [ ratapprox ] ,
we make the following conjecture : [ ratcurve ] let @xmath90 be an algebraic variety defined over @xmath5 , and @xmath3 any ample divisor on @xmath90 .
let @xmath0 be any @xmath5-rational point on @xmath90 .
assume that there is some rational curve @xmath54 on @xmath90 , defined over @xmath5 , such that @xmath91 .
then a sequence of best approximation to @xmath0 on @xmath90 with respect to @xmath3 exists , and may be chosen to lie along some rational curve through @xmath0 .
it is easy to see that if we remove the hypothesis that @xmath0 lie on a rational curve , then the conjecture is false . consider the case of a simple abelian variety @xmath90 of dimension greater than one , and
@xmath0 any @xmath5-rational non - torsion point .
then @xmath0 is certainly a limit point of the set @xmath92 in the archimedean topology , but any curve through @xmath0 must have geometric genus at least two , which by faltings theorem can contain only finitely many @xmath5-rational points .
the following theorem is quite useful : [ curve ] let @xmath93 be any irreducible rational curve of degree @xmath94 defined over @xmath5 , and let @xmath91 be any @xmath5-rational point .
let @xmath95 be the normalization map , and let @xmath96 be the maximum multiplicity of a branch of @xmath97 over @xmath0 .
then a sequence of best approximation to @xmath0 on @xmath54 has constant of approximation @xmath98 .
_ proof : _ if @xmath0 is a smooth point of @xmath54 , then the result follows trivially from the observation that @xmath97 changes distances to @xmath0 only by a multiplicative function bounded away from zero and infinity , and
@xmath97 raises heights to the power @xmath94 ( up to a multiplicative function bounded away from zero and infinity ) .
if @xmath0 is a singular point of @xmath54 , then up to a multiplicative function bounded away from zero and infinity , we have @xmath99 near @xmath0 , where @xmath21 is some point in the finite set @xmath100 .
even though the point @xmath21 depends on the point @xmath37 near @xmath0 , it nevertheless provides only a finite number of alternatives for @xmath101 , so that any sequence of best approximation to @xmath0 must have an infinite subsequence which is a sequence of best approximation to some element of @xmath100 .
thus , let @xmath102 be any point , and let @xmath96 be the multiplicity of the branch of @xmath97 through @xmath21 . assume without loss of generality that @xmath103 .
near @xmath21 , in affine coordinates , the function @xmath97 can be written as @xmath104 , where @xmath93 and the @xmath105 are rational functions in @xmath106 .
the multiplicity @xmath107 of @xmath97 at @xmath21 is equal to @xmath108 , where @xmath109 denotes the largest power of @xmath106 which divides the numerator of @xmath105 ( in lowest terms ) .
let @xmath10 be some index which achieves the minimum , and write @xmath110 .
then as @xmath111 approaches @xmath103 , the distance between @xmath37 and @xmath21 is @xmath112 , and the distance between @xmath113 and @xmath114 is ( up to a bounded multiplicative constant ) @xmath115 .
the result now follows from theorem [ ratapprox ] .
theorem [ curve ] explains why conjecture [ ratcurve ] does not refer specifically to rational curves of minimal @xmath3-degree .
a rational point @xmath0 may be better approximated along a curve of higher degree on which it is a cusp of high multiplicity than along a smooth curve of much lower degree .
[ cusp ] let @xmath54 be the plane curve @xmath116 , and @xmath0 the cusp @xmath117 $ ] . for @xmath118 , it is not hard to see that a sequence of best approximation to @xmath0 along @xmath54 is given by the sequence @xmath119 $ ] , which has constant of approximation @xmath120
however , let @xmath95 be the normalization map
. then @xmath121 is @xmath122 .
a sequence of best @xmath3-approximation to @xmath0 along @xmath62 has constant @xmath123 , not @xmath120 , since @xmath3 corresponds to an embedding of @xmath62 in @xmath124 as a twisted cubic curve .
this is because in definition [ appconst ] , we define @xmath40 in terms of an embedding of @xmath62 , not an arbitrary morphism , whereas in this example , we deal with the distance inherited from @xmath125 . in other words ,
if @xmath126 is the normalization of a subvariety of a smooth variety @xmath90 , then the function @xmath127 may not agree with the pullback of the function @xmath128 to @xmath129 .
if @xmath130 is a curve , then the proof of theorem [ curve ] describes how these two functions differ , but if @xmath130 has higher dimension , then one must be more careful . despite the unpleasant possibility described by example [ cusp ]
, we can relate conjecture [ ratcurve ] to the problem of finding accumulating curves through a point @xmath0 .
let @xmath131 be any set of @xmath5-rational points .
the counting function for @xmath8 is defined to be : @xmath132 where @xmath133 denotes the standard height function on @xmath4 .
let @xmath134 be an algebraic variety defined over @xmath5 , and let @xmath135 be any proper closed subset .
then @xmath130 is said to be an accumulating subvariety of @xmath90 if and only if : @xmath136 where by @xmath137 and @xmath138 we mean @xmath139 and @xmath140 , respectively .
roughly speaking , an accumulating subvariety of @xmath90 is a proper closed subset @xmath130 such that asymptotically , there are more rational points of bounded height on @xmath130 than there are off @xmath130 . let @xmath134 be a smooth algebraic variety defined over @xmath5 , and let @xmath141 be any @xmath5-rational point .
assume that conjecture [ ratcurve ] is true for @xmath0 on @xmath90 that is , assume that there is a rational curve @xmath54 through @xmath0 which contains a sequence of best approximation to @xmath0 .
assume further that @xmath54 has only ordinary singularities at @xmath0 .
if @xmath0 lies on an accumulating curve of @xmath90 , then that curve contains @xmath54 as an irreducible component .
_ proof : _ first , note that since @xmath90 contains a rational curve @xmath54 through @xmath0 , it follows that any accumulating curve containing @xmath0 must be a rational curve , and moreover any component containing @xmath0 must be a rational curve of minimal degree through @xmath0 . by theorem [ curve ] ,
since @xmath54 has only ordinary singularities through @xmath0 , it is clear that any rational curve through @xmath0 must have degree at least @xmath142 .
thus , @xmath54 is a rational curve of minimal degree through @xmath0 , and therefore any accumulating curve through @xmath0 must contain @xmath54 as an irreducible component .
next , we consider the question of rational approximation of rational points on more general varieties .
we begin with a straightforward but surprisingly useful result on products of varieties : [ product ] let @xmath1 and @xmath2 be smooth algebraic varieties defined over a number field @xmath5 , and let @xmath0 and @xmath37 be @xmath5-rational points on @xmath1 and @xmath2 , respectively .
let @xmath143 and @xmath144 be divisors on @xmath1 and @xmath2 , respectively , whose corresponding height functions @xmath145 and @xmath146 are bounded below by a positive constant in some zariski open neighbourhoods of @xmath0 and @xmath37 , respectively .
let @xmath7 ( respectively @xmath147 ) be a sequence of best @xmath143-approximation ( respectively best @xmath144-approximation ) for @xmath0 ( respectively @xmath37 ) , with constant of approximation @xmath46 ( respectively @xmath148 ) .
then either @xmath149 or @xmath150 is a sequence of best @xmath51-approximation for @xmath151 on @xmath152 , where @xmath153 ( @xmath154 is the projection onto the @xmath10th factor ) .
moreover , any sequence @xmath155 whose constant of approximation is less than @xmath156 must have either @xmath157 or @xmath158 for all but finitely many @xmath10 . _
proof : _ for any @xmath5-rational point @xmath159 on @xmath152 , we have @xmath160 , where @xmath145 and @xmath146 are heights on @xmath1 and @xmath2 with respect to the divisors @xmath143 and @xmath144 , respectively . to prove the theorem , note that the non - negativity of @xmath46 and @xmath148 implies that it suffices to prove the last claim .
let @xmath155 be any sequence of points in @xmath161 with @xmath162 .
we know that if @xmath163 is infinite then @xmath164 is unbounded for any @xmath165 .
similarly , if @xmath166 is infinite then @xmath167 is unbounded for any @xmath165 .
thus , if there are infinitely many @xmath10 for which both @xmath168 and @xmath169 , then we deduce that the following function is unbounded : @xmath170 we may now deduce that the following function is unbounded : @xmath171 which is to say that @xmath155 has constant of approximation at least @xmath156 .
we have the following immediate corollary : [ divsum ] let @xmath1 be a variety defined over @xmath5 , and let @xmath41 be any @xmath5-rational point .
let @xmath172 and @xmath173 be two divisors on @xmath90 with height functions @xmath174 and @xmath175 bounded below by a positive constant in a neighbourhood of @xmath41 .
assume that @xmath176 is a curve of best @xmath177-approximation to @xmath0 for each @xmath10 , and let @xmath178 be any positive linear combination of @xmath172 and @xmath173 .
* if @xmath179 , then @xmath54 is also a curve of best @xmath3-approximation to @xmath0 . *
if @xmath180 , then either @xmath181 is a curve of best @xmath3-approximation to @xmath0 , or @xmath182 and @xmath183 is a curve of best @xmath3-approximation to @xmath0 . the following theorem will be used in the proofs of theorem [ four ] and theorem [ onefibre ] .
[ addeffective ] let @xmath1 be a variety defined over @xmath5 , and let @xmath3 and @xmath184 be divisors on @xmath1 .
let @xmath41 be a rational point , and @xmath54 a curve of best @xmath3-approximation to @xmath0 .
assume that @xmath42 and @xmath185 are bounded below by positive constants in some neighbourhood of @xmath0 , that @xmath184 is effective , and that @xmath186
. then @xmath54 contains a sequence of best @xmath187-approximation to @xmath0 . _
proof : _ since @xmath186 , we can find a real , positive constant @xmath46 such that @xmath188 $ ] for all @xmath189 .
thus , the sequence of best @xmath3-approximation to @xmath0 along @xmath54 will have the same constant of approximation with respect to @xmath190 as with respect to @xmath3 .
however , @xmath191 is bounded from below away from @xmath184 , and @xmath192 .
it follows that for any sequence @xmath147 of rational points on @xmath90 with @xmath193 , if the quantity @xmath194 is unbounded , then the quantity @xmath195 will also be unbounded .
the theorem follows .
these theorems , surprisingly , give an immediate proof of conjecture [ ratcurve ] for any split , geometrically minimal surface @xmath1 over @xmath5 .
we say that a surface @xmath1 is split over @xmath5 if and only if the inclusion of nron - severi groups @xmath196 is an isomorphism that is , if every algebraic cycle on @xmath197 is numerically equivalent to some @xmath5-rational cycle .
the proof relies crucially on the classification of minimal rational surfaces over @xmath198 , which can be found in , for example , @xcite .
[ minimal ] conjecture [ ratcurve ] holds for any split , geometrically minimal rational surface defined over @xmath5 . in particular
, it holds for all the hirzebruch surfaces @xmath199 for @xmath200 .
_ proof : _ the classification shows that any minimal rational surface over @xmath198 is either @xmath125 , @xmath201 , or a hirzebruch surface @xmath199 for some integer @xmath202 . the case of @xmath125 is proven in theorem [ ratapprox ] .
the case @xmath203 is immediate from theorem [ product ] .
the case of @xmath199 follows from theorem [ product ] and corollary [ divsum ] as follows .
let @xmath204 be the closure in @xmath205 of the cone of ample divisors on @xmath199 , where @xmath206 is the nron - severi group of @xmath199 .
then @xmath204 is generated by the two divisor classes @xmath207 and @xmath208 , where @xmath209 is a fibre of the map @xmath210 and @xmath8 is the unique section of @xmath211 with self - intersection @xmath212 .
the class @xmath3 is @xmath213 , where @xmath97 is the contraction of @xmath8 to the vertex of the cone @xmath54 over a rational normal curve of degree @xmath57 and @xmath51 is a line of the ruling of the cone .
consider the classes @xmath3 , @xmath208 , and @xmath214 .
for any rational point @xmath0 on @xmath199 , it is clear that a curve of best @xmath208-approximation to @xmath0 is a fibre of @xmath211 , and a curve of best @xmath3-approximation to @xmath0 is the preimage of a line through @xmath215 on the cone @xmath54 ( if @xmath216 ) or @xmath8 ( if @xmath217 ) . if @xmath218 , then these are the same curve , so we are done by corollary [ divsum ] .
if @xmath219 , then we simply note that the divisor @xmath214 corresponds to an embedding of @xmath199 in projective space such that fibres of @xmath211 are all lines , and @xmath8 is also a line , so they both contain sequences of best @xmath220-approximation to @xmath0 .
we again conclude by corollary [ divsum ] , since every very ample divisor on @xmath199 is a positive linear combination either of @xmath3 and @xmath214 or @xmath208 and @xmath214 .
[ simplefibres ] let @xmath221 , and let @xmath1 be the blowup of the hirzebruch surface @xmath199 at @xmath222 points , no two of which lie in the same fibre of the map to @xmath62 .
conjecture [ ratcurve ] is true for @xmath1 .
_ proof : _ first , it clearly suffices to assume that none of the blown up points lies on the @xmath223-section on @xmath199 , since blowing up points on this section will only increase @xmath57 .
let @xmath8 be the class of the strict transform of the @xmath223-section on @xmath199 , and let @xmath208 be the class of the total transform of a fibre of the map from @xmath199 to @xmath62 . for each of the @xmath5 reducible fibres , let @xmath224 and @xmath225 be the two components , of which @xmath225 is the one which intersects @xmath8 .
let @xmath226 be any vector , and define a divisor @xmath227 on @xmath1 by : @xmath228 where the @xmath229 denotes a formal dot product .
the effective cone of @xmath1 is generated by the divisors @xmath224 , @xmath225 , and @xmath8 .
the nef cone of @xmath1 is generated by the divisors @xmath227 and @xmath208 , where @xmath46 ranges over all of @xmath230 .
_ proof : _ the picard rank of @xmath1 is @xmath231 , so an arbitrary divisor on @xmath1 can be written as @xmath232 .
if @xmath3 is to be ample , then its intersection with @xmath8 , @xmath224 , and @xmath225 must be positive , giving : @xmath233 these inequalities define an open cone in @xmath234 let @xmath54 be the closure of this cone
. then @xmath54 is finitely generated , and the generating rays of @xmath54 are all intersections of @xmath235 of the hyperplanes @xmath236 , @xmath237 , and @xmath238 .
let @xmath239 be a generator of @xmath54 . if @xmath240 , then for some @xmath241 , @xmath242 , since @xmath239 must be contained in at least @xmath235 of the listed hyperplanes .
thus , @xmath239 is contained in the hyperplane @xmath243 . furthermore ,
since @xmath239 is an extremal ray of @xmath54 , there must be some set of exactly @xmath235 of the @xmath244 and @xmath245 whose intersection is the space generated by @xmath239 . since the vector @xmath208 is contained in the intersection of all the @xmath244 and @xmath245 , it follows that @xmath239 is a positive multiple of @xmath208 .
thus , we may assume that @xmath246 .
if @xmath242 for some @xmath241 , then @xmath239 is again contained in @xmath243 , and thus also in @xmath247 .
but there must also be some @xmath248 such that @xmath249 , and hence @xmath250 and @xmath251 .
this is clearly a contradiction , and so for all @xmath10 , @xmath252 .
let @xmath226 be the vector whose @xmath10th component is 0 if @xmath253 , and 1 if @xmath254 .
then @xmath255 if and only if @xmath253 , and similarly for @xmath245 .
since @xmath256 , the independence of the @xmath235 hyperplanes shows that @xmath239 is a positive multiple of @xmath227 .
thus , the two cones described in the claim are indeed dual to one another .
to conclude the proof of the claim , it suffices to show that every positive linear combination of the @xmath227 and @xmath208 is ample .
first , it is clear that @xmath208 and @xmath227 are basepoint free , since @xmath208 corresponds to a morphism to @xmath62 , and @xmath227 corresponds to the morphism from @xmath1 to a cone , blowing down @xmath8 and exactly one component of each reducible fibre .
we next note that @xmath257 and @xmath258 , and : @xmath259 which is positive . therefore , by the nakai - moishezon criterion ( @xcite , theorem v.1.10 ) , any positive linear combination of @xmath208 and the @xmath227 is ample , and thus the claim is proven .
we now prove conjecture [ ratcurve ] for @xmath1 .
if @xmath0 does not lie on @xmath8 or any reducible fibre , then for each @xmath46 , a sequence of best @xmath227-approximation to @xmath0 is clearly contained in the component @xmath54 of @xmath208 , and certainly a sequence of best @xmath208-approximation is also contained in @xmath54 .
thus , for any ample @xmath3 , @xmath54 is a curve of best @xmath3-approximation to @xmath0 .
if @xmath0 lies on exactly one @xmath224 or @xmath225 ( and not @xmath8 ) , then the same analysis shows that this same @xmath224 or @xmath225 is a curve of best @xmath3-approximation to @xmath0 for any ample @xmath3 .
if @xmath0 lies on @xmath8 but no @xmath224 or @xmath225 , then notice that @xmath260 for any @xmath46 , so we conclude that @xmath8 is a curve of best @xmath227-approximation to @xmath0 for all @xmath46 .
since @xmath261 , and since @xmath262 contracts no curves through @xmath0 , we conclude from theorem [ projlines ] that @xmath8 and @xmath54 are both curves of best @xmath263-approximation to @xmath0 , for any @xmath46 . since any ample divisor @xmath8 is either a positive linear combination of elements of the set @xmath264 , or a positive linear combination of elements of the set @xmath265 , we conclude from corollary [ divsum ] that either @xmath8 or @xmath54 is a curve of best @xmath3-approximation to @xmath0 for any ample @xmath3 .
if @xmath266 for some @xmath10 , then a similar analysis shows that for any ample divisor @xmath3 , either @xmath8 or @xmath224 is a curve of best @xmath3-approximation to @xmath0 . in particular , if we divide the nef cone of @xmath1 into two subcones according to which of @xmath8 or @xmath224 has smaller degree , then it is straightforward to check that the curve of smaller degree has degree zero or one with respect to each generator of the corresponding subcone , and so by theorem [ projlines ] and corollary [ divsum ] , we conclude that the curve of smaller @xmath3-degree is always a curve of best @xmath3-approximation to @xmath0 .
if @xmath267 for some @xmath10 , a similar calculation shows that either @xmath224 or @xmath225 whichever has smaller @xmath3-degree is always a curve of best @xmath3-approximation to @xmath0 .
theorem [ simplefibres ] can be used to show that any @xmath5-split rational surface of picard rank at most three satisfies conjecture [ ratcurve ] . [
three ] let @xmath1 be a @xmath5-split rational surface , of picard rank at most three .
then conjecture [ ratcurve ] is true for @xmath1 .
_ proof : _ if the picard rank is one , then @xmath1 is isomorphic to @xmath125 .
if the picard rank is two , then either it s geometrically minimal , or else it s @xmath125 blown up at a point @xmath0 . in this case , however , @xmath268 , and the result follows from corollary [ minimal ] .
if the picard rank of @xmath1 is three , then it must be the blowup of some @xmath199 at some point @xmath0 , with @xmath200 .
if @xmath269 , then since any blowup of @xmath201 is also a blowup of @xmath174 , we may instead take @xmath270 .
if @xmath271 , then we can apply theorem [ simplefibres ] directly .
thus , we assume that @xmath270 , and therefore that @xmath1 is the blowup of @xmath125 at two different points .
we begin by finding generators for the effective cone of @xmath1 .
let @xmath272 be the blowing down map , and let @xmath273 and @xmath274 be the two exceptional divisors .
let @xmath8 be the strict transform of the line in @xmath125 which joins the two blown up points .
if we write @xmath275 , then @xmath51 is the class of @xmath276 .
then @xmath277 and @xmath278 correspond to morphisms @xmath279 and @xmath280 to @xmath62 .
we have the following well known description of the effective and nef cones of @xmath1 .
the curves @xmath8 , @xmath273 , and @xmath274 generate the closed cone @xmath281 of effective divisors on @xmath1 , and the classes @xmath51 , @xmath277 , and @xmath278 generate the nef cone of @xmath1 .
we now prove conjecture [ ratcurve ] for @xmath1 .
let @xmath41 be any point , and let @xmath3 be any ample divisor .
if @xmath282 , let @xmath176 be the unique irreducible component of the fibre of @xmath283 through @xmath0 .
each of @xmath51 , @xmath284 , and @xmath285 either contracts @xmath176 , or else contracts no curves through @xmath0 and maps @xmath176 to a line .
( note that if @xmath0 lies on @xmath8 , then @xmath286 . )
thus , by theorem [ projlines ] and corollary [ divsum ] , we conclude that @xmath176 is a curve of best @xmath3-approximation to @xmath0 for any @xmath3 in the cone generated by @xmath51 , @xmath284 , and @xmath287 .
this covers all cases except @xmath266 and @xmath3 is a positive linear combination of @xmath51 , @xmath284 , and @xmath287 . in this case ,
notice that by theorem [ product ] , both @xmath8 and @xmath224 are curves of best approximation to @xmath0 with respect to @xmath288 and @xmath289 . since @xmath51 contracts @xmath224 , this means that if @xmath3 lies in the cone generated by @xmath51 , @xmath288 , and @xmath289 , then @xmath224 is a curve of best @xmath3-approximation to @xmath0 by corollary [ divsum ] . and
since @xmath284 and @xmath287 both contract @xmath8 , we conclude again by corollary [ divsum ] that @xmath8 is a curve of best @xmath3-approximation to @xmath0 if @xmath3 is a positive linear combination of @xmath284 , @xmath288 , @xmath287 , and @xmath289 .
this concludes the proof of conjecture [ ratcurve ] for @xmath1 .
proceeding to the other extreme from theorem [ simplefibres ] , we now prove a theorem about the case in which the map to @xmath62 has exactly one reducible fibre .
[ onefibre ] let @xmath1 be a smooth rational surface obtained by a succession of blowups of the hirzebruch surface @xmath199 , and let @xmath290 be the associated map .
assume that @xmath211 has only one reducible fibre , with @xmath96 components , and assume @xmath291 .
if every multiple component of the reducible fibre intersects at least two other components then conjecture [ ratcurve ] is true for every point of @xmath1 . _
proof : _ if any of the blown - up points of @xmath199 lie on the @xmath223-section @xmath8 , then the strict transform of @xmath8 will have strictly smaller self - intersection than @xmath212 , and @xmath1 can be obtained as a blowup of @xmath292 for some @xmath293 .
thus , if we choose @xmath57 large enough , we may therefore assume that none of the blown - up points of @xmath199 lie on @xmath8 .
let us establish some notation : * the components of the reducible fibre of @xmath211 are denoted by @xmath294 , and we write @xmath295 for the unique @xmath223-section of @xmath211 . * the classes @xmath296 are a basis for the nron - severi group @xmath297 , and we denote by @xmath298 the dual basis with respect to the intersection pairing .
that is , @xmath299 .
* we write @xmath300 , so that @xmath301 is the multiplicity of @xmath224 as a component of the reducible fibre of @xmath211 .
we proceed with a technical lemma .
[ inductivestep ] assume without loss of generality that @xmath302 and @xmath303 , and let @xmath304 be the map which blows down @xmath305 .
let @xmath306 be the map satisfying @xmath307 , let @xmath308 be the @xmath223-section of @xmath89 , and let @xmath309 be the components of the unique reducible section of @xmath89 , ordered so that @xmath310 .
let @xmath311 be the dual basis to @xmath312 .
then @xmath313 if @xmath314 .
furthermore , @xmath305 can intersect either one or two other components @xmath224 .
if @xmath305 intersects one component @xmath224 , then @xmath315 . if @xmath305 intersects @xmath224 and @xmath316 , then @xmath317 .
_ proof of lemma : _ the fact that @xmath305 can intersect at most two components @xmath224 follows from the fact that no point of @xmath2 can lie on more than two components of the reducible fibre .
the @xmath177 are defined by the property that @xmath299 , where we define @xmath295 .
it is a straightforward matter to verify this equality for all @xmath10 and @xmath241 .
the next step is to compute the effective cone of @xmath1 .
[ effectcone ] assume that for all @xmath10 , some multiple of the divisor @xmath177 is basepoint free .
then the effective cone of @xmath1 is generated by the components @xmath318 of the reducible fibre and the @xmath223-section @xmath8 of @xmath211 .
furthermore , we have @xmath319 and @xmath320
. _ proof of lemma : _ let @xmath321 be the set of components of the reducible fibre , where @xmath273 is the unique component which intersects @xmath8 .
let @xmath322 be the intersection matrix of the @xmath8 and @xmath224 ; that is , let @xmath323 , where @xmath295 .
the only entry of @xmath287 which is not @xmath324 or @xmath325 is @xmath326 .
we therefore may regard @xmath287 as a specialization of the matrix @xmath327 with entries in @xmath328 $ ] , where @xmath329 except that @xmath330 .
specializing @xmath331 reduces @xmath327 to @xmath287 . to check that @xmath332 generates the effective cone , it suffices to check that the dual cone is the nef cone .
since @xmath332 form a basis for the vector space @xmath333 , the dual cone will be generated by the vectors @xmath334 , where @xmath335 is the inverse of the matrix @xmath327 , @xmath336 , and @xmath295 . to show that this dual is the ample cone , it suffices to show that every positive linear combination of the @xmath337 is ample . by the nakai - moishezon criterion ,
since we assume that some multiple of each @xmath177 is basepoint free , it suffices to show that @xmath338 for all @xmath10 and @xmath241 , and that for each @xmath10 , there is some @xmath241 for which @xmath339 . first , note that @xmath319 , the class of a fibre of @xmath211 . to see this , note that @xmath340 for all @xmath341 , and therefore @xmath342 for some @xmath343 by the non - degeneracy of the intersection pairing .
since @xmath344 , we conclude that @xmath319 .
in particular , the first row and column of @xmath345 are independent of @xmath75 , and @xmath346 for all @xmath347 .
now let @xmath341 .
since the only entry of @xmath327 depending on @xmath75 is @xmath330 , cramer s rule implies that the @xmath10th row @xmath348 of @xmath327 is @xmath349 , where @xmath350 and @xmath351 are independent of @xmath75 . specializing to @xmath331
, we obtain @xmath352 .
( note that @xmath353 for all @xmath347 , so that the @xmath354-cofactor matrix of @xmath327 has nontrivial kernel , and therefore zero determinant . we therefore conclude that the determinant of @xmath327 is @xmath355 , and in particular independent of @xmath75 . )
furthermore , we know that if @xmath356 denotes that @xmath241th column of @xmath357 , then @xmath358 is independent of @xmath75 for all @xmath241 , and hence @xmath359 for all @xmath347 . thus , @xmath360 for some @xmath361 . clearly @xmath320 , since @xmath362 , @xmath363 if @xmath364 , and @xmath365 .
thus , @xmath366 . furthermore , since @xmath367
, this means that @xmath368 is just @xmath57 times the multiplicity of @xmath316 in the fibre @xmath208 in particular , @xmath369 is a positive integer .
it therefore remains only to show that the coefficient of @xmath316 in @xmath370 is at least @xmath371 , and that at least one coefficient is strictly greater than @xmath371 .
( recall that @xmath96 is the number of components in the reducible fibre . )
we proceed by induction on @xmath96 .
if @xmath372 , the result follows immediately from the preceding calculations , since @xmath373 and @xmath172 will be the full list of @xmath177 s . for general @xmath96 , note that there must be some @xmath224 with @xmath374 and @xmath375 without loss of generality , we may assume that @xmath376 .
let @xmath304 be the map that blows down @xmath305 .
then @xmath2 is a smooth rational surface which admits a map @xmath377 such that @xmath378 , and the unique singular fibre of @xmath89 has @xmath379 components .
let @xmath380 for @xmath381 , and let @xmath382 .
denote the dual basis of @xmath383 by @xmath384 . by induction ,
the coefficient of @xmath385 in @xmath386 is at least @xmath387 , since the multiplicity of @xmath385 in @xmath388 is also @xmath369 . if @xmath381 , then by lemma [ inductivestep ] , the coefficient of @xmath316 in @xmath370 is also at least @xmath389 . if @xmath376 , then the coefficient of @xmath316 in @xmath177 is at least @xmath390 .
thus , if @xmath391 , then @xmath392 for all @xmath10 and @xmath241 . since the pairing is non - degenerate , for each @xmath10 there is some @xmath241 for which @xmath339 .
this concludes the proof of the lemma .
[ strictremark ] note that the proof of lemma [ effectcone ] also shows that @xmath339 for all @xmath10 and @xmath241 except @xmath393 .
the next step is to prove that for each @xmath10 , some multiple of @xmath177 is basepoint free .
together with lemma [ effectcone ] , this will complete the computation of the effective cone of @xmath1 .
define a graph @xmath394 with vertex set @xmath395 and edge set @xmath396 . since @xmath397 , this defines a simple graph , and moreover , it is a tree .
root this tree at @xmath273 ; that is , define a partial order on @xmath90 by @xmath398 if and only if there is a simple path ( that is , a path with no repeated vertices ) from @xmath273 to @xmath316 which contains the vertex @xmath224 . call @xmath224 a leaf of @xmath399 if and only if @xmath398 implies @xmath400 .
the hypothesis of the theorem is equivalent to demanding that the leaves of @xmath399 have multiplicity one in the fibre @xmath208 .
[ leafbasepointfree ] let @xmath224 be a leaf of @xmath399 .
then @xmath177 is basepoint free . _
proof : _ we proceed by double induction on @xmath57 and @xmath96 . if @xmath270 there there is nothing to prove . if @xmath401 , then @xmath1 is the hirzebruch surface @xmath199 and the lemma is clear . if @xmath402 , then we must have @xmath372 , and the result is again clear .
note also that if @xmath403 , then the sum of the multiplicities of the @xmath224 with @xmath374 is always at least two , and that this property will be preserved by blowing up . now consider a general @xmath404 , and let @xmath224 be a leaf of @xmath399 .
since @xmath224 has multiplicity one by hypothesis , there must be some other @xmath405-curve @xmath316 .
( it is possible that @xmath406 . )
let @xmath407 be the map that blows down @xmath316
. then @xmath2 will also be a rational surface with a map @xmath306 with at most one reducible fibre .
the unique @xmath408-section of @xmath89 will satisfy @xmath409 , and the reducible fibre will have @xmath379 components . therefore , by induction , the dual divisor @xmath410 to @xmath411 is basepoint free .
but @xmath412 , so @xmath177 is basepoint free as well .
[ multiplegens ] assume that @xmath413 .
then @xmath414 is an effective divisor supported on the reducible fibre of @xmath211 .
furthermore , if @xmath224 and @xmath316 intersect , then @xmath415 .
_ proof : _ since @xmath416 , it follows that @xmath414 is supported on the reducible fibre of @xmath211 .
thus , it remains only to show that it is effective .
it suffices to prove the result in the case that @xmath417 ; a simple induction will prove the general case from there .
furthermore , if we blow down a @xmath405-curve @xmath418 on @xmath1 to obtain @xmath2 , then by lemma [ inductivestep ] , the truth of the lemma for @xmath224 and @xmath316 on @xmath1 will be equivalent to the truth of the lemma for the images of @xmath224 and @xmath316 on @xmath2 , unless @xmath10 or @xmath241 equals @xmath248 .
thus , by induction on the number @xmath96 of components of the reducible fibre , we may assume that @xmath224 or @xmath316 is a @xmath405-curve on @xmath1 , that any @xmath405-curve in the reducible fibre is either @xmath224 or @xmath316 , and that @xmath375 ( because @xmath419 and @xmath420 ) .
say @xmath224 is a @xmath405-curve of multiplicity one .
then it must be a leaf of @xmath399 , so @xmath316 must be the unique other component which intersects @xmath224 . in that case , @xmath316 must also have multiplicity one , so let @xmath2 be the surface obtained by blowing down @xmath224 . by applying lemma [ inductivestep ]
, we see that @xmath421 , and the lemma follows .
now assume that @xmath224 is a @xmath405-curve of multiplicity @xmath422 .
then by hypothesis @xmath224 can not be a leaf , so it must intersect two curves @xmath316 and @xmath418 , with @xmath423 .
we have @xmath424 , so let @xmath304 be the map that blows down @xmath224 . then @xmath2 is also a rational surface with @xmath223-section and a single reducible fibre with @xmath379 components , so @xmath425 are generators of the effective cone of @xmath2 , with @xmath382 and @xmath426 for any @xmath427 .
let @xmath311 be the dual basis of @xmath428 in @xmath429 .
by lemma [ inductivestep ] , we have @xmath430 . by induction on @xmath96
, we have : @xmath431 since the multiplicity of @xmath432 is the same as the multiplicity of @xmath433 , for any @xmath434 . by lemma [ inductivestep ] again , pulling back to @xmath1 via @xmath97 gives : @xmath435 and therefore : @xmath436 as desired .
finally , we treat the case that @xmath316 is the only @xmath405-curve in the reducible fibre .
it is not a leaf , since @xmath413 . if @xmath406
, then since @xmath437 has multiplicity one , then since @xmath403 ( @xmath438 ) , there must be at least one other @xmath405-curve
@xmath439 on @xmath1 .
therefore @xmath440 , and let @xmath224 and @xmath418 be the two components which intersect @xmath316 .
( note that @xmath440 because @xmath441 . )
we have @xmath442 , @xmath443 , and @xmath444 . by induction and lemma [ inductivestep ] ,
we compute : @xmath445 and therefore : @xmath446 and the lemma is proven .
[ basepointfree ] for any @xmath10 , some positive multiple of @xmath177 is basepoint free .
_ proof of lemma : _ we have already proven this in the case that @xmath447 ( proof of lemma [ effectcone ] ) or @xmath224 is a leaf ( see lemma [ leafbasepointfree ] ) . by lemma [ multiplegens ] ,
we know that some multiple of each @xmath177 is effective , and in particular , the base locus of @xmath177 is supported on the set @xmath448 .
it therefore suffices to find an effective divisor @xmath184 , linearly equivalent to some multiple of @xmath177 , supported on the set @xmath449 .
we claim that for all @xmath10 , there is an effective divisor @xmath176 and a positive rational number @xmath450 such that @xmath451 , @xmath176 is supported on the set @xmath452 , and the base locus of a suitable multiple of @xmath453 is a subset of @xmath8 .
in particular , we will show that @xmath454 , where @xmath455 denotes the number of edges in the shortest path in the graph @xmath399 from @xmath273 to @xmath224 . if @xmath456 , this is trivial .
we proceed by induction on @xmath455 , the number of edges in the shortest path from @xmath273 to @xmath224 in the rooted tree .
let @xmath316 be the parent of @xmath224 that is , let @xmath316 be the unique component with @xmath457 and @xmath458 .
then by induction we can write @xmath459 , where @xmath460 is effective and supported on the set @xmath461 , and the base locus of a suitable multiple of @xmath462 is a subset of @xmath8 . by lemma [ multiplegens ] ,
we can write : @xmath463
we therefore deduce : @xmath464 set @xmath465 .
since @xmath466 , it follows that @xmath467 .
we have @xmath468 for all @xmath10 , so a suitable multiple of @xmath469 will be basepoint free away from @xmath8 , as desired .
this means that for each @xmath10 , the base locus of a suitable multiple of @xmath177 is contained in the union of @xmath8 and @xmath470 .
but we have already proven that the base locus of a suitable multiple of @xmath177 is contained in the union @xmath448 .
since these two sets are disjoint , we conclude that a large enough multiple of @xmath177 is basepoint free , as desired .
we now conclude the proof of theorem [ onefibre ] .
choose any @xmath471 .
[ vertexproof ] for any @xmath10 , a sequence of best @xmath177-approximation to @xmath0 can be found on the fibre of @xmath211 through @xmath0 ( this includes possibly along some @xmath316 ) , or else on @xmath8 . _
solution : _ if @xmath447 , then @xmath472 , so the claim is trivial .
if @xmath456 , then @xmath473 , and against the claim is clearly true . for general @xmath10 , if @xmath0 lies on a curve @xmath54 with @xmath474 , then clearly @xmath54 will be a curve of best @xmath177-approximation to @xmath0 .
however , such a curve @xmath54 must be linearly equivalent to a nonnegative linear combination of @xmath475 , where as usual @xmath295 . if @xmath54 is not @xmath8 or some @xmath224 , then @xmath54 must be nef , and so must be a positive linear combination of the @xmath177 . since @xmath291 , remark [ strictremark ] shows that this is impossible . thus ,
if @xmath474 , then @xmath54 is either @xmath8 or some @xmath224 for @xmath476 .
assume that @xmath224 is a leaf with @xmath477 .
then @xmath177 is basepoint free , and @xmath478 , so if @xmath479 is the morphism associated to @xmath177 , then @xmath480 maps the fibre of @xmath211 through @xmath0 to a line . by theorem [ projlines ] , this means that a sequence of best @xmath177-approximation to @xmath0 either lies along the fibre of @xmath211 through @xmath0 , or else along some irreducible curve @xmath54 with @xmath474 , which as noted must either be @xmath8 or some @xmath224 .
if @xmath224 is not a leaf , then there is some @xmath241 for which @xmath316 is a leaf and @xmath398 . by lemma [ multiplegens ]
, we can write @xmath481 .
if @xmath0 does not lie on any @xmath439 with @xmath482 , then theorem [ addeffective ] implies that either the fibre of @xmath211 through @xmath0 or @xmath8 is a curve of best @xmath177-approximation to @xmath0 . if @xmath0 lies on some @xmath439 for @xmath483 , then @xmath439 is a curve of best @xmath177-approximation to @xmath0 because @xmath484 and some multiple of @xmath177 is basepoint free .
if @xmath0 lies on @xmath224 but not on any other @xmath439 , then as in the proof of lemma [ basepointfree ] , we can write @xmath485 , where @xmath46 is a positive rational number and @xmath184 is an effective divisor whose support does not include @xmath224 .
since @xmath224 is a curve of best @xmath208-approximation to @xmath0 , lemma [ multiplegens ] implies that @xmath224 is also a curve of best @xmath177-approximation to @xmath0 . in all cases ,
the claim is proven .
the claim , together with corollary [ divsum ] , shows that if @xmath3 is any ample divisor on @xmath1 , then @xmath3 is a nonnegative linear combination of the @xmath177 , and so a sequence of best @xmath3-approximation to @xmath0 can be found on the fibre of @xmath211 which contains @xmath0 , unless @xmath0 lies on @xmath8 .
if @xmath0 lies on @xmath8 , then @xmath8 is a curve of best @xmath177-approximation to @xmath0 for all @xmath10 , and so @xmath8 is also a curve of best @xmath3-approximation to @xmath0 .
this concludes the proof of theorem [ onefibre ] .
these same techniques will also prove conjecture [ ratcurve ] for split rational surfaces of picard rank four . [
four ] conjecture [ ratcurve ] is true for any split rational surface @xmath1 of picard rank four . _
proof : _ every such rational surface is a blowup of a hirzebruch surface , and so @xmath1 admits a surjective morphism @xmath290 whose generic fibre is irreducible .
the nron - severi group has rank four , generated by the class @xmath208 of a fibre of this morphism , the class @xmath8 of a section , and two more classes , corresponding to irreducible components of reducible fibres , or equivalently , exceptional divisors of the blowing down map to some hirzebruch surface .
there are three possible configurations of reducible fibres ( see table [ fibretable ] ) : .configurations of reducible fibres [ cols="^,^,^ " , ] let @xmath0 be any point on @xmath1 .
we first assume that @xmath0 does not lie on any of the 16 @xmath405-curves on @xmath1 . in that case
, we will show that for any ample divisor @xmath3 , a sequence of best @xmath3-approximation to @xmath0 lies along a curve in the class @xmath486 for some @xmath10 , or in the class @xmath487 for some @xmath10 . to do this , we introduce 10 subcones of the nef cone of @xmath1 .
for @xmath10 between 1 and 5 , we define the cone @xmath488 to be the cone of ample divisors for which the divisor @xmath489 has minimal degree amongst the classes @xmath490 and @xmath460 , and we define the cone @xmath491 to be the cone of ample divisors for which the divisor @xmath176 has minimal degree amongst the classes @xmath490 and @xmath460 .
we will show that for every divisor in @xmath488 , a sequence of best approximation to @xmath0 lies along a curve in @xmath487 , and similarly that for any divisor in @xmath491 , a sequence of best approximation to @xmath0 lies along a curve in @xmath486 .
we first consider @xmath488 . for simplicity , we will consider only @xmath492 ; the other four cases differ from this one only by a permutation of the indices .
the cone @xmath492 has 26 generators : @xmath493 where in all cases we assume that @xmath375 and @xmath440 . of these generators , only @xmath494 is not big .
let @xmath208 be the unique curve through @xmath0 satisfying @xmath495 .
since we have assumed that @xmath0 does not lie on a @xmath405-curve , it follows that if a curve through @xmath0 has intersection zero with one of these generators , then the curve in question must be @xmath208 .
it suffices to show that for any generator of @xmath492 , a sequence of best approximation to @xmath0 lies on @xmath208 . for @xmath51 , @xmath494 , @xmath496 , @xmath497 , @xmath498 , and @xmath499 ,
this follows from theorem [ projlines ] because @xmath208 has degree one or zero . in all other cases ,
@xmath208 has degree two , and we can write them as follows : @xmath500 for each generator @xmath3 in the above list , we have written @xmath501 , where @xmath478 .
thus , by theorem [ product ] , @xmath208 contains a sequence of best @xmath3-approximation to @xmath0 unless @xmath0 lies on some curve of smaller @xmath3-degree .
since we have assumed that @xmath0 does not lie on a @xmath405-curve , and since @xmath3 lies in @xmath492 , it follows that @xmath0 can not lie on any curve of smaller @xmath3-degree than @xmath494 .
( note that the only candidates for a curve of smaller degree than two are those which intersect one of @xmath172 or @xmath173 trivially , and such curves are either @xmath405-curves or else of the form @xmath489 or @xmath176 for some @xmath10 . )
we conclude that a sequence of best @xmath3-approximation to @xmath0 lies on @xmath208 . now to @xmath491 .
again , we will consider only @xmath502 , the cone of divisors with respect to which @xmath181 has minimal degree amongst the @xmath490 and @xmath460 .
like @xmath492 , @xmath502 has 26 generators : @xmath503 where again we assume that @xmath375 and @xmath440 .
let @xmath208 be the curve in @xmath504 containing @xmath0 .
as before , it suffices to show that for any generator of @xmath502 , a sequence of best approximation to @xmath0 lies on @xmath208 . for @xmath181 , @xmath505 , @xmath506 , @xmath507 , and @xmath508 ,
this follows from theorem [ projlines ] because @xmath208 has degree one or zero , and all the generators contract no curves through @xmath0 other than @xmath208 . for the remaining nine generators
, we can write : @xmath509 in each case , we have written the generator @xmath3 as @xmath501 , where @xmath478 .
thus , by theorem [ product ] , we conclude that since @xmath0 does not lie on a @xmath405-curve , a sequence of best @xmath3-approximation to @xmath0 lies on @xmath208 for any generator @xmath3 of @xmath502 , and by corollary [ divsum ] also for any @xmath3 in @xmath502 .
it remains only to treat the case where @xmath0 lies on a @xmath405-curve .
since no three @xmath405-curves meet at a point , @xmath0 lies either on exactly one or exactly two @xmath405-curves .
let s first assume that @xmath0 lies on exactly one @xmath405-curve .
since the automorphism group of @xmath1 acts transitively on the set of @xmath405-curves ( see for example theorem 2.1 of @xcite ) , we may assume that this curve is @xmath184 .
consider the cone @xmath488 .
we will subdivide @xmath488 into two subcones .
the first is @xmath510 , generated by the following twenty - two divisor classes : @xmath511 where @xmath241 and @xmath248 are assumed to be different from @xmath10 .
the classes @xmath51 , @xmath489 , @xmath512 , @xmath513 , @xmath514 , and @xmath515 all either contract the component @xmath225 of @xmath489 through @xmath0 , or else contract no curves through @xmath0 and map @xmath225 to a line .
thus , by theorem [ projlines ] @xmath225 is a curve of best approximation to @xmath0 with respect to any of these divisor classes . for @xmath514 , we see that @xmath516 , and so theorem [ product ] implies that if @xmath225 is not a curve of best @xmath517-approximation to @xmath0 , then a sequence of best approximation to @xmath0 can be found on a curve contracted by either @xmath51 or @xmath460 .
the only such curve through @xmath0 is @xmath184 , and we check that @xmath518 , so @xmath225 is a curve of best @xmath517-approximation to @xmath0 ( as is @xmath184 ) .
this only leaves @xmath519 , which can be rewritten @xmath520 for any @xmath241 . since @xmath521 for any @xmath522
, we may apply theorem [ product ] to argue that if @xmath225 is not a curve of best @xmath523-approximation to @xmath0 , then by theorem [ product ] a sequence of best approximation to @xmath0 must lie on some curve contracted by @xmath51 , @xmath490 , or @xmath460 .
the only such curves through @xmath0 are @xmath184 and the component of @xmath490 through @xmath0 .
none of these have @xmath523-degree less than @xmath123 ( which is the degree of @xmath225 ) , so we conclude that @xmath225 is a curve of best @xmath523-approximation to @xmath0 . thus , by corollary [ divsum ] , it follows that @xmath225 is a curve of best @xmath287-approximation to @xmath0 for any @xmath287 in @xmath510 .
if @xmath287 lies in @xmath488 but not @xmath510 , then it must lie in the cone @xmath524 , generated by : @xmath525 where @xmath241 and @xmath248 range over every index different from @xmath10 .
all but the last two divisors in this list either contract @xmath184 , or else contract no curves through @xmath184 and map @xmath0 to a line . for the last two , we see that @xmath184 has the same degree as @xmath225 , and we already know that @xmath225 is a curve of best @xmath523- and @xmath517-approximation to @xmath0 .
thus , we conclude that @xmath184 is a curve of best @xmath287-approximation to @xmath0 for every @xmath287 in @xmath524 .
consider the cone @xmath491 .
it is straightforward to check that except for @xmath526 , every generator of @xmath491 either contracts @xmath184 or else contracts no curves through @xmath0 and maps @xmath184 to a line .
we can rewrite @xmath527 for any @xmath522 , and note that @xmath528 , so that by theorem [ product ] , @xmath184 is a curve of best @xmath529-approximation to @xmath0 unless @xmath0 lies on some curve of smaller @xmath529-degree which is contracted by @xmath51 or @xmath176 .
since this is not the case , we conclude by theorem [ projlines ] and corollary [ divsum ] that @xmath184 is a curve of best @xmath287-approximation to @xmath0 for any @xmath287 in @xmath491 .
to finish the proof of conjecture [ ratcurve ] for @xmath1 , we just need to consider the case that @xmath0 is the intersection point of two @xmath405-curves .
up to automorphisms of @xmath1 , we may assume that one of these @xmath405-curves is @xmath184 .
since the only @xmath405-curves intersecting @xmath184 are the @xmath224 , we may assume that @xmath530 for some @xmath10 .
consider the cone @xmath531 , generated by the following divisor classes : @xmath532 where @xmath241 and @xmath248 range over all values different from @xmath10 .
each of these divisors either contracts the component @xmath225 of @xmath489 through @xmath0 , or else contracts no curves through @xmath0 and maps @xmath225 to a line . by theorem [ projlines ] and corollary [ divsum ] ,
we conclude that conjecture [ ratcurve ] is true for any @xmath287 in @xmath531 .
next , consider the cone @xmath533 , generated by : @xmath534 where @xmath241 and @xmath248 range over all indices different from @xmath10 and each other .
thus , the cone @xmath533 is generated by 41 divisor classes .
each of these divisors either contracts @xmath224 , or else contracts no curves through @xmath0 and maps @xmath224 to a line . by theorem [ projlines ] and corollary [ divsum ]
, we conclude that conjecture [ ratcurve ] is true for all @xmath287 in @xmath533 .
finally , if @xmath287 does not lie in @xmath531 or @xmath533 , then it must lie in @xmath535 , generated by : @xmath536 where again @xmath241 and @xmath248 are different from each other and @xmath10 , for a total of 41 divisor classes in the list .
each class either contracts @xmath184 , or else contracts no curves through @xmath0 and maps @xmath184 to a line .
thus , it follows from theorem [ projlines ] and corollary [ divsum ] that conjecture [ ratcurve ] is true for any divisor @xmath287 in @xmath535 .
this concludes the proof of conjecture [ ratcurve ] for @xmath1 .
these techniques will probably be able to prove conjecture [ ratcurve ] for more split rational surfaces , but they will probably not suffice to prove the conjecture for a general rational surface .
for example , let @xmath1 be a blowup of @xmath125 at six @xmath5-rational points in general position , embedded in @xmath124 as a smooth cubic surface .
the family @xmath537 of plane cubic curves passing through the set of blown up points in @xmath125 is a three - dimensional ( projective ) linear subspace of the @xmath538 of plane cubic curves .
the zariski closure @xmath539 of the set of cuspidal cubics is a closed subset of dimension 7 .
therefore , @xmath540 has dimension at least one , so if the six points are chosen to lie on some cuspidal cubic whose cusp is not one of the six blown up points ( which will generically be the case ) , there will be a one - dimensional family @xmath541 of cuspidal cubic curves on @xmath1 .
let @xmath0 be the cusp of a cuspidal cubic @xmath54 , and assume that @xmath0 is @xmath5-rational .
then one can find a sequence on @xmath54 which approximates @xmath0 with constant of approximation @xmath120 .
in particular , no rational curve of minimal degree through @xmath0 is a curve of best approximation to @xmath0 , since that curve will generally be a conic ( @xmath0 can not lie on any of the 27 lines on @xmath1 ) .
this suggests that any proof of conjecture [ ratcurve ] for @xmath1 will be beyond the techniques of this paper .
[ vstrong ] let @xmath1 be a smooth algebraic variety defined over a number field @xmath5 , with canonical divisor @xmath542 .
let @xmath8 be a finite set of places of @xmath5 .
let @xmath51 be a big divisor on @xmath1 , and let @xmath3 be a normal crossings divisor on @xmath1 .
choose height functions @xmath543 and @xmath544 for @xmath542 and @xmath51 , respectively , and define a proximity function @xmath545 for @xmath3 with respect to @xmath8 , where @xmath546 is a local height function for @xmath3 at @xmath14 .
choose any @xmath165 .
then there exists a nonempty zariski open set @xmath547 such that for every @xmath5-rational point @xmath548 , we have the following inequality : @xmath549 assume that @xmath1 is as in the conjecture , and has non - negative kodaira dimension .
let @xmath0 be any fixed @xmath5-rational point , lying on a rational curve @xmath553 , and let @xmath3 be any normal crossings divisor containing @xmath0 .
since @xmath1 has non - negative kodaira dimension , there is a non - empty open set @xmath554 such that there is a constant @xmath17 satisfying @xmath555 for all @xmath556 .
furthermore , if @xmath8 is the set of archimedean places of @xmath5 , we may write : @xmath557 where the implied constant is independent of @xmath37 .
( if @xmath37 lies on the support of @xmath3 , then @xmath558 is infinite , but this will not affect our proof . ) let @xmath94 be the constant of approximation for @xmath0 on @xmath54 . for any @xmath165 ,
let @xmath559 be the intersection of @xmath560 with the non - empty open set provided by conjecture [ vstrong ] .
then for any point @xmath561 , conjecture [ vstrong ] implies : @xmath562 if we choose @xmath563 , then it s clear that any sequence of best approximation to @xmath0 must eventually lie in the complement of @xmath559 . if @xmath1 is a surface , then this complement is a union of curves , in which case the truth of conjecture [ ratcurve ] is clear .
if @xmath1 is of higher dimension , then a simple noetherian induction shows that it suffices to assume that @xmath1 contains no subvariety @xmath90 whose kodaira dimension is negative but whose dimension is at least two .
slightly more generally , we have therefore proven the following theorem : let @xmath90 be a smooth algebraic variety with non - negative kodaira dimension .
let @xmath141 be any rational point with a rational curve @xmath564 on @xmath90 through @xmath0 , and assume that every subvariety of @xmath90 with negative kodaira dimension satisfies conjecture [ ratcurve ] for @xmath0 .
assume that conjecture [ vstrong ] is true for @xmath90 .
then conjecture [ ratcurve ] is true for @xmath0 on @xmath90 .
one can also use our techniques to prove conjecture [ ratcurve ] for certain special points on some varieties of nonnegative kodaira dimension .
for example , let @xmath1 be a smooth quartic in @xmath124 containing a line @xmath51 .
this @xmath1 is a k3 surface , and @xmath51 is a smooth rational curve on @xmath1 with self - intersection @xmath565 . for a general such @xmath1 , the effective cone of @xmath1
is spanned by an elliptic curve @xmath184 and @xmath51 .
( indeed , one can choose @xmath184 so that @xmath566 is a hyperplane section of @xmath1 .
see section 5 of @xcite for details . )
the closure of the nef cone of @xmath1 is therefore spanned by the divisors @xmath184 and @xmath567 .
the former corresponds to a morphism from @xmath1 to @xmath62 giving @xmath1 the structure of an elliptic surface , and @xmath3 corresponds to the contraction of the @xmath565 curve @xmath51 .
let @xmath0 be a point on @xmath51 .
then clearly any sequence of best @xmath3-approximation to @xmath0 lies on @xmath51 , and by theorem [ ratapprox ] , a sequence of best @xmath568-approximation to @xmath0 also lies along @xmath51 .
therefore , for any ample divisor @xmath287 in the positive span of @xmath3 and @xmath569 , a sequence of best @xmath287-approximation to @xmath0 must lie along @xmath51 , by theorem [ divsum ] . | in this paper , we examine how well a rational point @xmath0 on an algebraic variety @xmath1 can be approximated by other rational points .
we conjecture that if @xmath0 lies on a rational curve , then the best approximations to @xmath0 on @xmath1 can be chosen to lie along a rational curve .
we prove this conjecture for a wide range of examples , and for a great many more examples we deduce our conjecture from vojta s main conjecture .
[ section ] [ thm]lemma [ thm]definition [ thm]corollary [ thm]conjecture [ thm]claim [ thm]example [ thm]remark p v o |
in @xcite , baker and norine developed a theory of divisors on finite graphs in which they uncovered many parallels between finite graphs and riemann surfaces .
in particular , they stated and proved a graph theoretical analogue of the classical riemann - roch theorem .
see @xcite for background on the interplay between divisors on graphs , curves and tropical curves . as observed in @xcite , there is also close connection between divisor theory and the chip - firing game of bjrner - lovsz - shor @xcite .
see @xcite for the connection to the abelian sandpile model and biggs dollar game @xcite .
an important parameter associated to a connected graph @xmath1 , in the context of divisor theory , is the ( divisorial ) gonality @xmath0 of @xmath1 .
this is the smallest degree of a positive rank divisor . in terms of the chip - firing game , the gonality
can then be expressed has at least 3 nodes . ] as @xmath2 , where @xmath3 is the maximum number of chips that can be placed on the graph @xmath4 such that adding a chip at an arbitrary node will result in a finite game .
there are only few graph classes for which the gonality is known .
examples include trees , complete multipartite graphs and @xmath5 grid graphs , see @xcite .
a conjectured upper bound is @xmath6 matching the classical brill - noether bound , see @xcite . in @xcite
, it was show that the treewidth of @xmath1 is a lower bound on @xmath0 , as conjectured in @xcite . in @xcite , a lower bound @xmath7
is given in terms of the smallest nonzero eigenvalue @xmath8 of the laplacian and the maximum degree @xmath9 .
for the erds - rnyi graph , it is known that asymptotically almost surely @xmath10 ( provided that @xmath11 ) , see @xcite . in this note
we show that computing the gonality of a graph is np - hard , see theorem [ maintheorem ] .
let @xmath4 be a connected graph .
we allow graphs to have parallel edges , but no loops .
for @xmath12 , we denote by @xmath13 the set of edges with an end in @xmath14 and an end in @xmath15 and by @xmath16:=e[a , a]$ ] the set of edges with both ends in @xmath14 . for distinct nodes
@xmath17 , we use the abbreviation @xmath18 for the set of edges between @xmath19 and @xmath20 . by @xmath21
we denote the degree of a node @xmath19 .
the _ laplacian _ of @xmath1 is the matrix @xmath22 defined by @xmath23 a vector @xmath24 is called a _ divisor _ on @xmath1 and @xmath25 is its _
degree_. a divisor @xmath26 is _ effective _ if @xmath27 .
two divisors @xmath26 and @xmath28 are _ equivalent _ , written @xmath29 , if there is an integer vector @xmath30 such that @xmath31 .
this is indeed an equivalence relation and equivalent divisors have equal degrees as the entries in every column of @xmath32 sum to zero .
let @xmath26 be a divisor .
if @xmath26 is equivalent to an effective divisor , then we define @xmath33 if @xmath26 is not equivalent to an effective divisor , we set @xmath34 .
observe that equivalent divisors have the same rank .
answering a question of h.w .
lenstra , it was shown in @xcite that computing the rank of a divisor is np - hard .
finally , we define the _ gonality _ of @xmath1 to be @xmath35 observe that in the definition of gonality , we can restrict ourselves to _ effective _ divisors @xmath26 .
hence @xmath0 is the minimum degree of an effective divisor @xmath26 such that for every node @xmath20 there is an effective divisor @xmath36 such that @xmath37 .
also note that @xmath38 since taking @xmath39 for all @xmath40 gives a divisor of positive rank . to facilitate reasoning about equivalence of effective divisors ,
we denote by @xmath41 the incidence vector of a subset @xmath42 of @xmath43 .
when @xmath26 and @xmath28 are effective divisors and @xmath44 , we say that @xmath28 is obtained from @xmath26 by firing on the set @xmath42 .
if we think of @xmath45 as the number of chips on a node @xmath20 , then firing on @xmath42 corresponds to moving one chip along each edge of the cut @xmath46 in the direction from @xmath42 to @xmath47 . in particular
, we must have that @xmath48 for every @xmath49 as @xmath50 .
hence , we can not fire on any set @xmath42 for which the cut @xmath46 has more than @xmath51 edges .
the following lemma from @xcite shows that for equivalent effective divisors @xmath26 and @xmath28 , we can obtain @xmath28 from @xmath26 by successively firing on sets .
[ chain ] let @xmath26 and @xmath28 be equivalent effective divisors satisfying @xmath52 . then there is a chain of sets @xmath53 such that @xmath54 is effective for every @xmath55 and such that @xmath56 . for an effective divisor @xmath26 on @xmath1 , we define the following equivalence relation @xmath57 on @xmath43 : @xmath58 an edge @xmath59 for which the ends are equivalent is called _
@xmath26-blocking_. note that if @xmath26 and @xmath28 are equivalent effective divisors , then @xmath60 . we need the following observations .
[ equiv ] let @xmath27 be a divisor on @xmath1 and let @xmath17
. then @xmath61 if and only if for some effective divisor @xmath36 we can fire on a subset @xmath42 with @xmath62 , @xmath63 . in particular , @xmath64 if every @xmath19@xmath20 cut has more than @xmath51 edges .
this follows directly from lemma [ chain ] . [ blocking ]
let @xmath27 be a divisor on @xmath4 .
let @xmath65 be the set of @xmath26-blocking edges and let @xmath42 be a component of @xmath66 .
then for every effective divisor @xmath36 we have @xmath67 .
this follows as no chips can move along a @xmath26-blocking edge .
to prove hardness of computing @xmath0 , we use the following construction .
let @xmath4 be a graph .
let @xmath68 .
construct a graph @xmath69 in the following way .
start with a single node @xmath70 .
for every @xmath40 add three nodes : @xmath71 . for every edge @xmath72 , make two nodes @xmath73 and @xmath74 .
the edges of @xmath69 are as follows . for every @xmath72 , add an edge between @xmath73 and @xmath74 , add @xmath75 parallel edges between @xmath19 and @xmath73 and @xmath75 parallel edges between @xmath74 and @xmath20 . for every @xmath40 , add three parallel edges between @xmath76 and @xmath77 , @xmath75 parallel edges between @xmath20 and @xmath76 and @xmath75 parallel edges between @xmath77 and @xmath70 . observe that @xmath75 is equal to the number of nodes of @xmath69 plus one .
see figure [ f1 ] for an example .
we first show that @xmath80 .
let @xmath26 be an effective divisor on @xmath69 with @xmath81 .
consider the equivalence relation @xmath82 on @xmath83 . clearly , @xmath84 .
hence , by lemma [ equiv ] , the @xmath75-fold edges are @xmath26-blocking .
it follows that @xmath70 is equivalent to all nodes @xmath85 , and every node @xmath40 is equivalent @xmath76 and to all nodes @xmath74 . by lemma
[ blocking ] , the numbers of chips on each of the sets @xmath86 , @xmath87 , @xmath88 and @xmath89 are constant over all effective @xmath36 .
hence , form the fact that @xmath90 it follows that now we show that equality can be attained . for this ,
let @xmath104 be an independent set in @xmath1 of size @xmath105 .
take the partition @xmath106 where @xmath107 and @xmath108 are singletons .
every edge of @xmath1 has endpoints in two distinct sets @xmath109 and @xmath110 with @xmath111 .
we orient the edge from @xmath109 to @xmath110 .
we now define the effective divisor @xmath26 on @xmath69 as follows .
@xmath112 it is easy to check that @xmath113 .
now for every @xmath114 let @xmath115 and let @xmath116 then for every @xmath117 , @xmath118 is effective
. furthermore , @xmath119 for every @xmath120 and @xmath121 for every edge with head in @xmath109 .
it follows that @xmath26 has positive rank .
since computing @xmath105 is np - hard , it follows from lemma [ gonalpha ] that also computing @xmath0 is np - hard . in @xcite
it was shown that computing @xmath105 is in fact apx - hard even on cubic graphs . since for cubic graphs
@xmath1 we have @xmath122 , and @xmath123 , it follows that computing @xmath0 is apx - hard . | the gonality @xmath0 of a graph @xmath1 is the smallest degree of a divisor of positive rank in the sense of baker - norine . in this note
we show that computing @xmath0 is np - hard by a reduction from the maximum independent set problem .
the construction shows that computing @xmath0 is moreover apx - hard . |
since the first inflationary cosmological models were proposed @xcite most calculations have been performed using one of a variety of approximations . while these are usually sufficient and often unavoidable , there is considerable interest in scalar field cosmologies that can be solved exactly . in this paper
we introduce a number of exact solutions for robertson walker universes that contain a single scalar field , @xmath0 , with the potential , @xmath1 potentials of this type characteristically arise when a higher - dimensional theory is compactified to @xmath2-spacetime , including various supergravity and superstring models @xcite . in particular , as zer and taha @xcite point out , the potential @xmath3 where @xmath4 is a constant , is motivated by a perturbation expansion in superstring theories . in scalar field cosmologies with exponential potentials @xmath4 is usually a free parameter and in this paper we discuss solutions with a variety of different values of @xmath4
. however , string theory makes the generic prediction that @xmath5 which must be incorporated into any fully realistic theory .
cosmological models based directly upon the superstring action have been examined by casas , garca - bellido and quirs @xcite .
brustein and steinhardt @xcite demonstrate that there are severe difficulties in implementing a realistic cosmology based on the conventional formulation of superstring theories , but this is not being attempted here .
exact solutions for potentials containing a single exponential term with restrictions either on the initial conditions or the choice of model parameters have been discussed by several authors @xcite .
the full solution was obtained by salopek and bond @xcite and generalised by lidsey to a two - field model @xcite . solutions for potentials of the type @xmath6
have been found by de ritis _ et .
_ @xcite .
powerlaw inflation @xcite , where @xmath7 and @xmath8 , is normally driven by a scalar field with @xmath9 .
we generalise this potential to models where @xmath10 has the form of equation ( [ vgeneral ] ) . in sections 3 and 4
we give several exact solutions in curved spacetime for a potential which has two exponential terms .
the flat spacetime case is discussed in section 5 and here we find a diverse range of models .
for instance , examples where powerlaw growth is preceeded by a period of exponential expansion are presented , as well as models where powerlaw inflation commences after a finite amount of non - inflationary growth .
like virtually all exact solutions , those presented here must be viewed as toy models rather than attempts to construct a realistic cosmology . in this sense ,
the inflationary solutions in this paper are analagous to intermediate inflation @xcite which is based on an unlikely looking potential but has received attention because it can be solved exactly . also , by studying exact solutions which widen the repertoire of possible inflationary behaviour we create new possibilities which can then be incorporated into more complicated models .
in this section we present the equations which govern the evolution of a robertson walker universe containing a single scalar field , @xmath0 , with effective potential @xmath10 . the density , @xmath11 , is @xmath12 expressed in natural units , the einstein field equations for a robertson walker universe , are @xcite @xmath13 where @xmath14 is the scale factor and @xmath15 is the hubble parameter .
the semi - classical equation of motion for @xmath0 is @xmath16 differentiating @xmath11 and comparing the result with equation ( [ eofmotion ] ) gives @xmath17 while taking the time , @xmath18 , as the independent variable is the most natural choice , the mathematical complexity of exact solutions to equations ( [ rhodef ] ) to ( [ eofmotion ] ) is often reduced by parametrising the motion in terms of the field , @xmath0 .
this formalism was introduced by muslimov , salopek and bond @xcite but the form of it used here is due to lidsey @xcite . in this paper
, we will find it useful to make the extra substitution , @xmath19 where @xmath20 is a constant .
the shift between @xmath21 and @xmath18 is obtained by rearranging equation ( [ rhodot ] ) @xmath22 employing equation ( [ drhodx ] ) we find a set of equations , expressing @xmath11 , @xmath23 , @xmath24 , @xmath18 and @xmath25 as functions of @xmath21 . from equation ( [ rhodot ] ) it follows that @xcite @xmath26 where @xmath27 and the dash denotes differentiation with respect to @xmath21 . comparing equations ( [ hdef ] ) and ( [ hxdef ] ) yields a first order differential equation for @xmath28 , @xmath29 the general solution of this equation is @xmath30 } \left\{a_{0}^2 + \frac{6k}{\xi^2}\int_{x_0}^x \exp{\left[\frac{2}{\xi^2 } \int_{x_0}^{x}\frac{\rho}{x^2\rho'}dx\right ] } \frac{dx}{x^2\rho'}\right\ } , \label{chisol}\ ] ] where @xmath31 and @xmath32 and the values of @xmath24 and @xmath21 when @xmath33 . for any given density , @xmath11 ,
the potential is @xmath34 or , equivalently , @xmath35 the time , @xmath36 , is calculated by integrating equation ( [ drhodx ] ) , @xmath37 finally @xmath38 , the ratio between the density and the critical density , is @xcite @xmath39 not all exact scalar field cosmologies are inflationary . during inflation the comoving coordinate volume contained inside
the horizon decreases as the universe expands or , equivalently , @xmath40 @xcite . in terms of the parameter @xmath41
, inflation is taking place when @xmath42 . using equation ( [ drhodx ] ) we write @xmath43 in terms of @xmath21 , @xmath44 the method we use to find exact solutions
is to specify a particular form of the density , @xmath45 , and then run the field equations backwards to derive the potential that produced it .
the choice of @xmath45 is guided by our interest in potentials that are of the general form , equation ( [ vgeneral ] ) .
however , as long as the integrals in equations ( [ chisol ] ) and ( [ vfromrho ] ) can be performed then this method will work for an arbitrary @xmath45 .
while the exact solutions presented here could have been written down without any comment as to how they were found , it will hopefully be helpful to the reader if the methodology is made explicit .
other authors have obtained exact solutions by constraining the field equations in some way and then solving the restricted problem . in particular , ellis and madsen @xcite
derive several exact solutions by specifying the functional form of the scale factor , @xmath14 , and then computing @xmath10 . while the calculation involved here
is similar , they emphasise the desired expansion whereas we are seeking exact solutions when the potential is of the generic form , equation ( [ vgeneral ] ) .
virtually all the exact solutions that are to be found in the literature do not hold over the complete range of initial conditions , and those discussed in this paper are no exception . also , by characterising the motion with the field ,
@xmath0 we are implicitly assuming that the @xmath0 is strictly increasing or decreasing .
when this assumption breaks down , we can describe the solution in a piecewise way .
an example of this is given in section 4 .
we begin by considering models in robertson walker spacetimes with positive curvature ( @xmath46 ) .
zer and taha @xcite give two exact solutions for the potential @xmath47 which they designate solutions i and ii . while their solutions are distinct from one another when the time is chosen as the independent variable , they both have the density , @xmath48 where @xmath49 and @xmath50 depend on the initial conditions . by substituting @xmath45 into equations ( [ chisol ] ) and ( [ vfromrho ] )
we derive the most general potential that produces a density of the form equation ( [ rhopos ] ) , @xmath51 we can carry out this procedure for other choices of @xmath45 and @xmath20 .
however , this case is special in that for other relatively simple forms of @xmath45 in curved spacetime the corresponding potential is extremely complicated . when @xmath52 or @xmath53 the potential , equation ( [ trialv ] ) , simplifies to the form of equation ( [ vpos ] ) .
we consider these two special cases in turn .
first , setting @xmath54 in equation ( [ rhopos ] ) and using equations ( [ chisol ] ) and ( [ timex ] ) gives the following solution , parametrised by @xmath21 : @xmath55 , \label{tax}\end{aligned}\ ] ] where we have identified @xmath56 from equations ( [ vpos ] ) and ( [ vax ] ) .
inverting gives @xmath14 and @xmath57 and generalises solution i of zer and taha .
@xmath58 } , \label{phiat } \\
a(t ) & = & { a_{o}}\sqrt{d{a_{o}}^2{x_o}^2 + \left(\frac{t}{{a_{o } } } + \sqrt{1 - d{a_{o}}^2{x_o}^2}\right)^2}. \label{aat}\end{aligned}\ ] ] the only constraint is the requirement that @xmath59 when @xmath60 , or @xmath61 . setting @xmath62 recovers zer and taha s result . for this solution @xmath63 , equation ( [ epsilondef ] ) ,
is @xmath64 when @xmath65 , @xmath66 and @xmath42 so this solution is always inflationary and non - singular , as zer and taha @xcite point out . at large negative times
, @xmath14 is decreasing towards a finite minimum size , after which it will expand indefinitely . while this is an inflationary solution , it is not asymptotically flat , since @xmath67 and @xmath68 at small @xmath21 , or large times .
further examples of inflationary models where @xmath69 can be found in ellis _ et .
if @xmath70 then @xmath71 at some finite , negative time and the resulting model universe does contain an initial singularity . in this case inflation
never begins , since @xmath72 at all times .
now put @xmath73 , giving the solution @xmath74 } , \label{phibt } \\
a(t ) & = & { a_{o}}\left[\frac{5}{c{a_{o}}^2{x_o } } + \frac{c{x_o}}{5}\left(\frac{t}{\sqrt{2 } } + \frac{5}{c}\sqrt{\frac{c}{5{x_o } } - \frac{1}{{a_{o}}^2{x_o}^2}}\right)^2 \right ] \label{abt}\end{aligned}\ ] ] where @xmath75 . to ensure that @xmath59 , we need @xmath76 . setting @xmath77 recovers solution ii of zer and taha .
this is a non - singular model and at large times @xmath78 and @xmath79 , so the solution approaches powerlaw inflation in flat spacetime .
the next possibility we consider is the existence of exact models in a robertson walker universe with negative curvature ( @xmath80 ) .
we look for the potential that gives a density of the form @xmath81 from equations ( [ chisol ] ) , ( [ vfromrho ] ) and ( [ rhoneg ] ) we obtain @xmath82 choosing either @xmath83 or @xmath53 simplifies the potential , equation ( [ trialvneg ] ) , to the generic form given by equation ( [ vgeneral ] ) . as is the case for positively curved spacetime , other choices of @xmath45 result in a very complicated potential .
putting @xmath53 gives the solution @xmath84 } , \label{phibtneg } \\
a(t ) & = & { a_{o}}\left [ \frac{a{x_o}}{6}\left(\frac{t}{\sqrt{2 } } + \frac{6}{a}\sqrt{\frac{1}{{a_{o}}^2{x_o}^2 } + \frac{a}{6{x_o}}}\right)^2 - \frac{6}{a{a_{o}}^2{x_o}^2}\right ] . \label{atnegb}\end{aligned}\ ] ] this universe begins with an initial singularity ( @xmath85 ) and expands forever .
it is always inflationary and at large times @xmath86 , so it is asymptotically flat . when @xmath87 the solution is markedly different from those others we have examined so far .
the scale factor @xmath14 initially increases but the expansion comes to a halt at a finite time , after which the universe contracts towards @xmath88 .
we have to patch two solutions together , one for the expanding phase , and one for the contracting phase , with the hubble parameter , @xmath25 , having the appropriate sign in equation ( [ timex ] ) . defining @xmath89 , we find @xmath90 , \quad h > 0 .
\label{tnegax1}\end{aligned}\ ] ] if @xmath91 , where @xmath92 , then @xmath36 acquires a complex portion
. however @xmath93 and @xmath94 is the maximum value of @xmath95 .
so when @xmath96 , @xmath97 and the universe is contracting . in this case @xmath98 and @xmath95 are the same as those calculated above but @xmath99 , \quad h < 0
. \label{tnegax2}\ ] ] choosing the initial conditions in equation ( [ tnegax2 ] ) to be @xmath100 , @xmath101 and @xmath102 gives @xmath36 during the contraction phase of the universe described by equations ( [ vnegax ] ) to ( [ anegax ] ) , @xmath103 , \quad h < 0 .
\label{tnegax3}\ ] ] these equations may now be inverted , giving @xmath104 } , \label{phinegat } \\
a(t ) & = & { a_{o}}\sqrt{d{a_{o}}^2{x_o}^2 - \left(\frac{t}{{a_{o } } } - \sqrt{d{a_{o}}^2{x_o}^2 -1}\right)^2}. \label{anegat}\end{aligned}\ ] ] this is not a viable inflationary model .
however , it is an example of an exact scalar field cosmology in a curved spacetime and it serves as a reminder that not all exact solutions will be inflationary .
in curved spacetime , we could only find a few exact solutions where the potential , @xmath105 , had a relatively simple form and there was typically some connection between the allowable range of initial conditions and one of the coefficients in the potential .
in flat spacetime , however , @xmath106 and @xmath107 so equations ( [ chisol ] ) to ( [ timex ] ) and ( [ epsilondef ] ) become @xmath108 } \label{axflat } \\
v(x ) & = & 3h^2 -2\xi^2x^2h'^2 \label{vxflat } \\
t(x ) & = & -\frac{1}{2\xi^2}\int_{{x_o}}^x \frac{1}{x^2h'}dx \label{txflat } \\
\epsilon(x ) & = & 2\left(\xi x \frac{h'}{h}\right)^2 . \label{epsilonflat}\end{aligned}\ ] ] when @xmath106 it is simpler to specify @xmath109 rather than @xmath45 .
any @xmath109 that is a polynomial in @xmath21 will , upon substitution into equation ( [ vxflat ] ) , give a potential that is automatically of the form equation ( [ vgeneral ] ) so we can find any number of exact inflationary models in flat spacetime .
the simplest choice of @xmath109 is @xmath110 substituting equation ( [ hxflat1 ] ) into equation ( [ vxflat ] ) gives the potential , @xmath111 the parametric solution for @xmath14 is @xmath112 this solution is given by muslimov @xcite but we quote it for convenience as many of the models discussed later tend to this solution as the time , @xmath18 , becomes large .
inverting , to make @xmath18 the independent variable , @xmath113 for large @xmath18 , @xmath114 where @xmath115 , which is powerlaw inflation if @xmath116 .
while this is not the general solution , it is an attractor @xcite .
we now turn our attention to new , exact solutions .
we start with @xmath117 which results in @xmath118 from equations ( [ axflat ] ) to ( [ txflat ] ) it follows that @xmath119^{1/4\xi^2 } \label{axflat2 } \\
t(x ) & = & \frac{1}{a\xi^2}\left[\frac{1}{2 } \left(\frac{1}{x}-\frac{1}{{x_o}}\right ) + \frac{b}{a}\log{\left(\frac{{x_o}(a-2bx)}{x(a-2b{x_o})}\right)}\right ] . \label{txflat2}\end{aligned}\ ] ] this solution can not be easily inverted , so we will work with the parametric form .
we treat this case in some detail as much of the analysis here will apply with minor modifications to the other exact solutions discussed in this section .
the potential @xmath105 , equation ( [ vxflat2a ] ) , possesses turning points at @xmath120 since @xmath21 is always positive , @xmath121 is only significant if it is greater than zero . for all values of @xmath20 , @xmath122 is a local maximum of @xmath105 .
however , the potential is only bounded below when @xmath123 and has a range of different forms , depending on the value of @xmath20 . like most exact scalar field cosmologies this solution applies to a restricted set of initial conditions , specifically @xmath124 as @xmath125 , so the field is always evolving away from the unstable equilibrium at @xmath122 .
formally , @xmath126 for all negative times and this solution lacks an initial singularity but small perturbations in @xmath0 or @xmath127 when @xmath21 is very close to @xmath122 render the solution classically unstable in this region .
however the behaviour of the exact solution at later times is representative of a class of models that approximate these initial conditions and we discuss the solution in this context .
if @xmath128 then @xmath36 increases with @xmath21 .
when @xmath129 we find @xmath130 , \quad a = 0.\ ] ] and @xmath95 becomes zero after a finite time .
thus this model universe collapses into a singularity for some choices of the initial conditions .
we now focus on the case where @xmath131 , and the expansion continues indefinitely .
for this solution , @xmath132 if @xmath116 then @xmath42 for all @xmath21 and inflation continues forever . for larger values of @xmath20 the inflationary phase will cease when @xmath133 , or @xmath134 furthermore , this solution has the capacity for both powerlaw and exponential inflation . during quasi - exponential expansion , @xmath25 must be relatively constant on timescales of @xmath135 , during which the universe expands by a single e - folding .
this requirement is satisfied when @xmath136 .
so if @xmath137 ( but not so close to the local maximum that the solution is unstable ) then @xmath14 is approximately exponential . thus this solution exhibits the properties of the two major inflationary models at different stages in its evolution . if @xmath138 and @xmath137 then a large amount of exponential expansion is possible , with the number of e - foldings depending critically on @xmath20 and the ratio @xmath139 . in figures ( [ vphi_fig ] ) and ( [ logavt_fig ] ) @xmath10 and @xmath14 are plotted for a representative set of parameter values .
physically , since @xmath122 is a local maximum , the potential is approximately flat when @xmath140 .
thus both @xmath21 and @xmath105 are changing slowly , leading to an era of quasi - exponential expansion .
as @xmath21 moves further away from @xmath122 the potential becomes steeper and the era of powerlaw expansion begins . for most values of @xmath20 ,
the coefficients of only two of the three coefficients in @xmath105 , equation ( [ vxflat2a ] ) are independent . however ,
when @xmath141 , @xmath142 or @xmath143 one of the terms in @xmath105 drops out , leaving only two nonzero terms which may be chosen arbitrarily .
in particular , when @xmath144 , the potential is @xmath145 which is the first two terms in the superstring motivated perturbation expansion , equation ( [ vgenerala ] ) , with @xmath146 .
the next case we treat is superficially similar to the last , with @xmath147 but the evolution we derive is markedly different .
the corresponding solution is @xmath148^{1/4\xi^2 } \label{axflat3 } \\
t(x ) & = & \frac{1}{a\xi^2}\left[\frac{1}{2 } \left(\frac{1}{x}-\frac{1}{{x_o}}\right ) - \frac{b}{a}\log{\left(\frac{{x_o}(a+2bx)}{x(a+2b{x_o})}\right)}\right ] \label{txflat3}\end{aligned}\ ] ] for this choice of @xmath109 , @xmath149 as @xmath150 and @xmath151 . for other values of @xmath20 ,
@xmath105 is negative for large @xmath21 .
when @xmath152 the potential has a local maximum at @xmath153 but for @xmath154 , @xmath155 for all allowable values of @xmath21 .
however , @xmath14 is always increasing and there are no solutions for which the universe reaches a maximum size and then contracts back to a future singularity .
this solution starts from a singularity though , since in the limit @xmath129 , @xmath156 , \quad a = 0.\ ] ] in this instance , the logarithmic term in @xmath95 , never dominates and the expansion is always approximately powerlaw . because this exact solution implicitly requires the value of @xmath127 to be large when it is in the region containing the local maximum , the conditions that gave rise to a period of exponential expansion for the solution given by equations ( [ axflat2 ] ) and ( [ txflat2 ] ) are not be satisfied in this case .
of course the possibility of a period of exponential expansion in the general solution to the potential , equation ( [ vxflat3a ] ) , is not ruled out . for this solution
@xmath157 if @xmath158 , @xmath72 and there is no inflationary phase . alternatively ,
if @xmath159 the solution is inflationary at all times . for intermediate values of @xmath20 , the inflationary era will be preceeded by a period of powerlaw expansion , @xmath160 , but with @xmath161 .
again , there are three special cases for which one of the terms in @xmath105 is zero , and a potential with two terms and independent coefficients results .
starting from @xmath162 , we find the potential , @xmath163 which has the form of the perturbation expansion suggested by superstring theory , equation ( [ vgenerala ] ) , for all values of @xmath20 . @xmath164^{1/6\xi^2 } \label{axflat4 } \\
t(x ) & = & \frac{1}{2a\xi^2}\left\{\frac{1}{x } - \frac{1}{{x_o } } + \frac{c}{2}\log{\left[\frac{(1-cx)(1+c{x_o})}{(1-c{x_o})(1+cx)}\right]}\right\ } \label{txflat4}\end{aligned}\ ] ] where @xmath165 . for all @xmath20
there is a local maximum at @xmath166 . for @xmath167
, @xmath21 will increase as the universe evolves , and it will eventually reach a maximum size and collapse towards a singularity . for @xmath168 , the expansion will continue indefinitely . if @xmath169 then @xmath14 will initially be approximately exponential , giving way to powerlaw behaviour at later times . again
, there are three special values of @xmath20 for which one of the terms drops out of the potential . in particular , when @xmath170 , the potential is @xmath171 and @xmath5 , the value derived from superstring theory , although the term in @xmath172 is missing .
setting @xmath173 gives similar results to those found for @xmath174 .
the potential and the scale factor are found simply by changing the sign of @xmath50 in equations ( [ vxflat4a ] ) and ( [ axflat4 ] ) .
the time , @xmath36 is @xmath175 \label{txflat5}\ ] ] this model with begins with a singularity , expands indefinitely and can have a mixture of non - inflationary and powerlaw expansion , depending on the value of @xmath20 .
we have found a number of new , exact scalar field cosmologies by deriving the potential that produces a specified form of the density , @xmath11 , or the hubble parameter , @xmath25 .
the technique used here typically does not produce a complete solution for a given potential .
in particular , the solutions presented in this paper do not allow the initial value of @xmath127 to be chosen arbitrarily .
however , it is easy to apply and a large number of potentials can be examined and interesting examples isolated , which can then be studied with more complicated analytic or numerical methods . in curved robertson walker spacetimes , we find a mixture of exact inflationary and non - inflationary models .
we generalise solutions previously found by zer and taha @xcite and develop analagous results for a universe with negative spatial curvature . in flat
spacetime we have presented exact solutions that have several distinct phases , distinguished by the properties of the scale factor , @xmath14 . as well as models with a smooth transition between inflationary and non - inflationary expansion , we have found instances where the inflationary growth is a combination of exponential and powerlaw expansion .
we are not aware of any other exact scalar field cosmologies that are of a comparable complexity and diversity .
if a realistic model possessed these properties , the resulting spectrum of primordial density perturbations would be comparatively complex , assuming that there is not so much powerlaw growth that the fluctuations produced during the era of exponential expansion are shifted beyond the present horizon .
we have concentrated on solutions with relatively simple potentials but the method used here can be used to generate an arbitrary number of exact solutions where the potential consists of exponential terms . while we can find only a handful of such solutions in curved spacetime , the simple relationship between @xmath25 and @xmath105 when @xmath106 , equation ( [ vxflat ] ) , means that large numbers of exact solutions can be obtained in flat spacetime and we have not exhausted their possibilities in this paper . | we present a number of new , exact scalar field cosmologies where the potential consists of two or more exponential terms .
such potentials are motivated by supergravity or superstring models formulated in higher dimensional spacetimes that have been compactified to ( 3 + 1)-dimensions . we have found solutions in both curved and flat robertson walker spacetimes .
these models have a diverse range of properties and often possess several distinct phases , with a smooth transition between ordinary and inflationary expansion .
while exponential potentials typically produce powerlaw inflation , we find models where the inflationary period contains eras of both powerlaw and exponential growth . |
in recent years , there is quite a bit activity on searching dibaryons both theoretically and experimentally . because it is commonly believed that the radius of the dibaryon is less than @xmath8 , studying dibaryon would enhance our knowledge about the short - range behavior of the quantum chromodynamics ( qcd ) , provide the information about the nonperturbative qcd ( npqcd ) and its appropriate treatment .
moreover , the existence of dibaryon is a direct evidence of quarks in hadrons and hadronic systems .
since jaffe predicted h @xcite , there have been lots of efforts devoted to the dibaryon study .
the most intensive studied dibaryons are h , @xmath9 and @xmath10 . up to now
, the most believed value of h is around the @xmath11 theshold @xcite , which agrees with the experimental finding , the lower limit of @xmath12 @xcite .
the theoretical prediction for @xmath13 varies model by model , but most believed value is several tens of mev @xcite .
the theoretical result of @xmath10 is still away from the expected value extracted from double - charge - exchange reactions @xcite .
therefore , it is reasonable to propose studies on the systems with multi - strangeness . with this idea , @xmath14 and
@xmath15 were studied @xcite . in this talk
, we would demonstrate a systematic study of the possible @xmath0wave dibaryons .
because npqcd still can not directly be solved , one has to use model to approximately treat the npqcd effect . in order to reliably study dibaryon ,
the employed model should at least satisfy two - fold requirements : ( 1 ) the ground state properties of baryons , and the available experimental data of the nucleon - nucleon and the nucleon - hyperon ( @xmath16 ) scattering can reasonably be reproduced in the dynamical calculation . ( 2 ) in the extended application , no additional parameters are required .
the su(3 ) chiral quark model is just one of the models which meets the requirements @xcite .
the details of the model can be found in ref.@xcite .
the model parameters are fixed before the calculation @xcite .
the set of values used in ref.@xciteis called model i ( set i ) . in order to further reduce the number of free parameters , we choose the masses of strange scalar mesons , @xmath17 and @xmath18 , to be the values of the mesons with the same quantum numbers in the particle data table .
moreover , in order to test the effects of various chiral fields , we also establish the extended su(2)chiral quark model ( model ii ) , where @xmath19 , @xmath17 and @xmath18 are disregarded , and the su(2 ) chiral quark model ( model iii ) , where only @xmath20 and @xmath21 are presented .
the bound state problem of the six - quark system is solved by using the resonating group method ( rgm ) . in rgm
, the trial wave function of the system can be written as @xmath22,\ ] ] where @xmath23 describes the trial wave function of of relative motion between baryon clusters @xmath24 and @xmath25 and @xmath26 is the antisymmetrizer . in terms of the variational method , one can solve @xmath23 and consequently the binding energy of the system .
as is well known , the matrix element @xmath27 , where the superscript @xmath28 implies that @xmath26 acts in the @xmath29 space only , is an important measure of the action of the pauli principle on the two - baryon state .
it also specifies the symmetry character of the system .
the symmetry character of the two - baryon system could generally be divided into three categoyies . in the first category ,
@xmath30 . this feature would be enormously beneficial to form a state with the @xmath31_{orbit}$ ] symmetry . if the inter - baryon interaction further demonstrates the attractive character with certain strength , it is possible to form a smaller - sized bound state . in the second category , @xmath32 .
thus , the pauli blocking effect here are very small so that two baryons are basically independent with each other .
then , if the inter - baryon meson - exchange interaction can provide large enough attraction , the system would be bound .
this result may also be deduced in terms of a model in the baryon - meson degrees of freedom .
the states with @xmath33 belong to the third category .
the pauli blocking effect between baryons are incredibly serious so that the state with @xmath31_{orbit}$ ] is almost a forbidden state , namely , it is very hard to form a bound state .
we present the calculated binding energies and the corresponding root - mean - square radii ( @xmath34 ) for the possible @xmath0wave states in the first and second categories in tables 1@xmath35 and @xmath36 , respectively .
[ table:1 ] @lllllll + + & system & @xmath27 & model i ( set i ) & model i ( set ii ) & model ii & model iii + & & & @xmath37 & @xmath37 & @xmath37 & @xmath37 + 0 & @xmath38 & @xmath39 & 22.2 //1.01 & 18.5 //1.05 & 64.8//0.84 & 62.7//0.86 + & @xmath40 & @xmath39 & 16.0 //1.10 & 13.5 //1.14 & 6.3//1.25 & 13.2//1.11 + -1 & @xmath41 & @xmath39 & 24.6 //0.99 & 19.0 //1.04 & 11.7//1.13 & 21.5//0.99 + & @xmath42 & @xmath39 & 25.9 //0.95 & 29.3 //0.93 & 78.0//0.79 & 76.3//0.80 + -5 & @xmath43 & @xmath39 & 92.4 //0.71 & 76.5 //0.72 & 58.8//0.75 & 52.1//0.76 + -6 & @xmath14 & @xmath39 & 113.8 //0.66 & 98.5 //0.67 &
72.6//0.70 & 51.9//0.73 + + + & system & @xmath27 & model i ( set i ) & model i ( set ii ) & model ii & model iii + & & & @xmath37 & @xmath37 & @xmath37 & @xmath37 + -3 & @xmath44 & 1 & 3.5 //1.18 & 12.7 //0.98 & 31.8 //0.81 & 49.5 //0.74 + & @xmath45 & 1 & 4.4 //1.15 & 14.2 //0.96 & 34.3 //0.80 & 49.6 //0.74 + -4 & @xmath46 & @xmath47 & 4.1 //1.17 & 0.4 //1.30 & -0.5 //1.33 & 3.1 //1.18 + -5 & @xmath48 & & 32.9 //0.78 & 32.6 //0.77 & 29.3 //0.79 & 17.6
//0.86 + + the units for @xmath49 and rms are in @xmath50 and @xmath51 , respectively .
_ s _ stands for the strangeness . in order to see the effects of various chiral fields and give a possible range of predicted bound states
, we also study these states in terms of models ii and iii and tabulated the result in table 1 . from table 1@xmath35
, one finds that in the @xmath52 and @xmath53 cases , the systems in model i are bounder than in models ii and iii , because the strong attraction provided by @xmath21 and strange mesons @xmath17 and @xmath18 .
if one believes that the su(3 ) model is more suitable for the multi - strangeness system , @xmath1_{(0,0)}$ ] and @xmath2_{(0,1/2)}$ ] should be deeply bound states with narrow widths .
on the other hand , in the @xmath54 and @xmath55 systems , models i and ii are favored by the low - spin and high - spin systems , respectively . because the color - magnetic interaction ( cmi ) shows repulsive and attractive features in the @xmath56 and @xmath57 systems , respectively , the binding energies of the @xmath58 systems are generally higher than those of the @xmath56 systems . moreover , due to the strong decay modes of both composed baryons , the widths of these bound states are broad .
the results in table 1@xmath36 shows another type of character . because , like @xmath59 , the exchange effect is almost negligible , the resultant binding energy is very sensitive to the contributions from chiral fields .
it is shown that the high - strangeness system is bounder in model i and the low - strangeness system has larger binding energy in model iii .
nevertheless , these systems are weakly bound . finally , to ensure the predicted bound states are reliable and stable
, we also examine their model - parameter dependencies .
the @xmath60- and @xmath61-dependencies are demonstrated in table 1 by sets i and ii .
it is shown that the qualitative binding behavior does not change although the binding of most systems become slightly weaker due to the overall weaker attractive feature from @xmath17 and @xmath18 .
moreover , the @xmath62- and @xmath63-dependencies are plotted in fig .
[ fig:1 ] .
it is shown that the binding energy always becomes smaller with increasing @xmath63 and decreasing @xmath62 , but the qualitative binding behavior does not change .
therefore , our predicted bound states are stable and reliable .
we investigate the possible @xmath0wave baryon - baryon bound states with the su(3 ) chiral quark model , and test the reliabilities and stabilities of results by using different models and model parameters .
it is shown that the symmetry property , characterized by the expectation value of antisymmetrizer , of the system is a major factor to affect the binding behavior .
sometimes , the contributions from various chiral fields demonstrate very important roles .
we arrange all the bound states into three types .
the first type states , such as @xmath1_{(0,0)}$ ] and @xmath2_{(0,1/2)}$ ] should be deeply bound states with narrow widths , or dibaryons.the states of second type , such as @xmath3_{(0,5/2)}$ ] , @xmath3_{(3,1/2)}$ ] , @xmath4_{(0,3)}$ ] and @xmath4_{(3,0)}~(d^{*})$ ] are bound states with broad widths .
the states like @xmath5_{(1,1/2)}$ ] , @xmath6_{(0,1)}$ ] , and @xmath7_{(2,1/2)}$ ] are third types states . like @xmath59 , they generally the weakly bound states , and their binding behaviors
very much depend on the contributions from the chiral fields .
we would call great attention from physicists on the states of the first and second types and wish that these states can be searched in the future experiment .
99 r.l.jaffee , pyhs.rev.lett .
38 ( 1977 ) 195 .
k.imai , hadrons and nuclei with strangeness- proceedings of the first sino - japan symposium on strangeness physics , singapore , 1999 , and references therein .
k.yazaki , prog.theor.phys .
suppl.91 ( 1987 ) 146 , and references therein .
p.n.shen , z.y.zhang , y.w.yu , s.q.yuan , s.yang , j.phys .
g ( 1999 ) .
f.wang , j.l.ping , g.h.wu , et al . ,
c51 ( 1995 ) 3411 .
x.q.yuan , z.y.zhang , y.w.yu and p.n.shen , phys.rev .
c ( 1999 ) .
g.wagner , l.ya.glozman , a.j.buchmann and a.faessler , y.w.yu , z.y.zhang and x.q.yuan , commun.theor.phys . 31
( 1999 ) 1 .
q.b.li and p.n.shen , chin.phys.lett .
z.y.zhang , y.w.yu , p.n.shen and et al .
, nucl.phys .
a625 ( 1997 ) 59 . | in the framework of the su(3 ) chiral quark model , the @xmath0wave baryon - baryon bound states are investigated .
it is found that according to the symmetry character of the system and the contributions from chiral fields , there are three types of bound states .
the states of the first type , such as @xmath1_{(0,0)}$ ] and @xmath2_{(0,1/2)}$ ] are deeply bound dibaryon with narrow widths .
the second type states , @xmath3_{(0,5/2)}$],@xmath3_{(3,1/2)}$ ] , @xmath4_{(0,3)}$ ] and @xmath4_{(3,0)}$ ] are also bound states , but with broad widths .
@xmath5_{(1,1/2)}$ ] , @xmath6_{(0,1)}$ ] , and @xmath7_{(2,1/2)}$ ] are third type states .
they , like _ d _ , are weakly bound only if the chiral fields can provide attraction between baryons . |
the goal of this paper is to discuss an unexpected connection between the parameter spaces of two families of one - dimensional dynamical systems , and establish an explicit correspondence between the bifurcation parameters for these families .
the family @xmath2}$ ] of @xmath0-continued fraction transformations , defined in @xcite , is a family of discontinuous interval maps , which generalize the well - known gauss map . for each @xmath3 $ ] ,
the map @xmath1 from the interval @xmath4 $ ] to itself is defined as @xmath5 and , for @xmath6 , @xmath7 where @xmath8 is a positive integer . for every @xmath9 $ ] at most two of the infinitely many branches of @xmath1
fail to be surjective , hence @xmath1 has infinite topological entropy . on the other hand , for all @xmath10 the map @xmath1 admits a unique invariant probability measure @xmath11 which is absolutely continuous with respect to lebesgue measure , and one can consider the metric entropy @xmath12 with respect to this measure ; @xmath13 can be computed by rohlin s formula and is finite ( see @xcite ) .
several authors have studied the variation of @xmath13 as a function of the parameter @xmath0 .
numerical evidence that the entropy @xmath13 is continuous but non - monotone was first produced in @xcite .
subsequently , @xcite found out that the entropy is monotone over intervals in parameter space for which the orbits of the two endpoints collide after a finite number of steps ( see equation on page ) .
this analysis is completed in @xcite , where intervals of parameters for which such relation holds are classified , and it is proven that the union of all such intervals has full measure .
the complementary set , denoted by @xmath14 , is the set of parameters across which the combinatorics of @xmath1 changes , hence it will be called the _
bifurcation set_. the second object we consider is the family of _ unimodal maps _ , i.e. smooth maps of the interval with only one critical point , the most famous example being the logistic family .
to any such map one can associate a _ kneading invariant _
@xcite which encodes the dynamics of the critical point and determines the combinatorial type of the map .
using an appropriate coding ( see @xcite , @xcite ) the set of all kneading invariants which arise from unimodal maps can be represented as the points of a set @xmath15 defined in terms of the _ tent map _ @xmath16 @xmath17 : t^k(x ) \leq x \ \forall
k \in \mathbb{n } \}\ ] ] the set @xmath15 is uncountable , totally disconnected and closed ; the topology of @xmath15 reflects the variation in the dynamics of the corresponding maps as the parameter varies : for instance families of isolated points in @xmath15 correspond to period doubling cascades ( see section [ pedou ] ) .
this set also parametrizes the real slice of the boundary of the mandelbrot set @xmath18 ( see section [ mandel ] ) , because for each admissible kneading sequence there exists exactly one real parameter on the boundary with that kneading sequence .
the key result of this paper ( section [ main ] ) is the following correspondence between the bifurcation parameters of the two families : [ mth ] the sets @xmath19 and @xmath20 are homeomorphic .
more precisely , the map @xmath21 \rightarrow [ \frac{1}{2},1]$ ] given by @xmath22 is an orientation - reversing homeomorphism which takes @xmath20 onto @xmath23 . as a consequence , intervals in the parameter space of @xmath0-continued fractions where a matching between the orbits of the endpoints occurs are in one - to - one correspondence with real hyperbolic components of the mandelbrot set , i.e. intervals in the parameter space of real quadratic polynomials where the orbit of the critical point is attracted to a periodic cycle . in terms of entropy , intervals
over which the metric entropy of @xmath0-continued fractions is monotone ( and conjecturally smooth ) are mapped to parameter intervals in the space of quadratic polynomials where the topological entropy is constant ( see figure [ nakadavsmandelbrot ] ) .
for instance , the matching interval @xmath24 , [ 0 ; \overline{2 , 1}])$ ] , identified in @xcite and @xcite , corresponds to the `` airplane component '' of period @xmath25 in the mandelbrot set .
this dictionary illuminates many connections between seemingly unrelated objects in real dynamics , complex dynamics , and arithmetic , which we will explore in the rest of the paper .
for example , the following properties of @xmath15 follow immediately by using the combinatorial tools developed in @xcite : 1 .
the set @xmath15 can be constructed via a bisection algorithm ( section [ bisec ] ) 2 .
the sequences of isolated points in @xmath20 observed experimentally in ( @xcite , section 4.2 ) are the images of the well - known period doubling cascades for unimodal maps ( section [ pedou ] ) 3 .
the derived set @xmath26 is a cantor set ( section [ top ] ) 4 .
the hausdorff dimension of @xmath15 is @xmath27 ( section [ hd ] ) [ nakadavsmandelbrot ] from the above properties of @xmath15 we derive some consequences on corresponding sets .
for example , ( iv ) implies that the set of external rays of the mandelbrot set which land on the real axis has full hausdorff dimension ( section [ mandel ] ) , a result first obtained in @xcite .
moreover , in section [ univ ] we give an arithmetic interpretation of @xmath15 .
namely , aperiodic elements of @xmath15 correspond to binary expansions of _ univoque numbers _ , i.e. the numbers @xmath28 such that @xmath27 admits a unique representation in base @xmath29 .
univoque numbers have been studied by erds , horvth and jo @xcite and many others ( see section [ univ ] for references ) , and once again the translation of statements ( i)-(iv ) immediately yields several results previously obtained by different authors .
0.3 cm moreover , our new characterization of the set @xmath14 ( see lemma [ characterizee ] ) shows that it is essentially the same object appearing in a paper of cassaigne as the spectrum of recurrence quotients for cutting sequences of geodesics on the torus ( @xcite , theorem 1.1 ) .
indeed , our results on @xmath14 answer some questions raised in @xcite , such as the computation of hausdorff dimension . finally , we would like to emphasize that the correspondence of theorem [ mth ] opens up many questions , and it is especially natural to ask to what extent results in the well - developed theory of unimodal maps can be translated into the continued fraction setting .
for instance , it would be interesting to identify an analogue of renormalization for the @xmath0-continued fractions , and to further explore the relation between the entropy of that family ( @xcite , @xcite , @xcite , @xcite ) and the topological entropy of unimodal maps ( @xcite and @xcite among others ) .
we wish to thank j - p .
allouche and h. bruin for their helpful comments .
g.t . wishes to thank c. mcmullen for many useful discussions .
is sponsored by project miur - prin2009 variational and topological methods in the study of nonlinear phenomena " , italy , and by the distinguished scientist fellowship program ( dsfp ) " , king saud university , riyadh , saudi arabia .
the lower half of figure 1 has been generated using the software mandel by w. jung .
let @xmath30 denote the _ tent map _ , the _ farey map _ and the _ gauss map _ of @xmath31 $ ] , given by and @xmath32 denote the integer and the fractional part of @xmath33 , respectively , so that @xmath34 . ]
@xmath35 \displaystyle 2(1-x ) & { \rm if}\ \frac{1}{2 } \leq x\leq 1 \end{array } \right . \
f(x):=\left\ { \begin{array}{cl } \displaystyle \frac{x}{1-x } & { \rm if}\ 0\leq x < \frac{1}{2 } \\[0.5 cm ] \displaystyle \frac{1-x}{x } & { \rm if}\ \frac{1}{2 } \leq x\leq 1 \end{array } \right.\ ] ] and @xmath36 , @xmath37 . the action of @xmath38 and @xmath39 can be nicely illustrated with different symbolic codings of numbers .
given @xmath40 $ ] we can expand it in ( at least ) two ways : using a _ continued fraction expansion _ , i.e. @xmath41\quad , \quad a_i\in { \mathbb n}\ ] ] and a _ binary expansion _ ,
i.e. @xmath42 the action of @xmath39 on binary expansions is as follows : for @xmath43 , [ tenda ] t(0 . 0 ) = 0 . ,
t(0.1 ) = 0 . where @xmath44 and @xmath45 , @xmath46 .
the actions of @xmath38 and @xmath47 are given by @xmath48 ) = [ 0 ; a_1 - 1,a_2,a_3,\dots ] $ ] if @xmath49 , while @xmath50)=[0 ; a_2,a_3, ... ]$ ] , and @xmath51 ) = [ 0 ; a_2,a_3,\dots]$ ] .
as a matter of fact @xmath47 can be obtained by @xmath38 by inducing on the interval @xmath52 $ ] , i.e. g(x ) = f^1/x ( x ) , x0 now , given @xmath53 $ ] , one may ask what is the number obtained by interpreting the partial quotients @xmath54 as the lengths of successive blocks in the dyadic expansion of a real number in @xmath31 $ ] ; this defines _
minkowski s question mark _
function @xmath55 \rightarrow [ 0,1]$ ] ?
( x)=_k1(-1)^k-1 2 ^ -(a_1++a_k-1 ) = 0 . _
a_1 - 1 the questionmark function @xmath56 has the following properties ( see @xcite ) : * it is strictly increasing from @xmath57 to @xmath27 and hlder continuous of exponent @xmath58 ; * @xmath33 is rational iff @xmath56 is of the form @xmath59 , with @xmath60 and @xmath61 integers ; * @xmath33 is a quadratic irrational iff @xmath56 is a ( non - dyadic ) rational ; * @xmath56 is a singular function : its derivative vanishes lebesgue - almost everywhere ; * it satisfies the functional equation @xmath62 .
we shall see later that the question mark function @xmath63 plays a key role in the main result of this paper ( theorem [ mth ] ) because it conjugates the actions of @xmath38 and @xmath39 . for a more general analysis of the role played by @xmath63 in a dynamical setting see @xcite .
let @xmath64 be a rational number , and let @xmath65 $ ] be its continued fraction expansion , with @xmath66 .
the _ quadratic interval _ associated to @xmath67 will be the open interval @xmath68 with endpoints @xmath69 \quad \textup{and } \quad [ 0 ; \overline{a_1 , \dots , a_{n-1 } , a_n - 1 , 1}]\ ] ] where @xmath70 $ ] denotes the real number with periodic continued fraction expansion of period @xmath71 . the number @xmath67 is called the _ pseudocenter _ of @xmath68 .
we also define the degenerate quadratic interval @xmath72 $ ] , where @xmath73 $ ] is the golden number .
the _ bifurcation set _ for continued fractions is @xmath74\setminus \bigcup_{r \in ( 0 , 1]\cap \mathbb{q } } i_r\ ] ] quadratic intervals which are not properly contained in any larger quadratic interval are called _ maximal _ quadratic intervals . by ( @xcite ,
section 2.2 ) , if two quadratic intervals are not disjoint , they are both contained in a maximal quadratic interval ; thus , the connected components of the complement of @xmath14 are precisely the maximal intervals . for all parameters belonging to a maximal quadratic interval , the @xmath0-continued fraction transformation @xmath1 satisfies a _ matching condition _ between the orbits of the endpoints , with fixed combinatorics : let @xmath68 be a maximal quadratic interval , and @xmath65 $ ] with @xmath75 even .
let @xmath76 then for all @xmath77 , @xmath78 as a consequence , by @xcite , the metric entropy @xmath79 is locally monotone on the complement of @xmath14 .
the set @xmath20 can be given the following new characterization : [ characterizee ] let @xmath40 $ ] .
the following are equivalent : 1 .
@xmath80 2 .
@xmath81 , @xmath82 3 .
@xmath83 , @xmath82 4 .
@xmath84 @xmath85 , where @xmath86 is an inverse branch of @xmath38 .
_ let @xmath87 $ ] be the continued fraction expansion of @xmath33 .
fix @xmath89 , so @xmath90 $ ] for some @xmath75 and @xmath91 .
= g^{n-1}(x ) \geq x\ ] ] @xmath93 .
let @xmath94 be the subsequence given by the partial quotients sums , i.e. @xmath95 , and note that @xmath96 $ ] . if @xmath97 then @xmath98 .
otherwise , @xmath99 = \psi_1(f^k(x))$ ] for some @xmath100 so that @xmath101 \leq [ 0 ; a_{n+1 } , a_{n+2 } , \dots ] = f^k(x ) \rightarrow f^{k}(y ) \leq y\ ] ] @xmath102 .
fix @xmath89 , and let @xmath103 , so that @xmath104 .
now , @xmath105 \leq [ 0 ; 1 , a_1 , a_2 , \dots ] = \psi_1(x)\ ] ] therefore @xmath106 = g^k(x ) \geq x = [ 0 ; a_1 , \dots]$ ] . @xmath107 .
let @xmath108 be the endpoint of a maximal interval , hence @xmath109 $ ] .
once you fix @xmath110 , then @xmath111 $ ] with @xmath112 . now , by @xcite , proposition 4.5 @xmath113 \geq [ 0 ; \overline{a_1 , \dots , a_l , a_{l+1 } , \dots , a_n } ] = x\ ] ] i.e. @xmath114 , and hence @xmath115 . since @xmath20 has empty interior ( @xcite , theorem 1.2 ) , every point in @xmath20 is a limit point of a sequence of endpoints of maximal intervals , so inequality @xmath115 extends to all of @xmath20 by continuity of @xmath38 .
if @xmath117 , then @xmath33 belongs to a maximal interval @xmath118 .
let @xmath119 = [ 0 ; a^+]$ ] be the two continued fraction expansions of the rational number @xmath120 , in such a way that @xmath121 $ ] and @xmath122 $ ] . here
@xmath123 is a finite string of integers of odd length @xmath124 , and @xmath125 is a finite string of integers of even length @xmath126 .
the sequence @xmath127 < [ 0 ; a^+a^+ ] < \dots < [ 0 ; ( a^+)^n ] < [ 0 ; ( a^+)^{n+1 } ] < \dots\ ] ] tends to @xmath128 = \alpha^+$ ] , hence for any @xmath129 there exists @xmath130 such that @xmath131 \leq x < [ 0 ; ( a^+)^{n+2}]\ ] ] therefore @xmath132 \leq x$ ] . if instead @xmath133 $ ] , then using the fact that @xmath124 is odd @xmath134 so @xmath135
. @xmath136 we now introduce a set @xmath15 which encodes the topological dynamics of unimodal maps in a universal way .
several similar approaches are possible ( @xcite , @xcite among others ) ; we follow @xcite , @xcite .
we recall that a smooth map @xmath137 \to [ 0,1]$ ] is called _ unimodal _ if it has exactly one critical point @xmath138 , and we are going to assume @xmath139 . if @xmath140 $ ] never maps to @xmath141 under @xmath142 , let us define the _ itinerary _ of @xmath143 $ ] to be the sequence @xmath144 if @xmath33 eventually maps to @xmath141 , let us define @xmath145 where the limit is taken over non - precritical points .
the _ kneading sequence _ of a unimodal map is the binary sequence @xmath146 , where @xmath147 is the critical value .
kneading sequences encode the combinatorics of the orbits and determine the topological entropy . a nice way to decide whether a given binary sequence @xmath148 is
the kneading sequence of some unimodal map is the following : we first associate to @xmath149 the number @xmath150 $ ] defined as [ eq : enc ] = 0.t_1t_2 =_k1 t_k 2 ^ -k , t_k = _
i=1^k s_i ( mod 2 ) one readily checks that the following diagram is commutative : [ diagram ] [ 0,1 ] @>i > > \{0,1}^n@ >
> > [ 0,1 ] + @vf vv @vv v @vv t v + [ 0,1 ] @>i > > \{0,1}^n@ >
> > [ 0,1 ] + 0.5 cm where @xmath39 is the tent map ( [ tenda ] ) and @xmath151 is the shift map .
since the critical value is the highest value in the image and the map @xmath152 is increasing ( @xcite , lemma 1.1 ) , one has the characterization a binary sequence @xmath153 is the kneading sequence of a continuous unimodal interval map if and only if @xmath154 satisfies @xmath155 one is thus led to study the set : = \ { x : t^k(x)x , kn } by considering @xmath156 one immediately realizes that @xmath157 $ ] .
note moreover that , if the map is @xmath158 , the extremal situations in which @xmath159 for some @xmath160 correspond to periodic attractors ( and thus periodic kneading sequence ) of the corresponding smooth unimodal maps ( see [ dmvs ] , chapter iv ) .
note that all kneading sequences which arise from unimodal maps can be actually realized by real quadratic polynomials @xmath161 , @xmath162 $ ] . by density of hyperbolicity ( @xcite , @xcite )
, aperiodic kneading sequences are realized by exactly one map @xmath163 . on the other hand ,
the set of parameters which realize a given admissible periodic kneading sequence is an interval in parameter space , and among those maps there is exactly one @xmath163 which has a parabolic orbit . in conclusion
, @xmath15 parametrizes the set of topologically unstable parameters , hence it will be called the _ binary bifurcation set_.
we are now ready to prove theorem [ mth ] , namely that the map @xmath21 \rightarrow [ \frac{1}{2},1]$ ] given by @xmath164 \mapsto \varphi(x ) = 0.\underbrace{11\ldots1}_{a_1}\underbrace{00\ldots0}_{a_2}\underbrace{11\ldots1}_{a_3}\ldots $ ] is an orientation - reversing homeomorphism which takes @xmath20 onto @xmath23 . _ proof of theorem [ mth ] . _
the key step is that minkowski s question mark function conjugates the farey and tent maps , i.e. @xmath165\ ] ] note that @xmath166 , hence @xmath167 is a homeomorphism of @xmath31 $ ] onto the image of @xmath168 , i.e. @xmath169 $ ] .
moreover , by lemma [ characterizee]-(d ) , @xmath170 : f^k(x ) \leq x \}\ ] ] and by @xmath171 : t^k(x ) \leq x \ } = \lambda\setminus\{0\}. \qquad \qed\ ] ] in the following subsections we will investigate a few consequences of such a correspondence . a direct consequence of the theorem is the following algorithm to construct @xmath15 , which mimics the one used to construct @xmath172 : let @xmath173 be a binary word of length @xmath75 ( ending with @xmath27 ) and @xmath174 the corresponding dyadic rational . setting @xmath175 , we see that @xmath176 has exactly two representations : @xmath177 where , if @xmath178 is any finite binary word , @xmath179 denotes the infinite sequence obtained by repeating @xmath178 indefinitely .
now define the _ dyadic interval _ associated to @xmath176 as the open interval @xmath180 whose endpoints are the ( non - dyadic ) rationals corresponds to the string @xmath125 which defines the upper endpoint of the interval in @xcite , section 2.2 . ]
@xmath181 0.2 cm example .
take @xmath182 .
since the binary expansions of @xmath176 are @xmath183 and @xmath184 , we get @xmath185 and @xmath186 .
( see also the example in section [ pedou ] . )
[ lambda_complement ] we have = [ 0,1 ) _
d(0,1]q_2 j_d with @xmath187 . from theorem [ mth ] and equation
. 0.1 cm note that in the previous construction there is substantial overlapping among the @xmath188 .
it is possible , however , to directly produce all connected components of the complement of @xmath15 by a bisection algorithm : namely , one can produce a new component in the gap between two previously computed components .
let @xmath189 \subseteq [ 0,1]$ ] be a closed interval , @xmath190 , and let the binary expansions or @xmath191 are dyadic numbers , take the expansion of @xmath120 which terminates with infinitely many zeros and the expansion of @xmath191 with infinitely many ones ] of @xmath120 and @xmath191 be @xmath192 let @xmath193 .
the _ binary pseudocenter _ of @xmath194 is the dyadic rational [ psece ] d(i ) : = 0.a_1 a_n-1 1 by translating one gets the following description of @xmath15 : let @xmath195 $ ] , and for any @xmath75 define recursively @xmath196 @xmath197 then @xmath198 from proposition 2.11 of @xcite and theorem [ mth ] .
period doubling bifurcation is a well known phenomenon which occurs with a universal structure in families of unimodal maps .
period doubling cascades are in one - to - one correspondence with _ periodic windows _ consisting of isolated points in the set @xmath19 , as shown in @xcite , @xcite .
we will see how similar windows of isolated points occur in @xmath14 , and in both cases the combinatorial pattern can be understood in terms of the thue - morse sequence @xmath199 . recall that @xmath200 can be defined in terms of the power series expansion [ eq : tm ] _
k0 ( 1-z^2^k)= 1-z - z^2+z^3+= _ n0 ( -1)^t_nz^n define the map @xmath201 on the set of finite binary words as [ uno ] ( ) : = 1 moreover , for any finite binary word @xmath178 , let @xmath202 be the rational number @xmath203 , and let for any @xmath204 [ due ] _
j():=_0(^j ( ) ) _ ( ) = _ j _ j ( ) the _ periodic window _ associated to @xmath178 is @xmath205 let @xmath206 be a point with periodic binary expansion , and let @xmath207 with @xmath208 minimal period . then the periodic window associated to @xmath178 contains exactly the sequence of isolated points just constructed : @xmath209 let us call _ generating number _ of the window @xmath210 the pseudocenter @xmath211 of the first interval @xmath212 $ ] . 0.4 cm examples .
if @xmath213 , the empty word , we obtain the sequence @xmath214 which converges to @xmath215 , and the generating number of @xmath216 is @xmath217 .
note that the binary expansion of @xmath218 is the thue - morse sequence .
taking @xmath219 we get the sequence @xmath220 which converges to the irrational number @xmath221 and belongs to the next - to - largest window @xmath222 in @xmath15 .
its generating number is @xmath223 .
0.3 cm windows of isolated points are present also in the set @xmath172 and can be generated with a similar procedure , introduced in @xcite and @xcite without connection to bifurcations of unimodal maps .
indeed , let @xmath68 be a maximal quadratic interval , and let @xmath224 = [ 0 ; s_1]$ ] be the continued fraction expansions of its pseudocenter , in such a way that @xmath225 has even length and @xmath226 has odd length .
let us define @xmath227 , @xmath228 and @xmath229 is generated from @xmath230 via the substitutions @xmath231 , @xmath232 . if we denote @xmath233 \qquad \alpha_\infty(r ) : = \lim_{n \to \infty } \alpha_n(r)\ ] ] it follows immediately from @xcite , proposition 3.9 that for each @xmath68 maximal quadratic interval of pseudocenter @xmath67 , @xmath234 = \bigcup_{n \geq 0 } \alpha_n(r)\ ] ] limit points @xmath235 of these cascades have an aperiodic continued fraction expansion with a common combinatorial structure which is again related to the thue - morse sequence @xmath236 : indeed , for any @xmath67 @xmath237\ ] ] where the sequence @xmath238 is such that @xmath239 .
for instance , if @xmath240 , then coincides with the first difference sequence of @xmath241 defined as @xmath242 . ]
@xmath243 \cong 0.38674997 ... $ ] is the accumulation point considered in @xcite and @xcite .
we end this section with a precise characterization of the points @xmath244 in @xmath15 and of their accumulation point @xmath245 , which we prove to be a transcendental number .
[ lemmino ] let @xmath246 be a finite binary word of length @xmath247 such that @xmath248 and define recursively the integers : * @xmath249 * @xmath250 * @xmath251 then @xmath252 for every @xmath204 , and @xmath253 .
moreover , the binary pseudocenters are given by @xmath254 we need the following elementary identity [ bina ] 0 .
i=1^s for any binary word @xmath255 .
@xmath256 is the integer whose binary expansion is @xmath257 , hence from @xmath258 .
moreover , if @xmath259 is the integer whose binary expansion is @xmath260 , with @xmath261 a binary string , then the number whose binary expansion is @xmath262 is @xmath263 . by the recursive formulas and and identity one has @xmath264 for all @xmath265
. finally , the binary pseudocenter of the interval @xmath266 $ ] is @xmath267 hence the lemma follows by taking @xmath268 .
furthermore , let us consider the generating function ( cfr .
equation ( [ eq : tm ] ) ) ( z ) : = _ k0 ( 1-z^2^k ) this function satisfies the functional equation @xmath269 , from which we see that all its zeroes lie on the unit circle and are actually dense there ( see also @xcite ) .
[ xi ] let @xmath270 be a finite binary word of length @xmath247 such that @xmath248 .
then @xmath271 in particular , @xmath245 is transcendental .
setting @xmath253 and using lemma [ lemmino ] we get for @xmath272 @xmath273 therefore @xmath274 where we used the relation @xmath275 given by lemma [ lemmino ] .
the claimed identity follows by taking the limit .
finally , mahler proved in ( @xcite , page 363 ) that @xmath276 is transcendental for every algebraic number @xmath120 , with @xmath277 .
examples of @xmath245 are @xmath278 0.5 cm a corresponding transcendence result can be given for @xmath20 .
namely , accumulation points of period doubling cascades @xmath235 have continued fraction expansion given by the substitution rule explained above , hence they are transcendental by @xcite .
it has been widely conjectured that all numbers with non - eventually periodic bounded partial quotients are transcendental .
let us point out that such a statement is equivalent to the transcendence of all non - quadratic irrationals in @xmath20 .
similarly , one might ask whether all irrational points in @xmath15 are transcendental as well . by using the bisection algorithm and proposition [ xi ]
it is immediate to determine the topology of bifurcation sets , namely [ topology ] let @xmath279 be the set of rational numbers with odd denominator
. then 1 .
the set @xmath280 is dense in @xmath15 .
2 . the derived set
@xmath26 ( i.e. @xmath15 minus its isolated points ) is a cantor set . as a corollary , @xmath26 and @xmath15 have cardinality @xmath281 .
3 . let @xmath282 be the usual cantor middle - third set , let @xmath283 , @xmath89 be the connected components of the complement @xmath284 . for each @xmath60 , pick a countable set of points @xmath285 which accumulate only on @xmath286 .
then @xmath15 is homeomorphic to @xmath287 .
any open neighbourhood of an element @xmath288 contains a subset of @xmath15 which is homeomorphic to @xmath15 itself ( fractal structure ) .
note that , even though the proposition was stated for @xmath15 , the same result holds for @xmath20 since they are homeomorphic .
the only change is that you have to replace @xmath279 by the set of quadratic irrationals with purely periodic continued fraction expansion . 1 .
@xmath15 is closed with no interior by proposition [ lambda_complement ] , hence the endpoints of the connected components of its complement are dense in it .
the connected components of the complement of @xmath15 are intervals of type @xmath188 , so their endpoints have purely periodic binary expansion , hence they belong to @xmath279 .
2 . by virtue of the homeomorphism @xmath167 of theorem [ mth ] ,
it is equivalent to prove the same result for @xmath20 .
let @xmath289 denote the points of @xmath14 which are periodic under @xmath47 @xmath290 and let us define the set of _ primitive elements _ as @xmath291 + [ isolated ] the set of isolated points of @xmath20 is precisely @xmath292 .
+ if @xmath0 is not a limit point of @xmath20 then it is the separating element between two adjacent maximal quadratic intervals @xmath293 and @xmath294 , so @xmath0 is the left endpoint of the rightmost interval and has odd minimal period ( see @xcite ) .
+ moreover , if @xmath295 , we denote with @xmath296 the limit point of the period doubling cascade generated ( @xcite , page 24 ) . ] by @xmath297 and let @xmath298 be the corresponding period doubling window .
we also set @xmath299 , hence by lemma [ isolated ] + @xmath300 now , no two intervals of the form @xmath301 with @xmath302 are adjacent to each other , since the right endpoint @xmath297 is always a quadratic irrational while @xmath296 is always transcendental ( see proposition [ xi ] ) , therefore @xmath303 has no isolated points and it is a cantor set .
3 . every closed subset of the interval with no interior and no isolated points is homeomorphic to the usual cantor middle - third set via a homeomorphism of the ambient interval .
such an extension can be chosen so that it maps any period doubling cascade to some @xmath304 .
4 . immediate from 3 . let us remark that similar results have been obtained for the set of univoque numbers in @xcite , @xcite , and by allouche and cosnard for a similar number set ( @xcite , @xcite , @xcite )
. for details on the relations between these sets see section [ univ ] . moreover , @xmath14 shares some features with the markov spectrum , since the gauss map is the first - return map on a section of geodesic flow on the modular surface .
even though the two sets are not homeomorphic , property 2 also holds for the markov spectrum @xcite . in @xcite
it is proved that @xmath305 by providing estimates on the hausdorff dimension of its segments .
more precisely , for every @xmath306 we consider @xmath307 , \ \
b_k:=\{x=[0;a_1 , a_2 , a_3 , ... ] : a_i\leq k \
\forall i\in { \mathbb n}\ } , \ \
\psi_k(x ) : = \frac{1}{k+x}\ ] ] and it is easy to check that @xmath308 therefore . ] @xmath309 .
we can use our dictionary to obtain analogous results in the linear setting , where in fact one can explicitely compute the hausdorff dimension of segments of @xmath15 .
let @xmath310 $ ] , @xmath311 and @xmath312 : \
\underline{x } \mbox { does not contain sequences of } k+1 \mbox { equal digits } \};\ ] ] using the correspondence of theorem [ mth ] , the inclusions become @xmath313 , and thus @xmath314 the set @xmath315 is self - similar , therefore its hausdorff dimension can be computed by standard techniques ( see @xcite , theorem 9.3 ) .
more precisely , if @xmath316 is the number of binary sequences of @xmath75 digits whose first digit is @xmath27 and do not contain @xmath317 consecutive equal digits , one has the following linear recurrence : @xmath318 which implies [ pisot ] for any fixed integer @xmath319 the hausdorff dimension of @xmath315 is @xmath320 , where @xmath321 is the only positive real root of the characteristic polynomial @xmath322 as a consequence , a simple estimate on the unique positive root of @xmath323 yields @xmath324
the goal of this section is to explore the relations between the bifurcation sets we described before and other well - known sets for which a combinatorial description can be given , namely the real slice of the mandelbrot set and the set of univoque numbers .
let us recall that the mandelbrot set encodes the dynamical properties of the family of quadratic polynomials @xmath325 namely it can be defined as @xmath326 a combinatorial model of the mandelbrot set can be constructed by using the theory of invariant quadratic laminations , as developed by thurston @xcite .
consider the unit disk in the complex plane with the action of the doubling map @xmath327 on the boundary .
@xmath328 is a simple curve embedded in the interior of the circle joining two points on the boundary .
it is usually represented as a geodesic for the hyperbolic metric in the poincar disk model .
one can extend the action of @xmath142 to the whole lamination : the image of a leaf @xmath328 is , by definition , the leaf which connects the images of the endpoints of @xmath328 .
we define the _ length _ @xmath329 of a leaf to be the ( euclidean ) length of the shortest arc of circle delimited by its endpoints , measured on the boundary circle and normalized in such a way that the whole circumference has length @xmath27 ( hence , for any leaf @xmath328 , @xmath330 ) .
a leaf is _ real _ if it is invariant with respect to complex conjugation . the doubling map @xmath142 preserves the set of real leaves .
a _ lamination _ is a closed subset of the disk which is a union of leaves , and such that any two leaves in the lamination can intersect only on the boundary of the disk .
a _ quadratic invariant lamination _ is a lamination whose set of leaves is completely invariant with respect to the action of the doubling map ( we refer to @xcite , page 66 for a complete definition of this invariance ) .
given an invariant quadratic lamination , its longest leaves are called its _ major leaves _ , and their image the _ minor leaf_. there can be at most @xmath331 major leaves , and they both have the same image . in the well - known theory of douady , hubbard and thurston , the leaves of an invariant quadratic lamination define an equivalence relation on the boundary circle and the quotient space is a model for the julia set of a certain quadratic polynomial . by ( @xcite , theorem ii.6.8 ) ,
different minor leaves do not cross , and @xmath332 is indeed a lamination . similarly to the julia set case , the space obtained by identifying points on the disk which belong to the same leaf of @xmath332 is a model of the classical mandelbrot set .
we will consider the set @xmath334 of _ real leaves _ inside the quadratic minor lamination : under the previous correspondence , they correspond to real points on the boundary of the mandelbrot set .
moreover , let @xmath335 be the set of endpoints of all leaves in @xmath336 .
since @xmath337 is symmetric , it is enough to consider the `` upper half '' @xmath338 $ ] .
_ proof of proposition [ simple ] . _
the first ingredient is the fact that the action induced by the doubling map on the lengths of leaves is essentially given by the tent map , namely for any leaf @xmath328
@xmath343 let @xmath344 $ ] be the endpoint of a minor leaf , let @xmath126 be the minor leaf and @xmath345 be one of the corresponding major leaves .
by maximality of @xmath345 and invariance of the quadratic lamination , @xmath346 for all @xmath347 , hence by @xmath348 from this and the fact that @xmath349 is a fixed point for @xmath39 it follows that the major leaf is longer than @xmath350 , hence @xmath351 moreover , by symmetry of the minor leaf with respect to complex conjugation , @xmath352 hence @xmath353 and by @xmath354 hence @xmath355 belongs to @xmath15 .
conversely , if @xmath356 , one can construct the real leaf @xmath126 which joins @xmath33 with @xmath357 .
the fact that @xmath126 is a minor leaf follows from the criterion ( @xcite , page 91 ) : a leaf @xmath126 is the minor leaf of an invariant quadratic lamination if and only if 1 .
all forward images have disjoint interior ; 2 . no forward image is shorter than @xmath126 ; 3 .
if @xmath126 is non - degenerate , then @xmath126 and all leaves on the forward orbit can intersect the ( at most 2 ) preimage leaves of @xmath126 of length at least @xmath350 only on the boundary .
a real number @xmath28 is called _ univoque _ if @xmath27 admits a _ unique _ expansion in base @xmath29 , i.e. if there exists a _ unique _
sequence @xmath358 such that @xmath359 the set @xmath360 of all univoque numbers is called the _
univoque set_. in the 90 s , @xcite discovered that there are infinitely many univoque numbers .
in fact @xcite showed that , given any binary sequence @xmath361 , then equation defines a univoque number if and only if the following condition is met @xmath362 where the minor and major signs refer to the lexicographical order and @xmath363 as usual .
binary sequences which satisfy equation are called _
admissible_. it is not difficult to establish ( by means of the so called _ greedy _ algorithm ) that equation defines an order preserving bijection between @xmath364 and the set of all admissible sequences .
the properties of @xmath365 were studied by various authors , leading in particular to the determination ( in @xcite ) of the smallest element , @xmath366 , which is not isolated in @xmath365 and whose corresponding admissible sequence is the shifted thue - morse sequence @xmath367 .
the geometric structure of @xmath365 is specified by the following properties : in a series of papers ( see @xcite , @xcite , @xcite ) , j .-
p . allouche and
m. cosnard introduced the number set : = \ { x : 1-x \{2^k x } x , kn } and recognized ( @xcite , proposition 1 ) admissible binary sequences to be in a one - to - one correspondence with the elements of @xmath368 which have non - periodic expansions .
it is not difficult to realize that the set @xmath368 essentially coincides with @xmath15 , namely let @xmath369 be a binary expansion of @xmath370 $ ] . from ( [ tenda ] )
it follows that @xmath371 define @xmath372 then @xmath373 : s_k(x ) \leq x , \ ; \forall k \in { \mathbb n}\}$ ] and since @xmath374 we have @xmath375 .
let now @xmath376 , and let @xmath377 ( @xmath378 since @xmath6 ) . given @xmath379 , either @xmath380 or @xmath381 . if @xmath380 , then @xmath382 . if instead @xmath381 , then there exists a unique @xmath60 s.t .
@xmath383 . thus @xmath384 and hence @xmath385 [ 1to1 ] there is a one - to - one correspondence between @xmath365 and @xmath386 .
more specifically , a number @xmath387 with non - periodic binary expansion belongs to @xmath15 if and only if @xmath388 for some @xmath389 and the expansion is unique .
it is an easy consequence of condition that a sequence @xmath390 is admissible iff @xmath391 in other words , @xmath390 is admissible iff the real value @xmath387 is a nonperiodic element of @xmath368 ; thus our claim follows from lemma [ gl ] .
it has been shown in @xcite that @xmath392 is transcendental .
the argument used there can be adapted to show that the image in @xmath365 of all the numbers @xmath393 dealt with in proposition [ xi ] are transcendental as well .
let @xmath178 be a finite binary word of length @xmath396 , and write @xmath397 and @xmath398 , with @xmath399 .
define the formal power series @xmath400 and @xmath401 . by equation @xmath402 ,
hence @xmath403 as formal power series , since @xmath404 has a unique binary expansion being transcendental . as a consequence , @xmath405 , and
since @xmath406 , @xmath407 .
if @xmath29 is algebraic , then @xmath408 is algebraic , hence also @xmath409 and @xmath410 are algebraic , contradicting mahler s criterion ( @xcite , page 363 ) . from our construction
it follows that rational elements of @xmath15 correspond to algebraic elements in @xmath411 ( and are in fact dense in @xmath411 @xcite ) ; rephrasing the question posed at the end of section [ pedou ] , one might ask whether these are the only non - transcendental elements of @xmath365 .
other geometric properties of the set @xmath364 also follow easily from our correspondence .
let @xmath412 $ ] be the map which associates to @xmath387 the unique value @xmath29 which satisfies the relation .
it is not difficult to check that @xmath413 is locally hlder - continuous and extends to a homeomorphism between @xmath26 and the closure @xmath414 of @xmath364 in @xmath415 $ ] , hence @xmath414 is also a cantor set .
moreover , one can prove that the hlder exponent of @xmath413 restricted to the segments @xmath416 ( considered in section [ hd ] ) is greater than the hausdorff dimension of @xmath416 , thus @xmath417 for all @xmath418 and @xmath414 has zero lebesgue measure .
, _ topological entropy of unimodal maps : monotonicity for quadratic polynomials _ , in _ real and complex dynamical systems ( hillerd , 1993 ) _ , nato adv .
464 * , 6587 , kluwer , dordrecht , 1995 . , _ characterization of the unique expansions @xmath420 and related problems _ bull .
france * 118 * ( 1990 ) , no .
3 , 377390 . , _ fractal geometry - mathematical foundations and applications _ , wiley , 2003 . | in this paper we construct a correspondence between the parameter spaces of two families of one - dimensional dynamical systems , the @xmath0-continued fraction transformations @xmath1 and unimodal maps .
this correspondence identifies bifurcation parameters in the two families , and allows one to transfer topological and metric properties from one setting to the other . as an application
, we recover results about the real slice of the mandelbrot set , and the set of univoque numbers . |
computation of the so - called exchange operator is often the most expensive step in electronic structure models applicable to large systems , such as the hybrid@xcite density functional theory ( dft ) and hartree - fock ( hf),@xcite the latter of which serves as the starting point for electron correlation treatment with modern reduced - scaling many - body methods.@xcite the exchange matrix , @xmath0 , is given by @xmath1 where @xmath2 , @xmath3 , @xmath4 , @xmath5 ,
denote basis functions , @xmath6is the density matrix , and @xmath7 are the electron repulsion integrals ( eris ) in the mulliken `` bra - ket '' notation .
here we consider atom - centered basis functions @xmath8 , typically represented by ( contracted ) gaussian - type orbitals .
although @xmath9 decays with the bra - ket distance @xmath10 only as @xmath11 ( i.e. , slowly ) , the decay of the density matrix @xmath12 in a finite system ( and , generally , any system with a nonzero band gap ) is exponential with the distance between @xmath4 and @xmath5.@xcite therefore , as first noted by almlf,@xcite the number of nontrivial elements of the exchange matrix @xmath0grows linearly with system size that is , @xmath0has @xmath13significant elements ( @xmath14 system size ) .
a number of linear - scaling methods for the computation of @xmath0have been investigated over the years , with pioneering work done in chemistry in the mid-1990s by schwegler , challacombe , and head - gordon@xcite and burant , scuseria , and frisch.@xcite several related algorithms have been developed , e.g. onx by challacombe et al and linkby ochsenfeld , white , and head - gordon@xcite .
the aforementioned @xmath13algorithms for exchange take advantage of the element - wise sparsity of the eri and density matrix tensors in eq . .
some effort has gone into utilization of rank sparsity of the eri tensor , as revealed by the multipole expansion of the coulomb operator,@xcite by the pseudospectral and related factorizations of eri,@xcite or , as done here , by the density fitting approximation of eris.@xcite the density fitting approximation@xcite ( df , also called resolution of the identity ) , expresses the two - center products in the bra and the ket of the eri tensor ( often called `` densities , '' not to be confused with the density matrix @xmath6 ) as linear combinations of one - center functions from an auxiliary basis : @xmath15 where @xmath16 , @xmath17 , are indices in an auxiliary basis set . to obtain the coefficients @xmath18 , one must left - project with the auxiliary basis to obtain a system of linear equations : @xmath19 the solution of these equations in their full form requires 3 effort and is thus untenable for large molecular systems . moreover ,
substitution of the form in eq . into the exchange matrix expression in eq .
does not reduce the formal scaling of the exchange matrix build .
consequently , significant work has been done on methods of reducing the scaling of eq . , principally by localizing the fitting basis.@xcite
significantly less research has been done on the use of these localized decompositions for the exchange matrix build , eq . , particularly in the context of the linear
scaling exchange work of the late 1990s .
indeed , only sodt and head - gordon@xcite have investigated a linear scaling exchange construction with local df .
also , neese and coworkers@xcite presented a linear scaling algorithm for hf exchange in the context of their `` chain - of - spheres '' exchange ( cosx ) method , which borrows elements from both df and the pseudo - spectral method of friesner , et al.@xcite herein , we present a linear scaling hf exchange algorithm and implementation using the concentric atomic density fitting ( cadf ) approach , based on a concept that has been around for some time@xcite and was recently revived by merlot , et al.@xcite and the present authors.@xcite in section [ sec : theory ] , we briefly review the theory behind cadf and develop a diagrammatic scheme for further discussion of linear scaling methods .
section [ sec : link ] presents a progression of algorithms that lead to the final cadf - link method , and section [ sec : impl ] discusses implementation details .
section [ sec : results ] gives the results of some benchmark computations , and section [ sec : conc ] contains some concluding remarks .
the idea behind what we call cadf has been around since the early days of df:@xcite expand a product density @xmath20 using only auxiliary functions centered on the same atom as the composite functions @xmath2 and @xmath3 .
the fitting equations can thus be written @xmath21 where @xmath22 , @xmath23 ,
are indices of atom centers , @xmath24 indicates that the function @xmath2 is centered on atom @xmath22 , and @xmath25 indicates that the function @xmath17 is centered on either atom @xmath22 or atom @xmath23 . in this context , the use of the simple expansion of the integral tensor @xmath26 yields errors in @xmath27that are too large to be useful for hf theory computations .
however , in 2000 dunlap@xcite noted that this expansion is linear in the density error ( unless the full , non - local coulomb metric is used ) , and a more robust expansion correct to second order in the density error can be constructed . in the context of cadf , this robust expansion can be written @xmath28 the use of this robust expansion yields reasonable errors for hf theory on the order of 2 - 5 times the errors arising from conventional density fitting , in most cases.@xcite however , eq
. breaks the positive semidefiniteness of the @xmath27tensor,@xcite which can cause convergence issues in rare cases .
merlot , et al .
and our group both suggested different approaches@xcite to correcting this issue . while we consider the convergence issues with cadf to be eminently solvable , to throughly test the cadf approximation we need to be able to apply cadf - based hartree - fock and density functional theory methods to large molecules .
thus , in this paper the focus is on developing a practical @xmath13cadf - based exchange matrix algorithm , and hence the issues of positivity are set aside herein . due to the rapid decay of the overlap between the gaussian basis function functions @xmath2 and @xmath3 , only @xmath13products @xmath20 result in significant eris .
thus , the four - center eri tensor @xmath27contains 2 significant entries .
hser and ahlrichs@xcite noted that the cauchy
schwarz inequality holds for the eri tensor , and thus this exponential decay could be exploited for prescreening using the inequality @xmath29 using the nomenclature of neese and coworkers,@xcite we say that @xmath2 forms an `` s - junction '' with @xmath3 if and only if @xmath30 is greater than some pairing threshold @xmath31 . to illustrate the screening of significant contributions to the exchange matrix
, we introduce a simple graphical method for showing contributing factors to the sparsity of a given tensor or tensor contraction expression .
tensor indices are shown as vertices of a diagram .
each of these indices has a range of @xmath13 , hence a diagram with @xmath32 vertices denotes to @xmath33 total contributions to a tensor or a tensor contraction .
an edge connecting two vertices denotes that only @xmath13 _ pairs _ of these indices are significant for any finite precision in the limit of infinite system size .
in other words , for each first index value there is a @xmath34 list of significant second index values .
for example , the eri tensor screened using eq . can be represented as @xmath35 ( mu ) @xmath2 ; ( la ) [ right = of mu ] @xmath4 ; ( nu ) [ right = of la ] @xmath3 ; ( si ) [ right = of nu ] @xmath5 ; ( mu )
node[above , font= ] s ( la ) ; ( nu )
node[above , font= ] s ( si ) ; [ d : g ] from this diagram , it is immediately apparent that the screening in eq .
requires 2 integrals to be computed for @xmath27 , since there are a constant number of @xmath4 for each @xmath2 and a constant number of @xmath5 for each @xmath3 , but no connection between @xmath2 and @xmath3 .
therefore , the diagram must be fully connected for a tensor or contraction to have @xmath13 significant contributions .
@xmath13 exchange construction algorithms take advantage of the aforementioned rapid decay of the density matrix.@xcite diagrammatically , the conventional exchange matrix expression @xmath36 can be represented as @xmath37 ( mu ) @xmath2 ; ( la ) [ right = of mu ] @xmath4 ; ( si ) [ right = of la ] @xmath5 ; ( nu ) [ right = of si ] @xmath3 ; ( mu )
node[above , font= ] s ( la ) ; ( nu )
node[above , font= ] s ( si ) ; ( la )
node[above , font= ] p ( si ) ; [ eq : kdiag ] where the connection between @xmath4 and @xmath5 is called a `` p - junction '' in the nomenclature of neese , et al.@xcite diagram is fully connected , which suggests the existence of a linear scaling algorithm .
note also that the path length between two indices in the result suggests ( in a qualitative sense ) some aspects of the performance of the associated linear scaling algorithm .
since there are a bounded number of @xmath4 for a given @xmath2we will use the notation @xmath38as well as a bounded number of @xmath5 given @xmath4 and a bounded number of @xmath3 given @xmath5 , the prefactor of the linear scaling algorithm will be bounded from above by the product @xmath39 , because one could , at worst , use the direct product of these index sets to construct the tensor . in practice , the scaling will be much better than this ( since , e.g. , the product of medium - sized contributions to each connection may still be quite small ) , though there may not necessarily be a reasonable algorithm to access this scaling .
nevertheless , the diagrammatic approach presents a concise picture of the nature of the factors that necessarily contribute to the performance of any linear scaling algorithm for the construction of @xmath0 .
it is clear from the form of the gaussian product rule ( gpr ) that @xmath38 and @xmath40 will depend on the number of basis functions per atom and the diffuseness of those functions . from physical reasoning
, we note that @xmath41 will further depend on the band gap of the system and the degree of delocalization of electrons in the system . given these properties of s and p junctions ,
we conclude that a linear scaling algorithm associated with this diagram would probably perform poorly for large basis sets or systems with small band gaps .
while this is not a particularly profound insight ( both of these are well - known issues with virtually all linear scaling methods ) , it suggests by contrast what a method that improves on these limitations might look like : such a method would have to provide an alternative path between @xmath2 and @xmath3 that does not rely on the square of the s junction constant ( for large basis sets ) or the p - junction constant ( for smaller band - gap systems ) .
our task herein is to identify useful additional pathways between @xmath2 and @xmath3 .
cadf introduces another kind of junction between indices , that of concentricity . using the diagrammatic approach ,
we can write the coefficient tensor as a union of two diagrams : @xmath42[font= ] ( lpic ) ( mu ) @xmath24 ; ( la ) [ right = of mu ] @xmath43 ; ( x ) [ below = of mu ] @xmath44 ; ( mu ) node[above , font= ] ( s ) s ( la ) ; ( x )
node[left , font= ] ca ( mu ) ; ; ( un ) [ font= , right = of lpic ] @xmath45 ; ( rpic ) [ right = of un ] ( mu ) @xmath24 ; ( la ) [ right = of mu ] @xmath43 ; ( x ) [ below = of la ] @xmath46 ; ( mu ) node[above , font= ] ( s ) s ( la ) ; ( x )
node[left , font= ] ca ( la ) ; ; ( eq ) [ font= , right = of rpic ] @xmath47 ; ( epic ) [ right = of eq ] ( mu ) @xmath24 ; ( la ) [ right = of mu ] @xmath43 ; ( mid ) at ( @xmath48 ) ; ( x ) [ below = of mid.south west ] @xmath49 ; ( mu )
node[above , font= ] ( s ) s ( la ) ; ( x )
node[left , font= ] ca ( mid.south west ) ; ; where `` ca '' is short for `` concentric atomic , '' indicating that the connected indices are centered on the same atom , and the squiggled lines are used to introduce a shorthand for the left - hand side of the equals sign . by plugging the robust cadf approximation to eri , eq . , into eq . and relabeling some indices , we can write @xmath0as @xmath50 \sum_{{\ensuremath{\sigma}\xspace}_dx_{(cd)}}\sum_{{\ensuremath{\lambda}\xspace}_b}\left[{\ensuremath{\left ( \mu_a{\ensuremath{\lambda}\xspace}_b \middle| x_{(cd ) } \right ) } } \vphantom { - \tfrac12 \sum_{y_{(ab)}}c_{\mu_a{\ensuremath{\lambda}\xspace}_b}^{y_{(ab)}}{\ensuremath{\left ( y_{(ab ) } \middle| x_{(cd ) }
\right ) } } } \right . \\
. - \tfrac12 \sum_{y_{(ab)}}c_{\mu_a{\ensuremath{\lambda}\xspace}_b}^{y_{(ab)}}{\ensuremath{\left ( y_{(ab ) } \middle| x_{(cd ) } \right)}}\right]d^{{\ensuremath{\lambda}\xspace}_b{\ensuremath{\sigma}\xspace}_d } c_{\nu_c{\ensuremath{\sigma}\xspace}_d}^{x_{(cd)}}. \end{multlined}\end{aligned}\ ] ] diagrammatically , this can be written @xmath51[font= ] ( lpic ) ( nu ) @xmath52 ; ( si ) [ left = of nu ] @xmath53 ; ( mid3 ) at ( @xmath54 ) ; ( x ) [ below = of mid3.south ] @xmath55 ; ( la ) [ left = of si ] @xmath43 ; ( mu ) [ left = of la ] @xmath24 ; ( mu )
node[above , font= ] ( s ) s ( la ) ; ( si )
node[above , font= ] ( p ) p ( la ) ; ( si ) node[above , font= ] ( s ) s ( nu ) ; ( x )
node[left , font= ] ca ( mid3.south ) ; ; ( minus ) [ font=,right = of lpic ] @xmath56 ; ( rpic ) [ right = of minus ] ( nu ) @xmath52 ; ( si ) [ left = of nu ] @xmath53 ; ( mid ) at ( @xmath54 ) ; ( x ) [ below = of mid.south west ] @xmath55 ; ( la ) [ left = of si ] @xmath43 ; ( mu ) [ left = of la ] @xmath24 ; ( mid2 ) at ( @xmath48 ) ; ( y ) [ below = of mid2.south west ] @xmath25 ; ( mu )
node[above , font= ] ( s ) s ( la ) ; ( si )
node[above , font= ] ( p ) p ( la ) ; ( si )
node[above , font= ] ( s ) s ( nu ) ; ( y )
node[left , font= ] ca ( mid2.south west ) ; ( x ) node[left , font= ] ca ( mid.south west ) ; ; both terms in the diagram are connected , hence a linear - scaling algorithm can be designed .
however , it is possible to significantly reduce the computational cost of the first term by screening the three - center two - electron integrals .
three - center eris can be screened more efficiently than the standard four - center eris because the potential created by a solid - harmonic gaussian @xmath58 of angular momentum @xmath59 decays as @xmath60 , rather than the @xmath11 decay of a general product @xmath61 .
the decay of the coulomb operator does not matter as much for the four - center eri - based exchange construction because the density decays so rapidly . in the cadf - based approach ,
efficient screening of the three - center eris is important and can not be achieved using only the schwartz bound .
recently we developed a tight estimator , called sqv@xmath57 , for three - center coulomb integrals that takes into account the correct asymptotic decay of the potential : @xmath62 r >
\operatorname{ext}_{\kappa\eta } + \operatorname{ext}_{\theta } \\ \text{and } s_{\kappa\eta}/q_{\kappa\eta } > { \ensuremath{\vartheta_{\mathrm{sq}}}\xspace}\end{multlined } \\
\frac{\pi\sqrt2\,\beta_{\ell_\theta}\!(\zeta_\theta)\,q_{\kappa\eta}}{(\zeta_\kappa+\zeta_\eta)^{\frac14}r^{\ell_\theta+1 } } & \begin{multlined}[c ] r > \operatorname{ext}_{\kappa\eta }
+ \operatorname{ext}_{\theta } \\ \text{and } s_{\kappa\eta}/q_{\kappa\eta } \leq { \ensuremath{\vartheta_{\mathrm{sq}}}\xspace}\end{multlined } \\ \ q_{\kappa\eta}q_\theta & r \leq \operatorname{ext}_{\kappa\eta } + \operatorname{ext}_{\theta } , \end{dcases } \label{eq : sqvl}\end{aligned}\ ] ] where @xmath63 @xmath64 , @xmath65 , and @xmath66 are the exponents of the various primitives , @xmath67 is the overlap , @xmath68 are cfmm extents , and @xmath69 and @xmath70 are user - defined thresholds ( see ref . for details , including generalization to contracted basis functions ) . practically speaking
, the ratio of prefactors from the first two cases is folded into the @xmath71 ratio , and the second case is used if this ratio is less than one .
effectively , the far field estimator uses the minimum of the first two cases when the @xmath71 ratio is greater than @xmath72 , and only the second case when the ratio is less than @xmath72 .
this latter detail will be important for the use of this estimator in section [ sec : link ] .
the sqv@xmath57 estimator allows us to add a crucial edge to the cadf exchange screening diagram : @xmath73 ( nu ) @xmath52 ; ( si ) [ left = of nu ] @xmath53 ; ( mid ) at ( @xmath54 ) ; ( x ) [ below = of mid.south ] @xmath55 ; ( la ) [ left = of si ] @xmath43 ; ( mu ) [ left = of la ] @xmath24 ; ( mid2 ) at ( @xmath48 ) ; ( mu )
node[above , font= ] ( s ) s ( la ) ; ( si )
node[above , font= ] ( p ) p ( la ) ; ( si )
node[above , font= ] ( s ) s ( nu ) ; ( x )
node[left , font= ] ca ( mid.south ) ; ( x ) .. controls ( x-|la ) and ( x-|mid2.south ) .. node [ below , near start , font= ] @xmath74 ( mid2.south ) ; [ d : sqvl ] note that the efficiency of the sqv@xmath57 screening will increase for kets with higher angular momenta ; hence , the more expensive the three - center eri , the more likely it will be screened out .
like four - center eri - based linear scaling exchange algorithms,@xcite our linear scaling exchange algorithm based on concentric atomic density fitting ( which we will call cadf - link ) relies on the construction of prescreening lists , @xmath75 and @xmath76 , to drive the exchange matrix build . both sets of lists are computed with linear effort .
@xmath75 specifies the list of @xmath4 for a given @xmath77 pair for which @xmath78 must be computed .
it is defined as follows : @xmath79 where @xmath80 , \label{eq : cexpand } \end{multlined } \\
{ { \ensuremath{\tilde r_{\mu{\ensuremath{\lambda}\xspace}}^{x}}\xspace}}&\equiv { \ensuremath{\left ( x \middle| x \right)}}^{-1/2 } i_{\mathrm{{sqv\ensuremath{\ell}\xspace}}}(\mu,{\ensuremath{\lambda}\xspace},x ) \end{aligned}\ ] ] and @xmath81 is the sqv@xmath57screening factor from eq . .
( keep in mind that though the tensor - like notation @xmath82is used for compactness , this quantity is intended to be computed on the fly , so a notation like @xmath83 would be conceptually more appropriate . )
we have loosely used the notation @xmath84 to represent the block frobenius norm of a matrix or tensor expression . in these expressions and the following , the orbital basis set ( obs ) and density fitting basis set ( dfbs ) are distinguished by their respective abbreviations , and the notation @xmath85 , for instance , represents the set of all basis functions from the obs centered on atom @xmath86 . the density fitting coefficients @xmath87 can be obtained in linear effort after a schwarz - based prescreening of obs pairs , as discussed in the original cadf paper.@xcite we immediately note that ( theoretically , at least ) the @xmath88 ( `` d - bar '' ) intermediate can be constructed in linear scaling time , since @xmath89 ( nu ) @xmath52 ; ( si ) [ left = of nu ] @xmath53 ; ( mid ) at ( @xmath54 ) ; ( x ) [ below = of mid.south ] @xmath55 ; ( la ) [ left = of si ] @xmath43 ; ( si )
node[above , font= ] ( p ) p ( la ) ; ( si )
node[above , font= ] ( s ) s ( nu ) ; ( x )
node[left , font= ] ca ( mid.south ) ; is a fully connected diagram .
indeed , the construction of @xmath88 is quite straightforward if the density matrix @xmath6is stored in a sparse data structure .
the construction of @xmath90 can also be performed in linear scaling effort , since the norm in the second term of eq .
need only loop over the @xmath91 that have significant overlap with @xmath53 . with this in mind , a procedure for the linear scaling construction of @xmath92 is given in algorithm [ alg : l3 ] .
similar to the original linkalgorithm,@xcite the key to linear scaling in algorithm [ alg : l3 ] is the list ordering that allows the program to exit the loops early in lines [ al : exit1 ] and [ al : exit2 ] .
the auxiliary list set @xmath93 is composed of @xmath16 for which @xmath94 is significant ; each list is then sorted by decreasing @xmath94 value .
the other helper list set , @xmath95 , is simply the @xmath2 that form schwarz pairs with @xmath4 , sorted by decreasing value of @xmath96 .
@xmath97 @xmath98 with @xmath99 sort @xmath93 by decreasing @xmath100 compute distance factor @xmath82 add @xmath4 to @xmath75 * break * out of @xmath2 loop [ al : exit1 ] * break * out of @xmath16 loop [ al : exit2 ] using only the 3 lists , a quadratic scaling @xmath0build algorithm can be devised .
such a procedure is outlined in algorithm [ alg : krepl ] .
note that the subscripted indices ( e.g. , @xmath24 ) are descriptive rather than prescriptive : the subscript @xmath22 merely indicates that the center on which the function @xmath2 is located has the index @xmath22 ; it does not indicate anything specific about the iteration range of @xmath24 .
the algorithm loops over the significant @xmath101 integrals and immediately contracts these contributions into the @xmath102intermediate .
the quadratic scaling of this algorithm arises because the loop in line [ al : bloop ] is not restricted ; the restriction of this loop requires @xmath103 , discussed in section [ sec : lstb ] .
form 3x for significant @xmath104 compute @xmath105 @xmath106 @xmath107 @xmath108
[ al : bloop ] [ al : kbegin ] @xmath109[al : kpart1 ] @xmath110[al : kpart2 ]
[ al : kend ] @xmath111 several important aspects of this quadratic scaling algorithm merit further discussion before the introduction of the linear scaling version .
first , note that the summation of @xmath112 into @xmath113 ( lines [ al : kbegin][al : kend ] of algorithm [ alg : krepl ] ) is split into two loops with one index restricted in each loop .
this restriction is made possible by the cadf coefficient definition , which precludes coefficients with dfbs functions that do not share a center with the obs pair being fit .
this loop structure also makes it possible to store the coefficients in an ordering more efficient for vectorization : @xmath114 is stored as @xmath115 \rightarrow { \ensuremath{\sigma}\xspace}_d$ ] ( from slowest- to fastest - running index ) , which is still a @xmath116 data structure .
secondly , note that the algorithm is integral direct , and that the algorithm s storage requirements never exceed 2 .
the intermediate @xmath112 requires linear storage in this algorithm , and the integrals @xmath78 need not be stored outside of the inner loop in which they are computed . in order to further improve the scaling of algorithm [ alg : krepl ] to linear , another set of prescreening lists , @xmath103 , must be formed . for a given @xmath2 and @xmath16 ,
the list @xmath76 from the set b is defined as @xmath117 where @xmath118 as with @xmath119 , the formation of @xmath120 in linear effort is relatively trivial if @xmath121 is stored in a sparse data structure , and particularly if @xmath75 is sorted by decreasing @xmath82 .
an algorithm for the linear scaling formation of b is outlined in algorithm [ alg : lb ] .
the procedure is greatly simplified by the fact that a list of significant @xmath77 pairs is already known from the construction of 3 .
the algorithm is little more than a sparse - sparse matrix multiply .
if @xmath90 is stored in a sparse data structure , line [ al : lc ] is already part of the manifestation of @xmath90 in memory .
similarly , the conditional in line [ al : lbif ] is essentially the procedure that a sparse - sparse matrix multiply undergoes to determine the nontrivial entries in the product matrix .
nonetheless , we have included the b build algorithm here for completeness .
@xmath122 with @xmath123
[ al : lc ] [ al : lbif ] add @xmath5 to @xmath76 the use of b to create a @xmath0build procedure that scales linearly requires only a few modifications to algorithm [ alg : krepl ] .
the revised procedure is given in algorithm [ alg : kb ] .
the main changes involve restricting the loops in lines [ al : blr ] and [ al : klr ] .
the restriction in line [ al : blr ] amounts to excluding from @xmath102any @xmath124 where the contraction over the full set of significant @xmath43 for a given @xmath125 is negligible .
these indices can then also be excluded from the summation in line [ al : klr ] , since the @xmath126 are trivial for these @xmath53 .
further optimization could be made by restricting the loop in line [ al : blr ] to include only @xmath53 for a given @xmath43 where @xmath127 is significant .
this optimization would arise naturally if a sparse data structure were used for @xmath6 , but in the present implementation dense 2 data structures were used ( see section [ sec : impl ] ) .
form 3x for significant @xmath104 form bx for significant @xmath104 [ al:3lr ] compute @xmath105
[ al : intc ] @xmath106 @xmath107
[ al : blr ] @xmath108 [ al : bcon ] [ al : klr ] @xmath109 [ al : kcon1 ] [ al : k2out ] [ al : schwloop ] @xmath110 [ al : kcon2 ] @xmath111
the algorithms detailed in section [ sec : link ] were implemented in a development version ( commit tag 3.0.0-cadflink of the localdf branch ) of the massively parallel quantum chemistry ( mpqc)@xcite quantum chemistry package .
while the algorithms were implemented predominantly as written here , several small details differ from the ideal linear scaling implementation .
most prominently , since integrals are computed more efficiently in shell blocks , all of the indices in the algorithms actually represent shell blocks rather than individual basis functions . in other words , for instance , @xmath128 where @xmath129 is a function index and @xmath16 is a shell index .
the upshot of this is that the prescreening algorithms and list formations are significantly less expensive than the main computation in practice , even though in principle the onset of linear scaling is later for these portions of the computation .
( the prescreening is more expensive in principle because the exit conditions of the loops in algorithm [ alg : l3 ] use the schwarz estimate of the integrals rather than the distance - including estimate , since the latter can not be easily ordered ) .
the quantities used in the screening process are replaced by their shell block frobenius norm analogs .
indeed , if a scenario were to arise in which the screening portions of the algorithm began to dominate the cost , the concept of shell blocks could be generalized further to arbitrary blocks and a multi - tiered prescreening approach could be used .
another difference in our implementation is that dense data structures were used to store all 2 quantities .
the primary reason for this is that the scf solver in mpqc is currently based on an 3 eigensolve ; alternative @xmath13 scf solvers are well known ( e.g. , density matrix minimization@xcite ) but are not yet implemented in our program .
it is difficult to measure the practical scaling of an algorithm outside of the context of its implementation details and execution environment .
thus , many authors@xcite opt to present their algorithmic scaling in terms of integral counts , contraction sizes , or other implementation- and execution - independent metrics . here
we use several such metrics corresponding to particular sections of algorithm [ alg : kb ] .
our implementation of cadf - link has been heavily optimized to run well on massively parallel computers and to take advantage of both thread and process parallelism .
however , a thorough discussion of the challenges involved in the parallelization of cadf - link has been reserved for a separate paper in the interest of both saving space and appealing to a more general audience .
for our cost metrics , we chose a series of one dimensional molecules ( linear alkanes ) and a series of three dimensional systems ( water clusters ) .
cartesian coordinates are included in the supplemental information .
we used the basis set pairs def2-svp@xcite
/ def2-svp / jk,@xcite cc - pvtz@xcite / cc - pvtz / jk,@xcite and cc - pvqz / cc - pvqz / jk . for these series of molecules and basis sets , we recorded the number of three - center integrals computed ( line [ al : intc ] ) , the number of multiplies in the contraction to form @xmath112 ( line [ al : bcon ] ) , and the number of multiplies in the contractions to form @xmath130 ( lines [ al : kcon1 ] and [ al : kcon2 ] ) .
these data were averaged over the first three scf iterations ; later iterations were omitted to minimize variability due to convergence acceleration procedures . for the computations in the remainder of this work ,
an scf convergence criterion of @xmath131 was used .
our method has no difficulty converging further than this , but since the emphasis is on per iteration performance , we chose a relatively loose convergence threshold .
an initial screening threshold @xmath132 of @xmath131 was used throughout as well .
however , this number was varied for differential density iterations as follows : the screening threshold for a given iteration was taken to be the initial threshold if the full density was used or the ratio of differential density to full density frobenius norm times the initial threshold if a differential density was used , down to a minimum threshold of @xmath133 . for the sqv@xmath57estimate ,
@xmath134 was used for both @xmath69 and @xmath70.@xcite this use of a scaled screening threshold allowed us to use a quite aggressive initial threshold without sacrificing as much accuracy in the final result .
this amounts to a simplified version of a previous variable precision scf approach@xcite that yields excellent results for large molecules .
indeed , we were usually able to converge scf computations to a root mean squared density change of @xmath135 or less using an initial threshold of @xmath131 .
this thresholding scheme also spreads the workload across iterations in a fairly uniform manor : in almost all cases , no iteration costs more than a factor of 2 different from any other iteration ( without threshold scaling , factors of 5 - 10 were often observed ) .
the top plot uses a linear range , and the bottom plot shows the same data on a logarithmic range ( subtracting 1 to show proximity to linear scaling ) .
, scaledwidth=50.0% ] to analyze the asymptotic scaling of a positive computational cost metric @xmath136 with the system size parameter @xmath137 it is convenient to introduce of an effective scaling exponent : @xmath138 this amounts roughly to a `` two - point fit '' to the form @xmath139 , for constants @xmath22 and @xmath23 , at the points @xmath140 and @xmath141 . by definition , the effective exponent of the cost of a @xmath13algorithm should approach 1.0 in the limit of large @xmath137 .
figure [ fig : scaling ] shows the effective scaling exponents for the key cost metrics of our algorithm when applied to our one- and three - dimensional series of molecules .
it is clear that all scaling exponents are less than 2 for @xmath142 , and decrease monotonically with @xmath137 .
also , the @xmath0contraction step has the worst scaling ( highest effective exponent ) .
this is anticipated due to the relatively weak restriction of the loop in line [ al : schwloop ] of algorithm [ alg : kb ] , which only utilizes schwarz screening . however , since the non - link behavior of the @xmath143 contraction is better than other parts of the algorithm ( 3 with no screening compared to 4 for the @xmath112 contraction without screening ) , the @xmath0contraction still costs less in terms of cpu time than the @xmath102contraction , even for our largest computations . nevertheless , the scaling with respect to the @xmath0contraction metric is good , reaching @xmath144 around 2500 basis functions for linear alkanes with a small basis and @xmath145 around 4000 and 10000 basis functions for a triple zeta and quadruple zeta basis , respectively . even for three dimensional systems ,
the scaling of the @xmath0contraction is still around @xmath146 by about 4000 and 15000 basis functions for the def2-svp / def2-svp / jk and cc - pvtz / cc - pvtz / jk basis sets , respectively .
the scaling behavior of the more expensive sections , the integral computation and @xmath102contraction , is even better .
the scaling reaches @xmath144 around 1300 , 3500 , and 8700 basis functions , respectively , for linear alkanes with the three basis set pairs in our study . for our three dimensional systems ,
the scaling of the @xmath102contraction is about @xmath146 by around 2400 and 5800 basis functions for def2-svp / def2-svp / jk and cc - pvtz / cc - pvtz / jk , respectively . while the scaling exponent functions are less smooth for water clusters compared to linear alkanes due to the less systematic growth of the former , the overall trends in the data are similar for both one- and three - dimensional systems . obviously , the decay of the scaling exponents is much more rapid for one dimensional systems and for smaller , less diffuse basis sets . in the worst case three dimensional water clusters with the cc - pvtz / cc - pvtz / jk basis set
the scaling of the most computationally intense sections of the algorithm is around @xmath147 for a system with 756 atoms . given that in the context of hf and hybrid ks dft the typical basis sets are smaller and less diffuse than cc - pvtz , it completely reasonable to conclude that cadf - link will closely approach linear scaling behavior in the vast majority of use cases .
figure [ fig : errors ] shows the errors in absolute energies resulting from the cadf - link approximation , relative to cadf - scf with schwarz screening only ( the coulomb matrix was computed in the same manner in both sets of computations ) . as with the cadf approximation and hartree
fock theory itself , the absolute energy errors arising from the cadf - link approximation scale linearly with the size of the system
. however , these errors are still about one or two orders of magnitude smaller that the errors arising from the cadf approximation for molecules of comparable size ( see ref . ) , which are in turn smaller than the basis set errors or the errors of the hartree
fock method itself .
importantly , as demonstrated by the bottom plot in figure [ fig : errors ] , the absolute error per electron does not increase significantly beyond a point for the molecules we tested . the larger error per electron for three dimensional structures compared to one dimensional ones
can be heuristically rationalized as follows .
as two basis functions are separated from each other , their pair - wise contribution ( roughly speaking ) to the energy is excluded at some distance .
a three dimensional structure will have more basis functions at or near this distance than a one dimensional structure .
furthermore , in the specific case of water clusters compared to linear alkanes , the latter have a significantly larger number of covalent bonds , meaning that a much smaller portion of the contributions to the energy are likely to have separations near this critical exclusion length . another notable feature of figure [ fig : errors ] is the error behavior when the distance factor ( from the sqv@xmath57estimator ) is omitted . as the original sqv@xmath57paper noted,@xcite the estimator performs worse for the def2-@xmath16vp / def2-@xmath16vp / jk basis sets than the cc - pv@xmath16z / cc - pv@xmath16z / jk basis sets , because the former contains more contracted basis functions , particularly in the fitting basis , than the latter . since the sqv@xmath57estimator is tightest for uncontracted integrals , the error arising from its use with contracted basis functions
is expected to be larger than with basis sets containing fewer contracted functions .
the data in figure [ fig : errors ] show this , but they also show that the increase in error from the use of the sqv@xmath57estimator is relatively small : in the def2-svp / def2-svp / jk case with linear alkanes , the sqv@xmath57estimator causes roughly a factor of two increase in the cadf - link error , and for water clusters the increase is even less .
figure [ fig : speedup ] shows the speedups for the three cost metrics we examined relative to a cadf - based exchange build with schwarz screening only that is , the ratios of each cost metric for cadf exchange with schwarz screening only to the cost metric for cadf - link for a given system size .
the most substantial speedups as much as 250 times are seen in the @xmath102contraction , which is expected given that both the inner and outer loops are restricted by cadf - link ( by the b@xmath148 and 3@xmath148 lists , lines [ al : blr ] and [ al:3lr ] respectively in algorithm [ alg : kb ] ) , whereas the 3 center integrals and the @xmath0contraction are only restricted by one list each .
indeed , the only significant additional restriction of the contraction in line [ al : kcon2 ] relative to schwarz screening comes from the outermost loop over @xmath104 pairs ( since the stronger concentricity restriction of the loop in line [ al : k2out ] is imposed even without the formation of bx ) .
consequently , the least substantial speedups are seen in the @xmath0contraction , though again this section is the least computationally intense of the three even for the largest computations we performed .
these data clearly show that the small increase in error ( see section [ sec : err ] ) is , in most contexts , more than compensated for by a massive decrease in computational effort .
another noteworthy trend in the data from figure [ fig : speedup ] is that for the first two cost metrics ( 3 center integrals and the @xmath102contraction ) , larger speedups were seen for three dimensional systems , while for the @xmath0contraction , larger speedups were seen for linear alkanes .
this is attributed to the more prominent role of the distance factor in the 3x definition compared to the bx definition . for 3x ,
the distance factor @xmath149 contributes directly to the thresholding for the inclusion of @xmath4 , whereas for bx , the distance factors for all @xmath4 connected by a p - junction to a given @xmath5 contribute to the thresholding for the inclusion of @xmath5 .
heuristically , the 3 center integrals cost metric is most influenced by the efficiency of 3x , and the @xmath0contribution is most influenced by the efficiency of bx ( the @xmath102cost metric is influenced by both ) . since we have already noted in section [ sec : err ] that the three - dimensional structures are more susceptible to the effects of the distance - including estimator , it makes sense that more substantial gains would be seen with water clusters for the 3 center integrals and the @xmath102contraction and with linear alkanes for the @xmath0contraction .
the more interesting trend in figure [ fig : speedup ] is the contribution to the speedup from the distance - dependent integral screening . for large water clusters with the def2-svp / def2-svp / jk basis pair , nearly tenfold speedups
are observed compared to cadf - link without the sqv@xmath57estimator .
this is remarkable , given that in the context of four - center eri - based link the use of distance - dependent screening results in speedups of about 2.0.@xcite this further demonstrates that there is more to be gained from screening three - center eris than from screening their four - center counterparts .
we have presented a linear scaling algorithm for computing the exchange matrix using the concentric atomic density fitting approximation .
our algorithm has been shown to perform well for basis sets of all sizes , and we have carried out some of the largest triple- and quadruple - zeta basis computations ever to demonstrate this point .
errors in absolute energies from the cadf - link approximation are substantially smaller than other sources of error and have been shown to grow linearly with basis set size . even for large basis sets ,
our method shows near linear scaling for systems of less than 1000 atoms , and for smaller basis sets the onset of linear scaling is even more rapid .
not only does our algorithm serve as a highly efficient way to compute the exchange matrix in the context of hartree - fock and dft methods , but also it offers a blueprint for the use of the concentric atomic density fitting in many - body methods .
the research by dsh and efv was supported by nsf grants che-0847295 , che-1362655 , and aci-1047696 , and a camille dreyfus teacher - scholar award .
the research by dsh and hfs was supported by nsf grant che-1361178 .
this work used resources of the national energy research scientific computing center , which is supported by the office of science of the u. s. department of energy under contract no . de - ac02 - 05ch11231 .
49ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) `` , '' ( , , ) p. @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , vol .
( ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | a linear - scaling algorithm is presented for computing the hartree fock ( hf ) exchange matrix using concentric atomic density fitting .
the algorithm utilizes the stronger distance dependence of the three - center electron repulsion integrals along with the rapid decay of the density matrix to accelerate the construction of the exchange matrix .
the new algorithm is tested with computations on systems with up to 1536 atoms and 15585 basis functions , the latter of which represents , to our knowledge , the largest quadruple - zeta hf computation ever performed .
our method handles screening of high angular momentum contributions in a particularly efficient manner , allowing the use of larger basis sets for large molecules without a prohibitive increase in cost . |
the original manuscript should be modified as follows : in section 2.2 , the equation ( 12 ) that describes the curvature effects on the shock velocity of the supernova remnant - cloud interaction : @xmath4 where @xmath5 should be more precisely integrated up to : @xmath6 this equation has an exact solution that should replace the approximate one given in the original manuscript by figure 2 . with the integration limits above the exact solution of eq .
( 12 ) is given by : @xmath7 where @xmath8 @xmath9 \nonumber\end{aligned}\ ] ] and @xmath10 @xmath11 @xmath12}\ ] ] @xmath13 @xmath14 the solution above to @xmath15 will produce slight changes on the multiplying factors that appear in the equations ( 13 ) to ( 26 ) of the original manuscript as indicated below : @xmath16 @xmath17 @xmath18 @xmath19 @xmath20 @xmath21 @xmath22 @xmath23 @xmath24 @xmath25 in section 3.2 , @xmath26 , should be replaced by @xmath27 , since ( as remarked in the original manuscript ) @xmath28 is a more representative average value of the time scale at which the density of the shocked material of the cloud drops by a factor of two after the impact according to radiative cooling numerical simulations ( melioli et al . 2005 ) . also , the exponent of @xmath29 in eq .
( 27 ) should be replaced by @xmath30 , so that the correct equation reads : @xmath31 this implies modifications also in the @xmath29 exponents of eqs .
( 28 ) and ( 29 ) , as described below , respectively by : @xmath32 and @xmath33 in section 3.3 , there is a typo in eq .
the 9/16 factor should be replaced by 25/8 , i.e , : @xmath34 considering this correction and the one due to @xmath35 in eq .
( 12 ) , the multiplying factor 75 in eq . ( 32 ) should be replaced by 160 , or :
@xmath36 the modifications above , particularly those in eqs .
( 27 - 28 ) and eqs.(31 - 32 ) result slight modifications in the diagrams of figure 3 which should be replaced by the figure below .
= 1 pc ; second panel from top : @xmath29 = 5 pc ; third panel : @xmath29 = 10 pc ; and bottom panel : @xmath29 = 20 pc .
solid ( black ) line : upper limit for complete cloud destruction after an encounter with an adiabatic snr derived from eq .
28 ; dashed ( red ) line : upper limit for the shocked cloud to reach the jeans mass derived from eq .
25 for an interaction with an adiabatic snr ; dotted ( blue ) lines : upper limits for the shock front to travel into the entire cloud before being decelerated to subsonic velocities derived from eqs . 30 to 34 for different values of the cooling function @xmath37 ( lower curve ) , @xmath38 ( middle curve ) , and @xmath39 erg @xmath40 s@xmath41 ( upper curve ) ; dotted - dashed ( pink ) line : maximum radius reached by a snr in an ism with density n=0.05 @xmath3 and temperature t=@xmath42 k derived from eq .
the shaded area defines the region where star formation can be induced by a snr - cloud interaction ( between the solid , dashed and dotted lines ) .
the dark shaded zone is bounded by the @xmath37 erg @xmath40 s@xmath41 dotted ( blue ) curve , while the light shaded zone is bounded by the @xmath43 erg @xmath40 s@xmath41 dotted ( blue ) curve .
the crosses in the panels indicate the initial conditions assumed for the clouds in the numerical simulations described in section4.2 of the original manuscriptce , title="fig:",width=226 ] = 1 pc ; second panel from top : @xmath29 = 5 pc ; third panel : @xmath29 = 10 pc ; and bottom panel : @xmath29 = 20 pc .
solid ( black ) line : upper limit for complete cloud destruction after an encounter with an adiabatic snr derived from eq .
28 ; dashed ( red ) line : upper limit for the shocked cloud to reach the jeans mass derived from eq .
25 for an interaction with an adiabatic snr ; dotted ( blue ) lines : upper limits for the shock front to travel into the entire cloud before being decelerated to subsonic velocities derived from eqs . 30 to 34 for different values of the cooling function @xmath37 ( lower curve ) , @xmath38 ( middle curve ) , and @xmath39 erg @xmath40 s@xmath41 ( upper curve ) ; dotted - dashed ( pink ) line : maximum radius reached by a snr in an ism with density n=0.05 @xmath3 and temperature t=@xmath42 k derived from eq .
9 . the shaded area defines the region where star formation can be induced by a snr - cloud interaction ( between the solid , dashed and dotted lines ) .
the dark shaded zone is bounded by the @xmath37 erg @xmath40 s@xmath41 dotted ( blue ) curve , while the light shaded zone is bounded by the @xmath43 erg @xmath40 s@xmath41 dotted ( blue ) curve .
the crosses in the panels indicate the initial conditions assumed for the clouds in the numerical simulations described in section4.2 of the original manuscriptce , title="fig:",width=226 ] = 1 pc ; second panel from top : @xmath29 = 5 pc ; third panel : @xmath29 = 10 pc ; and bottom panel : @xmath29 = 20 pc .
solid ( black ) line : upper limit for complete cloud destruction after an encounter with an adiabatic snr derived from eq .
28 ; dashed ( red ) line : upper limit for the shocked cloud to reach the jeans mass derived from eq .
25 for an interaction with an adiabatic snr ; dotted ( blue ) lines : upper limits for the shock front to travel into the entire cloud before being decelerated to subsonic velocities derived from eqs . 30 to 34 for different values of the cooling function @xmath37 ( lower curve ) , @xmath38 ( middle curve ) , and @xmath39 erg @xmath40 s@xmath41 ( upper curve ) ; dotted - dashed ( pink ) line : maximum radius reached by a snr in an ism with density n=0.05 @xmath3 and temperature t=@xmath42 k derived from eq .
9 . the shaded area defines the region where star formation can be induced by a snr - cloud interaction ( between the solid , dashed and dotted lines ) .
the dark shaded zone is bounded by the @xmath37 erg @xmath40 s@xmath41 dotted ( blue ) curve , while the light shaded zone is bounded by the @xmath43 erg @xmath40 s@xmath41 dotted ( blue ) curve .
the crosses in the panels indicate the initial conditions assumed for the clouds in the numerical simulations described in section4.2 of the original manuscriptce , title="fig:",width=226 ] = 1 pc ; second panel from top : @xmath29 = 5 pc ; third panel : @xmath29 = 10 pc ; and bottom panel : @xmath29 = 20 pc .
solid ( black ) line : upper limit for complete cloud destruction after an encounter with an adiabatic snr derived from eq .
28 ; dashed ( red ) line : upper limit for the shocked cloud to reach the jeans mass derived from eq .
25 for an interaction with an adiabatic snr ; dotted ( blue ) lines : upper limits for the shock front to travel into the entire cloud before being decelerated to subsonic velocities derived from eqs . 30 to 34 for different values of the cooling function @xmath37 ( lower curve ) , @xmath38 ( middle curve ) , and @xmath39 erg @xmath40 s@xmath41 ( upper curve ) ; dotted - dashed ( pink ) line : maximum radius reached by a snr in an ism with density n=0.05 @xmath3 and temperature t=@xmath42 k derived from eq .
9 . the shaded area defines the region where star formation can be induced by a snr - cloud interaction ( between the solid , dashed and dotted lines ) .
the dark shaded zone is bounded by the @xmath37 erg @xmath40 s@xmath41 dotted ( blue ) curve , while the light shaded zone is bounded by the @xmath43 erg @xmath40 s@xmath41 dotted ( blue ) curve .
the crosses in the panels indicate the initial conditions assumed for the clouds in the numerical simulations described in section4.2 of the original manuscriptce , title="fig:",width=226 ] few remarks are in order with regard to the figure : 1 . in the original manuscript ,
the dotted ( blue ) curves of the diagrams were built only for one value of the radiative cooling function of the shocked material , i.e. @xmath44 which is valid for a diffuse gas with a temperature @xmath45 100 k and ionization fraction @xmath46 10@xmath47 .
considering that the constraint established by the dotted ( blue ) curve in the diagrams is highly sensitive to the parameter ( through eq .
31 ) which in turn , can vary by two orders of magnitude depending on the value of the ionization fraction of the cloud gas , we have presently plotted in the diagrams three different dotted ( blue ) curves in order to cover a more realistic range of possible ionization fractions from 0.1 to 10@xmath47 , corresponding to @xmath37 ( lower dotted curve ) , @xmath48 ( middle curve ) , and @xmath39 erg @xmath40 s@xmath41 ( upper dotted curve ) , respectively ( see dalgarno & mccray 1972 ) . the lower dotted ( blue ) curve ( larger ionization fraction )
bounds the dark shaded zone , while the upper one ( smaller ionization fraction ) bounds the light shaded zone of the diagrams .
the middle dotted ( blue ) curve corresponds to the average value of @xmath49 in the range above , @xmath48 erg @xmath40 s@xmath41 , and could be taken as a reference .
2 . in the solution presented in the original manuscript for the cloud with @xmath50 1 pc ( top panel of the figure ) , there was no permitted shaded zone where induction of star formation by snr - cloud interaction would be allowed . according to the present corrections and modifications
, we see that a very thin shaded `` star - formation unstable '' zone appears now when the cooling function @xmath49 has values which are smaller than @xmath48 erg @xmath40 s@xmath41 , or ionization fractions @xmath51 .
the cross labeled in the bottom panel of the figure for a @xmath50 20 pc diffuse cloud corresponds to the initial conditions of the numerical simulations presented in figure 6 of the original manuscript ( that is , for a snr at a distance @xmath52 42 pc from the surface of the cloud ) . in the original manuscript , that cross lies outside the unstable shaded zone just above the upper limit for a complete shock penetration into cloud ( the dotted , blue line of the diagram ) . with the present modifications ,
the cross now lies near the upper limit of the dark shaded unstable zone for values of the cooling function @xmath53 erg @xmath40 s@xmath41 , or ionization fractions @xmath54 .
this result remarks how sensitive the analytical diagrams are to the choice of @xmath49 ( or the ionization fraction ) for a given initial temperature cloud .
according to the radiative cooling chemo - hydrodynamical simulations of figure 6 of the original manuscript ( which corresponds to the cross in the diagram ) , the snr shock front really stalls within the cloud before being able to cross it completely and after all the shocked cloud material does not reach the conditions to become jeans unstable , as predicted , but the dense cold shell that develops may fragment and later generate dense cores , as suggested in the original manuscript .
this points to an ambiguity of the results due to their sensitivity to @xmath49 and the real ionization fraction state of the gas .
we should also remark that the constraint established by the dotted ( blue ) curves in the diagrams is actually only an upper limit for the condition of penetration of the shock into the cloud . in order to estimate the density of the shocked material at the time @xmath55 when the shock stalls within the cloud ( see eq .
31 of the original manuscript or eq .
15 of this erratum ) , we have assumed pressure equilibrium between the shocked and the unshocked cloud material .
a quick exam of the numerical simulations of figure 6 , however , shows that the shock front stalls before this balance is attained .
this implies that the time @xmath55 should be smaller and therefore , the dotted ( blue ) curves in the diagrams should lie below the location predicted by eq .
( 31 ) . in spite of the important alterations above in the diagrams of figure 3 , the main results and conclusions of the original manuscript of melioli et al .
( 2006 ) , particularly those regarding the young stellar system @xmath0 pictoris , remain unchanged .
nonetheless , the present changes will be significant when compared with similar diagrams built taking into account the effects of the magnetic fields in the clouds .
these will be presented in a forthcoming manuscript ( leo , de gouveia dal pino and melioli 2007 , in prep . ) .
c.m . , e.m.g.d.p , and m.r.m.l .
acknowledge financial support from the brazilian agencies fapesp and cnpq .
99 dalgarno , a. , mccray , r. a. , 1972 , ar&aa , 10 , 375 leo , m. r. m. , de gouveia dal pino , e. m. , melioli , c. , 2007 , in prep .
melioli , c. , de gouveia dal pino , e.m .
, de la reza , r. & raga , a. , 2006 , mnras , 373 , 811 melioli , c. , de gouveia dal pino , e.m . , & raga , a. , 2005 , a&a , 443 , 495 | this is an erratum to the article entitled `` star formation triggered by sn explosions : an application to the stellar association of @xmath0 pictoris '' by c. melioli , e. m. de gouveia dal pino , r. de la reza and a. raga which was published in this journal ( mnras , 373 , 811 - 818 , 2006 ) with the following original abstract : in the present study , considering the physical conditions that are relevant interactions between supernova remnants ( snrs ) and dense molecular clouds for triggering star formation we have built a diagram of snr radius versus cloud density in which the constraints above delineate a shaded zone where star formation is allowed .
we have also performed fully 3-d radiatively cooling numerical simulations of the impact between snrs and clouds under different initial conditions in order to follow the initial steps of these interactions .
we determine the conditions that may lead either to cloud collapse and star formation or to complete cloud destruction and find that the numerical results are consistent with those of the snr - cloud density diagram .
finally , we have applied the results above to the @xmath1 pictoris stellar association which is composed of low mass post - t tauri stars with an age of 11 myr .
it has been recently suggested that its formation could have been triggered by the shock wave produced by a sn explosion localized at a distance of about 62 pc that may have occurred either in the lower centaurus crux ( lcc ) or in the upper centaurus lupus ( ucl ) which are both nearby older subgroups of that association ( ortega and co - workers).using the results of the analysis above we have shown that the suggested origin for the young association at the proposed distance is plausible only for a very restricted range of initial conditions for the parent molecular cloud , i.e. , a cloud with a radius of the order of 10 pc , a density of the order of 10@xmath220 @xmath3 , and a temperature of the order of 10@xmath2100 k .
[ firstpage ] stars : star formation ism : clouds - supernova remnants . |
in a review article of 1991 @xcite scully , englert and walther show that `` the information contained in a functioning measuring apparatus changes the outcome of the experiment '' . in the experiment under consideration
, a beam of two - level atoms incident upon a two - slit arrangement can show interference pattern .
before to reach the interference region however each atom , initially in the excited state , passes trough either one of two maser cavities which are both in the vacuum state , and makes a transition to the lower state , so that the emitted microwave photon leaves a which - way information in the cavity .
it turns out that the atom - cavity correlations are responsible for a loss of coherence which destroys the interference fringes . following a previous idea of scully and drhl @xcite (
also related to the delayed - choice experiments suggested by wheeler @xcite ) , they show how it is possible to retrieve the interference effects by removing ( erasing ) the _ welcher weg _ information .
+ the experimental arrangement of the optical stern - gerlach model @xcite can be used for a similar quantum optical test of complementarity .
in fact , the interaction hamiltonian of this model gives rise to entangled states in which the translational variables of the atomic center - of - mass are correlated with the cavity and the internal atomic variables . in this case as well as in the case analyzed in ref.@xcite , the presence or the loss of coherence in the atomic spatial distribution out the cavity ,
can be connected to the peculiar correlations between the measuring apparatus ( the cavity field ) and the system being observed .
diffraction and interference effects induced by field measurements have been analyzed by storey , collett and walls @xcite in the raman - nath regime , for virtual atomic transitions in the presence of very large detuning with respect to the rabi frequency . evoking an old question posed by popper @xcite , namely ,
if knowledge alone is sufficient to give uncertainty , they show that the localization of the atom trough the field measurement may be thought as a creation of a virtual slit ( or slits ) and their conclusion is that knowledge alone is sufficient to create uncertainty . according to this vision , here
we show that the knowledge of the field state erases the which - way information concerning two mutually exclusive quantum paths the system can follow . on the contrary ,
the mere possibility of acquiring this information which is stored in the atom - cavity correlations precludes any interference pattern . + a phase space analysis is carried out in sec .
iv in terms of the the wigner distribution for the atomic translational variables .
this analysis allows us to establish a well defined relation between the determinant of the covariance matrix of the wigner function and the distinguishability of the two quantum paths the .
the interaction of a two - level atom with a resonant mode of an optical cavity is usually described , in the rotating wave approximation ( rwa ) , by the well known jaynes - cummings hamiltonian @xmath0 , where @xmath1 and @xmath2 account for the internal atomic dynamics , while @xmath3 and @xmath4 are the annihilation and creation operators for the photons of the resonant mode of the cavity standing wave , and @xmath5 is the coupling constant . in this model
one assumes that the atom travels the cavity along a direction which is orthogonal to the cavity axis , with a velocity sufficiently large to treat classically this translational degree of freedom .
however , even if this condition is satisfied the translation degree of freedom along the cavity axis can not be overlooked .
in fact , this dynamics correlates with the dynamics describing the energy exchange between the two - level atom and the cavity field , giving rise , for example , to the optical stern - gerlach deflection , to the non dissipative damping of the rabi oscillations @xcite , and so on .
let us therefore consider the optical stern - gerlach model @xmath6 where the dynamics along the x - direction of the cavity axis is described by the position observable @xmath7 of the atomic center - of - mass and by its conjugate momentum @xmath8 .
we have assumed that the atom , of mass @xmath9 , enters the cavity near a nodal point of the cavity k - mode , with a spatial distribution narrow with respect to wavelength @xmath10 of the resonant mode ( @xmath11 ) .
we also assume that , at @xmath12 , the state of the global system is given by a factorized form , @xmath13 where the dressed states @xmath14 are eigenstates of the excitation number operator , @xmath15 .
they are also eigenstates of the interaction energy , @xmath16 with opposite eigenvalues , while @xmath17 describes the atomic translational dynamics .
the kets @xmath18 and @xmath19 denote the atom - cavity states in which the single excitation of the system pertains to the atom or to the cavity , respectively . by using the time evolution operator @xmath20 @xcite
one may write , for @xmath21 ( @xmath22 is the atomic flight time inside the cavity ) [ psit ] @xmath23\,,\;\;\label{subeq2}\end{aligned}\ ] ] where @xmath24\,e^{-i\,\omega\,t}{\left\vert\varphi(0)\right\rangle}.\ ] ] for a gaussian initial wave packet of minimum uncertainty , centered in @xmath25 , with zero mean velocity along the cavity axis and with a width @xmath26 , and getting rid of an irrelevant global phase factor we have @xmath27^{\frac{1}{2}}\exp(\mp \frac{i}{\hbar } m\,a\,x\,t)\cdot\nonumber \\ \cdot\exp\{-\frac{(x - x^{\pm}_{t})^{2}}{4\,\beta(t)}\}\end{aligned}\ ] ] where @xmath28 , @xmath29 , @xmath30 , while in the p - representation @xmath31 where @xmath32 , and @xmath33 is the fourier transform of @xmath34 .
+ the eq.s ( [ psit ] ) show that overlap and phase relation between the two components @xmath35 of the atomic translational state , may play a decisive role in the behavior of the internal dynamics . on the other hand
, they also show that interference effects in the atomic spatial distribution can appear in agreement with the record of the photon state in the cavity . + for @xmath36 , while @xmath37 , the spatial branches @xmath38 evolve according to the free hamiltonian , and one has @xmath39^{2}}{4\,\beta(t ) } \mp \frac{i}{\hbar}m\,a\,t\,x\right\}.\ ] ] as shown by the eq.s ( [ psit ] ) the state of the system splits up into a coherent superposition of two branches travelling in opposite directions and encoding correlations between the internal and the translational dynamics .
this suggests to associate the two orthogonal eigenstates @xmath40 and @xmath41 of the interaction energy to two mutually exclusive quantum paths , the two paths actually differing the one from the other for the opposite direction of the exchanged momentum between the atom and the cavity mirrors ( eq .
( [ phitrapprp ] ) ) .
+ to clarify this point , let us suppose to perform experiments measuring the atomic momentum out of the cavity . for @xmath42 sufficiently large the probability density of finding a particular momentum @xmath43 ( @xmath44 ) will be different from zero , as shown by eq.([phitrapprp ] ) , only inside a range @xmath45 around @xmath46 ( @xmath47 ) .
in fact for @xmath43 @xmath48 and for @xmath44 @xmath49 , while for @xmath50 the probability density is practically zero . in each case
( @xmath43 , @xmath44 ) the measured atomic momentum has been acquired by the atom from the mirrors of the cavity by means of a certain number of photon exchanges with the field . for a random sequence of positive and negative momentum exchanges ,
the output distribution of the atomic momentum would be different from zero in a range @xmath45 around @xmath51 .
the final evolution @xmath52 for the initial state @xmath18 is on the contrary compatible with a model in which , in a single experiment where , as above , the atomic momentum is measured , the atom entering the cavity has first @xmath53% probability of exchanging positive or negative momentum with the mirrors , but after the first choice , the system keeps to exchange momentum in the same direction . for either path the atoms exchange momentum with the mirrors , in such a way as to change their average momentum from @xmath54 to @xmath55 , respectively .
as in the double - slit experiment of young , we may observe interference fringes if there is no way to acquire information on the quantum paths . in our case
this information erasure can be accomplished , for example , by measuring the cavity photon state .
in fact , the states @xmath18 , and @xmath19 , are associated to the atom going through both the quantum paths , with definite relative phase , different for the two states .
+ on the contrary , if the above measurement is not made , from eq.([subeq1 ] ) , using the orthonormality of the dressed states , we get a spatial atomic distribution , both inside and out of the cavity , @xmath56 which does not exhibit interference terms since the correlations between the cavity and the atom cause a loss of coherence which destroys the interference .
one can also say that because of the entanglement , the atom imprints in the cavity information about the followed path , which is encoded in the interaction energy and that could also be read by finding out the direction of the exchanged momentum with the mirrors .
this information serves as a which - way identification and leads , as will be shown later , to a classical - like probability distribution in the phase space . in this case
, it is the mere possibility of recovering the information on the direction side of the exchanged momentum that makes distinction between the two paths .
+ let us now suppose that a measurement of the photon field in the cavity is made which records zero photons in the cavity , or one photon .
the atomic state out of the cavity , after the measurement , will be proportional , respectively , to [ psimesuret ] @xmath57 + if we repeat the measurement for many atoms entering the cavity in the same conditions , the interference patterns of the atoms correlated to the two different records will consequently be given respectively by [ pmesuredx ] @xmath58 where @xmath38 are given by the eq .
( [ phixtt ] ) .
+ eq.s ( [ psimesuret ] ) and ( [ pmesuredx ] ) show that the measurement of the photon field in the cavity plays the role of quantum eraser " which removes the which way information from the cavity and retrieves the interference terms .
it is worth to observe that for the two different records different interference effects will appear in form of fringes " and antifringes " patterns . + it is however to point out that the distinguishability of the two quantum paths depends on the momentum distribution as induced by the field interaction . in particular , the measurement of a positive or negative momentum exchange with the cavity could be possible only if the distance between the two main peaks is distinctly larger than the uncertainty @xmath59 , that is @xmath60 .
since in our case we assume @xmath61 , we have @xmath62 and the two paths can be distinguished if @xmath63 .
when this condition is not satisfied , the impossibility of distinguishing the two paths gives rise , out of the cavity , to spatial atomic distribution which exhibit diffraction rather than interference patterns ( see fig .
[ fig2 ] ) . + to analyze the spatial atomic distribution given by the eq.s ( [ pmesuredx ] ) , it is convenient to examine first the interference term @xmath64=\frac{2}{\sqrt{2\,\pi}\delta x_{t}}\exp\left\{-\frac{(x - x_{0})^{2}}{2\,\delta x_{t}^{2}}-\frac { \left[a\,t ( t - t/2 ) \right]^{2}}{2\,\delta x_{t}^{2 } } \right\}\cos\{2\,\varepsilon\,t\,k[x-(x - x_{0})\eta(t , t)]\}\ ] ] where @xmath65 the eq .
( [ 2rephiupphidown ] ) shows that for short flight time out of the cavity ( @xmath66 ) and for used values of the parameters , the interference term is of the same order of the main peaks @xmath67 of the spatial distribution , and fringe patterns can be observed . for increasing free flight time @xmath68 and @xmath69
, the main peaks travel far away the one from the other , while the interference term is damped to zero and the fringe patterns disappear .
in fact for @xmath68 , the exponential factor of eq .
( [ 2rephiupphidown ] ) gives rise to a damping for @xmath70^{-1 } > 1\ ] ] this implies that both the terms in the bracket be smaller than @xmath71 , that is , the main peaks must be well separated in both the position and momentum space .
the condition on the momentum space does not depend on @xmath72 and implies @xmath73 if this last condition is satisfied , the damping takes place for @xmath74 and consequently , for the parameters used in fig.s [ fig3 ] , @xmath75 .
the loss of the interference fringes is in these cases only due to the increasing distance between the two paths whose superposition is still coherent . as a consequence ,
as will be shown in the next section , the probability distribution in the phase space contains oscillating terms which show disagreement with a classical like behavior .
+ the spatial atomic distributions ( [ pmesured0x ] ) and ( [ pmesured1x ] ) are visualized in fig.s [ fig2]a , [ fig3]a , [ fig4]a and [ fig2]b , [ fig3]b , [ fig4]b , respectively , as a function of @xmath76 , for different values of @xmath42 .
fig.s [ fig2 ] show that for @xmath77 and @xmath78 , the probability of finding the atom in the excited state is enforced by the interference term , while the complementary probability of finding it in the ground state is strongly weakened .
the spatial distributions for @xmath68 do not change sensitively .
+ for @xmath79 and @xmath78 , the interference term gives rise to fringes ( fig.[fig3]a ) and antifringes ( fig.[fig3]b ) patterns which are evident up to @xmath80 , while for much longer times disappear .
+ for @xmath81 , the oscillations in @xmath82 of the interference term occur in closer succession and the resulting patterns are shown in fig.s [ fig4]a and [ fig4]b .
some properties of the atomic spatial distribution and of its correlation to the internal dynamics can be better understood by means of the wigner function @xmath83 @xcite .
+ we first analyze the wigner function of the reduced density operator @xmath84,\end{aligned}\ ] ] where @xmath52 is given by eq .
( [ subeq1 ] ) and @xmath85 indicates the trace on the internal atomic - cavity states .
the orthogonality of the dressed states leads to a reduced operator which appear as the incoherent sum of two terms each one related to a particular component of the translational state . using the characteristic function associated to @xmath86 , @xmath87\right\}\ ] ] and @xmath38 as given by eq .
( [ phixtt ] ) , we derive the wigner function @xmath88 as fourier transform of @xmath89 and we obtain ( for @xmath90 ) @xmath91 where @xmath92^{2}}{\delta x_{t}^{2}}+\frac{(p\pm m a t)^{2}}{\delta p_{0}^{2}}\right]}{2(1-\rho_{xp}^{2})}\right\}\cdot\nonumber \\
\cdot\exp\left\{\frac{\rho_{xp}\frac{(p\pm m\,a\,t)}{\delta p_{0}}\frac{[x\pm a\,t(t - t/2)]}{\delta x_{t}}}{2(1-\rho_{xp}^{2})}\right\},\;\;\nonumber\end{aligned}\ ] ] @xmath93 is the free particle correlation coefficient of the atomic translational variables @xmath94 + and we have put @xmath95 . + as in eq.([px ] ) , any phase relation between @xmath96 and @xmath97 has been lost after the trace operation on the internal dynamics , and as a consequence the wigner function assumes only positive values and can figure a classical - like joined probability distribution in the phase space .
+ the classicality of the probability distribution has been here recovered by releasing information on the internal dynamics and could be considered as a natural ending up of the distribution if a decay time of the photon in the cavity would be taken into account , after which , the information on the cavity state is irreversibly lost .
+ the positive definition of the wigner function allows to comment on the generalized uncertainty area @xmath98 ( the error box @xcite ) of the whole spatial distribution in the phase space and to relate it to the distance between the gaussian distributions which describe the different paths . the square uncertainty @xmath99 is evaluated as the determinant of the distribution covariance matrix @xmath100 and for @xmath101 it is given by @xmath102 where @xmath103 is the adimensional distance in the phase space between the average positions of the two gaussian centered in @xmath104 and @xmath105 , measured in units of @xmath26 and @xmath59 along @xmath82 and @xmath106 , respectively .
it is to notice that @xmath107 and then @xmath108 .
in fact , for @xmath90 the atom evolves freely and the uncertainty area ( or the determinant of the covariance matrix ) is constant in time as it happens for particle subjected to a potential at most quadratic .
since @xmath109 is the square of the initial uncertainty area , the enlargement of the square area in time is only due to the distance @xmath110 .
+ the distance @xmath110 in the phase space plays a fundamental role in determining the distinguishability of the two quantum paths related to the two translational branches @xmath111 and @xmath112 .
in fact , the following relation holds @xmath113 which shows how the two atomic translational paths become mutually exclusive for increasing values of @xmath114 .
+ we want now investigate the effect of a measurement of the photon field in the cavity on the wigner function .
the translational system state after the measurement will be proportional to one of the states ( [ psimesuret ] ) : [ psiaftermes ] @xmath115\label{psiam0}\\ { \left\langlex\vert\psi^{1}(x , t)\right\rangle}=\frac{1}{\sqrt{n_{1}(t)}}\left[\phi^{+}(x , t)-\phi^{-}(x , t)\right],\label{psiam1}\end{aligned}\ ] ] where [ normalization ] @xmath116,\label{n0}\\
n_{1}(t)=2\left[1-e^{-d(t)^{2}/8}\cos(2 \varepsilon k x_{0 } t ) \right],\label{n1}\end{aligned}\ ] ] ( for seek of simplicity , we will assume again that @xmath95 ) . the eq.s ( [ psimesuret ] ) show that after the measurement of the photon field , the spatial atomic state is given by a linear superposition of two gaussians , which , as well known @xcite , gives rise to negative contributions to the quasi probability distribution . by using in this case a more commune definition of the wigner function , @xmath117 one
easily gets for the two states ( [ psiaftermes ] ) @xmath118,\:\,\label{w0am}\\ w_{1}(x , p)=\frac{2}{n_{1}(t)}\left[w^{+}(x , p)+w^{-}(x , p)-w^{q}(x , p)\right],\:\,\label{w1am}\end{aligned}\ ] ] where @xmath119\,2\,\cos\left\{2\,\varepsilon\,k\,t\left[x+\frac{p}{m } \left(t-\frac{t}{2}\right)\right]\right\},\ ] ] gives negative contributions and testifies a precise phase relation between the two components @xmath120 , and the consequent possibility of interference fringes in the spatial atomic distribution .
@xmath121 is of the same order of magnitude of the two main peaks which travel in opposite direction in the phase space , even when they are absolutely disconnected , that is even in a `` macroscopical limit '' in which the scalar product given by the eq . ( [ scalar ] ) is practically zero ( @xmath122 ) . + supposing @xmath54 photons in the cavity or @xmath71 photon as the output of the measure , the following expressions for the square of the generalized uncertainty area is obtained : @xmath123\,e^{-d(t)^{2}/8}+\left(1-\frac{d(t)^{2}}{4}\right)\,e^{-d(t)^{2}/4 } \right\}\ ] ] the error box of the `` classical like '' distribution given in the eq .
( [ deltad ] ) is smaller or larger than @xmath124 or @xmath125 , respectively .
however , for @xmath126 , both @xmath127 and @xmath128 tend to the same value of the uncertainty area related to the `` classical like '' distribution of eq .
( [ wignerr ] ) .
it has been shown that in the optical stern - gerlach effect , if the information on the internal dynamics is released , the atom leaves the cavity in a state which results an incoherent superposition of two spatial paths related to two virtual slits . however , the measurement of photon presence in the cavity erases the which path information and spatial atomic interference may occur from the virtual slits , which exhibit fringe and antifringe patterns in accord to the lack or the presence of the photon . the wigner quasi - probability atomic distribution in the phase space , exhibits in the first case a `` classical like '' behavior , characterized by a superposition of two incoherent distributions related to the two paths .
the wigner function after the quantum erasure shows on the contrary the positive - negative oscillations typical of a coherent superposition of two quantum paths .
quite intriguing is however the fact that the error box of this second wigner distribution tends to the `` classical like '' one when the adimensional distance in the phase space between the two paths is much larger than @xmath71 ( one could say in a sort of macroscopic limit ) .
in fact , when the oscillation lengths in the momentum space ( or in the co - ordinate space ) of the wigner distribution are much smaller than the initial uncertainties of the atomic packet , the oscillating part does not contribute to the evaluation of the second moments of both @xmath82 and @xmath106 .
scully , b .-
englert , and h. walther , _ nature _ * 351 * , 111 ( 1991 ) m.o .
scully and k. drhl , _ phys . rev . a _ * 25 * , 2208 ( 1982 ) j.a .
wheeler , _ in mathematical foundations of quantum theory ( ed .
marlow ) 9 - 48 , academic , n.y . _
( 1978 ) t. sleator , t. pfau , v. balykin , o. carnal , and j. mlynek , _ phys .
lett . _ * 68 * , 1996 ( 1992 ) ; c. tanguy , s. reynaud , and c. cohen - tannoudji , _ j. phys .
b _ * 17 * , 4623 ( 1984 ) ; m. freyberger and a. m. herkommer , _ phys . rev
. lett . _ * 72 * , 1952 ( 1994 ) ; a. vaglica , _ phys .
a _ * 54 * , 3195 ( 1996 ) | in the optical stern - gerlach effect the two branches in which the incoming atomic packet splits up can display interference pattern outside the cavity when a field measurement is made which erases the which - way information on the quantum paths the system can follow . on the contrary
, the mere possibility to acquire this information causes a decoherence effect which cancels out the interference pattern .
a phase space analysis is also carried out to investigate on the negativity of the wigner function and on the connection between its covariance matrix and the distinguishability of the quantum paths . |
reconciling the cold dark matter ( cdm ) predictions of supersymmetric models with the stringent constraints from the wilkinson microwave anisotropy probe ( wmap ) has been one of the challenges within the particle physics community in recent years .
the amount of cdm deduced from wmap data @xcite puts severe constraints on possible dark matter candidates , inluding the lightest supersymmetric particle ( lsp ) .
additional constraints on the model parameters are obtained by imposing yukawa unification , and by taking into account the bounds from flavour changing neutral currents ( fcnc ) .
in addition to the above , the neutrino data of the past years provided evidence for the existence of neutrino oscillations and masses , pointing for the first time to physics beyond the standard model . as expected ,
the additional interactions required to generate neutrino masses also affect the energy dependence of the couplings of the mssm , and thus modify the yukawa unification predictions .
a first observation had been that the additional interactions of neutrinos , which affect the tau mass , may spoil bottom - tau unification for small @xmath0 @xcite .
subsequently , however , it has been realised that large lepton mixing naturally restores unification , and even allows unification for intermediate values of @xmath0 that were previously disfavoured @xcite .
this is done by making the simple observation that the @xmath4 equality at the gut scale refers to the @xmath5 entries of the charged lepton and down quark mass matrices , while the detailed structure of the mass matrices is not predicted by the grand unified group itself .
it is then possible to assume mass textures , such that , after diagonalisation at the gut scale , the @xmath6 and @xmath7 entries are no - longer equal . in the current work , based on @xcite ,
we revisit the issues of dark matter and yukawa unification taking into account the effects of massive neutrinos and large lepton mixing in see - saw models , and extending previous results to large @xmath0 .
we find that the effects on the allowed parameter space are significant and , in fact , it turns out that yukawa unification in the presence of neutrinos is also compatible with a negative @xmath1 , unlike what happens if the effects of neutrinos are ignored . passing to the relic density of neutralinos
, we study the consequences of large lepton mixing in the @xmath8 coannihilation region and in resonances in the @xmath9 annihilation channels , finding sizeable effects , particularly in the latter case .
it is interesting to note that for the cosmologically favoured area , it is also possible to observe tau flavour violation at the lhc , in the framework of non - minimal supersymmetric grand unification @xcite .
in the presence of massive neutrinos , the predictions for @xmath10 and unification clearly get modified .
radiative corrections from the neutrino yukawa couplings have to be included in renormalisation group runs from @xmath11 to @xmath12 ( scale of the heavy right - handed neutrinos ) .
below @xmath13 , right - handed neutrinos decouple from the spectrum and an effective see - saw mechanism is operative ; the relevant equations are given in @xcite .
in addition , if the gut scale lies significantly below a scale @xmath14 , at which gravitational effects can no longer be neglected , the renormalization of couplings at scales between @xmath14 and @xmath11 may induce additional effects to the running and the simplest example is provided by minimal su(5 ) @xcite ( however , modifications to soft masses are in this simplest case proportional to the @xmath15 mixing @xcite , and thus are significantly suppressed ) .
nevertheless , it has been realized that the influence of the runs above the gut scale on the dark matter abundance can be very sizeable @xcite , due to changes in the relation between @xmath16 and @xmath17 , which is crucial in the coannihilation area .
this we discuss in a subsequent section . in supersymmetric models
, unification is very sensitive to the model parameters @xcite , particularly the higgs mixing parameter , @xmath1 . to correctly obtain pole masses within this framework , the standard model and supersymmetric threshold corrections have to be included ;
for the bottom quark , these corrections result to a @xmath18 that can be very large , particularly for large values of @xmath0 @xcite . constraints from br(@xmath19 ) are also included in the analysis . before passing to the results ,
however , let us summarise a few facts on the possible range of the mass of the bottom quark : the 2-@xmath20 range for the @xmath21 bottom running mass , @xmath22 , is from 4.1 - 4.4 gev .
moreover , @xmath23 , and the central value of @xmath24 corresponds to @xmath25 from 2.82 to 3.06 gev . in fig .
[ mb1 ] , we summarize the predictions for @xmath10 in msugra and in the presence of massive neutrinos . in order to discuss the dependence of @xmath26 on @xmath0
, we consider the set of soft parameters : @xmath27 gev , @xmath28 , @xmath29 gev .
the figure exhibits the well - known fact that in the absence of phases or large trilinear terms , @xmath18 is positive for @xmath1 positive , and therefore the theoretical prediction for the @xmath30 quark pole mass is too high to be reconciled with @xmath4 unification . on the other hand , for @xmath31 , @xmath18 is negative and the theoretical prediction for the b quark mass can lie within the experimental range for values of @xmath0 between roughly 25 and 45 ; clearly , for a large @xmath0 it is mandatory to take into account the large supersymmetric corrections to @xmath10 @xcite . versus @xmath32 assuming @xmath33 at the high scale in the absence of lepton mixing .
we work with the following set of parameters : @xmath34 gev , @xmath28 gev , @xmath35 gev .
the experimental range of @xmath10 ( horizontal lines ) is also shown , the black thick ( blue thin ) lines have @xmath24 = 0.1212 ( 0.1132 ) , the dot - dash lines are obtained within the mssm , the solid lines include neutrino dirac yukawa up to the scale @xmath36 .
the upper ( lower ) set of lines corresponds to @xmath37 .
_ , width=302,height=264 ] [ mb1 ] the analysis in the presence of massive neutrinos takes into account only the third generation couplings , from the @xmath11 to the scale of the right handed neutrino masses , and evolve the light neutrino mass operator from this scale down to @xmath38 . a large value of the dirac - type neutrino yukawa coupling , @xmath39 at the gut scale may arise naturally within the framework of grand unification , and its value is determined by demanding a third generation low energy neutrino mass of @xmath40 ev .
the predictions for @xmath26 using the lower and upper bounds of the 2-@xmath20 experimental range of @xmath24 and the correponding range for @xmath26 after the evolution of the bounds on @xmath22 are shown for a scale @xmath41 gev .
we observe that for @xmath42 the prediction for @xmath26 is always very large , despite its dependence on the soft terms through @xmath18 .
for the values of the soft terms considered in fig.[mb1 ] , the allowed range of @xmath0 shrinks from @xmath43 to @xmath44 when we introduce the effect of see - saw neutrinos .
it is also shown that the influence of runs above @xmath11 is too small to have any significant effect .
the results are significantly modified once we consider the effects of lepton mixing in the diagonalisation and running of couplings from high to low energies . in order to show this , we focus on @xmath4 unification within the framework of su(5 ) gauge unification and flavour symmetries that provide consistent patterns for mass and mixing hierarchies , and naturally reconcile a small @xmath15 mixing with a large charged lepton one . taking into account the particle content of su(5 ) representations ( with symmetric up - type mass matrices , and down - type mass matrices that are transpose to the ones for charged leptons ) , one finds the approximate relations @xmath45 which , after diagonalization , lead to @xmath46 where @xmath47 parametrises the flavour mixing in the ( 2,3 ) sector . in the left pannel of fig.[deltb ]
we show the change of @xmath10 as a function of @xmath0 , when the effects from large lepton mixing are appropriately considered .
comparing with the previous plots , we see how solutions with positive @xmath1 are now viable , for the whole range of @xmath0 .
the appropriate size of the parameter @xmath47 in each case can be determined by imposing the relation @xmath48 at @xmath49 and investigating the values that are required in order to obtain a correct prediction for @xmath26 .
this is shown in the right pannel of fig.[deltb ] , where we demand a value of @xmath26 at the center of its experimental range , for the central value of @xmath24 . as a function of @xmath0 , when including lepton mixing effects .
in the right panel , we show the required values @xmath47 for consistent unification .
we use the set of soft terms of fig.[mb1 ] , @xmath50 and impose @xmath51 gev .
the upper ( lower ) line corresponds to @xmath42 ( @xmath31 ) .
_ , title="fig:",width=302,height=264 ] as a function of @xmath0 , when including lepton mixing effects . in the right panel , we show the required values @xmath47 for consistent unification .
we use the set of soft terms of fig.[mb1 ] , @xmath50 and impose @xmath51 gev .
the upper ( lower ) line corresponds to @xmath42 ( @xmath31 ) .
_ , title="fig:",width=302,height=264 ]
in msugra ( or the cmssm ) for choices of soft terms below the tev scale , the lsp is bino like and the prediction for @xmath52 is typically too large for models that satisfy the experimental constraints on susy .
in fact , the values of wmap can essentially be obtained in two regions : * @xmath53 coannihilation region that occurs for @xmath54 .
* resonances in the @xmath9 annihilation channel , which occur for @xmath55 .
since the above areas are `` fine - tuned '' , they will inevitably be sensitive to the changes induced by gut unification and sizeable mixing in the charged lepton sector .
the runs corresponding to @xmath56 have a big impact on the neutralino relic density .
the large values of the gauge unified coupling @xmath57 tend to increase the values of @xmath16 , even if we start with small @xmath58 at @xmath14 . , @xmath42 , @xmath59 , @xmath60 gev and @xmath61 for the same set of parameters as in fig.[mb1 ] , without ( left ) and with ( right ) the @xmath62 running .
the solid ( dash ) line corresponds to @xmath63 gev ( br(@xmath19)=@xmath64 ) . in the lower ( red ) area
the lsp is a stau .
_ , title="fig:",width=283,height=264 ] , @xmath42 , @xmath59 , @xmath60 gev and @xmath61 for the same set of parameters as in fig.[mb1 ] , without ( left ) and with ( right ) the @xmath62 running .
the solid ( dash ) line corresponds to @xmath63 gev ( br(@xmath19)=@xmath64 ) . in the lower ( red ) area
the lsp is a stau .
_ , title="fig:",width=283,height=264 ] but for @xmath65 @xcite . here , the parameter @xmath66.__,title="fig:",width=283,height=264 ] but for @xmath65 @xcite
. here , the parameter @xmath66.__,title="fig:",width=283,height=264 ] we see that the consideration of mixing effects , in combination with the inclusion of effects from the runs above @xmath11 , significantly enhances the allowed parameter space ( green area ) , an effect that is more visible for large @xmath0 .
areas with different colours are excluded @xcite .
the reduction of the allowed parameter space for smaller values of @xmath0 , is already evident for @xmath67 .
the @xmath68 picture has been discussed in @xcite , where it was shown that the wmap allowed region can be compatible with observable flavour violation at the lhc .
the case with @xmath31 and a more detailed discussion of lepton mixing effects are presented in @xcite .
we revisited the wmap dark matter constraints on yukawa unification in the presence of massive neutrinos .
large neutrino mixing , as indicated by the data modifies the predictions for the bottom quark mass , and enables yukawa also for large @xmath0 and for positive @xmath1 that were previously disfavoured .
a direct outcome is that the allowed parameter space for neutralino dark matter also increases , particularly for areas with resonant enhancement of the neutralino relic density . for completeness
, we also note that for the cosmologically favoured parameter region , we found lepton flavour violating rates very close to the current experimental bounds @xcite
. finally , interesting effects may arise in the case of non - universal soft terms .
these are also discussed in detail in @xcite .
cm + * acknowledgements * the research of s. lola and p. naranjo is funded by the fp6 marie curie excellence grant mext - ct-2004 - 014297 . the work of m.e.g , c.p and j.r.q is supported by the spanish mec projet fpa2006 - 13825 and the projet p07fqm02962 funded by `` junta de andalucia '' .
f. vissani and a. y. smirnov , phys .
b * 341 * , 173 ( 1994 ) , a. brignole , h. murayama and r. rattazzi , phys .
b * 335 * , 345 ( 1994 ) g. k. leontaris , s. lola and g. g. ross , nucl .
b * 454 * ( 1995 ) 25 .
b. grzadkowski and m. lindner , phys . lett .
b * 193 * ( 1987 ) 71 ; yu . f. pirogov and o. v. zenin , eur .
j. c * 10 * ( 1999 ) 629 ; n. haba , n. okamura and m. sugiura , prog .
* 103 * ( 2000 ) 367 , eur .
j. c * 10 * ( 1999 ) 677 .
j. hisano and d. nomura , phys .
d * 59 * , 116005 ( 1999 ) .
l. calibbi , y. mambrini and s. k. vempati , jhep * 0709 * , 081 ( 2007 ) . m. e. gomez , g. lazarides and c. pallis , nucl
. phys .
b * 638 * , 165 ( 2002 ) , phys .
d * 67 * , 097701 ( 2003 ) .
l. j. hall , r. rattazzi and u. sarid , phys .
d * 50 * , 7048 ( 1994 ) m. s. carena , m. olechowski , s. pokorski and c. e. m. wagner , nucl .
b * 426 * , 269 ( 1994 ) m. e. gomez , s. lola , p. naranjo and j. rodriguez - quintero , in preparation . | we revisit the wmap dark matter constraints on yukawa unification in the presence of massive neutrinos .
the large neutrino mixing indicated by the data modifies the predictions for the bottom quark mass , and enables yukawa also for large @xmath0 , and for positive @xmath1 that were previously disfavoured . as a result
, the allowed parameter space for neutralino dark matter also increases , particularly for areas with resonant enhancement of the neutralino relic density .
@xmath2 departamento de fsica aplicada , university of huelva , 21071 huelva , spain + @xmath3 department of physics , university of patras , 26500 patras , greece _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ proceedings of 4th international workshop on the dark side of the universe ( dsu 2008 ) , cairo , june 2008
. published by american institute of physics . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
buoyancy or density gradients drive flows in the atmosphere and interiors of planets and stars , as well as in electronic devices and industrial appliances like heat exchangers , boilers , etc .
accordingly , scientists ( including geo- , astro- , atmospheric- and solar physicists ) and engineers have been studying buoyancy - driven flows for more than a century . an important unsolved problem in this field is how to quantify the spectra and fluxes of kinetic energy and entropy ( @xmath8 and @xmath9 respectively , where @xmath10 and @xmath11 are the velocity and temperature fluctuations ) of buoyancy - driven flows @xcite . in this paper , we will study these quantities and respective nonlinear fluxes using direct numerical simulations , and show that the kinetic energy ( ke ) spectrum differs from kolmogorov s theory when buoyancy is strong . flows driven by buoyancy can be classified in two categories : ( a ) convective flows in which hotter and lighter fluid at the bottom rises , while colder and heavier fluid at the top comes down ; ( b ) stably stratified flows in which lighter fluid rests above heavier fluid .
the convective flows are unstable ; but the stably stratified flows are stable , hence their fluctuations vanish over time . therefore , a steady state of a stably stratified flow is achieved only when it is driven by an external force .
even though both types of flows are driven by density gradients , the properties of such flows are quite different , which we decipher using quantitative analysis of energy flux and energy supply rate by buoyancy . for a stably stratified flow ,
bolgiano @xcite and obukhov @xcite first proposed a phenomenology , according to which the ke flux @xmath0 of a stably stratified flow is depleted at different length scales due to a conversion of ke to `` potential energy '' via buoyancy ( @xmath12 ) . as a result
, @xmath5 decreases with wavenumber , and the energy spectrum is steeper than that prediced by kolmogorov theory @xmath13 , where @xmath6 is the wavenumber ) ; we refer to the above as _ bo _ phenomenology or scaling . according to this phenomenology ,
for @xmath14 , where @xmath15 is the bolgiano wavenumber @xcite , the ke spectrum @xmath4 , entropy spectrum @xmath16 , @xmath0 , and entropy flux @xmath17 are : @xmath18 where @xmath19 , @xmath20 , and @xmath21 are the thermal expansion coefficient , acceleration due to gravity , and the entropy dissipation rate respectively , and @xmath22 s are constants . for the wavenumbers in the range @xmath23 , @xmath24 , and @xmath25 , where @xmath26 is the ke dissipation rate , and @xmath27 is the wavenumber after which dissipation starts .
we remark that many researchers describe the stably stratified flows in terms of density fluctuation @xmath28 , which leads to an equivalent description since @xmath29 .
several research groups studied the properties of stably stratified flows using numerical simulations .
kimura and herring @xcite observed _ bo _ scaling in a narrow band of wavenumbers in their @xmath30 decaying buoyancy - dominated simulation . in 2012 , using @xmath31 simulations , kimura and herring @xcite showed that waves and vortex exhibit @xmath32 energy spectra at large wavenumbers , but for sufficiently strong stratification , the corresponding spectra are @xmath33 and @xmath34 , respectively , at small wavenumbers .
the terrestrial atmosphere exhibits @xmath35 energy spectrum for @xmath36 , and @xmath32 spectrum for @xmath37 .
lindborg @xcite and brethouwer et al .
@xcite attempted to explain this observation by studying quasi two - dimensional stratified flow ( horizontal distance @xmath38 vertical distance ) .
they performed a series of periodic box simulations and showed that the horizontal kinetic and potential energy spectra follow @xmath39 scaling , while the kinetic energy spectrum of the vertical velocity , and the potential energy spectrum follow @xmath40 scaling .
vallgren et al .
@xcite , and bartello and tobias @xcite observed similar scaling in their numerical simulations .
it is important to note that all these work are under the regime of strong stratification . using theoretical arguments , procaccia and zeitak @xcite
, lvov @xcite , lvov and falkovich @xcite , and rubinstein @xcite proposed that the _ bo _ scaling would also be applicable to rayleigh - bnard convection ( rbc ) . the numerical and experimental results of rbc , however , have been largely inconclusive . based on simulations with periodic boundary conditions , borue and orszag @xcite and kandera et al
@xcite reported kolmogorov - obukhov ( referred to as _ ko _ ) scaling , in which @xmath41 , and @xmath24 .
mishra and verma @xcite reported the _
ko _ scaling for zero- and low prandtl number flows . using numerical simulations ,
verzicco and camussi @xcite however reported the _ bo _ scaling for the frequency spectrum , which was computed using the data collected by real space probes .
calzavarini et al .
@xcite reported the _ bo _ scaling in the boundary layer , and the _ ko _ scaling in the bulk .
the experimental results @xcite are more divergent with some reporting the _ bo _ scaling , and some others reporting the _ ko _ scaling . in this paper
we simulate the stably stratified and rbc turbulence , and analyse the spectra and fluxes of the ke as well as the entropy .
we show that for the stratified flow , the ke flux and spectrum follow the _ bo _ scaling ( eqs .
( [ eq : eu]-[eq : pi ] ) ) when buoyancy is strong , but they follow the _ ko _ scaling for weak buoyancy .
the ke flux in rbc however increases at small wavenumbers , but remains flat for a narrow wavenumber band in the intermediate regime where the energy spectrum follows the _ ko _ scaling .
the outline of the paper is as follows . in sec .
[ sec : gov_eqns ] , we describe the parameters and equations used , as well as our assumptions . in sec .
[ sec : method ] , we discuss the numerical method of our simulations . results of our numerical simulations are discussed in sec . [ sec : results ] .
we present our conclusions in sec .
[ sec : conclusion ] .
the dynamical equations that describe the buoyancy - driven flows under the boussinesq approximation are @xmath42 where @xmath43 is the velocity field , @xmath11 and @xmath44 are the temperature and pressure fluctuations , respectively , with reference to the conduction state , @xmath45 is the buoyancy direction , @xmath46 is the external force field , @xmath47 is the temperature difference between two layers kept apart by a vertical distance @xmath48 , and @xmath49 , @xmath50 , and @xmath51 are fluid s mean density , kinematic viscosity , and thermal diffusivity respectively . for rbc , temperature of the top plate is lower than the bottom one , hence @xmath52 , but for the stably stratified flows , the gradient is opposite , i.e. @xmath53 .
it is easy to verify that eqs .
( [ eq : u_dim],[eq : th_dim ] ) conserve the volume integral @xmath54 in limit when @xmath55 and @xmath56 . for rbc
, the temperature gradient provides energy to the system , and a steady state is reached after some time ( approximately after a thermal diffusive time ) ; for such flows we can take @xmath57 .
however , stably stratified flows are stable , and the fluctuation die out if @xmath57 .
therefore , for obtaining a steady state in a stably stratified flow , we force the flow at small wavenumbers with random forcing prescribed by kimura and herring @xcite . in this paper , we contrast the scaling relations of stably stratified flow and rbc in a single formalism . for the same , we use temperature fluctuations
@xmath11 as a variable . however , this scheme is equivalent to usage of @xmath28 , the density fluctuations from the linear density profile @xmath58 ; the variable @xmath28 is often used for stably stratified flows .
we can rewrite eqs .
( [ eq : u_dim]-[eq : inc_dim ] ) in terms of @xmath28 using the following relations : @xmath59 thus , the two sets of equations are equivalent .
it is convenient to work with nondimensionalized equations , which is achieved by using @xmath48 as the length scale , @xmath60 as the velocity scale , and @xmath47 as the temperature scale .
therefore , @xmath61 , @xmath62 , @xmath63 , and @xmath64 , where primed variables are nondimensionalized .
when we use the density gradient @xmath65 , the velocity scale is @xmath66 , and the time scale is @xmath67 . in terms of the nondimensionalized variables ,
the equations are @xmath68 where the prandtl number is defined as @xmath69 the rayleigh number is defined as @xmath70 where @xmath71 is the usual definition taken from rbc , but @xmath72 , a modified form of @xmath71 , is in terms of density gradient and brunt visl frequency , which is defined as @xmath73 physically , brunt visl frequency is the frequency of the gravity waves in a stably stratified flow .
it is important to note that larger @xmath74 or @xmath75 implies stronger _ stability _ for a stably stratified flow , but larger @xmath71 implies stronger _ instability _ for rbc .
also , it has been shown that the available potential energy ( ape ) " , @xmath76 , matches with @xmath77 where @xmath78 @xcite .
the other important nondimensional numbers are as follows .
the reynolds number is defined as @xmath79 where @xmath80 is the rms velocity of the flow , computed as the volume average of the magnitude of the velocity field , and @xmath81 is the corresponding quantity in dimensionless form .
the richardson number , which is a ratio of the buoyancy and the nonlinear term @xmath82 , is defined as @xmath83 the froude number @xmath84 , which is the ratio of the characteristic fluid velocity and gravitational wave velocity , is defined as @xmath85 thus , the froude number is the rms velocity of the fluid in the dimensionless form .
note that the froude number is meaningful only for stably stratified flows .
also , small @xmath84 implies strongly stratified flow , while strong @xmath86 indicates strong buoyancy
. note that in later discussion we will focus our discussions on eqs .
( [ eq : u_ndim]-[eq : inc_ndim ] ) . for convenience
, we drop the primes from the variables in the subsequent discussions . in some of the earlier studies on
_ strongly _ stratified flows , e.g. lindborg @xcite , brethouwer et al .
@xcite , bartello and tobias @xcite , the equations have been written for horizontal and vertical components of the velocity field in terms of the froude number and reynolds number ( see appendix a ) . however , we use eqs .
( [ eq : u_ndim]-[eq : inc_ndim ] ) for our analysis since they help us contrast stably stratified flows and rbc in a single formalism . in the following discussion we contrast our assumptions and equations with those used for _ strongly _ stratified flows ( see appendix [ sec : appendix ] ) : a. a large number of earlier work , e.g. lindborg @xcite , brethouwer et al .
@xcite , bartello and tobias @xcite focus on strongly stratified flows .
a signature of such flows is that their froude number is much less than unity .
our focus is on moderately stratified flows , which is achieved by setting the froude number to unity or higher , or @xmath87 ( see eq .
( [ eq : fr ] ) ) .
however , @xmath88 for such flows . in sec .
[ subsec : stratified ] we will show that for @xmath89 , a buoyancy dominated flow , we obtain the _ bo _ scaling .
however for @xmath90 , a weakly buoyant flow , we obtain the _ ko _ scaling since the nonlinearity is weak for this case .
b. the strongly stratified flows ( @xmath91 ) are quasi two - dimensional and strongly anisotropic , hence they employ @xmath92 ( here @xmath93 are the lengths of the box along @xmath94 directions respectively ) @xcite .
these flows are expected to model the atmosphere of the earth .
our flows , however , are three - dimensional and weakly - anisotropic since @xmath95 .
therefore , we simulate the flows in geometries where @xmath96 .
the latter configurations are suitable for testing bolgiano - obukhov scaling , which is formulated as an isotropic spectrum . c. for the non - dimensionalized eqs .
( [ eq : u_ndim]-[eq : inc_ndim ] ) , the brunt - visl frequency @xmath75 is unity , implying that the time scale of the gravity waves is of the same order as the eddy turnover time of the large eddies .
d. our flows are turbulent , i.e. , @xmath97 .
e. a large number of stably stratified flow simulations ( e.g. , lindborg @xcite , brethouwer et al .
@xcite , vallgren et al .
@xcite , kimura and herring @xcite , and bartello and tobias @xcite ) employ periodic boundary condition ; this is to simulate the bulk flow away from the boundaries .
also , the bolgiano and obukhov scaling , as well as kolmogorov phenomenology , are strictly applicable for homogeneous and isotropic turbulence , for which a periodic box is a good geometrical configuration . keeping these aspects in mind , we employ the periodic boundary condition for simulating stably stratified flows .
+ boundary walls and thermal plates play an important role in the flow dynamics of rbc . in our present study , at the top and bottom plates , we employ the free - slip boundary condition for the velocity field , and the conducting boundary condition for the temperature field .
we apply the periodic boundary condition at the side walls .
we simulate the stably stratified flow and rbc by solving eqs .
( [ eq : u_ndim]-[eq : inc_ndim ] ) numerically for the aforementioned boundary conditions . after
that we study kinetic energy spectrum and flux , as well as other diagnostics tools like energy supply rate by buoyancy ; we will discuss these tools in the next section . in fourier space , the equation for the kinetic energy
is derived using eq .
( [ eq : u_ndim ] ) as @xcite @xmath98 where @xmath4 is the kinetic energy of the wavenumber shell of radius @xmath6 , @xmath99 is energy transfer rate to the shell @xmath6 due to nonlinear interactions , and @xmath100 is total energy supply rate to the shell from the forcing functions , both buoyancy and external forcing @xmath46 : @xmath101 where the first term is due to buoyancy , while the second term is due to the external random forcing . the term @xmath102 of eq .
( [ eq : dek_dt ] ) is the viscous dissipation rate , and is given by @xmath103 which is always positive .
the nonlinear interaction term @xmath99 is related to the kinetic energy flux @xmath5 as @xmath104 which is computed using the following formula @xcite @xmath105 [ { \bf u^*(k ) \cdot u(p ) } ] ) \label{eq : pi_verma}\ ] ] the energy flux @xmath106 is interpreted as the kinetic energy leaving a wavenumber sphere of radius @xmath107 . using eqs .
( [ eq : dek_dt],[eq : pi_def ] ) , we deduce that @xmath108 under a steady state ( @xmath109 ) , we obtain @xmath110 or @xmath111 equation ( [ eq : pi_k_fk_dk ] ) is obvious , but it provides us important clues on the energy spectrum and flux of the buoyancy - driven flows . here
we list three possibilities for the inertial range ( @xmath112 ) , where @xmath113 is the forcing wavenumber , and @xmath27 is the dissipation wavenumber :
1 . for the inertial range of fluid turbulence , @xmath114 and @xmath115 , hence @xmath116 and @xmath7 , which is the prediction of kolmogorov s theory .
[ enu_1 ] 2 . for the stably stratified flows ( @xmath53 in eq .
( [ eq : th_ndim ] ) ) , as argued by bolgiano and obukhov , the buoyancy converts kinetic energy of the flow to potential energy , i.e. , @xmath117 for @xmath118 . therefore , eq .
( [ eq : pi_k_fk_dk ] ) predicts that @xmath5 will decrease with @xmath6 in this wavenumber range , as shown in fig .
[ fig : sch_flux](a ) . in the wavenumber range ,
@xmath23 , buoyancy becomes weaker , hence @xmath119 , and kolmogorov s spectrum is expected . in the present paper , using numerical simulation , we demonstrate _ bo _ scaling in the @xmath118 regime ; the demonstration of @xmath7 for @xmath23 requires larger resolution than that used in this paper .
[ enu_2 ] 3 . for rbc ( @xmath120 in eq .
( [ eq : th_ndim ] ) ) , buoyancy feeds energy to the kinetic energy , hence @xmath121 .
therefore , the sign of @xmath122 depends crucially on @xmath102 .
first , for @xmath123 , @xmath124 since @xmath125 , then for the intermediate wavenumbers @xmath126 where @xmath127 , we expect @xmath128 . finally in the dissipative range ( @xmath129 ) , @xmath130 since @xmath131 .
here @xmath132 is the transition wavenumber shown in fig .
[ fig : sch_flux](b ) . consequently , as shown in fig .
[ fig : sch_flux](b ) , the flux @xmath5 first increases , then flattens , and finally decreases , in the three wavenumber bands discussed above . in the intermediate band , @xmath126 , we observe kolmogorov s @xmath32 spectrum due to a constant ke flux . + since the flux does not decrease due to buoyancy ( see eq .
( [ eq : pi ] ) ) , the _ bo _ scaling is not applicable to rbc turbulence , contrary to the predictions by procaccia and zeitak @xcite , lvov @xcite , lvov and falkovich @xcite , and rubinstein @xcite .
+ [ enu_3 ] there is another useful flux called the entropy flux @xmath17 , which is defined as @xmath133 [ { \bf \theta^*(k ) \cdot \theta(p)}]).\ ] ] both , the _ ko _ and _ bo _ , phenomenologies predict a constant @xmath17 .
in this paper we simulate stably stratified flows and rbc , and compute the kinetic energy and entropy spectra , as well as fluxes .
we also compute @xmath134 , and @xmath122 , and show that our results are in good agreement with the arguments of items [ enu_2 ] and [ enu_3 ] discussed above .
for stably stratified flows , the _ bo _ scaling is observed when @xmath89 , but the kolmogorov scaling @xmath7 is observed when @xmath90 , or when buoyancy is negligible .
rbc flows , however , exhibit the kolmogorov scaling @xmath7 for a narrow band of wavenumbers .
we perform direct numerical simulation of stably stratified flows and rbc in a three - dimensional box by solving eqs .
( [ eq : u_ndim]-[eq : inc_ndim ] ) using pseudospectral code tarang @xcite .
we employ fourth - order runge - kutta ( rk4 ) method for time stepping , courant - freidricks - lewey ( cfl ) condition for computing time step @xmath135 , and @xmath136 rule for dealiasing . for the stratified flows , we employ the periodic boundary conditions on all sides of a cubic box of size @xmath137 .
to obtain a steady turbulent flow , we apply a random force to the flow in the band @xmath138 using the scheme of kimura and herring @xcite .
the parameters chosen for our simulations are @xmath139 ( close to that of air ) , and richardson numbers @xmath140 , and @xmath141 .
the grid resolution for @xmath142 is @xmath31 , which is one of the largest grids for such simulations .
the resolutions for @xmath143 and @xmath141 are @xmath144 grids .
the parameters of our runs are listed in table [ table : simulation_details ] .
all our simulations are fully resolved since @xmath145 , where @xmath146 is the maximum wavenumber of the run , and @xmath147 is the kolmogorov length scale .
we simulate rbc of a fluid in a unit box with @xmath144 grid .
the parameters of the simulation are @xmath148 and rayleigh number @xmath149 . for the horizontal plates , we employ free - slip boundary condition for the velocity field , and conducting boundary condition , i.e. @xmath150 , for the temperature field . for the vertical walls ,
we apply periodic boundary condition for both the fields .
simulation details of rbc simulation are listed at the bottom row of the table [ table : simulation_details ] . in the next section
we will compute the the spectra and fluxes of the kinetic energy as well as that of entropy .
p1.0 cm p1.4 cm p1.2 cm p1.2 cm p1.6 cm p1.6 cm p1.6 cm p1.2 cm p1.0 cm p1.0 cm p1.6 cm + grid & @xmath86 & @xmath151 & @xmath152 & @xmath84 & @xmath26 & @xmath21 & @xmath153 & @xmath154 & @xmath15 & @xmath135 + + @xmath144 & @xmath141 & @xmath155 & @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 + @xmath31 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 + @xmath144 & @xmath174 & @xmath175 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath170 & @xmath180 & @xmath181 & @xmath182 + + @xmath144 & @xmath183 & @xmath184 & @xmath185 & na & @xmath186 & @xmath187 & @xmath188 & @xmath189 & na & @xmath190 + [ table : simulation_details ]
we compute the the spectra and fluxes of the kinetic energy as well as that of entropy using the steady - state data .
we will also compute @xmath134 , and @xmath122 for the flows .
these results will be discussed below .
first , we simulate stably stratified flows for @xmath191 and @xmath192 on a @xmath31 grid , and compute the spectrum and flux using the steady state data .
[ fig : spectra_0_01](a ) illustrates the normalized ke spectra , @xmath193 for the _ bo _ scaling , and @xmath194 for the _ ko _ scaling .
the numerical data fits with the _ bo _ scaling quite well for approximately a decade , thus confirming the phenomenology of bolgiano and obukhov .
the normalized entropy spectra , @xmath195 ( _ bo _ scaling ) and @xmath196 ( _ ko _ scaling ) , illustrated in fig .
[ fig : spectra_0_01](b ) also show that the _ bo _ scaling is preferred for @xmath192 stably stratified flow .
we cross check our spectrum results with the ke and entropy fluxes , which are plotted in fig .
[ fig : flux ] . clearly , the ke flux , @xmath5 , is positive , and it decreases with @xmath6 .
however @xmath197 is almost flat , thus @xmath198 , same as eq .
( [ eq : pi ] ) .
we also observe that @xmath17 is a constant in the inertial range [ eq .
( [ eq : pi_theta ] ) ] ; thus flux results are consistent with the _ bo _ predictions .
we also compute the bolgiano wavenumber @xmath15 @xcite using the numerical data , and find that @xmath199 .
our plots on spectra and fluxes show that @xmath199 is only 3 to 4 times smaller than @xmath27 , wavenumber where the dissipation range starts .
therefore a clear - cut crossover from @xmath200 to @xmath32 is not observed in our simulations .
we are in the process of performing simulations on even higher resolution to probe the dual spectra ( @xmath200 and @xmath32 ) .
we also compute energy supply rate by buoyancy , @xmath201 , @xmath102 , and @xmath122 using the numerical data , and plot them in fig .
[ fig : deriv_1 ] .
the figure illustrate that @xmath202 , as argued in item [ enu_2 ] of sec .
[ sec : gov_eqns ] .
the negative @xmath100 implies that @xmath5 decreases with @xmath6 even without @xmath102 , which is a crucial ingredient for the _ bo _ scaling .
note that the kinetic energy flux is depleted by both @xmath100 and @xmath102 , and they satisfy the relation of eq .
( [ eq : dpik_dk ] ) .
interestingly , for small @xmath6 , @xmath203 ( the black line of fig . [
fig : deriv_1 ] ) , consistent with @xmath3 .
we also performed @xmath144 grid simulations for @xmath204 and @xmath174 with @xmath191 .
the normalized ke spectra for these two cases are exhibited in figs .
[ fig : ke_spec](a ) and [ fig : ke_spec](b ) respectively .
our results show that _ bo _ scaling is valid for @xmath205 , but _
ko _ scaling ( with a constant @xmath5 ) is valid for @xmath206 , which is as expected since buoyancy is significant only for moderate and large @xmath207 s . we compute @xmath100 , @xmath102 , and @xmath122 for @xmath208 and @xmath209 , and plot them in figs .
[ fig : deriv_2](a , b ) respectively . in the inertial range , @xmath210 for both the cases , just like @xmath142 .
the behaviour of @xmath100 , @xmath102 , and @xmath122 for @xmath208 is very similar to that of @xmath142 , except that @xmath100 for @xmath208 is a bit smaller than that for @xmath142 . for @xmath211 ,
buoyancy is weak , hence @xmath100 is much smaller than that for @xmath142 , which leads to an approximately constant @xmath5 , and kolmogorov s spectrum for the kinetic energy .
recall that we employ periodic boundary condition for the stably stratified flows in the vertical direction , thus eliminating the effects of boundary walls . in fig .
[ fig : temp_pro ] we plot the plane - averaged ( over @xmath212 plane ) mean temperature profile @xmath213 . since @xmath214 is linear , a constant temperature gradient @xmath215 ( hence buoyancy ) acts in the whole box .
therefore , _ bo _
scaling is expected everywhere .
it is important to contrast the above profile with that for rayleigh - bnard convection in which most of the temperature drop takes place in the narrow thermal boundary layers at the plates @xcite , while the bulk flow has @xmath216 .
thus we expect _ bo _ scaling in the boundary layers , and _
ko _ scaling in the bulk , as reported by calzavarini et al .
@xcite . in the next subsection
we will discuss the results of rayleigh bnard convection .
borue and orszag @xcite , and skandera et al .
@xcite simulated rbc flow under periodic boundary condition .
they observed the _
ko _ scaling for both velocity and temperature fields , consistent with the arguments presented in sec .
[ sec : gov_eqns ] .
a shell model approximates the turbulence in a periodic box quite well ; a recent shell model of rbc flow @xcite also yields _ ko _ scaling , consistent with the numerical results of borue and orszag @xcite , and skandera et al .
@xcite . in a typical rbc flow , however , a fluid
is confined between two horizontal conducting plates that are maintained at constant temperatures , with the bottom plate hotter than the top one .
earlier , mishra and verma @xcite showed that zero- and small prandtl number rbc exhibit kolmgorov s spectrum for the kinetic energy , but their results were inconclusive for moderate prandtl number rbc . in this subsection
, we will investigate this issue for @xmath191 . to explore
which of the two scaling ( _ ko _ or _ bo _ ) is applicable for rbc turbulence with plates , we perform rbc simulations for @xmath191 and @xmath217 , and compute the spectra and fluxes of the ke as well as the entropy for the steady state data . in fig .
[ fig : ke_specra_rbc](a ) , we plot the normalized ke spectra for the _ bo _ and the _ ko _ scaling .
the plots indicate that the _ ko _ scaling fits better than the _ bo _ scaling for a narrow band of wavenumbers ( the shaded region , @xmath218 ) .
we plot the ke and entropy fluxes in fig [ fig : ke_specra_rbc](b ) .
we also plot a zoomed view of the energy flux in fig .
[ fig : flux_details ] , according to which ke flux increases till @xmath219 , and then starts to decrease . in the logarithmic scale ,
the ke flux is an approximate constant for the wavenumbers @xmath218 , a band where @xmath7 .
thus we claim that convective turbulence exhibits kolmogorov s power law for a narrow band of wavenumbers .
interestingly , the energy spectrum of rbc exhibits stronger fluctuations than that of stably stratified turbulence ; this feature is possibly due to the plumes " emanating from the plates .
this feature as well as a larger range of wavenumber exhibiting _ ko _ scaling may be visible in a large resolution simulation , which is planned as a future study .
further investigations of @xmath100 , @xmath5 , and @xmath122 provide stronger evidence for the _ ko _ scaling in rbc .
we plot these quantities in figs .
[ fig : flux_details ] and [ fig : deriv_rbc ] , according to which @xmath220 , consistent with the discussion of sec . [ sec : gov_eqns ] and fig .
[ fig : sch_flux](b ) .
in addition , for the wavenumber band @xmath221 , @xmath125 , hence , according to eq .
( [ eq : dpik_dk ] ) , @xmath222 . therefore , @xmath5 increases in this band of wavenumbers , as illustrated in fig .
[ fig : flux_details ] .
but for @xmath223 , we find that @xmath224 leading to @xmath225 , therefore , @xmath5 decreases with @xmath6 for this range of @xmath6 . however , for a narrow band of wavenumbers @xmath218 , @xmath127 , hence @xmath226 or @xmath227 .
the constancy of @xmath5 yields @xmath7 , consistent with the energy spectrum plots of fig .
[ fig : ke_specra_rbc ] . note that many simulations , including mishra and verma @xcite , reported that @xmath3 for moderate @xmath228 , but the decrease of @xmath5 in their work is essentially due to @xmath102 , not due to buoyancy .
thus , the flux and energy supply due to buoyancy reveal that convective turbulence follows _ ko _ scaling , at least for a narrow range of wavenumbers .
scaling is ruled out for rbc since @xmath220 .
the entropy @xmath229 is a useful quantity in rbc . the entropy flux , illustrated in fig .
[ fig : ke_specra_rbc](b ) , is constant for the narrow inertial range @xmath230 . in fig .
[ fig : pe_specra_rbc ] , we plot the entropy spectrum that exhibits dual branch , with the upper branch scaling as @xmath231 . mishra and verma @xcite , and pandey et al .
@xcite showed the dominant temperature modes @xmath232 , which are approximately @xmath233 where @xmath234 is an integer , constitute the @xmath231 branch of the entropy spectrum .
they showed that @xmath232 modes are responsible for the steep temperature variations in the thermal boundary layers of the plates .
interestingly , the temperature modes in both the branches of the entropy spectrum participate to yield a constant entropy flux in the inertial range .
we performed large resolution simulations of stably stratified flows and rayleigh bnard convection , and studied the spectra and fluxes of the kinetic energy and entropy .
we also compute the energy supply rate due to buoyancy that provide important clues on the underlying turbulence phenomena . for stably stratified turbulence ,
we show that the kinetic energy spectrum @xmath1 , the energy flux @xmath3 , the entropy spectrum @xmath2 , and the entropy flux @xmath235 , in agreement with the prediction of bolgiano and obukhov , referred to as _
bo _ scaling .
we also compute the energy supply rate by buoyancy , and find that to be negative , signalling the buoyancy - induced conversion of kinetic energy to potential energy . for the rayleigh bnard convection
, the energy supply rate due to buoyancy , @xmath100 , is positive .
hence the kinetic energy flux @xmath236 first increases with @xmath6 , and then flattens for a narrow band of wavenumbers , and finally decreases with @xmath6 ; the three regimes correspond to @xmath125 , @xmath127 , and @xmath131 , respectively , where @xmath102 is the dissipation spectrum .
we observe kolmogorov s spectrum ( @xmath32 ) for wavenumbers where @xmath127 or @xmath227 .
thus , a detailed investigation of the kinetic energy flux , the energy supply due to buoyancy , and the dissipation spectrum provide valuable inputs that rule out _ bo _ scaling for rbc , contrary to the predictions of procaccia and zeitak @xcite , lvov @xcite , lvov and falkovich @xcite , and rubinstein @xcite .
the entropy flux for rbc is constant in the inertial range , but the entropy spectrum exhibit dual branch , whose origin is related to the thermal boundary layer . in summary ,
stably stratified flows exhibit _ bo _ scaling in buoyancy dominated regime .
turbulent convection however exhibits kolmogorov s spectrum , rather than _ bo _ spectrum .
a recent shell model of buoyancy - driven flows @xcite shows similar results .
more work , specially very large resolution simulations , are required to explore dual spectra predicted by bolgiano and obukhov .
our numerical simulations were performed at centre for development of advanced computing ( cdac ) and ibm blue gene p shaheen " at kaust supercomputing laboratory , saudi arabia .
this work was supported by a research grant serb / f/3279/2013 - 14 from science and engineering research board , india .
we thank ambrish pandey , anindya chatterjee , pankaj mishra , and mani chandra for valuable suggestions .
many researchers , e.g. @xcite , have nondimensionalized eqs . ( [ eq : u_dim]-[eq : inc_dim ] ) as the following .
they choose the characteristic horizontal velocity @xmath237 as the horizontal velocity scale , the horizontal length @xmath238 and the vertical height @xmath239 as the horizontal and vertical length scales respectively , @xmath240 as the time scale , @xmath241 as the vertical velocity scale where @xmath242 is the aspect ratio , and @xmath243 as the density scale . in terms of non - dimensional variables ,
the equations are @xmath244 where @xmath245 here @xmath246 is the horizontal froude number , and @xmath247 is the brunt - visl frequency . | using high - resolution direct numerical simulation and arguments based on the kinetic energy flux @xmath0 , we demonstrate that for stably stratified flows , the kinetic energy spectrum @xmath1 , the entropy spectrum @xmath2 , and @xmath3 , consistent with the bolgiano - obukhov scaling .
this scaling arises due to the conversion of kinetic energy to the potential energy by buoyancy . for weaker buoyancy ,
this conversion is weak , hence @xmath4 follows kolmogorov s spectrum with a constant energy flux . for rayleigh bnard convection ,
we show that the energy supply rate by buoyancy is positive , which leads to an increasing @xmath5 with @xmath6 , thus ruling out bolgiano - obukhov scaling for the convective turbulence .
our numerical results show that convective turbulence for unit prandt number exhibits a constant @xmath5 and @xmath7 for a narrow band of wavenumbers . |
in quantum computing , elementary operations are operations that act on only a few ( usually one or two ) qubits .
for example , cnots and one - qubit rotations are elementary operations .
a quantum compiling algorithm is an algorithm for decomposing ( compiling " ) an arbitrary unitary matrix into a sequence of elementary operations ( seo ) .
a quantum compiler is a software program that implements a quantum compiling algorithm .
one measure of the inefficiency of a quantum compiler is the number of cnots it uses to express an unstructured unitary matrix ( a unitary matrix with no special symmetries ) .
we will henceforth refer to this number as @xmath4 .
although good quantum compilers will also require optimizations that deal with structured matrices , unstructured matrices are certainly an important case worthy of attention . minimizing the number of cnots is a reasonable goal , since a cnot operation ( or any 2-qubit interaction used as a cnot surrogate ) is expected to take more time to perform and to introduce more environmental noise into the quantum computer than a one - qubit rotation .
ref.@xcite proved that for matrices of dimension @xmath5 ( @xmath6 number of bits ) , @xmath7 .
this lower bound is achieved for @xmath8 by the 3 cnot circuits first proposed in ref.@xcite .
it is not known whether this bound can be achieved for @xmath9 . [ cols=">,>,>",options="header " , ] suppose @xmath0 is an @xmath1-bit unstructured unitary matrix that we wish to compile . as the above table illustrates , compiling @xmath0 is hopeless for @xmath2 unless we approximate @xmath0 .
we need a method for finding a unitary matrix that : ( 1 ) approximates @xmath0 well , and ( 2 ) is expressible with fewer cnots than @xmath0 .
the purpose of this paper is to propose one such approximation method . the use of approximations in quantum compiling dates back to the earliest papers in the field .
for example , refs.@xcite and @xcite contain discussions on this issue .
as ref.@xcite points out , even when compiling a highly structured matrix like the discrete fourier transform matrix , some gates that contribute negligibly to its exact seo representation can be omitted with impunity .
refs.@xcite@xcite and @xcite discuss a quantum compiling algorithm that decomposes an arbitrary unitary matrix into a sequence of @xmath3-multiplexors , each of which is then decomposed into a seo .
other workers have proposed@xcite@xcite alternative quantum compiling algorithms that also generate @xmath3-multiplexors as an intermediate step .
the strategy proposed in this paper for approximating @xmath0 is to approximate the intermediate @xmath3-multiplexors whose product equals @xmath0 . in this paper
, we will show how one can approximate a @xmath3-multiplexor by another @xmath3-multiplexor ( the approximant " ) that has fewer controls , and , therefore , is expressible with fewer cnots .
we will call the reduction in the number of control bits the * bit deficit * @xmath10 .
fig.[fig - approx ] is emblematic of our approach .
it shows a @xmath3-multiplexor with 3 controls being approximated by either a @xmath3-multiplexor with 2 controls or one with 1 control .
in this section , we will define some notation that is used throughout this paper . for additional information about our notation
, we recommend that the reader consult ref.@xcite .
ref.@xcite is a review article , written by the author of this paper , which uses the same notation as this paper .
let @xmath11 . as usual
, let @xmath12 represent the set of integers ( negative and non - negative ) , real numbers , and complex numbers , respectively . for integers
@xmath13 , @xmath14 such that @xmath15 , let @xmath16 . for @xmath17
equal to @xmath18 or @xmath19 , let @xmath20 and @xmath21 represent the set of positive and non - negative @xmath17 numbers , respectively .
for any positive integer @xmath22 and any set @xmath23 , let @xmath24 denote the cartesian product of @xmath22 copies of @xmath23 ; i.e. , the set of all @xmath22-tuples of elements of @xmath23 .
for any ( not necessarily distinct ) objects @xmath25 , let @xmath26 denote an ordered set . for some object @xmath14 , let @xmath27 .
let @xmath28 be the empty set .
for an ordered set @xmath23 , let @xmath29 be @xmath23 in reverse order .
we will use @xmath30 to represent the truth function " ; @xmath30 equals 1 if statement @xmath23 is true and 0 if @xmath23 is false .
for example , the kronecker delta function is defined by @xmath31 . for @xmath32 , _ k=0 ^
1(-1)^kx=2 ( x,0 ) . [ eq - bool - delta - fun ] for any positive integer @xmath33 , we will use @xmath34 where @xmath35 to denote the standard basis vectors in @xmath33 dimensions ; i.e. , @xmath36_j = \delta(i , j)$ ] for @xmath37 .
@xmath38 and @xmath39 will represent the @xmath22-dimensional unit and zero matrices . for any matrix @xmath40 and positive integer @xmath22 , let a^n = _
n a , a^n = _ n a . for any matrix @xmath41 and @xmath42 , @xmath43 will represent the @xmath44-norm of @xmath40 , and @xmath45 its frobenius norm .
see @xcite for a discussion of matrix norms .
let @xmath46 .
as is customary in the physics literature , @xmath47 will also be denoted by @xmath48 and called the magnitude of @xmath49 . for any complex matrix @xmath40
, we will use @xmath50 to denote the matrix that is obtained from @xmath40 by replacing each of its entries by its absolute value .
in other words , @xmath51_{ij}= abs(a_{ij } ) = |a_{ij}|$ ] .
( careful : ref.@xcite and many other mathematical books call @xmath52 what we call @xmath50 ) .
any @xmath53 can be expressed as a doubly infinite power series in powers of a base @xmath54 : @xmath55 .
this expansion can be represented by : @xmath56 , which is called the base @xmath57 representation of @xmath58 .
the plus or minus in these expressions is chosen to agree with the sign of @xmath58 .
it is customary to omit the subscript @xmath59 when @xmath60 .
for example @xmath61 suppose @xmath62 .
note that division by 2 shifts the binary representation of @xmath58 one space to the right : @xmath63 . likewise , multiplication by 2 shifts the binary representation of @xmath58 one space to the left : @xmath64 . in general , for any @xmath65 , @xmath66 .
define the action of an overline placed over an @xmath67 by @xmath68 @xmath69 . call this bit negation .
define the action of an oplus placed between @xmath70 by @xmath71 . call this bit addition .
one can extend the bit negation and bit addition operations so that they can act on non - negative reals .
suppose @xmath72 , and @xmath73 are non - negative real numbers .
then define the action of an overline over @xmath58 so that it acts on each bit individually ; i.e. , so that @xmath74_\alpha = \overline{x_\alpha}$ ] .
this overline operation is sometimes called bitwise negation .
likewise , define the action of an oplus placed between @xmath58 and @xmath75 by @xmath76 .
this oplus operation is sometimes called bitwise addition ( without carry ) . for any @xmath53 ,
the floor function is defined by @xmath77 , and the ceiling function by @xmath78 .
for example , if @xmath72 , then @xmath79 .
we will often use @xmath1 to denote a number of bits , and @xmath80 to denote the corresponding number of states .
we will use the sets @xmath81 and @xmath82 interchangeably , since any @xmath83 can be identified with its binary representation @xmath84 . for any @xmath85 ,
define @xmath86 ; i.e. , @xmath87 is the result of reversing the binary representation of @xmath58 .
suppose @xmath88 is a 1 - 1 onto map .
( we use the letter @xmath89 to remind us it is a permutation ; i.e. , a 1 - 1 onto map from a finite set onto itself ) . one can define a permutation matrix @xmath90 with entries given by @xmath91 for all @xmath92 .
( recall that all permutation matrices @xmath90 arise from permuting the columns of the unit matrix , and they satisfy @xmath93 . ) in this paper , we will often represent the map @xmath89 and its corresponding matrix @xmath90 by the same symbol @xmath89 .
whether the function or the matrix is being alluded to will be clear from the context .
for example , suppose @xmath40 is an @xmath94 dimension matrix , and @xmath89 is a permutation on the set @xmath82 .
then , it is easy to check that for all @xmath95 , @xmath96 and @xmath97 .
suppose @xmath98 is a 1 - 1 onto map ( i.e. , a bit permutation ) .
@xmath99 can be extended to a map @xmath100 as follows .
if @xmath101 , then let @xmath102_\alpha = x_{\pi_b(\alpha)}$ ] for all @xmath103 .
the function @xmath100 is 1 - 1 onto , so it can be used to define a permutation matrix of the same name .
thus , the symbol @xmath99 will be used to refer to 3 different objects : a permutation on the set @xmath104 , a permutation on the set @xmath82 , and an @xmath94 dimensional permutation matrix .
all permutations on @xmath104 generate a permutation on @xmath82 , but not all permutations on @xmath82 have an underlying permutation on @xmath104 .
an example of a bit permutation that will arise later is @xmath105 ; it maps @xmath106 for all @xmath107 and @xmath108 for all @xmath109 .
in this section , we will review some well known facts about gray code@xcite .
( gray code was named after a person named gray , not after the color . ) for any positive integer @xmath1 , we define a * grayish code * to be a list of the elements of @xmath81 such that adjacent @xmath1-tuples of the list differ in only one component . in other words ,
a grayish code is a 1 - 1 onto map @xmath110 such that , for all @xmath111 , the binary representations of @xmath112 and @xmath113 differ in only one component . for any @xmath114 , there are many functions @xmath115 that satisfy this definition .
next we will define a particular grayish code that we shall refer to as the " gray code and denote by @xmath116 .
the gray code for @xmath117 is : 0&0&0 + 1&1&1 + 0&00&00 + 1&01&01 + 2&10&11 + 3&11&10 + 0&000&000 + 1&001&001 + 2&010&011 + 3&011&010 + 4&100&110 + 5&101&111 + 6&110&101 + 7&111&100 + .
the * gray code * can be defined recursively as follows .
let @xmath118 .
for @xmath119 , let @xmath120 equal the set @xmath81 ordered in the gray code order .
in other words , @xmath121 . then , _ + 1 = \{0 _ , 1^r_}_ord for @xmath122.(see section [ sec - notation ] for ordered set notation . ) from the recursive definition of the gray code , it is possible to prove that if @xmath123 and @xmath124 are nonnegative integers such that @xmath125 , then [ eq - gk - comps ] g_= k_k_+1 , [ eq - g - in - k - comps ] for all @xmath126 .
( for all @xmath127 , @xmath128 ) .
eq.([eq - g - in - k - comps ] ) specifies @xmath1 linear equations for the @xmath1 components of @xmath129 expressed in terms of the @xmath1 components of @xmath130 .
these equations can be easily inverted using gauss elimination to get : k_= g_g_+1 g_+2 g_+3 .
[ eq - k - in - g - comps ] eqs.([eq - gk - comps ] ) can also be written in terms of the floor function : [ eq - gk - floor ] g = k , k = g .
as in section [ sec - notation ] , suppose @xmath99 represents a permutation on @xmath104 which generates a permutation on @xmath82 of the same name . clearly , @xmath131 is a grayish code .
indeed , @xmath131 is a 1 - 1 onto map , and permuting bits the same way for all elements of a list preserves the property that adjacent @xmath1-tuples differ in only one component .
( note , however , that it is easy to find @xmath99 s such that @xmath132 is not a grayish code .
hence , to preserve grayishness , one must apply the bit permutation after @xmath116 , not before ) .
in this section , we will review some well known facts about the so called hadamard , paley and walsh matrices ( a.k.a .
transforms ) @xcite . for any positive integer @xmath1
, we define the @xmath1-bit hadamard matrix by ( h_)_k , r= ( -1)^ _ = 0 ^ -1 k_r _ , the @xmath1-bit paley matrix by ( p_)_k , r= ( -1)^ _ = 0 ^ -1 _ = 0 ^ -1 k_r__+ ^-1 , and the @xmath1-bit walsh matrix by ( w_)_k , r= ( -1)^ _ = 0 ^ -1 _ = 0 ^ -1 k_r _ , where @xmath133 , @xmath123 and @xmath134 .
we will often omit the subscript @xmath1 from @xmath135 in contexts where doing this does not lead to confusion .
note that @xmath136 are real symmetric matrices . for @xmath137 ,
define the reversal " function @xmath138 , and the negation " function @xmath139 .
the function @xmath140 for @xmath1-bit gray code has been defined previously .
the functions @xmath141 , @xmath142 and @xmath140 are 1 - 1 onto so they can be used to define permutation matrices of the same name ( see section [ sec - notation ] . )
we will often write @xmath105 , @xmath143 and @xmath116 instead of @xmath141 , @xmath142 and @xmath140 in contexts where this does not lead to confusion .
note that @xmath105 and @xmath143 are symmetric matrices but @xmath116 is nt .
next we will show that the @xmath1-bit hadamard , paley and walsh matrices all have the same columns , except in different orders .
more specifically , @xmath144 are related to each other by the following equations : h_r = p , p_g = w .
[ eq - h - p - w - relations ] a more pictorial way of expressing eqs.([eq - h - p - w - relations ] ) is : h p w .
taking the transpose of both sides of eqs.([eq - h - p - w - relations ] ) leads to _ r h = p , _
g^t p = w .
[ eq - h - p - w - transp - relations ] in the last equation , we have used the fact that matrices @xmath145 , @xmath146 are symmetric but @xmath116 is nt . comparing eqs.([eq - h - p - w - relations ] ) and ( [ eq - h - p - w - transp - relations ] )
, we see that h_r = _ r h , [ eq - hs - r - sym ] and p_g = _ g^t p .
[ eq - ps - g - sym ] in fact , eq.([eq - hs - r - sym ] ) can be generalized as follows .
suppose @xmath99 is a bit permutation on @xmath104 .
then ( _ b^t h _ b)_kr & = & h__b(k),_b(r ) + & = & ( -1)^ _ = 0 ^ -1 [ _ b(k ) ] _ _ + & = & ( -1)^ _ = 0 ^ -1 k _ ( ) r _ ( ) + & = & h_kr , so h_b = _ b h . [ eq - hs - b - sym ] eq.([eq - hs - b - sym ] ) becomes eq.([eq - hs - r - sym ] ) when @xmath147 . to prove eqs.([eq - h - p - w - relations ] ) , note that ( h_r)_ik&= & _ j=0 ^ -1 h_ij(_r)_jk + & = & _ j(-1)^_i_j _ ( j = k^r ) + & = & p_ik . similarly , ( p_g)_ik&= & _
j=0 ^ -1 p_ij(_g)_jk + & = & _ j(-1)^ _ , i_j__+^-1 ( j=_g(k ) ) + & = & ( -1)^ _ , i_(k_k_+1 ) _ + ^-1 + & = & ( -1)^ _ i_(k_-1- + k_- ) + & = & w_ik .
the square of @xmath148 and @xmath149 is one . indeed , using eq.([eq - bool - delta - fun ] ) we get ( h^2)_ik&= & _
j=0 ^ -1 h_ijh_jk + & = & _ j_-1=0 ^ 1
_j_1=0 ^ 1 _
j_0=0 ^ 1 ( -1)^ _ = 0 ^ -1 ( i_+k_)j _ + & = & _ _ i_^k _ = _ i^k , p^2=(_r h)(h_r)=1 , and w^2=(_g^t p)(p_g)=1 . since their square equals one , and they are real symmetric matrices , @xmath148 and @xmath149 are also orthogonal matrices . from the definitions given above for @xmath150
, one can prove by induction on @xmath1 that these matrices obey the following recursive equations : _ r(0)=1 , _
r(+1)= , _
n ( ) , and _ g(0)=1 , _ g(+1)= . similarly , from the definitions given above for @xmath136 , one can prove by induction on @xmath1 that these matrices obey the following recursive equations : h_0 = 1 , h_+1 = h _ , p_0 = 1 , p_+1= , and w_0 = 1 , w_+1= . by virtue of eqs.([eq - h - p - w - relations ] ) , w = h _ r _ g , [ eq - w - fun - h ] eq.([eq - w - fun - h ] ) means that the permutation @xmath151 will permute the columns of @xmath152 to give @xmath149 . expressing eq.([eq - w - fun - h ] ) in component form , we find w_ij & = & _ r , k h_ir ( _ r)_rk ( _ g)_kj + & = & _ r , k h_ir ( r=_r(k ) ) ( k=_g(j ) ) + & = & h_i,_r_g(j ) .
thus , if we denote the columns of @xmath153 and @xmath154 by @xmath155 and @xmath156 , respectively , then _ j = _ _ r_g(j ) , [ eq - wi - def ] for @xmath157 .
in this section , we will define a property of vectors called constancy .
the columns of @xmath153 can be conveniently classified according to their constancy . consider the 3-bit hadamard matrix : h_3 = h_1 ^ 3= & & & & & & & + & & & & & & & + & & & & & & & + & & & & & & & + & & & & & & & + & & & & & & & + & & & & & & & + & & & & & & & + , [ eq - had3 ] where we have labelled the columns of @xmath158 by @xmath155 , where the index @xmath159 is given in its binary representation . according to eq.([eq - wi - def ] ) , to get @xmath160 from eq.([eq - had3 ] ) , one can simply reorder the columns of @xmath158 in bit - reversed gray code . the columns of @xmath158 ( and of @xmath160 )
can be classified according to their constancy .
we define the * constancy * @xmath161 of a vector @xmath162 to be the smallest number of identical adjacent entries of @xmath162 . for example , @xmath163^t)=1 $ ] and @xmath164^t)=2 $ ] .
the next table gives the constancy of the columns of @xmath158 , with the columns listed in the order in which they appear in @xmath160 .
000&000&000&8 + 001&001&100&4 + 010&011&110&2 + 011&010&010&2 + 100&110&011&1 + 101&111&111&1 + 110&101&101&1 + 111&100&001&1 + .
[ tab - const ] it is clear from eq.([tab - const ] ) that the columns of @xmath160 are listed in order of non - increasing constancy , and that the constancies of the columns of @xmath160 are all powers of 2 .
the literature on walsh matrices often refers to the index that labels the columns of @xmath149 as the * sequency * of that column .
thus , as sequency increases , constancy decreases or stays the same .
sequency and constancy are analogous to frequency and period , respectively , in fourier analysis
. note that given any matrix @xmath40 , more than one of the columns of @xmath40 may have the same constancy .
we will refer to : the number of columns of @xmath40 with the same constancy @xmath165 , as : the * multiplicity * * of the constancy * @xmath165 * in the matrix * @xmath40 , and denote it by @xmath166 . in this paper , we are only concerned with the case where @xmath40 equals the @xmath1-bit hadamard matrix so we will henceforth omit the subscript @xmath40 from @xmath166 .
sometimes we will also abbreviate @xmath167 by @xmath168 , if doing this does not lead to confusion .
the next table gives the multiplicity of the constancy @xmath165 in the @xmath1-bit hadamard matrix : l|lllll ( = k)&k=1 & k=2 & k=4 & k=8 & + = 1 & 1 & 1 & 0 & 0 & + = 2 & 2 & 1 & 1 & 0 & + =3 & 4 & 2 & 1 & 1 & + & & & & & + .
[ tab - multi ] it is also convenient to define ( )= _ k ( k ) .
[ eq - cum - multi ] we shall call this the * cumulative multiplicity of the constancy*. the next table can be easily obtained from eq.([tab - multi ] ) and eq.([eq - cum - multi ] ) .
it gives @xmath169 for the @xmath1-bit hadamard matrix .
l|lllll ( )&=2 ^ 0 & =2 ^ 1 & =2 ^ 2 & =2 ^ 3 & + = 1 & 2 & 1 & 0 & 0 & + = 2 & 4 & 2 & 1 & 0 & + =3 & 8 & 4 & 2 & 1 & + & & & & & + .
[ tab - cum - multi ] it is clear from eq.([tab - cum - multi ] ) that @xmath170 , where @xmath171 .
in this section , we discuss some symmetries of exact decompositions of @xmath3-multiplexors . for simplicity , we will first consider the @xmath172-multiplexors used in ref.@xcite .
ref.@xcite uses @xmath3-multiplexors that are more general than the @xmath172-multiplexors used in ref.@xcite . at the end of the paper
, we will discuss how to generalize our results for @xmath172-multiplexors so that they apply to the more general multiplexors used in ref.@xcite .
below , we will present some quantum circuit diagrams . besides the circuit notational conventions defined in refs.@xcite and @xcite , the circuits below will use the following additional notation . a square gate with an angle @xmath173 below the square will represent @xmath174 applied at that wire " . typically , we will consider a seo consisting of alternating one - qubit rotations and cnots .
the seo will always have a one - qubit rotation at one end and a cnot at the other .
the angle for the one - qubit rotation that either begins or ends the seo will be denoted by @xmath175 . given two adjacent angles @xmath176 and @xmath177 in the seo , @xmath178 and @xmath179 will differ only in one component , component @xmath180 , where @xmath180 is the position of the control bit of the cnot that lies between the @xmath176 and @xmath177 gates .
if we take the hermitian conjugate of the multiplexor @xmath181 , and then we replace the angles @xmath182 by their negatives ( and also the angles @xmath176 , hadamard transforms of the @xmath183 , by their negatives ) , we get the same multiplexor back .
henceforth , we will refer to this symmetry transformation as * time reversal*. thus , an @xmath172-multiplexor is invariant under time reversal .
suppose @xmath99 is a bit permutation on @xmath184 bits .
if we replace @xmath182 by @xmath185 ( and also @xmath176 by @xmath186 ) and @xmath187 by @xmath188 in the multiplexor @xmath181 , we get the same multiplexor back .
henceforth , we will refer to this symmetry transformation as * bit permutation*. thus , an @xmath172-multiplexor is also invariant under bit permutation .
fig.[fig - sym - ops ] shows how time reversal and bit permutation act on a sequence of one - qubit rotations and cnots .
more examples of the application of these transformations will be given below .
recall from ref.@xcite our definition of a general multiplexor with @xmath189 control qubits @xmath190 and @xmath191 target qubits @xmath192 : @xmath193 . in a multiplexor
whose matrices @xmath194 are independent of the @xmath180 component @xmath195 of @xmath14 , we can sum @xmath196 over @xmath197 to get 1 .
such a multiplexor acts as the identity on qubit @xmath180 .
when representing such a multiplexor in a circuit diagram ,
we can omit its half - moon node on qubit line @xmath180 .
fig.[fig - free - half - moon ] shows a very special case of this principle , a special case that will be used in the circuit diagrams below .
fig.[fig-1control ] shows two possible ways of decomposing an @xmath172-multiplexor with one control .
the decomposition ( a ) in fig.[fig-1control ] is equivalent to : ( i_bbool _ bp_b)=
e^i_0(1 ) ( 1)^n(0 ) e^i_1(1 ) ( 1)^n(0 ) .
[ eq - decomp-1control ] let lhs and rhs stand for the left and right hand sides of eq.([eq - decomp-1control ] ) . recall that @xmath198 and @xmath199 .
eq.([eq - decomp-1control ] ) can be proven as follows : [ eq - decomp-1control - proof ] @xmath200\}}\\ & = & lhs \label{eq - step - subst - theta } \;.\end{aligned}\ ] ] to arrive at eq.([eq - step - subst - theta ] ) , we expressed @xmath201 in terms of @xmath202 using = . if we take the hermitian conjugate of both sides of eq.([eq - decomp-1control ] ) , and then we replace the angles @xmath182 and @xmath203 by their negatives , we get ( i_bbool _ bp_b)= ( 1)^n(0 ) e^i_1(1 ) ( 1)^n(0 ) e^i_0(1 ) . [ eq - decomp-1control - rev ] eq.([eq - decomp-1control - rev ] ) is equivalent to decomposition ( b ) in fig.[fig-1control ] .
thus , decompositions ( a ) and ( b ) in fig.[fig-1control ] transform into each other under time reversal .
fig.[fig-2controls ] shows four possible ways of decomposing an @xmath172-multiplexor with two controls .
fig.[fig-2controls ] was obtained by applying the results of figs.[fig - free - half - moon ] and [ fig-1control ] . in fig.[fig-2controls ] , note that decompositions ( a ) and ( b ) transform into each other under time reversal .
decompositions ( c ) and ( d ) do too .
furthermore , decompositions ( b ) and ( c ) transform into each other under bit permutation .
the decompositions exhibited in fig.[fig-2controls ] can also be expressed analytically .
for example , decomposition ( b ) is equivalent to : @xmath204 eq.([eq - decomp-2-controls ] ) can be proven using the same techniques that were employed in eqs.([eq - decomp-1control - proof ] ) to prove eq.([eq - decomp-1control ] ) .
the proof requires that we assume : = .
[ fig-3controls ] ( ditto , [ fig-4controls ] ) shows one of several possible decompositions of an @xmath172-multiplexor with 3 ( ditto , 4 ) controls . in general , decompositions for multiplexors with @xmath189 controls
can be obtained starting from decompositions for multiplexors with @xmath205 controls .
in this section , we finally define our approximation of multiplexors . we give the number of cnots required to express the approximant , and an upper bound to the error incurred by using it .
so far we have used @xmath1 to denote a number of bits , and @xmath206 to denote the corresponding number of states .
below , we will use two other numbers of bits , @xmath207 and @xmath208 , where @xmath209 and @xmath210 .
their corresponding numbers of states will be denoted by @xmath211 and @xmath212 .
define an @xmath213 dimensional matrix @xmath214 by v = h _ b _ g , [ eq - v - fun - h ] where @xmath99 is an arbitrary bit permutation on @xmath207 bits .
eq.([eq - v - fun - h ] ) is a generalization of eq.([eq - w - fun - h ] ) .
both equations define a new matrix ( either @xmath214 or the walsh matrix @xmath149 ) by permuting the columns of the hadamard matrix @xmath152 .
@xmath214 becomes @xmath149 if we specialize the bit permutation @xmath99 to @xmath105 .
if we denote the columns of @xmath214 by @xmath215 for @xmath216 , then _
j=__b_g(j ) , [ eq - vi - def ] which is the counterpart of eq.([eq - wi - def ] ) . in ref.@xcite ,
the decomposition of an @xmath172 multiplexor starts by taking the following hadamard transform : = h _ , [ eq - theta - fun - phi ] where @xmath217 and @xmath218 .
the vectors @xmath219 constitute an orthonormal basis for the space @xmath220 in which @xmath221 lives , so @xmath221 can always be expanded in terms of them : = _ i=0 ^ -1 _ i ( _ i^ ) .
now suppose that we truncate this expansion , keeping only the first @xmath222 terms , where @xmath223 and @xmath224 .
let us call @xmath225 the resulting approximation to @xmath221 : = _ i=0^-1 _ i ( _ i^ ) .
[ eq - def - phi - prime ] define @xmath226 , an approximation to @xmath227 , as follows : = h _ .
[ eq - def - theta - prime ] if we let @xmath228 denote the standard basis vectors , then h__i= _ _ b_g(i ) = _ _ b_g(i ) .
therefore , = _ i=0^-1 _ _ b_g(i ) ( _ i^ ) .
[ eq - theta - prime ] by virtue of eq.([eq - theta - prime ] ) , if we list the components @xmath229 of @xmath226 in the grayish code order specified by the map @xmath230 , then the items in the list at positions from @xmath222 to the end of the list are zero .
consider , for example , fig.[fig-2controls ] , which gives the exact decompositions for a multiplexor with 2 controls .
suppose that in one of those decompositions , the angles @xmath203 s in the second half ( i.e. , the half that does not contain @xmath231 ) of the decomposition are all zero .
then the one - qubit rotations in the second half of the decomposition become the identity .
then the three cnots in the second half of the decomposition cancel each other in pairs except for one cnot that survives .
the net effect is that the decomposition for a multiplexor with 2 controls degenerates into a decomposition for a multiplexor with only 1 control .
the number of control bits is reduced by one in this example . in general
, we can approximate a @xmath3-multiplexor by another @xmath3-multiplexor ( the approximant " ) that has fewer controls , and , therefore , is expressible with fewer cnots .
we will call the reduction in the number of control bits the * bit deficit * @xmath10 . hence , @xmath232 .
if @xmath4 denotes the number of cnots in an approximant with bit deficit @xmath10 , then it is clear from figs.[fig-1control ] , [ fig-2controls ] , [ fig-3controls ] and [ fig-4controls ] that : n_cnot= & 2 ^ -1 & 2 ^ -2 & 2 ^ -3 & & 2 & 0 + .
[ eq - n - cnots ] hence , for @xmath233 , @xmath234 , but for @xmath235 , @xmath236 .
the bit permutation @xmath99 on which the approximation of a multiplexor depends can be chosen according to various criteria .
if we choose @xmath237 , then our approximation will keep only the higher constancy components of @xmath221 .
such a smoothing , * high constancies approximation * might be useful for some tasks . similarly ,
if we choose @xmath238 , then our approximation will keep only the lower constancy components of @xmath221 , giving a * low constancies approximation*. alternatively , we could use for @xmath99 a bit permutation , out of all possible bit permutations on @xmath207 bits , that minimizes the distance between the original multiplexor and its approximant .
such a * dominant constancies approximation * is useful if our goal is to minimize the error incurred by the approximation .
the error incurred by approximating a multiplexor can be bounded above as follows .
let @xmath239 denote the @xmath172-subset of an @xmath172-multiplexor @xmath240 and @xmath241 that of its approximant @xmath242 .
call @xmath243 the error of approximating @xmath240 by @xmath242 .
note that [ eq - error ] @xmath244 to arrive at step eq.([eq - step - ineq ] ) , we used the results of appendix [ ap - ineq ] .
we will sometimes refer to @xmath245 as the linearized error , to distinguish it from the error @xmath243 .
a simple picture emerges from all this .
the error @xmath246 and the number of cnots @xmath4 are two costs that we would like to minimize .
these two costs are fungible to a certain extent . given a multiplexor @xmath240 , and an upper bound @xmath247 on @xmath246 , we can use eqs.([eq - n - cnots ] ) and ( [ eq - error ] ) to find the approximant @xmath242 with the smallest @xmath4 .
similarly , given a multiplexor @xmath240 , and an upper bound @xmath248 on @xmath4 , we can use eqs.([eq - n - cnots ] ) and ( [ eq - error ] ) to find the approximant @xmath242 with the smallest @xmath246 . at this point
we encourage the reader to read appendix [ ap - computer - results ] .
it discusses the output of a computer program that calculates @xmath225 from @xmath221 via eq.([eq - def - phi - prime ] ) .
next we will show that eq.([eq - def - phi - prime ] ) can be simplified considerably by taking into account the explicit values of the column vectors @xmath215 . to get a quick glimpse of the simplification we seek ,
consider first the special case @xmath249 .
we have h_2 = & & & + & & & + & & & + & & & + , w_2 = & & & + & & & + & & & + & & & + .
define a matrix @xmath250 by = ( cc 1 & 1 + 1 & 1 ) . for any matrix @xmath40 ,
let @xmath251 be the submatrix of @xmath40 obtained by keeping only its columns from @xmath252 to @xmath159 .
it is easy to check that w_2(:,0:3 ) w_2(:,0:3)^t = diag(1,1,1,1 ) , w_2(:,0:1 ) w_2(:,0:1)^t = ( cc & 0 + 0 & ) = ( ) , [ eq - w2-two - cols ] and w_2(:,0:0 ) w_2(:,0:0)^t = ( cc & + & ) = ( ) . in each case , we formed a decimated matrix " @xmath253 from @xmath149 , where @xmath223 .
then we showed that the projection operator @xmath254 onto the column space of the decimated matrix , is a matrix whose entries are all either 0 or @xmath255 , and these entries sum to one along each row ( or column ) .
given a set @xmath23 of real numbers , and given @xmath256 , call the average of the elements of @xmath257 a partial average " of the elements of @xmath23 . for example , if @xmath258 and @xmath237 , then @xmath259 and @xmath260 . from eq.([eq - w2-two - cols ] ) , the entries of @xmath225 are partial averages of the entries of @xmath221 .
next we show how to simplify eq.([eq - def - phi - prime ] ) for arbitrary @xmath207 , not just for @xmath249 .
define an @xmath213 dimensional matrix @xmath261 by ^ ( , )_qs= _ r=0^-1 ( w_)_qr ( w_^t)_rs .
[ eq - gamma - fun - w ] below , we will show that @xmath261 reduces to : ^ ( , )_qs= ( = ) .
[ eq - gamma - fun - floor ] for example , when @xmath262 , eq.([eq - gamma - fun - floor ] ) becomes = = c|cccc & 00&01 & 10&11 + & 1&1&0&0 + 01&1&1&0&0 + 10&0&0&1&1 + 11&0&0&1&1 .
eqs.([eq - gamma - fun - w ] ) and ( [ eq - gamma - fun - floor ] ) are given in component form .
the identical statements written in matrix form are , respectively , ^ ( , ) & = & w_(:,0:-1 ) w_(:,0:-1)^t + & = & w_w_^t , and ^ ( , ) = ( ^_b)^ .
eq.([eq - gamma - fun - floor ] ) can be proven from eq.([eq - gamma - fun - w ] ) as follows : ^ ( , )_qs&= & _ r=0^-1 ( -1)^ _ = 0 ^ -1 _ = 0 ^ -1 ( q_r_+ r_s _ ) ( _ + ^ -1 + _
+ ^ ) + & = & ( _ r_-1=0 ^ 1 _ r_1=0 ^ 1 _ r_0=0 ^ 1 ) ( r_-1 = r_-2 = = r_-_b=0 ) + & & ( -1)^ _ = 0 ^ -1 _ = 0 ^ -1 r_t _ ( _ + ^ -1 +
_ + ^ ) ( where t = qs ) + & = & ( _ r_-1=0 ^ 1 _ r_1=0 ^ 1 _ r_0=0 ^
1 ) ( r_-1 = r_-2 = = r_-_b=0 ) + & & ( -1)^ r_-1(t_0 + t_1 ) + r_-2(t_1 + t_2 ) + r_1(t_-2 + t_-1 ) + r_0(t_-1 ) + & = & 2 ^ -_b _ q_-1^s_-1 _ q_-2^s_-2 _ q__b^s__b [ eq - delta - fun - step ] + & = & ( = ) . to arrive at eq.([eq - delta - fun - step ] ) , we used eq.([eq - bool - delta - fun ] ) .
recall that @xmath149 and @xmath214 can both we obtained by permuting the columns of @xmath152 : w = h _ r _ g , and v = h _ b _ g . from these two equations and from the fact , proven earlier , that @xmath99 and @xmath105 commute with @xmath152
, we get v = _ b _ r h _ r _ g = _ b _ r w . thus , & = & v_v_^t + & = & _ b _ r w_w_^t _ r _ b^t . hence ,
if we define @xmath263 and @xmath264 by = _ r _ b^t , = _ r _ b^t , then = ^ ( , ) .
[ eq - phi - pr - fun - phi ] we see from eq.([eq - phi - pr - fun - phi ] ) that even when @xmath265 , the entries of @xmath264 ( which are the same as the entries of @xmath225 but in a different order ) are partial averages of the entries of @xmath263 ( which are the same as those of @xmath221 but in a different order ) . thus _
k(_k ) _ j _ k ( _ k ) , for all @xmath216 .
this last equation implies as we mentioned before , the quantum compiling algorithm of ref.@xcite uses @xmath3-multiplexors that are more general than the @xmath172-multiplexors considered above .
luckily , the above results for @xmath172-multiplexors are still valid , with minor modifications , for the more general ones .
indeed , the @xmath3-subset of the multiplexors used in ref.@xcite is of the form @xmath266 \sigma_w^{f(b)}\}_{\forall b}$ ] .
ref.@xcite defines vectors @xmath267 and @xmath268 from the parameters @xmath269 and @xmath270 , respectively .
it then defines @xmath271 and @xmath272 as hadamard transforms of @xmath273 and @xmath274 , respectively , just as eq.([eq - theta - fun - phi ] ) defines @xmath227 as a hadamard transform of @xmath221 .
we can define approximations @xmath275 and @xmath276 by replacing @xmath221 , @xmath227 , @xmath225 , @xmath226 by @xmath273 , @xmath271 , @xmath275 , @xmath276 , respectively , within eqs.([eq - def - phi - prime ] ) and ( [ eq - def - theta - prime ] ) .
we can define approximations @xmath277 and @xmath278 analogously .
the expansions of @xmath279 and @xmath280 in the @xmath281 basis can be truncated at the same @xmath222 .
the table given in eq.([eq - n - cnots ] ) for the number of cnots still applies , except that @xmath4 may change by 1 if we eliminate the @xmath282 gate as in ref.@xcite .
when a @xmath3-multiplexor @xmath240 with @xmath3-subset @xmath266 \sigma_w^{f(b)}\}_{\forall b}$ ] is approximated by a @xmath3-multiplexor @xmath242 with @xmath3-subset @xmath283 \sigma_w^{f(b)}\}_{\forall b}$ ] , one can show , using the results of appendix [ ap - ineq ] , that -_2 _ b , which is a generalization of eq.([eq - error ] ) .
in this appendix , we establish a well known(see ref.@xcite , page 574 ) upper bound for the distance ( measured in either the 2-norm or the frobenius norm ) between two @xmath284 matrices .
let @xmath285 be 3d real vectors .
define @xmath286 . if @xmath287 , then e^i -e^i_2&= & e^i e^-i -1_2 + & & i_2 + & = & || .
next , we will show that this approximation can be turned into an inequality .
consider first the special case where @xmath288 and @xmath289 both point in the y direction .
then e^i - e^i_2 & = & e^i-1_2 + & = & ( cc ( ) -1 & ( ) + - ( ) & ( ) -1 ) _ 2 + & = & 2 ( ) ( cc - ( ) & ( ) + - ( ) & - ( ) ) _ 2 + & = & 2 |()| + & & || . to find an upper bound for @xmath290 when either @xmath288 or @xmath289 does not point in the y direction , we will use the following identity . for @xmath291 and @xmath292 ,
e^(a+e)t - e^at = _
0^t ds e^a(t - s ) e e^(a+e)s .
[ eq - diff - eq - sol ] to prove eq.([eq - diff - eq - sol ] ) , let @xmath293 and @xmath294 stand for the left and right hand sides of eq.([eq - diff - eq - sol ] ) .
it is easy to verify that ( l - r)(0)=0 , = a(l - r ) .
this initial value problem has the unique solution @xmath295 . in eq.([eq - diff - eq - sol ] ) , set @xmath296 , and @xmath297 , and take the 2-norm of both sides .
this yields e^i ( + ) - e^i _ 2 & & _ 0 ^ 1 ds e^i ( 1-s)_2 i _ 2 e^i ( + ) s _ 2 + & = & ||
. one can also find an upper bound for the distance , in the frobenius norm , between two @xmath284 matrices . if @xmath298 and @xmath299 , then the eigenvalues of @xmath300 are @xmath301 , for some real number @xmath173 . likewise , the eigenvalues of @xmath302 are @xmath303 .
thus @xmath304 is real .
if we denote the eigenvalues of @xmath305 by @xmath306 with @xmath307 , then @xmath308 has a single eigenvalue @xmath309 with algebraic multiplicity 2 .
thus e^i -e^i_f = e^i -e^i_2 .
[ eq-2-f - norms ] but we ve already proven that @xmath290 is bounded above by @xmath310 so e^i -e^i_f | | .
in this appendix , we discuss a simple computer program that verifies and illustrates many of the results of this paper .
our program is written in the octave language .
octave is a gratis , open - source interpreter that understands a subset of the matlab language .
hence , our program should also run in a matlab environment with few or no modifications . ....
phi(1)= 0.133765891 phi(2)= 0.270447403 phi(3)= 0.307625920 phi(4)= 0.311291575 phi(5)= 0.452735037 phi(6)= 0.569045961 phi(7)= 0.653136015 phi(8)= 0.867156088 ----------------------- permutation 1 = ( 1,2,3 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.293 0.420 0.480 0.589 0.293 0.420 0.480 0.589 2 0.387 0.504 0.387 0.504 0.387 0.504 0.387 0.504 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 ----------------------- permutation 2 = ( 1,3,2 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.293 0.420 0.480 0.589 0.293 0.420 0.480 0.589 2 0.356 0.356 0.535 0.535 0.356 0.356 0.535 0.535 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 ----------------------- permutation 3 = ( 2,1,3 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.221 0.291 0.221 0.291 0.553 0.718 0.553 0.718 2 0.387 0.504 0.387 0.504 0.387 0.504 0.387 0.504 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 ----------------------- permutation 4 = ( 2,3,1 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.202 0.202 0.309 0.309 0.511 0.511 0.760 0.760 2 0.356 0.356 0.535 0.535 0.356 0.356 0.535 0.535 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 ----------------------- permutation 5 = ( 3,1,2 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.221 0.291 0.221 0.291 0.553 0.718 0.553 0.718 2 0.256 0.256 0.256 0.256 0.636 0.636 0.636 0.636 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 ----------------------- permutation 6
= ( 3,2,1 ) delta_b , phi_prime= 0 0.134 0.270 0.308 0.311 0.453 0.569 0.653 0.867 1 0.202 0.202 0.309 0.309 0.511 0.511 0.760 0.760 2 0.256 0.256 0.256 0.256 0.636 0.636 0.636 0.636 3 0.446 0.446 0.446 0.446 0.446 0.446 0.446 0.446 .... .... error as function of ( permutation\delta_b ) 0 1 2 3 1 1.110e-16 2.779e-01 3.627e-01 4.215e-01 2 1.110e-16 2.779e-01 3.324e-01 4.215e-01 3 1.110e-16 1.491e-01 3.627e-01 4.215e-01 4 1.110e-16 1.070e-01 3.324e-01 4.215e-01 5 5.551e-17 1.491e-01 2.316e-01 4.215e-01 6 5.551e-17 1.070e-01 2.316e-01 4.215e-01 .... in this example , @xmath311 so @xmath312 and @xmath313 .
the first 8 lines of ` out_phis.txt ` give the components of @xmath221 . in this case , the computer picked 8 independent random numbers from the unit interval , and then it sorted them in non - decreasing order . ` my_moo.m ` can be easily modified so as to allow the user himself to supply the components of @xmath221 . after listing @xmath221 , ` out_phis.txt ` lists the @xmath314 permutations @xmath99 of @xmath207 bits . for each @xmath99
, it prints the components of @xmath225 , listed as a row , for each value of @xmath10(=row label ) .
note that for @xmath315 , @xmath316 , and for @xmath317 , all @xmath318 are equal to the average of all the components of @xmath221 .
note also that for all values of @xmath10 and @xmath159 , @xmath319 $ ] .
the second output file , ` out_error.txt ` , gives a table of the linearized error @xmath245 as a function of permutation number(=row label ) and @xmath10(=column label ) .
as expected , the error is zero when @xmath10 is zero , and it is independent of the permutation @xmath99 when @xmath10 is maximum ( when the bit deficit @xmath10 is maximum , the approximant has no control bits , so permuting bits at positions @xmath320 does not affect the error . ) note that in the above example , the last permutation minimizes the error for all @xmath10 .
this last permutation is @xmath321 ( bit - reversal ) , and it gives a high constancies expansion . recall that for this example , ` my_moo.m ` generated iid ( independent , identically distributed ) numbers for the components of @xmath221 , and then it rearranged them in monotonic order .
when @xmath221 is chosen in this way , the graph @xmath322 has a high probability of lying close to a straight line , and a high constancy staircase is the best fit for a straight line .
for this reason , almost every time that ` my_moo.m ` is operated in the mode which generates iid numbers for the components of @xmath221 , the high constancies expansion minimizes the error for all @xmath10 .
however , this need not always occur , as the following counterexample shows . try running ` my_moo.m ` for @xmath323 , and for @xmath221 with its first 7 components equal to 0 and its 9 subsequent components equal to 1 . for this @xmath221 , and for @xmath324 , the high constancies expansion yields an error of 7/8 while some of the other expansions yield errors as low as 5/8 . note that although ` my_moo.m ` visits all @xmath314 permutations of the control bits , visiting all permutations is a very inefficient way of finding the minimum error .
in fact , the @xmath314 control bit permutations can be grouped into equivalence classes , such that all permutations in a class give the same error .
it s clear from fig.[fig - approx ] that we only have to visit @xmath325 ( recall @xmath326 ) equivalence classes of permutations . whereas @xmath327 is exponential in @xmath207 , @xmath328 is polynomial in @xmath207 for two very important extremes .
namely , when @xmath10 or @xmath208 is of order one whereas @xmath207 is very large . indeed , if @xmath329 or @xmath330 , then @xmath331 ; if @xmath332 or @xmath333 , then @xmath334 , etc . | a quantum compiling algorithm is an algorithm for decomposing ( compiling " ) an arbitrary unitary matrix into a sequence of elementary operations ( seo ) .
suppose @xmath0 is an @xmath1-bit unstructured unitary matrix ( a unitary matrix with no special symmetries ) that we wish to compile . for @xmath2 , expressing @xmath0 as a seo requires more than a million cnots .
this calls for a method for finding a unitary matrix that : ( 1)approximates @xmath0 well , and ( 2 is expressible with fewer cnots than @xmath0 .
the purpose of this paper is to propose one such approximation method .
various quantum compiling algorithms have been proposed in the literature that decompose @xmath0 into a sequence of @xmath3-multiplexors , each of which is then decomposed into a seo .
our strategy for approximating @xmath0 is to approximate these intermediate @xmath3-multiplexors . in this paper
, we will show how one can approximate a @xmath3-multiplexor by another @xmath3-multiplexor that is expressible with fewer cnots . |
as most of the major natural satellites of the solar system planets , titan is locked in a spin - orbit synchronous resonance , i.e. its rotation period is very near to its orbital period around its parent planet saturn ( see for instance richardson et al .
2004 @xcite ) .
this corresponds to an equilibrium state known as a cassini state .
thanks to the cassini fly - bys , titan s gravity field is known well enough to study the behaviour of titan s spin around the cassini state . in particular , we now know titan s oblateness coefficients @xmath13 and @xmath14 ( tortora et al .
@xcite ) . in a recent paper , noyelles et al .
( @xcite , hereafter cited as paper i ) give a first theory of titan s rotation , with 3 degrees of freedom . in that work , an analytical approach and a numerical one , more complete ,
are compared .
moreover , some aspects of titan s rotation are elucidated , especially a likely resonance involving the free libration of titan s wobble . in this paper ,
i first propose an improvement of the analytical model resulting in better agreement with the numerical results .
then i study the dynamics of the likely resonance , and discuss its implications .
i firstly recall how to obtain the hamiltonian of the problem , as explained for instance in paper i. 3 references frames are considered : the first one @xmath15 is centered on titan s mass barycenter and is in translation with the reference frame in which the orbital motion of titan is described .
this is a cartesian coordinate system refering to the equatorial plane of saturn and to the node of this plane with the ecliptic at j2000 .
the second frame @xmath16 is linked to titan s angular momentum @xmath17 , and the third one @xmath18 is rigidly linked to titan .
the first variables that are being used are andoyer s variables ( see andoyer 1926 @xcite and deprit 1967 @xcite ) , which are based on two linked sets of euler s angles .
the first set @xmath19 locates the position of the angular momentum in the first frame @xmath20 , while the second one @xmath21 locates the body frame @xmath18 in the second frame @xmath16 ( see fig . [
fig : angles ] ) .
the canonical set of andoyer s variables consists of the three angular variables @xmath22 , @xmath23 , @xmath24 and their conjugated momenta defined by the norm @xmath25 of the angular momentum and two of its projections : @xmath26 + unfortunately , these variables present two singularities : when @xmath27 ( i.e. the angular momentum is colinear to @xmath28 , there is no wobble ) , @xmath22 and @xmath23 are undefined , and when @xmath29 ( i.e. when titan s principal axis of inertia is perpendicular to its orbital plane ) , @xmath24 and @xmath23 are undefined .
that is why we use the modified andoyer s variables : @xmath30 with @xmath31 .
paper i details how to obtain the hamiltonian of the problem in the modified andoyer s variables .
it takes account of the free rotation of the body , and of the external torque due to saturn ( see eq.[equ : hamiltonclass ] ) .
@xmath32\big[\frac{\gamma_1+\gamma_2 } { 1-\gamma_1-\gamma_2}\xi_q^2+\frac{\gamma_1-\gamma_2}{1-\gamma_1+\gamma_2}\eta_q^2\big ] \\ + n\big(\frac{d_0}{d}\big)^3\bigg(1+\delta_s\big(\frac{d_0}{d}\big)^2\bigg)\big[\delta_1(x^2+y^2)+\delta_2(x^2-y^2)\big ] \end{split}\ ] ] the coefficients of the hamiltonian are defined as follows : @xmath33 + where @xmath34 is saturn s @xmath13 , @xmath35 the distance titan - saturn , @xmath36 the mean value of @xmath35 , @xmath37 the mean motion associated with @xmath36 .
@xmath38 and @xmath39 are the first two coordinates of saturn in the reference frame @xmath18 bound to titan .
if we call @xmath40 , @xmath41 and @xmath42 the coordinates of saturn s centre of mass in the reference frame @xmath20 defined above , we have @xmath43 with @xmath44 and @xmath45 at the exact cassini state , @xmath46 , @xmath47 , @xmath48 and @xmath49 , @xmath50 and @xmath51 being respectively titan s mean longitude and ascending node in the frame @xmath20 . in paper
i , henrard & schwanen s model ( i.e. , titan moving on a circular orbit with a constant inclination and a constant precession of the nodes ) has been used to obtain the `` obliquity '' at the equilibrium @xmath52 and the three fundamental periods of the free librations around the cassini state .
then a numerical integration has been performed over 9000 years with complete ephemeris for titan ( tass1.6 , vienne & duriez 1995 @xcite ) .comparison between the analytical and numerical results with @xmath53 , adapted from paper i. the differences are given here relative to the numerical values , but relative to the analytical ones in paper i. [ cols="<,^,^,^",options="header " , ] [ tab : comprambh ] tab.[tab : comprambh ] gives a comparison with previous studies for io and europa .
we can see a good agreement , except perhaps with the `` obliquity '' of io .
i think that the difference of about @xmath54 can be explained by the fact that lainey s ephemeris have been updated between henrard s study and this paper .
in particular , henrard used @xmath55 rad as given in lainey s phd thesis ( 2002 @xcite ) , which is about @xmath56 higher than the value used in this study .
these data have been computed assuming that the satellites are in hydrostatic equilibrium , especially for the values of @xmath57 .
since the values of the periods of libration depend on @xmath57 , observing these librations should help estimate this parameter and test the hypothesis of hydrostatic equilibrium .
unfortunately , the free librations should have been damped by dissipative effects , unless the amplitude associated is being forced by a resonance ( as could be the case for titan ) .
but no resonance appears clearly in tab.[tab : frekgalrhea ] ( perhaps except between ganymede s wobble @xmath58 and the node of europa , with a regression period of 30.22 years ) .
the general analytical model given in this paper permits a first 3-dimensional description of the rotation of every synchronous body .
more particularly , it gives the equilibrium position , the three fundamental frequencies of the free librations , and the main terms of the free solution .
it is applied successfully to titan , io and europa , in the sense that the results are consistent with the previous studies .
moreover , this paper presents a study of a possible resonance involving titan s wobble . with it ,
i show that if the capture into resonance occured , then the wobble could currently have a forced amplitude of several degrees , and so induce a significant internal tidal dissipation .
moreover , the forcing of the wobble could falsify the space - based detection of titan s period of rotation to appear a little faster than it actually is , in particular because titan s rotation axis would be significantly different from its pole axis of figure .
cassini results on titan s rotation should give some information about that .
even if the system is not in resonance , a quasi - resonant state may have a detectable effect on the free libration w. i am indebted to several colleagues for fruitful discussions , especially nicolas rambaux , anne lematre , jacques henrard , alain vienne and michael efroimsky .
i also warmly thank the two referees , a. dobrovolskis and an anonymous one , who pointed out a mistake in a previous version of this paper .
this work benefited from the financial support of a fundp postdoctoral research grant .
i here detail the hamiltonian given by ( eq.[eq : t2 ] ) , i.e. seidelmann p.k . ,
abalakin v.k . ,
bursa m. et al . , report of the iau / iag working group on cartographic coordinates and rotational elements of the planets and satellites : 2000 , cel .
astr . , 82 , 83 - 111 ( 2002 ) | in noyelles et al . ( 2008 , astron . astrophys . , 478 , 959 - 970 ) ,
a resonance involving the wobble of titan is hinted at .
this paper studies this scenario and its consequences .
the first step is to build an accurate analytical model that would help to find the likely resonances in the rotation of every synchronous body . in this model ,
i take the orbital eccentricity of the body into account , as well as its variable inclination with respect to saturn s equator .
then an analytical study using the second fundamental model of the resonance is performed to study the resonance of interest .
finally , i study the dissipative consequences of this resonance .
i find that this resonance may have increased the wobble of titan by several degrees .
for instance , if titan s polar momentum @xmath0 is equal to @xmath1 ( @xmath2 and @xmath3 being respectively titan s mass and radius ) , the wobble might be forced to 41 degrees .
thanks to an original formula , i find that the dissipation associated with the forced wobble might not be negligible compared to the contribution of the eccentricity .
i also suspect that , due to the forced wobble , titan s period of rotation may be somewhat underestimated by observers .
finally , i use the analytical model presented in this paper to compute the periods of the free librations of the four galilean satellites as well as the saturnian satellite rhea . for io and europa ,
the results are consistent with previous studies . for the other satellites , the periods of the free librations are respectively @xmath4 d , @xmath5 y and @xmath6 y for ganymede , @xmath7 y , @xmath8 y and @xmath9 y for callisto , and @xmath10 d , @xmath11 y and @xmath12 y for rhea .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
we adopt the method as in @xcite to calculate the singlet yield . for the self - completeness , here we present a simple outline of this method .
the hamiltonian for the system ( two electron spins , one of which is coupled with several surrounding nuclear spins ) is as follows @xmath117 in our calculations of the main text , we have neglected the zeeman interactions between the nuclear spins and the external magnetic field .
we have included these interactions , and verified that the induced difference is very small ( as the gyromagnetic ratio for a nucleus h and n is much smaller than @xmath118 ) .
the numbers of hyperfine couplings are take from ref.@xcite .
we calculate the singlet yield as @xcite @xmath119 where @xmath16 is the radical re - encounter probability distribution , and @xmath120 is the singlet fidelity for the electron spin state @xmath121 at time @xmath19 .
( [ expmodel ] ) can be obtained from the conventional haberkorn approach @xcite in the case that the singlet and triplet recombination rates are the same , i.e. @xmath14 . the singlet yield is calculated following the method in @xcite .
we first write the singlet fidelity as @xmath122\\ & = & \sum\limits_{m}\sum\limits_{n } \langle m|(\rho_{0}\bigotimes_{j } \frac{\mathbb{i}_{j}}{d_{j}})|n\rangle \cdot \langle n| ( |s\rangle \langle s| \bigotimes_{j } \mathbb{i}_{j})|m\rangle \cdot e^{-i(\omega_{m}-\omega_{n})t}\end{aligned}\ ] ] where @xmath123 is the initial state of the radical pair , and the initial state of the nuclear spins at room temperature can be approximated as @xmath124 , where @xmath11 is the dimension of the @xmath12th nuclear spin , and @xmath125 and @xmath126 denote the eigen states of the hamiltonian @xmath127 in eq.(1 ) .
after some calculations , we have the singlet yield as @xmath128 where @xmath129 , @xmath130 and @xmath131 . in the main text
, we consider the radical pair reaction with the same singlet and triplet recombination rate , i.e. @xmath14 .
the method we use to calculate the singlet yield is based on the haberkorn approach @xcite that describes the recombination of radical pairs @xmath132-\frac{k_{s}}{2}(q_{s}\rho+\rho
q_{s})-\frac{k_{t}}{2}(q_{t}\rho+\rho q_{t } ) \label{hab}\ ] ] where @xmath133 and @xmath134 are the projection operators for the singlet and triplet electronic states of the radical pair .
there are alternative master equations based on quantum measurement that have been proposed to describe the recombination of radical pairs @xcite . under the condition @xmath14 , these master equations @xcite can be written in the following form @xmath132-(k_{s}+k_{t})\rho+k_{s}q_{t}\rho q_{t}+k_{t}q_{s}\rho q_{s } \label{jh}\ ] ] from eq.([hab ] ) or eq.([jh ] )
, one can obtain the density matrix of the radical pair state @xmath121 at time @xmath19 and thus calculate the singlet yield as @xmath135 dt\ ] ] we compare the results of the singlet yield from these two approaches .
for the simplicity of calculation , we take the three most significant hyperfine interactions in fadh@xmath21 , i.e. those for the nitrogens n5 and n10 and the proton h5 @xcite .
it can be seem from fig .
[ compsinyvf ] ( for the long lifetime @xmath136 ) and fig .
[ compsinyvfsl ] ( for the short lifetime @xmath137 ) that the difference between the results from two approaches is very small ( around @xmath138 ) .
we assume that the gradient field on the acceptor @xmath139 is much larger than the hyperfine couplings and the weak magnetic field @xmath1 . to calculate the singlet yield , for simplicity
, the hamiltonian can be approximated as @xcite @xmath140 we denote the density matrix of the initial radical pair state as @xmath56 , and use the eigen states of @xmath141 ( @xmath94 ) , namely @xmath142 and @xmath143 , as the spin basis of the acceptor .
the initial state can then be written as @xmath144 the singlet fidelity at time @xmath19 is @xmath145|s\rangle + \langle s|\left [ |u_{1}\rangle \langle u_{1}|\otimes ( u_{d } \rho^{11}_{d}u_{d}^{\dagger}\right]|s\rangle\\\nonumber & + & e^{i2\gamma_{e}tl_{a}}\langle s|\left[|u_{0}\rangle \langle u_{1}|\otimes ( u_{d } \rho^{01}_{d}u_{d}^{\dagger})\right]|s\rangle + e^{-i2\gamma_{e}t l_{a } } \langle s|\left [ |u_{1}\rangle \langle u_{0}|\otimes ( u_{d } \rho^{10}_{d}u_{d}^{\dagger})\right]|s\rangle\end{aligned}\ ] ] where @xmath146 .
if @xmath36 is very large , the last two terms ( second line ) in the above equation oscillate very fast and make no effective contribution to the singlet yield due to time average , thus the singlet yield will be @xmath147|s\rangle + \langle s|\left
[ |u_{1}\rangle \langle u_{1}|\otimes ( u_{d } \rho^{11}_{d}u_{d}^{\dagger}\right]|s\rangle\ } dt\\ & = & \frac{1}{2}\int_{0}^{\infty } f(t ) \cdot \mbox{tr } ( \rho^{00}_{d}u_{d}^{\dagger}|u_{1}\rangle \langle u_{1}|u_{d}+ \rho^{11}_{d}u_{d}^{\dagger}|u_{0}\rangle \langle u_{0}|u_{d})dt\\ & = & \int_{0}^{\infty } f(t ) \cdot \mbox{tr}[(\frac{i}{4}-\frac{\hat{a}\otimes u_{d}^{\dagger } \hat{a}u_{d}}{4})\rho]dt\\ & = & \frac{1}{4}-\frac{1}{4}\langle \hat{a}\otimes \hat{v}\rangle\end{aligned}\ ] ] where @xmath93 and @xmath148 . by choosing @xmath0 in the direction of @xmath99 , @xmath100 , and @xmath101 , the corresponding operator @xmath102 will be @xmath103 , @xmath104 , @xmath105 respectively ( which are the pauli operators ) . as an example , we assume that @xmath149 , it can be seen that if we choose @xmath1 in the direction of @xmath99 , @xmath100 , and @xmath101 respectively , the corresponding operators of @xmath107 are @xmath150 , @xmath151 , @xmath105 , with @xmath152 $ ] and @xmath153 $ ] .
these operators are linear independent . from the singlet yields corresponding to these choices of @xmath108 and @xmath106
, we can have three independent equations , following which we can infer the spin correlations @xmath154 , @xmath155 , @xmath156 . in a similar way
, we can choose @xmath157 and obtain the spin correlations @xmath158 , @xmath159 , @xmath160 , and @xmath161 , @xmath162 , @xmath163 . | magnetic fields as weak as earth s may affect the outcome of certain photochemical reactions that go through a radical pair intermediate .
when the reaction environment is anisotropic , this phenomenon can form the basis of a chemical compass and has been proposed as a mechanism for animal magnetoreception . here , we demonstrate how to optimize the design of a chemical compass with a much better directional sensitivity simply by a gradient field , e.g. from a magnetic nanostructure .
we propose an experimental test of these predictions , and suggest design principles for a hybrid metallic - organic chemical compass .
in addition to the practical interest in designing a biomimetic weak magnetic field sensor , our result shows that gradient fields can server as powerful tools to probe spin correlations in radical pair reactions . _
introduction. _ recently , there has been increasing interest in quantum biology namely investigating quantum effects in chemical and biological systems , e.g. , light harvesting systems @xcite , avian compass @xcite and olfactory sense @xcite .
the main motivation is to understand how quantum coherence ( entanglement ) may be exploited for the accomplishment of biological functions . as a key step towards this goal , it is desirable to find tools that can detect quantum effects under ambient conditions .
the ultimate goal of practical interest in studying quantum biology is to learn from nature and design highly efficient devices that can mimic biological systems in order to complete important tasks , e.g. collecting solar energy and detecting weak magnetic field . as an example of quantum biology
, the radical pair mechanism is an intriguing hypothesis @xcite to explain the ability of some species to respond to weak magnetic fields @xcite , e.g. birds @xcite , fruit flies @xcite , and plants @xcite
. a magnetochemical compass could find applications in remote magnetometry , in magnetic mapping of microscopic or topographically complex materials , and in imaging through scattering media @xcite .
it was demonstrated that a synthetic donor - bridge - acceptor compass composed of a linked carotenoid ( c ) , porphyrin ( p ) , and fullerene ( f ) @xcite can work at low temperature ( 193 k ) .
it is surprising that such a triad molecule is the only known example that has been experimentally demonstrated to be sensitive to the geomagnetic field ( yet not at room temperature ) .
it is currently not known how one might construct a biomimetic or synthetic chemical compass that functions at ambient temperature . to be measured ( yellow ) and a strong magnetic gradient @xmath0 ( blue ) , due to e.g. a magnetic nanostructure .
the outcome of a reaction depends on the direction of the weak field @xmath1 .
right : the directions of @xmath1 and the gradient field at the location of the acceptor @xmath0 depicted in the molecular coordinate frame.,width=332 ] in this letter , we approach to the goals of studying quantum biology in the context of chemical compass by demonstrating that a suitably designed gradient field can significantly improve the performance of a model chemical compass ( apart from increasing the intersystem crossing rate @xcite ) , see fig . [ setup ] .
it also opens a possible route to probe spin correlations of radical pairs and thereby investigate the role of quantum effects in spin chemistry .
the gradient field is strong at the location of one spin , and approximately zero at the other .
such a field can be created in the vicinity of a hard ferromagnetic nanostructure @xcite , by applying a spatially uniform bias field that cancels the field of the nanostructure in a small region of space .
in essence , the strong gradient field at one spin can substitute for strong anisotropic hyperfine couplings required for a purely molecular compass .
this geometry provides a more significant anisotropy and thereby shows much larger directional sensitivity than does the conventional compass mechanism based only on anisotropic hyperfine couplings . without requiring extra nuclear spins , the present model can work merely with two electron spins and thereby much simplifies quantum simulation of a chemical compass with , e.g. quantum dots and nitrogen - vacancy centers in diamond . with more freedom to tune parameters in a better controllable environment , such kind of quantum simulations would be very helpful to understand the recombination process of radical pairs , in particular , whether and how quantum measurement and zeno effect take place @xcite .
_ chemical compass mechanism. _ many chemical processes involve a radical pair intermediate , in which each radical has an unpaired electron coupled to an external magnetic field and a few nuclei via the hamiltonian @xcite @xmath2 where @xmath3 is the electron gyromagnetic ratio , @xmath4 denote the hyperfine coupling tensors and @xmath5 , @xmath6 are the electron and nuclear spin operators respectively . in our model ,
the magnetic field consists of two parts : @xmath7 , where the directional information about @xmath1 is what one wants to infer from the radical pair reaction , and @xmath8 is the local gradient field applied to each radical and is independent of @xmath1 .
the spin relaxation and decoherence times resulting from the factors other than hyperfine interactions are assumed to be considerably longer than the radical pair lifetime @xcite , to maximize sensitivity to weak magnetic fields @xcite . in many photochemical processes ,
the radical pair is created in a spin - correlated electronic singlet state @xmath9 within the timescale of picoseconds .
the nuclear spins start at thermal equilibrium , which under ambient conditions leads to an approximate density matrix as @xmath10 , where @xmath11 is the dimension of the @xmath12th nuclear spin and @xmath13 is the identity matrix .
the zeeman splitting from a magnetic field @xmath1 as weak as the geomagnetic field is much smaller than the thermal energy at ambient temperature . nonetheless , the field can influence the non - equilibrium electron spin dynamics and thereby determine the ratio of the chemical product from the singlet or triplet recombination as long as the thermalization time is longer than the reaction time . in experiments
, one may measure different quantities that are dependent on the weak magnetic field @xmath1 . here
we consider a simple first - order recombination reaction of the singlet radical pairs .
we note that there is some controversy over how to describe the radical pair reactions ( see e.g. @xcite ) . nevertheless , the conventional phenomenological density matrix approach @xcite works well in most cases , in particular when the singlet and triplet recombination rates are the same ( i.e. @xmath14 ) @xcite .
we adopt this method and calculate the singlet yield as @xmath15 , where @xmath16 is the radical reencounter probability distribution , and @xmath17 is the singlet fidelity for the electron spin state @xmath18 at time @xmath19 .
the integration of @xmath20 was performed following the method in @xcite .
_ gradient enhancement of magnetic field sensitivity. _ we starts from an optimally designed hyperfine compass model , one radical has strong and anisotropic hyperfine interactions , and the other radical has no hyperfine couplings @xcite .
we arbitrarily choose to call the first radical the acceptor , a , and the second the donor , d , though nothing that follows depends on this designation .
ritz and coworkers proposed that the radical pair fadh@xmath21-o@xmath22 meets this criterion , and they further speculated that this radical pair may be responsible for the magnetoreception of european robins @xcite , see also @xcite . without loss of the essential physics , we take the hyperfine couplings ( @xmath23 g ) from fadh@xmath21-o@xmath22 @xcite for our calculations . as a function of the angle @xmath24 of the weak magnetic field @xmath1 ( @xmath25 g ) with different gradient field strengths on the acceptor ,
i.e. @xmath26 0 g ( red , @xmath27 ) , 20 g ( blue , @xmath28 ) , 40 g ( green , @xmath29 ) , 80 g ( purple , @xmath30 ) , while @xmath31 . the recombination rate @xmath32 .
( b ) visibility @xmath33 as a function of the radical pair lifetime @xmath34 .
the direction of the gradient field @xmath0 is set as @xmath35 .
the same values of @xmath36 are used as in ( a).,title="fig:",width=170 ] as a function of the angle @xmath24 of the weak magnetic field @xmath1 ( @xmath25 g ) with different gradient field strengths on the acceptor , i.e. @xmath26 0 g ( red , @xmath27 ) , 20 g ( blue , @xmath28 ) , 40 g ( green , @xmath29 ) , 80 g ( purple , @xmath30 ) , while @xmath31 . the recombination rate @xmath32 .
( b ) visibility @xmath33 as a function of the radical pair lifetime @xmath34 .
the direction of the gradient field @xmath0 is set as @xmath35 .
the same values of @xmath36 are used as in ( a).,title="fig:",width=166 ] we define the molecular frame as the coordinate system , and the weak magnetic field @xmath1 can be represented as @xmath37 , @xmath38 .
the gradient field induces different local fields on two radicals .
we assume that the gradient field on the acceptor radical is @xmath39 while @xmath40 for the donor radical .
the strength of the weak magnetic field to be detected is the same as the geomagnetic field , i.e. @xmath25 g. to demonstrate the basic idea , we first consider @xmath41 , and then generalize to arbitrary @xmath42 . in fig .
[ sinyvf ] ( a ) , we plot the singlet yield as a function of the angle @xmath24 of the weak magnetic field @xmath1 with different gradient field strengths @xmath260 g , 20 g , 40 g , 80 g on the acceptor . in the case of @xmath43 ,
the directionality comes only from hyperfine anisotropy .
the gradient field clearly enhances the amplitude of the direction - dependent component of the magnetic field effect ( mfe ) .
to quantify the directional sensitivity , we use the magnetic visibility defined as @xcite @xmath44 as the gradient field becomes larger , the sensitivity will increase and approach to a saturate best value .
fig .
[ sinyvf ] ( b ) shows that for long radical pair lifetimes , the visibility with the gradient field @xmath2640 g is almost twice the visibility without the gradient field .
usually , the radical pair lifetime should be very long ( microseconds ) to maximize the effect of weak magnetic field @xcite , and hence performance , of the chemical compass [ fig .
[ sinyvf ] ( b ) ] .
this requirement places a severe constraint on the chemistry ; in typical radical pair reactions the lifetime is less than 100 ns @xcite . by increasing the overall magnitude of the visibility
, gradient - enhancement broadens the range of candidate reactions for a chemical compass .
_ liquid crystal experiment. _ in a uniaxially oriented sample , the mfe is averaged over all values of the angle @xmath42 .
such a sample is prepared by , for instance , freezing the molecules in a nematic liquid crystal in the presence of a strong magnetic field @xcite .
the ensemble - averaged mfe depends on @xmath24 only and is characterized by @xmath45 it can be seen from fig .
[ avsbsin ] ( a ) that the enhancement of the sensitivity can still be observed with the average signal @xmath46 by choosing appropriate values of @xmath47 . to induce the gradient field as above
, one feasible way is to use magnetic nanostructures @xcite .
we model the nanocrystal as a uniformly magnetized sphere , in which case the external magnetic field is the same as that of a point dipole of magnetic moment @xmath48 located at the center of the sphere @xcite .
we denote the position relative to the center of the sphere by the vector @xmath49 , and assume that both @xmath49 and @xmath48 lie along the z - axis .
the magnetic field at @xmath49 is @xmath50 where @xmath51n@xmath52a@xmath53 is the permeability of free space , the magnetic moment @xmath54 with @xmath55 the specific magnetization , @xmath56 is the material density , @xmath57 is the volume of the particle and @xmath58 is its radius .
the parameters for the typical magnetic material f@xmath59o@xmath60 are @xmath61 a@xmath52m@xmath62 kg@xmath63 , @xmath64 kg@xmath52m@xmath65 @xcite . for molecules with a separation @xmath66 between two radicals a few nanometers @xcite ,
it is sufficient for a nanoparticle to induce a large local field imbalance ( @xmath67 g ) on the donor and acceptor .
for example , using a f@xmath59o@xmath60 nanoparticle with the radius @xmath68 nm , it is possible to induce the local field difference as large as @xmath69 g between the position @xmath70 nm and @xmath71 nm ( assuming @xmath72 nm ) . by generating an additional homogenous field to compensate the field at the position @xmath73
, we can effectively obtain the gradient field on the donor and acceptor molecule as @xmath74 g and @xmath75 g respectively . as a function of the angle @xmath47 of the gradient field @xmath76 g. the blue dashed curve represents the visibility without the gradient field .
( b ) ensemble average of the singlet yield @xmath77 in eq .
( [ eavsiny ] ) from a monte carlo simulation of @xmath78 samples as a function of the angle @xmath24 of the weak magnetic field @xmath1 .
the gradient field is @xmath79 g with @xmath800 ( purple , @xmath30 ) , while @xmath31 . the fluctuations of the local fields @xmath81 and @xmath36 are characterized by the 3-dimensional gaussian distributions with the variance @xmath82 g and @xmath83 g. for comparison , we plot the singlet yield with no gradient field ( red , @xmath27 ) , and the one with the gradient field @xmath79 g ( @xmath800 ) without fluctuations ( blue , @xmath29 ) . in both panels
the radical pair lifetime is @xmath84 and @xmath25 g.,title="fig:",width=166 ] as a function of the angle @xmath47 of the gradient field @xmath76 g. the blue dashed curve represents the visibility without the gradient field .
( b ) ensemble average of the singlet yield @xmath77 in eq .
( [ eavsiny ] ) from a monte carlo simulation of @xmath78 samples as a function of the angle @xmath24 of the weak magnetic field @xmath1 .
the gradient field is @xmath79 g with @xmath800 ( purple , @xmath30 ) , while @xmath31 .
the fluctuations of the local fields @xmath81 and @xmath36 are characterized by the 3-dimensional gaussian distributions with the variance @xmath82 g and @xmath83 g. for comparison , we plot the singlet yield with no gradient field ( red , @xmath27 ) , and the one with the gradient field @xmath79 g ( @xmath800 ) without fluctuations ( blue , @xmath29 ) . in both panels
the radical pair lifetime is @xmath84 and @xmath25 g.,title="fig:",width=166 ] to see whether the effect of the gradient field shown above can manifest with experimental imperfections , we take into account the fluctuations of @xmath0 and @xmath85 by modeling the fluctuation as the three - dimensional gaussian distribution @xmath86 ( @xmath87 ) with @xmath82 g and @xmath83 g. therefore , the ensemble average of @xmath88 in eq .
( [ asiny ] ) is @xmath89 where @xmath90 is the average singlet yield when the local fields on the acceptor and donor molecules are @xmath91 respectively .
we have used monte carlo simulations to calculate the above ensemble average in eq.([eavsiny ] ) . in fig .
[ avsbsin ] ( b ) , we see that the enhancement from the gradient field can still be observed . _ probe spin correlations in a chemical compass. _ besides the significant enhancement of the directional sensitivity offered by gradient fields , we now examine how they can provide new insights into the quantum dynamics of radical pair reactions . for the present model chemical compass , if the gradient field on the acceptor @xmath0 dominates over the hyperfine couplings and the weak magnetic field @xmath1 , the singlet yield can be written as @xcite @xmath92 where the expectation value is calculated over the initial state , and @xmath93 ( with @xmath94 the eigen states of @xmath95 ) , @xmath96 with @xmath97 and the average taken over time weighted by @xmath98 . by choosing @xmath0 in the direction of @xmath99 , @xmath100 , and @xmath101 , the corresponding operator @xmath102 will be @xmath103 , @xmath104 , @xmath105 respectively ( which are the pauli operators ) . moreover , for each @xmath102 , one can choose @xmath106 also in the direction of @xmath99 , @xmath100 , and @xmath101 such that the operators of @xmath107 ( as a linear combination of pauli operators ) are linear independent , see @xcite .
the singlet yields corresponding to these choices of @xmath108 and @xmath106 lead to nine independent equations , from which we can infer the spin correlations @xmath109 for the radical pair state , where @xmath110 @xmath103 , @xmath104 or @xmath105 . with these correlations , one may check whether the radical pair state violates bell inequalities @xcite ; or obtain lower entanglement bounds of the radical pair state , see ref .
@xcite .
the above idea can be extended to monitor the dynamics of spin correlations suppose one can switch on gradient fields during the reaction . as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=37 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=37 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=37 ] as a function of the angles ( @xmath24 , @xmath42 ) of the weak magnetic field @xmath1 with @xmath25 g. the gradient field on the acceptor is @xmath26 80 g , on the donor it is @xmath31 .
the angle of the gradient field @xmath47 with respect to the z - axis is @xmath111 ( upper ) , @xmath112 ( middle ) and @xmath113 ( lower ) .
the patterns of the singlet yield over ( @xmath24 , @xmath42 ) from the initial singlet ( left ) and classically correlated state ( right ) are quite similar for @xmath35 , but are very different for @xmath114 and @xmath113 .
the radical pair lifetime is chosen as @xmath84.,title="fig:",width=158 ] as an example , we show that gradient fields can distinguish the singlet and the classically correlated initial state @xmath115 . for systems where the radical pair lifetime is much longer than the decoherence time , the conventional hyperfine - mediated mfe does not strongly depend on the initial states and thus can not allow one to achieve this goal , see e.g. @xcite .
if the gradient field is along the z - axis , the singlet yields are quite similar for the singlet and the classically correlated state [ see fig .
[ vtf ] ( upper ) ] . however , if we vary the direction of the gradient field @xmath0 , then the visibility for the singlet state will be much larger than for the classically correlated initial state [ fig .
[ vtf ] ( middle , lower ) ] .
in particular , for the classically correlated state , the singlet yield is insensitive to the angles ( @xmath24 , @xmath42 ) for @xmath116 while for the singlet state the angular sensitivity persists [ fig .
[ vtf ] ( lower ) ] .
the difference originates from the essential boundary between classical and quantum correlation ( entanglement ) .
the large gradient field can be viewed as a measurement of the acceptor spin along @xmath0 : the singlet state demonstrates perfect anti - correlation of the spins for any direction of @xmath0 , while for the classically correlated state this is true only in a certain direction of @xmath0 ( i.e. the @xmath101 direction ) . _
summary. _ we have demonstrated that a gradient field can lead to a significant enhancement of the performance of a chemical compass .
the gradient field also provides us a powerful tool to investigate quantum dynamics of radical pair reactions in spin chemistry . in particular
, it can distinguish whether the initial radical pair state is in the entangled singlet state or in the classically correlated state , even in the scenarios where such a goal could not be achieved before .
these phenomena persist upon addition of partial orientational averaging and addition of realistic magnetic noise .
the effects predicted here may be detectable in a hybrid system compass composed of magnetic nanoparticles and radical pairs in an oriented liquid crystalline host .
our work offers a simple method to design / simulate a biologically inspired weak magnetic field sensor based on the radical pair mechanism with a high sensitivity that may work at room temperature . _
acknowledgements. _ we are in debt to nan yang and adam cohen for valuable suggestions and beneficial communications .
we thank kiminori maeda , gian giacomo guerreschi and otfried ghne for helpful discussions , and hans briegel for continuous support in this work .
the work is supported by fwf ( sfb foqus ) .
99 g. s. engel , _ et al .
_ nature * 446 * , 782 ( 2007 ) ; h. lee , _ et al .
_ science * 316 * , 1462 ( 2007 ) ; e. collini and g. d. scholes , science * 323 * , 369 ( 2009 ) ; m. mohseni , _ et al .
_ j. chem .
phys . * 129 * , 174106 ( 2008 ) ; m. b. plenio , s. f. huelga , new j. phys .
* 10 * , 113019 ( 2008 ) ; f. caruso , a. w. chin , a. datta , s. f. huelga , and m. b. plenio , j. chem . phys .
* 131 * , 105106 ( 2009 ) ; m. sarovar , a. ishizaki , g. r. fleming , k. b. whaley , nature physics * 6 * , 462 ( 2010 ) .
j .- m .
cai , g. g. guerreschi , and h. j. briegel , _ phys .
rev .
lett . _ * 104 * , 220502 ( 2010 ) arxiv : 0906.2383 ( 2009 ) .
e. gauger , e. rieper , j. j. l. morton , s. c. benjamin , v. vedral , phys .
rev .
lett .
106 , 040503 ( 2011 ) .
i. k. kominis , _ phys .
rev .
e _ * 80 * , 056115 ( 2009 ) .
j. a. jones , p. j. hore , _ chem .
phys .
lett .
_ * 488 * , 90 ( 2010 ) .
l. turin , j. theor .
biol .
216(3 ) , 367 ( 2002 ) ; j. c. brookes , f. hartoutsiou , a. p. horsfield , a. m. stoneham , phys . rev .
lett . * 98 * , 038101 ( 2007 ) .
k. schulten , c. e. swenberg and a. weller , _ z. phys
. chem _ * nf111 * , 1 ( 1978 ) .
r. wiltschko and w. wiltschko , _ bioessays _ * 28 * , 157 ( 2006 ) . s. johnsen and k. j. lohmann , _ nature rev .
neurosci _ * 6 * , 703 ( 2005 ) . c. t. rodgers and p. j. hore , _ proc .
natl .
acad .
sci _ * 106 * , 353 ( 2009 ) .
t. ritz , s. adem and k. schulten , _ biophys .
j _ * 78 * , 707 ( 2000 ) .
w. wiltschko and r. wiltschko , _ j. exp .
biol _ * 204 * , 3295 ( 2001 ) .
t. ritz _ et .
al , _ _ nature _ * 429 * , 177 ( 2004 ) .
r. j. gegear , a. casselman , s. waddell and s. m. reppert , _ nature _ * 454 * , 1014 ( 2008 ) .
r. j. gegear , l. e. foley , a. casselman and s. m. reppert , _ nature _ * 463 * , 804 ( 2010 ) .
m. ahmad , p. galland , t. ritz , r. wiltschko and w. wiltschko , planta , * 225 * , 615 ( 2007 ) .
n. yang , y. tang , and a. e. cohen , _ nano today _ , * 4 * , 269 ( 2009 ) .
k. maeda _ et .
al , _ _ nature _ * 453 * , 387 ( 2008 ) .
a. e. cohen , _ j. phys
. chem .
a _ * 113 * , 11084 ( 2009 ) .
u. e. steiner , t. ulrich , _ chem .
rev _ * 89 * , 51 ( 1989 ) .
t. ritz _
et al .
_ , _ biophys .
j _ * 96 * , 3451 ( 2009 ) .
k. l. ivanov , m. v. petrova , n. n. lukzen and k. maeda , j. phys .
chem .
a , * 114 * , 9447(2010 ) .
a. i. shushin , j. chem . phys . * 133 * , 044505 ( 2010 ) . in this case , the singlet yields calculated from different approaches are expected to be very similar @xcite , see also @xcite for a comparison with the present model .
b. brocklehurst , _ j. chem
. sot .
faraday trans . _ * 72 * , 1869 ( 1976 ) .
j .- m .
cai , the calculation details are presented in the following appendix .
i. a. solovyov , k. schulten , _ biophys .
j _ * 96 * , 4804 ( 2009 ) .
f. cintolesi , t. ritz , c. w. m. kay , c. r. timmel , p. j. hore , _ chem .
phys . _ * 294 * , 385 ( 2003 ) .
d. j. griffiths , _ introduction to electrodynamics _ , 3rd edition , prentice hall inc .
( new jersey 1999 ) .
m. di valentin , a. bisol , g. agostini and d. carbonera , _ j. chem . inf . model . _ * 45 * , 1580 - 1588 ( 2005 ) .
r. horodecki , phys .
lett .
a. * 210 * , 223 ( 1996 ) .
k. m. r. audenaert , m. b. plenio , new j. phys . * 8 * , 266 ( 2006 ) . |
chiral perturbation theory @xcite provides a systematic low energy expansion of the qcd green functions in the sector of light mesons .
calculations up to two loops were done recently for the scattering amplitudes and the electroweak form factors in the chiral @xmath2 limit @xcite-@xcite .
a full @xmath3 @xmath4 calculation was also performed @xcite for a specific combination of form factors which does not involve arbitrary renormalization constants .
the result allowed to estimate the symmetry breaking corrections to a low energy theorem proposed by sirlin @xcite , which generalizes the ademollo - gatto theorem @xcite .
the fundamental properties of causality and unitarity are important ingredients in chpt .
dispersion relations for scattering amplitudes and form factors or their inverses were used in performing low order calculations @xcite , @xcite , or as a method of effective resummation of the high order terms @xcite , @xcite . in the present paper
we shall consider an alternative dispersive method suitable especially for form factors @xcite,@xcite , which uses as input an information about the modulus along the unitarity cut , leading to constraints on the values inside the analyticity domain .
the technique was for the first time combined with the ope expansion of a qcd polarization function in @xcite .
recently , this formalism was successfully applied to the form factors describing the weak semileptonic decays of heavy mesons @xcite-@xcite .
quite strong constraints on the shape of these form factors near zero recoil have been obtained by combining the dispersive formalism with the predictions of the heavy quark effective theory ( hqet ) @xcite , @xcite .
the predictive power of the technique was considerably increased @xcite , @xcite , @xcite by the simultaneous treatment of several form factors with different unitarity thresholds , connected among them near zero recoil by hqet .
it is of interest to apply this formalism to the form factors of the light pseudoscalar mesons , for which chiral symmetry predicts definite correlations for certain kinematical points .
a first investigation of the weak form factors of the @xmath5 decay in this framework was performed in @xcite .
similar techniques were applied in @xcite in order to parametrize the modulus of the pion electromagnetic form factor in the time like region . in the present paper
we perform an analysis of the electromagnetic and weak form factors of the @xmath0 and @xmath1 mesons , with emphasis on their low energy expansions .
the purpose of the investigation is to see whether the dispersive approach is useful for testing the rigorous predictions of chiral symmetry and for constraining the free parameters of chpt . to this end
we present generalizations of the standard technique suitable for incorporating additional information provided by experiments or chpt .
the paper is organized as follows . in the next section we review the dispersive formalism : first we present the form factors of interest and some of their properties , then we describe the physical input of the method and the standard mathematical techniques used for optimally exploiting this input . in section 3 we illustrate the simultaneous treatment of several form factors with different unitarity branch points by performing a test of a low energy theorem of chpt . in section 4 we derive constraints on the taylor coefficients at zero momentum of the pion electromagnetic form factor .
we first give the simple unitarity bounds obtained from the standard formalism and then show how they are improved using additional information on the phase and the modulus of the form factor along a part of the unitarity cut . in the last section we present some conclusions .
we consider the electromagnetic form factors of the @xmath0 and @xmath1 mesons , defined as @xmath6 and the form factors describing the weak semileptonic decay @xmath7 : @xmath8 where @xmath9 , @xmath10 and @xmath11 ( @xmath12 ) is the electromagnetic ( weak ) current : @xmath13 the form factors defined in ( [ defe ] ) and ( [ defw ] ) are analytic functions of real type in the complex plane @xmath14 cut along the real axis from the threshold of two particle production to infinity .
the cut starts at @xmath15 for the electromagnetic form factors @xmath16 and @xmath17 , and at @xmath18 in the case of the weak form factors @xmath19 and @xmath20 . at @xmath21 the conservation of the electromagnetic current and the ademollo - gatto theorem @xcite
give @xmath22 if we define the function @xmath23from ( [ norm ] ) it follows that @xmath24 this relation is valid up to terms quadratic in the chiral symmetry breaking parameters @xcite .
the ademollo - gatto theorem was extended by sirlin @xcite to values @xmath25 near the origin .
sirlin s theorem requires in particular the vanishing of the derivatives of the function @xmath26 at @xmath21 : @xmath27 in chpt the renormalization constants cancel in the combination of the form factors entering sirlin s function .
a recent @xmath28 calculation in full chiral @xmath4 perturbation theory gave the value @xcite @xmath29 where @xmath30 is the `` charge radius '' of the sirlin s form factor @xcite .
this result shows explicitely the breaking symmetry correction to sirlin theorem ( [ sirlin1 ] ) .
dispersive bounds on the above form factors are obtained by considering the vacuum polarization tensor : @xmath31 where @xmath32 denotes either the electromagnetic @xmath11 or the weak @xmath33 current ( the function @xmath34 vanishes in the electromagnetic case ) . from the asymptotic behaviour of qcd
it follows that the derivative @xmath35 of the amplitude @xmath36 satisfies the dispersion relation @xmath37 with the spectral function @xmath38 given by hadronic unitarity . using the definitions ( [ defe ] ) of the electromagnetic form factors and taking into account the positivity @xmath39
, we obtain the inequality : @xmath40\theta(t - t_k)\,,\end{aligned}\ ] ] where @xmath41 and @xmath42 are unitarity branch points . by inserting the inequality ( [ unitelm ] ) in the dispersion relation ( [ disprel ] ) for an euclidian point @xmath43
we obtain : @xmath44 \ , .
\end{aligned}\ ] ] a relation similar to ( [ condelm ] ) can be written for the polarization function @xmath45 of the weak current . using the definition ( [ defw ] ) we obtain from unitarity and positivity @xmath46 \theta(t - t_{\pi k})\,,\end{aligned}\ ] ] where @xmath47 , @xmath48 and @xmath49 is an isospin factor @xcite . by inserting ( [ unitw ] ) in a dispersion relation of the form ( [ disprel ] )
we obtain @xmath50 \ , .
\end{aligned}\ ] ] it is interesting to note that the inequalities ( [ condelm ] ) and ( [ condw ] ) can be added , leading to a single inequality @xmath51 \nonumber\\ & + & { \eta\over 48\pi^2}\int\limits_{t_{\pi k}}^\infty{{\rm d}t\over ( t+q^2)^2 } \left(1-{t_{\pi k}\over t}\right)^{1/2 } \left(1-{t_{\pi k}^{(-)}\over t}\right)^{1/2 } \nonumber \\ & & \times \left[\left(1-{t_{\pi k}\over t}\right ) \left(1-{t_{\pi k}^{(-)}\over t}\right ) \vert f_{\pi k}(t)\vert^2 + { \vert d_{\pi k}(t)\vert^2\over t^2}\right ] \,,\end{aligned}\ ] ] where @xmath52 in the euclidian region @xmath53 , the functions @xmath54 and @xmath55 can be calculated by applying renormalization group improved perturbative qcd , with nonperturbative corrections included by means of operator product expansions ( ope ) .
we used the expressions given in @xcite : @xmath56 \nonumber\\ & & -\left({1\over 6}+{3\over 4}\right){1\over \pi^2 q^4 } m_s^2(-q^2)+\left({1\over 18}+{1\over 12}\right){1\over q^6 } \langle{\alpha_s\over \pi}\bar g g\rangle + \left({10\over 9}+1\right ) { 1\over q^6 } m_q\langle \bar\psi\psi\rangle\ , , \end{aligned}\ ] ] corresponding to @xmath57 flavours .
the first and the second numerical constant in front of each term indicate the separate contribution of @xmath58 and @xmath59 , respectively . to evaluate these quantities we used the two - loop expressions of the running coupling @xmath60 and the running mass @xmath61 of the @xmath62-quark @xcite , the perturbative parameters @xmath63 @xcite and the condensates @xmath64 and @xmath65 @xcite . as concerns the value of the euclidian point @xmath66 at which we calculate the polarization amplitude , it must be on one hand high enough to ensure the validity of the ope expansion , and on the other hand small enough to provide a strong constraint on the form factors through the dispersion relation ( [ disprel ] ) .
in fact , the results turn out to be rather stable when @xmath66 is varied in the range @xmath67 .
the results reported in this work are obtained with the choice @xmath68 , for which @xmath69 and @xmath70 . with the l.h.s . of the relations ( [ condelm ] ) , ( [ condw ] ) or ( [ condit ] ) known , these inequalities are integral conditions for the sum of the moduli squared of the corresponding form factors .
it is useful to point out that an alternative integral condition for the electromagnetic form factors can be obtained by using as input , instead of ope in the euclidian region , a lower bound on the hadronic muon anomaly .
this method was first proposed in @xcite , and starts from the the vacuum polarization contribution of the hadronic part of the muon anomaly @xmath71 where @xmath72 by introducing ( [ amu ] ) in the unitarity relation ( [ unitelm ] ) we obtain the inequality @xmath73{\rm d}t\,,\end{aligned}\ ] ] which restricts the electromagnetic form factors along the unitarity cut . in the calculations we adopted the conservative lower bound @xmath74 @xcite , based on the experimental value , the weak radiative corrections and an estimate of the hadronic box diagram . in this subsection
we briefly review the standard techniques for exploiting in the optimal way the conditions ( [ condelm ] ) , ( [ condw ] ) , ( [ condit ] ) or ( [ condit1 ] ) in order to obtain constraints on the form factors and their derivatives inside the analyticity domain .
it is useful to introduce a compact notation , defining the form factors @xmath75 as @xmath76 and denoting by @xmath77 the unitarity branch points : @xmath78 we now apply to each integral with the lower limit @xmath77 appearing in the conditions ( [ condelm ] ) , ( [ condw ] ) , ( [ condit ] ) or ( [ condit1 ] ) the following conformal mapping @xmath79 by this transformation the complex @xmath80 plane is mapped onto the unit disk in the complex plane @xmath81 , such that @xmath82 and the unitarity cut @xmath83 becomes the boundary @xmath84 . for simplicity
we denote @xmath85 , where @xmath86 is the inverse of the mapping @xmath87 .
then the conditions ( [ condelm ] ) , ( [ condw ] ) , ( [ condit ] ) or ( [ condit1 ] ) can be written in the canonical form : @xmath88 where @xmath89 or @xmath90 . in eq.([l2norm ] ) @xmath91 are analytic functions without zeros inside the unit disk , their modulus square on the boundary are equal to the weight functions appearing in front of the form factors , multiplied by the jacobian of the conformal mapping ( [ mapp ] ) . in mathematical books
these functions are called `` outer functions '' and are defined in terms of their modulus on the boundary as @xcite @xmath92\ , , \quad\zeta={\rm exp}(i\theta)\,.\ ] ] in our case the functions @xmath91 have simple explicit expressions @xcite-@xcite . for the global condition ( [ condit ] ) which involves all the form factors , the outer functions @xmath91 are @xmath93 @xmath94 where @xmath95 and @xmath96 for the alternative condition ( [ condit1 ] ) the functions @xmath91 are replaced by : @xmath97 where @xmath98 is an outer function defined as in ( [ wouter ] ) in terms of its modulus @xmath99 on the boundary .
the inequality ( [ l2norm ] ) is a standard boundary condition in @xmath100 norm for the functions @xmath101 @xcite , from which one obtains constraints on the size and shape of these functions at interior points .
the results are expected to be stronger if one keeps in ( [ l2norm ] ) the contributions of all the functions @xmath101 and exploits in addition the correlations among them provided by symmetries at some kinematical points .
this technique proved to be very useful in the case of the weak form factors of heavy mesons , using hqet near zero recoil . in principle , a difficulty in applying the formalism to the excited states is the presence of the unphysical cuts below the unitarity branch points . in the case of heavy mesons ,
the unphysical cuts are well approximated by a few narrow resonances of known masses , which can be treated by adequate techniques with no assumption about the residua @xcite-@xcite .
as we shall see in the next section , the situation is more complicated in the case of the light mesons .
in applying the relation ( [ l2norm ] ) we recall that the functions @xmath102 and @xmath103 , _ i.e. _ the electromagnetic form factor of the pion and the weak form factors are analytic below their unitarity branch points , while the kaon electromagnetic form factors , denoted here as @xmath104 and @xmath105 , have an unphysical cut along the region @xmath106 , below the unitarity threshold of the @xmath107 production . in the variable @xmath81
this cut is placed inside the unit disk @xmath108 , along the segment @xmath109 , where @xmath110 denotes the image of the two pion production threshold by the conformal transformation ( [ mapp ] ) for the kaon form factors : @xmath111 the dispersive formalism can be applied to this case only if some assumptions about the unphysical cut are adopted .
for instance , if the phase of the kaon form factors along the unphysical cut is supposed to be known , the problem can be treated by means of a suitable modification of the standard techniques .
then one can use the inequalities ( [ condelm ] ) or ( [ condit ] ) to correlate the low energy expansions of the electromagnetic and weak form factors or to test some predictions of chiral symmetry .
in this section we will illustrate the procedure by testing the consistency of the relations ( [ sirlin1 ] ) proposed by sirlin @xcite , with the requirements of unitarity and perturbative qcd introduced in the condition ( [ condit ] ) .
we adopt for the phase of the kaon electromagnetic form factors a model similar to the one proposed in @xcite for the pion form factor , and generalized to include a narrow @xmath112 resonance , with the su(3 ) relation @xmath113 between the residua .
we assumed therefore that the phases of the form factors @xmath114 and @xmath115 along the unphysical region are @xmath116\,,\quad t_\pi < t
< t_k\,\ ] ] and @xmath117\,,\quad t_\pi < t < t_k\,.\ ] ] we used the @xmath118 width @xmath119 given by the resonance chiral effective theory @xcite , @xcite : @xmath120 where @xmath121 is the pion decay constant .
we shall use now the information on the phase in order to remove the unphysical cut of the kaon form factors . to this end
we shall use a so - called omns function @xcite , which exactly compensates the known phase of the form factors along this part of the cut , leaving us with functions analytic below the unitarity threshold . in what follows
we describe briefly the method .
since we treat all the form factors simultaneously it is convenient to introduce a collective notation , defining @xmath122\,,\quad i=2,3\ , , \end{aligned}\ ] ] where the functions @xmath123 are defined as @xmath124 for @xmath125 , and are extended for @xmath126 as lipschitz continuous functions @xcite .
of course , there is a large arbitrariness in such an extension , but the results are independent of the particular values of @xmath123 for @xmath126 .
the reason is that , being analytic and without zeros in the @xmath80-plane cut along @xmath127 , and with regular values on the boundary , @xmath128 and @xmath129 are outer functions , so by multiplying a class of functions with them , the class is not changed @xcite . by construction ,
the phase of @xmath130 coincides with @xmath123 .
we also use the notation @xmath131 to denote these functions in terms of the variable @xmath81 defined in ( [ mapp ] ) .
the functions @xmath131 are analytic in @xmath132 , excepted for a cut along the segment @xmath133 , where the phase of @xmath134 is equal to @xmath135 ( we recall that the conformal transformation ( [ mapp ] ) maps the upper half @xmath80-plane onto the lower semidisk in the @xmath81-plane ) .
we define the set of functions @xmath136 as @xmath137^{-1}\,,\quad i=1, .. \,,\,5\,,\ ] ] where @xmath101 are the form factors defined in ( [ notation ] ) . from ( [ f_i ] )
it is easy to see that the complex phases of the form factors @xmath138 and @xmath139 along the cut @xmath140 are compensated by the phases of the omns functions @xmath131 .
therefore , all the functions @xmath136 are analytic of real type inside the unit disk @xmath132 .
by introducing ( [ f_i ] ) in the @xmath100 norm condition ( [ l2norm ] ) we have @xmath141 where @xmath91 are the outer functions defined in ( [ w123 ] ) and ( [ w45 ] ) , and @xmath142 are additional outer functions which satisfy the relations @xmath143 they can be calculated in terms of their modulus on the boundary using the standard formula @xcite given in ( [ wouter ] ) , which is equivalent to @xmath144\,,\quad i=1, .... 5\,.\ ] ] it is convenient to introduce the new functions @xmath145 as @xmath146 where @xmath147^{-1}\,.\ ] ] now , in terms of @xmath145 the inequality ( [ l2norm1 ] ) takes the canonical form @xmath148 by construction the functions @xmath145 are analytic inside the disk @xmath132 and can be expanded as power series @xmath149 with real coefficients @xmath150 . by inserting these expansions in ( [ l2norm2 ] )
we obtain @xmath151 as we discussed in section 2 , the generalization of the ademollo - gatto theorem proposed by sirlin requires the vanishing of the first derivatives at @xmath21 of the function @xmath26 defined in ( [ delta ] ) , in the exact chiral limit . in order to test the consistency of this condition ,
let us put @xmath152 where the derivatives are with respect to @xmath80 , and @xmath153 are given numbers of the order of magnitude of chiral symmetry breaking @xcite . using ( [ g_i ] )
we write @xmath26 as @xmath154^{-1}\ , g_i(z)\,,\ ] ] where @xmath155 therefore the derivatives entering ( [ deltak ] ) become @xmath156^{-1 } \over { \rm d}t^{n}}\right]_{t=0}\ , \left[{{\rm d}^{k - n } g_i(z ) \over { \rm d}t^{k - n}}\right]_{t=0 } \,,\ ] ] and finally the conditions ( [ deltak ] ) can be written as @xmath157 where the numbers @xmath158 can be obtained in a straightforward way using eq .
( [ deltak1 ] ) , the definition ( [ omega_i ] ) of the functions @xmath159 and the connection ( [ mapp ] ) between the variables @xmath80 and @xmath81 .
of course , for each @xmath160 only the taylor coefficients @xmath161 with @xmath162 contribute to @xmath163 .
our objective is to test the consistency of the relations ( [ deltak ] ) with the inequality ( [ l2norm ] ) provided by the dispersive formalism .
the problem can be easily solved , since we expressed both these conditions in terms of the taylor coefficients @xmath161 , transforming in this way a functional problem into an algebraic one .
we notice that the first coefficients @xmath164 can be computed using the relations ( [ norm ] ) and ( [ omega_i ] ) .
the remaining coefficients @xmath161 , @xmath165 are free . the relations ( [ deltak2 ] ) can be viewed as a set of @xmath166 linear constraints for these coefficients , which satisfy in addition the quadratic condition ( [ l2norm3 ] ) . a simple way to test the consistency of these relations for a given set of numbers @xmath153
, @xmath167 is to define the quantity @xmath168 where the minimization is with respect to the coefficients @xmath161 , @xmath169 which satisfy the set of linear conditions ( [ deltak2 ] ) .
it is clear that , if the minimal value @xmath170 is greater than one , the conditions ( [ l2norm3 ] ) and ( [ deltak2 ] ) can not be satisfied simultaneously , while if @xmath170 is less than one the conditions are satisfied by at least one set of coefficients @xmath161 .
a more detailed reasoning along these lines @xcite shows that the inequality @xmath171 is a necessary and sufficient condition for the consistency of the relations ( [ l2norm3 ] ) and ( [ deltak2 ] ) . in order to solve the minimization problem ( [ mu_0 ] ) with the constraints ( [ deltak2 ] ) we apply the technique of lagrange multipliers .
the lagrangean of the problem is : @xmath172\,,\ ] ] where @xmath173 , @xmath174 are lagrange multipliers . in ( [ lagrange ] )
we indicated separately the coefficients @xmath175 which are known .
moreover , for convenience we extended the sum over @xmath176 in the last term up to @xmath177 , setting @xmath178 for @xmath179 .
the lagrangean @xmath180 is a convex function of the coefficients @xmath161 , its minimal value being given by the equations @xmath181 the solution of these equations is @xmath182 by introducing this solution in the conditions ( [ deltak ] ) we obtain the following system of equations for the lagrange multipliers @xmath173 : @xmath183 where the matrix @xmath184 is defined as @xmath185 and the numbers @xmath186 are related to the derivatives of the sirlin function @xmath153 defined in ( [ deltak ] ) by @xmath187 the system ( [ system ] ) has the solution @xmath188 where @xmath189 is the inverse of the matrix @xmath184 . by introducing these expressions in ( [ c_n ] )
we obtain the optimal coefficients @xmath161 for @xmath190 as @xmath191 with these coefficients for @xmath192 and @xmath193 for @xmath194 according to ( [ c_n ] ) we obtain after a straightforward calculation the following compact form the solution @xmath170 of the minimization problem ( [ mu_0 ] ) : @xmath195 and the consistency condition ( [ consis ] ) becomes @xmath196 we recall that in this inequality the coefficients @xmath175 are known according to ( [ c_0i ] ) from the normalization of the form factors at @xmath21 , and the matrix @xmath184 can be explicitely computed using ( [ u ] ) and the numbers @xmath158 appearing in ( [ deltak2 ] ) .
we mention that these numbers include the physical input on unitarity and qcd , contained in the outer functions @xmath91 , and the phase of the kaon form factors along the unphysical region entering the outer functions @xmath197 and @xmath198 . as for the quantities @xmath186 , they are related to the derivatives @xmath153 of the sirlin function , according to ( [ tildes_k ] ) . with this input the inequality ( [ consis1 ] )
is a constraint for the derivatives @xmath153 of the sirlin function @xmath26 at @xmath21 , defined in ( [ deltak ] ) . for illustration we evaluated the quantity @xmath170 for @xmath199 , using the value @xmath200 obtained recently in @xcite and several choices for the higher derivatives @xmath201 and @xmath202 .
a few results are listed in table 1 .
the sets of values for which @xmath170 is less than 1 are consistent with the dispersive bounds ( in particular , this is true for @xmath203 and small deviations from these values , showing that sirlin s theorem is verified within the present formalism ) .
we give also some pairs @xmath204 for which the quantity @xmath170 is greater than one , and therefore the inequality ( [ consis1 ] ) is violated .
[ tab2 ] .
the quantity @xmath170 defined in ( [ minnorm1 ] ) for @xmath200 @xcite and several values of the derivatives @xmath201 and @xmath202 of the sirlin function .
the values consistent with qcd and analyticity correspond to @xmath205 . [ cols="^,^,^ " , ] the results show an adequate consistency between the requirements of sirlin s theorem on one hand , and qcd and analyticity on the other .
however , in order to exploit the dispersive inequalities we had to adopt a model for the phase of the kaon form factors along the unphysical region .
the presence of the unphysical cut reduces therefore the model independence of the conclusions .
in this section we shall apply the dispersive formalism to the pion electromagnetic form factor @xmath206 .
we consider the taylor expansion of this function around the origin @xmath21
@xcite @xmath207 where @xmath208 is the radius of the charge distribution . in chpt
the calculation of the taylor coefficients requires the evaluation of higher pion loops , which introduce arbitrary renormalization constants @xcite , @xcite .
one - loop chpt @xcite predicts @xmath209 . at two - loop level
the coefficient @xmath210 can not be calculated as it depends on an arbitrary renormalization constant , and @xmath211 @xcite .
a fit of the aleph data @xcite on the hadronic @xmath212 decay rate with a gounaris - sakurai formula @xcite ( equivalent to the pad version of the one - loop chpt ) gives @xmath213 .
similar values @xmath214 were obtained in @xcite by usual dispersion relations .
other values proposed in the literature are : @xmath215 @xcite , and @xmath216 @xcite . in what follows we shall prove that the dispersive formalism imposes nontrivial constraints on the taylor expansion ( [ taylor ] ) , especially when combined with additional information about the form factor along a part of the time like region . for the present purpose we keep only the contribution of the pion form factor in the unitarity inequality ( [ unitelm ] ) , neglecting the positive terms due to the @xmath1 mesons . in this case the inequality ( [ l2norm ] ) becomes @xmath217 where the outer function @xmath218 is @xmath219 with @xmath220 defined in ( [ di ] ) .
if we use as an alternative input the lower bound on the hadronic contribution to the muon anomaly , the function @xmath218 is @xmath221 where @xmath98 was defined below ( [ wamu ] ) .
we consider now the function @xmath222 which is real analytic in the unit disk @xmath132 and can be expanded as @xmath223 with real coefficients @xmath224 .
then the condition ( [ l2pi ] ) becomes @xmath225 a constraint on the first @xmath166 coefficients appearing in the taylor expansion ( [ taylor ] ) is immediately obtained by noticing that ( [ l2pi1 ] ) implies @xmath226 the first @xmath166 coefficients @xmath227 can be expressed in a straightforward way in terms of the first @xmath166 taylor coefficients appearing in ( [ taylor ] ) .
the explicit relations are obtained easily using the definition ( [ g_pi ] ) and the conformal mapping ( [ mapp ] ) .
for instance , @xmath228 , @xmath229 etc , where the function @xmath218 is defined in ( [ wpi ] ) and the derivatives are with respect to @xmath81 .
if we use as input the lower bound on the muon anomaly , the function @xmath218 has the expression ( [ wamupi ] ) .
in particular , for @xmath230 and a fixed value of the charge radius @xmath231 the inequality ( [ l2pin ] ) defines an allowed domain in the plane of the coefficients @xmath210 and @xmath232 appearing in the taylor expansion ( [ taylor ] ) . in fig.1 , the interior of the larger ellipse is the domain obtained using the standard value @xmath233 and the qcd condition ( [ condit ] ) .
this domain is very large , but one can see a certain correlation among the values of the taylor coefficients . this feature becomes more stringent when higher derivatives of the form factor are taken into account .
for comparison we indicated also in fig .
1 the allowed domain obtained from the condition ( [ condit1 ] ) on the muon anomaly .
this domain is smaller , and in particular it excludes a pair of values for @xmath210 and @xmath232 proposed recently in @xcite . in the next subsections
we shall improve the above bounds by implementing informations about the phase and the modulus of the pion form factor along a part of the unitarity cut . according to watson theorem @xcite , along the elastic region @xmath234 the phase of the pion electromagnetic form factor coincides with the phase @xmath235 of the @xmath236 partial wave amplitude of the @xmath237 scattering .
this amplitude was calculated in the frame of chpt up to two loops @xcite .
the expansion upon chiral loops is assumed to describe correctly the phase up to @xmath238 . on the other hand ,
precise experimental data are available above @xmath239 @xcite .
new planned experiments on @xmath240 decay @xcite , @xcite , @xcite will provide accurate information about the low energy @xmath241 scattering amplitudes .
it is of interest to incorporate in the dispersive formalism for the pion form factor the additional information on the phase .
unlike the situation encountered in section 3 , where the knowledge of the phase allowed us to remove the unphysical cut below the unitarity threshold , in the present case we must implement the phase along a part of the unitarity cut .
we treat the problem adapting mathematical techniques used in @xcite-@xcite .
we assume that @xmath242=\delta_1 ^ 1(t)\,,\quad t_\pi\le t \le t_{in}\,,\ ] ] where @xmath243 is a known function and @xmath244 denotes the threshold of the inelastic cut or a certain point up to which the phase is supposed to be known . as we shall discuss below , the technique can be easily adapted to more general situations , for instance when the phase is given along the range @xmath245 with @xmath246 , or along a region consisting of two disjoint parts , @xmath247 and @xmath248 , with @xmath249 .
we first express the condition ( [ phase ] ) in the variable @xmath81 defined in ( [ mapp ] ) . to this end
we recall that the branch point @xmath250 is mapped onto the point @xmath251 , and the upper edge of the unitarity cut in the @xmath80 plane becomes the lower unit semicircle in the @xmath81 plane .
we denote by @xmath252 the image of the point @xmath253 in the @xmath81-plane , and let @xmath254 .
then the condition ( [ phase ] ) becomes @xmath255&=&-\delta_1 ^ 1(\theta)\,,\quad \theta_{in}\le \theta\le \pi\,,\nonumber\\ \lim_{r\to 1 } { \rm arg}[f_\pi(r{\rm e}^{i\theta})]&=&\delta_1 ^ 1 ( \theta)\,,\quad \pi\le \theta\le 2\pi-\theta_{in}\,,\end{aligned}\ ] ] where we denoted @xmath256 , with @xmath257 as follows from ( [ mapp ] ) .
we took into account the fact that the form factor is an analytic function of real type , which means that its phase is an odd function of @xmath258 , _
i.e. _ it satisfies the relation @xmath259 .
it is useful to note that the same property holds also for the imaginary part of the form factor , while the real part and the modulus are even functions of @xmath258 . in what follows
we shall derive the allowed domain for the taylor coefficients of the expansion ( [ taylor ] ) , taking into account the inequality ( [ l2pi ] ) provided by the dispersive formalism , and the additional relations ( [ phasez ] ) .
as usual , the information about the phase is implemented by means of an omns function @xcite .
we can use the definition given in ( [ o_i ] ) , or equivalently we define @xmath260\,,\ ] ] where @xmath261 is a lipschitz continuous function such that @xmath262 for @xmath263@xmath264 for @xmath265 , @xmath261 being arbitrary outside @xmath266 .
we shall see below that the results are independent on the choice of @xmath267 outside @xmath266 . using the plemelj - privalov relation @xcite @xmath268 where the last integral denotes the principal value , one can show that along the interval @xmath266 the phase of the function @xmath269 coincides with the phase of the form factor .
therefore , by multiplying @xmath270 with @xmath271^{-1}$ ] the phases compensate each other and the product is real along the elastic part of the cut . in the variable @xmath81
this condition has the form @xmath272 = 0\,,\quad \theta\in \gamma\,.\ ] ] by recalling the definition ( [ g_pi ] ) of the function @xmath273 and its power expansion ( [ g_piseries ] ) , we write the condition ( [ imag ] ) in the form @xmath274= \nonumber\\\sum\limits_{n=0}^\infty c_n\ , { \rm im}\lim_{r\to 1}\left[[w(r{\rm e}^{i\theta})]^{-1}\ , r^n\,{\rm e}^{i n\theta}\right ] = 0\,,\quad \theta\in \gamma\ , , \end{aligned}\ ] ] where @xmath275 is defined as @xmath276 the allowed domain of the taylor coefficients of the pion form factor , which satisfy the conditions ( [ l2pi ] ) and ( [ imag ] ) can be found , like in section 3 , by means of an optimization problem .
we first recall that the conditions ( [ l2pi ] ) and ( [ imag ] ) were written in the equivalent forms ( [ l2pi1 ] ) and ( [ imag1 ] ) respectively , in terms of the taylor coefficients @xmath227 .
we consider then the quantity @xmath277 where the minimization is with respect to the coefficients @xmath227 which satisfy the condition ( [ imag1 ] ) . as in section 3
one can show that the inequality ( [ consis ] ) is a necessary and sufficient condition for the consistency of the relations ( [ l2pi1 ] ) and ( [ imag1 ] ) .
we solve the constrained minimization problem ( [ mu_0pi ] ) with the generalized lagrange theory of optimization , based on hahn - banach theorem @xcite .
the lagrangean is @xmath278^{-1}\ , r^n{\rm e}^{i n\theta}\right]{\rm d}\theta \,,\ ] ] where the function @xmath279 is a generalized lagrange multiplyer @xcite .
we can assume without losing generality that it is an odd function , since its product with another odd function ( the imaginary part of an analytic function of real type ) is integrated along a symmetric interval @xmath266 .
we denoted for simplicity @xmath280 .
the numerical factor in front of the integral in ( [ l_pi ] ) and the modulus @xmath281 inside the integral were introduced explicitely for convenience .
we assume now that the first @xmath166 coefficients @xmath227 have prescribed values , and perform the minimization of the lagrangean with respect to the remaining coefficients @xmath227 , @xmath282 . the minimum condition @xmath283 gives the optimal coefficients @xmath284{\rm d}\theta
\,,\quad n\ge n+1\,.\ ] ] this expression can be written equivalently as @xmath285 where we added a term which has a vanishing contribution by parity arguments . in order to find the lagrange multiplier @xmath286 we introduce the optimal coefficients
@xmath227 in the constraint ( [ imag1 ] ) .
recalling that the first @xmath166 coefficients @xmath227 are assumed to have prescribed values , we write the condition ( [ imag1 ] ) in the form @xmath287
-{\rm im}\lim_{r\to 1}\left[{i\over \pi}{1\over w(\theta ) } \int\limits_\gamma \lambda(\theta')\ , { |w(\theta')|\over w^*(\theta ' ) } \,{{\rm e}^{i(n+1)(\theta-\theta')}\over 1-r { \rm e}^{i(\theta-\theta')}}\,{\rm d}\theta'\right]=0\,,\,\,\theta\in \gamma\,.\ ] ] it is useful to write @xmath288 where from ( [ w ] ) it follows that @xmath289 here @xmath290 is the phase of the outer function @xmath218 , and @xmath291 is connected to the phase of the pion form factor as explained below ( [ omnes ] ) . by inserting ( [ phasew ] ) in the relation ( [ inteq0 ] ) and applying the plemelj - privalov relation ( [ plemelj ] ) we obtain after a straightforward calculation the equation @xmath292-\lambda(\theta ) \nonumber\\ & & + { 1\over 2\pi } \int\limits_\gamma \lambda(\theta')\ , { \sin\left[\left(n+{1\over 2}\right)(\theta-\theta')-\phi(\theta ) + \phi(\theta')\right ] \over \sin \left[{\theta-\theta'\over 2}\right]}\,{\rm d}\theta'=0\,,\,\,\theta\in \gamma\ , , \end{aligned}\ ] ] where the last integral is defined as the principal value .
we obtained therefore a singular integral equation for the lagrange multiplier @xmath286 .
if the phase @xmath293 is lipschitz continuous , the equation is of fredholm type and can be solved by standard techniques . after solving this equation ,
the minimal norm @xmath170 can be computed by inserting in ( [ mu_0pi ] ) the prescribed values @xmath227 for @xmath294 and the optimal values ( [ c_npi1 ] ) of @xmath227 for @xmath282 .
taking into account the fact that @xmath224 we obtain @xmath295 by applying the plemelj - privalov relation ( [ plemelj ] ) and using the integral equation ( [ inteq1 ] ) satisfied by the function @xmath279 , we arrive finally at the expression @xmath296{\rm d}\theta\,.\ ] ] using the fact that @xmath279 and @xmath297 are odd functions , the integral equation ( [ inteq1 ] ) can be written in the form @xmath298-\lambda(\theta ) \nonumber\\ & & + { 1\over 2\pi } \int\limits_{\theta_{in}}^\pi \lambda(\theta')\ , \left[{\sin\left[\left(n+{1\over 2}\right)(\theta-\theta')-\phi(\theta ) + \phi(\theta')\right ] \over \sin \left[{\theta-\theta'\over 2}\right]}\right.\nonumber\\ & & -\left .
{ \sin\left[\left(n+{1\over 2}\right)(\theta+\theta'-2\pi)-\phi(\theta ) -\phi(\theta')\right ] \over \sin \left[{\theta+\theta'\over 2}-\pi\right]}\right]\ , { \rm d}\theta'=0\,,\quad \theta_{in}\leq\theta\leq \pi\,,\end{aligned}\ ] ] where @xmath299 moreover , the expression ( [ mu_0pi2 ] ) of @xmath170 can be written as @xmath300{\rm d}\theta\,.\ ] ] we notice that the first term in ( [ mu_0pi3 ] ) represents the unconstrained minimum , obtained from ( [ l2pin ] ) if the coefficients @xmath227 for @xmath282 were free .
the second term in the expression of @xmath170 is positive and represents the improvement brought by the knowledge of the phase of the form factor along the region @xmath301 .
it is important to emphasize that only the values of @xmath293 along the interval @xmath266 are required in the integral equation .
also , the modulus of the outer function does not appear ( it was absorbed in the definition of the lagrange multiplier @xmath279 ) , which means that the results are independent of the choice of the function @xmath302 outside the interval @xmath266 .
the equations ( [ inteq2 ] ) - ( [ mu_0pi3 ] ) provide a simple numerical procedure for finding the allowed domain of the coefficients of the taylor expansion ( [ taylor ] ) : one starts with a set of values for the first @xmath166 taylor coefficients . using the relations ( [ g_pi ] ) and ( [ g_piseries ] )
one finds the corresponding coefficients @xmath227 , which enter the integral equation ( [ inteq2 ] ) .
the solution @xmath279 of this equation is then used in ( [ mu_0pi3 ] ) to evaluate the quantity @xmath170 . recalling that the allowed domain of the taylor coefficients is described by the inequality ( [ consis ] ) , the values taken as input are accepted or rejected if @xmath170 is less or greater than unity , respectively . from the above derivation the generalization to the case where the region @xmath266 consists of two disjoint subintervals @xmath303 and @xmath304 is straightforward .
this case is of interest when using the phase given by chpt along @xmath305 and the experimental data on @xmath306 .
the resulting equations have the same form , the lagrange multiplier being defined on @xmath307 . to illustrate how the bounds on the taylor coefficients @xmath210 and @xmath232 are improved by the knowledge of the phase
, we took the expression @xcite @xmath308\,,\ ] ] where @xmath309 was defined in ( [ gamaro ] ) . at low energies above the threshold , this phase coincides with the one loop chpt expression @xcite @xmath310 while for @xmath311 it is in very good agreement with the experimental data @xcite .
we assumed that the phase of the pion form factor coincides with ( [ phaseunit ] ) along the region @xmath301 , with @xmath312 , which corresponds to @xmath313 .
as input for the bounds we use the ope expansion of the polarization amplitude , which led to the domain represented by the interior of the large ellipse in fig.1 . in this case the outer function @xmath218 is given in ( [ wpi ] ) and the function @xmath314 defined in ( [ phasew2 ] ) has the expression @xmath315 -\delta_1 ^ 1(\theta)\,,\quad \theta_{in}\leq \theta\leq \pi\,,\ ] ] with @xmath316 defined in ( [ di ] ) . in fig .
2 the interior of the ellipse is the allowed domain in the plane @xmath317 , for @xmath318 . compared with the large ellipse in fig.1
one can see the considerable improvement brought by the knowledge of the phase along a part of the unitarity cut . a more realistic calculation using chpt along an interval @xmath303 and the experimental data on @xmath235 along another interval @xmath304 is of interest and will provide precise model independent bounds on the taylor coefficients .
the modulus of the pion form factor along a part of the cut is known experimentally from the rate of @xmath319 annihilation into pions @xcite and the hadronic @xmath212 decay @xcite .
high precision data are available especially in the range @xmath320 . in this subsection
we derive the allowed domain of the taylor coefficients by including some information about the modulus of the pion form factor along a part of the cut .
it is a known mathematical fact that if the modulus and the phase of an analytic function are known exactly along a part of the boundary , then the function is in principle uniquely determined .
explicit formulas for recapturing an analytic function belonging to a certain class , from its restriction along a part of the boundary are available @xcite , @xcite ( see also @xcite for explicit expressions and further references ) .
these expressions are however very unstable numerically , reflecting the fact that the analytic continuation is an ill - posed problem @xcite .
more exactly , the formulas give the correct analytic continuation if the input values are known with infinite accuracy , but they lead to arbitrary predictions if these values are affected by errors . in what follows
we shall show that even a nonoptimal use of the input information about the phase and the modulus leads to a considerable improvement of the bounds on the taylor coefficients . unlike the case considered in the previous subsection , where the phase could be given along several disjoint intervals , the method described below requires the knowledge of the phase along the whole range @xmath321 .
we start by writing the inequality ( [ condelm ] ) in the form @xmath322 where @xmath323 assuming that the modulus @xmath324 is known along @xmath321 , the quantity @xmath325 can be evaluated numerically . to be conservative , one can use a lower bound on the modulus , which does not spoil the inequality ( [ condlb ] ) and leads to a larger allowed domain for the taylor coefficients . in order to incorporate the information about the phase we use the omns function defined in ( [ omnes ] ) , which can be written equivalently in the @xmath80 variable as @xmath326\,.\ ] ] the properties of the function @xmath327 were discussed below ( [ omnes ] ) , and we use it in order to remove the part of the cut below @xmath328 .
more precisely , writing as in ( [ f_i ] ) @xmath329 we see that the function @xmath330 is real on the real axis below @xmath244 , since the phase of the form factor along @xmath331 is compensated by the phase of @xmath332 . therefore @xmath330 is analytic in the @xmath80-plane cut for @xmath333 , and satisfies on the cut the condition @xmath334 which follows from ( [ condlb ] ) .
but this condition can be simply exploited with the standard techniques described in subsection [ s3 ] .
we first map the @xmath80 plane cut for @xmath333 onto the unit disk in the @xmath81-plane , using the conformal mapping @xmath335 we define as in ( [ omega1 ] ) the outer function @xmath336\,,\ ] ] such that @xmath337 for @xmath338 , and the outer function similar to ( [ wpi ] ) @xmath339 with @xmath340 by introducing a new function @xmath341 , defined as @xmath342^{-1 } f_\pi(z)\,,\ ] ] the inequality ( [ condlb1 ] ) takes the canonical form @xmath343 by construction the function @xmath341 is analytic in the disk @xmath132 and can be expanded as @xmath344 where the coefficients @xmath345 satisfy the inequality @xmath346 which follows from ( [ l2normpi3 ] ) . the first @xmath166 coefficients @xmath345 are connected in a straightfoward way through ( [ tildeg_pi ] ) to the first @xmath166 taylor coefficients of the form factor @xmath206 at @xmath21 .
therefore , the inequality @xmath347 which follows from ( [ l2pi3 ] ) , defines an allowed domain in the plane of these taylor coefficients . in principle
this domain can be further reduced : indeed , in addition to the boundary condition ( [ condlb1 ] ) , the function @xmath348 defined in ( [ f_pi ] ) has known error - affected values along the region @xmath321 , where it is holomorphic .
the inclusion of this additional information is not trivial , and we will not treat this problem here , using the simple inequality ( [ l2pi3n ] ) . in particular , for @xmath230 and @xmath208 fixed , this inequality gives the allowed domain of the parameters @xmath210 and @xmath232 of the expansion ( [ taylor ] ) . in a numerical application we used the same phase @xmath235 as in the preceeding subsection and the modulus @xmath349 from the aleph data @xcite . using @xmath350 and @xmath351 , we obtain for the quantity @xmath325 defined in ( [ m0 ] ) the estimate @xmath352 .
the result is not sensitive to the experimental uncertainties above the threshold @xmath250 , due to the phase space factor in the unitarity integral appearing in ( [ m0 ] ) . reducing the value of @xmath349 along @xmath321 by @xmath353 leads to an increase of @xmath325 by about @xmath354 and to bounds on the coefficients @xmath210 and @xmath232 weaker by about @xmath355 .
the interior of the ellipse shown in fig.3 indicates the allowed domain obtained in this conservative situation .
the combined information on the phase and the modulus is seen to restrict in an impressive way the values of the taylor coefficients of the pion form factor .
the allowed range of @xmath210 is @xmath356 , and for each @xmath210 the range of the values of @xmath232 is very narrow .
for instance , for @xmath357 the parameter @xmath232 is restricted to the interval @xmath358 .
the values @xmath359 adopted in @xcite are outside the allowed domain .
we recall that these strong bounds are obtained with no specific parametrization and without any assumption about the high energy behaviour of the form factor .
in the present paper we investigated the form factors of the light pseudoscalar mesons in a dispersive formalism which uses as input the ope expansion of the qcd polarization functions , combined with hadronic unitarity and analyticity .
we derived constraints on the size and shape of the form factors , which are of interest in particular for testing the low energy predictions of chpt .
generalizations of the original mathematical techniques , suitable for including additional informations about the form factors were developed . in section 3
we performed a test of sirlin s theorem @xcite , @xcite , which requires the vanishing of a certain combination of form factors which is free of arbitrary renormalization constants in chpt at @xmath360 . to this end
we treated simultaneously in the dispersive formalism all the electroweak form factors of the @xmath0 and @xmath1 mesons .
a difficulty is the large unphysical cut of the kaon electromagnetic form factors , which spoils the model independence of the results .
this is in contrast with the case of heavy mesons , where the weak form factors of the ground state and of the excited ones have very close branch points and the unphysical cuts are well approximated by a few number of poles . in that case the simultaneous treatment of several form factors related among them by heavy quark symmetry led to strong model independent bounds near zero recoil @xcite , @xcite , @xcite . in the present paper we adopted for the phase of the kaon form factors along the unphysical cut a model inspired from the resonance chiral effective theory , which turned out to be consistent with qcd and sirlin s theorem .
this subsection has however mainly a pedagogical character , showing how to combine in an optimal way the dispersive formalism with the omns functions and the technique of lagrange multipliers for the constraints at interior points .
the main results are presented in section 4 , where we derived model independent constraints on the taylor coefficients of the pion electromagnetic form factor .
we showed how to implement in the dispersive formalism the knowledge of the phase of the form factor along a part of the unitarity cut , even when this part consists of several disjoint intervals . to this end
we applied the lagrange theory for functional optimization , which led to an integral equation for the generalized lagrange multiplier .
the knowledge of the phase considerably improves the simple bounds on the higher taylor coefficients yielded by the dispersive formalism . in the present paper we illustrated this statement using a realistic model of the phase @xcite , which reproduces the one - loop chpt expression @xcite below @xmath361 and the present experimental data @xcite above @xmath362 .
the accurate data provided on the phase @xmath363 by the future @xmath240 experiments @xcite,@xcite will be a precious input to the formalism . in section 4
we obtained also improved constraints on the taylor coefficients if both the phase and the modulus of the form factor are given along a part of the cut , even if this information is used in an nonoptimal way .
the interior of the ellipse shown in fig .
3 represents the allowed domain of the coefficients @xmath210 and @xmath232 entering the taylor expansion ( [ taylor ] ) , obtained with the phase discussed above and the modulus from the hadronic @xmath212 decay rate @xcite along @xmath364 . once this input information is adopted , the bounds are model independent , since they are not based on specific parametrizations and are free of any assumption about the high energy behaviour of the form factors .
it is of interest to apply the techniques described in this paper to the higher taylor coefficients .
the results obtained so far suggest that strong correlations among these coefficients are expected , leading to restrictions on the arbitrary renormalization constants of chpt . 0.3 cm * acknowledgements : * i am pleased to thank prof
. h. leutwyler for very useful discussions and the kind hospitality at the institute of theoretical physics of the university of berne , and dr .
c. bourrely for many suggestions and a careful reading of the manuscript .
the support of the swiss national science foundation in the program csr ceec/ nis , contract no 7 ip 051219 is gratefully acknowledged .
0.2 cm 99 j. gasser and h. leutwyler , ann .
* 158 * , 142 ( 1984 ) ; nucl .
b250 * , 465 ( 1985 ) . j. gasser and u. meissner , nucl
b357 * , 90 ( 1991 ) .
j. bijnens , p. colangelo , g. ecker , j. gasser and m. e. sainio , phys .
* b374 * , 210 ( 1996 ) ; nucl . phys . *
b508 * , 263 ( 1997 ) ; ibid . * b517 * , 639 ( 1998 ) ( e ) . g. colangelo , m. finkemeier and r. urech , phys . rev .
* d54 * , 4403 ( 1996 ) .
j. bijnens , p. colangelo and p. talavera , jhep 9805:014 ( 1998 ) .
p. post , k. schilcher , phys .
* 79 * , 4088 ( 1997 ) .
a. sirlin , ann . phys . * 61 * , 294 ( 1970 ) ; phys . rev . lett .
* 43 * , 904 ( 1979 ) . m. ademollo , r. gatto , phys . lett . * 13 * , 264 ( 1964 ) . t. n. truong , phys .
lett . * 67 * , 2260 ( 1991 ) .
t. hannah , phys .
rev . * d55 * , 5613 ( 1997 ) .
meiman , sov.phys .
jetp * 17 * , 830 ( 1963 ) .
s. okubo , phys .
d * 3 * , 2807 ( 1971 ) ; + s. okubo and i. shih , phys .
d * 4 * , 2020 ( 1971 ) .
c. bourrely , b. machet and e. de rafael , nucl .
* b189 * , 157 ( 1981 ) .
rafael and j.taron , phys .
b282 * , 215 ( 1992 ) ; e.de rafael and j.taron , phys . rev .
* d50 * , 373 ( 1994 ) .
i.caprini , z.phys .
* c61 * , 651 ( 1994 ) .
i. caprini and c. macesanu , phys . rev . *
d54 * , 5686 ( 1996 ) .
c.g.boyd , b.grinstein and r.f.lebed , phys . lett .
* b353 * , 306 ( 1995 ) . i. caprini and m. neubert , phys . lett . *
b380 * , 376 ( 1996 ) .
c.g.boyd , b.grinstein and r.f.lebed , phys . rev . *
d56 * , 6895 ( 1997 ) .
i. caprini , l.lellouch and m. neubert , nucl . phys . *
b530 * , 153 ( 1998 ) .
n. isgur and m.b .
wise , phys .
b232 * , 113 ( 1989 ) ; * b237 * , 527 ( 1990 ) . m. neubert , phys . rep . * 245 * , 259 ( 1994 ) .
w.w . buck and r. f. lebed , phys . rev .
* d58 * , 056001 ( 1998 ) .
e.braaten , s. narison and a. pich , nucl . phys . *
b373 * , 581 ( 1992 ) .
m. a. shifman , a.i .
vainshtein and v. i. zakharov , nucl .
* b147 * , 385 ( 1979 ) .
m. gourdin and e. de rafael , nucl . phys . *
b10 * , 667 ( 1969 ) .
r. alemany , m. davier and a. hcker , eur . phys.j . *
c2 * , 123 ( 1998 ) .
i. caprini and p. dita , j.phys.a:math.gen .
* 13 * , 1265 ( 1980 ) .
r. barate et al ( aleph collaboration ) z. phys .
* c76 * , 15 ( 1997 ) .
gounaris and j.j.sakurai , phys .
* 21 * , 244 ( 1968 ) .
t. n. truong , hep - ph/9809476 .
m. davier , l. girlanda , h. hcher and j. stern , phys . rev . *
d58 * ; 096014 ( 1998 ) .
j. f. donoghue and e. s. na , phys . rev . * d56 * , 7073 ( 1997 ) .
watson , phys . rev . * 95 * , 228 ( 1954 ) .
protopopescu et al . , phys . rev .
* d7 * , 1279 ( 1973 ) .
+ p. estabrooks and a.d .
martin , nucl . phys .
* b79 * , 301 ( 1974 ) . j. loewe , in @xcite , p. 368 , and
hep - ph/9711361 .
d. pocani , in @xcite , p. 352 , and hep - ph/9801366 . ,
proceedings of the workshop held in mainz , germany , 1997 ( lecture notes in physics , vol .
513 , a. bernstein , d. drechsel and t. walcher , eds ) . c. bourrely , nucl .
b43 * , 434 ( 1972 ) .
g. auberson , nucl .
phys.b*98 * , 204 ( 1975 ) .
i. caprini , nuovo cim . *
49a * , 307 ( 1979 ) .
i. caprini , phys . rev . *
d27 * , 1479 ( 1983 ) .
n. i. mushelishvili , singular integral equations ( noordhoff , groningen , 1953 ) .
d. g. luenberger , optimization by vector space methods ( wiley , new york , 1969 ) .
( olya , cmd collaboration ) , nucl . phys . * b256 * , 365 ( 1985 )
. + d. bisello _ et al . _
( dme collaboration ) , phys .
b220 * , 321 ( 1989 ) .
krein and p.i .
nudelman , dokl.acad .
nauk sssr * 209 * , 537 ( 1973 ) .
m. rosenblum and j. rovnyak , proc .
soc . * 51 * , 335 ( 1975 ) .
s. ciulli , c. pomponiu and i. s. stefanescu , phys . rep . *
c17 * , 133 ( 1975 ) . | the form factors of the light pseudoscalar mesons are investigated in a dispersive formalism based on hadronic unitarity , analyticity and the ope expansion of the qcd green functions .
we propose generalizations of the original mathematical techniques , suitable for including additional low energy information provided by experiment or chiral perturbation theory ( chpt ) .
the simultaneous treatment of the electroweak form factors of the @xmath0 and @xmath1 mesons allows us to test the consistency with qcd of a low energy chpt theorem . by applying the formalism to the pion electromagnetic form factor
, we derive quite strong constraints on the higher taylor coefficients at zero momentum , using information about the phase and the modulus of the form factor along a part of the unitarity cut .
0.mm
* dispersive and chiral symmetry constraints on the light meson form factors * irinel caprini + national institute of physics and nuclear engineering , + pob mg 6 , bucharest , r-76900 romania |
coupled dynamical systems often arise in nature whenever a collective or cooperative phenomenon is favoured @xcite . in particular , the coupled map lattice with diffusive coupling ( cml ) provides a prototype model to study various features associated with the cooperative evolution of constituent systems @xcite .
one of the important properties of such cmls is that they exhibit size instability , that is , there is a critical size on the number of constituents for which stable synchronous chaotic state exists . increasing the number of constituents beyond this limit leads to the occurrence of spatially incoherent behaviour ( eg .
, high - dimensional chaos ) .
for example , bohr and christensen @xcite have studied such size instability behaviour in a two - dimensional coupled logistic lattice .
similar desynchronization has been found in arrays of coupled systems represented by nonlinear oscillators @xcite .
the stability of synchronous chaos in coupled dynamical systems plays an important role in the study of pattern formation , spatiotemporal chaos , etc .
@xcite . in general
, these studies on size instability are valid in most situations .
however , we have noted that in certain circumstances there is an ambiguity in dealing with these systems below the critical system sizes . to be specific , there exist certain nontrivial ranges of initial conditions for which the cml admits spatial and temporally periodic solutions in contrast to the usually expected stable synchronous chaos . in this brief report
, we show numerically the coexistence of such periodic states with the stable synchronous chaotic state well below the critical system size and explain the underlying mechanism .
consider an one - dimensional coupled map lattice with nearest - neighbour diffusive coupling @xcite @xmath0 , \label{cml}\end{aligned}\ ] ] where @xmath1 @xmath2 represents the lattice sites and @xmath3 is the system size , subject to periodic boundary conditions .
the stability analysis of the synchronized chaotic state defined by @xmath4 in the above cml using the procedure derived originally for coupled oscillators by heagy _ et al _
@xcite , gives the relation connecting transverse lyapunov exponents ( tles ) , in terms of the lyapunov exponent of single ( isolated ) map , @xmath5 as @xmath6 .
\label{tr_lya}\end{aligned}\ ] ] the synchronous state is stable only if the tles ( @xmath7 , @xmath8 ) are all negative . the above relation ( [ tr_lya ] )
can also be obtained by means of a direct perturbation of the form @xmath9 as considered by bohr and christensen @xcite .
the synchronous state loses its stability when the long waves ( lowest mode ) are unstable @xcite .
this means that for @xmath10 the synchronous state is unstable .
thus , substituting @xmath11 in eq .
( [ tr_lya ] ) , one obtains the maximum / critical lattice size @xmath12 that supports stable synchronous state as @xmath13 now , let us consider a coupled logistic lattice with diffusive coupling ( cll ) where each lattice site in eq .
( [ cml ] ) is occupied by the logistic map @xmath14 in particular for the choice @xmath15 , the lyapunov exponent of single ( isolated ) map is positive ( i.e. , @xmath16 ) and chaotic . in this case , for coupling strength @xmath17 , the critical lattice size ( @xmath18 ) is found to be @xmath19 from eq .
( [ lmax ] ) .
that is , upto the lattice size @xmath19 the cll exhibits synchronous chaos and for lattice size @xmath20 the synchronization is found to be lost , thereby confirming the size instability in the diffusively coupled logistic lattices .
when random initial conditions are assumed , in most cases , the above cll with diffusive coupling exhibits stable synchronous chaos for @xmath21 as predicted by eq .
( [ lmax ] ) . however , there are certain ranges of initial conditions for which cll shows some interesting asynchronous spatiotemporal patterns even for @xmath22 . for example , for @xmath23 and with the choice of initial conditions @xmath24 @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 or several nearby initial conditions , the cll exhibits a standing wave type pattern as shown in fig .
[ n8dyn_n](a ) .
but for the same lattice size , if we choose a different set of initial conditions , @xmath24 @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 or the nearby points , two standing waves with different amplitudes are produced within the cll as in fig .
[ n8dyn_n](b ) .
similarly a disturbed standing wave pattern as shown in fig .
[ n8dyn_n](c ) is possible to be exhibited by the cll for many choices of initial conditions . a synchronous chaos , as one would expect for @xmath23 from the theory , is also exhibited by the cll as depicted in fig . [ n8dyn_n](d ) , for most of the random choices of initial conditions .
this kind of multiple stable solutions is also observed in the cll for other lattice sizes , namely @xmath41 , @xmath42 , @xmath43 and @xmath44 which are also less than @xmath18 .
the occurrence of various spatiotemporal patterns for different lattice sizes of the cll and their percentage of occurrence are shown in table [ tab_pattern ] . in order to quantify the percentage of occurrence of different spatiotemporal patterns , we have used @xmath45 sets of random initial conditions ( i.c.s ) in the interval @xmath46 to @xmath47 and identified the number of i.c.s which lead to a specific pattern as indicated in table [ tab_pattern ] .
we have further confirmed our assertion by analyzing the same systems in different computing environments such as intel pentium 4 , sun sparc server / workstation and compaq alpha workstation .
in addition , by considering the whole cml of size @xmath3 as a single @xmath3-dimensional map we have verified that the above spatiotemporal periodic structures are essentially the stable fixed points of this map . for example , the periodic structure in fig .
[ n8dyn_n](a ) represents a fixed of point of period-2 of the eight - dimensional map and the eigenvalues of the corresponding jacobian matrix all are having magnitude less than unity . in a similar fashion one can verify that all the periodic structures are the stable fixed points of corresponding periods .
thus , in addition to the stable synchronized manifold , there exists other invariant sets corresponding to stable periodic structures .
.the possible spatiotemporal patterns for different lattice sizes and their percentage of occurrence for the coupled logistic lattice sampled over a set of @xmath48 random initial conditions ( ic s ) .
the parameters are fixed as @xmath15 in eq .
( [ fx : logi ] ) and @xmath49 in eq .
( [ cml ] ) . here ,
sw : standing wave , sync . : synchronized .
[ cols="^,<,>,^ , < , > " , ] [ table1 ] so , if it is possible to control the coupling force to fall in the region which corresponds to a periodic solution , then one can obtain standing wave patterns irrespective of the size of the lattices .
this is in fact possible by choosing appropriate initial conditions to each lattice point and this explains the occurrence of standing waves ( asynchronous ) as shown in figs .
[ n8dyn_n](a ) and [ n8dyn_n](b ) in the cll well below the critical lattice size ( @xmath18 ) .
similar explanation holds good for fig .
[ n8dyn_n](c ) .
the same principle is involved in the occurrence of standing waves even for @xmath50 @xcite .
from a careful numerical analysis , we have observed that the nodes and antinodes of standing waves are formed at or very close to the upos of the isolated logistic map . keeping this in mind
, we propose an expression for the standing wave pattern @xcite of the form @xmath51 where , @xmath52 where the discrete index @xmath53 corresponds to the lattice site , @xmath54 denotes the mode of the waves , and @xmath55 and @xmath56 represent the nodes and antinodes of different standing waves present in the pattern , which can take values from @xmath47 to @xmath57 and @xmath47 to @xmath58 , respectively and @xmath57 is the period of upo .
also in the above equation ( [ swave ] ) , @xmath59 s represent the values of the upos of the isolated logistic map at the node of the @xmath55-th standing wave , and @xmath60 and @xmath61 are the absolute values of the differences between upos at the node and upos at high and low amplitudes at the antinodes of the @xmath55th standing waves , respectively . the wave patterns with one and two number of standing waves obtained using eq .
( [ swave ] ) for @xmath62 are shown in fig .
[ n8dyn_a ] , whereas for the same lattice size , the numerically obtained pattern which have been discussed in sec .
[ sec3 ] is shown in figs .
[ n8dyn_n](a ) and [ n8dyn_n](b ) .
one observes that these two figures coincide very closely .
in this report , we have pointed out that coupled map lattices with diffusice coupling exhibit multiple stable states for the same set of parameters with respect to the initial conditions .
it has also been shown that by choosing appropriate initial conditions , one can obtain different standing wave type patterns for such coupled map lattices even for lattice size much less than the critical system size @xmath18 , where one normally would expect synchronized chaos .
in addition , we have proposed the mechanism behind the occurrence of such standing wave patterns .
this work has been supported by the national board for higher mathematics , department of atomic energy , government of india and the department of science and technology , government of india through research projects . | we analyze the size limits of coupled map lattices with diffusive coupling at the crossover of low - dimensional to high - dimensional chaos .
we investigate the existence of standing - wave - type periodic patterns , within the low - dimensional limit , in addition to the stable synchronous chaotic states depending upon the initial conditions .
further , we bring out a controlling mechanism to explain the emergence of standing - wave patterns in the coupled map lattices . finally , we give an analytic expression in terms of the unstable periodic orbits of the isolated map to represent the standing - wave patterns . |
in quantum physics , the state of the system can be unambiguously determined by measuring several select integrals of motion ( observables conserved in time evolution ) ; such observables are said to form a complete set of commuting observables ( csco ) @xcite .
mathematically , the spectra of the allowed values of the members of the set @xmath0 are the discrete sequences of real numbers , @xmath1 , that obey the following property : for any pair of indices @xmath2 , there exists at least one member of the csco ( say @xmath3 ) for which the corresponding elements are non - degenerate ( i.e. distinct , @xmath4 .
operationally , for any @xmath5 , the knowledge of the real - number sequence @xmath6 is sufficient to infer what @xmath5 was . in this paper
we consider a situation where the indistinguishability occurs due to an insufficient accuracy of the detection .
we assume that the spectra of the csco members are drawn from random processes , later chosen to be of the poisson type .
we further assume that for a given observable @xmath7 , two of its possible values , @xmath8 and @xmath9 , can _ not _ be resolved if they are separated by a distance less than the detection error , @xmath10 ; in this case , they are considered to be degenerate for all practical purposes .
the principal goal of this paper is to determine the probability for two observables , @xmath11 and @xmath12 , to form an csco ( see examples depicted at figs .
[ f : csco_novel_on ] and [ f : csco_novel_off ] ) . and @xmath12 , forming a complete set of commuting observables ( csco ) .
even if a particular pair of measured values of one of the observables is degenerate ( points in ovals ) , i.e. indistinct given the measurement error , it will not be degenerate vis - a - vis another observable .
if both observables are measured with respective errors @xmath13 and @xmath14 , the state of the system , @xmath15 , can be determined unambiguously . ] and @xmath12 are the only two observables that are measured , the states @xmath16 and @xmath17 are indistinct . ]
furthermore , we will concentrate on quantum systems with finite spectra of size @xmath18 , i.e. we will assume that a given system can be in any of @xmath18 available states , where @xmath18 is finite . in real - world applications
, @xmath18 grows with the system size . this study is partially inspired by a related issue of the functional dependence between the quantum observables ( i.e. the ability to predict the value of one observable having had measured the values of several others ) : while being a clear and extremely useful concept in classical physics , the practical significance of the notion of functional dependence in the quantum world is questionable at best . already in 1929 , von neumann showed @xcite that for any quantum system with @xmath18 states , one can construct a set of as many as @xmath19 conserved quantities each functionally independent from the other @xmath20 ; this is provided that any number of degeneracies is allowed . on the other hand , sutherland argued @xcite that any conserved quantity is functionally dependent on any other conserved quantity with a non - degenerate spectrum .
being able to determine the state of the system amounts to being able to predict the value of any other observable .
thus , from the functional dependence perspective we are looking for the probability that all conserved quantities be functionally dependent on two other , chosen beforehand .
consider two @xmath18-element - long finite ( ordered ) sequences of real numbers , @xmath21 both sequences are assumed to be @xmath18-event - long randomly reshuffled fragments of a poisson process .
let us elaborate on what this assumption means .
consider monotonically increasing permutations of the above sequences , @xmath22 according to the definition of the poisson process , the probability of finding exactly @xmath23 elements of the @xmath24 sequence in an interval of a length @xmath25 is @xmath26 $ } ] = \\ & \qquad\qquad \frac { e^ { -{\cal i } / \overline{\delta i^{(1 ) } } } ( { \cal i } / \overline{\delta i^{(1 ) } } ) ^k } { k ! } \,\ , , % \end{aligned}\ ] ] for any initial position of the interval @xmath27 .
here , @xmath28 is the expectation value of the interval between any two consecutive elements of the sequence : @xmath29 \,\ , .
% \end{aligned}\ ] ] likewise , @xmath30 $ } ] = \\ & \qquad\qquad \frac { e^ { -{\cal i } / \overline{\delta i^{(2 ) } } } ( { \cal i } / \overline{\delta i^{(2 ) } } ) ^k } { k ! } \\ &
\overline{\delta i^{(2 ) } } \equiv \mbox{{\it mean}}[\tilde{i}^{(2)}_{n+1}-\tilde{i}^{(2)}_{n } ] \,\ , .
% \end{aligned}\ ] ] an important particular property of the poisson processes is that the intervals between two successive elements of the sequence are mutually statistically independent , and they are distributed according to the _ exponential _ law : @xmath31 = e^{-{\cal i}/\overline{\delta i^{(1 ) } } } \\ & \mbox{{\it prob}}[\tilde{i}^{(2)}_{n+1}-\tilde{i}^{(2)}_{n } > { \cal i } ] = e^{-{\cal i}/\overline{\delta i^{(2 ) } } } \label{exponential_distribution } \end{aligned } \,\ , . % \end{aligned}\ ] ] let us now return to the original sequences @xmath32 and @xmath33 that are random permutations of the @xmath24 and @xmath34 : @xmath35 \\ & \left ( i^{(2)}_{1},\,i^{(2)}_{2},\,\ldots,\,i^{(2)}_{n},\,\ldots,\,i^{(2)}_{n } \right ) = \\ & \quad \mbox{random permutation } [ \left ( \tilde{i}^{(2)}_{1},\,\tilde{i}^{(2)}_{2},\,\ldots,\,\tilde{i}^{(2)}_{n},\,\ldots,\,\tilde{i}^{(2)}_{n } \right ) ] \,\ , .
% \end{aligned}\ ] ] it also follows from the properties of the poisson processes that the @xmath32 and @xmath33 ) sequences can be approximately generated by dropping @xmath18 uniformly distributed random numbers on respective intervals of lengths @xmath36 and @xmath37 . for a given sequence ,
say @xmath32 , call a pair of elements , @xmath38 and @xmath39 , _ degenerate _ if the difference between them is less than a given number , @xmath13 , called the _ detecting error _
: @xmath40 assume that the sequence @xmath41 is known , and a device ( emulating a process of quantum measurement ) produces a particular element of it , @xmath42 .
an observer is allowed to measure it in an attempt to determine what the index @xmath15 was ( the analogue of a quantum state index ) , using the sequence as a look - up table . for a perfect measurement device and in the absence of coinciding elements in the sequence , this task can be easily accomplished .
if however the accuracy of the measurement is limited , and the values separated by less than @xmath13 can not be distinguished , the value @xmath42 may happen to belong to one of the degenerate pairs , @xmath43 .
if , further , there is no other degenerate pair @xmath42 belongs to , @xmath15 can only be said to be equal to _ either _ @xmath44 _ or _
. if there are other relevant degenerate pairs , the uncertainty can be even greater .
likewise , degenerate pairs of elements of the second sequence is determined through @xmath46 where @xmath14 is the error of the detector that measures the @xmath12 observable .
now , both @xmath42 and @xmath47 are produced , and both detectors are used .
when both measurements are allowed , several ambiguities in determining the index can be removed . however ,
if for a given pair of indices , @xmath44 and @xmath45 , both @xmath43 and @xmath48 constitute the respective degenerate pairs , the indices @xmath44 and @xmath45 can never be resolved .
the goal of this work is to determine the probability that no above ambiguities exist : @xmath49 = ? \,\ , .
% \end{aligned}\ ] ] physically , we are interested in computing the probability that the observables @xmath11 and @xmath12 constitute a complete set of commuting observables ( csco ) .
consider a length-@xmath18 fragment of a poisson process , @xmath50 with the mean spacing @xmath51 . in many respects , studying monotonically increasing sequences is easier than their random - order counterparts .
this is fortunate since several important conclusions about the latter can be drawn using the former . as a model for a length-@xmath18 fragment of a poisson process with the mean spacing @xmath51 we use @xmath18 pseudorandom numbers uniformly distributed on an interval between 0 and @xmath52 .
when ordered in the order it has been generated , the sequence serves as a model for a random permutation of a poisson sequence , @xmath53 , the primary object of interest . when rearranged to a monotonically increasing sequence , a model for a poisson sequence per se , @xmath54 ,
is generated . to test the numerical method , we compare , in fig .
[ f:_r__probabilityforapairtobedegeneratei ] , numerical and analytical predictions for the probability @xmath55 of two subsequent elements of the monotonically increasing sequence to form a degenerate pair , with respect to a detection error @xmath56 .
the analytic prediction , @xmath57 = \\ & \qquad\qquad\qquad\qquad 1-e^{-\delta i/\overline{\delta i } } \stackrel{\delta i \ll \overline{\delta i}}{\approx } \frac{\delta i}{\overline{\delta i } } \,\ , , \label{p } \end{aligned } % \end{aligned}\ ] ] follows directly from one of the properties of the poisson sequences ( [ exponential_distribution ] ) .
the numerical and analytical predictions agree very well .
( i.e. to be degenerate ) .
@xmath51 is the mean spacing between consecutive elements .
red dots : numerical model , uncorrelated , uniformly - distributed random numbers on an interval , subsequently rearranged in a monotonically increasing succession .
green line : theoretical prediction ( [ p ] ) .
numerically , the spectrum consists of 100 elements ; the theoretical prediction is universal .
numerical points result from averaging over 1000 monte carlo realizations . ] from ( [ p ] ) it also follows that the average number of degenerate pairs is @xmath58 where @xmath59 is the number of pairs of indices with two neighboring elements . in what follows ,
the analytical predictions will be greatly simplified thanks to our ability to assume that degenerate pairs do not share elements , i.e. that a given element can either be isolated or belong to only one degenerate pair .
it is clear that the probability of forming a degenerate cluster must be much smaller than the one for an isolated pair , because of the small probability of forming the latter . yet ,
a quantitative assessment is due . assuming that that the individual spacings between the subsequent values of @xmath27 are statistically independent , we get : @xmath60 = ( 1-p)p^2 \,\ , .
\label{p_c } \end{aligned } % \end{aligned}\ ] ] note that the above probability is the probability for the index @xmath15 to be at the left end of a cluster of degenerate pairs containing more than one pair .
[ f:_r__probabilityforapairtobeleftpairofdegenerateclusteri ] compares the above prediction with _ ab initio _ numerics : the agreement is remarkable .
the overall conclusion is that for a small probability @xmath55 of forming a degenerate pair , one can safely assume that all the degenerate pairs are isolated . .
red dots : numerics .
green line : theoretical prediction ( [ p_c ] ) .
it does not depend on the sequence length @xmath18 .
the rest of the data is the same as in fig .
[ f:_r__probabilityforapairtobedegeneratei ] . ] this result will be heavily used in a subsequent treatment of the more physical random - order sequences .
it is instructive to also compute the probability of having no degenerate clusters at all , for two reasons .
( a ) this is the first measure in this study that characterizes the @xmath54 as a set , with no reference to the order of the elements .
the probability of no clusters we will obtain will therefore be identical to the one for the @xmath53 sequence , randomly ordered .
on the other hand , the monotonically increasing sequences are conceptually simpler , and they allow for more intuition in designing analytical predictions .
( b ) using this study we will show that while the spacings between neighboring members of the monotonic sequence are statistically independent , the degenerate pairs they produce are not statistically independent from the point of view of the randomly permuted sequence . even though the spacings between the consecutive elements of the @xmath61 sequence are statistically independent , the appearance of a cluster with the leftmost edge at an index @xmath15 will be preventing an appearance of another cluster at @xmath62 .
thus , appearances of clusters at different points are not statistically independent .
nevertheless , for a very low cluster probability , a typical distance between the clusters becomes much larger than their typical length , and the above correlations can be neglected .
our estimate for the probability of having no clusters at all then reads : @xmath63 \approx \\ & \qquad ( 1-p_{c})^{n_{p } } \stackrel{p\to 0,\ , n\to\infty,\ , p^2 n \to \mbox{const}}{\approx } e^{-p_{c } n } \stackrel{p\to0}{\approx } e^{-p^2 n } \,\ , , \label{p_no_clusters_correct } \end{aligned } % \end{aligned}\ ] ] where @xmath64 is given by ( [ n_p ] ) . above
, we neglected limitations on the cluster length closer to the right edge of the spectrum , since we are mainly interested in the @xmath65 limit .
as we have mentioned above , the result ( [ p_no_clusters_correct ] ) applies for both @xmath61 and @xmath27 sequences , being a characteristics of these sequences as a set , not as sequences with a particular order .
it is tempting to reinterpret it as the probability that @xmath66 pairs chosen uniformly at random from @xmath67 pairs of elements of an @xmath18-member - strong set have no elements in common .
the number of ways to choose @xmath66 pairs with no elements in common is @xmath68 the number of ways to chose any set of @xmath66 pairs is @xmath69 dividing the former by the latter we get the following expression : @xmath70 where @xmath66 is given by ( [ m_bar ] ) . comparing ( [ p_no_clusters_incorrect ] ) to ( [ p_no_clusters_correct ] )
one notices that the former overestimates the probability of cluster formation by a factor of two .
an interpretation of this simple relationship requires future research .
figure [ f:_r__probabilityofnoclustersi ] illustrates this difference . .
red dots : numerics .
green line : theoretical prediction ( [ p_no_clusters_correct ] ) .
blue line : a naive combinatorial hypothesis ( [ p_no_clusters_incorrect ] ) . at small detector window widths , the formula involves factorials of large numbers and becomes numerically unreliable .
purple line : the same as the blue line but using an asymptotic expression .
the rest of the data is the same as in fig .
[ f:_r__probabilityforapairtobedegeneratei ] . ]
the principal goal of our paper is to assess the ability of two quantum conserved quantities to serve as a csco set .
the physical realization of our model will involve two conserved quantities in an integrable system .
those are indeed known to realize poisson processes @xcite , but only after a permutation to a monotonically increasing sequence . for two observables , these permutations are completely unrelated .
thus a physically more relevant model would involve involve sequences that are represented by a random permutation of a poisson sequence , @xmath71 \,\ , .
% \end{aligned}\ ] ] numerically , those are modeled by sets of real numbers randomly independently distributed over an interval , with _ no subsequent reordering_. to make a theoretical prediction for the probability for a given pair of two subsequent elements of the sequence to be degenerate , we first assume that the probability for degenerate pairs to share elements ( _ i.e . form clusters of degenerate pairs _ ) is low ( see fig .
[ f:_r__probabilityforapairtobeleftpairofdegenerateclusteri ] and the expression ( [ p_c ] ) ) . in this case
, one can assume that each degenerate pair constitutes a pair of consecutive elements in the monotonically increasing counterpart of the sequence .
the probability of the later is @xmath72 ( with @xmath64 given by ( [ n_p ] ) ) .
this probability must then be multiplied by the probability of a neighboring pair in the monotonic sequence to be degenerate , @xmath55 .
we get @xmath73 = \\ & \qquad\qquad \frac{2}{n_{p } } p \stackrel{n\to\infty}{\approx } \frac{2}{n } p \,\ , .
\label{p_randomized } \end{aligned } % \end{aligned}\ ] ] fig .
[ f:_r__probabilityforapairtobedegenerateiunordered ] demonstrates that this prediction is indeed correct .
( i.e. to be degenerate ) .
@xmath51 is the mean spacing between consecutive elements before the random permutation .
red dots : numerical model , uncorrelated uniformly distributed random numbers on an interval , with _ no further rearrangement_. green line : theoretical prediction ( [ p_randomized ] ) . numerically , the spectrum consists of 100 elements .
numerical points result from an averaging over 1000 monte carlo realizations . ]
we are now in the position to assess the ability of a single observable @xmath7 to serve as a csco .
the probability of that is given by the probability of having no degenerate pairs at all . since this probability is a property of the sequence as a set , it can be estimated using the monotonically increasing counterpart .
there , the probability of not having degenerate pairs is the probability that none of the @xmath19 pairs of consecutive indices are degenerate .
we get @xmath74 = \\ & \qquad ( 1-p)^{n_{p } } \stackrel{p\to 0,\ , n\to\infty,\ , p n \to \mbox{const}}{\approx } e^{-p
n } \,\ , .
\label{p_csco_1 } \end{aligned } % \end{aligned}\ ] ] fig .
[ f:_r__probabilityofnodegeneratepairsi ] shows that this probability approaches unity only for a detector error as low as the inverse sequence length @xmath18 .
furthermore , the expression ( [ p_csco_1 ] ) shows that even if for a given @xmath18 the given observable does form a csco , for larger @xmath18 this property dissappears . .
physically , this probability corresponds to the probability for a given observable to form a complete set of commuting observables ( csco ) .
red dots : numerics .
green line : theoretical prediction ( [ p_csco_1 ] ) .
the rest of the data is the same as in fig .
[ f:_r__probabilityforapairtobedegeneratei ] . ]
we are finally ready to address the principal question posed : what is the probability that two observables with poisson spectra , subsequently independently randomly permuted , @xmath11 and @xmath12 , constitute a complete set of commuting observables ( csco ) .
mathematically , the corresponding probability is : @xmath75 = \\ & \qquad\qquad\qquad ( 1-p_{\mbox{\scriptsize randomized}}^{(1)}p_{\mbox{\scriptsize randomized}}^{(2)})^{{\cal n}_{p } } \\ & \qquad\qquad\qquad\qquad \stackrel{n\to\infty}{\approx } e^{-2 p^{(1)}p^{(2 ) } } \,\ , , \label{p_csco_2 } \end{aligned } % \end{aligned}\ ] ] where @xmath76 are the respective probabilities for a given pair to be degenerate ( see ( [ p_randomized ] ) ) , @xmath77 is the analogue for a monotonic sequence ( see ( [ p ] ) ) , and @xmath78 is the respective detection error .
we assume the same spacing , @xmath51 , for both sequences .
a naive combinatorial interpretation of this probability is the ratio between the number of ways in which @xmath79 pairs can be chosen from two @xmath18-element - long sets in such a way that no two pairs coincide and its unrestricted analogue : @xmath80 where @xmath81 is the number of pairs of @xmath18 objects ( [ cal_n_p ] ) , and @xmath82 is the number of pairs of neighboring elements in a sequence of length @xmath18 ( see ( [ n_p ] ) ) . in this particular case
the naive model generates the correct prediction .
note that in this case , no correlations between the degenerate pairs within a particular sequence are involved , thus a better result .
figure [ f:_r__probabilityofnojointdegeneratepairsunordered ] allows one to compare the various predictions made above , for @xmath83 .
notice that ( a ) already at @xmath84 , csco is reached ; ( b ) from the expression ( [ p_csco_2_2 ] ) one can see that for large @xmath18 , the probability ( [ p_csco_2_2 ] ) does not depend on the length of the spectrum @xmath18 , and thus csco persist for large systems . a particular illustration of this phenomenon is presented in fig . [ f:_r__probabilityofnojointdegeneratepairsunordered_vs_n ] .
) , a naive combinatorial hypothesis ( [ p_csco_2_2 ] ) , and its large @xmath18 asymptotic behavior , all three mutually indistinct .
note that the former and the latter are analytically shown to coincide . ] . as in fig .
[ f:_r__probabilityofnojointdegeneratepairsunordered ] , the multicolored line corresponds at the same time to the theoretical prediction ( [ p_csco_2 ] ) and to the large-@xmath18 asymptotic behavior of the combinatorial expression ( [ p_csco_2_2 ] ) , which are shown to coincide .
the corresponding detector window sizes are @xmath85 and @xmath86 . ]
in this work we show that with a large probability , differing from unity only by the expression in ( [ p_csco_2_2 ] ) ( see also figs .
[ f:_r__probabilityofnojointdegeneratepairsunordered]-[f:_r__probabilityofnojointdegeneratepairsunordered_vs_n ] ) , two integrals of motion with poisson distributed spectra form a complete set of commuting observables ( csco ) ; i.e. they can be used to unambiguously identify the state of the system , the detection error notwithstanding . for a given detector error , this probability converges to a fixed value as the number of states in the spectrum increases .
the above result is contrasted to an analogous result for a single observable . there , ( a ) only for a very small detector error can this observable be used as a csco , and ( b ) for a fixed detector error , the probability of a csco falls to zero exponentially as the number of states in the spectrum increases .
poisson spectra constitute a popular model for energy spectra of integrable ( i.e. exactly solvable ) quantum systems with no true degeneracies @xcite .
they can also be used to model spectra of other integrals of motion of integrable systems , provided they are substantially functions of at least two quantum numbers .
note that in generic non - integrable quantum systems , energy levels repel each other @xcite .
there , even a poor energy detector , with an error as large as the energy level spacing , would be able to identify a state : energy can thus serve as a csco .
therefore , the physical implication of our principal result can be formulated as follows : given a reasonably small ( i.e. smaller than the mean spacing between levels ) detector error , two generic integrals of motion of an integrable quantum system can be used as a complete set of commuting observables , no matter how large the system is . on the other hand ,
if only one integral of motion is used , the appearance of unidentifiable states is unavoidable for large systems . from the functional dependence perspective
, we showed that in poorly measured integrable systems , all conserved quantities are ( with a close - to - unity probability ) functionally dependent on any two a priori chosen generic conserved quantities .
results obtained in our article may find applications beyond the quantum state identification .
they should be generally applicable in a standard pattern recognition setting where an unknown object must be identified indirectly by one , two , or more attributes .
for example , using our results , we can show that chances of an ambiguity in identifying a person who was born on a particular date @xmath87 in a particular town @xmath88 within a specific time interval it and who is currently living on a particular street @xmath89 in another town @xmath90 depends neither on the length of the time interval @xmath91 , nor on the number of the streets in @xmath90 , but solely on the number of birth per day in @xmath88 , probability of a further migration to @xmath90 , percentage of @xmath88-born citizens in @xmath90 , and the average street population in @xmath90 ; in the other words , this probability depends only on the relative measures , the absolute measures being irrelevant .
the value for this probability is trivially obtainable from the central result of this paper .
more generally , _ for a large enough set of n patterns with two poisson distributed numerical identifiers , each measured with a finite error , the probability @xmath92 of having no unidentifiable patterns depends only on the probability @xmath93 of a given value of an identifier to be indistinct from its neighbor above and the analogous probability @xmath94 for another identifier and not on size of the set @xmath18_. at the same time , if only a date of birth is known , the procedure will start frequently produce ambiguous results for long enough interval of time considered : here , the probability of finding two people born in a given town on the same date within this interval of time approaches unity .
generally , a _
single poisson identifier becomes unreliable for larger sets of patterns_.
authors shared equal amount of work on the probabilistic models .
e.m . provided the necessary numerical support , along with overseeing the overall logistics of the project .
m.o . was solely responsible for the development of the combinatorial models .
we are grateful to vanja dunjko and steven g. jackson for helpful comments and to maxim olshanii ( olchanyi ) for providing the seed idea for this project . | in this article , we revisit the century - old question of the minimal set of observables needed to identify a quantum state : here , we replace the natural coincidences in their spectra by effective ones , induced by an imperfect measurement .
we show that if the detection error is smaller than the mean level spacing , then two observables with poisson spectra will suffice , no matter how large the system is .
the primary target of our findings is the integrable ( i.e. exactly solvable ) quantum systems whose spectra do obey the poisson statistics .
we also consider the implications of our findings for classical pattern recognition techniques . |
let @xmath0 be a finite group . in some @xmath1-graded @xmath0-modules @xmath2
, we have a phenomenon called `` coincidence of dimensions , '' i.e. , some integers @xmath3 satisfy the equations @xmath4 for all @xmath5 and @xmath6 .
we call a datum @xmath7 a _ representation - theoretical presentation _ for the phenomenon if a subgroup @xmath8 of @xmath0 and @xmath8-modules @xmath9 satisfy @xmath10 for all @xmath5 and @xmath6 , where @xmath11 denotes the induced module
. morita and nakajima gave representation - theoretical presentations to coincidences of dimensions of the springer modules @xmath12 @xcite .
the springer modules @xmath12 are graded algebras parametrized by partitions @xmath13 . as @xmath14-modules ,
@xmath12 are isomorphic to the cohomology rings of the variety of flags fixed by a unipotent matrix with jordan blocks of type @xmath15 .
( see @xcite .
see also @xcite for algebraic construction . )
we recall the case where @xmath15 is an @xmath3-partition , where an @xmath3-partition means a partition whose multiplicities are divisible by @xmath3 .
let @xmath16 denote the submodule @xmath17 of the springer module @xmath12 .
in this case , we have @xmath18 for all @xmath5 and @xmath6 , i.e. , @xmath12 has a coincidence of dimensions .
let @xmath19 be the semi - direct product @xmath20 of the young subgroup @xmath21 and an @xmath3-th cyclic group @xmath22 .
( see section [ notationsec ] for their definitions . ) for @xmath23 , let @xmath24 denote one - dimensional representations of @xmath19 mapping @xmath25 to @xmath26 and @xmath27 to @xmath28 , where @xmath29 denotes a primitive @xmath3-th root of unity .
then @xmath30 is a representation - theoretical presentation for this case , i.e. , @xmath31 for all @xmath5 . to prove it
, morita and nakajima described the values of the green polynomials at roots of unity , and showed that the characters of the submodules @xmath32 coincide with those of the induced modules @xmath33 in @xcite .
our motivation in this paper is to construct isomorphisms from the induced module @xmath33 to the submodule @xmath34 in a combinatorial manner . for this purpose , we consider the problem in a more general situation . in section [ mainsec ] ,
we consider the induced module @xmath35 of a one - dimensional @xmath36-module @xmath37 and another realization @xmath38 of @xmath35 , where @xmath36 is a subgroup of a finite group @xmath0 .
we show lemmas about the condition for an element of @xmath38 to generate @xmath38 . in special cases ,
these lemmas are equivalent to the classical schur s lemma .
we give a necessary and sufficient condition for a map from @xmath35 to @xmath38 to be an isomorphism .
we also give a procedure to construct isomorphisms . in subsection
[ appsec ] , we apply the main results to the case of representation - theoretical presentations for the springer modules corresponding to @xmath3-partitions @xmath15 . in subsection
[ tworowsec ] , we explicitly give isomorphisms in the case of @xmath39
in this section , we consider a subgroup @xmath36 of a finite group @xmath0 and a one - dimensional representation @xmath40 of @xmath36 .
let @xmath41 be the induced module of @xmath40 , and @xmath38 a @xmath0-module isomorphic to @xmath41 .
we show a condition for a map from @xmath41 to @xmath38 to be an isomorphism of @xmath0-modules .
we prove theorems [ modschulemmaiii ] , [ mainlemma ] and [ lemmagroupring ] in subsection [ proofsec ] . throughout this section ,
we use the following notation and assumption .
let @xmath42 be a field .
we consider representations over @xmath42 .
we assume the complete reducibility of representations .
let @xmath0 be a finite group , @xmath43 the unit of @xmath0 , @xmath36 a subgroup such that @xmath44 , @xmath37 a one - dimensional vector space over @xmath42 , @xmath45 a representation acting on @xmath37 , and @xmath35 the induced module of @xmath37 .
[ genericdef ] for a @xmath0-module @xmath38 , @xmath46 is said to be _ generic in @xmath38 _ if there exists an irreducible decomposition @xmath47 satisfying the following : * @xmath48 is isomorphic to @xmath49 if and only if @xmath50 for some @xmath51 . * for @xmath52 ,
let @xmath53 denote the image of the projection of @xmath54 to the irreducible component @xmath55 .
for @xmath56 , let @xmath57 be an isomorphism from @xmath55 to @xmath49 .
then @xmath58 is @xmath42-linearly independent . by schur s lemma , definition [ genericdef ]
does not depend on choices of @xmath57 .
we can describe a necessary and sufficient condition for an element of @xmath38 to generate @xmath38 as a @xmath0-module .
[ modschulemmaiii ] for a @xmath0-module @xmath38 , @xmath59 generates @xmath38 if and only if @xmath60 is generic in @xmath38 . in the case where @xmath61 and @xmath62 is a simple @xmath0-module ,
this theorem is equivalent to the classical schur s lemma .
let @xmath41 and @xmath38 be @xmath0-modules .
let @xmath63 generate @xmath41 as a @xmath0-module . for @xmath46
, we define @xmath64 to be the @xmath65$]-linear map such that @xmath66 whenever it is well - defined , where @xmath65 $ ] denotes the group ring of @xmath0 . for a representation @xmath67
, we define an element @xmath68 of the group ring @xmath69 $ ] to be @xmath70 .
we can describe a necessary and sufficient condition for a map from @xmath35 to @xmath38 to be an isomorphism .
[ mainlemma ] let @xmath38 be a realization of @xmath35 , @xmath71 a nonzero element , and @xmath72 a map from @xmath35 to @xmath38 .
then @xmath72 is an isomorphism from @xmath35 to @xmath38 if and only if @xmath73 for some @xmath74 which is generic in @xmath38 .
[ lemmagroupring ] the induced module @xmath35 is isomorphic to @xmath65{\boldsymbol{a}_{\zeta}}$ ] as @xmath0-modules . for a nonzero element @xmath71 , the map @xmath75 is an isomorphism from @xmath35 to @xmath65{\boldsymbol{a}_{\zeta}}$ ] . by theorem [ mainlemma ] ,
constructing isomorphisms is equivalent to finding elements @xmath54 such that @xmath74 are generic . by the definition of generic elements
, we have the following procedure to construct such elements . [ procelemma1 ] for all elements
@xmath54 given by the following procedure , @xmath74 are generic in @xmath38 @xmath76 1 .
fix an irreducible decomposition @xmath77 of @xmath38 such that @xmath55 is isomorphic to @xmath49 if and only if @xmath78 for some @xmath51 .
2 . fix an element @xmath79 of @xmath80 for each @xmath51 .
let @xmath81 .
3 . fix an isomorphism @xmath82 for each @xmath83 .
4 . fix a basis @xmath84 of @xmath85 .
let @xmath86 be the image @xmath87 .
5 . fix a maximal subset @xmath88 of @xmath89 such that @xmath90 is @xmath42-linearly independent .
fix a total ordering on @xmath88 .
let @xmath91 .
fix a total ordering on @xmath80 .
let @xmath92 .
let @xmath93 .
let @xmath54 be @xmath94 . by definition ,
any element @xmath95 such that @xmath74 is generic in @xmath38 is obtained from the procedure in theorem [ procelemma1 ] . in this section ,
we prove the main results .
first , in [ necessarysec ] , we prove theorem [ modschulemmaiii ] , an extended schur s lemma .
next , in [ welldefsec ] , we show a necessary and sufficient condition for @xmath64 to be injective and well - defined as a homomorphism of @xmath0-modules .
finally we prove theorems [ mainlemma ] and [ lemmagroupring ] . here
we prove theorem [ modschulemmaiii ] , which gives a necessary and sufficient condition for the surjectivity of @xmath96 .
[ modschulemmai ] for a simple @xmath0-module @xmath62 and a subset @xmath97 , @xmath98 generates @xmath99 if and only if @xmath100 is @xmath42-linearly independent .
first we prove the case where @xmath101 .
the case where @xmath102 or @xmath103 is clear .
suppose that @xmath104 and @xmath105 .
there exists @xmath106 $ ] such that @xmath107 in @xmath108 if and only if @xmath65 \set{f_1\oplus f_2 } = s \oplus s$ ] .
if @xmath109 , then @xmath110 .
hence there does not exist such @xmath111 . conversely we consider the case where there does not exist such @xmath111 .
since this implies @xmath112 for all @xmath113 , we have @xmath114 .
hence @xmath115 is a well - defined endomorphism of @xmath62 . by the classical schur s lemma , @xmath116 and @xmath117 are @xmath42-linearly dependent .
thus we have the lemma for @xmath101 .
next we prove the lemma in the case where @xmath118 .
assume that @xmath119 generates @xmath120 .
in this case , there exists @xmath121 such that @xmath122 for each @xmath123 . if @xmath124 is @xmath42-linearly dependent , then @xmath125 hence @xmath65\set{f \oplus f_k}\simeq k[g]\set{f } = \bigoplus_{i=1}^{k-1 } s$ ] .
conversely , we consider the case where @xmath65\set{f \oplus f_k } \simeq \bigoplus_{i=1}^{k-1 } s$ ] . in this case , @xmath126 . since @xmath127 @xmath128\set{p_i ( f \oplus f_k ) } \simeq k[g ] \set { f_i\oplus p_i f_k}\simeq s. \end{gathered}\ ] ] since it follows from the case where @xmath101 that there exists @xmath129 such that @xmath130 , @xmath131 since @xmath132 , @xmath133 hence @xmath124 is @xmath42-linearly dependent .
thus we have the lemma in the case where @xmath118 . in the case where @xmath101 , lemma [ modschulemmai ] is equivalent to the classical schur s lemma .
[ modschulemmaii ] let @xmath134 be a family of simple @xmath0-modules which are not isomorphic to one another . for subsets
@xmath135 , @xmath136 generates @xmath137 if and only if each subset @xmath138 is @xmath42-linearly independent . by lemma [ modschulemmai ]
, @xmath139 generates @xmath140 if and only if @xmath141 is @xmath42-linearly independent .
since @xmath85 is not isomorphic to @xmath142 for @xmath143 , @xmath136 generates @xmath137 if and only if each subset @xmath144 is @xmath42-linearly independent .
this follows from the definition of generic elements and lemma [ modschulemmaii ] .
here we show lemma [ symmetrizerlemma ] , which gives a necessary and sufficient condition for @xmath64 to be injective and well - defined as a homomorphism of @xmath0-modules .
then we prove theorems [ mainlemma ] and [ lemmagroupring ] .
[ symmetrizerlemma ] let @xmath0 be a group , @xmath36 a subgroup of @xmath0 , and @xmath145 a one - dimensional representation .
for a @xmath0-module @xmath146 , @xmath147 for @xmath148 , @xmath149 hence @xmath150 . for @xmath151 , @xmath152 hence @xmath153 .
thus we have @xmath154 for any @xmath40 , the operator @xmath68 is a projection , i.e. , @xmath155 .
now we prove theorems @xmath156 and @xmath157 .
let @xmath72 be a homomorphism from @xmath35 to @xmath38 .
since @xmath37 is one - dimensional , @xmath72 is determined by @xmath158 . by lemma @xmath159 ,
the homomorphism @xmath160 is well - defined .
conversely , if we take @xmath161 which can not be written as @xmath74 , then @xmath162 is not an isomorphism .
hence we have @xmath163 for some @xmath54 .
now @xmath160 is well - defined , and @xmath35 is isomorphic to @xmath38 .
since @xmath38 is finite dimensional , it is enough to prove the surjectivity @xmath160 . by theorem [ modschulemmaiii ] , @xmath65\set{{\boldsymbol{a}_{\zeta } } f}=r$ ] if and only if @xmath74 is generic in @xmath38 .
hence @xmath160 is surjective if and only if @xmath74 is generic in @xmath38 .
therefore we have theorem [ mainlemma ] . by lemma [ symmetrizerlemma ] , @xmath164 is isomorphic to @xmath37 as @xmath36-modules . by definition , @xmath165 if and only if @xmath166 . hence we have theorem [ lemmagroupring ] .
in this section , we apply the main results to the case of the springer modules . in subsection [ notationsec ] , we prepare some basic notation . in subsection
[ appsec ] , we rewrite the main results for the case of representation - theoretical presentations for the springer modules @xmath12 corresponding to @xmath3-partitions @xmath15 . moreover , in subsection [ tworowsec ] , we explicitly describe isomorphisms in the case where @xmath39 .
we identify a partition @xmath167 of @xmath168 with its young diagram @xmath169 with @xmath168 boxes .
if @xmath15 is a young diagram with @xmath168 boxes , we write @xmath13 and identify a young diagram @xmath15 with the array of @xmath168 boxes having left - justified rows with the @xmath123-th row containing @xmath170 boxes ; for example , @xmath171 for an integer @xmath3 , a young diagram @xmath15 is called an _ @xmath3-partition _ if multiplicities @xmath172 of @xmath123 are divisible by @xmath3 for all @xmath123 .
we call a map @xmath173 a _ numbering on a young diagram @xmath13 _ if @xmath173 is a bijection @xmath174 .
we call a map @xmath173 a _ semi - standard tableau on a young diagram @xmath175 with weight @xmath176 _ if @xmath177 satisfies @xmath178 , @xmath179 and @xmath180 for all @xmath181 .
a numbering which is a semi - standard tableau is called a _
standard tableau_. we identify a map @xmath173 from a young diagram @xmath15 to @xmath182 with a diagram putting each @xmath183 on the box in the @xmath184 position ; for example , @xmath185 for @xmath13 , @xmath186 denotes the numbering that maps @xmath187 , i.e. , a numbering obtained by putting numbers from @xmath28 to @xmath168 on boxes of @xmath15 from left to right in each row , starting in the top row and moving to the bottom row .
for example , @xmath188 let @xmath173 be a numbering on an @xmath3-partition @xmath13 .
we define @xmath193 to be the product @xmath194 of @xmath195 cyclic permutations of length @xmath3 . for example , @xmath196 .
we write @xmath25 for @xmath197 .
we define @xmath198 to be the @xmath3-th cyclic group @xmath199 .
we define @xmath200 to be the semi - direct product @xmath20 . for @xmath23 ,
let @xmath24 denote one - dimensional representations of @xmath19 that maps @xmath25 to @xmath26 and @xmath27 to @xmath28 , where @xmath29 denotes a primitive root of unity .
in this section , we apply the main results to the case of the springer modules @xmath12 corresponding to @xmath3-partitions @xmath15 .
we rewrite the main results for @xmath33 , @xmath191{\boldsymbol{z}_{\mu , k , l}}$ ] , and @xmath34 . 1 .
let @xmath216 be an irreducible decomposition of @xmath34 such that @xmath55 is isomorphic to the specht module @xmath85 for @xmath83 .
2 . let @xmath217 be an isomorphism from @xmath218 to @xmath48 for @xmath83 .
3 . let @xmath86 be the image @xmath87 of the specht polynomial @xmath219 .
4 . let @xmath89 be the set of standard tableaux on @xmath51 .
let @xmath88 be a maximal subset of @xmath89 such that @xmath220 is @xmath221-linearly independent .
[ youngsubstep ] 6 .
let @xmath222 be a maximal subset of @xmath88 such that @xmath223 is @xmath221-linearly independent .
let @xmath224 .
let @xmath225 .
let @xmath93 .
let @xmath54 be @xmath226 .
hence @xmath227{\boldsymbol{z}_{\mu , k , l } } \ni { \boldsymbol{z}_{\mu , k , l } } \mapsto { \boldsymbol{z}_{\mu , k , l } } f \in r_{\mu}(k;l)\\ \intertext{and } \varphi_{\varepsilon\otimes e , { \boldsymbol{z}_{\mu , k , l } } f } & : { \operatorname{ind}_{h_\mu(l)}^{s_m}{z_\mu(k;l ) } } \ni \varepsilon\otimes e \mapsto { \boldsymbol{z}_{\mu , k , l } } f \in r_{\mu}(k;l)\end{aligned}\ ] ] are isomorphisms of @xmath14-modules .
we consider the case where @xmath228 . in this case , @xmath229 is the coinvariant algebra of @xmath230
. we can decompose @xmath12 as follows : @xmath231 where @xmath232 is the image of the higher specht polynomial introduced in @xcite .
hence we have the decompositions @xmath233 for a standard tableau @xmath234 , the map from the specht module @xmath218 to @xmath235 that maps the specht polynomial @xmath236 to the higher specht polynomial @xmath237 for @xmath238 is an isomorphism . in this case ,
@xmath239 , where @xmath240 is @xmath241 . by direct calculations ,
we have @xmath242 let @xmath243 the polynomials @xmath244 are generic in @xmath245
. hence @xmath246{\boldsymbol{z}_{\mu , k,3 } } \ni { \boldsymbol{z}_{\mu , k,3 } } \mapsto { \boldsymbol{z}_{\mu , k,3 } } f^{(k ) } \in r_{\mu}(k;3)\\ \intertext{and } \varphi_{\varepsilon\otimes e , { \boldsymbol{z}_{\mu , k,3 } } f^{(k ) } } & : { \operatorname{ind}_{h_{\mu}(3)}^{s_m}{z_\mu(k;3 ) } } \ni \varepsilon\otimes e \mapsto { \boldsymbol{z}_{\mu , k,3 } } f^{(k ) } \in r_{\mu}(k;3)\end{aligned}\ ] ] are isomorphisms of @xmath230-modules .
we have the procedure to construct generic elements ( theorem [ procelemma ] ) . in this section ,
we show lemma [ onlysslemma ] , which makes the procedure simpler .
then we show theorem [ tworowscase ] , which explicitly constructs isomorphisms in the case of the springer modules corresponding to @xmath39 .
[ semisstadlemma ] let @xmath258 and @xmath259 be numbers lying in the same row of @xmath186 .
if @xmath258 and @xmath259 lie in the same column of @xmath250 , then @xmath260 acts on @xmath219 by @xmath261 this implies @xmath262 [ onlysslemma ] let @xmath51 and @xmath15 be partitions of @xmath168 .
let @xmath265 be a maximal subset of @xmath89 such that @xmath266 is @xmath221-linearly independent .
let @xmath267 .
then @xmath268 is a semi - standard tableau on @xmath51 with weight @xmath15 . if @xmath269 , then @xmath270 . for all @xmath5 , tableaux that are standardizations of semi - standard tableaux with weight @xmath283 on @xmath275
are only @xmath274 .
since @xmath284 , it follows from a direct calculation that @xmath285 .
hence we have the theorem . | the springer modules have a combinatorial property called `` coincidence of dimensions , '' i.e. , the springer modules are naturally decomposed into submodules with common dimensions .
morita and nakajima proved the property by giving modules with common dimensions whose induced modules are isomorphic to the submodules of springer modules .
they proved that the induced modules are isomorphic to the submodules , by showing the coincidence of their characters .
our aim is to construct isomorphisms between the induced modules and their corresponding submodules in a combinatorial manner . for this purpose ,
we show lemmas , which are equivalent to the classical schur s lemma in special cases . we also give a procedure to construct isomorphisms , and
explicitly construct isomorphisms in the case of the springer modules corresponding to young diagrams of two rows . |
actin network is one of the most relevant structural component inside cytoskeleton of eukaryotic cells , this have a highly dynamic mechanical properties in the cell .
its dynamics is essential to many process , such as cell adhesion , mechanosensing and mechanotransduction , motility , and cell shape among others . at the same time
the development of actin based biomimetic systems at micro and nano - scale demand a deep understanding of their mechanical properties at large deformation where the non - linear mechanics effect encode effects that allow the tuning of unexpected effects @xcite .
in the cell , the dynamics and mechanics of the actin cytoskeleton are regulated by upward sixty known actin - binding proteins ( abps ) defining different levels with emergents behaviour with strong implications in physiological and pathological conditions . in particular the calponin abp was discovered in smooth muscle cells and was studied as possible regulator of actomyosin interaction @xcite .
non - muscle calponin are known to be involved in actin stabilisation , stress - fibres formation and increase the tensile strength of the tissue under strain among others effect @xcite . despite the growing evidence on the role performed by calponin as a structural stabilisers the underlying micro - mechanics still unknown . to study the mechanical effect developed by calponin over the actin structure @xcite studied an in - vitro f - actin network crosslinked with calponin to gain insights about its mechanical properties using large deformation rheology .
they found that the networks with calponin are able to reach a higher failure stress and strain while reducing the pre - strain of the network .
calponin delays the onset of network stiffening , something observed in polymer networks with increased flexibility .
they also observed that the micro - structural origin of these behavior was related to the decrease on the persistence length at single filament level . in order to explain the observed effect reported by @xcite we develop a mathematical model within the framework of continuum mechanics . in this model
, we introduce the hypothesis by which the difference between the two networks , with and without calponin , is interpreted as a alterations in the pre - strain developed by the network entanglement and the regulation of the crosslinks adhesion energy . as a consequence , when the network have a higher internal pre - strain their crosslinks are also near the fluidisation transition . according to the observed stress - strain relationship at the concentrations of cross - linkers and actin used in the experiment
, it can be assumed continuum strain , with elasticity originating from the entropic nature of the individual polymers @xcite . in the first part of this work
we describe the mechanics of actin networks with rigid crosslinks , using the wormlike chain model in the form proposed by @xcite and further developed by @xcite , whereas the network is described using an homogenized continuum framework based on the eight chain network @xcite .
afterwards , we introduce the inelastic effect driven by the crosslinks as alterations in the contour length of the f - actin network . to define the phenomenological law that drives the changes in the contour length , we propose a simple model for the gelation process of the network based on the interactions between the entanglement and adhesion energy of crosslinkers .
finally we compare with experimental measurements performed by @xcite
let @xmath0 be a fixed reference configuration of the continuos body of interest ( assumed to be stress free ) .
we use the notation @xmath1 for the deformation , which transforms a typical material point @xmath2 to a position @xmath3 in the deformed configuration .
further , let @xmath4 be the deformation gradient and @xmath5 the local volume ratio . consider a multiplicative split of @xmath6 into spherical ( dilatational ) part , @xmath7 and a uni - modular ( distortional ) part @xmath8 .
note that @xmath9 . )
eight chains model .
@xmath10 ) semi - flexible bundle filaments in which the contour length is defined as the distance between the crosslinks .
@xmath11 ) unbinding probabiity.,width=226 ] we use the right and left cauchy - green tensors denoted by @xmath12 and @xmath10 , respectively , and their modified counterparts associated with @xmath8 , @xmath13 and @xmath14 , respectively . hence , @xmath15 @xmath16
@xmath17 @xmath18 the mechanical behavior of the f - actin cross - linked network is modelled by means of a strain energy function ( sef ) based on the wormlike chain model for semi - flexible filaments .
this model , proposed firstly by @xcite and later in a polynomial and homogenised form by @xcite , is defined in terms of four physical parameters related to the network architecture and network deformation ( see figure [ dynamic_net ] ) : i ) the contour length , @xmath19 ; ii ) the persistence length , @xmath20 ; iii ) the end - to - end length at zero deformation , @xmath21 , associated with the network prestress @xcite ; and iv ) the macroscopic network stretch from the condition of zero force , @xmath22 .
@xmath23 with @xmath24,\label{eq_w1}\end{aligned}\ ] ] where @xmath25 is the filament end - to - end distance , @xmath26 the first invariant of @xmath13 . using standard procedures from continuum mechanics , the cauchy stress , @xmath27 , can be derived from direct differentiation of eq .
[ eq_w ] with respect to @xmath12 @xmath28\left[\frac{\frac{l_{c}}{l_{p}}-6\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}{\frac{l_{c}}{l_{p}}-2\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}\right]\left(\bar{\mathbf{b}}-\frac{1}{3}\bar{i}_1\mathbf{i}\right)+p\mathbf{i } , \end{split}\ ] ] where @xmath29 is a constitutive relation for the dilatational part of @xmath27 . for a simple shear test , @xmath30 , and the stress - strain relation results @xmath31\left[\frac{\displaystyle{\frac{l_{c}}{l_{p}}}-6\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}{\displaystyle{\frac{l_{c}}{l_{p}}}-2\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}\right]\gamma . \\[2pt]\ ] ] as mentioned previously , in vivo or in vitro actin networks experience prestress during network formation and remodelling due to the entanglement and the formation and disruption of crosslinks . in order to account for these effects ,
we introduce a passive prestress , in the network model , through the parameter @xmath32
@xmath33 where @xmath34 is a dimensionless parameter associated with the passive prestress .
the network is buildup by the interaction between the actin filaments and the crosslinkers .
the nature of this interaction defines the mechanical properties of the structure .
if these interactions are stable ( for the stress and the time scales of the experiments ) , they provide a strong gelation process and the meshwork shows a solid - like behaviour under deformation .
if on the contrary , the crosslinks are not completely stable , but they are associated with a reaction that can proceed in both directions , folding / unfolding or flexible / rigid states of the crosslink , we then speak of a weak gelation process with the meshwork showing a fluid - like behaviour without manifesting a complete unbinding .
clearly if the level of deformation exerted over the crosslink exceeds a given threshold it will break irreversibly .
these effects are account within the model via the contour length , @xmath35 .
we propose the following expression for @xmath35 @xmath36 where @xmath37 defines the unfolding probability encompassing the states of unfolding or flexible cross - link ( see bellow ) , @xmath38 represents the contour length when @xmath39 , and @xmath40 represents the average increment of the contour length when the unbinding probability is one .
chemical crosslinks can be modelled as a reversible two - state equilibrium process @xcite : @xmath41 where @xmath42 the binding probability encompassing the states folding or rigid cross - link . since only these two states are possible , then @xmath43 .
the two - state model has the folded state as the preferred low free energy equilibrium state at zero force and the unfolded state as the high free energy equilibrium state at zero force .
@xmath44 represents the difference in the free - energy between these states , @xmath45 represents the external mechanical work that induce the deformation of the crosslink , and @xmath46 represents the thermal energy .
as we are only able to measure macroscopic quantities as stress and strain and we aim to develop a constitutive model in the mesoscale , we propose the next phenomenological expression , using the previous expression as motivation : @xmath47},\ ] ] where the main driving force is @xmath48 , the average stretch over the bundle . in order to simplify the mathematical treatment , we consider @xmath49 to be proportional to the free adhesion energy @xmath44 .
parameter @xmath50 gives us an idea of the sharpness of the transition between states and @xmath51 is the strain at which the probability of unbinding transition is 0.5 . if @xmath52 , the network is easy to be remodelled showing a more fluid - like behavior . on the contrary , if @xmath53 , the crosslink is more stable and the probability of transition is low .
consequently the network behaves as a solid - like structure .
showing an extension of the solid - like regime.,width=264 ] the model of f - actin network with weak crosslinks subjected to simple shear can be summarized as follows : @xmath31\left[\frac{\displaystyle{\frac{l_{c}}{l_{p}}}-6\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}{\displaystyle{\frac{l_{c}}{l_{p}}}-2\left(1-\frac{\lambda r_{0}}{l_{c}}\right)}\right]\gamma , \nonumber\ ] ] @xmath54 @xmath55}. \nonumber\ ] ] in order to describe qualitatively the behaviour of the coupled set of equations under alterations in the pre - strain and the adhesion energy of the crosslinks we evaluate them in the regime of semi - flexible response i.e @xmath56 .
figure [ quant_resp].a describes the effect of an increment on the pre - strain ( 1+@xmath57 ) on network response , with the remaining parameters kept constant . as can be observed , as the pre - strain increases , the network stiffness increases and is able to reach a higher level of stress ( higher yield point ) .
however , the yielding point ( fluidization of the network ) occurs earlier reducing the solid - like regime of the network .
figure [ quant_resp].b shows the response of the network for different values of @xmath51 .
contrary to the pre - strain , as @xmath51 increases the initial stiffness of the network remains unaltered while the yielding stress and strain increase , extending the solid - like regime .
this implies that as @xmath51 increases the crosslinks become more stable .
the proposed theory is used to describe the experiments conducted by @xcite on the artificially reconstituted f - actin networks crosslinked with filamin , where the network has an actin decorated with and without calponin .
we evaluate the proposed model for the set of parameters shown in table [ table1 ] identified by means of nonlinear least - square fit to experiments of monotonic large deformation stress - strain shear tests from @xcite .
the parameters of the model shown in table [ table1 ] can be divided in two types : ( i ) rigid - wormlike chain parameters @xmath38 , @xmath20 , @xmath40 and @xmath57 which are of the order of magnitude of the values used to describe in experiments of in - vitro f - actin networks and to keep on the regime the regime of semi - flexible entropic elasticity .
( ii ) parameters associated with the dynamics of the crosslinks @xmath50 and @xmath58 .
these parameters encode the transitions to induce fluidisation of the network and represent an indirect measure of the adhesion force of crosslinks .
while @xmath51 describes the transition point in the contour length of the network meshsize ( average distance between crosslinks ) , @xmath50 describes the sharpness of this transition .
these values are phenomenological and were identified in order to fit the experimental data . .
, width=283 ] we compared the results obtained by the model with the average curve of monotonic simple shear at a constant shear strain rate .
the network stiffening response begins at low levels of strain and continues almost linearly until reaching a maximum critical shear stress .
after that point , the response change towards a regime of stress - strain softening where the stress decreases slowly towards zero as the shear strain increases , and the structure flows . for low levels of calponin
the networks present a higher level of pre - strain as noted by the higher slope at low level of deformation , which can also can be observed in the figure [ res].c as a larger value of @xmath59
. for low levels of calponin strain stiffening occurs until @xmath60 , with a maximum stress @xmath61 ( see figure [ res].a ) . on the contrary , for higher levels of calponin , the response shows a lower level of pre - strain ( lower initial stiffness ) , but the yielding point occurs at a larger deformation @xmath62 and higher stress @xmath63 ( experimental values between 3 and 6 @xmath64 ) as shown in figure [ res].a , indicating that calponin extends the solid - like behaviour of the network .
this is a remarkable result with respect to pure f - actine networks for which a lower level of pres - train increases the solid - like regime , but reduces the yield stress significantly .
figure [ res].b shows the scaled stress - strain curve with respect to the yield point .
this figure demonstrates the ability of the model to describe the nonlinear effects .
in order to explain the effects of calponin in stabilisation of the crosslinked f - actin networks our propose that the observed effect which promotes by modification in the flexural rigidity , and as a consequence the persistence length , at single filament will integrate at network scale as a change in the internal pre - stress and the adhesion energy over the crosslinks .
we condense these observations in the figure [ calp_eff ] . during the gelation process
, the network entanglement induces pre - stress across the network which is propagated through the bundles to the chemical crosslinks @xcite .
the trapped stress in the structure is compensated by the deformation of the bundle and the chemical crosslinks . as a consequence ,
it is potentially able to induce conformational changes over the crosslink structure , as was described by @xcite .
the red and black dots in figure [ calp_eff ] describe the effect of the pre - stress on the energy landscape of the chemical crosslinks .
qualitatively , results easy to see that the energy gap is lower under the presence of pre - stress on the crosslink , where the adhesion energy increases , changing from @xmath65 to @xmath44 , with @xmath66 . this can be understood as a combined action of two mechanical regulation pathways over the bonds reaction , where : @xmath67 @xmath45 represents the mechanical work induced by the macroscopic deformation that propagates through the network down to the crosslink .
@xmath69 represents the mechanical work introduced during the entanglement also deforms the crosslink structure .
on one side , without calponin the bundles are less flexible , and therefore network deformation will deform more the crosslink , which in terms of the energy landscape , it appears as tilted down ( se figure [ calp_eff ] ) , and therefore facilitates the conformational change from the folded configuration to the unfolded configuration .
under the perspective of the proposed model , it implies that @xmath51 is shifted to the left , and the pre - strain @xmath70 is higher .
therefore , the network is closer to the fluidisation transition and the maximum stress ( yield stress ) is reduced . on the contrary , if the concentration of calponin is higher , the bundle flexibility increases and persistence length decrease implying that the effects associated with trapped stress will be more concentrated on the bundle deformation than on the crosslinks . in consequence the adhesion energy of the crosslink is higher ( more difficult to jump from the folded to the unfolded configuration ) and the level of shear deformation that the network is able to achieve before the solid - fluid transition results also is also higher . in summary , the proposed model provides arguments to describe the changes observed in the flexibility of the actin bundles encoded the effects at network scale that drives the increment of adhesion and the whole stabilisation of the structure .
this will help to gain better understanding in the complex mechanics behind of the cytoskeleton - like structural building blocks and synthetic bio - structures .
.model parameters to fit the experiments of @xcite [ cols= " < , < , < " , ] [ table1 ]
16 [ 1]#1 [ 1]`#1 ` urlstyle [ 1]doi : # 1 m. jensen , j. watt , j. hodgkinson , c. gallant , s. appel , m. el - mezgueldi , t. angelini , k. morgan , w. lehman , and j. moore .
effects of basic calponin on the flexural mechanics and stability of f - actin .
_ cytoskeleton _ , 690 ( 1):0 4958 , 2012 . h. lpez - menndez and j. f. rodrguez .
microstructural model for cyclic hardening in f - actin networks crosslinked by @xmath71-actinin .
_ journal of the mechanics and physics of solids _ , 91:0 2839 , 2016 .
jh shin , ml gardel , l mahadevan , p matsudaira , and da weitz .
relating microstructure to rheology of a bundled and cross - linked f - actin network in vitro .
_ proceedings of the national academy of sciences of the united states of america _ , 1010 ( 26):0 96369641 , 2004 . | the synthetic actin network demands great interest as bio - material due to its soft and wet nature that mimic many biological scaffolding structures .
the study of these structures can contribute to a better understanding of this new micro / nano technology and the cytoskeleton like structural building blocks .
inside the cell the actin network is regulated by tens of actin - binding proteins ( abp s ) , which makes for a highly complex system with several emergent behavior .
in particular calponin is an abp that was identified as an actin stabiliser , but its mechanism is poorly understood .
recent experiments using an in vitro model system of cross - linked actin with calponin and large deformation bulk rheology , found that networks with basic calponin exhibited a delayed onset and were able to withstand a higher maximal strain before softening . in this work
we propose a mathematical model into the framework of nonlinear continuum mechanics to explain the observation of jansen , where we introduce the hypothesis by which the difference between the two networks , with and without calponin , is interpreted as a alterations in the pre - strain developed by the network entanglement and the regulation of the crosslinks adhesion energy .
the mechanics of the f - actin is modelled using the wormlike chain model for semi - flexible filaments and the gelation process is described as mesoscale dynamics for the crosslinks .
the model has been validated with reported experimental results from literature , showing a good agreement between theory and experiments .
f - actin networks , calponin , chemical crosslinks , adhesion , non - linear rheology , allosteric materials |
in the hot and dilute astrophysical plasmas , transport is mainly due to the interaction of the charged particles with their collectively self - generated electromagnetic turbulence .
the latter replaces the binary particle collisions responsible for the transport in denser and colder plasmas .
different than those collision - dominated plasmas , the properties of the collisionless self - generated turbulence depend , however , on the parameters and configuration of the plasma . `` anomalous '' ( or turbulent ) transport depends , therefore , also on the plasma parameters and configuration . here , we focus on the role of anomalous ( turbulent ) transport in collisionless guide - field magnetic reconnection . in space and astrophysical
plasmas , reconnection ubiquitously converts magnetic energy into particle acceleration , bulk flows and heat . in many environments , like ,
e.g. , stellar atmospheres , magnetic reconnection develops in external , current - aligned , magnetic fields . because of this , we investigate , the influence of the strength of the external guide - magnetic - field on the generation of the turbulence in collisionless current sheet ( cs ) regions prone to magnetic reconnection .
note that guide field reconnection is now investigated in situ by the mms space mission@xcite and by laboratory experiments.@xcite .
so far , the properties of collisionless guide - field reconnection were analyzed theoretically and by numerical simulations using a number of different simplifying plasma models ( see , e.g. , refs . and references therein ) .
a number of micro - instabilities was identified which can provide , in principle , anomalous transport for collisionless reconnection .
its dependence on the guide field strength , however , has not been completely understood , yet .
this is particularly true with respect to regions outside the immediate vicinity of the x - line , near the separatrices and in the exhaust ( outflow ) region of reconnection .
also , it appeared to be necessary to include the consequences of the non - linear evolution of the unstable plasma waves beyond the limits of the quasi - linear theory . to understand these issues
, we carried out 2.5d particle - in - cell ( pic ) code simulations varying the strength of the external ( guide- ) field . to verify our approach
, we compare our results with those obtained in the limiting case of antiparallel reconnection .
we self - consistently obtain the non - linear evolution of micro - instabilities to turbulence and investigate its influence on the balancing of macroscopic electric fields in collisionless guide - field reconnection ( see sec .
[ sec : ohm ] ) . to further verify our findings , we independently calculate also the correlations of electromagnetic field and the turbulent plasma fluctuations ( see sec . [
sec : resistivity_fluctuations ] ) as well as the slowing - down rate of the mean current ( see sec . [
sec : momentum_exchange ] ) . since the 1970 s , different mathematical approaches and formalisms have been developed to describe the anomalous transport in collisionless plasmas .
they mainly assumed weak turbulence so that quasi - linear approximations could be applied ( see , e.g. , refs . , and references therein ) . according to these results ,
collective - collisionless - plasma - transport effects can be described by an effective anomalous resistivity @xmath0 relating linearly ( or tensorially ) electric field and current density as @xmath1 , providing a term for a resistive mhd - ohm s law . such anomalous resistivity term ( which includes @xmath0 ) of the mhd - ohm s law is meant to describe the consequences of the interaction of the charged particles with the self - generated turbulence , replacing the binary coulomb collisions which dominate in cold and dense plasmas.@xcite early simulations aiming to obtain self - consistently @xmath0 used simplified setups , designed to explicitly consider specific instabilities and to verify predictions of the quasi - linear theory ( see , e.g. , refs . ) .
the investigations of more realistic scenarios , including the consideration of the nonlinear development in the course of magnetic reconnection , had to wait until more computational power became available .
note that micro - turbulence can break the frozen - in condition of ideal mhd in the momentum equation and heat the plasma as collisional joule heating not only at the x - lines , but also near the separatrices and in the outflow region of reconnection .
cluster@xcite and more recently the mms space mission@xcite provided positive observational evidence for the role of micro - turbulence and anomalous resistivity in reconnection regions
. there is , however , a controversy that some authors claim the contribution of anomalous resistivity to the reconnection electric field is negligible , interpreting measurements of the polar@xcite , themis@xcite and also cluster@xcite missions .
evidence for anomalous resistivity was , however , found also in magnetic reconnection experiments such as mrx ( see refs . and references therein ) .
note that these investigations revealed that models of anomalous resistivity do not explain sufficiently well the experimental results.@xcite considering 2.5d configurations we reach similar conclusions . in this paper
we will apply a mean - field approach to derive terms of a generalized ohm s law .
let us first briefly review the mathematical framework for the calculation of the terms of such equation . in our mean - field approach ,
we split off the fast fluctuations of the evolution of the slowly varying mean quantities . and
in fact , macroscopically observable electromagnetic fields , plasma parameters and variables , are mean quantities slowly varying compared to fast fluctuations , which include microscopic variables .
hence , let us represent any macroscopically relevant physical quantity @xmath2 by a slowly varying and spatially averaged ( over the micro - scales of fast variation ) mean value @xmath3 , and a local and instantaneous deviation from it @xmath4 : @xmath5 in eq . , the brackets @xmath6 stand for an appropriate time and spatial averaging which reveals mean quantities .
the spatial and time scales of the averaging should be chosen in a way that they do not exceed the scales of approximate homogeneity and stationarity of the variation of @xmath2 .
on the other hand , they should be large and long enough , to average over the local and fast variations , including microscopic fluctuations , of the turbulence ( see sec .
[ sec : convergence_fluctuations ] ) . for a collisionless plasma , the vlasov equation describes the evolution of the single particle distribution function @xmath7 for every specie @xmath8 in the self - consistently determined electromagnetic fields @xmath9 and @xmath10 .
the vlasov equation includes the detailed variations including that of the fast turbulent fluctuations .
its right - hand - side ( r.h.s . ) vanishes if the particle collisions are not explicitly considered . from the vlasov equation
, the following boltzmann equation can be derived after writing each term in the form given by eq .
: @xmath11\langle f_{\alpha}\rangle
\\ \nonumber & = \left(\dfrac{\partial \langle f_{\alpha}\rangle}{\partial t}\right)_{\rm anom } , \\ & = -\frac{q_{\alpha}}{m_{\alpha}}\left\langle \left ( \delta \vec{e } + \vec{v } \times \delta \vec{b}\right)\cdot \frac{\partial \delta f_{\alpha}}{\partial \vec{v } } \right\rangle.\end{aligned}\ ] ] the left hand side ( l.h.s . ) of eq .
[ vlasov_anomalous ] contains the mean field quantities @xmath12 , @xmath13 , @xmath14 . the r.h.s . of this boltzmann equation
resembles a collision term , describing the higher order correlations between the split - off fluctuations of the electromagnetic fields ( @xmath15 , @xmath16 ) , the distribution function @xmath17 ( e.g. , density @xmath18 , bulk flow velocity @xmath19 , pressure tensor @xmath20 , etc . ) and its derivatives @xmath21 .
taking the first order velocity momenta of eq . and considering only first - order fluctuations ( see the articles ref . , the reviews refs . , or the textbook ref . and references therein ) one obtains , for an electron - ion plasma ( with subscripts `` e '' and `` i '' , respectively ) , a generalized two - fluid ohm s law for the mean field quantities : @xmath22 in eq . , @xmath23 and @xmath24 are summation indices , @xmath25 is the @xmath26component of the macroscopic convective electric field , @xmath27 is the electron pressure - tensor term while @xmath28 is usually called the `` ( electron ) inertial term '' .
in particular , the `` anomalous resistivity '' term @xmath29 can be written as:@xcite @xmath30 note that eq
. corresponds to the assumption that the electric field is linearly ( or tensorially ) related to the current density .
meanwhile , eqs . and
are directly derived from the averaged vlasov equation . in this work ,
we calculate the r.h.s .
term of eq .
for the numerically simulated quantities by appropriately averaging ( see above ) over the correlated fluctuations . while these quantities previously were determined mainly by quasi - linear calculations based on estimates of a ( weak ) turbulence saturation in idealized scenarios ( see the review ref . , the article ref . , or the textbooks ref . or ref . , and references therein )
, we obtain them directly and self - consistently .
the setup of our simulations is described in ref . .
we use the pic - code acronym.@xcite since we aim at the investigation of quasi-2.5-dimensional reconnection , we neglect variations along the @xmath31 direction .
two css are initialized forming a double harris - sheet equilibrium.@xcite the external guide - field strength @xmath32 along the @xmath31 direction ( perpendicular to the @xmath33@xmath34 reconnection plane ) , is varied in a range from @xmath35 ( for the antiparallel reconnection limit ) and @xmath36 to simulate scenarios like stellar coronae and laboratory experiments . here
we express the guide field strength normalized to the amplitude of the antiparallel component ( in the @xmath34 direction ) of the harris - cs @xmath37 ( i.e. , @xmath38 ) .
the cs parameters were : halfwidth @xmath39 , mass ratio @xmath40 , frequency ratio @xmath41 , temperature ratio of @xmath42 and a background plasma density @xmath43 .
these parameters give an electron thermal speed of @xmath44 , where @xmath45 is the speed of light .
other physical parameters are defined as : @xmath46 is the electron / ion plasma density of the current - carrying population at the center of the cs , @xmath47 is the ion / electron skin depth , @xmath48 the ion / electron plasma frequency calculated with the density @xmath49 and @xmath50 is the electron plasma frequency in the harris cs magnetic field amplitude @xmath37 .
the numerical parameters of our simulations were : @xmath51 and @xmath52 particles per cell for the current - carrying and background population , respectively ( for both electron and ion population ) ; a simulation box size of @xmath53 in the @xmath33@xmath34 plane , the grid spans @xmath54 grid points , periodic boundary conditions apply in the @xmath33 and @xmath34 directions , the spatial ( grid ) resolution was @xmath55 , where @xmath56 is the electron debye length in the center of the cs ( calculated with @xmath49 ) .
the timestep is chosen to satisfy the courant - friedrichs - lewy ( cfl ) condition for light wave propagation at a level of @xmath57 .
reconnection is initialized by a small - amplitude long - wavelength perturbation to quickly reach a fully developed stage with x - lines at @xmath58 and @xmath59 for every cs .
note that we illustrate our analysis in this paper using the results obtained for the cs centered around @xmath60 .
we define a reference electric field as @xmath61 used for the normalizations , with @xmath62 the alfvn speed calculated with the central density @xmath49 and the asymptotic in - plane magnetic field strength @xmath37 .
we calculate the mean quantities appearing in the generalized ohm s law eq . by means of a ( running ) time average with a time windows of @xmath63 , where @xmath64 is the lower hybrid frequency calculated in the total magnetic field .
this filtering damps out fluctuations with frequency significantly higher than the inverse of the time windows length .
the windows length is chosen to cover all relevant plasma frequencies while numerically caused fluctuations are removed .
more consequences of this choice are discussed in sec .
[ sec : convergence_fluctuations ] .
first , in fig . [ fig : ohm_lhs ] ,
we show the guide field dependence of the spatial distribution of the ( out - of - plane ) @xmath65component ( time averaged ) dc electric fields , @xmath66 , in the left hand side ( l.h.s . ) of the ( time averaged ) generalized ohm s law eq . .
this sum describes the mean non - ideal electric field .
it can , therefore , be used as signature of the non - ideal processes violating the frozen - in condition of the electron fluid to the plasma .
note that we use different times in each case to compensate for the delay in the reconnection saturation time for stronger guide - fields ( see , e.g. , refs . ) .
the plots are shown when the magnetic islands reach comparable sizes . in all these contours plots we use , as a post - processing step , a spatial ( low - pass ) gaussian filter with a small width @xmath67 to get rid of the pic shot noise with small wavelengths . fig .
[ fig : ohm_lhs](a ) shows that in antiparallel reconnection and for very small guide fields , the non - ideal terms are located mostly in the diffusion region near the x - line of reconnection .
this is in accordance with the previous findings that the non - ideal contributions due to the off - diagonal terms of the electron pressure tensor or the electron inertia terms balance the reconnection electric field @xmath68 ( see a review in , e.g. , ref . ) .
however , for guide fields of the order of @xmath69 ( fig .
[ fig : ohm_lhs](c ) ) , the deviations from non - ideal behavior shift to the separatrices , in particular the low - density one , with a `` patchy '' spatial distribution .
therefore , dissipation and violation of the frozen - in condition can also occur away from the traditional diffusion region near the x - line , as discussed in some previous works@xcite .
note that the low - density separatrix is the region around the antisymmetrically located upper right and lower left separatrix arms in our setup , characterized by a diminished density in comparison with the other separatrix , as typically seen in guide - field reconnection.@xcite for larger guide fields of the order of @xmath70 ( fig .
[ fig : ohm_lhs](d ) ) , the non - ideal contribution close to the x - line is much less important and spatially smaller than in the low density separatrix .
cases with guide field @xmath71 , such as @xmath36 ( fig .
[ fig : ohm_lhs](f ) ) , show a diminished non - ideal activity even in the separatrices .
the reason is because the ( streaming ) instability producing the non - ideal behavior is active only in the strong guide field regime @xmath72 .
the upper limit seems to be a result of the reduced 2.5d geometry used in the simulation , as discussed in our previous work.@xcite fig . [
fig : ohm_diff ] displays the difference between the l.h.s and r.h.s . of the @xmath65component of the mean generalized ohm s law eq . , excepting the a priori unknown anomalous electric field term .
any deviation can be attributed to non - ideal effects beyond the caused by the pressure or inertia term contributions to the electric field , and are thus equal to @xmath73 in eq . .
for the antiparallel @xmath35 ( fig .
[ fig : ohm_diff](a ) ) or small guide fields @xmath74 cases ( fig .
[ fig : ohm_diff](b ) ) , the deviations are small in comparison to the non - ideal electric field @xmath68 and can be attributed to the pic noise .
this is justified by comparing with the values of this difference at points in the inflow region , away from the cs . however , for stronger guide fields such as @xmath70 ( fig .
[ fig : ohm_diff](d ) ) , the deviations start to form localized and `` patchy '' structures near the separatrices .
they have smaller spatial extent than the structures seen for the l.h.s . of the ohm s law in fig .
[ fig : ohm_lhs](d ) .
this is also in agreement with previous works showing the relative contributions of the instantaneous ohm s law in the separatrices of guide - field reconnection.@xcite all these observations prove that that non - ideal behavior seen in the latter can be explained mostly by the two - fluid effects of pressure and inertia contributions to the electric field ( plots not shown here ) , being violated only very locally . in fig .
[ fig : ohm_diff_profile ] , we complement the information of fig .
[ fig : ohm_diff ] by analyzing the individual contributions to the balance of the mean ohm s law terms for a @xmath33-cut at the separatrices .
we have applied an additional savitzky - golay spatial filter with a small width to smooth out those profiles , without affecting significantly the values of the local maxima / minima . in the past , many works have investigated the relative contributions of the generalized ohm s law terms to the electric field in guide field reconnection near the x - line , in order to find the mechanism leading to the breaking of the frozen - in condition ( see , e.g. , refs . ) .
note , however , that most of those works used the instantaneous instead of the time averaged version of the generalized ohm s law used here , not allowing for additional terms coming from higher order fluctuations . for guide fields @xmath75 ( fig .
[ fig : ohm_diff_profile](a)-(c ) ) , most of the out - of - plane @xmath68 is balanced by the convective electric field , with small contributions from pressure or inertia terms .
note that even small guide fields ( @xmath74 , fig .
[ fig : ohm_diff_profile](b ) ) introduce an asymmetry with respect the cs mid - plane in all the electric field terms , with a significant contribution from the pressure term .
but it is only with guide fields stronger than unity ( e.g. , @xmath70 , fig .
[ fig : ohm_diff_profile](d ) ) that the inertia term becomes relevant , taking similar peak values to both pressure and convective electric fields , well above the noise level .
this is especially noticeably in the region close to the low density separatrix ( at @xmath76 ) . associated with this fact are the largest deviations in the balance of the mean field generalized ohm s law ( on the order of @xmath77 )
this implies that the anomalous term in eq . reaches the strongest values among all the cases analyzed here , with a corresponding largest anomalous resistivity @xmath78 . in this case of @xmath70
, we verified that among the three contributions to the inertia term , i.e. , @xmath79 , the most important is the ( eulerian ) time derivative ( third term ) .
this indicates that the out - of - plane electron current @xmath80 is changing quickly , signature of an efficient momentum transfer between electron and ions . indeed , by multiplying the mean field vlasov equation by @xmath81 and integrating in the velocity space , it is possible to relate directly the slowing - down rate of the ( mean ) electron current density with an effective collision frequency @xmath82 for each specie @xmath83 along the direction @xmath23,@xcite @xmath84 as a result , the effective resistivity along the @xmath23-direction is:@xcite @xmath85 this approach has been used , e.g. , by refs .
for the analysis of vlasov code simulations of current - driven instabilities , in refs . for 3d vlasov simulations of cs , and in refs .
for 2d pic simulations of cs . because the contribution from the ion quantities is much smaller than the electron ones,@xcite the @xmath86component of the anomalous electric field can be approximated as , @xmath87 j_{j}. % \end{aligned}\ ] ] then , @xmath88 is related to the time derivative of the electron current density and thus to the ( eulerian ) partial time derivative term @xmath89 of the inertia term of the mean generalized ohm s law eq . .
the difference is the contribution coming from the electron density @xmath90 and its time derivative , not appearing in the inertia term .
the spatial distribution of @xmath91 can be seen in fig .
[ fig : eanom_ddt_bgs ] .
we avoid the calculation in regions where the mean electron current density @xmath92 is very close to zero by assigning a zero value to the total quotient , in order to avoid unphysical diverging values of the electric field due to numerical noise .
[ fig : eanom_ddt_bgs](a)-(b ) show small contributions to the mean @xmath93 in the separatrix region for small guide fields . for these cases
, @xmath91 is mostly concentrated around the x - line .
larger guide fields ( @xmath69 , fig . [
fig : eanom_ddt_bgs](c ) ) localize the contributions of this term to a very small area close to the x - line , with a slight increase at the separatrices . again , only for @xmath94 ( fig .
[ fig : eanom_ddt_bgs](d ) ) , the contribution of this term mostly correlates with the locations at the separatrices where the balance of the mean field ohm s law is violated ( compare with fig . [
fig : ohm_diff](d ) ) .
note , however , that the spatial distribution of @xmath91 is more spread and with higher values than @xmath95 given by the difference in the mean field ohm s law .
therefore , it can not account for all the deviations in the mean ohm s law .
but it proves the existence , instead , of an effective transfer of momentum between electron and ions with a `` patchy '' spatial structure ( alternating sign ) .
stronger guide fields ( @xmath96 , fig . [
fig : eanom_ddt_bgs](e)-(f ) ) , reduce both values and `` patchiness '' of @xmath91 .
this correlates with the behavior shown by @xmath95 in fig .
[ fig : ohm_diff](e)-(f ) , due to the weakening of the instabilities producing the fluctuations .
in the remaining part of this paper , we are going to characterize in more detail the physics of the processes leading to the deviations in the mean field ohm s law for the case @xmath70 at the separatrices . the enhanced wave - particle interactions in the separatrices for the case @xmath70 do not only lead to an alternating effective resistivity in the @xmath65direction between @xmath97 and @xmath68 .
indeed , the relation between the @xmath33 and @xmath34 components of the current density and electric field also behaves in a very similar manner , and with similar values as well .
however , the deviations from the mean generalized ohm s law do not provide a clear proof of this , possibly because the typical values of @xmath98 and @xmath99 are about 20 times larger than @xmath68 .
but the time derivative of the electron current density in these ( in - plane ) @xmath100 and @xmath101 directions do provide an effective resistivity of the same order of magnitude as this for the @xmath65component .
this effective resistivity has a spatial distribution concentrated near the separatrices , while the noise is more dominant in the outflow region .
this can be seen in the contour plots of @xmath102 and @xmath103 shown in fig .
[ fig : einertia_bg3_components ] and comparing with @xmath91 in fig .
[ fig : eanom_ddt_bgs](d ) .
note that both @xmath100 and @xmath101 components are noisier than the @xmath65component , because @xmath104 and @xmath105 , appearing in the denominator of eq .
, are closer to zero in larger regions inside the cs than @xmath106 . in summary ,
the similar behavior of all the components of the anomalous electric field implies that the momentum exchange between electrons and ions does not have a clear preferential direction as one would expect in a guide field situation , but rather tends to be isotropic near the separatrices .
the locations of preferential momentum transfer at the separatrices for the case @xmath70 do not only alternate sign in space , but they also do in time .
this can be easily seen by tracking the values of @xmath107 at a point in the low density separatrix located at @xmath108 , @xmath109 .
the results for the three components of this anomalous electric field are shown in fig .
[ fig : edamping_timeseries_bg3 ] .
the values start to significantly increase after @xmath110 , with average peaks of about @xmath111 .
this corresponds to an equivalent collision frequency @xmath112 ( given by eq .
[ nu_anom_ddt ] ) with values around @xmath113 of the lower hybrid frequency @xmath114 ( plots not shown here ) .
note , however , that due to the quasi - periodic nature of these fluctuations , there is no net dissipation that can be associated to this anomalous electric field .
[ fig : e_anom_ddt_stack ] shows the time evolution of a @xmath33-cut of this anomalous electric field across the cs midplane during fully developed reconnection .
the fluctuations of @xmath107 move along the separatrices , outwards from the x - line , associated with the motion of the dominant electron beam producing the streaming instability originating all these processes .
the low density separatrix is located at @xmath115 for @xmath116 and reaches @xmath117 for @xmath118 , due to the growth of the magnetic island .
note that a long enough time average following a point near the low density separatrix ( with positive values of , e.g. , @xmath91 ) would have a net positive value , implying a net effective dissipation . in (
our ) previous work,@xcite we provided evidence that the instability leading to the development of electric field fluctuations at the separatrices ( and also in the exhaust region ) , is a streaming instability due to counterstreaming electron beams generated only for a regime of guide fields @xmath72 .
indeed , the entire region where this instability takes place shows significant deviations from gyrotropy . ref . showed that the typical fluctuation frequency is broadband up to @xmath119 , as expected from this kind of instability .
thus , fluctuations around the lower hybrid frequency can be characterized as an effective collision frequency of the same order of magnitude , even though they arise from very different physical processes .
the fact that the effective collision frequency increases with the lower hybrid frequency was also shown observationally , experimentally , and numerically ( see ref . and references therein ) .
we also found that in the same locations where the peak values of the anomalous resistivity ( either positive or negative ) are maximal / minimal , the traditional dissipation @xmath120 and the dissipation in the electron frame of reference@xcite @xmath121 are also maximal / minimal ( plots not shown here ) .
this further validates that there is an effective exchange of energy between the electromagnetic fields and the particles , a signature of local wave - particle interactions .
note , however , that there is no overall dissipation as it would be with only positive values of anomalous resistivity , and so no net collisionless transport during the whole duration of the reconnection process .
note that the two approaches mentioned before for the calculation of the anomalous resistivity the remaining term of the mean field generalized ohm s law and the slowing - down rate of the electron current involve only mean quantities .
now , we show the information contained in the so far neglected part : the fluctuations ( given by the r.h.s .
of eq . ) causing the anomalous resistivity in the separatrices for the case @xmath70 .
previous works have quantified mostly spatial fluctuations , in 2d@xcite and also 3d pic simulations of cs.@xcite these works used a spatial average in the equilibrium current direction ( @xmath122 ) , where turbulence due to micro - instabilities was assumed to be homogeneous .
instead , especially considering that a cs is inhomogeneous , we calculate these fluctuations as the rms fluctuation values , defined as the standard deviation of each point in the corresponding time series , with the mean taken as the running average mentioned before
. the time average can be equivalent to the spatial one if the turbulence is both stationary and homogeneous ( and both equivalent to an ensemble average if the processes are ergodic@xcite ) .
note that this time averaging approach is also practically universally used in spacecraft data analysis , since it is often the only measurement available .
the results of our time averaging procedure are shown in fig .
[ fig : location_fluctuations_bg3 ] , using the same time average @xmath123 as before ( @xmath63 ) .
the dominant in - plane electric field rms fluctuations @xmath124 and @xmath125 ( fig .
[ fig : location_fluctuations_bg3](a)-(b ) ) are located mostly in the separatrices .
density fluctuations @xmath126 ( fig .
[ fig : location_fluctuations_bg3](d ) ) are more significant only inside the exhaust region , less important near the separatrices . therefore ,
correlated fluctuations @xmath127 and @xmath128 are expected to be important only in the separatrices .
no significant level of in - plane electrostatic fluctuation is observed close to the x - line . on the other hand
, there is no significant fluctuation of the out - of - plane electric field @xmath129 at any location that can be clearly distinguished from the surrounding noise level .
note also that the typical rms values of this component are much smaller than those of the in - plane components @xmath130 ( by a factor of 2.5 - 3 ) .
therefore , there is no significant contribution from the correlated term @xmath131 to the mean reconnection electric field @xmath93 .
[ fig : fluctuations_timeseries_bg3 ] shows the time series of the electric field due to correlated fluctuations @xmath132 , at the same point in the separatrices used for @xmath107 in fig .
[ fig : edamping_timeseries_bg3 ] . comparing with the latter
, we can see the electric field due to fluctuations shows peaks differing significantly among all the components .
as expected , the component @xmath133 is very small compared to the in - plane components .
although the reason of such disagreement between the values of the anomalous resistivity given by @xmath107 and @xmath132 is not clear at the moment , we speculate that it might be due to the tensor nature of the effective resistivity , since we consider it as an scalar quantity in all our calculations . in such a way , the electric field might have contributions from all components of the current density , modifying their final values , especially for @xmath132 .
[ fig : fluctuations_timeseries_bg3 ] also shows that , in general , @xmath132 tends to have more positive than negative values for later times , providing a net transport due to anomalous resistivity , but mostly in the in - plane directions .
the other critical difference by comparing @xmath132 and @xmath107 is that the latter ( shown in fig .
[ fig : edamping_timeseries_bg3 ] ) has typical values larger by one order of magnitude .
this fact illustrates that different methods to calculate anomalous resistivity based on ( the same ) time averages do not necessarily agree each other , a fact that has not been considered properly in previous works .
this has to be taken into account when extrapolating conclusions based on this methods when applied to other scenarios with more significant anomalous resistivity . in any case , these very small values due to correlated fluctuations are not enough to explain the deviations in the mean field generalized ohm s law or to contribute significantly to the mean electric field .
this is in agreement with ref . , who also found very small values of anomalous resistivity when performing a time average , in comparison with a calculation based on a spatial average .
we also show in fig .
[ fig : fluctuations_timeseries_bg3 ] the relative contributions from the first and third term to the total electric field @xmath132 in eq . .
in general , the electromagnetic contributions from the fluctuations of the magnetic field @xmath134 with the electron current @xmath135 are one order of magnitude smaller than the electrostatic ones ( @xmath136 ) for the in - plane components .
this is to be expected , since the electromagnetic correlated fluctuations were shown to be smaller in low - plasma-@xmath137 conditions,@xcite like in our case with a strong guide field . in spite of this , we also find a significant contribution from the electromagnetic term for the @xmath65 component of the anomalous resistivity calculated from the correlated fluctuations , since @xmath138 is very small .
all the calculations shown before rely on the choice of a given length for the time windows used for splitting mean from fluctuating quantities , being somewhat arbitrary to define the separation between both .
the values of all quantities shown before will vary by changing this time windows .
this is because , in a kinetic approach , the separation of scales introduced by the mean and fluctuating quantities is not unique .
this is specially critical when there is no clear scale separation between average and fluctuating quantities in eq . , making eq . possibly invalid . as a consequence ,
the quasi - collision term in the boltzmann equation do vary for different choices , affecting the macroscopic description .
this fact was already considered since a long time ago@xcite ( see also ref . ) and it continues being discussed nowadays when applied to in - situ measurements in the solar wind.@xcite for example , the same measurement process of electron distributions functions is by definition non - ergodic,@xcite therefore affecting any kind of calculations involving the macroscopic momenta of this quantity . in all our previous calculations , we chose @xmath139 as a representative time average windows , since it is one of the smallest values where the fluctuations due to the noise are significantly smaller than the fluctuations of interest ( broadband up to the lower hybrid frequency ) . also , the mean and fluctuating quantities are not too sensitive to small changes around this value .
[ fig : ohm_diff_convergence ] show the difference between the l.h.s and r.h.s . of the @xmath65component of the mean generalized ohm s law eq .
( same quantity shown in fig .
[ fig : ohm_diff](d ) ) , for the case @xmath70 and varying time windows over one order of magnitude . smaller time windows ( @xmath140 , fig .
[ fig : ohm_diff_convergence](a ) ) are unreliable because they produce quantities with mostly fast fluctuations , in regions away from the current sheet , due to numerical noise and not due to relevant physical processes .
larger time windows ( @xmath141 , fig . [
fig : ohm_diff_convergence](c)-(d ) ) smooth out small scale structures in the separatrices .
we do not show even larger values of time windows , close to reconnection time scales , since they would produce unreliable results because the current sheet and the location of separatrices change location during the development of reconnection ( see fig .
[ fig : e_anom_ddt_stack ] ) . a large time average would be meaningful under conditions of stationary reconnection , which would require a much larger simulation box and/or non - periodic boundary conditions .
the profiles shown in fig .
[ fig : ohm_diff_profile_convergence ] reveals complementary information to that displayed in fig .
[ fig : ohm_diff_convergence ] .
note that , similarly to fig .
[ fig : ohm_diff_profile ] , we have also applied a spatial savitzky - golay filter .
the peak values of the remaining term in the mean ohm s law are always at the low density separatrix , and decrease linearly for increasing time windows . on the other hand , and at the same time , the anomalous electric field becomes more comparable to the peak values of the pressure and inertia term contributions to the mean electric field when using larger time windows , since their maximum values decrease faster than in a linear fashion .
this can also be understood in the sense that larger time windows take into account a broader spectrum of fluctuating frequencies , and then the relative contribution to the mean electric field should be larger than smaller time windows which only consider high frequencies not caused by streaming instabilities in the lower hybrid frequency range . in order to see the time evolution of the correlated fluctuations , in fig .
[ fig : anom_ez_timeseries_bg3](b ) we show the electric field due to this contribution @xmath142 . as expected , the peak values of this anomalous electric field do not contribute significantly to the mean electric field @xmath93 , shown in fig . [
fig : anom_ez_timeseries_bg3](a ) , or to the non - ideal electric field in the l.h.s . of the mean ohm s law eq .
( plot not shown here ) .
the difference is about two orders of magnitude .
those plots also confirm that larger time windows exhibit larger values of the anomalous resistivity , but in any case not varying significantly for later times , as well as more stable over large time periods and with a tendency toward positive values .
this same behavior is also shown , but even in a more strongly manner , by the @xmath143 component of the anomalous resistivity due to correlated fluctuations in fig .
[ fig : anom_ex_timeseries_bg3](a ) .
note that all the values are enhanced in more than one order of magnitude compared to the @xmath65component .
finally , we have also verified that the time average of all fluctuating quantities ( @xmath144 ) in the chosen range of time windows is close to zero , i.e. , the reynolds rules ( see , e.g. , appendix a2 in the textbook ref . )
are well satisfied ( plots not shown here ) .
we investigated the consequences of the self - generated plasma - turbulence for the transport of the current - carrying charged particles in two - dimensional guide - field reconnection . in order to derive an `` anomalous '' effective resistivity due to the turbulence
, we utilized a mean - field approach .
for this sake we split - off the fast , micro - turbulent fluctuations from the mean , slowly varying variables quantities describing the macroscopic evolution of the plasma .
we based our analysis on the results of fully kinetic 2.5d particle - in - cell ( pic ) code ( acronym ) simulations .
based on those data , we first calculated the contributions to the balanced slowly varying ( mean ) electric field in the framework of a generalized ohm s law description .
we then compared our findings first with the slowing - down rate of the macroscopic ( mean ) current and also by calculating the first order correlations of the micro - turbulent fluctuations .
the latter method provides a collision term due to the fluctuations in the right hand side of a boltzmann equation for the mean quantities , derived from the vlasov equation for the fully kinetic variables .
we found , by all the previous methods , that in two - dimensional collisionless guide - field reconnection , the self - generated kinetic - scale micro - turbulence does not significantly contribute to the balancing of the mean electric fields , neither near the reconnection x - line nor near the separatrices and the exhaust region of reconnection . for strong external guide magnetic fields ( @xmath72 )
, we found , however , that the self - generated turbulence near the `` low - density '' separatrices provides non - negligible contributions to the mean electric field .
there , the energy exchange between the particles and the mean , averaged over the turbulence , electric fields , is `` patchy '' in space and oscillating in time due to the unstable plasma waves caused by a counterstreaming electron beam instability@xcite .
part of the energy exchange is , however , irreversible and the corresponding slowly varying ( mean ) electric field can be related to be due to a `` anomalous '' resistivity produced by the turbulence nevertheless , for most of the reconnection regions in guide fields in the range @xmath72 , the electron inertia dominates the slowly varying , macroscopic ( mean ) electric field . in the ohm s law , this contribution is dominant due to the large partial eulerian time - derivative of the mean electron bulk - drift velocity .
note that only near the `` low density '' separatrices the inertia term is comparable to the pressure and the convective electric field terms . for smaller and vanishing guide fields @xmath145 , however , the contribution of the electron inertia to the mean electric field is negligibly small , not only near the reconnection x - line but also at the separatrices and in the exhaust region of reconnection .
an independent estimate for the effective resistivity is the slowing - down rate of the mean velocity of the main current carriers , the electrons .
the resulting `` anomalous '' resistive ( turbulent ) contribution to the electric field is spatially spread .
it slightly differs from the results for the balanced mean electric field obtained by calculating the terms of the generalized ohm s law .
however , the slowing - down rate of the mean electron current , associated to the electron - ion momentum transfer , is largest where the anomalous electric field via the calculation of the mean generalized ohm s law is strongest .
the resulting slowing - down rate of the current , if attributed to an effective `` collision frequency '' , ranges from the ion - cyclotron to the lower - hybrid frequency .
note that the electron inertia term is not directly proportional to the electron - current slowing - down rate .
in addition to the out - of - plane electric field field @xmath68 , we further investigated the contribution of the self - generated turbulence to the electric field balance and energy transfer via the in - plane electric field components @xmath98 and @xmath99 .
the momentum - exchange between the electron and ions in those directions appeared to be comparable to the one in the @xmath31 direction near the `` low density separatrices '' , i.e. , there the momentum transfer is isotropic .
a large portion of the energy exchange is reversible and quasi - periodic as indicated by a changing sign of @xmath120 .
a smaller part of the energy exchange corresponds to a net irreversible energy loss , i.e. , to dissipation .
note that the latter should be detected with a time averaging window which is large enough to take into account the broad turbulence spectrum below the lower - hybrid frequency , generated by the beam instability of electrons flowing along the separatrix .
we further verified the enhancement of the momentum transfer due to wave - particle interaction near the separatrices of strong - guide - field reconnection by another independent method , i.e. , calculating the first order correlations of field and plasma fluctuations .
we found that for strong guide fields , the electric field balanced due to the quasi - collision term of the resulting boltzmann equation is negligibly small near the x - line of 2d reconnection .
as obtained by the other methods , the contributions to the correlated field - plasma fluctuations are most important near the `` low density separatrices '' .
the relevant fluctuations are mainly electrostatic with a characteristic frequency less than the lower - hybrid frequency .
note that the contribution of the correlated fluctuations to the mean electric field is smaller and does not completely match the contributions obtained by the other two methods of determining the remaining term in the ohm s law and of calculating the rate of the slowing down of the electron current .
this disagreement is , perhaps , due to the tensorial nature of the `` anomalous '' resistivity due to micro - turbulence which involves all vector components of the current density .
further , we found the criterion for an appropriate choice of the sliding window width for the averaging procedure of the mean field approach . according to our results , one has to make sure that the window width has to include all the relevant frequency range of the turbulence for the anomalous transport , otherwise the result will be very sensitive on the chosen window - width .
our findings provide a number of macroscopic signatures allowing a quantification of the wave - particle interaction and the momentum transfer due to micro - turbulence in antiparallel and in collisionless guide - field magnetic reconnection .
in particular , we found that in guide - field magnetic reconnection , the self - generated turbulence is maximum near the separatrices rather than close to the x - line .
our findings can be used , e.g. , for the diagnostics of in - situ measurements in magnetic reconnection regions in space and in laboratory experiments .
since all such measurements are averaging over space and time , it is important to make sure which part of the turbulence is cut off , which may essentially contributes to the `` anomalous '' transport .
this can be done by coordinated measurements of the mean quantities readily available , such as the ohm s law terms and the time derivatives of the mean electron current .
such analysis became possible due to the availability of high resolution observations of fluctuations in space by the current mms mission and it would become possible in an even better way with the envisioned thor@xcite spacecraft mission . while we found that in the limit of two - dimensional collisionless guide field reconnection the self - generated micro - turbulence does not significantly contribute to the mean , slowly varying electric field near the x - line , the situation will change if three - dimensional turbulence becomes relevant .
our approach to obtain the turbulence - related contribution to the mean macroscopically observable electric field and the different terms of the generalized ohm s law will , however , still apply .
this is particularly true if the kinetic turbulence is caused by plasma micro - instabilities which can only arise in three dimensions including the direction of the main current flow .
then , it is expected that 3d turbulence will contribute more significantly to the `` anomalous '' transport terms in the generalized ohm s law .
we acknowledge the developers of the acronym code ( verein zur frderung kinetischer plasmasimulationen e.v . ) .
p.m. and j.b .
acknowledge the financial support by the max - planck - princeton center for plasma physics . p.k .
acknowledges support from the nrf and dst of south africa through the following disclosure : + this work is based upon research supported by the national research foundation and department of science and technology .
any opinion , findings and conclusions or recommendations expressed in this material are those of the authors and therefore the nrf and dst do not accept any liability in regard thereto .
73ifxundefined [ 1 ] ifx#1 ifnum [ 1 ]
# 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1126/science.aaf2939 [ * * , ( ) ] link:\doibase 10.1063/1.872336 [ * * , ( ) ] link:\doibase 10.1063/1.4861359 [ * * , ( ) ] in http://www.cambridge.org/us/academic/subjects/astronomy/astrophysics/reconnection-magnetic-fields-magnetohydrodynamics-and-collisionless-theory-and-observations[__ ] , ( , ) chap . , pp .
link:\doibase 10.1146/annurev - astro-082708 - 101726 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.603 [ * * , ( ) ] link:\doibase 10.3389/fphy.2013.00031 [ * * , ( ) ] link:\doibase 10.1007/978 - 3 - 319 - 26432 - 5 [ _ _ ] , astrophysics and space science library ( , ) link:\doibase 10.1063/1.861021 [ * * , ( ) ] link:\doibase 10.1063/1.861867 [ * * , ( ) ] link:\doibase 10.1088/0029 - 5515/17/6/017 [ * * , ( ) ] link:\doibase 10.1029/rg015i001p00113 [ * * , ( ) ] link:\doibase 10.1007/bf00172248 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.51.1 [ * * , ( ) ] @noop _ _ , ed .
( , , ) in @noop _ _ ( , ) pp .
link:\doibase 10.1103/physrevlett.25.706 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.27.1553 [ * * , ( ) ] link:\doibase 10.1063/1.862321 [ * * , ( ) ] link:\doibase 10.1029/ja086ia05p03357 [ * * , ( ) ] link:\doibase 10.1016/j.asr.2005.08.024 [ * * , ( ) ] link:\doibase 10.1029/2006gl026556 [ * * , ( ) ] link:\doibase 10.1029/2006ja012000 [ * * , ( ) ] link:\doibase 10.1002/2016gl069553 [ * * , ( ) ] link:\doibase 10.1029/2002gl016113 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.035001 [ * * , ( ) ] link:\doibase 10.1063/1.3647508 [ * * , ( ) ] link:\doibase 10.1029/2011ja017109 [ * * , ( ) ] link:\doibase 10.1063/1.4862039 [ * * , ( ) ] link:\doibase 10.1063/1.1927096 [ * * , ( ) ] link:\doibase 10.1029/2005ja011482 [ * * , ( ) ] link:\doibase 10.1007/s11214 - 006 - 9094-x [ * * , ( ) ] link:\doibase
10.1088/0741 - 3335/49/12b / s30 [ * * , ( ) ] link:\doibase 10.1142/p020 [ _ _ ] ( , ) @noop _ _ ( , , ) link:\doibase 10.1186/bf03353256 [ * * , ( ) ] https://www.worldcat.org/title/magnetic-reconnection-in-plasmas/oclc/43076847[__ ] ( , , ) link:\doibase 10.1063/1.4963773 [ * * , ( ) ] , in link:\doibase 10.1007/978 - 3 - 642 - 23869 - 7 [ _ _ ] , ( , , ) pp .
link:\doibase 10.1007/bf02733547 [ * * , ( ) ] link:\doibase 10.1063/1.872088 [ * * , ( ) ] link:\doibase 10.1063/1.1914309 [ * * , ( ) ] link:\doibase 10.1063/1.1768552 [ * * , ( ) ] link:\doibase 10.1063/1.1834592 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.135004 [ * * , ( ) ] link:\doibase 10.1063/1.871095 [ * * , ( ) ] link:\doibase 10.1029/2002ja009699 [ * * , ( ) ] link:\doibase 10.1063/1.4811123 [ * * , ( ) ] link:\doibase 10.1063/1.4942031 [ * * , ( ) ] link:\doibase 10.1029/2009ja014343 [ * * , ( ) ] link:\doibase 10.1038/nature10091 [ * * , ( ) ] link:\doibase 10.1063/1.2767623 [ * * , ( ) ] link:\doibase 10.1029/2001gl013451 [ * * , ( ) ] link:\doibase 10.1029/2003ja010092 [ * * , ( ) ] link:\doibase 10.1029/2004gl020028 [ * * , ( ) ] link:\doibase 10.1029/2004ja010793 [ * * , ( ) ] link:\doibase 10.1007/s11214 - 006 - 6542 - 6 [ * * , ( ) ] link:\doibase 10.1063/1.2209611 [ * * , ( ) ] link:\doibase 10.1086/590654 [ * * , ( ) ] link:\doibase 10.1088/0004 - 637x/720/1/771 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.195003 [ * * , ( ) ] , link:\doibase 10.1029/ja086ia07p05541 [ * * , ( ) ] link:\doibase 10.1126/science.1080333 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.185001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.265004 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.265004 [ * * , ( ) ] link:\doibase 10.1063/1.4882677 [ * * , ( ) ] , link:\doibase 10.1007/978 - 94 - 007 - 7180 - 2 [ _ _ ] ( , , ) http://www.cambridge.org/catalogue/catalogue.asp?isbn=9780521858977[__ ] ( , , ) link:\doibase 10.1002/2014ja020661 [ * * , ( ) ] link:\doibase 10.5194/angeo-29 - 909 - 2011 [ * * , ( ) ] link:\doibase 10.1007/b137536 [ _ _ ] , scientific computation ( , , ) link:\doibase 10.1017/s0022377816000775 [ * * , ( ) ] | transport in hot and dilute , i.e. , collisionless , astrophysical and space plasmas is called `` anomalous '' .
this transport is due to the interaction between the particles and the self - generated turbulence by their collective interactions .
the anomalous transport has very different and not well known properties compared to the transport due to binary collisions , dominant in colder and denser plasmas . because of its relevance for astrophysical and space plasmas , we explore the excitation of turbulence in current sheets prone to component- or guide - field reconnection , a process not well understood , yet .
this configuration is typical for stellar coronae and it is created in the laboratory for which a 2.5d geometry applies . in our analysis , in addition to the immediate vicinity of the x - line , we also include regions outside and near the separatrices .
we analyze the anomalous transport properties by using 2.5d particle - in - cell ( pic ) code simulations .
we split off the mean , slow variation ( in contrast to the fast turbulent fluctuations ) of the macroscopic observables and determine the main transport terms of the generalized ohm s law .
we verify our findings by comparing with the independently determined slowing - down rate of the macroscopic currents and with the transport terms obtained by the first order correlations of the turbulent fluctuations .
we find that the turbulence is most intense in the `` low density '' separatrix region of guide - field reconnection .
it is excited by streaming instabilities , it is mainly electrostatic , `` patchy '' in space and so is the associated anomalous transport .
parts of the energy exchange between turbulence and particles is reversible and quasi - periodic .
the remaining irreversible anomalous resistivity can be parametrized by an effective collision rate ranging from the local ion - cyclotron to the lower - hybrid frequency .
the contributions to the parallel and the perpendicular ( to the magnetic field ) components of the slowly varying , dc - electric fields , balanced by the turbulence , are similar .
this anomalous electric field is , however , smaller than the contributions of the off - diagonal pressure and electron inertia terms of the ohm s law .
this result can now be verified by in - situ measurements of the turbulence , in and around the magnetic reconnection regions of the earth s magnetosphere by the multi - spacecraft mission mms and in laboratory experiments like mrx and vineta - ii .
+ _ the following article has been submitted to physics of plasmas .
after it is published , it will be found at http://scitation.aip.org/content/aip/journal/pop_ |
ground- and space - based observations of @xmath2-rays from supernova remnants ( snrs ) prove that galactic cosmic rays are produced in these objects . it is shown that @xmath2-rays are likely leptonic in rx j1713.7 - 3946 @xcite and hadronic in ic 443 and w44 @xcite . in most other snrs ,
@xmath2-rays remain of unknown origin because theoretical models may explain observed @xmath2-ray spectra in both frameworks : either with relativistic electrons or with protons ( e.g. * ? ? ?
there are some hints , both theoretical @xcite and observational @xcite , that snrs having the shell - like morphology in @xmath2-rays are likely emitting through the leptonic mechanism .
there are many evidences that electrons are accelerated in snrs from observations in radio and nonthermal x - rays . instead , direct observational proofs about acceleration of protons in snr shocks are difficult to obtain because of small radiative ability of these particles .
photons emitted in hadronic collisions may be observed in @xmath2-rays only .
however , photons from inverse - compton interactions with electrons have the same energies .
hadronic @xmath2-rays should dominate in systems with rather high magnetic field ( in order to reduce efficiency of leptonic @xmath2-rays ) and with high density of cold protons which are targets for relativistic protons accelerated by the snr shock .
such conditions naturally appear when the shock moves in the medium with the high density , e.g. it enters into a molecular cloud .
therefore , snrs located near molecular clouds are promising sites to observe hadronic @xmath2-rays @xcite .
thus , there is a reason for a systematic search for such snrs @xcite .
the shock in snr interacting with the molecular cloud decelerates because of increased density . as a result ,
the temperature of the post - shock plasma decreases and radiative losses become important .
the shock gradually transforms from the adiabatic to the radiative one ( * ? ? ? * and references therein ) .
the physics of the shock and conditions for particle acceleration change considerably .
picture of snr evolution consists typically of the three stages free expansion , adiabatic and radiative each with its own features .
dissipation of very old snrs could be considered as the fourth stage ( e.g. * ? ? ?
it is often assumed that these phases change one another very rapidly comparing to the duration of the phases themselves .
however , the transition from the fully adiabatic to the fully radiative shock lasts almost the same time as duration of the adiabatic stage @xcite .
it has its own , different from other stages , features @xcite .
first , the presence of the dynamically important radiative losses does not allow the @xcite model to be used .
second , the thin cold radiative shell is not yet formed and , therefore , the solution for the radiative shock @xcite is not relevant .
therefore , there is the need to separately consider the post - adiabatic stage in a general scenario of the snr evolution , especially in snrs interacting with molecular clouds .
evolution of snr during the post - adiabatic stage is accompanied by increasing role of the radiative losses and therefore essential change of the flow structure . in particular
, the regions with high density and magnetic field ( mf ) should appear right after the shock .
it is expected that such flow reconstruction has to result in the prominent increase of the hadronic @xmath2-rays ( due to the high density of target protons ) and the decrease of the lepronic @xmath2-rays ( due to synchrotron losses of electrons in high mf ) .
thus , the post - adiabatic snrs could be sources of the hadronic emission .
hydrodynamic properties of the post - adiabatic flows are demonstrated in the numerical simulations of @xcite .
@xcite have developed an approximate semi - analytical method to describe the radiative shock without mf .
it bases on the main properties of the flow during transition from the adiabatic to the fully radiative regime and neglects minor effects .
this hydrodynamic method is applied to a study of the hadronic @xmath2-ray emission from snrs by @xcite .
magnetic field becomes an important factor of the snr evolution just in the post - adiabatic snrs .
radiative losses lead to increase of the shock compression and thus to increase of the mf strength .
the energy density of mf becomes comparable to the thermal energy density .
mf affects dynamics of the shock and the flow downstream @xcite .
in particular , mf limits the density of the radiative shell . in the simulations without mf , the ratio of the post - shock to the pre - shock density may reach few hundred @xcite . in the present paper ,
the evolution of the post - adiabatic snrs in the interstellar medium with magnetic field is studied .
we solve numerically the system of differential equations of the ideal mhd in order to see how does mf affect the evolution of the post - adiabatic snrs .
note , that other factors could influence the formation and compression of the radiative shell , namely , thermal conduction @xcite and cosmic ray pressure in case of efficient acceleration @xcite .
they are outside of the scope of the present paper .
the numerical code pluto @xcite is adopted for our simulations .
it is designed to describe the supersonic flows in the presence of strong shocks .
pluto integrates the system of the time - dependent conservation laws of the ideal mhd in the form @xmath3 where @xmath4 is the density , @xmath5 the momentum density , @xmath6 the flow velocity , @xmath7 the total ( thermal @xmath8 and magnetic @xmath9 ) pressure , @xmath10 the unit vector , @xmath11 the mf strength , @xmath12 represents the radiative losses , @xmath13 the total energy density .
the ideal gas equation of state is assumed with @xmath14 .
the total energy density is a sum of the thermal , kinetic and magnetic components : @xmath15 additionally , the divergence - free condition holds : @xmath16 .
the system ( [ postadiab : eq2 ] ) is integrated with the use of the finite volumes approach .
the solving strategy in pluto consists of the three main steps : polynomial interpolation of the cell - averaged values to the interfaces of cells , solving the riemann problem at the cell edges and the time evolution . for each of the steps
several possible algorithms are available in the code .
therefore , one can set up a number of computational schemas .
we have tried a series of different combinations of available alternative algorithms in order to choose the one which seems to be the most appropriate for our goals .
namely , we used the following set in our simulations : linear interpolation with min mod limiter , hll riemann solver and characteristic tracing algorithm for the time evolution . additionally , we used oned shock flattening algorithm for the numerical dissipation near the strong shocks and eight wave formulation for controlling @xmath16 condition .
such numerical setup provides the second order accuracy both in spatial and temporal integrations .
our one - dimensional spherically - symmetrical computations were carried on the static uniform grid .
the physical grid size is 32 pc with 60 000 computational zones .
we used the reflective boundary conditions ( variables are symmetric across the boundary and the normal components of vectors are antisymmetric ) at the beginning of the grid as well as the outflow ( zero gradient across the boundary ) at the end of the grid .
the time step was limited by the cfl condition with courant number 0.75 .
the simulations were carried until @xmath17 . there are known solutions of the system ( [ postadiab : eq2 ] ) .
namely , @xcite found the analytical solutions for @xmath18 and @xmath19 ( for approximations of the sedov solutions see * ? ? ?
. analytical description of mf in sedov snrs ( @xmath18 , @xmath20 ) are given by @xcite .
the numerical solutions for @xmath21 and @xmath19 are presented by @xcite .
the goal of the present paper is to study the case of the post - adiabatic shocks in the ambient mf , i.e. @xmath22 and @xmath23 .
the spherically symmetric problem is considered .
, @xmath24 . plots for other values of @xmath25 and the parallel shock are the same . ]
initially , the hydrodynamic ( hd ) module in the pluto code was tested on @xcite analytical solutions for uniform ism .
the self - similarity of the spatial profiles of the hydrodynamic parameters ( density , pressure and velocity ) was reproduced in the numerical simulations . on the next step ,
mhd simulations were carried out of the adiabatic snr in uniform ism and uniform ismf in order to reveal the influence of the magnetic field on the evolution of hd parameters behind the shock as well as to determine the postshock evolution of magnetic field .
the simulations demonstrate ( fig .
[ vb - fig2a ] ) that mf may be considered as ` test - like ' , i.e. it is not capable to change the distribution of the hydrodynamic parameters inside the adiabatic snr .
the reason is that the energy density of mf ( with the strengths which are expected in snrs ) is much smaller than the thermal energy density of the gas ( high plasma @xmath26 ) .
thus , in order to model the mf inside the adiabatic snr one could separate mf and hydrodynamics : one should first know the hd structure and then , on this hydrodynamic background , one may calculate the mf distribution .
actually , such an approach is already adopted in a number of publications in order to develop an analytical description of mf in sedov snr .
note , that the strong shock from the point explosion is spherical even in the presence of ambient mf , because the deviation from the sphericity @xmath27 @xcite .
the whole problem is axisymmetric if the ambient mf is uniform .
the changes of the uniform mf due to the strong point explosion in the ideal plasma with infinite conductivity and ` weak ' mf are considered by @xcite .
the equations @xmath28 , \quad \mathrm{div } \mathbf{b}=0\ ] ] transform to the equations for the radial @xmath29 and tangential @xmath30 components of mf in the spherical coordinates @xmath31 where @xmath32 is the radius , @xmath33 the angle measured from the direction of @xmath34 . at the shock front ,
@xmath35 @xmath36 where @xmath37 is the shock compression ratio , the index ` o ' refers to the pre - shock position , the index ` s ' to the immediately post - shock one .
the self - similar solution of eq .
( [ rad : eqmfsedov_sphericla ] ) may be found ,
after substitution ( [ rad : eqmfsedov_sphericla ] ) with the @xcite solution for the flow velocity @xmath38 , by separation of variables ( overline denotes the value normalized by the value of the same parameter at the shock , e.g. @xmath39 , @xmath40 , @xmath41 ) .
the expressions for the self - similar solution for mf are quite large @xcite .
an approximation of this solution is found assumming that @xmath42 in sedov solutions .
if so , the downstream profiles of both mf components would have the same shape @xcite : @xmath43 ) and tangential ( @xmath44 ) components of mf downstream of the sedov shock .
solid line shows the analytic expressions ( [ rad : mflagr ] ) , dots represent our numerical simulations .
approximate solution ( [ rad : appsolmf ] ) is shown by the dashed line . ]
the above solutions are derived in terms of the euler coordinates @xmath32 . in many situations
an analysis in terms of the lagrangian coordinates @xmath45 is preferable .
there are solutions of that problem as well @xcite and they are much simpler : @xmath46 these equations reflect the flux - conservation and freezing conditions . in order to have the dependence @xmath47 , one should use the known @xcite relations for @xmath48 and @xmath49 or their much simpler approximations ( * ? ? ?
86 , 87 ) . on the other hand , in order to have @xmath50 , one may use ( [ rad : mflagr ] ) with @xcite relations for @xmath51 and @xmath52 or their approximations ( * ? ? ?
a2 , a5 ) .
the mf strength at any place in sedov snr is @xmath53 the formula for the mf compression ratio at the shock for an arbitrary @xmath33 is therefore @xmath54 or @xcite @xmath55 fig .
[ vb - fig3 ] shows that the analytical solutions ( [ rad : mflagr ] ) derived under assumption of the ` weak ' field are in good agreement with our numerical simulations which are free of such simplification . at some time of snr evolution ,
radiative losses become high enough and the adiabatic condition no longer holds .
this marks the end of the adiabatic stage .
radiative energy losses of the plasma are described with the term @xmath56 where @xmath57 is the plasma temperature , @xmath58 is the equilibrium ( e.g. * ? ? ? * ) or nonequilibrium ( e.g. * ? ? ? * ) cooling coefficient .
we carried out the 1-d hd simulations ( @xmath19 ) of the spherical snr in the uniform ism . in all simulations in the present paper ,
we use the same initial parameters as in @xcite .
namely , the explosion energy @xmath59 , the hydrogen number density in ism @xmath60 , the temperature in ism @xmath61 . the nonequilibrium cooling function @xmath12 is taken from @xcite .
the results of @xcite were reproduced if the mean mass per particle , per hydrogen atom , per free electron in the terms of the proton mass are @xmath62 , @xmath63 , @xmath64 respectively . this
is demonstrated on fig .
[ tk - fig5bl ] .
plot for the expansion parameter @xmath65 ( i.e. @xmath66 ) on fig .
[ tk - fig5bl]a is useful in order to see the evolutionary stage of the shock .
the deviation of @xmath67 from @xmath68 marks the end of the adiabatic regime and beginning of the post - adiabatic phase .
it happens around the transition time @xcite @xmath69 the radiative stage changes the post - adiabatic one around the shell - formation time @xcite @xmath70 these reference times are @xmath71 and @xmath72 for adopted @xmath73 and @xmath74 .
the duration of the post - adiabatic stage is @xmath75 .
the accuracy of our simulations is demonstrated by fig .
[ tk - fig5bl]b : the total energy as a sum of components @xmath76 is equal to the explosion energy up to the final time of simulations .
( all these energies are calculated as an integrals of the respective local values over the whole spherical snr volume . )
the same plot shows that the energy of radiative losses increases , the thermal energy decreases while the kinetic energy is almost constant with time .
thus , the radiative losses affect mostly the thermal component of the total energy , in a way that the sum @xmath77 is approximately constant in time .
the semi - analytical approximate method for hd of the post - adiabatic snr is developed by @xcite .
it uses lagrangian coordinates and approximately reproduces the main features of the numerical simulations .
in order to reveal the role of mf in the post - adiabatic evolution of snrs , simulations were done for the two orientations of the ambient mf , parallel and perpendicular to the shock normal , and for few values of the ismf strength : @xmath78 , @xmath1 , @xmath79 and @xmath80 . in the simulations ,
the shock position was determined from the condition of the maximum of @xmath81 where @xmath8 is the thermal pressure . and @xmath82 in the simulations for @xmath83 ( solid lines ) and @xmath84 ( dashed lines ) .
] our simulations reveal that mf does not affect dynamics of the shock if it moves along the mf lines .
the reason is that the energy density of the radial mf is much smaller than the thermal energy density .
the situation is different for the perpendicular shock .
the strength of the tangential mf increases considerably after passage of the shock front .
[ fig_blast_dynamics]a shows the evolution of the shock radius @xmath85 and velocity @xmath86 for the parallel and perpendicular orientations of ismf .
if the ambient mf is quite high then it is even possible that dynamics of the snr radius is close to the extrapolation of the sedov formula ( fig .
[ fig_blast_dynamics]a , orange dashed line ) , despite the fact that radiation losses are significant ( fig .
[ fig_blast_dynamics]d , orange line ) .
effect of the strong mf is visible also on the plots for the shock velocity @xmath86 ( fig .
[ fig_blast_dynamics]a ) and the deceleration parameter @xmath67 ( fig .
[ fig_blast_dynamics]c ) : the shock decelerates slower for higher mf strengths ; mf works against of oscillations of @xmath67 . in addition , the value of @xmath67 on the radiative stage ( @xmath87 ) is higher than the value expected from the pure hydrodynamical model @xmath88 @xcite and could be close to the adiabatic value @xmath68 for large @xmath89 ( fig .
[ fig_blast_dynamics]b , orange dashed line ) . fig .
[ fig_blast_dynamics] and d demonstrate evolution of the components of the total energy for the perpendicular shock with @xmath83 and @xmath84 respectively .
the magnetic energy is calculated on the plots as @xmath90 i.e. it is the energy transformed to mf from the explosion energy .
: it is calculated with @xmath91 . ]
magnetic energy is small for strengths @xmath92 and dynamics of the energy components is similar to behavior in simulations ( fig .
[ tk - fig5bl ] ) without mf .
the evolution of @xmath93 and @xmath94 in the presence of the only parallel mf is the same as in purely hd simulations , for different strengths @xmath95 .
radiative losses affect mostly the thermal energy , even if mf is strong ( fig .
[ fig_blast_dynamics]c - d ) .
some decrease of kinetic energy with time is visible ; it transforms into the energy of the perpendicular mf .
thus , we observe almost independent channels of energy transformations : the thermal goes into the radiative and the kinetic into the mf energy .
this statement is illustrated clearly on fig .
[ fig_energysums ] : the sums @xmath77 and @xmath82 are approximately constant with time . since radiative losses come mostly from the thermal energy , the expressions ( [ rad - ttr ] ) for the transition and ( [ rad - tsf ] ) for the shell formation times are valid also in mhd models . mf energy increases with time and may reach prominent fraction of the total energy , e.g. for @xmath96 , it is about @xmath97 at @xmath98 ( fig .
[ fig_blast_dynamics]d ) .
comparison of fig .
[ tk - fig5bl]c , [ fig_blast_dynamics]c and d reveals that radiative losses are somehow smaller for stronger mf .
in fact , at @xmath98 , about 60 % of the explosion energy are lost through the thermal radiation if the shock moves in ism without mf or along the mf lines ( fig .
[ tk - fig5bl]c ) , whereas the fraction of radiated energy is 50 % for @xmath99 ( fig .
[ fig_blast_dynamics]d ) . fig .
[ fig_space_time_t ] and [ fig_space_time_rho ] shows the time evolution of the spatial distributions of the flow parameters in the region @xmath100 and in the time interval @xmath101 .
the appearance of the first flow elements which radiated their energy away may clearly be tracked on the spatial - temporal distributions of temperature ( fig .
[ fig_space_time_t ] ) .
radiative losses lead to the quick fall of the temperature behind the shock starting from @xmath102 .
the first elements to cool appear a bit downstream of the shock , in agreement with @xcite .
if mf is low , radiative losses are effective in the thin region behind the shock .
this region corresponds to the dense shell ( [ fig_space_time_rho]a - b ) which cools more and more effectively with increase of its density .
( solid lines ) as well as the maximum value of mf strength @xmath103 ( dotted lines ) , for few values of the strength of the ambient perpendicular mf . ] the shell is more extended for @xmath104 ( fig .
[ fig_space_time_rho]c ) .
the thin dense shell is not formed for @xmath83 ( fig .
[ fig_space_time_rho]d ) .
this is because the magnetic pressure is high enough ( comparing to simulations with small mf strength ) to prevent the huge compression of gas on the post - adiabatic and radiative phases .
[ fig_space_time_rho]a - b shows that the maximum density may be few hundred times the density in ism , if mf is zero or parallel to the shock normal .
in contrast , fig .
[ fig_space_time_rho]c demonstrates that the compression factor of the perpendicular shock with @xmath104 is smaller than 50 .
the density compression factor for the post - adiabatic snrs varies between 4 and 10 for @xmath105 ( fig .
[ fig_space_time_rho]d ) .
this effect is visible more clearly on fig .
[ fig_rho_max_log ] where the evolution of @xmath106 and @xmath103 are shown for different values of perpendicular mf ( @xmath107 and @xmath108 are the maximum values within the shell , its location within the shell varies with time ) .
it is interesting to note on this figure that the shock compression is smaller than 4 for @xmath109 ; so large mf contributes to the ambient pressure and the shock is not ` strong ' any more . the smaller the density in the shell the smaller the radiative losses . on the other hand , the region affected by these losses are larger because the shell is thicker , if mf is accounted
in fact , it is prominent on fig . [ fig_space_time_t]b and fig .
[ fig_space_time_t]c that the area behind the shock front , which is affected by the radiative losses , is wider in the case of perpendicular mf with @xmath110 , and increases with the value of the mf strength ( fig . [ fig_space_time_t]d ) .
large value of the perpendicular mf creates an additional pressure immediately downstream which prevents the flow from large compression ( which is present in the pure hd simulations ) .
it also pushes the shock to the larger distance .
this results in @xmath85 , @xmath86 and @xmath67 being closer to the sedov extrapolations , even in the radiative phase @xmath87 ( fig . [ fig_blast_dynamics]a - b , green and orange lines ) . in this respect ,
mf behaves like a ` compensator ' of pressure which is lost due to radiative losses .
[ fig_space_time_rho ] confirms that the thickness of the shell for the large strength of the perpendicular mf is not small anymore and increases with time .
therefore , the theoretical model of the ` snowplow ' may not be used .
it is known from the hydrodynamic simulations ( e.g. * ? ? ?
* ) that , at the post - adiabatic stage , the two shells form ( fig .
[ fig_plot_profile_530 ] ) .
they merge into one around time ( @xmath111 ) which corresponds to the first minimum of the deceleration parameter @xmath67 ( fig . [ fig_blast_dynamics]c ) ; this is visible on fig .
[ fig_space_time_rho ] for @xmath112 and @xmath113 .
two shells are formed because the effective cooling happens and develops a bit downstream .
the pressure decrease in the outer shell results in the shock deceleration ; this is reflected by the decrease of the parameter @xmath67 up to the first minimum ( fig . [ fig_blast_dynamics]b ) .
the internal pressure pushes the inner shell until both shells merge ( time of the minimum of @xmath67 ) and then it pushes the merged shell ( the first increase of @xmath67 , fig . [ fig_blast_dynamics]b ) .
the shell experiences the radiative losses and slows down ( the second decrease on @xmath67 around @xmath114 ) reducing the ram pressure of the incoming gas .
following oscillations of @xmath67 are similar , they are due to the interplay between the internal and the ram pressures regulated by the radiative losses .
the overall evolution of the structures due to the radiative cooling , in particular , the formation of secondary shocks , are similar to described in @xcite .
perpendicular mf introduces the additional , magnetic , pressure in this picture .
when @xmath67 approaches its first minimum , the magnetic pressure becomes comparable with the thermal one .
later , mf appears to be an intermediate chain in interactions between the internal and the ram pressures : the two peaks of the pressure are clearly separated by the mf pressure ( fig .
[ fig_plot_profile_700]c ; fig .
[ fig_plot_full ] ) .
this effect is visible after @xmath115 ( which corresponds to the first minumum of @xmath67 ) for @xmath110 ( cf .
[ fig_plot_profile_530 ] and [ fig_plot_profile_700 ] ) .
larger mf prevents the two shells from interactions during the whole post - adiabatic phase , even around @xmath115 ( fig .
[ fig_space_time_rho]d ) . and
eulerian @xmath32 coordinates . ] figs .
[ fig_plot_profile_530 ] , [ fig_plot_profile_700 ] and [ fig_plot_full ] correspond to the time before and after the first minimum of the deceleration parameter @xmath67 .
the figures demonstrate that the thickness of the whole radiative shell increases with increase of the strength of ambient mf .
mf reduces the density in the shell ; the density is smaller for larger mf strength ( fig .
[ fig_rho_max_log ] ) .
the shock goes forward to the larger extent if tangential mf is present ; this effect is more prominent for higher @xmath116 ( fig .
[ fig_blast_dynamics ] , [ fig_plot_profile_530]a , [ fig_plot_profile_700]a ) .
simulations with parallel orientation of mf show no influence of mf on the distribution of hydrodynamical parameters of the flow .
the distributions are the same as in the purely hydrodynamical simulations .
[ fig_mf_parallel ] shows distributions of mf strength for @xmath117 and @xmath118 for @xmath119 and @xmath120 .
the parallel mf is not compressed behind the shock , as expected ( fig .
[ fig_mf_parallel ] ) .
mf distribution is self - similar in respect of the mf strength , i.e. the mf profiles @xmath121 are the same for different values of @xmath122 but the shape of the profile varies in time .
after @xmath123 , there is a significant drop in the magnetic field right after the shock ; the drop increases with time .
in particular , at @xmath124 , the mf strength becomes @xmath125 .
such a behavior is related to the formation of the thin dense shell . fig .
[ fig_sxema ] is useful for understanding .
it shows the relation between the lagrangian @xmath45 and eulerian @xmath32 coordinates .
the coordinate @xmath32 is common spatial position .
the coordinate @xmath45 is like a number attached to the fluid element ; it is defined as @xmath126 , i.e. it is the shock radius at time @xmath127 when the given fluid element was shocked .
therefore , @xmath128 in the ambient medium . during the adiabatic stage the profile @xmath48 is self - similar ,
i.e. it has the same shape for any time .
the shell starts to form approximately at @xmath129 .
the radiative shell consists mostly of the swept up ism of the radius @xmath130 @xcite ; this contribution is neglected on the fig .
[ fig_sxema ] . ]
( e.g. * ? ? ?
therefore , the shell is thick in coordinates @xmath45 ; the thickness is @xmath131 and increases with time .
the hot gas behind the shell ( with @xmath132 ) evolves almost adiabatically , thus the relation @xmath48 for this region scales with time approximately as @xmath133 ( see fig . [ fig_sxema ] ) .
the pressure in the shell falls due to the radiative losses ; therefore , the gas with lagrangian coordinates @xmath134 is compressed in a dense shell which is quite thin in common coordinates @xmath32 : @xmath135 .
the mf components behave in accordance to the fundamental eqs .
( [ rad : mflagr ] ) which are valid either for adiabatic or for non - adiabatic regime .
let s consider a fluid element which is located downstream but quite close to the shock , say @xmath136 .
the coordinate @xmath45 which corresponds to this @xmath32 is considerably smaller : @xmath137 ( see fig . [ fig_sxema ] ) . in other words ,
the ratio @xmath138 is smaller than unity and decreases with time .
this is why @xmath139 experiences rapid fall behind the shock , as on fig .
[ fig_mf_parallel ] .
note , that this effect also explains why @xmath103 is typically larger than @xmath106 on fig .
[ fig_rho_max_log ] : the perpendicular component of mf @xmath140 .
the component of mf parallel to the shock normal does not affect the flow structure during the whole evolution of snr .
the parallel field depends on the hydrodynamic structure rather than vice versa .
[ fig_plasma_beta]a , c explains this effect : the plasma @xmath26 is less than unity everywhere inside snr .
spatial distributions of the perpendicular mf ( fig .
[ fig_mf_perpendicular ] ) correlate with distributions of density ( fig .
[ fig_plot_full ] ) because mf is frozen into the plasma . at the beginning of the post - adiabatic stage ,
when the radiative losses are not essential yet , the distributions for mf of the intermediate strength ( @xmath141 ) are close to the self - similarity in respect of the value of @xmath95 ( like in the case of the parallel field ) .
however , this property is quickly violated due to different behavior of the radiative loses in the flows modified by mf of the different strengths .
the tangential mf is important factor in dynamics of the flows with shocks , in accordance with the values of the plasma @xmath26 ( fig .
[ fig_plasma_beta]b , d ) .
the strength of the perpendicular mf may reach quite high values on the post - adiabatic and radiative stages , even for small values @xmath142 because if @xmath26 is small then mf does not prevent plasma from high compression and the perpendicular mf is compressed to similar values as density .
mf with higher @xmath116 limits the compression and the maximum value of the strength is smaller ( fig .
[ fig_rho_max_log ] ) .
the numerical code pluto was adopted and used for magneto - hydrodynamic simulations of flows with shock waves in supernova remnants with the radiative losses of plasma .
calculations are performed for the case of the one - dimensional flows under conditions of spherical symmetry , in the interstellar medium with the uniform distributions of density and magnetic field .
the role of mf in formation of the flow structure is revealed .
it is significant only in the post - adiabatic and radiative phases .
the pressure of the tangential mf prevents large compression of plasma and limits the level of maximum density in the radiative shell .
radiative shell is more extended behind the perpendicular shock .
these effects are more significant for larger values of mf strength .
parallel mf does not affect the flow dynamic because its energy density is much smaller than the thermal energy density .
radiative losses affect mostly the thermal energy .
the magnetic energy increases mostly by reducing the kinetic energy . at oblique shocks of the post - adiabatic snrs ,
the perpendicular component of mf increases and the parallel component decreases .
this mechanism may be responsible for the predominantly tangential orientation of mf in old snrs , as it is known from observations of the radio polarization @xcite . at late times ,
when the shock decelerates a lot and the mach number becomes small , efficiency of particle acceleration by the fermi mechanism considerably decreases . on the other hand , at these times the tangential mf dominates over most of the snr surface , even in the regions where the parallel mf was large at previous stages ( quasiparallel shocks ) .
this probably keeps more or less effective further acceleration of particles due to the drift mechanism . especially in the view of the rather small shock velocity
: the particles tied to mf lines might spend more times for multiple crossings of the ( quasi-)perpendicular shock until the mf line will be swept up downstream .
the old post - adiabatic snrs or regions of snrs interacting with the high - density medium may be promising sources of hadronic @xmath2-rays .
radiative losses in these objects essentially affect dynamics of the shock and structure of the flow .
they cause considerable increase of density and mf strength in the radiative shell .
therefore , prominent emission of hadronic @xmath2-rays should be generated there whereas the competing leptonic @xmath2-rays has to be substantially reduced under such conditions .
all simulations were performed on the computational cluster in institute for applied problems in mechanics and mathematics .
the research was partially supported by the program of ukrainian national academy of sciences `` grid infrastructure and grid technologies for science and applications '' ( grant 0115u002936 ) .
99 abdo a. a. et al .
, 2011 , apj , 734 , 28 acero f. et al . , a&a , 2010 , 516 , a62 acero , f. lemoine - goumard m. , renaud m. , ballet j. , hewitt j. , rousseau r. , tanaka t. , 2015 a&a , 580 , a74 ackermann m. et al . , 2013 , science , 339 , 807 bandiera r. , petruk o. , 2004 , a&a , 419 , 419 blondin j. m. , wright e. b. , borkowski k. j. , reynolds s. p. , 1998 ,
apj , 500 , 342 chen y. et al . , 2014 ,
iau symposium , 296 , 170 chevalier r. , 1974 , apj , 188 , 501 cioffi d. f. , mckee c. f. , bertschinger e. , 1988 , apj , 334 , 252 cox d. p. , 1972 ,
apj , 178 , 159 cox d. p. , 1986 ,
apj , 304 , 771 cox d. p , anderson p. , 1982
, apj , 253 , 268 cox d. p. , franco j. , 1981 , apj , 251 , 687 dubner g. , giacani e. , 2015 , astron .
rev . , 23 , 3 falle s. , 1975 , mnras , 172 , 55 fleishman g. , toptygin i. , 2013 , cosmic electrodynamics .
springer , new york hnatyk b. , petruk o. , telezhinsky i. , 2007 , kinematics and physics of celestial bodies , 23 , 137 innes d. , 1992 , a&a , 256 , 660 jiang b. et al .
, 2010 , apj , 712 , 1147 karlikov v. , korobeinikov v. , 1960 , dokl .
nauk sssr , 133 , 764 korobeinikov v. , 1964 , zhurnal prikladnoi mekhaniki i tekhnicheskoi fiziki , 4 , 113 korobeinikov v. , 1991 , problems of point blast theory .
springer - verlag , new york lee s .- h .
, patnaude d. , raymond j. , nagataki s. , slane p. , ellison d. , 2015 , apj , 806 , 71 mckee c. , ostriker j. , 1977 , apj , 218 , 148 mignone a. , bodo g. , massaglia s. , matsakos t. , tesileanu o. , zanni c. , ferrari a. , 2007 , apjs , 170 , 228 mignone a. , zanni c. , tzeferacos p. , van straalen b. , colella p. , bodo g. , 2012 , apjs , 198 , 7 milne d. k. , 1987 , aust .
j. phys . , 40 , 771 orlando s. , bocchino f. , reale f. , peres g. , pagano p. , 2008 ,
apj , 678 , 274 pasko v. , silich s. , 1986 , kinematics and physics of celestial bodies , 2 , 15 petruk o. , 2000 , a&a , 357 , 686 petruk o. , 2005 , journal of physical studies , 9 , 364 petruk o. , bocchino f. , miceli m. , dubner g. , castelletti g. , orlando s. , iakubovskyi d. , telezhinsky i. , 2009 , mnras , 399 , 157 raymond j. c. , cox d. p. , smith b. w. , 1976 , apj , 204 , 290 reynolds s. p. , 1998 ,
apj , 493 , 375 sedov l. , 1959 , similarity and dimensional methods in mechanics . academic press , new york slane p. , bykov a. , ellison d. , dubner g. , castro d. , space science reviews , 2015 , 188 , 187 slavin j. , cox d. , 1992 , apj , 392 , 131 sutherland r. s. , dopita m. a. , 1993 , apjs , 88 , 253 telezhinsky i. , hnatyk b. , 2007 , mod phys lett a , 22 , 2617 toptygin i. , 2004 , astr .
lett . , 30 , 393 | gamma - rays from hadronic collisions are expected from supernova remnants ( snrs ) located near molecular clouds .
the temperature on the shock interacting with the dense environment quickly reaches @xmath0 k. the radiative losses of plasma become essential in the evolution of snrs .
they decrease the thermal pressure and essentially increase the density behind the shock
. the presence of ambient magnetic field may considerably alter the behavior of the post - adiabatic snrs comparing to hydrodynamic scenario . in the present paper ,
the magneto - hydrodynamic simulations of radiative shocks in magnetic field are performed .
high plasma compression due to the radiative losses results also in the prominent increase of the strength of the tangential component of magnetic field behind the shock and the decrease of the parallel one .
if the strength of the tangential field before the shock is higher than about @xmath1 it prevents formation of the very dense thin shell .
the higher the strength of the tangential magnetic field the larger the thickness and the lower the maximum density in the radiative shell .
parallel magnetic field does not affect the distribution of the hydrodynamic parameters behind the shock .
there are almost independent channels of energy transformations : radiative losses are due to the thermal energy , the magnetic energy increases by reducing the kinetic energy .
the large density and high strength of the perpendicular magnetic field in the radiative shells of snrs should result in considerable increase of the hadronic gamma - ray flux comparing to the leptonic one .
[ firstpage ] ism : supernova remnants shock waves
ism : magnetic fields |
according to the theory of strong interactions , the quantum chromo dynamics ( qcd ) , hadrons are bound objects of quarks and gluons .
the properties of hadrons in vacuum are well known and confirmed by lattice qcd calculations @xcite while the properties of hadrons in a strongly interacting environment are subject of intensive research .
qcd inspired approaches as well as phenomenological models based on phase shifts and su(3 ) symmetry @xcite predict significant changes of hadron properties in a strongly interacting medium .
the results of the different models vary substantially .
it is therefore one of the challenges of novel experimental heavy - ion physics to study these in - medium modifications of hadrons . besides the in - medium properties of the antikaon , interesting also for astrophysical reasons ,
the vector mesons and especially the @xmath12 meson have been in the focus of the theoretical interest because the @xmath12 has the quantum numbers of a photon and can therefore disintegrate into an electron - positron pair .
having only electromagnetic interactions this pair may easily leave the reaction region without further collisions .
this allows to reconstruct the invariant mass of the decaying @xmath13 .
thus there is a hope that by measuring the dilepton invariant mass spectra the in - medium mass and width of the @xmath12 meson become experimentally accessible . for this
it is necessary to separate the background from known dilepton sources dalitz decays of baryonic and mesonic resonances as well as @xmath9 and @xmath5 bremsstrahlung . at sps energies of 40 and 158
@xmath0gev such an enhancement above the known background has been measured by the ceres @xcite and the na60 @xcite collaborations .
the experimental results are compatible with the assumption that in the medium the peak position of the @xmath12 meson mass distribution remains rather unchanged while the width increases considerably ( cf .
@xcite ) . at much lower energies , at energies around 1 @xmath0gev dileptons
have been measured in heavy - ion collisions at the bevalac in berkeley by the dls collaboration @xcite .
these data led to the so called dls puzzle because the dls dilepton yield in c+c and ca+ca collisions in the invariant mass range from 0.2 to 0.5 gev @xcite was about five times higher than the results from different transport models at that time using the conventional dilepton sources such as bremsstrahlung , @xmath14 and @xmath6 dalitz decays and direct decay of vector mesons ( @xmath15 ) @xcite .
this discrepancy remained even after including in the transport calculations the different scenarios for the in - medium modifications of vector meson properties , as dropping mass or collisional broadening of the @xmath12 and @xmath16 spectral functions @xcite . to solve this puzzle was one of the main motivations to build the hades ( high acceptance dilepton spectrometer ) detector at gsi in germany @xcite . in 2008
the hades collaboration confirmed the dls data @xcite for c+c at 1.0 @xmath0gev . in the mean time also the theoretical transport approaches as well as effective models for the elementary @xmath17 reactions have been further developed . as it has been suggested in ref .
@xcite , the dls puzzle can be solved when incorporating stronger @xmath4 and @xmath3 bremsstrahlung contributions in line with the updated one - boson - exchange ( obe ) model calculations from @xcite .
the previous obe approaches @xcite used in the old transport calculations for the analysis of the dls data , gave results close to the soft photon approximation .
as shown in ref .
@xcite the results of the hsd model ( off - shell microscopic hadron - string - dynamics ( hsd ) transport approach ) with enhanced bremsstrahlung cross sections agree very well with the hades experimental data for c+c at 1 and 2 @xmath0gev as well as with the dls data for c+c and ca+ca at 1 @xmath0gev , especially when one includes a collisional broadening in the vector - meson spectral functions .
similar results have been obtained by other , independent transport approaches iqmd @xcite and the rossendorf buu @xcite . despite of the fact that theory predicts that the vector meson properties are modified substantially already at energies as low as 1 - 2 @xmath0gev
, it is quite difficult to observe these changes experimentally .
the production yield of @xmath13 and @xmath18 mesons is small at these energies and the background from other dilepton sources like @xmath6-dalitz decay and @xmath9 bremsstrahlung is large in the mass range of interest , @xmath19 gev .
therefore the presently available data do not allow for a detailed investigation of the in - medium properties of vector mesons .
this focuses the interest of the present studies to the question whether the invariant mass spectrum below the @xmath20 peak depends on the system size and on the beam energy in a non - trivial way , i.e. whether the dilepton invariant mass spectra can be understood as a superposition of individual @xmath3 and @xmath4 interactions . in a first publication @xcite the hades collaboration found that the invariant mass spectra of dileptons , observed in 1 and 2 @xmath0gev c+c collisions , are practically coincident below @xmath21 gev if divided by the total number of observed @xmath22 and after subtracting the @xmath23 dalitz decay contribution
it is strongly suppressed at 1 @xmath0gev but becomes essential at 2 @xmath0gev due to the kinematical threshold for the @xmath23 production in @xmath17 collisions .
this scaling with the @xmath22 number can be interpreted as a scaling with the number of participants @xmath24 .
the hades collaboration , comparing the dilepton yield from the light c+c
systems with the elementary @xmath3 and @xmath25 interactions , albeit taken at different energies , has concluded that the dilepton invariant mass spectra in these light systems can be considered as a mere superposition of @xmath4 and @xmath3 collisions without any in - medium enhancement . in a more recent publication @xcite a heavier system , ar+kcl at 1.75 @xmath0gev ,
was investigated and the collaboration came to the conclusion that in this reaction the dilepton invariant mass yield between 0.2 and 0.6 gev is about 2 - 3 times larger than expected from a mere superposition of @xmath3 and @xmath4 collisions @xcite . from the analysis of the excess in the transverse - mass slope and from the angular anisotropies
the hades collaboration concluded that the excess of dileptons in the low invariant mass region scales with the system size very differently than the freeze - out yield of pions and @xmath23 and that the data are compatible with the assumption that they originate from @xmath6 dalitz decays , being suggestive of resonance matter @xcite .
it is the purpose of the present study to investigate this enhancement within the presently available transport codes hsd , iqmd and urqmd in order to explore whether the dilepton production in these systems can be reproduced by the theoretical approaches and to identify eventually the origin of this in - medium enhancement .
all these codes have been successfully employed to investigate a multitude of experimental observables .
they are , however , not of the same sophistication as far as the dileptons are concerned .
the dilepton part of the urqmd program is still under development and up to now bremsstrahlung is not included .
this limits the predictive power to parts of the spectra where bremsstrahlung is not essential . for the study of the in - medium enhancement we limit ourselves to hsd and iqmd calculations .
we start out in section ii with a short description of the hsd model and of the improvements made compared to the standard hsd 2.5 version used for the extended dilepton study in ref . @xcite .
then we come to a brief description of dilepton production in iqmd . in section iii the dilepton production in elementary reactions , measured by the hades and and dls collaboration , is compared with hsd and iqmd calculations .
the forth section is devoted to the study of dilepton production in heavy - ion collisions .
we discuss our calculations for all systems which have been measured by the hades collaboration and present also our predictions for au+au collision at 1.25 @xmath0gev which is presently analysed . after checking that the invariant mass spectra of dileptons in heavy - ion collisions are well reproduced by the hsd as well as by the iqmd approach we study in section v the enhancement of the dilepton production in heavy - ion collisions as compared to the elementary reactions and identify its origin . in section
vi we present our conclusions .
our analysis of the experimental results is carried out within the off - shell hsd transport model @xcite - based on covariant self energies for the baryons @xcite .
it has been used for the description of @xmath26 and @xmath27 collisions from sis to rhic energies .
we recall that in the hsd approach nucleons , @xmath6 s , n@xmath28(1440 ) , n@xmath28(1535 ) , @xmath29 , @xmath30 and @xmath31 hyperons , @xmath32 s , @xmath33 s and @xmath34 s as well as their antiparticles are included on the baryonic side whereas the @xmath35 and @xmath36 octet states are incorporated in the mesonic sector .
inelastic baryon baryon ( and meson - baryon ) collisions with energies above @xmath37 gev ( and @xmath38 gev ) are described by the fritiof string model @xcite whereas low energy hadron
hadron collisions are modelled using experimental cross sections . the dilepton production by the decay of a ( baryonic or mesonic ) resonance @xmath39 can be schematically presented in the following way : @xmath40 in a first step a resonance @xmath39 might be produced in baryon - baryon ( @xmath41 ) or meson - baryon ( @xmath42 ) collisions ( [ chbbr ] ) , ( [ chmbr ] ) . then this resonance can either couple directly to dileptons ( [ chrd ] ) ( e.g. dalitz decay of the @xmath6 resonance : @xmath43 ) or produces mesonic ( [ chrmd ] ) or baryonic ( [ chrprd ] ) resonances which then produce dileptons via direct decays ( @xmath44 ) or dalitz decays ( @xmath45 ) . with increasing energy hadrons
are created by non - resonant mechanisms or string decay .
this is also true for those which disintegrate into dileptons .
the electromagnetic part of all conventional dilepton sources
@xmath46 dalitz decays as well as direct decay of vector mesons @xmath47 and @xmath48 are calculated as described in detail in ref .
we note that we use here again ( as in early hsd dilepton studies @xcite ) the `` wolf '' model for the differential electromagnetic decay width of the @xmath6 resonance @xcite instead of the `` ernst '' description @xcite adopted at that time in @xcite . the treatment of the enhanced bremsstrahlung contribution from @xmath3 , @xmath4 as well as @xmath5 quasi - elastic scattering , based on the obe calculations by kaptari and kmpfer @xcite , is discussed in detail in ref .
@xcite ( cf .
section 2.6 there ) where also a discussion of the different models @xcite , which formulate bremsstrahlung in the elementary reactions , can be found .
we note here that the obe models mentioned above @xcite provide different results not only for the @xmath9 bremsstrahlung contribution ( which might be attributed to the different way to realize the gauge invariance ) but for the @xmath6-dalitz decay , due to the different form factors . in our transport analysis
we use only the bremsstrahlung contribution from @xcite avoiding the uncertainties in the @xmath6 channel in the obe models and neglecting the quantum mechanical interference between individual contributions which can not be treated consistently in transport approaches .
also we stress here again that in order to separate the bremsstrahlung ( @xmath49 ) from a vector - dominance like dilepton production via the @xmath12-meson ( @xmath50 ) , we do not employ a vector - dominance form factor when calculating the bremsstrahlung .
thus , the dilepton radiation via the decay of the virtual photon ( @xmath51 ) and the direct @xmath12 decay to @xmath52 are distinguished explicitly in the calculations . in the section
vi we discuss the model uncertainties concerning the treatment of @xmath6 s and bremsstrahlung .
the off - shell hsd transport approach incorporates the _ off - shell propagation _ for vector mesons as described in ref .
@xcite in extension of early buu transport models @xcite . in the off - shell transport description ,
the hadron spectral functions change dynamically during the propagation through the medium and evolve towards the on - shell spectral functions in the vacuum . as demonstrated in refs .
@xcite , the off - shell dynamics is important for resonances with a rather long lifetime in the vacuum but strongly decreasing lifetime in the nuclear medium ( especially @xmath16 and @xmath48 mesons ) and also be proven to be vital for the correct description of the dilepton decay of @xmath12 mesons with masses close to the two pion decay threshold . for a detailed description of the off - shell dynamics and the implementation of in - medium scenarios ( as a collisional broadening and/or dropping mass scenario ) in hsd as well as for an extension of the lund string model to include modified spectral functions we refer the reader to refs .
@xcite . for the present study we consider the scenario of a collisional broadening of the vector meson spectral functions .
this is also supported by experimental data in contrast to the dropping mass scenario ( cf .
we incorporate the effect of collisional broadening of the vector - meson spectral functions by using for the vector meson width @xmath53 here @xmath54 is the total width of the vector mesons ( @xmath55 ) in the vacuum . the collisional width in ( [ gammas ] ) is approximated as @xmath56 here @xmath57 are the velocity , 3-momentum and energy of the vector meson in the rest frame of the nucleon current and @xmath58 ; @xmath59 is the nuclear density and @xmath60 the meson - nucleon total cross section .
we use the broadening coefficients @xmath61 mev for the @xmath12 and @xmath62 mev for @xmath16 mesons as obtained in @xcite . for the further details we refer the reader to ref .
@xcite .
we use the time integration method to calculate dilepton spectra which means that vector mesons and resonances can emit dileptons from their production ( birth ) up to their absorption ( death ) .
this is especially important for the study of in - medium effects because this method takes the full in - medium dynamics into account .
we note that it is very important to have an adequate description of the elementary reactions , especially near the threshold where the cross sections grow very rapidly .
this rise has a big impact on the description of the experimental data .
the comparison of the latest experimental data from the hades collaboration on @xmath3 collisions at 3.5 gev @xcite with hsd calculations shows that the previous parametrizations of @xmath23-meson and of vector mesons ( @xmath63 ) production cross sections from ref .
@xcite overestimate the data .
the over - prediction of the dilepton yield at the @xmath12-peak has been already realized in ref .
@xcite from the comparison to the dls data for @xmath3 at 2.09 gev .
thus , we have modified the hsd model accordingly in order to obtain a better description of the existing experimental data on elementary reactions ( cf . the discussions in the next subsection ) .
we note that the hsd model is well tested with respect to the bulk observables at low energy and in light systems , relevant for present study .
the pion and eta production from c+c collisions at the energies considered here are shown in section 3 of ref .
@xcite . here
we describe the major changes / improvements made in the hsd model used here compared to the basic hsd version 2.5 used for the dilepton analysis in ref .
@xcite : the high energy part of the @xmath64 and the @xmath65 cross sections have newly been parametrized .
the new parametrization , compared to the experimental data , is shown in fig .
[ fig_xseta ] for @xmath3 and @xmath4 reactions as a function of the centre of mass energy above threshold @xmath66 .
the solid and dashed lines represent the inclusive @xmath67 and @xmath68 cross sections from the hsd model .
the experimental data are collected from refs .
@xcite : the full squares and the open star stand for the exclusive @xmath69 data , the dots for @xmath70 data . the full diamond and the full star show inclusive data for @xmath71 .
the open and full stars indicate the exclusive and inclusive cross sections extrapolated from the dilepton data by the hades collaboration @xcite .
it is important to note that in elementary reactions also a deuteron can be produced in the final state via @xmath72 .
the cross section for this channel is indicated by the dashed - dotted line in fig .
[ fig_xseta ] , whereas the dotted line shows the @xmath73 cross section including the @xmath72 contribution . the open triangles
show the experimental data for @xmath74 from refs .
the channel @xmath74 is not considered in the hsd calculations for @xmath75 and @xmath2 reactions because the probability for deuteron formation in the baryonic medium is negligibly small .
production cross section in @xmath3 and @xmath4 reactions as a function of the invariant energy above threshold @xmath66 .
the solid and dashed lines represent the inclusive @xmath67 and @xmath73 cross sections from the hsd model , the dashed - dotted line indicates the @xmath72 channel and the dotted line the @xmath73 cross section including @xmath76 .
the experimental data are collected from refs .
@xcite : the full squares and open star stand for the exclusive @xmath69 experimental data , the dots for @xmath70 and open triangles for @xmath74 experimental data ; the solid diamond and solid star show inclusive experimental data for @xmath71 .
the open and full stars indicate the exclusive and inclusive cross sections extrapolated from the dilepton data by the hades collaboration @xcite . , width=302 ] the hsd model has also been improved concerning the isospin separation of the vector meson production in baryon - baryon ( @xmath41 ) and secondary meson - baryon ( @xmath42 ) reactions .
the isospin averaged cross sections @xmath77
( @xmath78 ) and @xmath79 have been replaced by cross sections which take explicitly the isospin for each channel into account .
the new parametrization of the cross section as a function of the centre of mass energy , @xmath80 for the @xmath3 reaction is compared in fig .
[ fig_xsvm ] to the experimental data .
the solid lines represent the parametrizations of the inclusive @xmath81 cross sections while the dashed lines stand for the exclusive cross sections .
we denote these exclusive cross sections for the @xmath12-meson production as non - resonant since in this study we consider explicitly the possible contribution of the baryonic resonance @xmath82 to the sub - threshold @xmath13 production ( @xmath83 ) .
it is indicated as the dashed - dotted line on the left plot .
the dotted line shows the sum of the inclusive non - resonant and exclusive resonant contribution .
the experimental data @xcite are shown for exclusive @xmath84 ( dots ) and inclusive @xmath85 ( squares ) vector meson production .
the stars indicate the inclusive cross sections extrapolated from the dilepton data by the hades collaboration @xcite .
( left plot ( a ) ) and @xmath86 ( right plot ( b ) ) as a function of the centre of mass energy @xmath80 .
the solid lines represent the parametrizations of the inclusive @xmath87 cross sections while the dashed lines stand for the exclusive non - resonant cross sections .
the dashed - dotted line on the left plot shows the contribution from the @xmath82 resonance to the @xmath13 production via the process @xmath83 and the dotted line indicates the sum of the inclusive non - resonant and exclusive resonant contributions .
the experimental data @xcite are shown for exclusive @xmath88 ( dots ) and inclusive @xmath85 ( squares ) vector meson production .
the stars indicate the inclusive cross sections extrapolated from the dilepton data by the hades collaboration @xcite .
the vertical light blue lines show the threshold for meson production by string formation and decay ( @xmath89 gev ) as implemented in hsd for baryon - baryon channels .
, width=415 ] we note that we do not propagate explicitly the @xmath82 resonance in the hsd approach , rather we consider it as an excitation in the amplitude which enhances the @xmath12-meson production in @xmath17 and @xmath5 reactions at sub - threshold energies . the modelling of the @xmath82 production in @xmath17 collisions is based on a phase space model with a constant matrix element adopted from ref .
the contribution of the @xmath82 to the @xmath12 cross section is included in line with ref .
@xcite which has been used in our previous work @xcite .
the decay channels of the @xmath82 resonance are not well established .
especially the disintegration into a @xmath12 is estimated in between 15 -25% . including this contribution from the @xmath82 resonance decay presents an upper estimate for the @xmath12-meson production in @xmath17 and @xmath5 reactions at sub - threshold energies .
this model assumption can be checked experimentally via an observation of an enhancement of the dilepton yield near the @xmath12-peak in the elementary reactions at sub - threshold energies . in the case of heavy - ion collisions at low bombarding energies the contribution of the @xmath82 resonance to the dilepton spectra
can hardly be seen especially not in reactions of the light nuclei as c+c as measured by the hades collaboration at 1.0 @xmath0gev .
the fermi motion modifies the available energy for meson production and due to the rapid rise of the cross section at threshold the inclusive @xmath12-meson production mechanism starts to dominate , see fig .
[ fig_xsvm ] , even if the nominal energy is below threshold . on the other hand ,
the @xmath82 resonance can be excited by pion - baryon collisions and contribute to the @xmath12-meson production via the process @xmath90 .
the probability of such processes is larger for heavy nuclei collisions but the pion density is relatively small at sub - threshold energies where the possible contribution of @xmath82 plays a role .
consequently , the enhancement of the @xmath12-meson production by accounting for the @xmath82 channel is relatively small in the hsd model .
this differs from e.g. the urqmd model @xcite where a much larger cross sections for the @xmath82 production is used .
we will come back to this discussion in section iv .
+ we improved also the description of multi - meson production between the two pion production threshold and @xmath91 gev where we match the standard hsd description of particle production via strings .
close to the two - pion threshold the two pions are dominantly produced by the decay of 2 deltas created in nn collisions . with increasing energy
the available phase space is sufficient for multi - meson production and a lot of extra channels become open .
however , it is unknown whether the light mesons are produced by the decay of heavy baryonic resonances or directly from the excitation and decay of the strings . since there is very little experimental information on the exclusive channel decomposition in this intermediate energy range we used the fritiof lund string model as an event generator for the production of such multi - meson channels by adding them to the exclusive channels which are modeled in the hsd explicitly , such that we obtain the inelastic @xmath17 cross section , i.e. @xmath92 , where @xmath93 stands for the exclusive channels such as @xmath94 , @xmath95 and channels with strangeness production such as @xmath96 and @xmath97 . here
@xmath98 corresponds to the sum of the 2 pion production channels as @xmath99 and multi - meson production @xmath100 @xmath101 and channels with the final hyperons and strange mesons .
we note that since close to the threshold the fritiof model does nt provide the correct isospin decomposition for @xmath6 production , since e.g. an exclusive channel @xmath102 is missing , we have adjusted the fritiof model to correct for the isospin decomposition of produced @xmath6 s : for the exclusive channel @xmath103 we assume now that 3/4 of the produced @xmath6 are in the @xmath104 state and only 1/4 in the dilepton producing @xmath105 state .
this leads to a reduction of @xmath106 production and an enhancement of @xmath104 production , respectively .
the excitation function of the multiplicity of the different pions in hsd is shown in fig.[fig_xspi ] , on the left hand side for @xmath3 collisions , on the right hand side for @xmath4 reactions .
these multiplicities are compared with the available data which are very scarce for @xmath4 reactions .
additionally to the total pion multiplicity , the multiplicity of @xmath6 s themself is very important for the dilepton study because the @xmath6 resonances decay into pions as well as into dileptons whereas other sources of pions do not contribute to the dilepton yield .
the @xmath6 production in @xmath3 collisions in the hsd approach is shown in fig .
[ fig_xsdel ] and compared with the available experimental data . here
the production cross sections for the inclusive channels @xmath107 ( solid line ) and for the exclusive channel @xmath108 ( dashed line ) from hsd are presented as a function of the invariant energy @xmath80 .
the experimental data @xcite are shown for exclusive @xmath108 production .
the star indicates the extrapolation for the @xmath6 inclusive cross section from the dilepton spectra by the hades collaboration based on the pluto simulation program @xcite from ref .
one can see from fig .
[ fig_xsdel ] that the inclusive @xmath6 production dominates the exclusive one already at relatively low @xmath80 .
however , due to the lack of inclusive experimental data on @xmath6 production it is hard to justify the modeling of @xmath6 dynamics beyond the exclusive channels which are relatively well known experimentally and accurately modeled in transport approaches . .
the hsd results are shown in terms of lines with open symbols whereas the experimental data are indicated by the corresponding solid symbols , i.e. for @xmath109 : hsd - the solid line with open dots , experimental data - solid dots from refs .
@xcite ; for @xmath110 : hsd - the solid line with open stars , the hades data - full stars from refs .
@xcite ; for @xmath111 : hsd - the solid line with open triangles , experimental data - full triangles from refs .
right ( b ) : the production cross sections for @xmath112 from the hsd model , the hades data full star from the extrapolation in ref .
@xcite.,title="fig : " ] .
the hsd results are shown in terms of lines with open symbols whereas the experimental data are indicated by the corresponding solid symbols , i.e. for @xmath109 : hsd - the solid line with open dots , experimental data - solid dots from refs .
@xcite ; for @xmath110 : hsd - the solid line with open stars , the hades data - full stars from refs . @xcite ; for @xmath111 : hsd - the solid line with open triangles , experimental data - full triangles from refs .
right ( b ) : the production cross sections for @xmath112 from the hsd model , the hades data full star from the extrapolation in ref .
@xcite.,title="fig:",width=232 ] ( solid line ) and the exclusive channel @xmath108 ( dashed line ) from hsd as a function of the invariant energy @xmath80 . the experimental data @xcite are shown for exclusive @xmath108 production .
the star indicates the inclusive cross section extrapolated from the dilepton data by the hades collaboration @xcite.,width=359 ] as said , above a kinetic energy of 1.5 gev there is no experimental information available on whether resonance are involved in the production .
this is the reason way different parameterizations have been advanced .
for example in the resonance based gibuu model @xcite a lower inclusive @xmath106 production cross section is used as compared to hsd .
also the isospin relations are different which leads to a lower dilepton contribution from @xmath6 dalitz decay .
additionally the different parametrization for the differential dilepton decay width is employed which lowering the @xmath6 dalitz decay channal substantially compare to the hsd ( cf . discussions in section vi ) . in nucleon - nucleon collisions at low energies , i.e. below @xmath113 gev ,
very seldom more than one meson is produced .
the cross sections for these reactions have been measured experimentally ( cf .
@xcite ) and are used in the transport approach . above @xmath114 the multi - meson production starts to dominate , but the experimental information on inclusive as well as exclusive multi - meson production channels are very poor .
also it is not known whether the mesons are directly produced or whether they are decay products of intermediate resonances or strings .
the theoretical analysis of these data has not produced yet a consistent knowledge on the channel decomposition @xcite .
this introduces large uncertainties for the prediction of the dilepton yield in transport theories because it depends on the formation of specific intermediate resonances .
we note that in the urqmd model the production of mesons at intermediate energies is realized exclusively via excitation and decay of heavy baryonic resonances which are explicitly propagated in the transport model @xcite , whereas in hsd the string mechanism is used ( as discussed above ) for the description of the same final meson spectra .
thus , one needs more exclusive experimental information in order to differentiate between the models .
the iqmd model used for the calculations in this study is the same as introduced in the first iqmd paper on dilepton production @xcite . in this model
all pions are produced by the decay of @xmath6 resonances . because no higher mass resonances are included we limit the prediction to beam energies up to 2 @xmath0gev .
the excitation function of the pion yield for the ca+ca system , compared with the available data , is shown in fig .
[ fig_piiqmd ] .
we see that the pion multiplicity , the result of a complicated interplay between @xmath6 creation , absorption and decay , is quite reasonable reproduced by the iqmd approach @xcite .
this is also the case for heavier systems @xcite .
thus both , the iqmd as well as the hsd approach , describe the available pion data quite well , a prerequisite for an analysis of the dilepton spectra which are not only normalized to the pion yield but have an important contribution from the @xmath6 decay . for other models which are used to describe the dilepton production , like @xcite
, it remains to be seen whether they reproduce heavy - ion pion data .
for the calculations of the dilepton spectra the standard iqmd program @xcite has been supplemented with all elementary cross sections which are important for this process@xcite . for
that we have used the parametrizations of available experimental data , but for many channels , @xmath3 data are only available for low @xmath80 values and @xmath115 data are very scarce .
consequently , in heavy - ion collisions at beam energies larger than 1.5 @xmath0gev most of the particles which emit dileptons are produced using theoretically calculated cross sections . in ref .
@xcite we have studied how the uncertainties of the cross sections from elementary reactions influence the dilepton spectra in heavy - ion collisions . for these studies we use the set up in which the @xmath116 cross section is 5 times higher than the @xmath117 cross section .
this explains the difference between hsd and iqmd at dileption invariant masses close to the @xmath63 peak .
multiplicity per participating nucleon ( @xmath118 ) for ca+ca collisions using @xmath119 as @xmath120 .
data of the fopi collaboration are compared with data of harris et al . @xcite and predictions of the iqmd model @xcite.,width=359 ] in the iqmd approach
the dileptons are calculated perturbatively using the spontaneous decay method - contrary to the time integration method in hsd and urqmd .
it is based on the assumption that all hadrons which decay into dileptons and which are produced in the heavy - ion collision contribute to the dilepton yield as if they were produced in free space .
this implies that a possible later reabsorption of the hadrons is not taken into account . because in reality some of the @xmath6 s and of the other dilepton producing hadrons are reabsorbed , the iqmd calculations give an upper limit for the dilepton production in heavy - ion collisions .
consequently , the spontaneous decay method limits the approach to small systems contrary to the time integration method which follows the in - medium dynamics of all dilepton sources exactly .
however , for the systems studied here the spontaneous decay method is still acceptable . for the details of the cross sections for the creation of dilepton producing particles we refer to ref .
the first reaction considered here is the dilepton production in elementary reactions like @xmath122 , quasi - free @xmath123 and @xmath124 reactions .
we start our discussion with the hades and dls data a 1.25 gev .
[ fig_pp125 ] shows the differential cross section @xmath125 for dileptons as a function of the invariant mass m for @xmath3 ( left ) , @xmath4 ( middle ) and @xmath25 ( right ) reactions at 1.25 gev .
the hsd results are presented in comparison to the experimental data from the hades collaboration @xcite .
the different lines display the contributions from the various channels in the hsd calculations ( for the colour coding we refer to the legend ) .
we note here ( and that applies to all further plots ) that the theoretical calculations passed through the appropriate experimental acceptance filters and that the mass / momentum resolution is taken into account . as seen in the left part of fig .
[ fig_pp125 ] the @xmath3 dilepton yield is dominated by the @xmath6-dalitz decay while bremsstrahlung is sub - leading due to the destructive interference between initial and final state amplitudes in case of equal charges due to a different sign in the acceleration .
thus these hades data provide a solid constraint on the @xmath6 production whose control will be very important for a robust interpretation of the heavy - ion data . in @xmath4 collisions , however , bremsstrahlung is dominating as can be seen from the middle part of fig .
[ fig_pp125 ] .
because the form of the dilepton invariant mass spectrum from @xmath6 decay and from bremsstrahlung is not completely the same , the form of the @xmath3 and the @xmath4 spectra is not identical and we see in @xmath115 a slight enhancement close to the kinematic limit . in the right part of fig .
[ fig_pp125 ] we compare the hsd results for @xmath25 collisions with the so called quasi - free @xmath4 hades data , used later as the reference spectra @xmath126 for the interpretation of the heavy - ion data . experimentally the quasi - free @xmath4 events have been separated by measuring the proton spectator in the @xmath25 reactions at @xmath127 @xmath128 .
the comparison of the @xmath3 and the @xmath129 data shows clearly that in @xmath129 the proton does not scatter on a quasi free neutron . the kinematical limit for the invariant mass of the dilepton which is @xmath130 gev in @xmath3 and @xmath115 collisions is well exceeded in the @xmath25 collisions .
the largest invariant mass observed ( @xmath131 gev ) corresponds to the maximal invariant mass which is kinematically allowed in the _ three _ body @xmath25 system under the condition that the outgoing proton has at least a momentum of 1.6 @xmath128 .
therefore at the upper end of the invariant mass spectra we have a collision of the proton with the deuteron with a center of mass energy of @xmath132 .
this observation one has to keep in mind for the interpretation of dilepton production in heavy - ion collisions , when the @xmath25 results are used as a reference to discuss the in - medium enhancement of the dilepton yield . in semiclassical transport calculations , like hsd ,
one simulates the deuteron as a bound system of a proton and a neutron which are redistributed in coordinate and momentum space according to the wave function of the paris potential @xcite . the energy of each nucleon ( in the deuteron rest frame )
is taken as @xmath133 , where @xmath134 mev is the binding energy of the deuteron .
we use the energy - momentum relation for free particles to determine the effective mass of the nucleon and then the energy - momentum 4-vector to describe the collision .
an incoming nucleon scatters with one or subsequently with both nucleons of the deuteron but never with the two at the same time .
this gives another kinematics as compared to a true three - body collision and therefore hsd calculations underpredict the dilepton production close to the kinematical limit of @xmath25 collisions .
another problem with the quasi - free @xmath4 scattering is related to the possibility of deuteron formation in the final state .
this is not probable in heavy - ion collisions ( cf .
@xcite ) and not included in hsd . however , as seen from the fig .
[ fig_xseta ] , the process @xmath74 might be important for the @xmath23-production at threshold energies . thus , we include this contribution as an enhanced cross section for @xmath23 production in @xmath4 ( this was not included in our previous work @xcite ) but we do not treat the deuteron formation explicitly in the code . as seen from the right part of fig .
[ fig_pp125 ] in np collisions around @xmath135 gev the @xmath23 contribution turns out to be of the same order of importance as @xmath6-dalitz decays and bremsstrahlung . for @xmath3
( left plot ( a ) ) , @xmath4 ( middle plot ( b ) ) and @xmath25 ( right plot ( c ) ) reactions at 1.25 gev in comparison to the experimental data for @xmath3 ( left ) and quasi - free @xmath4 ( middle and right plots ) reactions from the hades collaboration @xcite .
the individual colored lines display the contributions from the various channels in the hsd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, width=491 ] figure [ fig_pp125 ] ( right ) shows that in @xmath25 collisions the hsd model underestimates the dilepton yield between @xmath136 gev , a region which is accessible in two - body collisions at this energy . a possible candidate to explain this enhancement is the contribution of sub - threshold @xmath12-meson production via excitation and decay of the @xmath82 resonance shown as the dashed - dotted line in fig .
[ fig_xsvm ] .
a very small contribution of this resonant @xmath12 production channel is even seen in @xmath3 collisions ( dotted line on the left plot )
. however , this contribution is not sufficient to describe the experimental data .
this is in line with a recent study by the gibuu group @xcite .
also iqmd calculations fail to describe this part of the spectrum .
[ fig_iqnn125 ] shows the iqmd predictions for @xmath3 and @xmath115 collisions as well compared to @xmath3 and @xmath25 hades data .
we see a very good agreement between hsd and iqmd predictions for the elementary @xmath3 and @xmath4 reactions . for @xmath3 ( left ( a ) ) and @xmath4 ( right ( b ) ) reactions at 1.25 gev in comparison to the experimental data for @xmath3 ( left ) and quasi - free @xmath129 ( right ) reactions from the hades collaboration @xcite .
the individual colored lines display the contributions from the various channels in the iqmd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, width=415 ] the cross section @xmath125 at 1.27 gev , calculated in the hsd model , is compared in fig .
[ fig_pp125dls ] to the @xmath3 ( left ) and @xmath25 ( right ) dls data @xcite .
the theoretical calculations passed through the corresponding dls acceptance filter and mass resolution .
while the agreement between hsd and the data looks reasonable , one has to keep in mind that due to the very broad mass resolution the spectra are strongly distorted at large invariant masses .
there seems to be an underestimation of the last experimental point for @xmath25 , however , the quality of the data does not allow for robust conclusions . for @xmath3 ( left plot ( a ) ) and @xmath25 ( right plot ( b ) ) at 1.27 gev in comparison to the dls data @xcite .
the hsd calculations passed through the corresponding dls acceptance filter and mass resolution . , width=566 ] the differential cross section @xmath125 from hsd calculations for @xmath52 production in @xmath3 reactions at bombarding energies of 2.2 gev in comparison to the hades data @xcite is presented in fig .
[ fig_mpp22 ] ( left ) . the right part of fig .
[ fig_mpp22 ] shows for the same reaction the hsd results for the differential transverse momentum cross sections for @xmath3 at 2.2 gev separated for different invariant mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.55 gev and @xmath139 0.55 gev . also at an energy of 2.2 gev
we see a quite satisfying agreement between theory and experiment .
[ fig_pp2dld ] shows the dilepton differential cross section @xmath125 for the @xmath3 ( left plot ) and @xmath25 ( right plot ) at 2.09 gev from hsd calculations in comparison to the dls data @xcite .
we see also here a good agreement and the fact that the dls as well as the hades data are reproduced with the same theory underlines the consistency of both data sets which have quite different acceptance cuts . from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 2.2 gev in comparison to the hades data @xcite .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( for the colour coding see legend ) .
right ( b ) : hsd results for the differential dilepton transverse momentum cross section for @xmath3 at 2.2 gev and for different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.55 gev and @xmath139 0.55 gev .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolution .
, title="fig:",width=302 ] from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 2.2 gev in comparison to the hades data @xcite .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( for the colour coding see legend ) .
right ( b ) : hsd results for the differential dilepton transverse momentum cross section for @xmath3 at 2.2 gev and for different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.55 gev and @xmath139 0.55 gev .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolution .
, title="fig:",width=302 ] for the @xmath3 ( left plot ( a ) ) and @xmath25 ( right plot ( b ) ) at 2.09 gev in comparison to the dls data @xcite .
the theoretical calculations passed through the corresponding acceptance filters and mass resolutions .
, title="fig:",width=302 ] for the @xmath3 ( left plot ( a ) ) and @xmath25 ( right plot ( b ) ) at 2.09 gev in comparison to the dls data @xcite .
the theoretical calculations passed through the corresponding acceptance filters and mass resolutions .
, title="fig:",width=302 ] finally we come to the hades pp data at 3.5 gev .
although hades has not measured heavy ion collisions at this energy we include these results for completeness .
[ fig_m35 ] shows the differential cross section @xmath125 from hsd calculations for dilepton production in @xmath3 reactions at a bombarding energy of 3.5 gev in comparison to the hades data @xcite .
we present the results including and excluding the bremsstrahlung contribution because at this energy there exist no solid bremsstrahlung calculations . the validity of our approach , to take the kaptari and kaempfer matrix element and to adjust only the phase space , as described in detail in ref .
@xcite , becomes questionable at such a high energy .
the thick lines , labelled in the legend as `` all wo brems '' , show the sum of all channels ( labelled as `` all '' ) without @xmath3 bremsstrahlung . for the distribution of the invariant masses of the dileptons
, bremsstrahlung does not play a major role at this energy in @xmath3 , as expected . in fig .
[ fig_y35 ] we compare the hsd results for @xmath3 at 3.5 gev and for 4 different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev to the hades data @xcite .
the upper 4 plots show the rapidity distribution and the lower 4 plots the transverse momentum spectra . as in fig .
[ fig_m35 ] the thick lines , labelled in the legend as `` all wo brems '' , show the sum of all channels ( labelled as `` all '' ) without @xmath3 bremsstrahlung .
we observe that the rapidity distribution is well described except for invariant masses around the @xmath12 peak where we overpredict the data by a constant factor .
also the transverse momentum distribution is well described by theory with the exception of a region around @xmath140 gev where our calculations overpredict the data .
we note that the present result is in a better agreement with the hades @xmath141 data as compared to the early hsd predictions @xcite due to the following reasons : a lowering of the @xmath23 dalitz dilepton contribution due to the reduction of the @xmath23 production cross section in line with the new hades data ( cf .
discussion in section ii.a.1(1 ) ) ; a lowering of the direct @xmath63 dilepton decay contributions due to the modification of the vector meson production cross section ( cf .
discussion in section ii.a.1(2 ) ) ; a lowering of the @xmath6 dalitz dilepton contribution due to the adjustment of the isospin decomposition in the exclusive channel @xmath142 from fritiof ( cf .
discussion in section ii.a.1 ( 3 ) ) .
the latter reduces the total ( inclusive ) @xmath106 production by a factor up to 1.4 at 3.5 gev and correspondingly the dilepton yield .
this reduction is even larger ( more then a factor of 3 ) for dileptons with high invariant masses and high @xmath141 since they stem dominantly from the dalitz decay of exclusive @xmath6 s simply due to kinematical reasons - a lower amount of associated particles leaves more energy for the generation of high mass @xmath6 s .
an addition reduction of the @xmath6 dilepton yield stems from the different parametrizations used for the differential electromagnetic decay width of the @xmath6 resonance ( cf .
discussion in section vi ) : presently - `` wolf '' @xcite instead of the original `` ernst '' description @xcite with a coupling constant @xmath143 instead of @xmath144 which is consistent with the photon ( @xmath145 ) limit . without these modification
the present hsd version reproduces the results of @xcite .
we speculate that hsd produces slightly too many @xmath6 at 3.5 gev . since the elementary cross section for inclusive @xmath6 production in @xmath3 reactions at this energy is not available ,
the repartition of the pion yield between @xmath6 resonances ( which produce dileptons ) and other resonances ( which do not produce dileptons ) is not well known and may be the origin of the deviation obtained in the @xmath141 spectra . for the mass bin @xmath146 gev
we see that above @xmath147 gev/@xmath148 bremsstrahlung is the dominating source of dilepton production .
we plot the sum of all contributions without bremsstrahlung as well . from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 3.5 gev in comparison to the hades data @xcite .
the individual colored lines display the contributions from the various channels in the hsd calculations ( see color coding in the legend ) .
the thick line , labeled as `` all wo brems '' , shows the total sum of all channels ( labeled as `` all '' ) without @xmath3 bremsstrahlung .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, width=283 ] at 3.5 gev and for 4 different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev in comparison to the hades data @xcite .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( see colour coding in the legend ) .
the tick lines , labelled in the legend as `` all wo brems '' , show the sum of all channels ( labelled as `` all '' ) without @xmath3 bremsstrahlung .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, title="fig:",width=359 ] + at 3.5 gev and for 4 different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev in comparison to the hades data @xcite .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( see colour coding in the legend ) .
the tick lines , labelled in the legend as `` all wo brems '' , show the sum of all channels ( labelled as `` all '' ) without @xmath3 bremsstrahlung .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, title="fig:",width=359 ] we are coming now to @xmath2 reactions .
[ fig_mpnb35 ] compares the differential cross section @xmath125 from hsd calculations for @xmath52 production in @xmath124 reaction at a bombarding energy of 3.5 gev to the hades data @xcite .
the upper part shows the case of the free vector - meson spectral functions while the lower part gives the result for the collisional broadening scenario .
again the thick lines , labeled in the legend as `` all wo brems '' , show the sum of all channels ( labeled as `` all '' ) without @xmath17 bremsstrahlung .
we display both cases since the treatment of bremsstrahlung using the extrapolation of the obe model to such high energy is questionable , as discussed above .
for the same reason the @xmath5 bremsstrahlung presented fig . in [ fig_mpnb35 ] has to be considered with care .
the collisional broadening scenario comes closer to the experimental results in the region around the @xmath12 peak .
we thus find a nice agreement between theory and experiment also for proton - nucleus collisions . from hsd calculations for @xmath52 production in the @xmath124 reaction at a bombarding energy of 3.5 @xmath0gev in comparison to the hades data @xcite .
the upper part ( a ) shows the case of free vector - meson spectral functions while the lower part ( b ) gives the result for the collisional broadening scenario .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( see colour coding in the legend ) .
the tick lines , labelled in legend as `` all wo brems '' , show the sum of all channels ( labelled as `` all '' ) without @xmath3 bremsstrahlung .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions .
, width=283 ]
now we come to the heavy - ion results and start with showing in fig .
[ fig_cc10 ] the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - of hsd calculations for c+c at 1.0 @xmath0gev in comparison to the hades data @xcite .
the hades collaboration has obtained the @xmath22 multiplicity by the average of the multiplicity of charged pions @xcite and we apply the same method for the theoretical calculations .
the upper part displays the results for free vector - meson spectral functions while the lower part shows the result for the collisional broadening scenario .
we note here , and this holds for all dilepton spectra normalized to the number of @xmath22 s , that the normalization is done by the total number of @xmath22 s in @xmath149 , i.e. without applying an experimental acceptance .
this allows for a direct comparison with the published hades results .
the @xmath6 dalitz decay and bremsstrahlung contributions are the dominant channels and contribute with about the same weight to the invariant mass spectra . for invariant masses
@xmath150 gev also the subthreshold @xmath23 channel contributes in an important way .
the different descriptions of the @xmath12 meson become important only at large invariant masses where no experimental data are available .
the figure shows as well the contribution from direct @xmath12 decays when including the @xmath151 resonance which may enhance the @xmath12 meson production at sub - threshold energies as discussed in section ii.b . as seen in the figure
, there is indeed a small contribution but not larger than the experimental error bars . at higher energies other channels
therefore the @xmath151 resonance is not an important source for dilepton production in heavy - ion reactions .
also the in - medium effects due to the collisional broadening of the spectral functions for @xmath12 and @xmath16 mesons is not visible in the final spectra due to the strong contributions from other dilepton sources at low invariant masses where this effect is most pronounced and partly due to the limited experimental mass resolution at high invariant masses which smears out the spectra . fig .
[ fig_iqcc10 ] shows the results of iqmd calculations , including acceptance in the same way as the hsd calculations .
it is remarkable that the two quite sophisticated transport theories predict results which are that similar .
even the channel decomposition is very similar what is all but trivial because the invariant mass spectra depend on many details of the reaction .
they include the @xmath6 dynamics in a nucleus , which we will discuss in section v in more detail , the number of collisions and hence of the spatial distributions of the nucleons in the colliding nuclei , the fermi momentum and the pauli blocking of reactions if final state nucleons would be placed in already occupied phase space regions .
[ fig_cc20 ] shows the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - from hsd calculations for c+c - at 2 @xmath0gev in comparison to the hades data @xcite .
the theoretical calculations passed through the corresponding hades acceptance filters and mass / momentum resolutions which leads to a smearing of the spectra at high invariant mass and particularly in the @xmath16 peak region .
the upper part shows again the case of free vector - meson spectral functions while the lower part presents the result for the collisional broadening scenario . also here
the difference between the in - medium scenarios is of minor importance , partly due to the limited mass resolution which smears out the spectra .
nevertheless , one can conclude that the free calculations predict an enhancement in the region of the @xmath12 mass which is not seen in the experimental data , which are more in favor to the collisional broadening scenario .
[ fig_iqcc20 ] compares the same data with the results from iqmd calculations for c+c - at 2 @xmath0gev which have been acceptance corrected in the same way as the hsd data .
again we see a very good agreement between the two theoretical approaches . only the different parametrizations of the @xmath16 cross section yield deviations at invariant masses around 0.77 gev .
multiplicity - from iqmd for c+c at 2 @xmath0gev in comparison to the hades data @xcite .
the different colour lines display individual channels in the transport calculation ( see legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolutions . ]
multiplicity - from iqmd for c+c at 2 @xmath0gev in comparison to the hades data @xcite .
the different colour lines display individual channels in the transport calculation ( see legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolutions . ]
[ fig_markcl ] displays the mass differential dilepton spectra - normalized to the @xmath22multiplicity - from hsd calculations for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite .
the upper part shows again the case of free vector - meson spectral functions while the lower part gives the result for the collisional broadening scenario .
also in this data set the enhancement around the @xmath12 mass is clearly visible .
for this heavier system the collisional broadening scenario shows a slightly better agreement with experiment than the free result and we expect that for larger systems the difference between the two approaches increases
. multiplicity - from iqmd for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite .
the individual colored lines display the contributions from the various channels in the iqmd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolutions . ]
multiplicity - from iqmd for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite .
the individual colored lines display the contributions from the various channels in the iqmd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolutions . ]
[ fig_iqarkcl ] , which presents the iqmd results for this reaction , shows that the agreement between both theories continues also for heavier systems .
again up to invariant masses of 0.7 gev both invariant mass spectra are almost identical and agree with data .
also the channel decomposition is rather similar . here
one can see again the overestimation of the dilepton yield by iqmd at @xmath152 peak which is related to the enhance @xmath16 production cross section in elementary @xmath4 collisions relative to @xmath3 collisions due to the isospin model used in iqmd ( cf .
section ii.c ) .
the transverse momentum spectra - normalized to the @xmath22 multiplicity - for ar+kcl at 1.75 @xmath0gev have been measured by the hades collaboration for 5 different mass bins @xcite : bin 1 : @xmath137 0.15 gev , bin 2 : 0.13 @xmath138 0.3 gev , bin 3 : 0.3 @xmath138 0.45 gev bin 4 : 0.45 @xmath138 0.65 gev and bin 5 : @xmath139 0.65 gev .
[ fig_mtarkcl ] presents the hades data in comparison with hsd calculations ; on the top without medium effect , on the bottom for the dropping mass scenario .
we see also here a good agreement between theory and experiment .
thus one can conclude that the agreement between theory and experiment ( fig .
[ fig_cc10 ] - fig .
[ fig_mtarkcl ] ) up to @xmath153 0.5 gev is of such a quality that we can use the theory to study the physical processes involved .
multiplicity - for ar+kcl at 1.75 @xmath0gev for 5 different mass bins : bin 1 : @xmath137 0.15 gev , bin 2 : 0.13 @xmath138 0.3 gev , bin 3 : 0.3 @xmath138 0.45 gev bin 4 : 0.45 @xmath138 0.65 gev and bin 5 : @xmath139 0.65 gev in comparison to the hades data @xcite . the upper part ( a ) shows the case of free vector - meson spectral functions while the lower part ( b ) gives the result for the collisional broadening scenario .
the individual coloured lines display the contributions from the various channels in the hsd calculations ( see colour coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter and mass / momentum resolutions . ]
the hades collaboration has recently measured also the dilepton invariant mass spectra for the reaction au+au at 1.25 @xmath0gev .
the analysis is not completed yet .
[ fig_mau125 ] presents the hsd predictions for the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - for this reaction .
the upper part shows the case of free vector - meson spectral functions while the lower part gives the result for the collisional broadening scenario . in this subsection we present the results from the urqmd ( v. 2.3 ) transport model @xcite .
in this model the dilepton afterburner does not contain bremsstrahlung .
it is , however , useful to verify whether it agrees with hsd and iqmd calculations as far as all hadronic dilepton sources are concerned . for the details of the dilepton treatment in urqmd at sis energies we refer the reader to refs .
@xcite .
[ fig_urcc20 ] shows the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - from urqmd calculations for c+c - at 2 @xmath0gev in comparison to the hades data @xcite and fig .
[ fig_urarkcl ] for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite . as one can see from figs .
[ fig_urcc20 ] and [ fig_urarkcl ] the urqmd v. 2.3 substantially overestimates the dilepton yield from the vector mesons .
the problem can be traced back to the description of @xmath12 production in elementary @xmath17 collisions which proceeds via an excitation and decay of heavy baryonic resonances @xmath154 .
their coupling to the @xmath12 channel is not well known and may therefore be overestimated . on the other hand
the dilepton yield at low invariant masses is underestimated for both systems .
this is , first of all , due to the lack of the bremsstrahlung contributions but also due to an underprediction of the @xmath23 yield in urqmd .
we note that the urqmd model is presently under improvement and extension , updated results for the dileptons at sis energies are expected soon @xcite . s - from urqmd for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite .
the individual colored lines display the contributions from the various channels in the hsd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter including mass / momentum resolutions . ]
s - from urqmd for ar+kcl at 1.76 @xmath0gev in comparison to the hades data @xcite .
the individual colored lines display the contributions from the various channels in the hsd calculations ( see color coding in the legend ) .
the theoretical calculations passed through the corresponding hades acceptance filter including mass / momentum resolutions . ]
the primary interest of measuring dilepton production in heavy - ion collisions is to see whether it is a mere superposition of the production in elementary @xmath155 collisions . of course in this threshold energy regime the fermi motion of the nucleons inside a nucleus plays an important role and therefore the question has to be formulated more precisely : is there an in medium enhancement beyond the fermi motion ?
this question we will address in this section .
the hades collaboration has measured the elementary reactions at different beam energies than the heavy - ion reactions , i.e. @xmath3 and quasi - free @xmath4 reactions at 1.25 gev whereas the c+c collisions at 1.0 and 2.0 @xmath0gev and ar+kcl at 1.75 @xmath0gev .
thus , a comparison of elementary reaction data with those of heavy ions at the same energy was not possible experimentally .
therefore , we also have to calculate the reference spectrum @xmath156 at 1.25 gev in order to compare with experimental @xmath157 ratios .
then we show the sensitivity of the ratio @xmath157 to the energy selection of reference spectra @xmath17 which finally might influence the interpretation of in - medium modifications in @xmath75 collisions relative to the @xmath17 . all
calculations presented here have been performed with free vector meson spectral functions .
[ fig_rcc1nn ] , left , shows the mass differential dilepton spectra - normalized to the multiplicity of @xmath22 s and after @xmath23 dalitz yield subtraction - from hsd calculations for c+c at 1.0 @xmath0gev ( solid line ) , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) and at 1.0 gev ( dashed line ) as well as for @xmath25 at 1.25 gev ( dot - dashed line ) .
these calculations are compared to the corresponding hades data from refs .
@xcite - for c+c at 1.0 @xmath0gev and the reference spectra taken as an averaged sum of @xmath3 and quasi - free @xmath129 ( denoted as @xmath158 ) measured at 1.25 gev .
the theoretical calculations passed through the hades acceptance filter for c+c at 1.0 @xmath0gev ( denoted as `` acc : cc@1agev '' ) and mass / momentum resolutions which smears out the high mass region .
the theoretical reference spectra is taken as the averaged sum of dilepton spectra from @xmath122 and free @xmath159 collisions . as seen from the figure
there is no essential difference between our theoretical @xmath25 and @xmath17 spectra up to @xmath160 gev and only for larger invariant masses the enhanced open phase space for @xmath25 compared to @xmath17 becomes important .
[ fig_rcc1nn ] ( r.h.s . ) shows the ratio of the dilepton differential spectra for c+c at 1.0 @xmath0gev to the isospin - averaged @xmath156 .
both spectra are normalized to the @xmath22 multiplicity and the @xmath23 dalitz yield has been subtracted .
the solid and short dashed line present the ratio of c+c at 1.0 @xmath0gev to @xmath17 at 1.25 gev in the acceptance region and in 4@xmath161 , respectively .
the dash - dotted and dashed lines are the corresponding ratios of c+c at @xmath162 to @xmath17 at 1.0 @xmath163 .
if we divide the spectra of c+c at 1.0 @xmath0gev by the @xmath17 spectra at 1.25 gev the ratio is quite flat , as the experiments show as well .
the enhancement in the theory at the upper end of the @xmath22 peak and hence around m=0.15 gev comes in about equal parts from bremsstrahlung and @xmath6 dalitz decay .
we observe as well that the acceptance cuts do not change the enhancement
. therefore we can discuss it in the next section using @xmath164 yields . in this figure
we display as well that the true enhancement , obtained by comparing c+c and @xmath17 at the same energy , is much larger . for @xmath165 gev
it is about a factor of two .
we note that the hades collaboration used @xmath3 and quasi - free @xmath129 spectra at 1.25 gev as a reference @xmath166 spectrum for the ratios of the dilepton yields from @xmath27 to @xmath17 . in order to avoid the additional uncertainties of dilepton production in @xmath25 collision , a system which can not be modeled reasonably good in semi - classical approaches
, we use the reference spectra @xmath126 . as fig .
[ fig_rcc1nn ] shows , both methods are equivalent up to invariant masses of m=0.4 gev . above this value
the ratio increases very fast because in the elementary reactions the limitation due to phase space is more severe than in heavy - ion collisions , where the fermi motion can provide larger invariant masses . these hsd results are confirmed by the iqmd calculations shown in fig .
[ fig_riqcc1nn ] in a form equivalent to fig .
[ fig_rcc1nn ] . multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 1.0 @xmath0gev ( solid line ) , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the hades acceptance filter for c+c at 1.0 @xmath0gev and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - of c+c at 1.0 @xmath0gev ( employing c+c at @xmath167 experimental ( `` acc : cc@1agev '' ) acceptance ) to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 1.0 @xmath0gev ( solid line ) , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the hades acceptance filter for c+c at 1.0 @xmath0gev and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - of c+c at 1.0 @xmath0gev ( employing c+c at @xmath167 experimental ( `` acc : cc@1agev '' ) acceptance ) to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 1.0 @xmath0gev ( solid line ) , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the hades acceptance filter for c+c at 1.0 @xmath0gev and mass / momentum resolutions .
( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - of c+c at 1.0 @xmath0gev ( employing c+c at @xmath167 experimental ( `` acc : cc@1agev '' ) acceptance ) to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev . , title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 1.0 @xmath0gev ( solid line ) , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the hades acceptance filter for c+c at 1.0 @xmath0gev and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - of c+c at 1.0 @xmath0gev ( employing c+c at @xmath167 experimental ( `` acc : cc@1agev '' ) acceptance ) to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev .
, title="fig:",width=321 ] now we step to the energy 2.0 @xmath0gev . in order to compare the experimental data for c+c , measured at two different energies 1.0 and 2.0 @xmath0gev , the hades collaboration transformed the c+c data measured at 2.0 @xmath0gev to the acceptance of c+c at 1.0 @xmath0gev by using - due to lack of statistics - a one - dimensional transformation ( see ref .
we denote this transformation as `` @xmath168 '' in order to distinguish it from the standard three dimensional filtering procedure using the `` 3d '' ( defined above as `` acc : cc@1agev '' ) experimental acceptance matrix ( which depends on @xmath169 and @xmath170 ) , provided by the hades collaboration @xcite for the filtering of theoretical @xmath149 results . fig .
[ fig_rcc2nn ] presents for c+c at 2.0 @xmath0gev the same quantities as fig .
[ fig_rcc1nn ] for c+c at 1.0 @xmath0gev .
the solid line on the left is the result of the hsd calculations , the short dashed line and the dashed dotted line are the isospin - averaged reference spectra @xmath156 at 1.25 gev and @xmath25 at @xmath171 . the dashed line is the reference @xmath17 spectrum at 2.0 gev ; the corresponding hades data are taken from refs .
@xcite . note , that the simulated hsd mass distribution for c+c at 2.0 @xmath0gev has been transformed to the corresponding acceptance in the same way as done for the experimental data using the `` @xmath168 '' transformation .
fluctuations introduced by this procedure result in part from the limited statistics of the relevant hades c+c data set and in part from the necessary re - binning of the latter . the right part of fig .
[ fig_rcc2nn ] shows the ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - of c+c at 2.0 @xmath0gev to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev , applying the c+c at 2.0 @xmath0gev `` @xmath168 '' experimental acceptance ( solid line ) , and in @xmath149 result with the default `` wolf '' differential electromagnetic width for @xmath6 dalitz decay ( short dashed line ) and `` krivoruchenko '' width ( dash - dot - dotted line ) in order to demonstrate the model uncertainties ( cf .
discussion in section vi ) .
also the hsd results for the ratio of c+c at 2.0 @xmath0gev to the reference @xmath17 spectra , taken at 2.0 gev , are shown , including the full `` 3d''- experimental acceptance ( dash - dotted line ) and in @xmath149 ( dashed line ) .
these results show that the experimental data measured up to an invariant mass of @xmath172 gev are compatible with a ratio of one and hence with no in - medium enhancement .
the theoretical results are more complicated . up to an invariant mass of @xmath173 gev theory
predicts a enhancement factor of about 1.8 for @xmath149 .
the ratio at the same nominal energy shows this enhancement even up to invariant masses of @xmath174 gev before the influence of the fermi motion sets in .
the iqmd calculations for c+c at 2.0 @xmath0gev are presented in fig .
[ fig_riqcc2nn ] which shows the same quantities as fig .
[ fig_rcc2nn ] .
we see that the both model agree quite well and the form of the ratio is identical in both approaches . multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 2.0 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the corresponding hades acceptance filter for c+c at 1.0 @xmath0gev ( `` 1d - acc : cc@1agev '' ) and mass / momentum resolutions ( see the discussion in the text ) .
right ( b ) : ratio of the dilepton differential spectra for c+c at 2.0 @xmath0gev - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev with experimental ( `` 1d - acc : cc@1agev '' ) acceptance for c+c at 1.0 @xmath0gev ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 2.0 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the corresponding hades acceptance filter for c+c at 1.0 @xmath0gev ( `` 1d - acc : cc@1agev '' ) and mass / momentum resolutions ( see the discussion in the text ) .
right ( b ) : ratio of the dilepton differential spectra for c+c at 2.0 @xmath0gev - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev with experimental ( `` 1d - acc : cc@1agev '' ) acceptance for c+c at 1.0 @xmath0gev ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 2.0 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the corresponding hades acceptance filter for c+c at 1.0 @xmath0gev ( `` 1d - acc : cc@1agev '' ) and mass / momentum resolutions ( see the discussion in the text ) .
right ( b ) : ratio of the dilepton differential spectra for c+c at 2.0 @xmath0gev - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev with experimental ( `` 1d - acc : cc@1agev '' ) acceptance for c+c at 1.0 @xmath0gev ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for c+c at 2.0 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the hades data @xcite .
the theoretical calculations passed through the corresponding hades acceptance filter for c+c at 1.0 @xmath0gev ( `` 1d - acc : cc@1agev '' ) and mass / momentum resolutions ( see the discussion in the text ) .
right ( b ) : ratio of the dilepton differential spectra for c+c at 2.0 @xmath0gev - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev with experimental ( `` 1d - acc : cc@1agev '' ) acceptance for c+c at 1.0 @xmath0gev ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for of ar+kcl at 1.76 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the corresponding hades data @xcite . the theoretical calculations for @xmath175 and for @xmath17 passed through the hades acceptance filter for ar+kcl and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 , taken at 1.25 gev , employing the ar+kcl experimental acceptance ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for of ar+kcl at 1.76 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the corresponding hades data @xcite .
the theoretical calculations for @xmath175 and for @xmath17 passed through the hades acceptance filter for ar+kcl and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 , taken at 1.25 gev , employing the ar+kcl experimental acceptance ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for of ar+kcl at 1.76 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the corresponding hades data @xcite .
the theoretical calculations for @xmath175 and for @xmath17 passed through the hades acceptance filter for ar+kcl and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 , taken at 1.25 gev , employing the ar+kcl experimental acceptance ( solid line ) .
, title="fig:",width=321 ] multiplicity and after @xmath23 dalitz yield subtraction - from iqmd calculations for of ar+kcl at 1.76 @xmath0gev ( solid line ) and for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) in comparison to the corresponding hades data @xcite .
the theoretical calculations for @xmath175 and for @xmath17 passed through the hades acceptance filter for ar+kcl and mass / momentum resolutions .
right ( b ) : ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 , taken at 1.25 gev , employing the ar+kcl experimental acceptance ( solid line ) .
, title="fig:",width=321 ] fig .
[ fig_rarknn ] ( l.h.s . ) displays the mass differential dilepton spectra for ar+kcl at 1.76 @xmath0gev ( solid line ) - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction .
we compare hsd calculations for ar+kcl at 1.76 @xmath0gev , for the isospin - averaged reference spectra @xmath156 at 1.25 gev ( short dashed line ) and at 1.76 gev ( dashed line ) as well as for @xmath25 at 1.25 gev ( dot - dashed line ) to the corresponding hades data , taken from ref .
the theoretical calculations for @xmath175 and for @xmath17 passed through the hades acceptance filter for ar+kcl and mass / momentum resolutions
. the right part of fig .
[ fig_rarknn ] shows the ratio of the dilepton differential spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev and employing the ar+kcl experimental acceptance ( solid line ) and in @xmath149 ( short dashed line ) .
we display as well the hsd results for the ratio of ar+kcl at 1.76 @xmath0gev to the reference @xmath17 spectrum at the same energy , including the experimental @xmath175 acceptance ( dash - dotted line ) and in @xmath149 ( dashed line ) .
these results show clearly that for invariant masses of @xmath176 gev the data as well as theory are not a mere superposition of the elementary spectra .
the comparison also excludes that this enhancement , observed in heavy - ion collisions , is due to acceptance since the results with acceptance and in @xmath149 are very similar . at larger invariant masses theory and data
do not agree because of the bump at the invariant masses around @xmath177 gev , seen in the experimental @xmath25 reactions , is not reproduced by theory . taking the reference spectra at the same nominal energy theory predicts that this enhancement is constant up to energies of @xmath177 gev .
then the fermi motion becomes important and yields a strong increase of the ratio .
consequently , the experimental ratios of the invariant mass spectra measured in heavy - ion collisions to the isospin - averaged reference spectra @xmath156 taken at 1.25 gev reveals an in - medium enhancement in ar+kcl collisions at 1.75 @xmath0gev whereas in @xmath178 collisions at 2 @xmath0gev this ratio is compatible with one and therefore no in - medium enhancement is seen .
the transport models show an enhancement in all heavy - ion reactions when the reference spectrum is taken at the same energy .
it shows as well that acceptance cuts do not modify this enhancement .
the origin of this enhancement will be discussed in the next subsection . in fig .
[ fig_riqarknn ] we display the same quantities as in fig . [ fig_rarknn ] but for iqmd calculations .
the enhancement of the experimental ratio is confirmed by iqmd calculations , which are in quantitative agreement with the hsd results .
in this section we present the energy and system size dependence of the dilepton yield in @xmath149 as predicted by the hsd calculations in order to study the question of a possible in medium enhancement and to identify eventually its physical origin .
mass differential dilepton spectra - normalized to the @xmath22 multiplicity - obtained in hsd calculations for @xmath4 ( left ( a ) ) and @xmath3 ( right ( b ) ) collisions at 1.0 , 1.25 , 1.75 , 2.0 and 3.5 gev . ,
width=472 ] fig .
[ fig_mnn ] shows the hsd calculations for the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - for @xmath4 ( left ) and @xmath3 ( right ) collisions at 1.0 , 1.25 , 1.75 , 2.0 and 3.5 gev in @xmath149 acceptance . whereas the normalization renders the low invariant mass part to one , independent of the beam energy ,
the spectra at high invariant masses show a strong beam energy dependence , as expected .
bremsstrahlung is not coupled to the number of pions ( or the number of participants which is often assumed to be proportional to the number of @xmath161 s ) but to the number of collisions .
also the production of heavier mesons increases at these energies close to the meson thresholds , either because it becomes easier to produce them directly or because the baryonic resonances which decay into these resonances are more frequently populated .
last but not least , the phase space limitation of the invariant mass changes with energy which makes ratios between invariant mass spectra at different energies complicated . due to the isospin dependence of different processes
the @xmath3 and @xmath4 invariant mass spectra differ in detail but are generally determined by phase space .
we can conclude from fig .
[ fig_mnn ] that the comparison of dilepton data of heavy - ions and of elementary reactions suffer substantially if both are measured at different energies .
this renders quantitative conclusions difficult .
multiplicity- obtained in hsd calculations for the minimal bias c+c , ar+kcl , au+au collisions and for the isospin - averaged reference spectra @xmath126 at 1.0 , 1.25 , 1.75 , 2.0 @xmath0gev in @xmath149 acceptance .
the upper plot ( a ) corresponds to the total dilepton @xmath75 spectra whereas the lower plot ( b ) shows the dilepton spectra after @xmath23 dalitz yield subtraction .
the thick lines on the lower plot stand for the @xmath75 dilepton yields whereas the thin lines show the @xmath17 spectra at the same energy .
, width=359 ] fig .
[ fig_maa ] displays the results of hsd calculations for the @xmath149 mass differential dilepton spectra - normalized to the @xmath22 multiplicity - for the minimal bias symmetric heavy - ion collisions as compared to the isospin - averaged reference spectra @xmath126 .
we display calculations for c+c , ar+kcl , au+au at 1.0 , 1.25 , 1.75 , 2.0 @xmath0gev .
the upper plot corresponds to the total dilepton @xmath75 spectra whereas the lower plot shows the dilepton spectra after @xmath23 dalitz yield subtraction .
the thick lines on the lower plot stand for the @xmath75 dilepton yields whereas the thin lines show the @xmath17 spectra at the same energies .
we see clearly that the dilepton spectra do not scale with the @xmath22 multiplicity for invariant masses
@xmath179 gev .
there is a strong energy and system size dependence of this invariant mass region due to the complicated dynamics of baryon resonances and mesons .
generally the invariant mass spectra in @xmath75 collisions are smoother due to the fermi motion .
[ fig_raa ] presents the ratio @xmath180 of the mass differential dilepton spectra - normalized to the @xmath22 multiplicities - obtained in hsd calculations .
displayed are the ratios of minimal bias c+c , ar+kcl , au+au collisions and of the isospin - averaged reference spectra @xmath126 at the same energy .
the lower plot depicts the same ratios but for the dilepton spectra after @xmath23 dalitz yield subtraction .
clearly we see a quite complex structure .
we start with the energy dependence of the ratio which decreases with energy . including the @xmath23 production this can be clearly seen by comparing the au+au collisions at 1.75 and at 1.25 @xmath0gev as well as by comparing the c+c system at different energies ; @xmath23 subtraction modifies some details but does not change the tendency .
it is also obvious that the ratio increases with the system size .
the ratio for au+au at 1.25 @xmath0gev is about 4.5 , that of c+c at the same energy around 2.5 .
we study now the origin of this enhancement in detail . of the invariant mass differential dilepton @xmath149
spectra - normalized to the @xmath22 multiplicity - from hsd calculations for minimal bias @xmath75 collisions : we display c+c , ar+kcl , au+au collisions in comparison to the isospin - averaged reference spectra @xmath126 at 1.0 , 1.25 , 1.75 , 2.0 @xmath0gev . lower plot ( b ) : the same ratios but for the dilepton spectra after @xmath23 dalitz yield subtraction .
, width=359 ] in fig .
[ fig_rdbr ] we display the enhancement factor in heavy - ion collisions for two different processes : bremsstrahlung and @xmath181 dalitz decay .
we show the ratio @xmath180 of the dilepton yield from hsd calculations of the minimal bias @xmath75 collisions : c+c , ar+kcl , au+au and of the isospin - averaged reference spectra @xmath126 at the same energy .
the upper part shows the contribution from bremsstrahlung , the lower part that from the @xmath6 dalitz decay . we do not expect that bremsstrahlung , one of the dominant sources at beam energies around 1 @xmath0gev , scales with the number of pions ;
therefore the ratio should deviate from one .
it has to be systematically larger than one due to multiple collisions of incoming nucleons in heavy - ion collisions .
we see that the ratio depends on the mass but little on the energy of the system . in au+au collisions where the number of elementary collisions is large
the enhancement can reach a factor of 3 . at higher energies
the bremsstrahlung contribution is not really settled because there are no reliable calculations for the elastic and inelastic elementary channels .
.ratio of @xmath22 mesons and the integrated dilepton yield @xmath182 from @xmath6 dalitz decays for c+c and au+au at b=0.5 fm and 1 @xmath0gev and that from the elementary @xmath17 reactions for different scenarios : with / without fermi motion ( fermi m. ) , with / without secondary meson - baryon collisions ( mb col. ) [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] the other dominant source for dilepton production at beam energies around 1 @xmath0gev is @xmath6 dalitz decay .
one may assume that the @xmath6 dalitz decay scales with the number of pions because the relative ratio is given by the branching ratio but this is not the case .
first of all , we are here in a threshold region where the fermi momentum only can lead to a substantial enhancement of the production .
secondly , pions from @xmath6 decay can be reabsorbed by nucleons and can form again a @xmath6 which may later disappear in a @xmath183 collisions .
this process is even important in systems as small as c+c .
dileptons , on the contrary , can not be reabsorbed and are seen in the detector .
table i shows quantitatively the consequences of these processes for reactions at 1 @xmath0gev .
we compare there the pion and dilepton yield for c+c and au+au for different conditions . if there is neither a fermi momentum ( fermi m. ) nor meson absorption on baryons ( mb col . )
the ratio of @xmath22 s to dileptions corresponds to the branching ratios and the enhancement factor ( last column ) is one , independent of the system size of the heavy - ion reaction .
the fermi motion alone increases the pion yield ( @xmath184 column ) as well as the dilepton yield ( @xmath185 column ) by almost a factor of two . because in the ratio displayed in fig .
[ fig_rdbr ] one divides by the number of pions this ratio remains one for small invariant masses whereas the fermi motion makes the ratio explode for invariant masses close to the phase space boundary .
meson - baryon interactions ( mb coll ) lower the number of pions in heavy - ion collisions , by 15% in c+c collisions and by 47% in au+au collisions because they can lead to a disappearance of the pions if the @xmath186 collision is followed by a @xmath187 collision .
at the same time they enhance the dilepton yield because dileptons do not get reabsorbed and therefore every @xmath6 which is produced contributes to the dilepton yield .
the meson - baryon interactions are therefore the reason that dileptons behave differently than pions .
this cycle of @xmath6 production , @xmath6 decay and @xmath161 reabsorption in @xmath186 collisions , which leads in heavy system to the creation of several generations of @xmath6 s , has been studied already 20 years ago as one of the key elements to the pion dynamics in heavy - ion collision which allows the pions to equilibrate with the system and to serve as a measure of the number of participants @xcite .
the last two lines of table i show that the pion absorption enhances the dilepton production as compared to the pions by a factor of about 1.5 - 1.7 in c+c collisions and by a factor of 3.5 - 4.1 for au+au collisions , i.e. the enhancement grows with the size of the system . of the dilepton yield from the bremsstrahlung channel ( upper part ( a ) ) and @xmath6 dalitz decay ( lower part ( b ) )
- normalized to the multiplicity of @xmath22 .
we display hsd calculations for the ratio of the minimal bias c+c , ar+kcl , au+au collisions and the isospin - averaged reference spectra @xmath126 at the same energy . , width=359 ] the system size effect is demonstrated explicitly in fig .
[ fig_raa175 ] : the left plot shows the ratio of the mass differential dilepton spectra @xmath188 - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - from hsd calculations for the minimal bias c+c , ar+kcl , cr+cr , ti+pb , au+au collisions and of the isospin - averaged reference spectra @xmath126 at 1.75 @xmath0gev .
the right plot shows the same but for the @xmath6 dalitz decay contributions only .
we see also here that the different ratios are separated by a factor which is ( almost ) independent of invariant mass and depends basically on the size of the colliding nuclei since the effect of multiple @xmath6 regeneration increases with the atomic number of the colliding ions .
ratio @xmath180 of the mass differential dilepton spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - from hsd calculations for the minimal bias c+c , ar+kcl , cr+cr , ti+pb and au+au collisions to the isospin - averaged reference spectra @xmath126 at 1.75 @xmath0gev .
right ( b ) : same as the left plot but for the @xmath6 dalitz decay contributions , only .
, title="fig:",width=321 ] ratio @xmath180 of the mass differential dilepton spectra - normalized to the @xmath22 multiplicity and after @xmath23 dalitz yield subtraction - from hsd calculations for the minimal bias c+c , ar+kcl , cr+cr , ti+pb and au+au collisions to the isospin - averaged reference spectra @xmath126 at 1.75 @xmath0gev .
right ( b ) : same as the left plot but for the @xmath6 dalitz decay contributions , only .
, title="fig:",width=321 ] thus , the dilepton enhancement observed in fig . [ fig_raa ] ( and hence also in the experimental spectra ) is due to bremsstrahlung and due to the @xmath6 dynamics in the medium . both are not related to collective effects like the in - medium modifications of spectral functions but are a mere consequence of the presence of other nucleons in the nuclei .
they also appear if no potential but only collisional interactions between the nucleons exist .
this effect grows with the nuclear size which is directly related to an increase of the high baryon density phase from light to heavy - ion collisions . in units of the normal nuclear density @xmath189 for central ( @xmath190 ) au+au ( left plot ( a ) ) and ar+kcl ( right plot ( b ) ) at different energies - 1.25 , 1.7 , 2.0 and 3.5 @xmath0gev .
, width=491 ] this is demonstrated in fig .
[ fig_denst ] which shows the time evolution of the baryon density from hsd in the central cell @xmath191 in units of the normal nuclear density @xmath189 for central ( @xmath190 ) au+au ( left plot ) and ar+kcl ( right plot ) at different energies - 1.25 , 1.7 , 2.0 and 3.5 @xmath0gev . by comparing the ar+kcl and au+au density profiles one sees that the maximum density reached in the central cell is approximately the same in both cases - up to @xmath192 and only slightly grows with increasing energy . however , the high baryon density phase for the heavy au+au nuclei collisions is much longer than for the intermediate ar+kcl system which implies a longer reaction time and a stronger influence of secondary reactions on observables as discussed above .
now we come to the question - how the in - medium effects in vector meson production can influence the ratios .
the dilepton spectra for @xmath124 at 3.5 @xmath0gev , c+c at 1.0 , 2.0 @xmath0gev and for ar+kcl at 1.75 @xmath0gev within the collisional broadening scenario for the vector meson spectral functions have been already presented in sections iii and iv ( cf . figs .
[ fig_mpnb35],[fig_cc10],[fig_cc20],[fig_markcl ] ) in comparison to the hades data as well as our predictions for au+au at 1.25 @xmath0gev ( cf .
[ fig_mau125 ] ) . in fig .
[ fig_maa175 ] we display for reactions at 1.70 @xmath0gev the system size dependence of the 4@xmath161 mass differential dilepton spectra - normalized to the @xmath22 multiplicity - from hsd calculations for minimal bias @xmath75 reactions .
we display the result for the symmetric cr+cr and au+au systems as well as for the asymmetric ti+pb system .
the solid lines stand for the no medium effects scenario whereas the dashed lines show the dilepton yield for the collisional broadening scenario .
the lower plot is a zoom of the upper one for the mass range @xmath193 gev .
first of all we note the growth of the dilepton yield for @xmath194 gev when going from the intermediate cr+cr to the heavy system au+au .
the larger the system mass , the more important is the aforementioned @xmath6 reaction cycle and the more the dilepton production is enhanced as compared to pion production .
as we have discussed already in sections iii and iv , for the collisional broadening scenario one sees clearly the influence of the larger width of the vector meson resonances ( the peaks get smaller and broader ) .
ratio @xmath195 of the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - from hsd calculations for minimal bias au+au ( cr+cr ) collisions and c+c collisions .
this ratio is displaed for the no medium effects and for the collisional broadening scenario . ]
ratio @xmath195 of the mass differential dilepton spectra - normalized to the @xmath22 multiplicity - from hsd calculations for minimal bias au+au ( cr+cr ) collisions and c+c collisions .
this ratio is displaed for the no medium effects and for the collisional broadening scenario . , width=340 ] what would be the consequence of this in - medium effect on the dilepton ratio of @xmath27 spectra to the reference spectrum ? would this observable yield information on the underlying dynamical processes ?
previously we concentrated on the ratio @xmath196 where the reference spectrum is constructed as an average of @xmath3 and @xmath4 yields : @xmath156 .
however , such a ratio would not be well suited for studying in - medium effects in the vector meson mass region due to the limited open phase space in @xmath17 collisions relative to @xmath27 collisions - taken at the same energies - since the fermi motion in @xmath27 extends the kinematical limits , which leads to a fast rise of @xmath196 at larger invariant masses @xmath197 .
moreover , as has been discussed in section iii.a , there is a general problem with @xmath17 as a reference spectrum since experimentally @xmath4 are usually quasi - free @xmath25 reactions . for the beam energies discussed here , in the interesting invariant mass region , @xmath198 gev there are no quasi - free @xmath4 collisions anymore but genuine three - body @xmath25 collisions .
alternatively , the in - medium enhancement can be studied by comparing the yield of a heavy system to that of a light system .
fig [ fig_r17 ] displays for a beam energy of 1.7 @xmath0gev the ratio of the invariant mass differential dilepton spectra for intermediate cr+cr and heavy au+au nuclei and of the light nuclei c+c , which is chosen as a reference spectrum. we study two scenarios the no medium effects and the collisional broading scenario .
one clearly sees that the enhancement for @xmath199 gev due to the multiple @xmath6 production and bremsstrahlung persists when one compares collisions of heavy and light nuclei and can become as large as a factor of two .
thus , c+c collisions can also be used as reference spectra to study such nuclear effects .
moreover , we observe as well that the difference between the two scenarios is small for low invariant masses and becomes only noticeable at invariant masses close to the @xmath12 mass .
however , even there the differences remain moderate . therefore high precision data are required to study the question whether vector mesons are modified by the strongly interacting medium in this energy region . on the other hand
the ratio for auau / cc grows much faster ( for both - the no medium and the in - medium scenario ) than for crcr / cc .
this is due to the enhancement of the vector meson productions by secondary meson - baryon and meson - meson interactions in heavy system relative to the light system .
this effect is hence easy to observe experimentally .
in this section we discuss the different assumptions in different transport approaches due to the lack of experimental information and theoretical knowledge and the consequences for the prediction of these approaches .
the uncertainties related to the production cross sections in elementary reactions have been addressed already in section ii .
there are , however , other sources of uncertainties , in particular for the dilepton production by @xmath6 dalitz decay - the lack of knowledge of the electromagnetic decay width of the @xmath6 resonance , of the mass distribution of the @xmath6 resonance in elementary @xmath17 collisions and of its total decay width as well as different assumptions on how the total decay width is related to the @xmath6 life time .
the differential electromagnetic decay width of a @xmath6 resonance into dileptons of an invariant mass m , @xmath200 , can be related to the delta decay into a nucleon and a virtual photon , @xmath201 , by ( cf .
@xcite ) : @xmath202 where @xmath203 and @xmath204 is the current mass of the @xmath6-resonance . unfortunately there is no direct measurement of the @xmath201 width and starting from the pioneering work of jones and scadron @xcite there is a series of different models @xcite . in the present versions of the hsd , iqmd and urqmd transport
approaches the `` wolf '' model is employed for the electromagnetic decay width @xcite : @xmath205 \nonumber\\ & & m_l(m , m_\delta ) = ( e f g)^2 { m_\delta^2 \over 9 m_n } m^2 \cdot 4 ( m_\delta -m_n -q_0 ) , \ \
e^2=4\pi\alpha , \ \
g=5.44 \nonumber\\ & & m_t(m , m_\delta ) = ( e f g)^2 { m_\delta^2 \over 9 m_n } \left[q_0 ^ 2 ( 5m_\delta -3(q_0+m_n ) ) -m^2(m_\delta+m_n+q_0)\right ] \nonumber \\ & & f = -1.5 { m_\delta + m_n \over m_n ( ( m_n+m_\delta)^2-m^2 ) } \nonumber \\ & & q_0 = ( m^2+p_f^2)^{1/2 } \label{daldel1 } \\ & & p_f^2={(m_\delta^2 -(m_n+m)^2)(m_\delta^2 -(m_n - m)^2 ) \over 4 m_\delta^2 } \nonumber\\ & & \lambda ( m^2,m_n^2,m_\delta^2 ) = m^4+m_n^4+m_\delta^4 - 2 ( m^2 m_n^2 + m^2 m_\delta^2 + m_n^2 m_\delta^2 ) .
\nonumber\end{aligned}\ ] ] resonance to dileptons @xmath200 for different models denoted as `` wolf '' @xcite , `` zetenyi '' @xcite , `` ernst '' @xcite and `` krivoruchenko '' @xcite for different delta masses of 1.232 gev ( a ) , 1.5 gev ( b ) and 1.8 gev ( c ) . , width=472 ] there is a variety of models for the electromagnetic decay width of @xmath6 resonance to dileptons @xmath200 - cf .
figure [ fig_gamee ] shows @xmath206 for different models denoted as `` wolf '' @xcite , `` zetenyi '' @xcite , `` ernst '' @xcite and `` krivoruchenko '' @xcite and for three different delta masses 1.232 gev ( a ) , 1.5 gev ( b ) and 1.8 gev ( c ) .
one can see that in the low mass region , i.e. around the @xmath6 pole mass 1.232 , all approaches give similar results whereas with increasing @xmath204 the differences grow .
the models `` wolf '' and `` ernst '' lead to a similar dilepton yield which is , however , up to a factor of 3 higher then that from the models `` krivoruchenko '' and `` zatenyi '' .
this introduces a systematic error for the prediction of the dilepton yield for large mass dileptons .
the population of high mass delta s in nn reactions depends on the shape of the differential mass distribution which is given by the @xmath6 spectral function .
the spectral function of a @xmath6 resonance of mass @xmath204 is usually assumed to be of the relativistic breit - wigner form : @xmath207 with @xmath208 being the pole mass of the @xmath6 .
the factor @xmath209 is fixed by the normalization condition : @xmath210 where @xmath211 gev is chosen as an upper limit for the numerical integration .
the lower limit for the vacuum spectral function corresponds to the nucleon - pion decay , @xmath212 . in @xmath17 collisions
the deltas can be populated up to the @xmath213 and hence the available part of spectral function is defined by the beam energy .
the shape of spectral function ( and correspondingly the production of high mass delta s ) depends strongly on the total width @xmath214 . due to the lack of experimental information this total width
has to be assumed and different parametrizations exist . for the present hsd calculations we adopt the `` monitz '' model @xcite ( cf . also ref .
@xcite ) : @xmath215 in the urqmd model one employs the `` bass '' parametrization @xcite which differs from the `` monitz '' model ( [ widthdel ] ) by the formfactor ( [ widthdel1 ] ) : @xmath216 in the left part ( a ) of fig .
[ fig_gamt ] we show the mass dependence of the total width @xmath217 from different models : `` const '' - a constant width @xmath218 gev , `` monitz '' - from eq .
[ widthdel ] ( cf .
@xcite ) , `` bass '' - from [ fqurqmd ] ( cf .
@xcite ) as well as the parametrization used in the iqmd model @xcite denoted as `` iqmd '' .
we observe substantial differences between the models , especially for large mass @xmath6 .
these differences become more important at higher energies . for lower energies , especially for the 1 @xmath0gev data
, phase space limits the @xmath6 masses to @xmath219 gev .
( left plot ( a ) ) , life time ( middle plot ( b ) ) and the spectral function ( right plot ( c ) ) from different models : `` const '' - the constant width @xmath220 gev , `` monitz '' - from eq .
[ widthdel ] ( cf .
@xcite ) , `` bass '' - from [ fqurqmd ] ( cf .
@xcite ) , `` iqmd '' - from @xcite . , title="fig:",width=347 ] ( left plot ( a ) ) , life time ( middle plot ( b ) ) and the spectral function ( right plot ( c ) ) from different models : `` const '' - the constant width @xmath220 gev , `` monitz '' - from eq .
[ widthdel ] ( cf .
@xcite ) , `` bass '' - from [ fqurqmd ] ( cf .
@xcite ) , `` iqmd '' - from @xcite . , title="fig:",width=226 ] the total decay width is related to the life time of the resonances - an other important quantity for the transport approaches - by @xmath221 the lifetime as a function of @xmath204 is illustrated in the middle part ( b ) of fig .
[ fig_gamt ] .
the lifetime of large mass @xmath6 is in the `` bass '' parametrization up to three times lower than in the monitz parametrization .
@xmath6 s of such a high mass are rare , however , as can be see from the right part ( c ) of fig .
[ fig_gamt ] which shows the mass dependence of the spectral function for different parametrizations of the width .
now we show how the uncertainties in the modelling of the total @xmath6 width and of the electromagnetic decay width affect the final results for the dilepton yield .
we start out with a simple example : the dilepton yield from the @xmath6 dalitz decay is a convolution of the mass distribution of the @xmath6 resonances - which we take for our model study to be defined by the spectral function @xmath222 ( eq.([spfund ] ) ) - and the @xmath6 mass dependent branching ratio for the electromagnetic decay into dileptons which is defined as a ratio of electromagnetic partial width @xmath223 and the total width @xmath224 : @xmath225 where the expression ( [ dndmee ] ) has been re - written in terms of the @xmath6 life time using relation ( [ tau ] ) .
2 models for the partial electromagnetic decay width @xmath223 : 1 ) solid line : total width - `` bass '' , electromagnetic - `` wolf '' ; 2 ) dot - dashed line : total width - `` bass '' , electromagnetic - `` krivoruchenko '' 3 ) dashed line : total width - `` const '' , electromagnetic - `` wolf '' 4 ) dot - dot - dashed line : total width - `` const '' , electromagnetic - `` krivoruchenko '' .
right(b ) : the dilepton yield as a function of invariant dilepton mass for the 2 assumptions of the life time @xmath226 : solid line - `` bass '' , dashed line - `` const '' life time , while using the `` bass '' total width for spectral function and the `` wolf '' model for the partial electromagnetic decay width in both cases.,title="fig:",width=321 ] 2 models for the partial electromagnetic decay width @xmath223 : 1 ) solid line : total width - `` bass '' , electromagnetic - `` wolf '' ; 2 ) dot - dashed line : total width - `` bass '' , electromagnetic - `` krivoruchenko '' 3 ) dashed line : total width - `` const '' , electromagnetic - `` wolf '' 4 ) dot - dot - dashed line : total width - `` const '' , electromagnetic - `` krivoruchenko '' .
right(b ) : the dilepton yield as a function of invariant dilepton mass for the 2 assumptions of the life time @xmath226 : solid line - `` bass '' , dashed line - `` const '' life time , while using the `` bass '' total width for spectral function and the `` wolf '' model for the partial electromagnetic decay width in both cases.,title="fig:",width=321 ] in the left side ( a ) of fig .
[ fig_dndm ] we show the dilepton yield for two different assumptions for @xmath224 and for @xmath223 : 1 ) solid line : total width - `` bass '' , electromagnetic - `` wolf '' ; 2 ) dot - dashed line : total width - `` bass '' , electromagnetic - `` krivoruchenko '' ; 3 ) dashed line : total width - `` const '' , electromagnetic - `` wolf '' ; 4 ) dot - dot - dashed line : total width - `` const '' , electromagnetic - `` krivoruchenko '' .
the variation of @xmath224 changes the dilepton yield only marginally as long as the same electromagnetic decay width is used - cases 1),3 ) and 2),4 ) .
the reason can easily be seen from eq .
( [ dndmee ] ) : the total width @xmath224 enters in the numerator of spectral function ( eq . [ spfund ] ) and in the denominator of the branching ratio and thus cancels .
the only remaining dependence comes from the denominator of eq .
( [ spfund ] ) but far from the pole mass @xmath224 this term is small as compared to the other part of the denominator .
oppositely , for a fixed total width @xmath224 the variation of @xmath223 leads to differences of the dilepton yield up to the factor of 3 for high invariant masses - cases 1),2 ) and 3),4 ) .
thus we can conclude that different assumptions on the total width @xmath224 have little influence on the invariant mass distribution of dileptons whereas the lack of knowledge of @xmath223 introduces an uncertainty of up to a factor of three for the dilepton yield from @xmath6 decay at large invariant masses .
different assumptions have been made of how to relate @xmath224 to the lifetime of the @xmath6 .
we do not discuss here the rational behind the different approaches .
rather we concentrate on the consequences for the dilepton yield . in hsd the total width for the @xmath6 production ( i.e. that which enters the spectral function @xmath222 ) is the same as the width used to determine the life time eq .
( [ tau ] ) . in this case
, we have a cancellation of the total width in eq .
( [ dndmee ] ) as discussed above , which leads to the low sensitivity of dilepton spectra to different @xmath224 . in the urqmd model ( cf .
e.g. the corresponding discussion in ref .
@xcite , section 3.3.4 ) , on the contrary , the width used for the @xmath6 production differs from that in the life time definition ( [ tau ] ) , so there is no cancellation of the widths any more , rather the ratios of the two widths enters the eq .
( [ dndmee ] ) . in the right part ( b ) of fig .
[ fig_dndm ] we demonstrate the consequences of the different life time definitions .
we employ in all cases the `` wolf '' parametrization of @xmath227 and @xmath224 of `` bass '' but vary the description of the life time . the full blue line shows the dilepton yield under the assumption that @xmath224 of `` bass '' determines the lifetime ( eq .
( [ tau ] ) ) whereas the dashed red line shows the result assuming that for the calculation of the life time ( eq .
( [ tau ] ) ) a constant width of 120 mev is employed . for a constant width
we observe a strong enhancement which is mainly related to the large contribution of the high mass @xmath6 s to the dilepton yield .
this is illustrated in fig .
[ fig_dal_md ] where we show the contribution of @xmath6 s from different mass ranges to the dilepton yield .
the sum of all 4 bins gives the solid curve of fig .
[ fig_dndm ] .
one has to keep in mind , however , that in real @xmath17 collisions at low energies the high mass tail of the @xmath6 distribution is strongly suppressed due to the limitation of the phase space .
s from the 4 mass bins to the dilepton yield @xmath228 : 1 ) @xmath229 gev , 2 ) @xmath230 gev , 3 ) @xmath231 gev , 4 ) @xmath232 gev .
calculations are done using the `` bass '' total width for spectral function and `` wolf '' model for the partial electromagnetic decay width .
, width=321 ] now we extend our study of systematic errors to @xmath3 and heavy - ion calculations .
for this purpose we use the hsd model . mass distribution from hsd for the central c+c collisions at 2 @xmath0gev for the 2 model cases for the total width : solid line - `` monitz '' , short dashed - `` bass '' width .
the thin dash - dotted and dashed lines show the spectral function @xmath222 calculated with `` monitz '' and `` bass '' widths , correspondingly .
the dotted line stands for @xmath233 from hsd for c+c collisions at 5.0 @xmath0gev with `` bass '' width ( scaled to the maximum of @xmath233 at 2 @xmath0gev for easy comparison of the shape of mass distributions),width=321 ] .
[ fig_dndmd ] the @xmath6 resonances can be produced dominantly in @xmath17 or @xmath5 collisions .
the mass distribution of the produced @xmath6 s @xmath234 is defined by the spectral function @xmath222 ( eq.([spfund ] ) ) integrated over the corresponding phase space which depends on the invariant energy @xmath80 of the @xmath17 or @xmath5 collisions and the masses of the final associated particles @xmath235 ( e.g. @xmath236 ) . at low energies
the phase space leads to the suppression of high mass @xmath6 s .
we start with the time integrated @xmath6 mass distribution @xmath233 .
it is shown in fig .
[ fig_dndmd ] for central c+c collisions at 2@xmath0gev .
we display the mass distribution for 2 choices for the total width : the solid line displays the calculation for the `` monitz '' width , the short dashed line that for the `` bass '' width .
for comparison we also show the spectral function @xmath222 ( scaled to the maximum of @xmath233 ) for the both widths . due to
the limited available energy in low energy heavy - ion collisions only a part of the full spectral function can be explored ( the absorption and rescattering effects for c+c collisions do not distort the initial production shape of @xmath6 mass distribution too much ) .
this lowers the uncertainties of the predicted dilepton yields related to the very high mass tail of the distribution .
going to higher energies the phase space opens more and more and the high mass tail of spectral function can be populated .
this is shown by the dotted line which displays @xmath233 for c+c collisions at 5.0 @xmath0gev employing the `` bass '' width which we scaled to the maximum of @xmath233 for c+c at 2 @xmath0gev for easy comparison of the shape of corresponding mass distributions . from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 3.5 gev in comparison to the hades data @xcite .
right(b - e ) : the hsd results for the transverse momentum spectra for @xmath3 at 3.5 gev and for 4 different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev in comparison to the hades data @xcite .
the individual lines similar to the left part .
the dash - dot - dotted and the dashed lines shows the delta dalitz contribution and the corresponding total sum of all channels without @xmath3 bremsstrahlung as in figs .
[ fig_m35],[fig_y35 ] - for the `` wolf '' electromagnetic decay width .
the dash - dotted line stands for the parametrization using `` krivoruchenko '' width , the solid line is the corresponding sum . ,
width=325 ] from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 3.5 gev in comparison to the hades data @xcite .
right(b - e ) : the hsd results for the transverse momentum spectra for @xmath3 at 3.5 gev and for 4 different mass bins : @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev in comparison to the hades data @xcite .
the individual lines similar to the left part .
the dash - dot - dotted and the dashed lines shows the delta dalitz contribution and the corresponding total sum of all channels without @xmath3 bremsstrahlung as in figs .
[ fig_m35],[fig_y35 ] - for the `` wolf '' electromagnetic decay width .
the dash - dotted line stands for the parametrization using `` krivoruchenko '' width , the solid line is the corresponding sum . , width=325 ] [ fig_gameehsd ] multiplicity - from hsd calculations for central ( b=0.5 fm ) c+c collisions at 2 @xmath0gev in comparison to the hades data @xcite .
the dash - dot - dotted and the dashed lines shows the delta dalitz contribution and the corresponding total sum of all channels as in fig .
[ fig_cc20]- for the `` wolf '' electromagnetic decay width .
the dash - dotted line stands for the parametrization using `` krivoruchenko '' width , the solid line is the corresponding sum .
right(b ) : similar to the left part - the mass differential dilepton spectra as a function of invariant dilepton mass for the 2 model cases of life time @xmath226 : solid line - `` const '' , dashed line - `` bass '' life time , while using the `` bass '' total width for the @xmath6 production and dynamics and `` wolf '' model for the partial electromagnetic decay width in both cases . ,
title="fig:",width=321 ] multiplicity - from hsd calculations for central ( b=0.5 fm ) c+c collisions at 2 @xmath0gev in comparison to the hades data @xcite .
the dash - dot - dotted and the dashed lines shows the delta dalitz contribution and the corresponding total sum of all channels as in fig .
[ fig_cc20]- for the `` wolf '' electromagnetic decay width .
the dash - dotted line stands for the parametrization using `` krivoruchenko '' width , the solid line is the corresponding sum .
right(b ) : similar to the left part - the mass differential dilepton spectra as a function of invariant dilepton mass for the 2 model cases of life time @xmath226 : solid line - `` const '' , dashed line - `` bass '' life time , while using the `` bass '' total width for the @xmath6 production and dynamics and `` wolf '' model for the partial electromagnetic decay width in both cases . , title="fig:",width=321 ] similar to the left ( a ) part of fig .
[ fig_dndm ] we demonstrate in fig .
[ fig_pp35kriv ] the consequences of the variation of the electromagnetic decay width on the differential cross section @xmath125 ( left ( a ) ) and the transverse momentum spectra ( right ( b - e ) ) for 4 different mass bins ( @xmath137 0.15 gev , 0.15 @xmath138 0.47 gev , 0.47 @xmath138 0.7 gev and @xmath139 0.7 gev ) from hsd calculations for @xmath52 production in @xmath3 reactions at a bombarding energy of 3.5 gev .
the dash - dot - dotted and the dashed lines show the delta dalitz contribution and the corresponding total sum of all channels without @xmath3 bremsstrahlung as in figs .
[ fig_m35],[fig_y35 ] - for the `` wolf '' electromagnetic decay width .
the dash - dotted line stands for the parametrization using `` krivoruchenko '' width , the solid line is the corresponding sum .
we point out that we have selected the @xmath3 reaction at 3.5 gev here since at this high energy the open phase space is large enough to populate the high mass @xmath6 s .
thus , one expects a large deviation in the dilepton mass spectra coming from the high mass tail of the @xmath6 spectral function - as follows from figs .
[ fig_gamee ] and [ fig_dndm](left ) - compared to the low energy reactions where the available energy limits the production of heavy @xmath6 s .
furthermore , figure [ fig_pp35kriv ] ( right ) demonstrates the sensitivity of @xmath141 distribution to the form of the electromagnetic decay width . in spite
that the deviation is bigger for the bin with the largest dilepton masses ( @xmath237 gev ) , this effect is not visible in the final @xmath141 spectra due to the dominant contributions from the direct decay of vector mesons . for the lower mass bins ( 0.15 @xmath138 0.47 gev , and 0.47 @xmath138 0.7 gev ) the difference is better observed in the final @xmath141 spectra .
thus , the measurement of the @xmath141 distributions at various mass bins can help in distinguishing of different models .
we continue with the comparison of the final mass differential dilepton spectra for central c+c collision at 2 @xmath0gev fig .
[ fig_gameehsd ] ( left ( a ) ) .
the legend for the individual lines is the same as in fig .
[ fig_pp35kriv ] .
we observe similar deviations as obtained within the `` convolution '' model - see the left ( a ) part of fig .
[ fig_dndm ] .
thus , we conclude that the uncertainty of the electromagnetic decay width of the @xmath6 resonance translates to an uncertainty of about a factor of 1.5 in the dilepton yield from @xmath6 decays in heavy ion collisions . for large invariant masses of the @xmath6 this uncertainty reaches even a factor of 3 .
however , these large invariant masses are only populated at beam energies at which @xmath23 production becomes important with the consequence that the @xmath23 dalitz decay and bremsstrahlung are the dominant sources for dilepton production .
therefore the uncertainty of the dilepton yield from large mass @xmath6 decay has little influence on the measured total dilepton yield at large invariant masses as well as on the ratio @xmath196 - cf .
dash - dot - dotted line in fig .
[ fig_rcc2nn ] ( b ) .
the simulations for the different assumptions about the life time for the central c+c at 2 @xmath0gev are presented in the right part ( b ) of fig .
[ fig_gameehsd ] .
the assumptions correspond to that of the right ( b ) part of fig .
[ fig_dndm ] .
the solid line - `` const '' - displays the results assuming a constant life time whereas the dashed line show the result assuming the `` bass '' life time . in both cases the `` bass '' total width has been employed for the @xmath6 spectral function and the `` wolf '' model has been used for the partial electromagnetic decay width .
one can see that the two assumptions about the lifetime yield an uncertainty of a factor of 2 , slightly less than the factor we obtained for elementary reactions .
[ fig_dndmd ] ) .
we would like to stress here that the uncertainties in the electromagnetic decay width of @xmath6 resonance as well as that of total width / life time of the @xmath6 can be reduced by measuring the dilepton yield in @xmath5 reactions at different energies .
such a measurement would allow for preciser predictions than presently possible .
the introduction of the electromagnetic @xmath238 transition formfactor @xmath239 for the @xmath6 dalitz decay has been studied in ref .
@xcite within the gibuu transport model for @xmath17 and @xmath240 reactions . the model from refs
@xcite has been choosen for the @xmath238 transition formfactor which is based on the vector dominance model ( vdm ) model assuming that the virtual photon is converted first to a @xmath241 meson , i.e. the transition @xmath242 can be considered as @xmath243 . for that one
needs to extrapolate the @xmath238 transition formfactor from the space - like region to the time - like region where its strength is unknown experimentally .
resonance to dileptons @xmath200 using the `` krivoruchenko '' model for different @xmath6 masses of 1.232 gev ( a ) , 1.5 ( b ) and 1.8 gev ( c ) : the thick lines - with the @xmath238 formfactor @xmath239 from ref .
@xcite , the thin lines - without the @xmath238 formfactor .
right ( b ) : the dilepton yield as a function of the invariant dilepton mass using the `` krivoruchenko '' model for the electromagnetic decay width and the `` bass '' model for the total width : the solid lines - with including the @xmath238 formfactor @xmath239 , the dashed line - without the @xmath238 formfactor .
, title="fig:",width=283 ] resonance to dileptons @xmath200 using the `` krivoruchenko '' model for different @xmath6 masses of 1.232 gev ( a ) , 1.5 ( b ) and 1.8 gev ( c ) : the thick lines - with the @xmath238 formfactor @xmath239 from ref .
@xcite , the thin lines - without the @xmath238 formfactor .
right ( b ) : the dilepton yield as a function of the invariant dilepton mass using the `` krivoruchenko '' model for the electromagnetic decay width and the `` bass '' model for the total width : the solid lines - with including the @xmath238 formfactor @xmath239 , the dashed line - without the @xmath238 formfactor .
, title="fig:",width=283 ] multiplicity - for c+c at 2.0 @xmath0gev : the thick lines - with the @xmath238 formfactor @xmath239 from ref .
@xcite , the thin lines - without the @xmath238 formfactor . here
the dashed lines corresponds to the `` wolf '' model for the electromagnetic decay width , the dash - dotted lines - for the `` krivoruchenko '' model .
, width=321 ] in fig .
[ fig_iachel ] we demonstrate the effect of the electromagnetic @xmath238 transition formfactor : the left plot ( a ) shows the electromagnetic decay width of the @xmath6 resonance into dileptons @xmath200 using the `` krivoruchenko '' model for different @xmath6 masses of 1.232 gev ( a ) , 1.5 gev ( b ) and 1.8 gev ( c ) : the thick ( thin ) lines - with ( without ) the @xmath238 formfactor @xmath239 from ref . @xcite .
the right plot ( b ) displays the dilepton yield using the `` convolution '' model ( eq .
( [ dndmee ] ) ) with the `` bass '' total width .
one can see that the inclusion of the formfactor leads to an enhancement of the dilepton yield up to a factor of 10 at @xmath244 gev .
we have investigated the consequences of the electromagnetic @xmath238 transition formfactor for heavy - ion collisions for c+c reactions at 2 @xmath0gev where the @xmath6 channel is one of the dominant channels .
[ fig_cc20ich ] shows the results of the hsd calculations for 2 different models ( in line with our discussion above on the model uncertainties ) for the electromagnetic decay width - `` krivoruchenko '' ( dashed lines ) and `` wolf '' ( dash - dotted lines ) which provide the lower and upper limit for the effects of the form factor for the final spectra ( solid lower and upper lines ) .
one can conclude that the introduction of the @xmath238 transition formfactor @xcite leads to overestimation of the dilepton yield in heavy - ion collisions , i.e. is not in line with the hades data . here
we discuss the uncertainties related to the implementation of the electromagnetic pion formfactor @xmath245 motivated by the vector dominance model ( vdm ) model for the @xmath4 bremsstrahlung contribution as advocated in the obe model in ref .
@xcite . with the help of this formfactor one hopes to account for the dilepton radiation from the internal charged pion exchange line in @xmath246 processes assuming vector dominance , i.e. that the photon couples to dileptons via a @xmath13 meson .
this diagramm does nt exist for the @xmath3 reaction , so the enhancement should be seen only in @xmath4 dilepton yield .
there is a debate if the virtual photon converts fully ( i.e. by 100% ) to @xmath12 meson ( @xmath247 ) @xcite or by 50% only @xcite and the rest decays directly into dileptons ( @xmath248 ) . in ref .
@xcite the electromagnetic pion formfactor @xmath245 has been parametrized as @xmath249 where @xmath250 = 1.9 @xmath251 .
the width @xmath252 is given in ref .
@xcite as @xmath253 with @xmath254 .
the parameter @xmath255 fm is an interaction radius , @xmath256 gev .
according to the obe calculations of @xcite for the @xmath4 reaction at 1.25 gev the incorporation of the formfactor @xmath245 leads to the enhancement of the bremsstrahlung contribution and a better agreement with the hades data for quasi - free @xmath4 scattering . following ref .
@xcite we have performed a model study by including the formfactor from @xcite in our calculations of @xmath4 bremsstrahlung by simply multiplication of the parametrized obe results from @xcite used in hsd by the formfactor of eq .
( [ eqfpi ] ) .
indeed , this provides an upper limit since we can not distinguish the individual diagramms in our parametrization of the bremsstrahlung cross section .
we obtain a good agreement with the full obe calculations from @xcite , as shown in fig .
[ fig_pdfpi ] , which presents the hsd results for the dilepton differential cross section @xmath125 for @xmath4 ( left plot ( a ) ) and @xmath25 ( right plot ( b ) ) reactions at 1.25 gev including the electromagnetic pion formfactor for the bremsstrahlung channel ( denoted as `` brems .
nn with @xmath257 '' ) in comparison to the `` standard '' hsd calculations without formfactor ( denoted as `` brems .
nn w / o @xmath257 '' ) as in fig . [ fig_pp125 ] . the dashed line ( denoted as `` obe with @xmath258 '' )
shows the obe results for @xmath123 from ref .
@xcite . for @xmath4 ( left plot ( a ) ) and @xmath25 ( right plot ( b ) ) reactions at 1.25 gev including the electromagnetic pion formfactor for the bremsstrahlung channel ( denoted as `` brems .
nn with @xmath257 '' ) in comparison to the `` standard '' hsd calculations without formfactor ( denoted as `` brems .
nn w / o @xmath257 '' ) as in fig . [ fig_pp125 ] . the dashed line ( denoted as `` obe with @xmath258 '' )
shows the obe results for @xmath123 from ref .
] as seen form fig .
[ fig_pdfpi ] the inclusion of the formfactor does nt explain the experimental hades quasi - free @xmath123 data .
we check now how the formfactor will change the heavy - ion results where we have reliable experimental constraints from the hades measurements .
the hsd results for the dilepton differential cross section @xmath125 for c+c at 1.0 @xmath0gev ( left ( a ) ) 2.0 @xmath0gev ( middle ( b ) ) and ar+kcl ( right ( c ) ) reactions at 1.75 @xmath0gev are shown in fig .
[ fig_aafpi ] .
one can clearly see a sizeable overestimation of the dilepton yields for all systems which brings us to the conclusion that to include a form factor is not supported by the experimental data on heavy - ion collisions . for c+c at 1.0 @xmath0gev ( left ( a ) ) 2.0 @xmath0gev ( middle ( b ) ) and ar+kcl ( right ( c ) ) reactions at 1.75 @xmath0gev .
the lines description as in fig .
[ fig_pdfpi ] . ,
title="fig:",width=226 ] for c+c at 1.0 @xmath0gev ( left ( a ) ) 2.0 @xmath0gev ( middle ( b ) ) and ar+kcl ( right ( c ) ) reactions at 1.75 @xmath0gev . the lines description as in fig .
[ fig_pdfpi ] . ,
title="fig:",width=226 ] for c+c at 1.0 @xmath0gev ( left ( a ) ) 2.0 @xmath0gev ( middle ( b ) ) and ar+kcl ( right ( c ) ) reactions at 1.75 @xmath0gev . the lines description as in fig .
[ fig_pdfpi ] .
, title="fig:",width=226 ]
we have studied the production of dileptions in @xmath3 , @xmath4 , @xmath26 and @xmath27 collisions at energies between 1 and 3.5 @xmath0gev by comparing the results of three independent transport approaches - hsd , iqmd and urqmd - with all existing data in this energy domain .
these data allowed for the first time to study the cycle of creation and absorption of the @xmath6 resonance in heavy ion reaction which has been theoretically predicted since long . despite of common general ideas of transport approaches which are based on the modeling /
parameterizations of baryon - baryon , meson - baryon and meson - meson elementary reactions with further dynamical evolution including the propagation in a self - generated mean - field potential and explicit interactions , the models differ in the actual realization and underlying assumptions where no control from experimental data is available . especially for one of the dominant channels for the dilepton production in this energy domain , the @xmath6 dalitz decay , the experimental results do not allow for a robust parametrization of the input for these transport theories .
neither the spectral function of the @xmath6 nor the differential decay width of the @xmath6 into dileptons are well known . at energies around 1 @xmath0gev
the @xmath6 production is in agreement with the isospin model which assumes that the difference @xmath6 states are produced according to the isospin clebsch gordon coefficients at higher energies , where 2 pion channels contribute substantially , little information is available on population of the different @xmath6 states .
this situation will hopefully change with the planned experiments on dilepton production in @xmath5 collisions .
such an information would substantially improve the predictive power of transport theories for heavy ion results .
similar is the situation for the bremsstrahlung contribution which turns out to be the dominant channel for the low energy collisions at 1 @xmath0gev .
the present obe models provide different predictions as compared to the soft - photon - approximation and do not agree among each other .
more precise data , especially on dilepton production in elementary @xmath3 , @xmath4 and especially @xmath5 collisions for different energies are need to reduce this systematic error .
they would allow for a more reliable predictions of bremsstrahlung in transport approaches .
we stress the importance of providing such constraint form the experimental side since the transport models are the only reliable tool to study the physics of heavy ion collisions at those energies where neither thermal models nor hydrodynamic models are applicable because the created matter is far from equilibrium . despite of these uncertainties the results of different transport approaches for the final dilepton yield agree quite well among each other even if there are the deviations in the channel decompositions .
the data are in between the systematical error of the transport predictions .
we have started our investigation with the dilepton spectra from elementary reactions which can be described as a superposition of the emission from known dilepton sources . in @xmath3 collisions at energies of around 1 gev dileptons stem dominantly from the @xmath6 dalitz decay whereas in @xmath4 collisions the bremsstrahlung radiation becomes equally important . at higher energies the @xmath23 production sets in and
contributes to the invariant mass range @xmath259 gev . at
higher invariant masses the vector meson decays dominate but the data are presently not precise enough to allow for firm conclusions .
new experimental differential data would be very useful to check the underlying model assumptions .
our study demonstrates that in heavy - ion reactions the dilepton production for invariant masses below @xmath260 gev can not be interpreted as a simple convolution of the average dilepton yield from @xmath4 and @xmath3 collisions times the number of elementary collisions .
the presence of a nuclear medium manifests itself in several ways : first of all the fermi motion of nucleons in nuclei smears out the energy distribution of primary @xmath17 collisions substantially .
this has a big influence on the particle production at ( sub-)threshold energies .
the fermi motion enhances the pion as well as the dilepton yields in @xmath27 collisions at threshold energies by up to a factor of two .
the enhancement is , however , identical for pions and dileptons .
therefore , if the fermi motion would be the only difference between @xmath27 and @xmath17 collisions , one could expect that if one normalizes the dilepton yield to the pion multiplicity , as done in the experimental analysis , no enhancement would be observed .
the real situation is quite different : an enhancement of the dilepton yield in @xmath27 relative to @xmath17 is observed experimentally even if one normalizes the dilepton yield by the @xmath22 multiplicity .
the experimental enhancement is , however , plagued partly by the use of @xmath25 collisions instead of @xmath4 collisions because in the interesting kinematical regime the @xmath25 collisions are true three - body collisions and can not be interpreted as quasi - free @xmath4 reactions .
so the true enhancement as compared to elementary collisions can not be inferred from present data for invariant masses above 0.5 gev .
we have analysed this enhancement in detail and found two origins : the first reason is the bremsstrahlung radiation from @xmath4 and @xmath3 reactions which does not scale with the pion number ( i.e. the number of participants ) rather with the number of elementary elastic collisions .
the second reason is the shining of dileptons from the intermediate @xmath6 s , which take part in the @xmath261 and @xmath262 reaction cycle .
this cycle produces a number of generations of @xmath6 s during the reaction which increases with the size of the system . at the end
only one pion is produced but each intermediate @xmath6 has contributed to the dilepton yield because emitted dileptons do not get absorbed .
this leads to an enhancement of the dilepton yield as compared to the final number of pions .
thus , the enhancement confirms the predictions of transport theories that in heavy - ion collisions several generations of @xmath6 s are formed which decay and are recreated by @xmath186 reactions .
accordingly , the dilepton data from @xmath27 reactions shed light on the @xmath6 dynamics in the medium . in the investigated invariant mass range , @xmath263 gev , we do not find evidence that the observed enhancement of the dilepton yield in heavy - ion collisions over the elementary reactions requires the assumption of conventional in - medium effects like a modification of the spectral functions of the involved hadrons .
theory predicts such a modification for vector mesons and therefore for invariant masses @xmath198 gev . more precise data are needed to draw robust conclusions on the in - medium modifications in this invariant mass range .
we summarize with the final remark that the ratio of dilepton yields @xmath157 is a sensitive observable which allows to penetrate the intermediate phase of the heavy - ion reaction and sheds light on the @xmath6 dynamics which is not accessible by the hadronic observables .
thus , the hades data provide the first experimental constraint on this issue in heavy - ion collisions at sis energies .
moreover , by measuring this ratio at a low bombarding energy one can get access to the bremsstrahlung radiation since it becomes the dominant process there .
one expects to observe in this case the scaling of the ratio with the number of binary collisions rather than with the number of pions . a precise measurement of the ratio of dilepton yields in heavy - nuclei collisions ( as au+au or pb+pb ) and of that in the light systems ( as c+c ) for invariant masses @xmath198 gev will help to obtain information on the in - medium modification of the spectral function of vector mesons .
the authors are grateful for fruitful discussions with w. cassing , c. hartnack , s. endres , l. fabbietti , m. gumberidze , f. krizek , y. leifels , o. linnyk , a. rustamov , j. stroth , p. salabura , h. van hees , m. weber and j. weil .
our special thanks go to t. galatyuk and r. holzmann for the continuous help and many useful advices concerning the experimental data and filtering procedure .
e.b . and s.v .
acknowledge financial support through the
hic for fair " framework of the loewe " program .
z. fodor and c. hoelbling , rev .
* 84 * ( 2012)449 .
g. e. brown and m. rho , phys .
* 66 * ( 1991 ) 2720 .
g. e. brown , c. h. lee , m. rho and v. thorsson , nucl . phys . a * 567 * ( 1994 ) 937 .
t. hatsuda , s. h. lee and h. shiomi , phys .
c * 52 * ( 1995 ) 3364 .
m. asakawa , c. m. ko , p. lvai , x. j. qiu , phys .
rev . c*46 * ( 1992 ) r1159 .
r. rapp , g. chanfray and j. wambach , phys .
* 76 * ( 1996 ) 368 .
r. rapp , g. chanfray and j. wambach , nucl .
a * 617 * ( 1997 ) 472 .
w. peters , m. post , h. lenske , s. leupold , and u. mosel , nucl . phys .
a * 632 * ( 1998 ) 109 .
s. leupold , w. peters , u. mosel , nucl .
phys . a*628 * ( 1998 ) 311 .
h. van hees and r. rapp , j. phys .
g * 34 * ( 2007 ) s1051 . c. gale and s. turbide , nucl .
a * 783 * ( 2007 ) 35 .
t. renk and j. ruppert , arxiv : hep - ph/0605130 .
bratkovskaya , o. linnyk , and w. cassing , phys .
* b670 * ( 2009 ) 428 .
g. wolf , w. cassing and u. mosel , prog .
phys . * 30 * ( 1993 ) 273 ; nucl .
a * 552 * ( 1993 ) 549 ; nucl .
a * 545 * ( 1992 ) 139c .
e. l. bratkovskaya , w. cassing and u. mosel , phys .
b * 376 * ( 1996 ) 12 .
l. xiong , z. g. wu , c. m. ko and j. q. wu , nucl .
a * 512 * ( 1990 ) 772 . c. ernst , s. a. bass , m. belkacem , h. stoecker and w. greiner , phys .
c * 58 * ( 1998 ) 447 .
e. l. bratkovskaya , w. cassing , r. rapp , and j. wambach , nucl .
a*634 * ( 1998 ) 168 .
l. p. kaptari and b. kmpfer , nucl .
a * 764 * ( 2006 ) 338 .
m. schfer , t. s. biro , w. cassing and u. mosel , phys .
b * 221 * ( 1989 ) 1 .
r. shyam and u. mosel , phys .
c * 67 * ( 2003 ) 065202 .
f. de jong and u. mosel , phys .
b * 392 * ( 1997 ) 273 .
m. thomere , c. hartnack , g. wolf and j. aichelin , phys .
c * 75 * ( 2007 ) 064902 .
h. w. barz , b. kmpfer , g. .wolf and m. zetenyi , arxiv:0910.1541 [ nucl - th ] .
e. l. bratkovskaya and w. cassing , nucl .
a * 619 * ( 1997 ) 413 .
w. cassing and e. l. bratkovskaya , phys .
* 308 * ( 1999 ) 65 .
w. ehehalt and w. cassing , nucl .
a * 602 * ( 1996 ) 449 .
k. weber , b. blaettel , w. cassing , h. c. doenges , v. koch , a. lang and u. mosel , nucl .
a * 539 * ( 1992 ) 713 .
p. adlarson _ et al .
_ [ wasa - at - cosy collaboration ] , phys .
b * 706 * ( 2012 ) 256 .
l. alvarez - ruso , e. oset and e. hernandez , nucl .
a * 633 * ( 1998 ) 519 .
e. oset , j. a. gomez tejedor , f. cano , j. c. nacher , s. kamalov , l. alvarez - ruso and e. hernandez , few body syst .
suppl . * 11 * ( 1999 ) 275 .
x. cao , b. -s .
zou and h. -s .
xu , phys .
c * 81 * ( 2010 ) 065201 .
o. buss , t. gaitanos , k. gallmeister , h. van hees , m. kaskulov , o. lalakulich , a. b. larionov and t. leitner _
et al . _ ,
* 512 * ( 2012 ) 1 .
bass _ et al .
* 42 * ( 1998 ) 279 . c. hartnack , r. k. puri , j. aichelin , j. konopka , s. a. bass , h. stoecker and w. greiner , eur .
j. a * 1 * ( 1998 ) 151 . c. hartnack , h. oeschler , y. leifels , e. l. bratkovskaya and j. aichelin , phys .
rept . * 510 * ( 2012 ) 119 . j. w. harris , g. odyniec , h. g. pugh , l. s. schroeder , m. l. tincknell , w. rauch , r. stock and r. bock _ et al .
_ , phys .
* 58 * ( 1987 ) 463 .
w. reisdorf and y. leifels , private communication .
j. weil , h. van hees , and u. mosel , eur .
j. phys . * a 48 * ( 2012 ) 111 .
d. schumacher , s. vogel and m. bleicher , acta phys . hung . a * 27 * ( 2006 ) 451 .
k. schmidt , e. santini , s. vogel , c. sturm , m. bleicher , and h. stcker , phys .
c * 79 * ( 2009 ) 064908 . | we study the dilepton production in heavy - ion collisions at energies of 1 - 2 @xmath0gev as well as in proton induced @xmath1 and @xmath2 reactions from 1 gev up to 3.5 gev where data have been taken by the hades collaboration .
for the analysis we employ three different transport models - the microscopic off - shell hadron - string - dynamics ( hsd ) transport approach , the isospin quantum molecular dynamics ( iqmd ) approach as well as the ultra - relativistic quantum molecular dynamics ( urqmd ) approach . we find that the hsd and iqmd models describe very reasonably the elementary @xmath3 , @xmath4 and @xmath5 reactions despite of different assumptions on quantities like the excitation function of the @xmath6 multiplicity , where solid experimental constraints are not available .
taking these data on elementary collisions as input , the three models provide a good description of the presently available heavy ion data . in particular , we confirm the experimentally observed enhancement of the dilepton yield ( normalized to the multiplicity of neutral pions @xmath7 ) in heavy - ion collisions with respect to that measured in @xmath8 collisions .
we identify two contributions to this enhancement : a ) the @xmath9 bremsstrahlung which scales with the number of collisions and not with the number of participants , i.e. pions ; b ) the dilepton emission from intermediate @xmath6 s which are part of the reaction cycles @xmath10 and @xmath11 . with increasing system size more
generations of intermediate @xmath6 s are created .
if such @xmath6 decays into a pion , the pion can be reabsorbed , however , if it decays into a dilepton , the dilepton escapes from the system .
thus , experimentally one observes only one pion ( from the last produced @xmath6 ) whereas the dilepton yield accumulates the contributions from all @xmath6 s of the cycle .
we show as well that the fermi motion enhances the production of pions and dileptons in the same way .
furthermore , employing the off - shell hsd approach , we explore the influence of in - medium effects like the modification of self - energies and spectral functions of the vector mesons due to their interactions with the hadronic environment .
we find only a modest influence of the in - medium effects on the dilepton spectra in the invariant mass range where data with small error bars exist . |
the lisa gravity wave sensitivity goal requires that the residual acceleration of its test masses ( tms ) must be kept below 3 fm / s@xmath2/@xmath0 down to 0.1mhz @xcite .
this is obtained with a drag - free control , with the spacecraft that shields the tms from environmental disturbances and is centered about the tms geodesic motion according to a position sensor , also called `` gravity reference sensor '' ( grs ) .
the current design of the lisa grs foresees cubic 2 kg and ( 46 mm)@xmath3 tms , surrounded by an array of electrodes included in a capacitive readout and actuation scheme @xcite . in preparation for the flight - test of the lisa core elements , ltp @xcite ,
relevant upper limits on the most important grs related surface disturbances have been already set by means of torsion pendulums @xcite .
these experiments used a single lisa - like tm , hanging in axis from a torsion fiber and surrounded by one grs prototype : this configuration is extremely sensitive to torques arising in the grs but is insensitive to net forces and to any source acting along axes passing through the center of the tm .
upper limits on force noise sources have been thus set using model - dependent arm - lengths , to convert the torques into forces @xcite . to overcome this limit
, we are developing a new torsion pendulum facility , employing a four tm configuration @xcite , that allows for a direct measurement of force disturbances arising in the grs .
we present here results with a first prototype pendulum that allows an estimate of the intrinsic pendulum performance .
we discuss the characterization of the gravity gradient noise , one environmental noise source that could limit the apparatus performances , and finally report on the status of development of the facility .
on the left , a sketch of the experiment concept ( top view ) . on the right , a photograph of the facility : its main component , the vacuum chamber , is visible.,width=377 ] a sketch of the experiment concept is shown in fig.[facility : fig ] , with a photograph of the facility .
four lisa - like tms lay on the horizontal plane , are mounted on a cross - shaped support , and are suspended off - axis with respect to the pendulum torsion fiber , at distance @xmath4 bigger than the tm side - length .
one of the tms is included in a grs prototype : any net force arising inside the grs gets thus converted into a torque on the pendulum by the arm - length @xmath4 .
this pendulum has full sensitivity to disturbances arising inside the sensor , independently from their location , and the unique conversion arm - length @xmath4 allows to set model - independent force noise upper - limits .
use of hollow tms reduces the pendulum weight and allows for thinner torsion fibers .
this makes the pendulum mainly sensitive to surface forces ( however the most relevant envisioned by noise analyses ) but increases its force sensitivity , that can be further enhanced by enlarging the pendulum arms , in an optimal compromise with low weight and coupling to gravity gradients ( see below ) .
the direct measurement of grs relevant parameters like force gradients ( `` stiffness '' ) or stray electrostatic forces will provide then fully representative estimates for lisa . the main component of the facility is a 250 @xmath5 vacuum vessel ( see fig.[facility : fig ] ) , accomodating the pendulum and the grs , that sits on a concrete slab partially isolated from the laboratory floor .
a 90 cm long tube encloses the torsion fiber . a 400 @xmath6 turbo pump sets the residual pressure at @xmath7 mbar .
the entire apparatus is enclosed in a thermally insulated box whose temperature is controlled by a thermalized water bath stabilizing the air circulating inside a heat exchanger .
the experiment is equipped with manual micro - manipulators ( 4 dof for grs alignment + 2 dof for pendulum rotational / vertical alignment ) and motorized stages ( 2 translators for grs alignment and automatized stiffness measurements ) .
the pendulum motion is monitored by a commercial autocollimator , with 50 nrad resolution for both twist and tilt modes .
the pendulum tilt motion is suppressed with a magnetic eddy current damper ( decay time @xmath8 s ) .
finally , the facility is equipped with different environmental monitors , like thermometers for in - vacuum and in - air temperatures , two 3-axis flux - gate magnetometers for magnetic fields and field gradients and a 2-axis tilt - meter for the residual tilt motion of the apparatus .
left : a photograph of the first pendulum prototype , in the vacuum vessel : 1 .
cross - shaped support 3 .
capacitive sensors 5 . sensing electrodes 6 .
copper electrostatic shields .
right : typical force noise sensitivity of the apparatus , compared with lisa and ltp residual force noise goals.,height=207 ] as the main grs - related noise sources are expected to depend strongly on the tm - electrode separation ( gap ) , an estimate of the intrinsic pendulum noise floor has been performed with a prototype pendulum integrated with very large - gap sensors ( allowing us also to debug the facility and verify the fiber performances ) .
we employed ( 3 cm)@xmath3 solid uncoated al tms ( @xmath9 g ) , mounted on an al cross shaped support and centered at a distance of @xmath1089 mm from the fiber axis ( see fig.[facility : fig ] ) . a au - coated mirror was mounted along the fiber axis .
the total pendulum weight was @xmath11 g , roughly the same of the next pendulum , with ( 46 mm)@xmath3 tms .
two single axis capacitive sensors surrounded two of the tms ( electrically isolated ) at very large gaps ( 8 mm along the force sensitive axis ) , providing both a high sensitivity ( 200 nrad /@xmath0 ) readout of the pendulum twist motion and actuation authority .
we used a 95 cm long , 50 @xmath12 m diameter w fiber , with a measured mechanical quality factor @xmath13 . with the pendulum inertia
@xmath14 g m@xmath15 and a torsion constant @xmath16 nnm / rad , the torsion pendulum resonance was @xmath17 mhz .
a typical force sensitivity performance of the apparatus , obtained by dividing the measured pendulum torque noise by the arm - length @xmath18 mm , is shown in fig.[noise : fig ] .
below 1 mhz , the observed noise is within a factor 2 from the instrumental limit , set here by the pendulum mechanical thermal noise . above 1 mhz ,
the sensitivity is limited by the angular read - out noise . the performance shown in fig.[noise : fig ] would allow to place upper limit on the disturbances related to a grs prototype at the level of 800 fn/@xmath0 between 0.2 and few mhz , with a best sensitivity of 300 fn/@xmath0 at 1 mhz , a factor 50 from the lisa goal . the excess noise at low frequencies , that could be given by environmental temperature fluctuations as well as tilt motion of the apparatus , is still being investigated .
a possible drawback of this four mass pendulum configuration is its sensitivity to gravity gradients @xcite .
the gravity gradient torque induced on a pendulum by objects at a distance @xmath19 goes like @xmath20 , with @xmath4 and @xmath5 , respectively , the characteristic radius and degree of symmetry of the pendulum mass distribution about the torsion axis .
this pendulum symmetry is nominally high , with @xmath21 .
however , lower order terms , particularly quadrupole , could enter through machining and assembling imperfections , producing mass distribution asymmetries and tilting of the pendulum : this enhances the pendulum sensitivity to gravity gradients and avoids reaching its ultimate performances . in order to qualitatively characterize the role of gravity gradient noise of our experimental site
, we performed a test in which we purposely enhanced the pendulum quadrupole moment ( @xmath22 ) to detect any induced excess noise on the pendulum .
two of the four masses where moved towards the center of the pendulum by @xmath23 mm , as shown in fig.[noise : fig : nightday ] .
if the previous configuration had a `` nominally '' nulled quadrupole moment , i.e. @xmath24 , this produced a ( calculated ) mass quadrupole moment of the pendulum of order @xmath25 g m@xmath15 ( for comparison , a 1 cm displacement would produce @xmath26 g m@xmath15 , and a 2 mm displacement @xmath27 g m@xmath15 ) . left : scheme of the enhancement of the pendulum mass quadrupole .
right : comparison of the typical night and day force sensitivities of the unbalanced quadrupole pendulum configuration.,height=207 ] measured force noise spectra are shown in fig.[noise : fig : nightday ] . when human activity is present close to the experiment , in particular due to people working at few m from the apparatus , the observed force noise is about a factor 50 higher than the instrumental limit . however ,
when human activity inside and near the lab is reduced ( night and week - ends ) we did not observe any excess noise compared to the typical performance of the pendulum , as in fig.[noise : fig ] . with reasonable tolerances of the final ( 46 mm)@xmath3 tm assembly of order 100 @xmath12 m ( or better ) , the mass quadrupole moment of the pendulum
will be reduced at least by a factor 300 , compared to the one in fig.[noise : fig : nightday ] .
this would make gravity gradient noise to be substantially negligible for the pendulum performance . even if this test provides only a partial picture of the gravity gradient noise of our experimental site (
this is only one of the possible pendulum assembling imperfections , as for instance tilting with respect to the vertical axis , for which we did not perform a specific investigation ) , this test shows that quadrupole gravity gradient noise should not represent a relevant limitation for reaching the ultimate pendulum performance in terms of force sensitivity .
schematic of the new pendulum , currently under construction : ( a ) a 3d cartoon of the final assembly ; ( b ) top view schematic ; ( c ) a sketch of the `` stiffness compensator '' ; ( d ) one of the already delivered 46 mm tms ( the supporting shaft is visible ) ; ( e ) a replica of the ltp grs engineering model.,height=207 ] we are now working on the integration of a fully representative grs prototype to be tested in the facility .
a new pendulum , shown in fig.[newpend : fig ] , is currently under construction , employing four al ( @xmath28mm)@xmath3 hollow tms .
one tm will be included into a grs prototype that implements the ltp design .
an additional capacitive sensor , implemented with very large gaps and needed to compensate the tilt - twist coupling induced on the pendulum by the grs - related stiffness , will surround the opposite tm .
this `` stiffness compensator '' will also be the support for an optical read - out proposed for lisa by our infn collaborators @xcite .
these two tms will be electrically isolated , for the capacitive readout and actuation . for electrostatic homogeneity
, the remaining two tms will be electrically grounded to the rest of the pendulum and then to the fiber .
similarly , the whole pendulum will be au - coated , included tms and cross shaped support .
this facility will allow to test grs prototypes for ltp and lisa with full - size tms and highly representative conditions of the flight mission . by direct measurement of forces instead of torques
, it will allow stringent upper limits on lisa force noise , which will be independent of the model of the noise source and its location within the grs .
furthermore , the role of the sources that produce force noise on the lisa tms will be more straightforward to evaluate and model , leading to a more accurate verification of the lisa noise model .
99 bender p _
et al . _ , lisa esa - sci(2000)11 , 2000 weber w j _
et al._2002 _ spie proc . _ * 4856 * 31 dolesi r _ et al._2003 class .
grav.*20 * s99 anza s _ et al._2005 class .
quant . grav.*22
carbone l _
et al._2003 phys .
rev . lett.*91 * 151101 carbone l
_ et al._2005 class .
quant . grav.*22
* s509 carbone l
_ et al._2004 class .
quant . grav.*21
carbone l _
et al._proc . of 10@xmath29 marcel grossmann meeting on general relativity , gr - qc/0411049 y. su _
et al._1994 phys . rev .
* d * * 50 * , 3614 di fiore l. _ et al .
_ , this conference proceedings | a four mass torsion pendulum facility for testing of the lisa grs is under development in trento . with a lisa - like test mass suspended off - axis with respect to the pendulum fiber ,
the facility allows for a direct measurement of surface force disturbances arising in the grs .
we present here results with a prototype pendulum integrated with very large - gap sensors , which allows an estimate of the intrinsic pendulum noise floor in the absence of sensor related force noise .
the apparatus has shown a torque noise near to its mechanical thermal noise limit , and would allow to place upper limits on grs related disturbances with a best sensitivity of 300 fn/@xmath0 at 1mhz , a factor 50 from the lisa goal .
also , we discuss the characterization of the gravity gradient noise , one environmental noise source that could limit the apparatus performances , and report on the status of development of the facility .
address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy , [email protected] address = centro fisica degli stati aggregati , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy address = dipartimento di fisica , universit di trento , and i.n.f.n . , gruppo di trento , povo ( tn ) , italy submitted to _ proc . of the @xmath1 int .
lisa symposium _ ,
aip conference proceedings , 2006 |
top quark pair production in @xmath2 collisions is a sensitive probe of quantum chromodynamics at high energy . at lowest order in the standard model ( sm ) , quark pair production is symmetric under charge conjugation . at next - to - leading order ( nlo )
the interference of processes that differ under charge conjugation leads to a small forward - backward asymmetry of order @xmath11 in the @xmath0 rest frame @xcite .
an analogous effect is predicted at order @xmath12 in qed and is confirmed in measurements of @xmath13 @xcite .
study of the nlo qcd asymmetry in inclusive jet events is hampered by the difficulty of measuring the jet charge . in pair produced top quarks with one semi - leptonic decay , the top can be tagged according to the well - measured lepton charge , enabling a probe of the nlo qcd effect and a test of charge conjugation symmetry in strong interactions at high energy .
the cdf and d0 experiments have made initial measurements of the @xmath0 asymmetry in @xmath2 collisions at @xmath3 tev at the fermilab tevatron @xcite . in the @xmath14-invariant @xmath2 system
, the nlo qcd effect appears as a charge dependent forward - backward asymmetry of the top quark direction with respect to the proton direction . using data samples corresponding to @xmath15 and @xmath16 respectively , cdf and d0 report positive asymmetries that are consistent with the qcd prediction within large experimental uncertainties .
a number of theoretical papers suggest interesting new physics mechanisms including axigluons , diquarks , new weak bosons , and extra - dimensions that can all produce forward - backward @xmath0 asymmetries @xcite .
the model building must accommodate the observed consistency of the @xmath0 cross - section and total invariant mass distribution with the sm qcd prediction @xcite .
we report here on a new study of the forward - backward asymmetry in @xmath2 collisions at @xmath3 tev , using data corresponding to an integrated luminosity of @xmath17 recorded with the cdf ii detector .
we study events with the lepton+jets topology , where either the @xmath4 or @xmath5 has decayed semileptonically .
the asymmetries are measured in two variables : @xmath18 , the rapidity of the hadronically decaying top quark , corresponding to the top rapidity in the laboratory ( lab ) frame , and @xmath6 , the difference of the rapidities of the leptonic and hadronic top systems , which measures the top quark rapidity in the @xmath0 rest frame .
we show that @xmath4 and @xmath5 asymmetries are consistent with @xmath14 conservation , and combine them to measure the total asymmetry in the sample .
we measure the inclusive asymmetries , and the functional dependence of the @xmath0 frame asymmetry on @xmath6 and on the total invariant mass of the @xmath0 system , @xmath7 .
we apply corrections for backgrounds , acceptance , and resolution to calculate parton level measures of the inclusive asymmetry in both the lab and @xmath0 rest frames , and in two regions of @xmath6 and @xmath7 in the @xmath0 frame .
cdf ii is a general purpose , azimuthally and forward - backward symmetric magnetic spectrometer with calorimetry and muon detectors @xcite .
charged particle trajectories are measured with a silicon - microstrip detector backed by a large open - cell drift chamber in a 1.4 t solenoidal magnetic field .
electromagnetic and hadronic calorimeters located outside the solenoid provide jet and missing energy reconstruction . outside the calorimeter are multilayer proportional chambers and plastic scintillator hodoscopes that provide muon identification in the pseudorapidity region @xmath19 .
we use a cylindrical coordinate system with origin at the detector center and z - axis along the proton direction @xcite .
this measurement uses @xmath20 candidate events in the `` lepton+jets '' topology , where one top quark decays semileptonically ( @xmath21 ) and the other hadronically ( @xmath22 ) @xcite .
we detect the lepton and four jets from top quark decays and quark hadronization , and an inferred neutrino based on the presence of missing energy .
the detector is triggered by a high transverse momentum electron(muon ) in the central portion of the detector with @xmath23 gev(gev / c ) and @xmath24 .
we require four or more hadronic jets with @xmath25 gev and @xmath26 , and a large amount of missing transverse energy , @xmath27 gev , consistent with the presence of an undetected neutrino .
the jets are reconstructed using a cone algorithm with @xmath28 , and calorimeter signals are corrected for detector inefficiencies and for the energy scale factor .
the secvtx algorithm @xcite is used to find displaced @xmath29-decay vertices using the tracks within the jet cones , and at least one jet must contain such a `` @xmath29-tag '' .
jets with @xmath29-tags are restricted to @xmath24 .
a total of 1260 events pass this selection .
the size of the non-@xmath0 background processes in the lepton+jets+@xmath29-tag selection is derived in precision measurements of the @xmath0 production cross - section @xcite . the estimated background in our sample @xmath30 events .
the predominant backgrounds are from qcd - induced w+multi - parton events containing either @xmath29-tagged heavy - flavor jets or errantly tagged light - flavor jets .
these are modeled using a simulation sample derived from the alpgen generator @xcite and a data driven technique that derives tagging efficiencies , mis - tagging rates and sample normalizations from direct measurement . a background component from qcd multi - jet events with fake leptons and mis - measured @xmath31
is modeled using multi - jet events with lepton candidates that are rejected by our cuts .
other small backgrounds from electroweak processes ( @xmath32 , single - top ) are reliably estimated using monte carlo generators .
further details on the sample selection and background modeling can be found in ref .
@xcite .
the reconstruction ( reco ) of the @xmath0 kinematics uses the measured momenta of the lepton , the @xmath31 , and the four leading jets in the event .
the jet - parton assignment and calculation of the @xmath0 four - vectors uses a simple @xmath33-based fit of the lepton and jet kinematics to the @xmath0 hypothesis , allowing the jet energies to float within their expected uncertainties , and applying the constraints that @xmath34 , @xmath35 , and @xmath29-tagged jets are associated with b - partons .
this algorithm is well understood in the context of precision top mass measurements , where the fit is performed without the top mass constraint @xcite , and other top physics studies that use the top mass constraint @xcite .
we study the reconstructed top quark rapidity and the difference in the reconstructed top and anti - top rapidities , from which we derive the forward - backward asymmetries in the @xmath2 ( laboratory ) rest frame and in the @xmath0 rest frame .
the validity of the analysis is checked at all steps by comparison to a standard prediction made using the pythia @xcite @xmath0 model , the cdf lepton+jets+@xmath29-tag background model , and a full simulation of the cdf - ii detector .
we use pythia 6.2.16 with cteq5l parton distribution functions @xcite and @xmath36 .
the background model developed in concert with the precision cross - section studies provides good measures of both the normalizations and shapes of the non-@xmath0 processes @xcite .
the veracity of the combined pythia plus background model , and in particular , its reliability for the estimation of systematic uncertainties , is well verified in many other top - physics studies at cdf @xcite . note
that because pythia does not include the nlo qcd charge asymmetry , the standard pythia prediction is not the sm prediction for the forward - backward asymmetry .
studies with the mc@nlo generator @xcite ( see sec . [
sec : mcnlo ] ) predict that the magnitude of reconstructed qcd asymmetry in our sample is smaller than the current experimental resolution .
symmetric pythia is thus a good approximation for sm @xmath0 and provides an unbiased control sample for many of our studies .
we will compare our measurements to the sm predictions of mc@nlo when appropriate .
in the lepton+jets decay topology of the @xmath0 pair , there is a leptonic decay , @xmath37 , and a hadronic decay @xmath38 . the complications of the central lepton acceptance and the reconstruction of the neutrino from the @xmath31 create a difference in the reconstruction resolution for the two different kinds of decay . in order to control effects of this kind
, our treatment of top rapidity variables maintains the distinction between the leptonic and hadronic decay systems , with the @xmath4 and @xmath5 assignments following in accordance with the lepton charge .
the most direct measurement of the top direction with respect to the beamline is the rapidity of the hadronic top system in the lab frame , @xmath18 , which has acceptance out to @xmath39 and good directional precision . in events with a negative ( positive ) lepton
, @xmath18 is the lab rapidity of the @xmath4 quark , @xmath40 ( @xmath41 quark , @xmath42 ) .
if @xmath14 is a good symmetry , the distributions of @xmath42 and @xmath40 are reflections of each other , and we can combine both samples , weighting with the lepton charge , to use @xmath43 as the rapidity of the @xmath4 quark in the lab frame , @xmath40 .
a frame independent measurement is available in the rapidity difference of the leptonic and hadronic systems @xmath44 .
after multiplication by the lepton charge @xmath45 , this variable measures the difference between the top and antitop rapidities : @xmath46 . the rapidity difference
@xmath6 is independent of the @xmath0 system longitudinal motion and is simply related to the top quark rapidity in the @xmath0 rest frame : @xmath47 . since the rapidity preserves the sign of the production angle , an asymmetry in @xmath6 is identical to the asymmetry in the top quark production angle in the @xmath0 rest frame .
since @xmath18 and @xmath50 are identified with either a @xmath4 or an @xmath5 by the sign of the lepton in the event , they are the primary variables for defining the charge dependence of the asymmetries and testing for @xmath14 invariance .
we define the forward - backward charge asymmetry in the @xmath0 rest frame to be : since these models have relatively low - lying octet masses near 2 tev/@xmath54 we expect a significant @xmath7-dependent asymmetry over our experimental range . the @xmath55 behavior for the two color - octet samples is shown in fig . [ fig : dadm_axig ] .
both show a smooth and significant rise of the asymmetry with increasing mass . in sec .
[ sec : hilowmass ] we discussed a simple representation of @xmath56 with two regions of low and high @xmath7 .
the question for that representation is how to choose the boundary mass between high and low .
table [ tab : avm_high_axi ] shows the asymmetry , uncertainty , and significance @xmath57 at high mass as a function of the mass threshold for both octet models .
the uncertainties are calculated assuming the data sample size of @xmath17 . for both samples , the significance of the asymmetry at high mass is maximum at reconstructed @xmath58 .
looking at fig .
[ fig : mttb ] , we see that this is reasonable : @xmath59 cuts off the bulk of the low mass peak while retaining good statistics on the tail .
[ fig : avm2bins_octet ] compares the @xmath6 distributions in the octeta sample and pythia when the events are divided into samples below and above @xmath58 .
the @xmath6 distribution is much broader at high mass , as expected , and shows a marked asymmetry .
the reconstructed asymmetries at high and low mass in the color octet samples are given in table [ tab : ahilowcharges_axig ] .
the uncertainties here reflect the monte carlo statistics . at high mass
the color octet samples have large asymmetries as seen in fig . [
fig : avm2bins_octet ] . at low mass ,
the models have small but significant asymmetries , especially in octetb .
we have checked that these asymmetries are charge asymmetries , reversing sign under interchange of lepton charge .
d. w. jung , p. ko , j. s. lee , and s. h. nam , phys .
b * 691 * , 238 ( 2010 ) ; e. alvarez , l. darold , and a. szynknam , arxiv:1011.6557 ; d. w. jung , p. ko , and j. s. lee , arxiv:1011.5976 ; c. h. chen , g. cvetic , and c. s. kim , arxiv:1009.5976 ; y. k. wong , b. xiao , and s. h. zhu , arxiv:1008.2685 ; m. bauer , f. goertz , u. haisch , t. pfoh , and s. westhoff , j. high energy phys . 11 ( 2010 ) 039 ; r. s. chivukula , e. h. simmons , and c. p. yuan , phys . rev .
d * 82 * , 094009 ( 2010 ; b. xiao , y. k. wang , and s. h. zhu , phys .
d * 82 * , 034026 ( 2010 ; ) q .- h .
cao , d. mckeen , j. rosner , g. shaughnessy , and c. wagner , phys .
d * 81 * , 114004 ( 2010 ) ; i. dorsner , s. fajfer , j. f. kamenik , and n. kosnik , phys .
d * 81 * , 055009 ( 2010 ) ; s. h. jung , h. muryama , a. pierce , and j d. wells , phys .
d * 81 * , 015004 ( 2010 ) ; k. cheung , w .- y .
keung , and t .- c .
yuan , phys .
b * 682 * , 287 ( 2009 ) ; p. h. frampton , j. shu , and k. wang , phys .
b * 683 * , 294 ( 2010 ) ; j. shu , t. m. p. tait , and k. wang , phys .
d * 81 * , 034012 ( 2010 ) ; a. arhrib , r. benbrik , and c .- h .
chen , ( 2009 ) , phys .
d * 82 * , 034034 ( 2010 ) ; d .- w .
jung , p. ko , j. s. lee , and s .- h .
nam , ( 2009 ) , arxiv:0912.1105 ; j. cao , z. heng , l. wu , and j. m. yang , ( 2009 ) , phys .
d * 81 * , 014016 ( 2010 ) ; v. barger , w .- y .
keung , and c .- t .
yu , ( 2010 ) , phys .
d * 81 * , 113009 ( 2010 ) .
p. ferrario , g. rodrigo , phys . rev .
d * 78 * , 094018 ( 2008 ) .
the polar angle is @xmath60 and the azimuthal angle is @xmath61 . with total energy @xmath62 and momentum @xmath63 ,
the transverse energy is defined as @xmath64 and the transverse momentum is @xmath65 .
the missing transverse energy ( @xmath31 ) is defined by @xmath66 where @xmath67 is a unit vector perpendicular to the beam axis and pointing to the @xmath68 calorimeter tower .
the pseudorapidity is @xmath69 . | we present a new measurement of the inclusive forward - backward @xmath0 production asymmetry and its rapidity and mass dependence .
the measurements are performed with data corresponding to an integrated luminosity of @xmath1 of @xmath2 collisions at @xmath3 tev , recorded with the cdf ii detector at the fermilab tevatron .
significant inclusive asymmetries are observed in both the laboratory frame and the @xmath0 rest frame , and in both cases are found to be consistent with cp conservation under interchange of @xmath4 and @xmath5 . in the @xmath0 rest frame ,
the asymmetry is observed to increase with the @xmath0 rapidity difference , @xmath6 , and with the invariant mass @xmath7 of the @xmath0 system .
fully corrected parton - level asymmetries are derived in two regions of each variable , and the asymmetry is found to be most significant at large @xmath6 and @xmath7 . for @xmath8 , the parton - level asymmetry in the @xmath0 rest frame is @xmath9 compared to a next - to - leading order qcd prediction of @xmath10 . |
spin - orbit ( so ) coupled two - dimensional electron systems are of great interest , both as model systems and as the active component of devices that control electron spin with electric fields.@xcite unfortunately , the potential of the so interaction to control electron spin comes with a price - the so terms in the hamiltonian break su(2 ) spin symmetry .
the violation of su(2 ) means that electron spin polarization is not conserved , decaying instead with a characteristic spin memory time @xmath3 .
the mechanism by which so coupling leads to spin memory loss has been intensively investigated in two - dimensional electron gases ( 2degs ) in semiconductor quantum wells ( qws ) , as described in recent reviews.@xcite in gaas qws and related systems , breaking of inversion symmetry allows so coupling that is linear in the electron wave vector @xmath4.@xcite the so terms in the hamiltonian can be viewed as effective magnetic fields that act only on the electron spin , with magnitude and direction that vary with @xmath4 . the loss of spin memory in the effective magnetic field , @xmath5 , takes place through the dyakonov - perel ( dp ) mechanism.@xcite in this process the electron spin precesses during its ballistic motion between collisions ; each time it is scattered @xmath5 and consequently the precession vector , @xmath6 , change .
the net result is exponential decay of spin polarization at a rate approximately equal to @xmath7 , where @xmath8 is the mean time between collisions .
there exist two distinct contributions to @xmath5 , the rashba term@xcite arising from asymmetry of the confining potential and the dresselhaus term@xcite originating in the intrinsic inversion asymmetry of the gaas crystal structure . a prescription for lengthening spin lifetime in qws of iii - v semiconductors by tuning the rashba coupling strength ( @xmath9 ) to equal the linear dresselhaus coupling ( @xmath10 )
was proposed by schliemann _ _
et al.__@xcite recently it was recognized that this mechanism amounts to a restoration of su(2 ) symmetry even in the presence of anisotropic so interactions.@xcite the main purpose of this paper is to assess theoretically to what extent tuning so interactions can be expected to increase the distance over which electron spin polarization can propagate without decay . the potential to extend the spin propagation length despite dp spin memory decay is based on the strong correlation between the electron s displacement in space and the rotation of its spin on the bloch sphere .
an important step toward a quantitative theory of such correlations was made by burkov _
_ et al.__@xcite and mishchenko _ et al._,@xcite who derived equations of motion that describe the coupling of spin and charge current degrees of freedom in ( 001 ) gaas qws .
initially only the linear rashba so coupling was examined , subsequently bernevig _ _ et al.__@xcite and stanescu and galitski @xcite extended the theory to include the linear and cubic dresselhaus terms , respectively .
the equations of motion can be solved to obtain the normal modes of the coupled system , which are waves of mixed electrical current and spin polarization . there
exist four such modes , reflecting three spin degrees of freedom ( @xmath11 , and @xmath12 ) and the charge density , @xmath13 .
for wave vectors , @xmath14 , parallel to the directions @xmath15 $ ] and @xmath16 $ ] , the four modes decouple into two pairs ; in one the spin precesses in a plane containing @xmath14 and the normal direction @xmath17 , in the other the current is coupled to the component of in - plane spin polarization perpendicular to @xmath14 .
the spin precession mode is the one relevant to spin polarization memory .
for example , the decay rate of this mode at @xmath18 is precisely the dp decay rate , @xmath19 . in the absence of spin - space correlation
, the decay rate , @xmath20 , of a spin polarization wave would increase monotonically with @xmath21 , _
i.e. _ , @xmath22 , where @xmath23 is the spin diffusion coefficient . instead
, it was predicted @xcite that for rashba so coupling the minimum decay rate occurs at nonzero wave vector , at which point @xmath20 is approximately half the dp rate .
_ et al.__@xcite showed theoretically that the minimum @xmath20 is further reduced when both rashba and linear dresselhaus interactions are nonzero and vanishes when the strength of the two couplings is equal . the resulting
persistent spin helix " ( psh ) was shown to be a conserved quantity of a newly found su(2 ) symmetry that arises when @xmath24 and the cubic dresselhaus term ( @xmath25 ) is zero.@xcite however , stanescu and galitski @xcite showed that perfect su(2 ) is broken when @xmath26 , leading to large , but not infinite , psh lifetime . recently , using the transient spin grating technique , koralek _ _ et al.__@xcite observed the psh mode experimentally by independently tuning the rashba and linear dresselhaus couplings .
the question that arises is whether the psh effect can be exploited to lengthen the distance that a packet of spin polarization can propagate in an applied electric field . in this paper
we address this question by analyzing the effects of an in - plane electric ( @xmath27 ) field on the spin - precession modes .
we focus on @xmath28 , which is the orientation relevant to the drift of spin polarization .
to predict the spin memory length it is necessary to determine how the applied field modifies both the real ( @xmath29 ) and imaginary ( @xmath30 ) parts of the normal - mode frequency , @xmath31 of spin - polarization modes .
the real part is related to the drift velocity whereas the imaginary part is related to the lifetime .
the modification of @xmath32 is linear in @xmath27 ( to lowest order ) , whereas the affect of @xmath27 on @xmath33 is quadratic .
kleinert and bryksin @xcite recently have treated this to problem to linear order in @xmath27 , using quantum kinetic theory , and obtained results for @xmath32 . in this work ,
we derive and solve equations of motion to quadratic order in @xmath27 using a random walk ( rw ) approach that is different from previous treatments of this problem .
the advantages of our approach are physical transparency and mathematical simplicity .
we construct a semiclassical random walk model that tracks the electron s motion in real space and the propagation of its spin on the bloch sphere . in sec .
[ sec : rw ] , we introduce the random walk model , derive the equations of motion in the absence of an @xmath27 field , and solve for the spin - wave dispersion relations . we compare the results thus obtained with the earlier quantum kinetic theory approaches.@xcite in sec .
[ sec : e ] , we include an in - plane @xmath27 field , obtaining the equations of motion and the dispersion relations to quadratic order .
we use the dispersion relations to analyze the motion of a spin - polarization packet in the presence of the in - plane field , for different regimes of field strength .
we illustrate the results by focusing on representative so couplings : linear dresselhaus coupling only , the su(2 ) case where rashba and dresselhaus terms are equal , and the case of su(2 ) broken by a small cubic dresselhaus term .
a brief summary is given in sec .
[ summary and conclusion ] .
as mentioned above , as an electron propagates between scattering events , so coupling causes its spin to precess .
thus , as the electron performs an rw in real space , its spin performs an rw on the bloch sphere .
we consider a 2d electron gas with both structure and bulk inversion asymmetry .
the so hamiltonian for conduction band electrons in a iii - v semiconductor qw grown in the [ 001 ] direction ( taken as @xmath17-direction ) is given by , @xmath34 where , @xmath35v_y -{\boldsymbol}{\hat{y}}\left[\alpha+\beta_1-\frac{2\beta_3(v_x^2-v_y^2)}{v_f^2}\right]v_x\right\},\ ] ] @xmath36 is the electron spin , @xmath37 and @xmath38 are the components of velocity in the [ 110 ] and [ 1@xmath390 ] directions , @xmath9 , @xmath10 , and @xmath25 are dimensionless quantities describing the strength of the rashba , linear , and cubic dresselhaus spin - orbit couplings , respectively , and @xmath40 is the fermi wave vector .
spins precess about the effective so field according to @xmath41 we assume that the impurity potential is short range so that there is no correlation between the scattering events . in the absence of the @xmath27 field , electrons perform an isotropic 2d random walk with @xmath42 ( velocity between the @xmath13th and @xmath43th scattering events ) given by @xmath44 , where @xmath45 is a random two - dimensional unit vector with a uniform probability density @xmath46 . the displacement from @xmath13th to @xmath43th step
is given by @xmath47 where @xmath8 is the electron scattering time .
in the following we consider @xmath48 , the change in angle of the electron s spin between scattering events , as a small parameter . in this case
we can obtain from eq .
( [ times ] ) the change in the spin direction during the mean - free time as a series expansion in @xmath48 , @xmath49where we retain terms to second order .
let @xmath50 be the probability that after @xmath13 steps of random walk the electron arrives at position @xmath51 and @xmath52 be the conditional probability that given the electron is at @xmath51 , its spin is @xmath53 .
the joint probability @xmath54 satisfies the following recursion relation : @xmath55 where @xmath56 denotes average over @xmath57 , i.e. , @xmath58 .
once @xmath54 is determined , the magnetization can be obtained from the following integral on the bloch sphere : @xmath59 by substituting eq .
( [ recursion ] ) into eq .
( [ magnetization ] ) , we obtain , @xmath60 taylor series expansion on the right hand side of eq .
( [ m ] ) yields , @xmath61 \{p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})\\ & -{\boldsymbol}{v}_n\tau\cdot\nabla[p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s } ) ] + \frac{1}{2}{\boldsymbol}{v}_n\tau\cdot\nabla\nabla[p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})]\cdot{\boldsymbol}{v}_n\tau\}d\sigma\rangle .
\end{split}\ ] ] again retaining terms to second order , we can write , @xmath62where , @xmath63\\ & + \frac{1}{2}{\boldsymbol}{v}_n\tau\cdot\nabla\nabla[p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})]\cdot{\boldsymbol}{v}_n\tau\}d\sigma\rangle , \end{split}\ ] ] @xmath64 \{p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})-{\boldsymbol}{v}_n\tau\cdot\nabla[p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})]\}d\sigma\rangle,\]]and @xmath65 \{p_n({\boldsymbol}{r})d_{n}({\boldsymbol}{r};{\boldsymbol}{s})\}d\sigma\rangle.\ ] ] upon performing the average over @xmath57 , all terms that linear in @xmath42 or @xmath66 vanish by symmetry , leading to , @xmath67 @xmath68 @xmath69 where @xmath70 taking the continuum limit @xmath71 , @xmath72 , and substituting into eq .
( [ eq : i ] ) , we obtain the equation of motion for the magnetization vector .
resolving the vector equation into components yields three scalar equations , @xmath73
@xmath74 @xmath75 solving the equations of motion for eigenmodes with wave vector parallel to @xmath76 yields the dispersion relation , @xmath77 this dispersion relation corresponds to modes in which the spin polarization spirals in the @xmath78-@xmath79 plane .
note that @xmath31 is purely imaginary so that for all wave vectors the spin - polarization wave decays exponentially with time .
however , the dispersion relation differs from ordinary diffusion , where @xmath80 .
the difference can be traced to the terms in eq .
( [ eq : i2 ] ) that are proportional to the first derivative of spin density with respect to position - these terms are absent in the usual diffusion equation .
the coefficients of these additional terms are the cross - correlation functions , @xmath81 and @xmath82 , which shows explicitly that the anomalous diffusion is a consequence of the correlation between the electron s motion in real space and the propagation of its spin on the bloch sphere . in the su(2 )
case ( @xmath24 and @xmath83 ) , eq . ( [ dispersion1 ] )
simplifies to , @xmath84 where @xmath85 and @xmath86 .
the vanishing decay rate of the @xmath87 mode at @xmath88 indicates the appearance of a conserved quantity - a helical spin - polarization wave or
persistent spin helix".@xcite the dispersion relations obtained above for the spiral polarization waves are the same as those obtained previously , including the cubic dresselhaus term.@xcite we note , however , that while the rw approach accurately describes the spiral coupling of @xmath78-@xmath79 components of spin , it does not capture the coupling between charge current and the @xmath89 component of spin that appears in the quantum kinetic formulation .
this is because the rw approach does not include relaxation to the equilibrium state . in other words , between consecutive scattering events
the electron s spin precesses about @xmath5 but has no tendency to spiral in toward it .
thus the well - known current - induced spin polarization ( cisp ) effect @xcite is not predicted .
to recover cisp requires adding to eq .
( [ times ] ) a phenomenological gilbert damping term , @xmath90where @xmath91 is the damping parameter .
in this section , we explore how the spin dynamics change in the presence of an @xmath27 field parallel to the wave vector of the spin spiral . to include the effect of @xmath27 we add a drift term to the velocity at each random walk step , @xmath92 where @xmath93 is the drift velocity assumed to be a linear function of @xmath27 .
we assume further that the electric field does not change the shape of the impurity potential and therefore the scattering probability density is still uniform .
the drift velocity modifies the precession vector , adding a fixed precession @xmath94v_d,\ ] ] to @xmath66 at each step of the random walk .
substituting and following the same strategy as before , we obtain , @xmath95 @xmath96 @xmath97 where the @xmath98 are the quantities @xmath99 evaluated in the presence of the electric field .
the field alters the equations of motion in two ways .
first , new terms appear that are linear in @xmath27 .
the new term added to @xmath100 converts the time derivative of @xmath101 to the convective derivative , that is the time derivative in a frame moving with the drifting electrons .
the term added to @xmath102 indicates that the @xmath27 field introduces uniform precession about the @xmath103 axis , when viewed in the frame co - moving with @xmath104 .
the second type of modification is quadratic in @xmath27 ; the field increases @xmath105 by the additive factor @xmath106 and the mean - square velocity @xmath107 by the factor @xmath108 . solving for normal modes with wave vector parallel to @xmath76
, we obtain @xmath109 to linear order in @xmath27 , this dispersion relation is the same as that obtained by kleinert and bryksin.@xcite in the presence of the electric field @xmath31 acquires a real part , which describes the propagation of spin polarization . equation ( [ io0 ] ) also describes the modifications of the spin polarization lifetime that appear at second order in @xmath27 . in the following we discuss the spin dynamics that emerge from this dispersion relation for representative so hamiltonians .
for the case of @xmath110 , the dispersion relation simplifies to , @xmath111 where @xmath112 . to distinguish the lifetime and propagation effects we write the dispersion relation in the form
, @xmath113 where @xmath114 is the decay rate and @xmath115 is the rate of phase advance .
the real and imaginary parts of @xmath116 , corresponding to the longer lived of the two modes , are plotted in fig .
[ fig : psh ] . as is apparent from fig .
[ fig : ratepsh ] , the spin polarization lifetime , @xmath117 remains infinite at the psh wave vector , despite the presence of the electric field .
this result is consistent with the theoretical prediction that at the su(2 ) point the spin helix generation operators commute with all perturbation terms that are not explicitly spin dependent.@xcite however , the field increases the effective diffusion constant by the factor @xmath118 so that the decay rate for @xmath2 increases rapidly when the drift velocity approaches the thermal velocity of the electrons .
the spin helix generation operators wo nt commute with the hamiltonion if there exists a spatial disorder of so interactions.@xcite the rate of phase advance [ plotted in fig .
[ fig : phasepsh ] ] vanishes at @xmath88 , i.e. , the psh is stationary , despite the fact that the fermi sea of electrons is moving by with average velocity @xmath104 .
moreover , spin spirals with @xmath119 will appear to move backward , that is , opposite to the direction of electron flow .
although unusual , this property can be understood by considering the spin dynamics in a frame moving with velocity @xmath104 . in this frame
@xmath120 parallel to @xmath76 is perceived as a precession vector @xmath121 .
therefore in the moving frame @xmath122 .
transforming back to the laboratory frame then yields @xmath123 .
the nature of spin propagation at the su(2 ) symmetry point can be made more clear if we fourier transform from the wave vector to spatial domain .
if we inject a @xmath124-function stripe of @xmath79 polarized spins at @xmath125 , the space - time evolution of @xmath12 is proportional to the propagator , @xmath126 , where @xmath127 where @xmath128 and @xmath129 are the weighting factors for the passive and active modes , respectively and @xmath130 in the su(2 ) case . upon substituting the dispersion relations
@xmath131 , we obtain , @xmath132}.\ ] ] the spin propagator is the product of a gaussian envelope function and a static spin wave with wave vector @xmath0 .
the envelope function is the one - dimensional diffusion propagator with width proportional to @xmath133 and drift velocity @xmath93 .
an illustration of the space - time evolution described by this propagator is provided fig .
[ fig : waterpsh ] , for a drift velocity @xmath134 .
note that the phase of the spin wave modulated by the gaussian envelope remains stationary as the packet drifts and diffuses .
this contrasts with the more familiar wave packet , where the modulated wave and envelope functions both propagate , albeit with velocities that may differ . with a normalized @xmath124-function injection at @xmath135 , and drift velocity @xmath134 in the su(2 )
the spin polarization develops into a conserved stationary wave with a gaussian wave packet.,scaledwidth=60.0% ] when su(2 ) is exact , the integral of the gaussian envelope function is conserved , even in the presence of an @xmath27 field .
however , stanescu and galitski @xcite have shown theoretically that @xmath25 , which is nonzero in real systems , breaks su(2 ) .
@xcite verified experimentally that @xmath25 is indeed the factor that limits psh lifetime in experiments on ( 001 ) gaas quantum wells . in this section
we calculate the dispersion relation and spin packet time evolution in the presence of a small cubic dresselhaus term .
it was shown previously that when @xmath25 is small , the maximum lifetime occurs when the rashba interaction @xmath136 ( ref . ) .
we consider a qw with rashba coupling tuned to this value and assume that @xmath137 .
this condition is met in qws in the 2d limit , where @xmath138 ( @xmath139 is the well width ) . in this case
the dispersion relation in the presence of the electric field can be written as @xmath140where @xmath141 and @xmath142 . performing the fourier transform to obtain the space - time evolution of a spin packet ,
we obtain , @xmath143}.\]]in the presence of the cubic dresselhaus interaction the integral of the gaussian envelope is no longer conserved .
the decay rate can be written in the form , @xmath144 illustrating that although the decay rate is nonzero , it is reduced relative to the dp relaxation rate by a factor @xmath145 .
this ratio is expected theoretically,@xcite and has been verified experimentally,@xcite to be determined by the relation , @xmath146 for quite reasonable qw parameters a @xmath25 to @xmath10 ratio of 1:100 can be achieved , equivalent to a lifetime enhancement relative to the dp spin memory time on the order of @xmath147 . finally , we consider a fully symmetric well in which only the linear dresselhaus coupling exists
. to make comparison with the su(2 ) situation , we set the strength of the linear dresselhaus coupling be @xmath148 , so that the resonant wave vector is at @xmath149 .
the dispersion relations @xmath150 and @xmath151 obtained by substituting @xmath152 and replacing @xmath10 by @xmath148 in eq .
( [ io0 ] ) are plotted in fig .
[ fig : dresselhauss ] .
some qualitative features of the dispersion relations are similar to the su(2 ) case , in that @xmath150 has a global minimum and @xmath151 crosses zero at @xmath153 .
the most important difference is that the minimum @xmath150 does not reach zero , and therefore the spin spiral does decay . in the limit of low electric field ,
the lifetime of the spin spiral is only about a factor of 2 longer than the @xmath154 ( dp ) lifetime .
the propagation of a spin packet in the linear - dresselhaus - only case is illustrated in fig .
[ fig : waterdresselhaus ] , using the same initial condition and drift velocity as in su(2 ) case .
we performed numerical integration of eq .
( [ ft ] ) to obtain the propagator . as we have seen previously
, a drifting and diffusing envelope function modulates a spiral spin wave .
however , now the spiral spin fades very quickly .
the contrast between linear dresselhaus only and su(2 ) is illustrated in fig .
[ fig : area ] , which is a plot of the integral of the envelope as a function of time . after a rapid initial decay ,
the integral is constant in the su(2 ) case , whereas with only the linear dresselhaus interaction the integrated amplitude decays exponentially with rate @xmath155 .
figure [ fig : xt ] presents another way of visualizing the difference in propagation for the su(2 ) [ fig .
[ fig : sxt ] ] and linear - dresselhaus - only [ fig . [ fig : dxt ] ] hamiltonians .
the @xmath79 component of spin polarization is shown ( with color coded amplitude ) as a function of time on the vertical axis and position on the horizontal axis .
it is clear , from the vertical orientation of the contours that the positions of the nodes and antinodes of @xmath12 are fixed in space . in the linear - dresselhaus - only case with
the same initial condition and applied @xmath27 field as in the su(2 ) case .
the features are similar to those in the su(2 ) case , except the envelope function decays exponentially.,scaledwidth=60.0% ] is conserved after an initial decay ; while in the linear - dresselhaus - only case , @xmath156 decays exponentially.,scaledwidth=45.0% ]
we have developed a random walk model to describe the time evolution of electron spin in two dimensions in the presence of rashba and dresselhaus interactions . from the random walk model we derived equations of motion for spin polarization and obtained dispersion relations for @xmath14 parallel to one of the symmetry directions of the rashba / dresselhaus hamiltonian . in sec .
[ sec : rw ] , we showed that the dispersion relations for spin - polarization waves that spiral in the plane containing the surface normal and the wave vector are identical to those obtained from previous analyses.@xcite the random walk approach is instructive in showing , in a simple but explicit way , how anomalous spin diffusion and the persistent spin helix arise from nonvanishing correlations between the velocity and spin precession vectors . in sec .
[ sec : e ] , we obtained dispersion relations for spin - polarization waves that include the effects of an electric field parallel to @xmath14 , to second order in @xmath27 .
the terms linear in @xmath27 are equivalent to those obtained from the quantum kinetic approach.@xcite to first order in @xmath27 , the field introduces a precession vector in the plane of the 2deg and perpendicular to @xmath120 .
the precession about the @xmath89 axis gives rise to an unusual behavior in that the spiral with wave vector @xmath0 is stationary in space despite the motion of electrons in the field ; waves with @xmath157 propagate in the same direction as the drifting electrons while those with @xmath119 propagate backward . "
the terms that are second order in @xmath27 affect the decay rate of spin polarization without changing the velocity .
the solutions obtained when these terms are included point to the special properties of waves with wave vector @xmath0 , whose lifetime turns out to be unchanged by the field . however , the decay rate of the all other waves increases , in proportion to @xmath1 .
we illustrated these results by considering three representative spin - orbit hamiltonians : su(2 ) symmetric or @xmath24 and @xmath83 ; su(2 ) broken by a small but nonzero @xmath25 ; and linear dresselhaus coupling only or @xmath152 . in order to show the nature of spin propagation more clearly , we fourier transformed the solutions from wave vector to real space and obtained the dynamics of spin - polarization packets . in all cases
the spin packets move at the electron drift velocity . in the su(2 ) case the integrated amplitude of the spin spiral is conserved , while in the linear - dresselhaus - only case the amplitude decays with a rate @xmath158 . when su(2 ) is weakly broken by small , but nonzero @xmath25 , the integrated amplitude decays at a rate @xmath159 .
the conclusions reached by our analysis of the rw model are consistent with a recent monte carlo study of a specific 2deg system , a ( 001 ) @xmath160 quantum well with carrier density @xmath161 @xmath162 ( ref . ) . in this study spin polarization dynamics were calculated under conditions of steady state injection from a ferromagnetic contact . for @xmath163 ratios that are close to unity , the spin polarization is conserved over several wavelengths of the psh , despite the fact that transport takes place in the diffusive regime .
moreover , the polarization is not diminished with increasing electric field .
the authors point out that the psh effect can be used to achieve a novel variation of the datta - das spin - field - effect transistor ( ref . )
in which a gate electrode modulates the @xmath9 to @xmath10 ratio only slightly away from unity .
this has the effect of varying the wavelength of the psh without significantly reducing its lifetime .
thus small changes in gate voltage can in principle lead to large changes in source to drain conductance .
whether such a device can actually be realized depends on two factors : fabricating ferromagnetic injectors and analyzers with high figures of merit , and demonstrating that the psh effects that have been observed at temperatures below @xmath164 100 k ( ref . ) can be realized at room temperature .
this work was supported by the director , office of science , office of basic energy sciences , materials sciences and engineering division , of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 . | we introduce and solve a semi - classical random walk ( rw ) model that describes the dynamics of spin polarization waves in zinc - blende semiconductor quantum wells .
we derive the dispersion relations for these waves , including the rashba , linear and cubic dresselhaus spin - orbit interactions , as well as the effects of an electric field applied parallel to the spin polarization wave vector . in agreement with calculations based on quantum kinetic theory [ p. kleinert and v. v. bryksin , phys .
rev .
b * 76 * , 205326 ( 2007 ) ] , the rw approach predicts that spin waves acquire a phase velocity in the presence of the field that crosses zero at a nonzero wave vector , @xmath0 . in addition , we show that the spin - wave decay rate is independent of field at @xmath0 but increases as @xmath1 for @xmath2 .
these predictions can be tested experimentally by suitable transient spin grating experiments . |
in the spatial multiplexing scheme , @xmath6 spdc sources are pumped by a single pulsed laser whose power is split equally for each spdc source by a series of beam splitters ( bss ) .
these spdc sources are then coupled by fast photon routers and directed to a single output .
each individual spdc source has the probability to generate one pair , @xmath0 .
the pump power input is chosen low enough that the generation of more than one pair , @xmath1 , is much smaller than @xmath0 . with sufficiently large @xmath6 and feed - forward operation of the fast photon router
, the probability of obtaining one pair emission in this array , @xmath7 , can approach unity and hence be on demand .
the detection signal of the heralding ( trigger ) photon of an individual source is used to identify which source has produced a pair of photons and to control the photon routers to direct the successfully created photon to the single output of this @xmath6-spdc array .
et al . _
proposed a modular configuration by using @xmath8 routers @xcite , where two inputs can be routed to a single output . as shown in fig .
[ fig1](a ) , in a 1-spdc hsps , one nonlinear crystal ( nlc ) cut for collinear type - ii phase matching is pumped by laser pulses , where a pair of orthogonally polarized photons is generated .
their quantum state is a product state in the polarization degree of freedom : @xmath9 . @xmath10 and @xmath11 denote the vertical and horizontal polarization states of photons 1 and 2 respectively
. then they are separated by a polarizing beam splitter ( pbs ) .
the detection of the vertically polarized photon heralds the existence of the horizontally polarized photon , which is the output of this 1-spdc source .
two such 1-spdc sources can be coupled with a photon router and integrated into a 2-spdc module , as also shown in fig .
[ fig1](a ) .
it is straightforward to increase the size of the network .
two 2-spdc modules can build a 4-spdc module and so on [ fig .
[ fig1](b ) ] .
the advantages of this multiplexing are as follows : first , one can enhance the one - pair - generation probability by a gain factor of ( see appendix a ) @xmath12second , the signal - to - noise ratio of an @xmath6-spdc array is the same as that of a 1-spdc source : @xmath13 .
@xmath14 is the probability of obtaining more than one pair in this array . in fig .
[ fig1](c ) , the gain @xmath15 of a multiplexed system with @xmath6 sources ( blue ) is shown and compared with a single 1-spdc source ( red ) at @xmath16 . for @xmath17 close to 1 ( @xmath3 close to 0 ) , it first increases linearly with @xmath6 .
the gain saturates at the value @xmath18 for an infinite number of sources .
the black dashed line in fig .
[ fig1](c ) shows this saturation limit at @xmath16 .
the dependence of the gain on the mean photon pair number per spdc source @xmath3 and the numbers of the sources @xmath6 is shown in fig .
[ fig1](d ) . intuitively , in order to obtain high - quality and high - rate single photons
, one should operate in a regime , in which @xmath3 is low for each spdc , in order to reduce the higher order emission per source , and @xmath6 is large , so that the chance of one source firing in this array is high .
this is the reason why the optimal operating combination of @xmath3 and @xmath6 in terms of maximizing @xmath15 lies in the upper - left corner in fig .
[ fig1](d ) , given that @xmath6 is limited to 10 .
( color online ) scheme and calculations of generating on - demand single photons via multiplexing with active feed forward from spontaneous parametric down conversion ( spdc ) .
( a ) a 2-spdc module consists of two 1-spdc sources coupled with a photon router .
each 1-spdc source produces a pair of photons probabilistically from a nonlinear crystal ( nlc ) pumped by laser pulses whose power is split by a beam splitter ( bs ) .
the signal of each avalanche photodiode ( apd ) detector , indicated by the black arrows , is feed - forwarded and controls the photon router to direct one of the successfully created photons to a single output .
this output single photon along with the output feed - forward signal can then be used in the next level .
suitable individual optical delays ( not shown ) are incorporated in order to erase the output photons temporal distinguishability from different sources .
( b ) similarly , a 4-spdc multiplexing system can be realized by coupling two 2-spdc modules with another photon router .
an @xmath6-spdc multiplexing system requires @xmath6 spdc sources and @xmath19 routers .
( c ) the calculated gain , _ g _ , of using a multiplexed system is plotted as a function of the number of spdc sources , @xmath6 , where only integer values of @xmath6 is relevant ( blue solid line ) .
the trivial gain of using a 1-spdc source is shown for comparison ( red solid line ) .
the mean photon pair number per spdc source is @xmath16 , indicated with the vertical dashed line in ( d ) , and the horizontal black dashed line represents the saturation limit @xmath20 .
in addition , we show the gain when using multiplexing with a 0.95 transmission per path router and 0.97 of the polarization router ( blue dashed line ) in order to illustrate loss effects . for details , see the main text and appendix a. * d * , calculation result of the gain , _ g _ as a function of both the numbers of spdc sources @xmath6 and the mean photon pair number per spdc source @xmath3.,scaledwidth=49.0% ]
initially , we created photon pairs from a @xmath21-barium - borate crystal ( bbo ) via noncollinear type - ii phase matching @xcite .
as shown in fig . 2a , photons 1 and 2 were generated from bbo1 and their quantum state was @xmath22 . in comparison with
the product state generated from the collinear phase matching ( @xmath23 ) , this non - collinear scheme is equivalent to having two 1-spdc sources ( @xmath24 and @xmath25 ) , each subject to half of the pump power
. additionally , the non - collinear type - ii phase matching overlaps the spatial modes of these two sources as schematically shown by the dashed pbs in fig .
by using a polarization switch and a polarizer , these two 1-spdc sources form a 2-spdc module . in comparison with the scheme proposed in ref . [
16 ] , our noncollinear scheme has two advantages : the amounts of crystals and beam splitters needed for building an @xmath6-spdc array are reduced from @xmath6 to @xmath26 and there is no need to spatially overlap the photons generated from two 1-spdc sources .
an identical 2-spdc module subsequently generated photons 3 and 4 from bbo2 .
see appendix b for details .
( color online ) experimental setup .
( a ) two pairs of the polarization entangled photons ( 1 and 2 , 3 and 4 ) were generated from two @xmath21-barium borate crystals ( bbo1 and bbo2 ) via spdc , where half - wave plates ( @xmath27/2 ) and compensating bbo crystals were used to counter walkoff effects in the down - conversion crystal , and single - mode fibers and interference filters ( ifs ) were used to clean their spatial and spectral modes , respectively .
photons 1 and 4 entered into spatial modes a and d respectively .
photons 2 and 3 entered into spatial modes b and c and were delayed in single - mode fibers ( @xmath28 ) with respect to photons 1 and 4 .
this allowed the logic and electronics to have enough time to control the eoms in the router .
due to the sequential configuration of the sources ( see appendix a ) , appropriate individual fiber delays are implemented to eliminate the output photons temporal distinguishability from different sources .
the layout of the ultrafast photon router is shown in ( b ) .
see text and appendix cfor details and notations .
( c ) the experimental scheme of a 4-spdc multiplexed system with active feed - forward .
sources i and ii , iii and iv generated photon pairs from bbo 1 ( bbo 2 ) .
sources i and ii ( iii and iv ) were inherently overlapped via the noncollinear phase matching , as explained in the text .
the polarization states of photons 1 and 4 were measured with polarizing beam splitters ( pbs ) and single - photon counting modules ( spcms ) .
the detection signals of photons 1 and 4 were fed into the logic and then used to control the ultrafast photon router . depending on whether photon 1 or photon 4 was detected ,
the ultrafast photon router would be switched on or off .
thus , all the photons exited from mode b@xmath29 and no photons exited from c@xmath29 .
the quality of the output single photons was quantified via measurements of the correlation function , @xmath30.,scaledwidth=48.0% ] these two 2-spdc modules form a 4-spdc multiplexed system with an ultrafast photon router , which satisfies two main criteria : ( 1 ) short response time , which is necessary for the feed - forward operations in a large system , and ( 2 ) high routing visibility , which is needed for a high quality hsps . as shown in fig .
2(b ) , we employed a mach - zehnder interferometer ( mzi ) to realize the ultrafast photon router .
it consisted of two 50:50 bss , mirrors , and most importantly two electro - optic modulators ( eoms ) with one in each arm of the mzi , where two rubidium titanate phosphate ( rtp ) crystals were used as the electro - optic material .
this design of the router has several advantages : ( 1 ) it can route incoming photons independent of their polarizations , which is crucial for certain tasks . in the schemes of polarization - entanglement - based quantum cryptography @xcite and quantum computation with polarization encoded photonic qubits @xcite , any polarization - dependent routing will destroy the polarization entanglement and hence defeat the original goal of achieving communication security and computation speed - up provided by the entanglement .
for instance , in polarization - entanglement - based quantum cryptography described in ref .
@xcite , the source is located at alice s laboratory , and out of each pair of entangled photons , one is sent to alice and the other to bob . in a multiplexed scenario , many sources produce polarization - entangled photon pairs .
the trigger photons are sent to and measured by alice .
depending on the outcome of her polarization measurement , she sends an electronic signal to the routers for multiplexing the photons which will be sent to bob . multiplexed systems with our router can enhance the key rate while keeping the quantum bit error rate ( due to the multiple pair generations ) constant .
whatever critical key rate is needed at the receiver , @xmath6-spdc multiplexing extends the possible communication distance between the two parties by the additional distance @xmath31 compared to 1-spdc , where @xmath32 is the @xmath33 decay distance in the transmission channel ( valid for both free space and fiber ) . in a 1550 nm fiber - based cryptography system
, a 4-spdc multiplexing system could increase the available distance by @xmath34 .
note however , whether one needs entangled photons depends on what kind of quantum - information processing task the photon will be used for later on .
one does not need entanglement for the operation of the multiplexing procedure itself or for using multiplexing in the bennett and brassard quantum cryptography protocol @xcite , as the latter is not based on entanglement . then one can also use a mixture of polarization - correlated product states as the input of the multiplexing scheme . ( 2 )
a similar design has been widely used in the photonic industry to implement intensity modulators @xcite .
therefore , the scheme demonstrated here has the potential to be implemented with the mature technology of integrated optics .
the optical axes of both rtp crystals in the eoms were oriented along @xmath35 and the voltages applied to them were always of the same amplitude but with opposite polarities .
if the opposite half - wave voltages ( hwvs ) were applied to the eoms , a @xmath36 phase shift would be introduced to the mzi .
we locked the phase of the mzi at @xmath37 when the eoms were off .
this 0 phase is defined via the condition that all the photons that enter from input b exit into c@xmath29 , and c into b@xmath29 ( fig .
therefore , when the eoms were on and were applied with opposite hwvs , the photons that entered from input b exited into b@xmath29 , and c into c@xmath29 .
the rising and falling times of our router for a @xmath36-phase modulation are each about @xmath38 ns . ( this could be reduced to tens of picoseconds with state - of - art integrated photonics devices @xcite . )
the recharger time of the high - voltage driver of our eoms is about @xmath39 ns . the traveling time in the used coaxial cables is also a few nanoseconds .
together , this in principle allows a repetition rate of our router of approximately @xmath40 mhz .
( note that the dead time of trigger apds is about @xmath41 ns , which will limit the operating frequency of small - scale multiplexing systems . )
the visibility of @xmath36-phase modulation ( routing visibility ) , @xmath42 , was above 95%
. it is defined as @xmath43 for spatial mode b as the input , where @xmath44 and @xmath45 are the output intensities of spatial mode b@xmath29 of phase @xmath37 and @xmath36 , respectively .
similar results were obtained for spatial mode c as the input .
therefore , this photon router meets the two requirements : short response time ( compared to the recharging time of our eoms ) and high routing visibility .
see appendix c for further details of this ultrafast photon router . the 4-spdc multiplexing system [ fig .
2(c ) ] was operated as follows .
we detected photons 1 and 4 in the modes of 1h , 1v , 4h and 4v , and sent the electrical signals from the detectors through a custom - built programmable logic to the router .
when the detector on mode 1h or 1v fired , the eoms in the mzi were switched on with hwv and thus the phase of the mzi was switched to @xmath36 .
therefore , photon 2 was routed to spatial mode b@xmath29 .
when the detector on mode 4h or 4v fired , the eoms in the mzi were switched off , and the phase of the mzi remained at 0 . therefore , photon 3 was routed to spatial mode b@xmath29 . in the last step ,
an additional electro - optical modulator ( eom3 ) placed in spatial mode b@xmath29 acted as a polarization router and performed an active polarization rotation on the output photon , before it was finally cleaned by using a pbs .
however , due to limited resources , we bypassed eom3 and performed this operation with post - selection .
note that introducing eom3 would only reduce the routing visibility by at most 1% ( with a polarization switching visibility of at least 99% ) @xcite .
in order that our _ m_-spdc source is useful in photonic quantum - information processing , it is most important that it performs well in hong - ou - mandel ( hom ) two - photon interference @xcite , which is at the heart of many protocols and has been demonstrated with different systems and schemes .
for instance , interference between consecutive photons emitted from the same quantum dots @xcite , between photons from different donor impurities @xcite and molecules @xcite , as well as between two independent spdc sources @xcite have been shown . at present , in terms of the count rate and quality of the independent single photons , femtosecond - pulsed spdc is still better than other schemes and has been successfully used in multiphoton experiments in the past . a single - photon source based on our scheme is suitable for high quality hom two - photon interference and enables multiphoton experiments @xcite .
the present multiplexing scheme includes the photon router as the only additional complexity which was not involved in earlier experiments .
since the path length difference of the mzi can be controlled very accurately , the router will not introduce any temporal distinguishability , even though it is based on first - order interference which is wavelength sensitive ( nm range ) .
in contrast , most of the photonic quantum - computation experiments ( c - phase gate @xcite , entanglement swapping @xcite , etc . ) rely on second - order interference , which is coherence length sensitive ( depends on the context , but at least more than @xmath46 nm ) . in order to empirically exclude possible degradations of the indistinguishability due to the router , we have performed hom two - photon interference experiments with the router .
experimentally , we use a pair of photons generated from one source and send it to the router .
we vary the path - length difference between these two photons with a motorized translation stage mounted on one of the fiber coupling stages and measure the twofold coincidences between two detectors placed directly behind two outputs of the router [ b@xmath29 and c@xmath29 in fig .
[ fig2](b ) ] .
the phase of the mzi and hence the reflectivity ( and transmissivity ) of the router is varied by applying different voltages on the eoms . in case of a @xmath47 phase ,
the router becomes a balanced bs and the distinguishability of two input photons spatial modes is erased . in consequence ,
the minimum of the coincidence counts occurs for the optimal temporal overlap ( with the help of suitable individual fiber delays ) of the two photons , and hom two - photon interference with a visibility of @xmath48 can be observed ( red circles in fig .
[ fighom ] ) . in case of a @xmath37 phase
, the whole router represents a highly transmissive bs ( routing visibility above 95% ) and the two photons remain distinguishable in their spatial modes . correspondingly , one obtains path - length - difference insensitive coincidence counts ( black square in fig .
[ fighom ] ) .
this is in agreement with complementarity , where in principle no hom interference can be observed .
in addition , we have compared the results when using the router with those when using a normal free - space bs : quantitatively identical results have been obtained ( within the error bars ) . these hom experiments together with the earlier works @xcite demonstrate the suitability of the multiplexed photons for scalable quantum information processing .
( color online ) experimental hom two - photon interference performed with the photon router .
the two - fold coincidences are measured between two detectors placed directly behind two outputs of the photon router ( b@xmath29 and c@xmath29 in fig . [
fig2]b ) and are plotted over the relative delay of the interfering photons [ realized by changing the length of path c in fig .
2(b ) ] . by adjusting the voltage applied on the eoms , we set the phase of the mzi and hence varied the reflectivity ( and transmissivity ) of the router
when the phase of the mzi is @xmath47 , hom two - photon interference shows up ( red circles ) .
this is because the distinguishability of the two input photons spatial modes has been erased by the router .
consequently , the minimum coincidence counts occur for the temporal overlap of the two photons , i.e. zero time delay .
the red solid line is a gaussian fit to the data with a visibility of @xmath48 .
when the phase is 0 , path length difference insensitive coincidence counts are obtained ( black squares ) and no hom two - photon interference shows up , because the two photons remain distinguishable in their spatial modes . the black solid line is the average of the coincidence counts for 0 phase setting .
background has not been subtracted .
error bars indicate @xmath49 standard deviation.,scaledwidth=49.0% ] we quantified the quality of the multiplexed single photons with the second - order correlation function at zero time delay , @xmath30 .
the smaller @xmath30 is , the higher the quality of the single photons becomes .
the correlation function can be measured in good approximation with a 50:50 bs and two more detectors @xcite , as shown in fig .
it is defined as @xmath50 , where @xmath51 , @xmath52 , @xmath53 , and @xmath54 are the coincidence counts among trigger , transmission , and reflection of the beam splitter , the coincidence counts between trigger and transmission , the coincidence counts between trigger and reflection , and the single counts of the trigger , respectively .
figure [ fig3 ] shows @xmath30 versus the counts per 350 seconds with 1-spdc , 2-spdc and 4-spdc multiplexing systems .
( color online ) experimental results .
the correlation function at zero delay @xmath30 is plotted versus the counts in 350 s for 1-spdc ( black circles ) , 2-spdc ( red triangles ) and 4-spdc ( blue squares ) multiplexed systems , where all of them contain the router circuitry .
the @xmath30 function saturates at the value 1 for large count rates , i.e. for large pump powers .
the @xmath30 data shown are in linear regime of low count rates and hence fitted with lines through the origin . using multiplexed systems
, one can increase the count rate while keeping constant the signal - to - noise ratio and hence the @xmath30 function .
in other words , increasing the number of spdc sources enables one to increase the count rate with the same signal - to - noise ratio or ( if one turns down the pump power ) improves the quality of the single - photon output while maintaining the count rate of the output .
error bars indicate @xmath49 standard deviation.,scaledwidth=46.0% ] we performed the experiments with various levels of pump power ( 100 , 250 , 500 , and 1000 mw ) to demonstrate the heralded single - photon quality and generation rate trade - off of the spdc sources . with a 1-spdc source ( black circles in fig .
[ fig3 ] ) , the counts increased as we increased the pump power .
the quality of the output single photon decreased as @xmath30 was increased because of higher order emissions . at a constant mean photon pair number @xmath3 per source , with single - photon multiplexing , 2-spdc ( red triangles in fig .
[ fig3 ] ) and 4-spdc multiplexed ( blue squares in fig . [ fig3 ] ) systems enhanced the counts by factors of 2 and 4 , respectively , while @xmath30 remained the same as the 1-spdc . on the other hand
, one can understand fig .
[ fig3 ] also from the following perspective : by accordingly decreasing @xmath3 per source , one can employ multiplexed systems while keeping the counts constant and increase the quality of the output single photon via decreasing @xmath30 . in the regime of small
@xmath3 , the counts scale proportionally to @xmath6 at fixed @xmath30 , and @xmath30 scales inversely proportional to @xmath6 at fixed counts . because the coupling efficiencies of the sources are different , the count rates and correlation functions are also different .
the values of the counts and the correlation function shown in fig .
[ fig3 ] were averaged over four spdc sources ( see appendix b ) .
the transmission throughput of the single - photon router is an important parameter for the viability of a multiplexed system , and next we provide an analysis of losses in our system . in our experiment , the output photon s transmission of the 4-spdc multiplexing system is @xmath55 , where the main loss is due to single - mode - fiber to single - mode - fiber coupling in the router . to have an advantage of 4-spdc multiplexing over a 1-spdc without router circuitry would necessitate @xmath56 or , equivalently ,
@xmath57 if the routers are all the same and each have a transmission of @xmath58 .
the gain due to multiplexing is @xmath59 if the mean photon pair number is small ( see appendix a ) .
since the transmission of 0.1 of our router circuitry is a technical issue , our experiment is a proof - of - concept demonstration of single - photon multiplexing and demonstrates its advantage .
based on current technology and theoretical analysis of the fiber coupling efficiency , it may be feasible to improve the transmission per router to about 95% . this number is obtained from experimentally achievable fresnel losses on each optical surface and theoretically predicted fiber coupling efficiency @xcite .
then , the total gain of a 4-spdc source with routers compared to a 1-spdc source without router would be approximately @xmath60 .
( color online ) dependence of the photon correlation @xmath61 vs the heralding output probability @xmath62 , determined in a numerical simulation of spdc based hspss with different types of detectors ( see appendix d and ref .
@xcite ) . the heralding output probability @xmath63 is defined as the ratio of the twofold coincidence counts ( between trigger and output ) and the laser repetition rate .
the blue curve is for a 1-spdc source with a standard bucket photon detector ( d ) with 60% detection efficiency for the trigger photon .
the red curve is assuming a photon - number - resolving detector ( pnrd ) with 95% efficiency .
remarkably , as shown by the green curve , a 4-spdc multiplexed system with standard detectors is able to outperform a 1-spdc heralding source realized with a highly advanced prnd.,scaledwidth=47.0% ] practically , the repetition rate of multiplexing is only limited by the repetition rate of the pulsed laser ( in our case 80 mhz , but femtosecond lasers with 1 ghz are commercially available @xcite ) , the jitter of the detector ( 500 ps in our case , but those with 50 ps are commercially available @xcite ) and the switching time of the modulators ( 5.6 ns in our case , but less than 100 ps has been shown @xcite ) .
in addition , by employing state - of - the - art photonic structures , low insertion loss of the modulators can be achieved @xcite .
it is thus possible to realize a gigahertz on - demand single - photon source with this scheme . since in our experiment the photons
are generated by the spdc process pumped with a femtosecond - pulsed laser , which has been successfully used in multiphoton experiments in the past , a single photon source based on our scheme is suitable for high - quality multiphoton interference @xcite , which is at the heart of the quantum repeater @xcite and linear optical quantum computation @xcite .
it is very interesting to note that a 4-spdc source such as demonstrated here based on standard bucket single - photon detectors , would potentially outperform a heralded photon source based on a highly efficient photon - number - resolving detector ( pnrd ) such as a superconducting transition - edge sensor ( tes ) @xcite in specific multi - photon experiments @xcite ( see fig .
[ fig4 ] and appendix d for details ) .
this shows , that a 4-spdc source would have important practical implications for advancing existing linear optical quantum computing experiments .
it is conceivable that a scheme which combines the multiplexing technique with superconducting detectors as trigger detectors will be advantageous .
moreover , the scheme we have presented here is universal for any photon - pair source , including spdc , spontaneous four - wave mixing , and so on . for the future realization of a multiplexed system with many levels , stabilizing the interferometers will be challenging for bulk optics but less for integrated optics .
moreover , a 60% transmission throughput of an integrated photonic quantum circuit has already been reported @xcite , which includes the fiber - coupling efficiency and is well above the loss threshold of @xmath64 mentioned above .
therefore , together with state - of - art micro - optics technology , ( for instance , integrated photonic quantum circuits on a silicon chip @xcite , faster modulators @xcite , and on - chip single - photon detectors @xcite ) , it is possible to develop a compact , scalable and nearly on - demand single - photon source for photon - based quantum science and technology following our scheme .
we thank w. naylor , s. ramelow , c. schff and p. walther for discussions .
we acknowledge support from the european commission , project qap ( no . 015848 ) , q - essence ( no . 248095 ) , erc advanced grant , sfb - foqus , jtf and the doctoral program coqus of the austrian science foundation ( fwf ) .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , & _ _ * * , ( ) .
_ et al .
_ _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , , &
_ _ * * , ( ) . , , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , , &
_ _ * * , ( ) . , , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , , , & _ _ * * , ( ) . , , , & _ _
* * , ( ) .
_ et al . _
_ et al . _
_ _ * * , ( ) . & _ _ * * , ( ) . , & _ _
* * , ( ) . , &
_ _ * * , ( ) . , &
_ _ * * , ( ) . , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . & in _ _ , ( ) . , , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , & _ _ * * , ( ) . , , , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , , , & _ _ ( ) .
_ et al . _
_ _ * * , ( ) . , , , & _ _ * * , ( ) . , , & _ _ * * , ( ) .
_ _ ( , ) . & _ _ * * , ( ) .
http://www.gigaoptics.de http://www.idquantique.com/. , , & _ _ * * , ( ) . ,
& _ _ * * , ( ) . , & _ _ ( ) .
, , , & _ _ * * , ( ) . , , & _ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) . , , , & _ _ * * , ( ) . , , , & , _ _ ( )
in an array , @xmath6 _ sequentially _ pumped hspss , based on spdc , are coupled with @xmath65 photon routers .
since the array of spdc sources is pumped sequentially , the source with the shortest path provides the first chance for getting a single photon , the next longer provides the next chance , and so on .
the routers are operated in a way such that once they are activated by the trigger detector fired in a particular path , they block the outputs of all the other sources . therefore , the source with the shortest path has `` priority '' in the array .
the overall probability of obtaining one and only one output single photon out of the whole array is as follows : @xmath66where @xmath67 is the probability of generating @xmath37 output photons from an individual spdc source .
analogously , the probability of obtaining more than one output photon is : @xmath68therefore , the signal to noise ratio of obtaining one output single photon from this array equals that of a single spdc source : @xmath69the probability of obtaining one output single photon is enhanced by a gain factor of : @xmath70where the approximation at the end holds for @xmath71 . in any experimental setup , photon losses in the routers
decrease the performance of the multiplexing scheme .
we now analyze the influence of these losses quantitatively .
we consider the case where the mean photon pair number @xmath3 is much smaller than 1 .
then , @xmath17 is close to 1 and thus the gain becomes @xmath72 [ as long as @xmath73 . due to the cascaded setup
, however , using @xmath6 spdc sources means that the output photon has to travel through @xmath74 path routers ( assuming @xmath75 is an integer number ) and hence has an increasing chance of being lost .
the output photon s transmission through all @xmath76 path routers be denoted as @xmath77 . in order to have an advantage over a single photon source , the gain @xmath15 has to overcompensate the loss such that the total gain
, @xmath78 is larger than 1 .
this leads to the generic condition @xmath79 .
then , @xmath80 is increasing with @xmath6 and there is no upper bound . in case
the transmission through all path routers is the same , and each single router transmission is denoted as @xmath58 , the output photon s total transmission through the routers is @xmath81 .
the total gain @xmath82 has to be larger than 1 , which leads to the condition that the transmission per path router must fulfill @xmath57 .
therefore , multiplexing is advantageous over a single - photon source if and only if the loss of each individual path router is smaller than @xmath64 ( under the assumption of small mean photon pair number ) .
losses in the routers do build up but are then overcompensated by the gain of the multiplexing . in our hybrid experimental scheme ,
two different kinds of routers are needed , because we employ the polarization and the path degrees of freedom . in an @xmath6-spdc arrangement utilizing polarization and path ,
there are only @xmath83 path routers ( in our case , only one mzi because @xmath84 ) , and it is sufficient to have one polarization router ( in our scheme , eom3 ) at the end of the array .
therefore , every output photon has to travel through @xmath85 path routers ( assuming @xmath85 is an integer number ) and one polarization router .
when we denote the transmission of each path router by @xmath86 and the transmission of the polarization router by @xmath87 , the total transmission of the hybrid multiplexed system is @xmath88 .
the total gain therefore becomes @xmath89 demanding the total gain has to be larger than 1 , leads to the condition @xmath90{2 t_{\textrm{pol}}}}$ ] . in the case of a 4-spdc multiplexed system with high - transmission polarization routing , @xmath91 , and
the requirement for the transmission of the path router is @xmath92 .
[ [ the - pulsed - four - photon - spontaneous - parametric - down - conversion - sources ] ] the pulsed four - photon spontaneous parametric down conversion sources ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ we used ultraviolet ( uv ) femtosecond pulses of an up - converted mode - locked ti : sapphire laser to pump a 2 mm type - ii bbo crystal at a wavelength of 404 nm with a pump power of 1400 mw , a pulse duration of 180 fs , and a repetition rate of 80 mhz . the uv pump power was adjusted by varying the up - conversion efficiency via moving the up - conversion crystal in or out of the focus of the lens for the fundamental pulse .
the uv pulses successively passed through two bbo crystals to generate two polarization entangled photon pairs ( photons 1 and 2 , and photons 3 and 4 ) in spatial modes a and b , and c and d via type - ii spontaneous parametric down - conversion respectively , as shown in fig .
after proper filtering , maximally polarization entangled states of the form @xmath93 were generated .
@xmath94 is the phase difference between horizontal and vertical polarization due to birefringence in the crystal , which could be corrected by placing a half wave plate ( @xmath27/2 ) oriented at @xmath35 and 1-mm compensation bbo crystals in each photon s path @xcite . by tilting the compensation bbo crystals , we set the phases to be 0 for both sources , and hence we obtained the two polarization - entangled states described in the text .
to exactly define the spatial and spectral properties of the emitted four photons , we coupled each photon into single - mode fibers and filter them with 3 nm ( full width at half maximum , fwhm ) bandpass interference filters ( if ) for photons 1 and 4 , and 1 nm ifs for photons 2 and 3 . in the hom interference experiment , we used 3 nm ifs for both photons .
the photons were detected by single - photon counting modules ( spcms ) , and the coincidences were registered with a coincidence logic . at the pump power of @xmath95 mw ,
the average trigger rate is about @xmath96 khz and the mean photon pair number generated per pulse is approximately @xmath97 .
note that the coupling efficiency of the sources from bbo 1 is about @xmath98% , and that from bbo 2 is about @xmath99% .
this difference eventually gives the difference in the correlation functions and the count rates of each source .
for clarity , the 1-spdc data shown in fig . [ fig3 ] were obtained by averaging the @xmath30 functions and the counts of the four individual 1-spdc sources .
the 2-spdc ( 4-spdc ) data were obtained by averaging the @xmath30 functions and summing the counts of two ( four ) 1-spdc sources .
we used a mzi to realize the ultrafast photon router .
the performance of this router strongly depends on the phase stability of the mzi . to achieve that
, we built the interferometer in an enclosed box made by acoustic isolation materials in order to stabilize the phase passively .
additionally , an active phase stabilization system has also been implemented .
this was accomplished by using an auxiliary beam from a power - stabilized he - ne laser counter - propagating through the whole mzi with a little transversal displacement from the signal beam , which picked up the phase fluctuation of the interferometer . after passing through the interferometer ,
the intensity fluctuation of the he - ne laser beam was measured with a silicon photon detector and the signal was fed into an analog proportional - integral - derivative ( pid ) regulator .
a ring piezotransducer attached to one of the mirrors in the mzi was controlled by this pid regulator and compensated the phase fluctuation actively .
the insertion loss of this router and fiber delays was about 90@xmath100 , which was mainly due to the single - mode fiber coupling , lossy narrow - band interference filters ( in spatial modes b@xmath29 and c@xmath29 , 1 nm fwhm , about 70% peak transmission ) and fresnel loss on various optical elements . the most straight - forward step to improving the transmission of the router is to use high peak transmission and broadband interference filters . with a 95% peak transmission and 3-nm fwhm interference filter ,
we have measured the total transmission throughput of the router , including the delay fibers and correlation function measurement setup , to be more than 25% .
note that one can engineer the interactions between the pump and nonlinear crystals and match the group velocities of the pump and down - converted photons @xcite . by employing this novel technique
, one can generate pure single photons without using narrow - band filters , which potentially allows us to further enhance the transmission throughput to be more than 50% .
[ [ comparing-4-spdc - to - a-1-spdc - with - different - types - of - photon - detectors ] ] comparing 4-spdc to a 1-spdc with different types of photon detectors ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in order to estimate the required performance of hsps in practical applications , we implemented a numerical simulation of the quantum optics circuits .
thereby , every photon mode is represented in a fock space from 0 to 4 excitations , and all unitary operations are connected hamiltonians expressed within the full hilbert space .
hence , all higher order photon terms are intrinsically taken into account , up to the natural cutoff from the fock space . through numerical simulation
we studied the output probability for the hsps versus the observed second - order correlation function for zero time delay , @xmath101 , for various pumping strengths @xmath102 of the spdc operator , @xmath103 where @xmath104 and @xmath105 are the two photon modes .
these results are shown in fig .
interestingly , the 4-spdc source based on 60@xmath100 efficiency bucket detectors ( for detecting trigger photons ) and optical transmission throughput of 95% per router , which should be reasonable to reach with an optimized optical system , can outperform a hsps built from 1-spdc heralded with the currently best possible detector technology , such as a pnrd system ( for detecting trigger photons ) with total detector efficiency of 95@xmath100 .
note that we assume the coupling efficiency from the source to the detectors to be 70% for both cases , because that is independent of the detectors . | an on - demand single - photon source is a fundamental building block in quantum science and technology .
we experimentally demonstrate the proof of concept for a scheme to generate on - demand single photons via actively multiplexing several heralded photons probabilistically produced from pulsed spontaneous parametric down - conversions ( spdcs ) . by utilizing a four - photon - pair source , an active feed - forward technique , and an ultrafast single - photon router
, we show a fourfold enhancement of the output photon rate .
simultaneously , we maintain the quality of the output single - photon states , confirmed by correlation measurements .
we also experimentally verify , via hong - ou - mandel interference , that the router does not affect the indistinguishability of the single photons .
furthermore , we give numerical simulations , which indicate that photons based on multiplexing of four spdc sources can outperform the heralding based on highly advanced photon - number - resolving detectors .
our results show a route for on - demand single - photon generation and the practical realization of scalable linear optical quantum information processing .
numerous applications for on - demand single - photon sources have been proposed in the field of quantum - enhanced science and technology @xcite . the commonly studied methods to generate on - demand
single photons are based on emissions from molecules @xcite , atoms @xcite , color centers in diamond @xcite , quantum dots @xcite , and donor impurities @xcite
. however , each of these methods has certain challenges to overcome . for atomic systems , the repetition rates , the collection efficiency , and the complexity of the experimental setup are the main obstacles . for color centers , quantum dots and donor impurities , it is difficult to achieve a time - bandwidth limited emission spectrum @xcite and indistinguishable output photons from different sources @xcite .
an alternative approach is the so - called heralded single - photon source ( hsps ) based on the process of spontaneous parametric down - conversion ( spdc ) , where a pair of photons is created from a nonlinear crystal pumped by a laser .
the detection of one trigger photon indicates , or heralds , the presence of its twin . because of the phase - matching condition and energy conservation , a hsps from spdc has some appealing features
the linear momentum ( direction ) , polarization , and wavelength of that heralded photon are well defined via the measurement of its twin .
the scheme faces two main challenges : ( 1 ) the conversion process is inherently random and hence there is no prior knowledge of when the heralding event will occur .
( 2 ) because of the nature of the spdc process , in addition to the probability of generating one pair of photons , @xmath0 , there is also a finite probability of generating more than one pair , @xmath1 , via higher order emissions , which decreases the quality of the hsps .
this latter generation probability increases nonlinearly with the interaction strength of the pump laser and the nonlinear crystal . to surmount the random production of photon pairs from spdc ,
a pulsed laser can be used to pump the nonlinear crystals , and hence the photon pairs can be created only at certain times . when the pulse duration of the pump laser is much shorter than the coincidence measurement time , the probability of generating _
n _ pairs of photons per pulse can be represented as a poisson distribution : @xmath2where the mean number of pairs , @xmath3 , depends on the pump power and parameters of the crystals .
note that it has been shown that to use bose ( thermal ) distribution , the analysis presented below would yield essentially the same results @xcite .
an ideal hsps would have exactly one photon generated from each pulse , which requires @xmath4 .
however , for @xmath4 , the probability to observe more than one photon is 0.26 due to the poisson distribution .
this significant higher order emissions probability , @xmath1 , drastically reduces the quality of the hsps . for this reason ,
pulsed spdc is normally operated at low power ( thus @xmath5 ) and obeys a trade - off between the count rate and quality of the hsps .
this is the reason why a single spdc source is fundamentally limited .
this limitation can be overcome by using spatial or temporal multiplexing of several sources @xcite .
although this idea has attracted significant attention , as highlighted in recent review articles @xcite , it has not yet been experimentally realized .
here we follow and extend the proposal in ref . [
16 ] .
we report an experimental realization of a 4-spdc single - photon multiplexing system and demonstrate a proof - of - principle enhancement over a 1-spdc single - photon source . |
the growing interest in the interdisciplinary physics of complex systems , has focussed physicists attention on agent - based modeling @xcite of social dynamics , as a very attractive methodological framework for social sciences where concepts and tools from statistical physics turn out to be very appropriate @xcite for the analysis of the collective behaviors emerging from the social interactions of the agents .
the dynamical social phenomena of interest include residential segregation @xcite , cultural globalization @xcite , opinion formation @xcite , rumor spreading @xcite and others .
the question that motivates the formulation of axelrod s model for cultural dissemination @xcite is how cultural diversity among groups and individuals could survive despite the tendencies to become more and more alike as a result of social interactions .
the model assumes a highly non - biased scenario , where the culture of an agent is defined as a set of equally important cultural features , whose particular values ( traits ) can be transmitted ( by imitation ) among interacting agents .
it also assumes that the driving force of cultural dynamics is the `` homophile satisfaction '' , the agents commitment to become more similar to their neighbors .
moreover , the more cultural features an agent shares with a neighbor , the more likely the agent will imitate an uncommon feature s trait of the neighbor agent . in other words , the higher the cultural similarity , the higher the social influence . the simulations of the model dynamics show that for low initial cultural diversity , measured by the number @xmath0 of different traits for each cultural feature ( see below ) , the system converges to a global cultural state , while for @xmath0 above a critical value @xmath1 the system freezes in an absorbing state where different cultures persist .
the ( non - equilibrium ) phase transition @xcite between globalization and multiculturalism was first studied for a square planar geometry @xcite , but soon other network structures of social links @xcite were considered , as well as the effects of different types of noise ( `` cultural drift '' ) @xcite , external fields ( modeling _ e.g. _ influential media , or information feedback ) @xcite , and global or local non - uniform couplings @xcite . in all those extensions of axelrod s model mentioned in the above paragraph , the cultural dynamics occurs on a network of social contacts that is fixed from the outset . however , very often social networks are dynamical structures that continuously reshape . a simple mechanism of network reshaping is agents mobility , and a scenario ( named the axelrod - schelling model ) where cultural agents placed in culturally dissimilar environments are allowed to move has recently been analyzed @xcite . in this model ,
new interesting features of cultural evolution appear depending on the values of a parameter , the ( in-)tolerance , that controls the strength of agents mobility . a different mechanism of network reshaping has been considered in @xcite , where a cultural agent breaks its link to a completely dissimilar neighbor , redirecting it to a randomly chosen agent . at variance with the mobility scenario of the axelrod - schelling model , that limits the scope of network structures to clusters configurations on the starting structure ( square planar lattice , or others ) , the rewiring mechanism allows for a wider set of network structures to emerge in the co - evolution of culture and social ties . in this paper
we introduce in the scenario of network rewiring a tolerance parameter @xmath2 controlling the likelihood of links rewiring , in such a way that the limit @xmath3 recovers the case analyzed in @xcite , where only links with an associated null cultural overlap are broken .
lower values of @xmath2 correspond to less tolerant attitudes where social links with progressively higher values of the cultural overlap may be broken with some probability that depends on these values .
the results show a counterintuitive dependence of the tolerance @xmath2 on the critical value @xmath1 . on one hand , as expected from @xcite , rewiring promotes globalization for high values of the tolerance , but on the other hand , very low values of @xmath2 ( which enhance the rewiring probability ) show the higher values of @xmath1 .
indeed , a non monotonous behavior is observed in @xmath4 : our results unambiguously show that for some intermediate values of the tolerance @xmath2 , cultural globalization is disfavored with respect to the original axelrod s model where no rewiring of links is allowed . in other words , rewiring does not always promote globalization . on the other hand ,
the resulting network topology depends on @xmath0 , changing from a poisson connectivity distribution @xmath5 to a fat tailed distribution for @xmath6 .
as in axelrod s model , the culture of an agent @xmath7 is a vector of @xmath8 integer variables @xmath9 ( @xmath10 ) , called cultural _ features _ , that can take on @xmath0 values , @xmath11 , the cultural _ traits _ that the feature @xmath12 can assume .
the @xmath13 cultural agents occupy the nodes of a network of average degree @xmath14 whose links define the social contacts among them .
the dynamics is defined , at each time step , as follows : * each agent @xmath7 imitates an uncommon feature s trait of a randomly chosen neighbor @xmath15 with a probability equal to their _ cultural overlap _
@xmath16 , defined as the proportion of common cultural features , @xmath17 where @xmath18 denotes the kronecker s delta which is 1 if @xmath19 and 0 otherwise .
the whole set of @xmath13 agents perform this step in parallel . *
each agent @xmath7 disconnects its link with a randomly chosen neighbor agent @xmath15 with probability equal to its _ dissimilarity _ @xmath20 , provided the dissimilarity @xmath21 exceeds a threshold ( _ tolerance _ ) @xmath2 , @xmath22 and rewires it randomly to other non - neighbor agent .
the tolerance @xmath23 is a model parameter .
first we note that the initial total number of links in the network is preserved in the rewiring process , so the average degree @xmath14 remains constant .
however , the rewiring process allows for substantial modifications of the network topological features , _
e.g. _ connectedness , degree distribution , etc . in that respect , except for the limiting situation of very low initial cultural diversity @xmath0 and a very high tolerance @xmath2 ( where the likelihood of rewiring could be very low ) , one should expect that the choices for the initial network of social ties have no influence in the asymptotic behavior of the dynamics .
when the threshold tolerance @xmath2 satisfies @xmath24 , only those links among agents with zero cultural overlap are rewired , so the model becomes the one studied in @xcite .
on the other hand , when the tolerance takes the value @xmath25 , there is not rewiring likelihood and the original axelrod s model is recovered .
when @xmath26 rewiring is always possible provided the cultural similarity is not complete , _
@xmath27 , so that it corresponds to the highest intolerance .
the usual order parameter for axelrod s model is @xmath28 , where @xmath29 is the average ( over a large number of different random initial conditions ) of the number of agents sharing the most abundant ( dominant ) culture , and @xmath13 is the number of agents in the population .
large values of the order parameter characterize the globalization ( cultural consensus ) regime .
we also compute the normalized size @xmath30 of the largest network component ( _ i.e. _ , the largest connected subgraph of the network ) .
we have studied networks of sizes @xmath31 , @xmath32 ; averaging over @xmath33 - @xmath34 replicas .
the considered cultural vectors have @xmath35 cultural features , each one with a variability @xmath36 @xmath37 - @xmath38 .
we studied different values of the tolerance threshold @xmath39 and different values of the average connectivity @xmath40 .
each simulation is performed for @xmath13 , @xmath8 , @xmath14 , @xmath2 , and @xmath0 fixed .
for the sake of comparison with previous results @xcite , we will present results for @xmath41 .
the behavior of the order parameter for different values of @xmath2 is seen in fig .
[ k4_diff_z ] . like in @xcite ,
three different macroscopic phases are observed with increasing values of @xmath0 , namely a monocultural phase , with a giant cultural cluster , a multicultural one with disconnected monocultural domains , and finally a multicultural phase with continuous rewiring .
the nature of the latter phase has been successfully explained in @xcite : at very large values of the initial cultural diversity @xmath0 , the expected number of pairs of agents sharing at least one cultural trait becomes smaller than the total number of links in the network , so that rewiring can not stop .
here we will focus attention on the first two phases and the transition between them . in figure [ histog ]
we show the size distribution of the dominant culture over different realizations , measured for different values of @xmath0 , at a particular fixed value of the tolerance @xmath42 . in the region of @xmath0 values near the transition from globalization to multiculturalism ,
the distribution is double peaked , indicating that the transition is first order , as in the original axelrod s model .
the transition value , @xmath1 , may be roughly estimated as the @xmath0 value where the peaks of the size distribution are equal in height .
the estimates of the transition points for different values of the tolerance @xmath2 are shown in fig .
[ qc_vs_z ] .
the non monotonous character of the graph @xmath4 seen in this figure reveals a highly non trivial influence of the tolerance parameter on the co - evolution of cultural dynamics and the network of social ties .
let us first consider the ( most tolerant ) case @xmath43 that , except for the system size @xmath13 , coincides exactly with the situation considered in @xcite , _
i.e. _ , only links with zero cultural overlap are rewired . as discussed in @xcite , for @xmath0 values larger than the critical value for a fixed network ( @xmath44 )
, rewiring allows redirecting links with zero overlap to agents with some common cultural trait ( compatible agents ) , so reinforcing the power of social influence to reach cultural globalization .
once all links connect compatible agents , rewiring stops @xcite . from there on , the network structure will remain fixed , and globalization will be reached with the proviso that the network has so far remained connected .
this is the case for most realizations ( for @xmath31 ) up to values of @xmath45 .
increasing further the cultural diversity @xmath0 , increases the frequency of rewiring events and slows down the finding of compatible agents , favoring the topological fragmentation into network components before rewiring stops . under these conditions ,
the asymptotic state will consist of disconnected monocultural components .
on one hand , network plasticity allows to connect compatible agents , so promoting globalization ; but on the other hand it may produce network fragmentation , so favoring multiculturalism .
what we have seen in the previous paragraph is that for @xmath43 the first effect prevails over the second one up to @xmath46 .
going from there to less tolerant situations ( decreasing @xmath2 ) , increases the likelihood of rewiring , making easier that network fragmentation occurs before rewiring stops .
this has the effect of decreasing the critical value @xmath1 .
in fact , from fig
. [ qc_vs_z ] we see that for @xmath47 and @xmath48 multiculturalism prevails for cultural diversities where the original axelrod s model shows cultural globalization . in these cases
network plasticity promotes multiculturalism in a very efficient way : agents segregate from neighbors with low cultural similarity and form disconnected social groups where full local cultural consensus is easily achieved , for @xmath0 values low enough to allow a global culture in fixed connected networks . for very low values of the tolerance parameter ,
though network fragmentation occurs easily during the evolution , fig .
[ qc_vs_z ] shows that globalization persists up to very high values of the initial cultural diversity @xmath0 . to explain this seemingly paradoxical observation
, one must realize that network fragmentation is not an irreversible process , provided links connecting agents with high cultural overlap have a positive rewiring probability . under these circumstances , transient connections among different components
occur so frequently so as to make it possible a progressive cultural homogenization between components that otherwise would have separately reached different local consensuses .
[ time.evolution ] illustrates the time evolution for @xmath49 and different values of @xmath2 .
panel ( a ) shows an example of cultural evolution where network fragmentation reverts to a connected monocultural network for @xmath50 . panel ( b ) , that corresponds to @xmath51 , shows that social fragmentation persists during the whole evolution , while in panel ( c ) , which corresponds to the most tolerant situation ( @xmath43 ) , the network remains connected all the time .
the degree distribution of the network is poissonian centered about @xmath14 for all @xmath0 values , except for @xmath52 where it becomes fat tailed , with several lowly connected ( and disconnected ) sites .
for very high @xmath0 values , in the dynamical phase , the network rewiring is esentially random , so @xmath53 is again poisson like , centered around @xmath14 .
in this paper we have generalized the scenario for co - evolution of axelrod s cultural dynamics and network of social ties that was considered in @xcite , by introducing a tolerance parameter @xmath2 that controls the strength of network plasticity .
specifically , @xmath2 fixes the fraction of uncommon cultural features above which an agent breaks its tie with a neighbor ( with probability equal to the cultural dissimilarity ) , so that , the lower the @xmath2 value , the higher the social network plasticity .
our results show that the network plasticity , when controlled by the tolerance parameter , has competing effects on the formation of a global culture .
when tolerance is highest , network plasticity promotes cultural globalization for values of the initial cultural diversity where multiculturalism would have been the outcome for fixed networks . on the contrary , for intermediate values of the tolerance , the network plasticity produces the fragmentation of the ( artificial ) society into disconnected cultural groups for values of the initial cultural diversity where global cultural consensus would have occurred in fixed networks . for very low values of the tolerance ,
social fragmentation occurs during the system evolution , but the network plasticity is so high that it allows the final cultural homogenization of the transient groups for very high values of the cultural diversity .
intermediate tolerances promote multiculturalism , while both extreme intolerance and extreme tolerance favor the formation of a global culture , being the former more efficient than the latter .
this work has been partially supported by micinn through grants fis2008 - 01240 and fis2009 - 13364-c02 - 01 , and by comunidad de aragn ( spain ) through a grant to fenol group .
y. m. was partially supported by the fet - open project dynanets ( grant no .
233847 ) funded by the european commission and by comunidad de aragn ( spain ) through the project fmi22/10 . | starting from axelrod s model of cultural dissemination , we introduce a rewiring probability , enabling agents to cut the links with their unfriendly neighbors if their cultural similarity is below a tolerance parameter . for low values of tolerance , rewiring promotes the convergence to a frozen monocultural state
. however , intermediate tolerance values prevent rewiring once the network is fragmented , resulting in a multicultural society even for values of initial cultural diversity in which the original axelrod model reaches globalization . |
the sine - gordon model is a paradigm for relativistic integrable models in @xmath0 dimensions ( e.g. , see @xcite .
) its multi - soliton spectrum is well known and consists not only of multi - kink scattering configurations but also of bound states , the simplest of which is the so - called breather .
it may be obtained formally by analytically continuing the kink - antikink configuration in its relative velocity variable , @xmath1 , and oscillates periodically in time .
a systematic procedure for deriving the integrability features of the sine - gordon model relates it to the self - duality equations of su(2 ) yang - mills theory in @xmath2 dimensions @xcite . in a light - cone gauge
, these equations follow from the nair - schiff action @xcite .
a first and straightforward dimensional reduction produces ward s modified chiral sigma - model action for su(2)-valued fields in @xmath3 dimensions @xcite .
a second dimensional reduction then generates an abelian sigma model equivalent to the sine - gordon theory .
to arrive there , one must prescribe a particular dependence on one spatial coordinate ( rather than trivial independence ) and also algebraically restrict the field from su(2 ) to a u(1 ) subgroup .
the remaining phase @xmath4 turns out to be ruled by the sine - gordon equation , with the coupling or mass @xmath5 appearing as a parameter of the dimensional reduction . for several years now the moyal deformation of integrable field theories has been of some interest . in particular , the ward model @xcite and the sine - gordon model @xcite have been generalized to the noncommutative realm .
the key insight for the latter case was that the extension of su(2 ) to u(2 ) , necessary for implementing the moyal deformation in the yang - mills theory , should be retained under the dimensional and algebraic reduction , so that the noncommutative sigma - model field takes its values in u(1)@xmath6u(1 ) rather than u(1 ) .
the deformed sine - gordon model so obtained @xcite features _ two _ scalar fields ( phases ) @xmath7 and @xmath8 , whose noncommutative abelian wzw actions are coupled in a simple way . in the commutative limit , the average @xmath9 of these phases produces the standard sine - gordon field while their difference @xmath10 decouples as a free field . since
the powerful techniques for constructing multi - soliton solutions in integrable models have been shown to survive the noncommutative deformation , it is straightforward ( but may be tedious ) to work out such configurations for the moyal - deformed sine - gordon model as well .
the basic strategy was already outlined in @xcite but has been applied only to the simple kink so far @xcite . however , owing to the relativistic invariance , a one - kink configuration depends only on its single co - moving coordinate @xmath11 , and so it can not get deformed . only multi - lumps with relative motion should be affected by noncommutativity .
the first instances are the two - kink , kink - antikink and breather solutions . in this letter
we apply the moyal deformation to the two latter cases .
it is important to verify the effect of noncommutativity , since the tree - level computations of @xcite had suggested that perhaps the entire moyal deformation of the sine - gordon model might be ficticious . here
, we demonstrate this _ not _ to be the case , by working out the first - order ( in the noncommutativity parameter ) correction to the ` classical ' kink - antikink and breather configurations .
it turns out that this leading correction affects only the would - be free field @xmath10 ; the generalized sine - gordon field @xmath12 gets modified at second order onwards , as does the energy density . only the substantial calculational effort prevented us from evaluating higher orders ,
but we present the starting - point equations for doing so . as an exact result ,
the temporal periodicity of the breather is unchanged by the deformation .
the integrable noncommutative sine - gordon model introduced in @xcite involves _ two _ u(1)-valued fields @xmath13 and may be defined via its action @xmath14 \= s_{\textrm{wzw}}[g_+]\ + \ s_{\textrm{wzw}}[g_-]\ + \
\alpha^2\!\int\!\diff{t}\,\diff{x}\;\bigl(g_+^\+g_-+g_-^\+g_+-2\bigr ) \quad,\ ] ] where @xmath15 is the abelian wzw action @xmath16\= - \sfrac12\int\!\diff t\,\diff x\ ; \bigl ( \pa_t g^\+\,\pa_t g - \pa_x g^\+\,\pa_x g ) \
-\ \int\!\diff t\,\diff x\int_0 ^ 1\!\!\diff \la\ \hat g^\+\pa_{[t}\hat g\star\hat g^\+\pa_{x]}\hat g\star\hat g^\+\pa_\la\hat g\ ] ] with a homotopy path @xmath17 connecting @xmath18 and @xmath19 and a moyal star product @xmath20_\star = \im\,\theta \quad.\ ] ] in light - cone variables @xmath21 the corresponding equations of motion read @xmath22 \pa_v\bigl(g_+^\+\star\pa_u g_+-g_-^\+\star\pa_u g_-\bigr ) & \= 2\a^2\bigl(g_+^\+\star g_- - g_-^\+\star g_+ \bigr ) \quad , \end{aligned}\ ] ] which in the commutative limit @xmath23 simplifies to @xmath24 hence , the identification of the standard sine - gordon field @xmath12 with mass @xmath25 is made via @xmath26 for later use , we embed the u(1)-valued fields into u(2 ) , @xmath27 g_+ { - } g_- & \ \ g_+ { + } g_- \end{matrix}\biggr ) \quad\buildrel{\theta\to0}\over{\longrightarrow}\quad \ep^{\frac{\im}{4}(\p_+-\p_- ) } \biggl(\begin{matrix } \
\cos\sfrac{\vp}{2 } & \im\sin\sfrac{\vp}{2 } \\[4pt ] \im\sin\sfrac{\vp}{2 } & \
\cos\sfrac{\vp}{2 } \end{matrix}\biggr ) \quad.\ ] ]
the breather solution may be obtained from a kink - antikink configuration with relative velocity @xmath28 by an analytic continuation @xmath1 .
the co - moving coordinates for the kink and antikink read @xmath29 respectively , where @xmath30 , @xmath31 a convenient way to construct the kink - antikink solution employs the dressing method . for the case at hand
, it yields @xcite @xmath32_\star^{-1}\star\tdag_1 \nonumber \\[6pt ] & -\ 2\,\bigl ( { \mathbbm{1}}- \sfrac{1+\vel}{\vel}\,p_1 \bigr ) \star t_2\star \bigl[\tdag_2\star({\mathbbm{1}}{-}\s p_1)\star t_2\bigr]_\star^{-1}\star\tdag_2 \label{gconf } \\[12pt ] = \ { \mathbbm{1}}\ & -\ 2\,\bigl ( { \mathbbm{1}}+ \sfrac{1-\vel}{\vel}\,p_2 \bigr ) \star p_1 \star \bigl [ { \mathbbm{1}}{-}\s p_2\star p_1\bigr]_\star^{-1 }
-\ 2\,\bigl ( { \mathbbm{1}}- \sfrac{1+\vel}{\vel}\,p_1 \bigr ) \star p_2 \star \bigl [ { \mathbbm{1}}{-}\s p_1\star p_2\bigr]_\star^{-1 } \quad , \nonumber\end{aligned}\ ] ] where @xmath33 and we introduced hermitian projectors @xmath34 based on @xmath35 matrix - valued functions @xmath36 and @xmath37 related to the kink and antikink components of the configuration .
the @xmath38 are determined only up to right ( star- ) multiplication with an arbitrary invertible function and may be taken as @xmath39 \im\ep^{2\a\h_1 } \end{matrix}\biggr ) \und t_2 \= \biggl(\begin{matrix } 1 \\[4pt ] -\im\ep^{2\a\h_2 } \end{matrix}\biggr)\ ] ] by a suitable choice of the coordinate origin .
note that we have dropped the star index on the exponentials since each one depends on a single coordinate combination only . by inserting these @xmath38 into ( [ proj ] ) , the ensuing projectors into ( [ gconf ] )
one is in principle able to read off @xmath40 from ( [ embed ] ) and extract the noncommutative breather configuration @xmath41 .
before delving into the explicit computation , let us first retrieve the familiar commutative breather in the @xmath23 limit .
since a coordinate rescaling modifies the coupling @xmath42 we take the freedom to put @xmath43 in the following . dropping all stars ,
one first builds @xmath44 \im\ep^{\h_1 } & \ep^{2\h_1 } \end{matrix}\biggr ) \und p_2 \= \frac{1}{1+\ep^{2\h_2}}\ \biggl(\begin{matrix } 1 & \im\ep^{\h_2 } \\[4pt ] -\im\ep^{\h_2 } & \ep^{2\h_2 } \end{matrix}\biggr)\ ] ] and thus @xmath45^{-1 } \= \frac{1+\ep^{2\h_k } } { ( 1{-}\s)(1{-}\ep^{\h_1+\h_2})^2+(\ep^{\h_1}{+}\ep^{\h_2})^2 } \qquad\textrm{for}\quad ( j , k ) = ( 1,2)\ \textrm{or}\ ( 2,1 ) \quad.\ ] ] next , one obtains @xmath46 -\im\sfrac{1-\vel}{\vel}\ep^{\h_2 } & 1{+}\sfrac1\vel\ep^{2\h_2 } \end{matrix}\biggr ) \biggl(\begin{matrix } 1 & -\im\ep^{\h_1 } \\[4pt ] \im\ep^{\h_1 } & \ep^{2\h_1 } \end{matrix}\biggr ) \quad , \\[8pt ] \bigl({\mathbbm{1}}-\sfrac{1+\vel}{\vel}\,p_1 \bigr)\,t_2\,\tdag_2 \=&\ \frac{1}{1+\ep^{2\h_1}}\ \biggl(\begin{matrix } -\sfrac1\vel{+}\ep^{2\h_1 } & \im\sfrac{1+\vel}{\vel}\ep^{\h_1 } \\[4pt ] -\im\sfrac{1+\vel}{\vel}\ep^{\h_1 } & 1{-}\sfrac1\vel\ep^{2\h_1 } \end{matrix}\biggr ) \biggl(\begin{matrix } 1 & \im\ep^{\h_2 } \\[4pt ] -\im\ep^{\h_2 } & \ep^{2\h_2 } \end{matrix}\biggr ) \quad , \end{aligned}\ ] ] which combine to @xmath47 2\im\vel^{-1}(\ep^{\h_1}{+}\ep^{\h_2})(1{-}\ep^{\h_1+\h_2 } ) & \vel^{-2}(1{-}\ep^{\h_1+\h_2})^2{-}(\ep^{\h_1}{+}\ep^{\h_2})^2 \end{matrix}\biggr ) \nonumber \\[12pt ] & \= \frac{1}{\sinh^2\g\vel t + \vel^2\cosh^2\g x}\ \biggl(\begin{matrix } \sinh^2\g\vel t-\vel^2\cosh^2\g x & 2\im\vel\sinh\g\vel t\ \cosh\g x\\[8pt ] 2\im\vel\sinh\g\vel t\ \cosh\g x & \sinh^2\g\vel t-\vel^2\cosh^2\g \end{matrix}\biggr ) \label{gcomm}\end{aligned}\ ] ] with the help of @xmath48 comparing with ( [ embed ] ) we learn that , with @xmath49 , @xmath50 so that @xmath51 analytically continues via @xmath52 to @xmath53 with @xmath54 .
since @xmath55 is real we must have @xmath56 or @xmath57 .
the boundary condition @xmath58 for @xmath59 selects the second option , and shift @xmath60 , or else , put @xmath56 and shift @xmath61 . ] and we have recovered the celebrated breather configuration @xcite @xmath62
when attempting to repeat the above computation in the moyal - deformed case , one must account for the noncommutativity of the co - moving coordinates , @xmath63_\star \= \im\theta \qquad\longrightarrow\qquad [ \h_1\,,\,\h_2]_\star \= 2\im\theta\,\g^2\vel \= 2\im\theta\,\sfrac{\vel}{1-\vel^2 } \ = : \ \im\la
\quad,\ ] ] which leads to the fundamental intertwining relation , @xmath64 which ( for @xmath65 regular at zero ) implies @xmath66 again , we put @xmath43 for convenience .
the projectors ( [ projconf ] ) are unaffected by the deformation , but the star products become relevant when @xmath67 or @xmath68 meets @xmath69 or @xmath70 . as a basic ingredient in ( [ gconf ] ) ,
we first compute @xmath71 \!\!\im\s\,\ep^{\h_k } & 1{+}(1{-}\s)\ep^{2\h_k } \end{matrix } \biggr)\ , ( 1{+}\ep^{2\h_k})^{-\frac12 } \biggr ] \star \biggl ( \begin{matrix } 1 \\[8pt ] \im\ep^{\h_j } \end{matrix } \biggr ) \\[8pt ] \nonumber \=&\quad\ \ ; ( 1{+}\ep^{2\h_k})^{-\frac12 } \star ( 1{-}\s+\ep^{2\h_k } ) \star ( 1{+}\ep^{2\h_k})^{-\frac12 } \\
\nonumber & + \ ( 1{+}\ep^{2\h_k})^{-\frac12 } \star \s\,\ep^{\h_j+\h_k\mp\frac\im2\la } \star ( 1{+}\ep^{2\h_k\mp2\im\la})^{-\frac12 } \\ & + \
( 1{+}\ep^{2\h_k\pm2\im\la})^{-\frac12 } \star \s\,\ep^{\h_j+\h_k\pm\frac\im2\la } \star ( 1{+}\ep^{2\h_k})^{-\frac12 } \\
\nonumber & + \ ( 1{+}\ep^{2\h_k\pm2\im\la})^{-\frac12 } \star ( \ep^{\h_j}+(1{-}\s)\ep^{2\h_j+2\h_k } ) \star ( 1{+}\ep^{2\h_k\mp2\im\la})^{-\frac12 } \\[8pt ] \nonumber \=&\ ( 1{+}\ep^{2\h_k})^{-\frac12 } \star \bigl[(1{-}\s)(1{-}\ep^{\h_1+\h_2})^2 + ( \ep^{\h_1}{+}\ep^{\h_2})^2\bigr]\star ( 1{+}\ep^{2\h_k})^{-\frac12 } \ + \
o(\la^2 ) \quad.\end{aligned}\ ] ] in the last step , we dropped terms of @xmath72 in order to arrive at a manageable expression . inserting the above into ( [ gconf ] ) and abbreviating @xmath73 we find the matrix elements of @xmath74 up to @xmath72 ( denoted by ` @xmath75 ' ) , @xmath76\star n_2^{\frac12}\star d^{-1}\star n_2^{\frac12 } \\ & \quad\ , -\ 2\,n_1^{-1}\star \bigl [ \ep^{2\h_1}-\sfrac1\vel+\sfrac{1+\vel}{\vel}\ep^{\h_1+\h_2+\frac\im2\la } \bigr]\star n_1^{\frac12}\star d^{-1}\star n_1^{\frac12 } \quad , \end{aligned } \\[8pt ] & \begin{aligned } g_{12 } \
\simeq\ & \qquad\phantom { ( } 2\im\,n_2^{-1}\star \bigl [ \ep^{2\h_2}+\sfrac1\vel-\sfrac{1-\vel}{\vel}\ep^{\h_1+\h_2-\frac\im2\la } \bigr]\star n_2^{\frac12}\star d^{-1}\star n_2^{\frac12}\star\ep^{\h_1}\\ & \quad\ , - 2\im\,n_1^{-1}\star \bigl [ \ep^{2\h_1}-\sfrac1\vel+\sfrac{1+\vel}{\vel}\ep^{\h_1+\h_2+\frac\im2\la } \bigr]\star n_1^{\frac12}\star d^{-1}\star n_1^{\frac12}\star\ep^{\h_2 } \quad , \end{aligned } \\[8pt ] & \begin{aligned } g_{21 } \
\simeq\ & \quad\ , -2\im\,n_2^{-1}\star \bigl [ \ep^{\h_1}-\sfrac{1-\vel}{\vel}\ep^{\h_2}+\sfrac1\vel\ep^{\h_1 + 2\h_2-\im\la } \bigr]\star n_2^{\frac12}\star d^{-1}\star n_2^{\frac12}\\ & \quad\ , + 2\im\,n_1^{-1}\star \bigl [ \ep^{\h_2}+\sfrac{1+\vel}{\vel}\ep^{\h_1}-\sfrac1\vel\ep^{2\h_1+\h_2+\im\la } \bigr]\star n_1^{\frac12}\star d^{-1}\star n_1^{\frac12 } \quad , \end{aligned } \\[8pt ] & \begin{aligned } \label{g22 } g_{22 } \
\simeq\ & \ 1\
-\ 2\,n_2^{-1}\star \bigl [ \ep^{\h_1}-\sfrac{1-\vel}{\vel}\ep^{\h_2}+\sfrac1\vel\ep^{\h_1 + 2\h_2-\im\la } \bigr]\star n_2^{\frac12}\star d^{-1}\star n_2^{\frac12}\star\ep^{\h_1}\\ & \quad\ , -\ 2\,n_1^{-1}\star \bigl [ \ep^{\h_2}+\sfrac{1+\vel}{\vel}\ep^{\h_1}-\sfrac1\vel\ep^{2\h_1+\h_2+\im\la } \bigr]\star n_1^{\frac12}\star d^{-1}\star n_1^{\frac12}\star\ep^{\h_2 } \quad .
\end{aligned } \end{aligned}\ ] ] there is some pattern with respect to the interchange @xmath77 and regarding sign flips of @xmath78 and @xmath79 , but no obvious symmetry under @xmath80 .
we have chosen the positions of the @xmath81 such that their arguments are not shifted .
the equalities @xmath82 and @xmath83 are far from manifest .
note that , for the exact result , @xmath84 is to be inverted with respect to star multiplication .
however , since @xmath85 , we may take the ordinary inverse in ( [ g11])([g22 ] ) .
finally , the commutative limit collapses @xmath74 to ( [ gcomm ] ) , since all @xmath81 factors cancel and disappear .
we can also employ the last equation of ( [ gconf ] ) , which expresses @xmath74 in terms of projectors only .
after rescaling the projectors to @xmath86 we rewrite @xmath87_\star^{-1 } \star\bigl[1+\tp_2\bigr]\ -\ 2\,p_2\star\bigl[{\mathbbm{1}}-\tp_1\star\tp_2\bigr]_\star^{-1 } \star\bigl[1+\tp_1\bigr ] \\[4pt ] \= & { \mathbbm{1}}\ -\
\sfrac{2\vel}{1-\vel}\,\bigl [ \tp_1 + \tp_1{\star}\tp_2 + \tp_1{\star}\tp_2{\star}\tp_1 + \ldots \bigr ] \ + \ \sfrac{2\vel}{1+\vel}\,\bigl [ \tp_2 + \tp_2{\star}\tp_1 + \tp_2{\star}\tp_1{\star}\tp_2 + \ldots \bigr ] \quad .
\end{aligned}\ ] ] in the last line , we have traded the notorious star inverses for formal geometric series , @xmath88_\star^{-1 } \= \sum_{n=0}^\infty\,\bigl [ \tp_j\star\tp_k\bigr]^n_\star \qquad\textrm{with}\quad ( j , k ) = ( 1,2)\ \textrm{or}\ ( 2,1 ) \quad,\ ] ] which may be truncated in an approximation for large velocities @xmath89 . in this way , @xmath74 is given as a power series in words @xmath90 .
remembering ( [ proj ] ) and abbreviating also @xmath91 the ` projector words ' simplify to @xmath92 \= \bigl ( \begin{matrix } \ep^{-\frac12\h_j } \\
\pm\im\ep^{\frac12\h_j } \end{matrix } \bigr)\ , \tn_j^{-1}\star\tn_{jk}\star\tn_k^{-1}\star\tn_{kj } \star\cdots\star\tn_\ell^{-1}\ , \bigl ( \ep^{-\frac12\h_\ell } \quad { \pm}\im\ep^{\frac12\h_\ell } \bigr ) \quad , \end{aligned}\ ] ] where the last line is a symmetric rewriting with @xmath93 pulling all together , one arrives at @xmath94 \nonumber g_{12 } = & \quad\phantom{xr } 2\im n_1^{-1}\ep^{\h_1 } -2\im n_2^{-1}\ep^{\h_2 } + 2\im\sfrac{1+\vel}{\vel}n_1^{-1}{\star}n_{12}{\star}n_2^{-1}\ep^{\h_2 } + 2\im\sfrac{1-\vel}{\vel}n_2^{-1}{\star}n_{21}{\star}n_1^{-1}\ep^{\h_1 } + \ldots \quad , \\[8pt ] \nonumber g_{21 } = & \quad\ !
-2\im\ep^{\h_1}n_1^{-1 } + 2\im\ep^{\h_2}n_2^{-1 } + 2\im\sfrac{1+\vel}{\vel}\ep^{\h_1}n_1^{-1}{\star}n_{12}{\star}n_2^{-1 } + 2\im\sfrac{1-\vel}{\vel}\ep^{\h_2}n_2^{-1}{\star}n_{21}{\star}n_1^{-1 } + \ldots \quad , \\[8pt ] \nonumber g_{22 } = & \ 1 -2\ep^{\h_1}n_1^{-1}\ep^{\h_1 } -2\ep^{\h_2}n_2^{-1}\ep^{\h_2 } -2\sfrac{1+\vel}{\vel}\ep^{\h_1}n_1^{-1}{\star}n_{12}{\star}n_2^{-1}\ep^{\h_2 } + 2\sfrac{1-\vel}{\vel}\ep^{\h_2}n_2^{-1}{\star}n_{21}{\star}n_1^{-1}\ep^{\h_1 } + \!\ldots\end{aligned}\ ] ] with @xmath81 and @xmath95 to be taken from ( [ abbr1 ] ) and ( [ abbr2 ] ) , respectively .
this is an exact result .
no star - inverse needs to be taken , but we are left with infinite series , which may be summed in closed form only for @xmath96 .
the task is to extract the deformed breather configuration @xmath98 from ( [ g11])([g22 ] ) or from ( [ gseries ] ) , at least to subleading order in a @xmath97 expansion , @xmath99 keeping @xmath78 fixed and noticing that @xmath100 for any function @xmath101 , we have @xmath102 and thus @xmath103 \sfrac12(\p_+{-}\p_- ) \ & \simeq\ 2\pi + \ \im\la\,g_+^{(1)}\ep^{-\sfrac\im2\vp}\ + \ \im\la\,g_-^{(1)}\ep^{\frac\im2\vp } \= 2\pi + \ 2\im\la\,g_e^{(1)}\cos\sfrac{\vp}{2 } \ + \
2\la\,g_o^{(1)}\sin\sfrac{\vp}{2 } \end{aligned}\ ] ] since @xmath104 . from ( [ gseries ] )
one learns that , in the @xmath79 expansion , the even orders of @xmath105 and the odd orders of @xmath106 are real while the odd orders of @xmath105 and the even orders of @xmath106 are imaginary . because ( [ anglexp ] ) must be real equations for @xmath107 , this implies that @xmath108 and so the sine - gordon field @xmath109 gets deformed only at @xmath72 while the orthogonal combination @xmath110 is turned on at @xmath111 .
interestingly , the relation ( [ eorel ] ) is again the commutative one , thus @xmath112 g_o & g_e\end{matrix}\biggr ) \
\simeq\ \ep^{\im\pi+\im\la\chi}\ \biggl(\begin{matrix } \
\cos\sfrac{\vp}{2 } & \im\sin\sfrac{\vp}{2 } \\[4pt ] \im\sin\sfrac{\vp}{2 } & \
\cos\sfrac{\vp}{2 } \end{matrix}\biggr ) \qquad\textrm{with}\quad \chi = \im\,g_e^{(1)}/\cos\sfrac{\vp}{2 } = g_o^{(1)}/\sin\sfrac{\vp}{2 } \quad.\ ] ] for computing @xmath113 it suffices to look at any one of the @xmath74 matrix elements . in order to expand @xmath74 to @xmath111
we need the first subleading term in multiple star products , @xmath114 } f_j ) \cdots f_n \quad,\ ] ] where @xmath115}f_j)\equiv\frac{\pa f_i}{\pa\h_1}\frac{\pa f_j}{\pa\h_2 } -\frac{\pa f_i}{\pa\h_2}\frac{\pa f_j}{\pa\h_1}$ ] .
the products appearing in ( [ g11])([g22 ] ) take the forms @xmath116}d \bigr ) \quad,\\[4pt ] n_1^{-1}{\star}\ep^h{\star}n_1^{\frac12}{\star}d^{-1}{\star}n_1^{\frac12 } & \
\simeq\ \ep^h d^{-1}\ -\ \sfrac\im2\la\,\ep^h d^{-2 } \bigl ( 2d\,\pa_2 h\,\sfrac{n_1'}{n_1}-\pa_2 d\,\sfrac{n_1'}{n_1}-\pa_{[2}h\,\pa_{1 ] } d \bigr ) \quad,\\[4pt ] n_2^{-1}{\star}\ep^h{\star}n_2^{\frac12}{\star}d^{-1}{\star}n_2^{\frac12 } { \star}\ep^{\h_1 } & \ \simeq\ \ep^{h+\h_1}d^{-1 } \\ & \ + \ \sfrac\im2\la\,\ep^{h+\h_1 } d^{-2 } \bigl ( 2d\,\pa_1 h\,\sfrac{n_2'}{n_2}-\pa_1 d\,\sfrac{n_2'}{n_2}-\pa_{[1}h\,\pa_{2]}d + \pa_2 d - d\,\pa_2 h \bigr ) \quad , \nonumber\\[4pt ] n_1^{-1}{\star}\ep^h{\star}n_1^{\frac12}{\star}d^{-1}{\star}n_1^{\frac12 } { \star}\ep^{\h_2 } & \ \simeq\ \ep^{h+\h_2}d^{-1 } \\ & \ -\ \sfrac\im2\la\,\ep^{h+\h_2 } d^{-2 } \bigl ( 2d\,\pa_2 h\,\sfrac{n_1'}{n_1}-\pa_2 d\,\sfrac{n_1'}{n_1}-\pa_{[2}h\,\pa_{1]}d + \pa_1 d - d\,\pa_1 h \bigr ) \quad , \nonumber\end{aligned}\ ] ] with @xmath101 being linear in @xmath117 and @xmath118 .
collecting all terms and noticing cancellations we obtain @xmath119 -d^2 g_{12}^{(1 ) } & \= \ep^{\h_1 + 2\h_2}\bigl(\pa_1d(2{-}\sfrac{n_2'}{n_2})+\pa_2d-2d\bigr ) + \ep^{2\h_1+\h_2}(\pa_1d{-}2d)(1{-}\sfrac{n_2'}{n_2 } ) \ + \ ( 1\leftrightarrow2 ) \quad , \\[4pt ] d^2 g_{21}^{(1 ) } & \= \ep^{\h_1}\bigl((2d{-}\pa_1d)\sfrac{n_2'}{n_2}-\pa_2d\bigr ) + \ep^{\h_2}\pa_1d(1{-}\sfrac{n_2'}{n_2 } ) \ + \ ( 1\leftrightarrow2 ) \quad , \\[4pt ] \im\vel d^2 g_{22}^{(1 ) } & \= \ep^{2\h_1 + 2\h_2}(\pa_1d{-}2d)(2{-}\sfrac{n_2'}{n_2 } ) - \ep^{\h_1+\h_2}\bigl ( \pa_1d(1{-}\sfrac{n_2'}{n_2})+\pa_2d - d\bigr ) \ + \ ( 1\leftrightarrow2 ) \quad , \\[4pt ] \end{aligned}\ ] ] -12ptwhich further collapses to @xmath120 ^ 2 } \= -\frac{\im\vel}{2 } \frac{\sinh^2\g\vel t-\vel^2\cosh^2\g x}{[\sinh^2\g\vel t+\vel^2\cosh^2\g x]^2 } \quad , \\[8pt ] g_o^{(1 ) } & \= \ \
\,4\vel^2\ep^{\h_1+\h_2 } \frac{(\ep^{\h_1}+\ep^{\h_2})(1{-}\ep^{\h_1+\h_2 } ) } { [ ( 1{-}\ep^{\h_1+\h_2})^2+\vel^2(\ep^{\h_1}+\ep^{\h_2})]^2 } \,\= \ \vel^2 \frac{\sinh\g\vel t\ \cosh\g x}{[\sinh^2\g\vel t+\vel^2\cosh^2\g x]^2 } \quad
. \end{aligned}\ ] ] comparing to ( [ gcomm ] ) , we indeed confirm that @xmath121 , and hence @xmath122 it appears as if the sine - gordon field gets deformed via @xmath123 , but this is misleading .
this formula provides the explicit @xmath124 correction to the commutative kink - antikink configuration . to obtain the breather
, we still must analytically continue @xmath1 , which yields @xmath125 so that the leading correction to @xmath74 remains a phase factor .
this is the main result of this letter .
clearly , @xmath113 oscillates with twice the classical breather frequency @xmath126 .
more generally , our construction shows that the deformed breather frequency does not depend on @xmath97 at all .
below we illustrate the shapes of @xmath127 and @xmath128 for a typical value of @xmath78 .
ward , phil .
. lond . * a 315 * ( 1985 ) 451 ; + l.j . mason and g.a.j .
sparling , phys .
lett . * a 137 * ( 1989 ) 29 ; j. geom .
* 8 * ( 1992 ) 243 ; + m.j .
ablowitz , s. chakravarty and l.a .
takhtajan , commun .
* 158 * ( 1993 ) , 289 ; + t.a .
ivanova and a.d .
popov , phys .
lett . * a 205 * ( 1995 ) 158 [ hep - th/9508129 ] ; + theor . math .
* 102 * ( 1995 ) 280 [ teor . mat .
* 102 * ( 1995 ) 384 ] ; + s. chakravarty , s.l .
kent and e.t .
newman , j. math .
* 36 * ( 1995 ) 763 ; + m. legare , int .
j. mod .
phys . * a 12 * ( 1997 ) 219 .
v.p . nair and j. schiff ,
* b 371 * ( 1992 ) 329 .
ward , j. math .
* 29 * ( 1988 ) 386 ; commun . math . phys .
* 128 * ( 1990 ) 319 .
o. lechtenfeld and a.d popov , jhep * 0111 * ( 2001 ) 040 [ hep - th/0106213 ] ; + phys .
* b 523 * ( 2001 ) 178 [ hep - th/0108118 ] . o. lechtenfeld , fortschr . phys .
* 52 * ( 2004 ) 596 [ hep - th/0401158 ] .
c.s . chu and o. lechtenfeld ,
b * 625 * ( 2005 ) 145 [ hep - th/0507062 ] .
o. lechtenfeld , l. mazzanti , s. penati , a.d .
popov and l. tamassia , + nucl .
b * 705 * ( 2005 ) 477 [ hep - th/0406065 ] .
s. krkolu and o. lechtenfeld , jhep * 0709 * ( 2007 ) 020 [ arxiv:0708.1310 [ hep - th ] ] . | as shown in [ hep - th/0406065 ] , there exists a noncommutative deformation of the sine - gordon model which remains ( classically ) integrable but features a second scalar field .
we employ the dressing method ( adapted to the moyal - deformed situation ) for constructing the deformed kink - antikink and breather configurations .
explicit results and plots are presented for the leading noncommutativity correction to the breather .
its temporal periodicity is unchanged .
itp
uh18/08 + 2.0 cm * the noncommutative sine - gordon breather * andr fischer and olaf lechtenfeld + _ institut fr theoretische physik , leibniz universitt hannover + appelstrae 2 , 30167 hannover , germany _
+ emails : afischer , lechtenf @itp.uni - hannover.de |
motivated by a power - assignment problem in wireless networks ( see below for a short discussion of this application ) funke et al .
@xcite studied the following special case tsp(@xmath8 ) of the traveling salesman problem ( tsp ) which is specified by an integer @xmath16 and a real number @xmath17 .
the cities are @xmath18 points in @xmath19-dimensional space @xmath1 , and the distance between two points @xmath6 and @xmath7 is @xmath20 , where @xmath5 denotes the standard euclidean distance between @xmath6 and @xmath7 . *
the objective in problem tsp(@xmath8 ) is to find a shortest tour ( under distances @xmath21 ) that visits every city _ exactly _ once .
* in the closely related problem rev - tsp@xmath22 , the objective is to find a shortest tour that visits every city _ at least _ once ; thus the salesman is allowed to revisit cities .
note that tsp@xmath23 is the classical two - dimensional euclidean tspand that tsp@xmath24 is the so - called _ bottleneck tsp _ in @xmath1 , where the goal is to find a tour whose longest edge has minimum length .
we are , however , mainly interested in the case where @xmath25 is some small constant , and we will not touch the case @xmath26 . [ [ similarities - and - differences - to - the - classical - euclidean - tsp ] ] _ similarities and differences to the classical euclidean tsp _ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the classical euclidean tsp is np - hard even in two dimensions , but it is relatively easy to approximate . in particular , it admits a polynomial - time approximation scheme : given a parameter @xmath27 and a set of @xmath18 points in @xmath19-dimensional euclidean space , one can find in @xmath28 time a tour whose length is at most @xmath29 times the optimal length @xcite .
a crucial property of the euclidean tsp is that the underlying euclidean distances satisfy the triangle inequality .
the triangle inequality implies that no reasonable salesman would ever revisit the same city : instead of returning to a city , it is always cheaper to skip the city and to travel directly to the successor city .
all positive approximation results for the euclidean tsp rely heavily on the triangle inequality . in strong contrast to this , for exponents @xmath15 the distance function @xmath21 does not satisfy the triangle inequality .
thus the combinatorial structure of the problem changes significantly for example , revisits may suddenly become helpful and the existing approximation algorithms for euclidean tsp can not be applied .
another nice property of the classical euclidean problem tsp@xmath23 is that , sloppily speaking , instances with many cities have long optimal tours .
consider for instance a set @xmath0 of @xmath18 points in the unit square .
then there exists a tour whose euclidean length is bounded by @xmath30 @xcite .
this bound is essentially tight since there are point sets for which _ every _ tour has euclidean length @xmath31 .
interestingly , these results do not carry over to tsp@xmath32 with _ squared _ euclidean distances .
problem # 124 in the book by bollobs @xcite shows that there always exists a tour for @xmath0 such that the sum of the squared euclidean distances is bounded by @xmath33 , and that this bound of @xmath33 is best possible .
since , as a rule of thumb , large objective values are easier to approximate than small objective values , this already indicates a substantial difference in the approximability behaviors of tsp(2,1 ) and tsp(2,2 ) .
[ [ previous - work - and - our - results ] ] _ previous work and our results _ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + funke et al .
@xcite note that the distance function @xmath21 satisfies the so - called _
@xmath34-relaxed triangle inequality _ with parameter @xmath35 ( see section [ sec : tspalpha ] for a definition ) .
the classical tsp under the @xmath34-relaxed triangle inequality has been extensively studied @xcite , and all the corresponding machinery from the literature can be applied directly to tsp(@xmath8 ) .
for instance , andreae @xcite derives a @xmath36-approximation for the classical tsp under the @xmath34-relaxed triangle inequality ( @xmath37-tsp , for short ) .
this result translates into a @xmath38-approximation for tsp@xmath39 . for @xmath40 ,
it is better to apply bender and chekuri s @xmath41-approximation @xcite for @xmath37-tsp , which yields a @xmath42-approximation for tsp@xmath39 .
funke et al .
derive a @xmath43-approximation algorithm for tsp@xmath39 , which for the range @xmath44 is better than applying the known results @xcite .
the best result for @xmath45 is obtained by bckenhauer et al .
@xcite whose christofides - based @xmath46-approximation for @xmath37-tsp yields a @xmath47-approximation for tsp@xmath39 .
we will demonstrate in section [ sec : tspalpha ] that essentially _ every _ variant of the original @xmath48-algorithm by andreae and bandelt @xcite already gives a @xmath49-approximation for tsp@xmath22 .
the bottom - line of all this , and the actual starting point of our paper , is that the machinery around the @xmath34-relaxed triangle inequality only yields a bound of roughly @xmath50 .
this raises the following questions : how much can geometry help us in getting even better approximation ratios ?
can we beat the 6-approximation for tsp@xmath32 of funke et al . ?
we answer these questions affirmatively : we develop a new variant of the @xmath48-algorithm which we call the _ geometric @xmath48-algorithm_. an intricate analysis in section [ sec : two ] shows that this yields a 5-approximation for tsp@xmath32 .
we then extend our analysis to tsp@xmath12 with @xmath51 , and thus obtain a @xmath52-approximation ; see section [ sec : greatertwo ] .
this new bound is always better than the bound @xmath50 of funke et al . and of our analysis of the @xmath48-algorithm .
finally , in section [ sec : revtsp ] , we turn our attention to the following two questions : ( a ) how does the approximability of tspbehave when we make @xmath25 larger than one ? ( b ) does allowing revisits change the complexity or the approximability of the problem ? as we know , classical euclidean tsp ( that is , tsp@xmath53 ) is np - hard @xcite and has a polynomial - time approximation scheme ( ptas ) in any fixed number @xmath19 of dimensions @xcite . on the other hand ,
rev - tsp@xmath22 has to the best of our knowledge not been studied before . concerning question
( b ) , complexity behaves as expected : rev - tsp@xmath22 is np - hard for any @xmath16 and any , and our ( straightforward ) hardness argument also works for tsp@xmath22 . in terms of approximability , we show that whereas the two - dimensional problem rev - tsp@xmath12 still has a ptas for all values @xmath11 , the problem becomes apx - hard for all @xmath15 in three dimensions .
we were surprised that the apx - hardness proof , too , carried over to tsp@xmath54 for all @xmath15 .
this inapproximability result stands in strong contrast to the behavior of the classical euclidean tsp ( the case @xmath55 ) .
[ [ the - connection - to - wireless - networks ] ] _ the connection to wireless networks _ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider a wireless network whose nodes are equipped with omni - directional antennas .
the nodes are modeled as points in the plane , and every node can communicate with all other nodes that are within its transmission radius .
the power ( that is , the energy ) needed to achieve a transmission radius of @xmath56 is roughly proportional to @xmath57 for some real parameter @xmath25 called the _ distance - power gradient_. depending on environmental conditions , @xmath25 typically is in the range 2 to 6 ( * ? ? ?
* chapter 1 ) .
the goal is to assign powers to the nodes such that the resulting network has certain desirable properties , while the overall power consumption is minimized .
a widely studied variant has the objective to make the resulting network strongly connected @xcite .
other variants ( finding broadcast trees ; having small hop diameter ; etc ) have been studied as well .
funke et al .
@xcite suggest that it is useful to have a tour through the network , which can be used to pass a _ virtual token _ around .
the resulting power - assignment problem is tsp@xmath12 .
another setting related to tsp@xmath12 is the following . instead of omni - directional antennas ,
some wireless networks use directional antennas .
this achieves the same transmission radius under a smaller energy consumption @xcite . to model directional antennas ,
caragiannis et al .
@xcite assume that a node can communicate with other nodes in a circular sector of a given angle ( where the sector s radius is still determined by the power of the node s signal ) .
for directional antennas one not only has to assign a power level to each node , but also has to decide on the direction in which each node transmits .
if the opening angle tends to zero and the points are in general position , a strongly connected network becomes a tour .
hence , our results on tsp(@xmath58 ) may shed some light on the difficulty of power assignment for directional antennas with small opening angles .
in this section we lay the basis for our main contribution , a 5-approximation for tsp@xmath32 in section [ sec : two ] .
we review known algorithms for a related version of tsp , which can be applied to our setting . as it turns out
, these algorithms already yield the same worst - case bounds as the algorithm that funke et al .
@xcite gave recently .
we recall some definitions .
let @xmath59 be a set , let @xmath60 be a distance function on @xmath59 , and let @xmath61 .
we say that @xmath62 fulfills the _ @xmath34-relaxed triangle inequality _ if any three elements @xmath63 satisfy @xmath64 . recall that we denote by @xmath37-tsp the tsp problem on complete graphs whose weight function ( when viewed as a distance function on the vertices ) fulfills the @xmath34-relaxed triangle inequality .
the following lemma , which has been observed by funke et al .
@xcite , allows us to apply algorithms for @xmath37-tsp to our problem .
the proof relies on hlder s inequality .
[ lem : relaxed ] let @xmath17 be a fixed constant .
the distance function @xmath65 fulfills the @xmath34-relaxed triangle inequality for @xmath35 .
andreae and bandelt @xcite gave an approximation algorithm for @xmath37-tsp .
their @xmath48-algorithm is an adaptation of the well - known double - spanning - tree heuristic for tsp .
this heuristic finds a minimum spanning tree ( mst ) in the given graph @xmath66 , doubles all edges , finds an euler tour in the resulting multigraph , and finally constructs a hamiltonian cycle from the euler tour by skipping all nodes that have already been visited .
the weight of the mst is a lower bound for the length of a tsp - tour since removing any edge from a tour yields a spanning tree whose weight is at least the weight of the mst .
note that this statement holds for arbitrary weight functions .
if the triangle inequality holds , the heuristic yields a 2-approximation since then skipping over visited nodes never increases the length of the tour , which initially equals twice the weight of the mst . for the weight function @xmath21 , however , the heuristic can perform arbitrarily badly consider a sequence of @xmath18 equally - spaced points on a line .
the @xmath48-algorithm of andreae and bandelt also creates a hamiltonian tour by shortcutting the mst , but their algorithm never skips more than two consecutive nodes .
it is never necessary to skip more than two consecutive nodes because the cube @xmath67 of a tree @xmath68 is always hamiltonian by a result of sekanina @xcite .
recall that the cube of a graph @xmath66 contains an edge @xmath69 if there is a path from @xmath70 to @xmath71 in @xmath66 that uses at most three edges .
the proof of sekanina is constructive ; andreae and bandelt use it to construct a tour in @xmath72 .
the recursive procedure of sekanina @xcite to obtain a hamiltonian cycle in @xmath67 intuitively works as illustrated in fig .
[ fig : intuition_t3 ] ; for the pseudo - code , see algorithm [ alg : cycleincube ] .
the algorithm is applied to a tree @xmath68 and an edge @xmath73 of @xmath68 . removing the edge
@xmath74 splits the tree into two components @xmath75 and @xmath76 . in each component
@xmath77 ( @xmath78 ) , the algorithm selects an arbitrary edge @xmath79 incident to @xmath80 and recursively computes a hamiltonian cycle of @xmath77 that includes the edge @xmath81 .
the algorithm returns a hamiltonian cycle of @xmath68 that includes @xmath74 .
the cycle consists of the cycles in @xmath75 and @xmath76 without the edges @xmath82 and @xmath83 , respectively .
the two resulting paths are stitched together with the help of @xmath74 and the new edge @xmath84 . ,
@xmath73 ) ] note that different choices of the edge @xmath81 in line [ lin : pick ] give rise to different versions of the algorithm .
the standard @xmath48-algorithm takes an arbitrary such edge , while andreae s refined version @xcite makes a specific choice , which gives a better result .
( in the next section we will choose @xmath81 based on the local geometry of the mst , which will lead to an improved result for our problem . )
andreae s tour in mst@xmath85 has weight at most @xmath86 times the weight of the mst , which is worst - case optimal @xcite . combining his result with lemma [ lem : relaxed ] yields that the refined @xmath48-algorithm is a @xmath87-approximation for tsp@xmath39 .
we now improve on this with the help of a simple argument .
we will frequently use the following definition .
let @xmath68 be a tree and let @xmath88 be a simple path in @xmath68 .
then we call @xmath89 a
_ @xmath90-shortcut _ of @xmath68 .
we say that a shortcut @xmath91 _ uses _ an edge @xmath74 if @xmath74 lies on the path connecting @xmath71 and @xmath92 in @xmath68 .
it is not hard to see that the weight of a @xmath90-shortcut can be bounded as follows .
[ lem : nshortcut ] let @xmath2 and let @xmath74 be a @xmath90-shortcut using edges @xmath93 . then @xmath94 given a tree @xmath68 , the tour constructed by the @xmath48-algorithm consists of edges of @xmath68 and 2- and 3-shortcuts that use edges of @xmath68 .
note that in this tour each edge of @xmath68 is used exactly twice .
thus , for @xmath95 , the original @xmath48-algorithm does actually better than the bound we obtained above for the refined @xmath48-algorithm .
[ cor : t3 ] every version of the @xmath48-algorithm is a @xmath96-approximation for tsp@xmath39 .
note that our improved analysis of the @xmath48-algorithm yields the same result as the algorithm of funke et al .
@xcite .
bender and chekuri @xcite designed a @xmath41-approximation for @xmath37-tsp using a different lower bound : the optimal tsp tour is a biconnected subgraph of the original graph .
the weight of the optimal tsp tour is at least that of the minimum - weight biconnected subgraph .
the latter is np - hard to compute @xcite , but can be approximated within a factor of 2 @xcite .
moreover , the square of a biconnected subgraph is always hamiltonian . thus using only edges of the biconnected subgraph and _ two_-shortcuts
yields a @xmath41-approximation for @xmath37-tsp . combining the result of bender and chekuri with lemma [ lem : relaxed ]
immediately yields the following result , which is better than corollary [ cor : t3 ] for @xmath97 .
[ cor : bc ] the algorithm of bender and chekuri is a @xmath42-approximation for tsp@xmath39 .
in the previous section we have used graph - theoretic arguments to determine the performance of the @xmath48-algorithm . by corollary [ cor : t3 ] ,
the @xmath48-algorithm yields a 6-approximation for @xmath98 , independently of the dimension of the underlying euclidean space .
we now define what we call the _ geometric _ @xmath48-algorithm and show that it yields a 5-approximation for tsp@xmath32 .
the geometric @xmath48-algorithm simply chooses in line [ lin : pick ] of algorithm [ alg : cycleincube ] the edge @xmath81 that makes the smallest angle with the edge @xmath74 .
the idea behind taking advantage of geometry is as follows . in corollary [ cor : t3 ]
we have exploited the fact that each edge is used in two @xmath99-shortcuts .
the weight of a 3-shortcut is maximum if the corresponding points lie on a line . for the case of the euclidean mst it is well - known that edges make an angle of at least @xmath100 if they share an endpoint .
the same proof also works for the mst w.r.t .
this guarantees that in line [ lin : pick ] of algorithm [ alg : cycleincube ] , we can pick an edge @xmath81 that makes a relatively small angle with @xmath74if the degree of @xmath80 is larger than 2 . otherwise , it is easy to see that @xmath81 is used by a ( @xmath101)- and a ( @xmath102)-shortcut , which is favorable to being used by two 3-shortcuts , see lemma [ lem : nshortcut ] .
although the intuition behind our geometric @xmath48-algorithm is clear , its analysis turns out to be non - trivial .
we start with the following lemma that can be proved with some elementary trigonometry .
given two line segments @xmath103 and @xmath104 incident to the same point , we denote the smaller angle between @xmath103 and @xmath104 by @xmath105 and define @xmath106 .
[ lem:3-shortcut ] given a tree @xmath68 , the 3-shortcut @xmath107 that uses the edges @xmath108 , @xmath109 , @xmath110 of @xmath68 in this order has weight @xmath111 where @xmath112 if @xmath108 and @xmath110 lie on the same side of the line through @xmath109 , and @xmath113 if @xmath108 and @xmath110 lie on opposite sides .
moreover , @xmath114 .
lemma [ lem:3-shortcut ] ( illustrated in fig .
[ fig : cost3shortcut ] ) expresses the weight of a 3-shortcut in terms of the lengths of the edges and the angles between them .
now we show that if an edge @xmath108 is used in two 3-shortcuts , two of these angles are related .
note that the @xmath48-algorithm generates the two 3-shortcuts that use @xmath108 in two consecutive recursive calls , see fig .
[ fig : two3shortcuts_renumbered ] .
the @xmath48-algorithm is first applied to edge @xmath109 and then recursively to edge @xmath108 . in the recursive call , the shortcut @xmath115 is generated where @xmath19 is an edge incident to both @xmath108 and @xmath109 .
then the algorithm returns from the recursion and generates the 3-shortcut @xmath107 .
thus @xmath108 is the middle edge in one 3-shortcut and the first or last edge in the other 3-shortcut .
we rely on the following .
[ lem : related_angles ] if the geometric @xmath48-algorithm generates the two 3-shortcuts @xmath107 and @xmath115 in two recursive calls and @xmath19 is incident to both @xmath108 and @xmath109 , then @xmath116 .
now we are ready to prove the main result of this section .
[ thm : t3_5-app ] the geometric @xmath48-algorithm yields a 5-approximation for tsp@xmath32 .
we express the length of each shortcut @xmath103 of the @xmath48-tour in terms of the lengths of the mst edges that @xmath103 uses . changing the perspective , for each mst edge @xmath108
, we use contrib@xmath117 to denote the sum of all terms that contain the factor @xmath118 .
the edge @xmath108 is used in at most two shortcuts . bounding their lengths
yields an upper bound on contrib@xmath117 .
the sum of all contributions relates the length of the @xmath48-tour to that of the mst ( w.r.t .
@xmath21 ) , which in turn is a lower bound for the length of an optimal tsp tour . due to lemma [ lem : nshortcut ] ,
contrib@xmath119 if @xmath108 is used in a ( @xmath1202)-shortcut on one side and a ( @xmath1203)-shortcut on the other side .
so we focus on the case that @xmath108 is used in two 3-shortcuts , see fig .
[ fig : two3shortcuts_renumbered ] .
we rewrite the composite terms in the bound for @xmath107 in lemma [ lem:3-shortcut ] using young s inequality with @xmath121 , which , given @xmath122 and @xmath27 , states that @xmath123 .
let @xmath71 be the vertex that is incident to edges @xmath108 and @xmath109 .
if there are multiple 3-shortcuts that use edges that are incident to @xmath71 then the @xmath48-algorithm generates these in consecutive recursive calls .
we renumber the edges incident to @xmath71 such that the algorithm is first applied to @xmath124 , then recursively to @xmath125 etc .
then there is some @xmath126 such that @xmath127 and @xmath128 because the algorithm is first applied to @xmath109 and then recursively to @xmath108 .
we define @xmath129 .
we rewrite the term @xmath130 in the bound for @xmath131 in lemma [ lem:3-shortcut ] as follows . @xmath132 & f(|vv_{i+1}| , |vv_i| , \psi_i ) = \begin{cases } 0 & \text{if $ \psi_i { \geqslant}\frac{\pi}{2}$,}\\ |vv_i|^2 + |vv_{i+1}|^2\cos^2\psi_i & \text{if $ \psi_i < \frac{\pi}{2}$ and}\\ & \text{$\left(i=1\text { or } \left(i>1\text { and } \psi_{i-1 } { \geqslant}\frac{\pi}{2}\right)\right)$,}\\ \left(|vv_i|^2 + |vv_{i+1}|^2\right)\cos\psi_i & \text{if $ \psi_i < \frac{\pi}{2}$ and $ i>1 $ and $ \psi_{i-1 } < \frac{\pi}{2}$.}\notag\\ \end{cases } \end{aligned}\ ] ] the second case of inequality ( [ ineq : crossterms ] ) follows from young s inequality with @xmath133 and the third case from young s inequality with @xmath134 .
replacing @xmath135 in the bound for @xmath131 in lemma [ lem:3-shortcut ] is analogous .
together , the two replacements yield the bound @xmath136 we use ( [ ineq : abc_with_f ] ) to bound the weights of all 3-shortcuts .
the weight of the final tour is the sum of the weights of all shortcuts . in this sum
we can take the two occurrences of an edge @xmath137 together and analyze the contribution of @xmath108 to the tour .
note that the result of ( [ ineq : abc_with_f ] ) is still at most @xmath138 .
so if an edge @xmath108 is used in a ( @xmath1203)-shortcut on one side and a ( @xmath101)-shortcut on the other side , then we still have that contrib@xmath119 .
it remains to consider the case that @xmath108 is used in two 3-shortcuts .
let @xmath107 and @xmath115 be these 3-shortcuts .
the algorithm is first applied to edge @xmath109 and generates shortcut @xmath107 , where @xmath108 is the first or the third edge of the shortcut .
then the algorithm is recursively applied to edge @xmath108 and generates shortcut @xmath115 , where @xmath108 is the middle edge .
[ fig : two3shortcuts_renumbered ] shows how the vertices are numbered in this case .
let @xmath139 be a function that takes a sum of terms and returns the sum of all terms that contain @xmath118 .
we derive the following expression for contrib@xmath117 . @xmath140 by definition of @xmath141 we have to consider three cases in ( [ ineq : contriba ] ) for contrib@xmath117 .
* case i : * @xmath142 or @xmath143 .
we assume w.l.o.g . that @xmath142 .
then we know that @xmath144 and in the worst case @xmath145 .
thus we have that contrib@xmath146 * case ii : * @xmath147 and @xmath148 and ( @xmath149 or ( @xmath150 and @xmath151 ) ) . by definition of @xmath141
we have : @xmath152 \sigma_a(f(|vv_{i+2}|,|vv_{i+1}|,\psi_{i+1 } ) ) & = & \sigma_a\left((|vv_{i+1}|^2 + |vv_{i+2}|^2)\cos\psi_{i+1}\right ) & = & |a|^2\cos\psi_{i+1 } \end{array}\ ] ] lemma [ lem : related_angles ] states that @xmath153 .
we also know that @xmath154 by definition .
thus we have @xmath155 * case iii : * @xmath147 and @xmath156 and @xmath157 and @xmath158 .
it can be shown that this leads to a contradiction , see fig .
[ fig : three3shortcuts ] ( on page ) .
depicts the situation .
we use @xmath159 as shorthand for @xmath160 . note that @xmath161 . recall that the geometric @xmath48-algorithm proceeds with the edge that makes the smallest angle with the current edge .
the algorithm is first applied to edge @xmath162 and proceeds recursively with @xmath163 .
so we have that @xmath164 for some @xmath165 .
since @xmath158 we have @xmath166 and thus also @xmath167 .
this leaves little room for the other angles : the smallest angle between @xmath163 and @xmath168 is at most @xmath169 .
this means that at least one of @xmath159 and @xmath170 is at most @xmath171 and thus at least one of @xmath172 and @xmath173 is greater than @xmath171 , which contradicts the assumptions of case iii .
in cases i and ii , the contribution of any edge @xmath118 to the tour is at most @xmath174 .
the theorem follows by summing up the contributions of all edges . when using the mst as a lower bound in the analysis , there is not much room for improvement .
there are instances of tsp(2,2 ) where the @xmath48-algorithm yields a tour whose weight is @xmath175 times that of the mst ; see also ( * ? ? ?
* theorem 4.19 ) .
in this section we generalize the main result of the previous section to @xmath95 .
our new bound is always better than the bound @xmath176 of funke et al .
@xcite , see also corollary [ cor : t3 ] . for @xmath177
our bound is better than the bound @xmath42 that follows from the algorithm of bender and chekuri @xcite , see corollary [ cor : bc ] .
[ thm : t3_alpha ] the geometric t@xmath178-algorithm yields a @xmath179-approximation for tsp@xmath12 if @xmath95 . if an edge @xmath108 is used in a ( @xmath1202)-shortcut on one side and a ( @xmath1203)-shortcut on the other side then the total contribution of @xmath108 to the tour is at most @xmath180 by lemma [ lem : nshortcut ] .
so we will focus our analysis again on the case that @xmath108 is used in two 3-shortcuts .
for @xmath98 we can express the weight of a 3-shortcut by lemma [ lem:3-shortcut ] and rewrite the composite terms as in inequality ( [ ineq : crossterms ] ) .
for @xmath51 we apply hlder s inequality .
@xmath181\notag \end{aligned}\ ] ] we introduced the constants of type @xmath182 to shorten the expression .
note that the last inequality holds only if @xmath51 . in order to bound the contribution of an edge @xmath108 that is used in two 3-shortcuts
we follow the proof of theorem [ thm : t3_5-app ] .
since the assumptions of case iii in that proof led to a contradiction , it suffices to consider cases i and ii .
* case i * : @xmath142 or @xmath143 .
@xmath183 * case ii * : @xmath147 and @xmath148 and ( @xmath149 or ( @xmath150 and @xmath151 ) ) .
@xmath184 now we use the fact that the function @xmath185 \rightarrow { \ensuremath{\mathbb{r } } } , x \mapsto ( 2+\cos x)^k + ( 2 + \sin^2 x/2)^k$ ] attains its maximum value at @xmath186 .
thus @xmath187 also attains its maximum in the range @xmath188 in @xmath186 .
this yields @xmath189 in both cases we showed that contrib@xmath190 .
the theorem follows for @xmath51 by summing up the contributions of all edges .
the case @xmath98 corresponds to theorem [ thm : t3_5-app ] .
in this section we discuss complexity and approximability of tsp and its variant rev - tsp , where the salesman is allowed to revisit the cities .
recall that for any fixed dimension @xmath16 , tsp@xmath53 is np - hard @xcite and admits a ptas @xcite .
tsp@xmath22 and rev - tsp@xmath22 are np - hard for any @xmath16 and @xmath17 .
itai et al .
@xcite showed that , given @xmath18 points in the unit grid , it is np - hard to decide whether there is a tsp tour of euclidean length @xmath18 .
thus for both of our problems it is np - hard to distinguish between @xmath191 and @xmath192 .
tsp@xmath22 and rev - tsp@xmath22 are apx - hard for any @xmath14 and any @xmath15 .
we only discuss the case @xmath193 and @xmath98all other cases can be settled by slightly modified arguments tsp .
and we only consider rev - tsp ; a similar reduction can be used for tsp .
we reduce from @xmath194-tsp , the tsp on the complete graph where the weight of every edge is either 1 or 2 ; this problem is apx - hard @xcite .
an instance of @xmath194-tsp consists of the complete graph @xmath195 with vertex set @xmath196 , edge set @xmath197 where @xmath198 , and edge lengths that are specified by a weight function @xmath199 . given @xmath200 and @xmath92
, we construct a corresponding instance @xmath201 of rev - tsp@xmath202 .
we start our construction by introducing several auxiliary line segments . for each vertex
@xmath203 we define its _ spine _ to be the vertical line segment going from point @xmath204 to point @xmath205 . for each edge @xmath206 with @xmath207 , we define two corresponding line segments that are parallel to the @xmath208-plane and that are called _
bones_. the first bone connects point @xmath209 on the spine of @xmath210 to the point @xmath211 . the other bone connects point @xmath212 on the spine of @xmath213 to the point @xmath214 , where @xmath215 if @xmath216 and @xmath217 if @xmath218 . note that
these two bones do not quite touch ; they are separated by a gap of length @xmath219 . in order to get the instance @xmath220 of rev - tsp@xmath202
, we subdivide every single ( spine or bone ) line segment introduced above by a dense , evenly distributed set of points we call these points _ cities _ from now on so that every unit - length piece receives @xmath221 cities . the distance between adjacent cities
is @xmath222 , and so the cost for going from one city to an adjacent city is @xmath223 .
all these cities together form instance @xmath220 , and this completes our construction .
since we have introduced line segments with a total length of at most @xmath224 , the overall number of cities is at most @xmath225 .
for @xmath226 we call the cities on the spine of @xmath210 and on all bones incident to this spine the _ city cluster _ of @xmath210 . traversing all cities within such a city cluster
is very cheap ; even if we visit every city twice , this costs at most @xmath227 for all cities in all city clusters together . in a traveling salesman tour ,
the only expensive steps occur when the salesman jumps from one city cluster to another city cluster . by the above definition of @xmath219 ,
when jumping from bone to bone across the gap corresponding to edge @xmath228 the incurred cost is exactly @xmath229 .
note that jumping from city cluster to city cluster in any other way would be much more expensive and would thus not reduce the total cost of the tour . finally , let us show that our reduction is approximation preserving .
fix an @xmath121 with @xmath230 .
consider an instance @xmath200 and @xmath92 of @xmath194-tsp , and assume without loss of generality that @xmath231 .
consider an optimal tour @xmath232 for this instance .
if @xmath232 uses @xmath233 edges of length @xmath234 and @xmath235 edges of length @xmath236 , then it has cost @xmath237 . given a ptas for rev - tsp ,
we show how to compute in polynomial time a tour of cost at most @xmath238 for @xmath200 and @xmath92 .
first note that the tour @xmath232 can be transformed into a tour @xmath239 through @xmath220 that makes @xmath240 jumps of cost 2 and @xmath235 jumps of cost 1 . that tour @xmath239 costs at most @xmath241 .
using our hypothetical ptas for rev - tsp , we can compute for any @xmath242 in polynomial time a tour @xmath243 through @xmath220 of cost at most @xmath244 , where @xmath245 is the cost of an optimal rev - tsp tour . the existence of @xmath239 yields @xmath246 .
the tour @xmath243 can be transformed into a tour @xmath247 through @xmath200 : just map the jumps of @xmath243 to the corresponding edges of @xmath200 . since this mapping can not increase the cost , tour @xmath247 costs at most @xmath248 .
choosing @xmath249 and using @xmath250 , we can bound the cost of @xmath247 from above by @xmath251 as desired . like @xmath243 , the tour @xmath247 may visit vertices more than once .
this can be fixed by greedily introducing shortcuts .
the shortcuts do not increase the cost of the tour since the weight function @xmath92 ( trivially ) fulfills the triangle inequality .
[ thm : ptas ] there exists a ptas for rev - tsp@xmath12 for any @xmath11 .
given a set @xmath0 of points in the plane , consider the _
gabriel graph _
@xmath252 that has a vertex for each point in @xmath0 .
there is an edge between points @xmath6 and @xmath7 , if the open disk with diameter @xmath253 is empty , in other words , if for all points @xmath254 , the angle @xmath255 is at most @xmath171 .
the weight of the edge is @xmath20 .
note that @xmath256 if @xmath255 is at least @xmath171 .
therefore , there is an optimal tsp tour with revisits through @xmath0 that only uses the edges of @xmath252 : indeed , if a tour uses an edge @xmath253 for which there is a point @xmath56 with @xmath257 , then replacing @xmath253 by @xmath258 and @xmath259 would shorten the tour .
such a replacement is feasible since revisiting city @xmath56 is allowed .
the gabriel graph is planar .
hence we end up with an instance of the tsp on weighted planar graphs , for which a ptas is known @xcite .
recall that a _
_ quasi-__ptas is an approximation scheme with running time @xmath260 , where @xmath18 is the size of the input .
the following result follows immediately from the facts that ( a ) the metric @xmath21 has bounded doubling dimension and ( b ) tsp on metrics of bounded doubling dimension admits a quasi - ptas @xcite .
there exists a quasi - ptas for rev - tsp@xmath22 for any @xmath261 $ ] and @xmath262 .
in order to construct considerably better approximation algorithms for tsp@xmath22 , we expect that substantially different methods of analysis have to be found . a result of van nijnatten ( * ? ? ?
* theorem 4.19 ) indicates that there is not much room left for improvement as long as we compare to the mst .
the approximability of rev - tsp@xmath12 for @xmath263 is an interesting open question .
we believe that a ( quasi)-ptas may be obtained using the framework of the ptas for weighted planar graph tsp by arora et al .
a simple reduction shows that deriving a ptas for our problem is at least as hard as deriving a ptas for weighted planar graph tsp .
assume we have a ptas for rev - tsp@xmath12 for some @xmath15 . given a weighted planar graph and a planar embedding
, we replace each edge by a dense set of points such that traversing a subedge basically costs zero . by making one subedge of each edge @xmath74 longer
, we can make the cost of that subedge ( and thus of @xmath74 ) in rev - tsp proportional to the weight of @xmath74 .
then , the costs of the optimal solutions of the two problems will be the same up to an arbitrarily small constant factor of @xmath29 .
such a reduction is polynomially bounded if all weights are polynomially bounded , which can be achieved by a standard rounding scheme . a ptas for rev - tsp@xmath12 for any @xmath15 would be an interesting generalization of the existing ptas s for weighted planar graphs .
ideally , one would have a ptas with running time independent of @xmath25 since it would contain both euclidean tsp and weighted planar graph tsp as special cases .
e. althaus , g. clinescu , i. mandoiu , s. prasad , n. tchervenski , and a. zelikovsky .
power efficient range assignment for symmetric connectivity in static ad hoc wireless networks .
, 12(3):287299 , 2006 .
bckenhauer , j. hromkovi , r. klasing , s. seibert , and w. unger . towards the notion of stability of approximation for hard optimization tasks and the traveling salesman problem .
, 285(1):324 , 2002 .
i. caragiannis , c. kaklamanis , e. kranakis , d. krizanc , and a. wiese .
communication in wireless networks with directional antennas . in f. m. auf der heide and n. shavit , editors , _ proc .
20th annu .
parallel algorithms architect .
_ , p. 344351
, 2008 .
s. funke , s. laue , r. naujoks , and z. lotker .
power assignment problems in wireless communication : covering points by disks , reaching few receivers quickly , and energy - efficient travelling salesman tours . in _ proc .
ieee conf .
distributed comput .
sensor systems _ , lncs 5067 , p. 282295 | let @xmath0 be a set of points in @xmath1 , and let @xmath2 be a real number .
we define the distance between two points @xmath3 as @xmath4 , where @xmath5 denotes the standard euclidean distance between @xmath6 and @xmath7 .
we denote the traveling salesman problem under this distance function by tsp(@xmath8 ) .
we design a 5-approximation algorithm for tsp(2,2 ) and generalize this result to obtain an approximation factor of @xmath9 for @xmath10 and all @xmath11 .
we also study the variant rev - tsp of the problem where the traveling salesman is allowed to revisit points .
we present a polynomial - time approximation scheme for rev - tsp@xmath12 with @xmath11 , and we show that rev - tsp@xmath13 is apx - hard if @xmath14 and @xmath15 .
the apx - hardness proof carries over to tsp@xmath13 for the same parameter ranges .
mark de berg fred van nijnatten ren sitters gerhard j. woeginger alexander wolff |
high - redshift quasars provide direct probes of the epoch when the first generation of galaxies and quasars formed .
the absorption spectra of these quasars reveal the state of the intergalactic medium ( igm ) close to the reionization epoch ( @xcite , @xcite , @xcite , @xcite ) .
the lack of a gunn - peterson trough ( shklovsky 1964 , scheuer 1965 , gunn & peterson 1965 ) in the spectrum of the luminous quasar sdssp j104433.04012502.2ddmmss.s , where `` p '' stands for the preliminary sdss astrometry , and the positions are expressed in j2000.0 coordinates .
the astrometry is accurate to better than @xmath30 in each coordinate .
] at @xmath15 ( sdss 10440125 for brevity , @xcite , for an updated redshift of this object , see @xcite and @xcite ) indicates that the universe was already highly ionized at that redshift . assuming that sdss 1044 - 0125 is radiating at the eddington luminosity , this object contains a central black hole of several billion solar masses .
the assembly of such massive objects in a timescale shorter than 1 gyr yields constraints on models of the formation of massive black holes ( e.g. , @xcite ) .
the abundance and evolution of such quasars can provide sensitive tests for models of quasar and galaxy evolution .
the sloan digital sky survey ( sdss ; @xcite ) uses a dedicated 2.5 m telescope and a large format ccd camera ( @xcite ) at the apache point observatory in new mexico to obtain images in five broad bands ( @xmath31 , @xmath32 , @xmath33 , @xmath0 and @xmath34 , @xmath32 , @xmath33 , @xmath0 and @xmath34 .
as the sdss photometric calibration system is still being finalized , the sdss photometry presented here is referred to as @xmath35 , @xmath36 , @xmath37 , @xmath38 and @xmath39 . ] centered at 3551 , 4686 , 6166 , 7480 and 8932 , respectively ; @xcite ) over 10,000 deg@xmath2 of high galactic latitude sky . over 200 quasars at @xmath40 have been discovered to date from the sdss multicolor imaging data ( fan et al . 1999a , 2000a , c , 2001a , @xcite , schneider et al .
2000 , 2001 , @xcite ) .
the inclusion of the reddest band , @xmath34 , in principle enables the discovery of quasars up to @xmath41 from the sdss data ( fan et al .
2000c ) . encouraged by the discovery of sdss 10440125 ,
we have carried out a systematic survey of @xmath0-dropout quasars ( @xmath3 , whose lyman @xmath11 forest is entirely in the @xmath0 band ) based on the sdss imaging data .
the goals of this survey are : ( 1 ) to define a color - selected flux - limited complete sample of quasars at @xmath42 in order to study the evolution of the quasar spatial density at high redshift ; ( 2 ) to provide multiple lines of sight for studying the nature of the high - redshift igm ; ( 3 ) to study the intrinsic properties of quasars and the metallicity of the quasar environment at high redshift . in this paper
, we present the first results of this survey , covering an area of 1550 deg@xmath2 . in 2 , we present our color selection procedures for @xmath43 quasars .
special software is developed to control the number of false single - band detections in the imaging data .
follow - up optical and near - infrared photometry is used to separate high - redshift quasar candidates from the more numerous cool dwarfs which have similar optical colors . in 3 ,
we describe the discovery of three new quasars at @xmath7 , 5.99 and 6.28 .
we discuss their spectral properties and the implications for the metallicity of the quasar environment and the status of the igm in 4 .
these three quasars , combined with sdss 10440125 , form a complete color - selected sample at @xmath5 .
we calculate the survey selection function and derive the spatial density of @xmath27 quasars in 5 . in 6 , we discuss the cosmological implications of the observations .
follow - up spectroscopic observations carried out with the keck telescope will be presented by becker et al .
( 2001 ; paper ii ) .
we present our results using two different cosmologies throughout the paper : a @xmath44-dominated universe with @xmath45 km s@xmath20 mpc@xmath20 , @xmath46 and @xmath47 ( @xcite , @xcite ) , which is referred to as the @xmath44-model ; and a einstein - de sitter universe with @xmath48 and @xmath49 = 50 km s@xmath20 mpc@xmath20 , which we refer to as the @xmath48 model in this paper . the age of the universe is 13.9 gyr and 13.0 gyr in the @xmath44-model and the @xmath48 model , respectively .
the observed colors of quasars evolve strongly with redshift , as the intrinsic emission and the intervening absorption spectral features move through the sdss filter system ( @xcite , @xcite ) .
the strong ly@xmath11 forest absorption blueward of the ly@xmath11 emission line enters the sdss @xmath32 , @xmath33 and @xmath0 bands at redshifts of about 3.6 , 4.6 and 5.5 , respectively . at @xmath50 ,
the ly@xmath11 emission line begins to move out of the sdss @xmath0 filter . with an average @xmath51
, these high - redshift quasars have undetectable flux in the bluer @xmath31 , @xmath32 and @xmath33 bands , and little flux ( sometimes also undetectable ) in @xmath0 band .
they become @xmath0-dropout objects with only one measurable color in sdss photometry .
figure 1 presents the @xmath52 vs. @xmath39 color - magnitude diagram for 50,000 stellar sources selected at random from the sdss imaging data .
it also shows the locations of all the @xmath0-dropout objects ( including l / t dwarfs and high - redshift quasars , see below ) selected in the 1550 deg@xmath2 survey area .
the median track of simulated quasars ( see 5.1 ) with absolute magnitude @xmath53 , representative of the quasars discussed in this paper , is also plotted .
a simple cut at @xmath54 selects quasar candidates at @xmath43 . at @xmath55 ,
the ly@xmath11 forest absorption enters the @xmath34 filter , and the @xmath39 flux decreases rapidly with redshift with increasing amount of absorption , while @xmath52 remains approximately constant , as the increasing absorption in the @xmath0 and @xmath34 bands are roughly equal . at @xmath56 ,
the ly@xmath11 emission line has redshifted to 9240 , well past the effective wavelength of the @xmath34 filter .
most of the @xmath34 flux is absorbed by the ly@xmath11 forest , and the object disappears from optical images altogether .
this is the upper limit to the redshift of objects detectable with sdss imaging data .
quasars at @xmath43 are extremely rare on the sky .
we found four @xmath0-dropout quasars in 1550 deg@xmath2 .
for comparison , over the same area , there are about 15 million objects detected in the @xmath34 band above 6-@xmath57 , and about 6.5 million cosmic ray hits in the @xmath34 band .
the quasars are faint , with low signal - to - noise ratio ( s / n ) photometry from the sdss imaging data .
therefore , the key to a successful selection process is an efficient elimination of potential contaminants .
we face three technical challenges : \(1 ) elimination of false @xmath34-band only detections , composed primarily of cosmic rays , but also including satellite trails , electronic ghost images and bleed trails from bright stars .
while the recognition of the majority of cosmic rays is relatively straightforward , even a tiny fraction of mis - classified cosmic rays in the @xmath34 band , which would then appear as @xmath0-dropout candidates , would dominate over the much rarer true quasars .
note that sdss scans the sky only once in the course of the survey except in a small fraction of overlapping area ; thus repeat observations can not be used to recognize cosmic rays . moreover ,
sdss uses thick ccds for the @xmath34 band .
a cosmic ray hit on these thick devices typically occupies more than a single pixel , making the separation of real detections from cosmic ray hits more difficult , especially when the seeing is good .
\(2 ) reliability of @xmath39 photometry .
as is evident from figure 1 , the @xmath0-dropout objects are distributed on the extreme tail of the @xmath52 distribution . under typical
seeing , @xmath58 for objects with @xmath59 .
objects in the tail of the error distribution could scatter into the selected region of color space , and pose a serious contamination when a large number of objects is searched .
( 3 ) separating quasars and cool dwarfs . as shown in 3 , the surface density of cool dwarfs , with spectral types ranging from mid - l to t , which have @xmath54 ( geballe et al .
2001 , leggett et al .
2001 ) , is @xmath60 times higher than that of @xmath43 quasars .
we use @xmath6-band photometry to separate these two classes of objects ( see below , also fan et al . 2000b , zheng et al .
2000 ) . because of the rarity of high - redshift quasars , and the overwhelming number of contaminants , our photometric selection procedure of @xmath43 quasar candidates is quite involved , and includes five separate steps : \1 .
selection of @xmath0-dropout sources from the sdss database .
\2 . running an improved cosmic ray classifier and visually inspecting all remaining candidates to reduce the number of false candidates due to cosmic rays .
\3 . matching the candidate list to the two - micron all sky survey ( 2mass , @xcite ) in the public release area of 2mass , to obtain @xmath61 photometry .
independent @xmath34 photometry to further eliminate false and biased @xmath34 detections .
@xmath6-band photometry of objects not in 2mass to separate quasar and cool dwarf candidates .
this then leaves us with a sample for follow - up spectroscopy . in this and the subsequent section
, we describe this process in detail .
the techniques and rationale for these various cuts are described in the present section , while the observations themselves are described in
[ sec : observations ] .
the sdss photometric data are processed by a series of automated pipelines to carry out astrometric and photometric measurements .
the photometric pipeline ( _ photo _ , @xcite ) reduces the data from the imaging camera and produces corrected images and object catalogs .
_ photo _ classifies detected objects either as `` star '' ( point source , consistent with the point spread function , psf ) or `` galaxy '' ( extended source ) .
however , we select all @xmath0-dropout objects irrespective of their star - galaxy classification , as the star - galaxy separation becomes less reliable at low s / n .
cosmic rays are recognized as objects with profile gradients significantly steeper than that of the psf , and are interpolated over ; they are not output explicitly by the pipeline . while _
photo _ classifies more than 99.9% of the cosmic rays correctly , visual inspection shows that a majority of objects with @xmath34-only detections that are classified as `` stars '' still have the appearance of cosmic rays .
this is further confirmed by comparing those areas of the sky with multiple observations .
therefore , we have recently improved the performance of the object classifier , as follows . for every object detected by the imaging pipeline
, we count the pixels in an @xmath62 pixel region ( @xmath63 ) centered on the object , for which @xmath64 \over i(0 ) - c*n[i(0)]}\end{aligned}\ ] ] where @xmath65 is the object s central intensity , @xmath66 is the intensity of the pixel in question , psf(0 ) and psf(@xmath67 ) are the equivalent quantities for the locally - determined model of the psf , the noise @xmath68 is the photon noise with an additional @xmath69 added in quadrature , and @xmath70 .
this criterion searches for objects with profile gradients significantly sharper than that of the psf .
if more than 3 pixels satisfy this condition , we consider the object to be a candidate cosmic ray , and reject it .
the photometric pipeline assigns detailed flags for each detected object , indicating objects whose photometry ( and therefore , colors ) may be in error .
we reject objects flagged as being saturated in any band , lying on the bleed trail of a saturated star , or overlapping the edge of the image boundary .
we pay close attention to objects blended with close neighbors , and reject deblended children with peakcenter , notchecked , or deblend_nopeak set in the @xmath0 and @xmath34 bands , which are typical signs of deblending problems . for a complete description of the definition of flags ,
see @xcite .
the flag - checking procedure used here is similar to that used in the sdss quasar target selection pipeline ( @xcite ) .
we have re - classified all the objects with @xmath54 using the new cosmic ray classifier . we found that depending on the seeing of a specific run , @xmath71 to @xmath72 of the @xmath0-dropout objects formerly classified as `` stars '' are now classified as cosmic rays , while a negligible fraction of the `` real '' objects ( based on multiple observations and objects with confirmed spectra ) are mis - classified as `` cosmic rays '' .
we then visually inspect the @xmath34 images of the remaining candidates , and reject a further 25% as cosmic rays and other artifacts .
after the improved cosmic ray rejection and visual inspection , we carry out independent follow - up @xmath34 photometry of all the remaining candidates ( of which there are 121 ; see below ) using the apache point 3.5 m and other telescopes ( see 3.1 ) .
this serves two purposes : ( a ) to confirm that the objects are indeed real ; and ( b ) to measure the @xmath39 magnitude more accurately .
since our @xmath0-dropout candidates have the reddest @xmath52 colors , we tend to pick up objects lying on the several @xmath57 tail of the @xmath39 error distribution ( note that the @xmath34-band only objects are very faint , and therefore are measured at low s / n ) ; they will thus have artificially red @xmath52 color and blue @xmath73 color , and are more likely to be selected as @xmath43 quasar candidates ( see below ) . to include those objects in the sample would lower our spectroscopic success rate considerably . as discussed by fan et al ( 2000b ) and zheng et al .
( 2000 ) , and illustrated in figure 1 , the major astrophysical contaminants of @xmath0-dropout quasars are very cool dwarfs with spectral types l and t ( strauss et al . 1999 ,
tsvetanov et al
. 2000 , fan et al .
2000a , leggett et al .
2000 , 2001 , geballe et al .
these have effective temperatures @xmath74 , and @xmath75 , and can not be distinguished from @xmath43 quasars using sdss optical photometry alone .
figure 2 presents the @xmath52 vs. @xmath73 color - color diagram of the @xmath0-dropout sample .
finlator et al .
( 2000 ) have matched the sdss photometric catalog in @xmath76 deg@xmath2 with that of 2mass in the @xmath61 bands .
the small dots in figure 2 show the sdss-2mass matches and indicate the location of normal stars on this color - color diagram .
the track of quasar colors shows that by using @xmath77 , @xmath0-dropout cool dwarfs and high - redshift quasars are clearly separated , as the @xmath78 color of high - redshift quasars is dominated by the blue power law continuum , while the cool temperature of l / t dwarfs results in very red @xmath79 color , typically larger than 2 .
therefore we determine whether the candidate has been detected by 2mass .
while the non - detection of a faint candidate by 2mass does not give a tight constraint on the spectral shape of the object , most of the brighter objects ( @xmath80 ) that are located in the area of the 2mass publicly released data are indeed detected by 2mass , with @xmath81 , indicating that most of them are cool dwarfs .
for those not matched with 2mass , we carry out follow - up @xmath6-band photometry ( see 3.1 ) .
the final photometric selection criteria for this survey are : @xmath82 we also require that the object not be detected in any of the other three bands @xmath83 and @xmath33 . the magnitudes and colors in eq .
( 1 ) are _ not _ corrected for interstellar extinction .
this is a small effect , which has been taken into account in the selection function calculation ( 5.1 ) . in the selection ,
only objects with s / n @xmath84 in the @xmath34 band are considered ( item [ b ] ) , as the @xmath85 color becomes very unreliable for the faintest sources and would have selected many more candidates with a more inclusive cut .
item ( c ) selects quasars with @xmath3 , and item ( d ) separates quasars from cool dwarfs .
note that the sdss photometry is on the ab magnitude system ( fukugita et al .
1995 ) , while the @xmath6-band photometry is on the vega - based system .
the selected area is illustrated as the shaded region in figure 2 .
table 1 lists the details of the 25 sdss photometric runs used in this survey .
the projection of these runs on the sky is illustrated in figure 3 .
the entire region is located in the northern galactic cap . in table 1 ,
the location of each run is presented in terms of survey stripe number ( @xmath86 ) and the range of survey great circle longitude ( @xmath87 and @xmath88 ) . when taking data , the sdss telescope moves along a series of great circles on the sky and the photometric camera drift - scans at the sidereal rate , with an effective exposure time of 54.1 seconds .
the survey coordinate system ( @xmath89 , @xmath90 ) is a spherical system with poles at @xmath91 , and @xmath92 , @xmath93 .
@xmath94 , @xmath95 is located at @xmath96 , @xmath97 ( york et al .
2000 , @xcite ) .
each drift scan tracks a survey stripe @xmath86 centered on a constant @xmath90 , given by @xmath98 two scans , or strips , one offset to the north and one to the south , are required to fill a stripe .
the 25 sdss imaging runs used in the @xmath0-dropout survey in this paper were taken between march 20 , 1999 ( run 745 ) and march 19 , 2001 ( run 2190 ) over a range of seeing conditions . the seeing of sdss images is characterized by the psfwidth parameter , defined as the effective width of the best fit double - gaussian psf model . for a single gaussian profile , psfwidth = 1.06 fwhm . only those runs , or parts of runs , with psfwidth in @xmath34 band in the 4th column of the sdss camera ( near the middle of the focal plane ) better than @xmath99
are used in this paper .
the median seeing values ( over all six columns of the camera ) in the @xmath0 and @xmath34 bands are also listed in table 1 . due to the complicated geometry of the survey and overlap between runs
, we estimated the total survey area by monte - carlo integration calculating the fraction of randomly distributed points on the celestial sphere falling into the boundary of any of the runs listed in table 1 .
the total area covered is 1550 deg@xmath2 , excluding regions with poor seeing .
the median psfwidth is @xmath100 in the @xmath0 and @xmath34 bands over the whole area . following the procedures described in 2 , we first search the sdss database and eliminate false detections ( steps 1 and 2 ) based on the sdss images .
this results in a total of 121 objects with @xmath54 that require further @xmath34 and @xmath6 band photometry .
the candidate list includes five known t dwarfs from the sdss ( strauss et al . 1999 ,
tsvetanov et al
. 2000 , leggett et al .
2000 ) , one known t dwarf from 2mass ( burgasser et al .
1999 ) , and the known @xmath15 quasar sdss 10440125 .
thirty - five objects are located in the 2mass first and second incremental data release area .
of these , 13 objects are detected in 2mass @xmath61 photometry and have colors of l to t dwarfs .
independent @xmath34 photometry ( step 4 in 2.2 ) was carried out using the seaver prototype imaging camera ( spicam ) in the sdss @xmath34 filter on the arc 3.5 m telescope at the apache point observatory in several nights between february and april 2001 .
spicam has a backside illuminated site @xmath101 pixel ccd with a field of view of @xmath102 .
the pixel scale is @xmath103 , and the typical seeing was @xmath104 .
the quantum efficiency peaks at 89% at 6500 , and is about 49% at 9000 .
the exposure times were between 90 and 180 seconds , depending on the weather conditions .
note that since the sdss photometry itself provides many local reference stars , photometric weather is not required for @xmath34 photometry .
the s / n is typically twice that of the sdss imaging .
the spicam images are reduced using normal iraf procedures .
sky flats are constructed from the scaled median of target images .
@xmath39 magnitudes are measured with aperture photometry with a radius of @xmath105 , and calibrated by local sdss standards .
follow - up @xmath6 band infrared photometry ( step 5 in 2.2 ) was carried out on photometric nights using several telescopes , between february and may 2001 .
the majority of the targets were observed with the grim ii instrument ( the near infrared grism spectrometer and imager ) , also on the arc 3.5 m .
it uses a @xmath106 hgcdte detector and covers a field of view of @xmath107 at f/10 .
jhk standards from persson et al .
( 1998 ) were monitored throughout the night .
we used similar iraf procedures to reduce the grim ii data .
a small number of objects were observed with the ufti instrument , a near ir camera with a @xmath108 hgcdte detector covering @xmath109 on the 3.8 m united kingdom infrared telescope ( ukirt ) on mauna kea , and with the magic instrument , a near - infrared imager and low - resolution infrared spectrograph at the 2.2 m telescope at calar alto observatory in spain .
the typical photometric error of our infrared photometry is 0.05 0.10 mag , and is dominated by calibration errors . since quasars and l / t dwarfs have @xmath73 colors that differ by more than 0.5 mag , this error has very little effect on the selection efficiency ( 5.2 ) .
we attempted to follow the sequence described in 2.2 to select @xmath0-dropout quasar candidates . but as we had to use photometric nights for the ir photometry , we changed the observing sequence in two cases : ( a ) on a dry photometric night , ir photometry had the highest priority ; ( b ) when the weather was non - photometric , for objects that had spicam @xmath34 photometry but did not yet have @xmath6 band photometry , we carried out spectroscopy directly , without pre - selection using @xmath73 color . in the latter case , we started from either the brightest or the reddest object , depending on weather .
it is worth noting that the rapid instrument change mechanism at the arc 3.5 m ( @xmath110 15 minutes ) and the excellent weather monitoring system at apo allows a very efficient use of telescope time . in the end
, we obtained photometry and/or spectroscopy of our full sample .
table 2 summarizes the classifications of the @xmath0-dropout sample . using 2mass matches and
follow - up @xmath34 and @xmath6 photometry , we found that among the 121 @xmath0-dropout candidates , 35 are false detections , most likely cosmic rays that still passed the first two steps of selection ; 71 have @xmath111 , and are classified as cool dwarfs .
ten objects are classified as cool dwarfs based on spectroscopy ; so no @xmath6 photometry was necessary .
the spectra of these dwarfs were obtained using the double imaging spectrograph ( dis ) on the arc 3.5 m between march and may 2001 , and using the echelle spectrograph and imager ( esi ) on keck ii telescope in march 2001 .
one of the candidates , sdssp j081948.97 + 420930.2 , has an intermediate @xmath73 color ( @xmath112 ) ; it is a low - ionization bal quasar at @xmath113 ( see becker et al .
1997 , hall et al .
the keck spectrum of this object is presented in hall et al .
finally , four objects have @xmath54 and @xmath77 , including the previously known @xmath15 quasar sdss 10440125 , and three new quasars at @xmath43 .
note that in this survey , _ all _ objects that satisfy the color selection criteria ( eq.[1 ] ) indeed turn out to be @xmath3 quasars from their spectra .
the magnitude and colors of these objects are plotted on figures 1 and 2 . among the cool dwarfs found in this survey , eleven are t dwarfs .
they form a complete flux - limited t dwarf sample .
six of these objects were previously known ; the optical - ir spectra of four more of them are presented in geballe et al .
( 2001 ) . in a separate paper
, we will present the photometry of l and t dwarfs in the @xmath0-dropout sample , the observations of the final t dwarf , and the analysis of the surface density and spatial densities of t dwarfs in the solar neighborhood .
the optical spectra of the three new quasars were obtained using the dis on the arc 3.5 m between march and may 2001 .
the instrument and data reduction procedure are described in detail by fan et al .
( 1999a ) .
figure 4 presents the @xmath34 band finding charts of the three new quasars , sdssp j083643.85 + 005453.3 ( @xmath7 , sdss 0836 + 0054 for brevity ) , sdssp j130608.26 + 035636.3 ( sdss 1306 + 0356 , @xmath8 ) and sdssp j103027.10 + 052455.0 ( sdss 1030 + 0524 , @xmath9 ) .
table 3 presents their photometric properties , and figure 5 shows the arc 3.5 m discovery spectra of the three objects .
the exposure time of each spectrum is 3600 seconds .
the spectra in figure 5 have been smoothed to a resolution of @xmath114 .
the flux calibration has been adjusted to match the sdss @xmath39 magnitudes .
the locations of the ly@xmath11 and nv emission lines are indicated in the figure .
the expected positions of ly@xmath115+ovi and the lyman limit are also shown .
a near - ir spectrum of sdss 1030 + 0524 was obtained on the night of 11 june 2001 , using the nirspec ( @xcite ) instrument on the keck ii telescope , as part of the gemini / nirspec service observing program , carried out by gemini staff scientists tom geballe and marianne takamiya .
observations were taken in grating mode , with the n2 blocking filter and a @xmath116 slit , which gives a wavelength coverage of 1.09 @xmath117 m 1.29 @xmath117 m , and a spectral resolution of @xmath118 .
a total of 10 exposures , each with 300 sec exposure time , were taken using an abba sequence , nodding along the slit .
the weather was photometric with @xmath119 seeing .
hip 48414 , a bright a1v star , was observed as an atmospheric calibration star at similar airmass to that of the quasar observation .
the nirspec data were reduced using the wmkonspec package , an iraf package developed at keck observatory to reduce nirspec data .
the flux calibration has been adjusted to match its measured @xmath6-band magnitude . the final reduced spectrum , binned to 4@xmath10pixel@xmath20 , is shown in figure 6 .
for comparison , the optical spectrum of sdss 1030 + 0524 is also shown , and the locations of the ly@xmath11 , nv and civ emission lines are indicated .
the discovery spectra of the three objects show unambiguous signatures of very high - redshift quasars , similar to that of sdss 10440125 : strong , broad and asymmetric ly@xmath11+nv emission lines , with sharp discontinuities to the blue side , due to the onset of very strong ly@xmath11 absorption .
the discovery spectra are of relatively low s / n .
accurate redshift determination from these spectra are difficult .
the ly@xmath11 emission line is severely affected by the ly@xmath11 forest absorption , and the nv line is blended with ly@xmath11 .
the low s / n does not allow us to use weaker lines such as oi+siii@xmath89 1302 or siiv+oiv]@xmath89 1400 to determine an accurate redshift .
the next strong emission line in the quasar spectrum is civ@xmath89 1549 , which is located beyond 1 @xmath117 m for @xmath120 , and requires near - ir spectroscopy .
therefore , we use the near ir spectrum to determine the redshift of sdss 1030 + 0524 .
the redshifts of the other two quasars were determined using high s / n keck spectra described in paper ii .
the uncertainties in all these redshifts are of order 0.02 . in this section , we first briefly comment on the major spectral signatures of each quasar ( 4.1 ) .
we use the emission lines to constrain the metallicity of the quasar environment ( 4.2 ) .
then we calculate the average absorption in the ly@xmath11 forest region , based on the discovery spectra , and comment on the possible detection of the gunn - peterson trough in the spectrum of sdss 1030 + 0524 ( 4.3 ) .
* sdssp j083643.85 + 005453.3 * ( @xmath7 ) .
this quasar has a similar redshift to sdss 10440125 .
the ly@xmath11 emission line is very broad and strong .
a separate nv@xmath891240 component is tentatively detected .
the rest frame equivalent width ( ew ) of the ly@xmath11+nv line is @xmath121 , quite typical of lower - redshift quasars ( fan et al .
the ly@xmath115+ovi emission line is also clearly detected at @xmath122 ; this line was also seen in the spectrum of sdss 1044 - 0125 .
it is evident that there is detectable flux blueward of the ly@xmath11 emission line .
the universe is still highly ionized at @xmath123 .
sdss 0836 + 0054 is an extremely luminous object . with @xmath124 and @xmath125
, it has an absolute magnitude @xmath126 ( @xmath48 model , see 5.1 ) , and is the most luminous quasar discovered at @xmath127 to date ; in particular , it is 0.5 mag more luminous than sdss 10441025 .
sdss 0836 + 0054 has a radio counterpart in the first radio survey ( @xcite ) at 20@xmath10 cm .
the radio source is unresolved with a total flux of @xmath128 mjy , and the positional match is better than @xmath104 .
this is the highest - redshift radio loud quasar known , and the second radio - loud quasar detected at @xmath129 ( the other one is sdssp j091316.56 + 591921.5 , at @xmath130 , with a 20@xmath10 cm flux of 18.1@xmath10mjy ; see anderson et al .
2001 ) .
* sdssp j130608.26 + 035636.3 * ( @xmath8 ) .
this is the second highest - redshift quasar known to date .
the ly@xmath11 emission line is centered at @xmath131 , and a separate nv component is also detected .
the rest frame equivalent width of ly@xmath11+nv is @xmath132 .
the ly@xmath11 forest is stronger than that in sdss 0836 + 0054 , but there is still detectable flux in the ly@xmath11 forest region , immediately blueward of the ly@xmath11 emission line .
the apparent absorption at @xmath133 is not real , but is due to imperfect telluric absorption correction in this noisy spectrum .
the object has @xmath134 and is detected at the 4-@xmath57 level in the sdss @xmath0 band .
it is also a luminous object , with @xmath135 in the @xmath21 model .
* sdssp j103027.10 + 052455.0 * ( @xmath9 ) .
this is the highest - redshift quasar known to date . at @xmath136 and @xmath137 , it is also the faintest quasar found in the current survey , and has @xmath138 in the @xmath21 model .
it is close to the flux limit of the survey , and is only detected at 10-@xmath57 in @xmath34 .
the object is very red , and is completely undetected in the sdss @xmath0 band .
the discovery spectrum shows very strong ly@xmath11+nv emission , with a rest - frame equivalent width of @xmath121 , and a tentative detection of a separate nv component .
the ly@xmath11 absorption is strikingly strong , with almost no detectable flux blueward of the ly@xmath11 emission .
a strong civ emission line is detected in the near ir spectrum ( figure 6 ) .
a gaussian fit to the civ line profile gives @xmath139 , with a rest frame ew = @xmath140 .
the continuum shape at 9000 @xmath141 13000 is consistent with @xmath142 , as indicated by the dashed line in figure 6 .
emission line ratios can be used to measure the metallicity of the gas in the broad emission line region ( belr ) .
there is growing evidence from those measurements that belr have roughly solar or higher metallicities even out to @xmath143 ( e.g. , @xcite ) .
we use the spectra of our quasars to estimate the quasar metallicity at @xmath27 . of various line ratios , the nv @xmath891240/civ @xmath891549 ratio and nv @xmath891240/heii @xmath891640 ratio are particularly useful abundance diagnostics ( @xcite , @xcite ) .
hamann & ferland ( 1993 ) examined the chemical evolution of belr gas by applying spectral synthesis and chemical enrichment models to the nv / heii and nv / civ line ratios .
they found that nominal belr parameters predict nv / heii near unity for solar abundance , increasing to @xmath144 for @xmath145 .
the sensitivity of nv / civ to metallicity is due to the fact that n is a secondary element , with abundance proportional to @xmath146 , while c is a primary element with abundance proportional to @xmath147 .
their calculations show that for the ism associated with a population with solar abundance , nv / civ @xmath148 0.1 .
luminous quasars at @xmath149 show nv / civ @xmath150 , indicating a metallicity of @xmath151 . from the keck / nirspec @xmath6-band spectrum of sdss 1030 + 0524
, we find that the rest frame ew of civ is 31.5 @xmath152 8.6 .
we do not detect heii in emission at @xmath153 , which indicates that the heii ew is smaller than @xmath154 at 3-@xmath57 .
the discovery spectra in figure 5 show tentative detections of nv emission in all three quasars .
the quality of the spectra does not allow an accurate fit to the ly@xmath11+nv line profile in order to derive the equivalent width of nv .
if we assume the ew of nv to be of the order 10 , and a power law continuum with @xmath142 , we find the flux ratios nv / civ @xmath155 0.4 , and nv / heii @xmath156
. both limits would imply super - solar metallicity of the belr region . in the calculation of hamann & ferland ( 1993 ) ,
these line ratios are consistent with the `` giant elliptical '' model with fast stellar evolution and a top - heavy initial mass function , with @xmath157 . a high resolution , high s / n spectrum is needed to fit the ly@xmath11+nv line profile and put better constraints on the line ratios ( paper ii )
. however , note that krolik & voit ( 1998 ) point out that nv can be excited by resonance scattering from the red wing of the ly@xmath11 emission line , thus biasing these metallicity calculations .
the luminous high - redshift quasars discussed in this paper are ideal targets for detailed studies of the intergalactic medium at high redshift .
we calculate average absorption in the ly@xmath11 forest region based on the discovery spectra .
the results are summarized in table 4 , which also includes the measurements for sdss 10440125 .
three quantities are calculated for each quasar .
we first estimate the average continuum decrements as : @xmath158 , where @xmath159 and @xmath160 are the observed and the unabsorbed continuum fluxes of the quasar , and @xmath161 and @xmath162 measure the decrements in the region between rest - frame ly@xmath11 and ly@xmath115 ( @xmath163 ) and between ly@xmath115 and the lyman limit ( @xmath164 ) , respectively ( @xcite ) . following fan et al .
( 2000b ) , we measure @xmath161 and @xmath162 assuming a power law continuum @xmath165 with @xmath166 .
the errors on @xmath167 and @xmath168 are dominated by the continuum determination .
the error bars in table 4 reflect the range of @xmath167 and @xmath168 , allowing @xmath11 to range from @xmath169 to + 0.5 .
the range of igm redshifts covered by the window over which @xmath161 is calculated is rather large ( @xmath170 for a quasar at @xmath171 ) .
therefore we also calculate the transmitted flux ratio over a smaller window ( of 0.2 ) centered on a redshift @xmath172 : @xmath173 we assume a power law continuum of @xmath142 .
the error bars in table 4 reflect only the photon noise of the spectra .
the quantity @xmath174 is more noisy than @xmath167 , but can be calculated up to within a redshift of 0.2 of the quasar ; closer than this , it would be affected both by the proximity effect from the quasar itself and the blue wing of the ly@xmath11 emission line ( @xmath175 , oke & korycansky 1982 ) . in table 4 and figure 7 , we show the @xmath174@xmath176 values for the redshifts @xmath177 . from table 4 and
figure 7 , it is evident that the amount of ly@xmath11 absorption increases rapidly with redshift , with @xmath178 at @xmath179 . for comparison ,
@xmath180 at @xmath181 ( songaila et al . 1999 ) . the most striking feature , however , is that based on the spectrum of sdss 1030 + 0524 , @xmath174@xmath182 , consistent with no flux detected at all .
the flux level is consistent with zero from 8400 to 8700 in the spectrum ( figure 5 ) , corresponding to @xmath183 .
the 1-@xmath57 _ lower limit _ on the flux decrement in this quasar is a factor of 50 . at slightly lower redshift , we find @xmath184 from the other three @xmath43 quasars in the sample
. the flux decrement observed in sdss 1030 + 0524 is at least a factor of 4 larger than that seen in any other quasar at @xmath129 .
there seems to be a drastic change in the amount of ly@xmath11 absorption at @xmath42 .
the absence of flux over a 300 region in the arc 3.5 m discovery spectrum indicates a possible first detection of the complete gunn - peterson trough .
this detection is still highly tentative , due to the uncertainty in sky subtraction in these low s / n spectra .
note that spectra taken on different nights show the same non - detection of flux in the ly@xmath11 forest region .
this result shows that the fraction of neutral hydrogen has increased substantially between @xmath185 and @xmath17 , and , if confirmed by high s / n spectroscopy , would show that the universe is approaching the epoch of reionization . in paper ii
, we use the high - resolution keck spectrum of sdss 1030 + 0524 to place stronger constraints on the status of the igm at @xmath42 , and discuss its implication on the reionization of the universe in more detail . for a uniformly distributed igm ,
the gunn - peterson ( 1965 ) optical depth is : @xmath186 if the igm were mostly neutral , @xmath187 , and the gunn - peterson opacity would be @xmath188
. even a tiny fraction of neutral hydrogen ( of the order of @xmath189 ) in the igm could result in a large optical depth , and undetectable flux in the ly@xmath11 forest region .
therefore , the existence of the gunn - peterson trough by itself does not indicate that the object is observed prior to the reionization epoch , as it takes only a tiny fractional abundance of neutral hydrogen to create a trough .
indeed , both semi - analytic models ( miralda - escud et al .
2000 ) and hydro - dynamical simulations ( @xcite ) of structure formation show that the gunn - peterson trough should begin to appear at @xmath27 , while the epoch of overlap of hii regions could have occurred at a higher redshift .
the three new @xmath43 quasars presented in this paper , plus sdss 10440125 , comprise a complete color - selected flux - limited sample of @xmath43 quasars at @xmath5 .
the color selection criteria are listed in eq .
( 1 ) , and the total sky coverage is 1550 deg@xmath2 , using the sdss photometric runs in table 1 .
the continuum properties of this sample are given in table 5 .
following fan et al .
( 2000c ) , the quantity @xmath190 is defined as the ab magnitude of the continuum at rest - frame 1280 , after correcting for interstellar extinction using the map of @xcite .
we extrapolate the continuum to rest - frame 1450 , assuming a continuum shape @xmath142 to calculate @xmath191 .
table 5 also lists the absolute magnitudes @xmath192 and @xmath193 in both the @xmath44 and @xmath48 models . in the next subsections , we first calculate the selection function of our color - selected sample , then derive the total spatial density of luminous quasars at @xmath42 . the selection function @xmath194 is defined as the probability that a quasar of a given @xmath192 and @xmath34 will satisfy the selection criteria ( eq . [ 1 ] ) .
we calculate this using a monte - carlo simulation of quasar colors , based on the quasar spectral model described in fan ( 1999 ) .
we follow the procedures described by fan et al .
( 2001a ) , with several modifications : \(1 ) @xmath6
band magnitudes are included in the calculation .
we use the @xmath6 band filter curve measured for the filters used in the grim ii instrument on the arc 3.5 m ( we assume that the the ufti and magic instruments have similar @xmath6 band filter curves ) .
it is very similar to the filter used in the 2mass survey .
the detailed filter curve is available from the authors .
since the @xmath6 band flux of the quasar is dominated by the power law continuum , any small differences between the @xmath6 band filter curves of different instruments has little effect on the selection function .
\(2 ) sdss photometric errors , as a function of seeing , are modeled in more detail .
the selection criteria include @xmath195 , which is a strong function of the @xmath39 magnitude and the observed seeing .
we first construct the distribution of @xmath0 and @xmath34 seeing for the entire survey area . in the simulation
, we randomly draw the @xmath0 and @xmath34 seeings from this distribution and use them to calculate the photometric error terms of the sdss magnitudes .
\(3 ) the selection is based on the @xmath39 magnitudes without correcting for interstellar extinction .
we therefore construct the @xmath196 distribution of the entire survey area based on the dust map of @xcite , and assign values of @xmath196 drawn at random from this distribution , to the quasars in the simulation . following fan et al .
( 2001a ) , we model the quasar intrinsic spectral energy distributions ( seds ) using a power law continuum @xmath197 , with @xmath198 , plus a series of broad emission lines , with the rest - frame equivalent width of ly@xmath11+nv with a mean of 69.3 and a standard deviation of 18.0 .
we use the quasar absorption line models in fan ( 1999 ) to simulate the ly@xmath11 absorption , which is critical to predicting the colors of high - redshift quasars . in this model , we assume that the number density of ly@xmath11 forest lines evolves as @xmath199 .
the results of the selection function are summarized in figure 8 for both cosmologies .
the heavy lines in the figure represent the 5% contour , indicating the survey limit in absolute magnitude at each redshift . from the figure , we notice : \(1 )
the survey is sensitive to the redshift range @xmath200 .
the lower limit is due to the cut @xmath54 .
the upper limit is due to both the cut @xmath201 , and the fact that quasars begin to disappear from @xmath34 band images because of ly@xmath11 absorption in the @xmath34 band ( 2.1 ) .
\(2 ) at @xmath202 , the survey reaches @xmath203 , and is rather complete ( @xmath204 ) at the bright end . at @xmath205
, it becomes increasingly incomplete even for @xmath206 .
the survey limiting absolute magnitude as a function of redshift changes much faster than the change in distance modulus would indicate , due to increasing ly@xmath11 absorption in the @xmath34 band .
figure 8 also plots the locations of the four quasars in the sample .
three of them are located in the region of high selection probability @xmath207 , while the highest - redshift and faintest object , sdss 1030 + 0524 , has a selection probability of only @xmath208 .
these calculations are based on the statistics of the quasar emission line , continuum and absorption systems observed at much lower redshift , @xmath209 . to test the sensitivity of our conclusions to our imperfect knowledge of quasar seds at @xmath27
, we repeated the calculations with various continuum slopes and emission line strengths . if we assume a different slope for the power - law continuum , @xmath210 rather than 0.79 , we find that the selection probabilities of the quasars change by an average of 3% ; similarly , when we change the average equivalent width of the ly@xmath11+nv blend from 69 to 39 , the selection probabilities change by 2% .
thus , given the small size of the quasar sample , the error in the estimated spatial density ( 5.3 below ) is dominated by small number statistics rather than the assumptions made in the quasar color simulation . however , using these simulations we find that the selection criteria in eq .
( 1 ) will not select quasars with undetectable emission lines ( ew(ly@xmath11 ) @xmath110 0 , e.g. , sdss 1533 - 0039 , @xcite ) .
such objects will have bluer @xmath85 and redder @xmath211 colors , and are located much closer to the stellar locus .
they can , however , be selected using a more relaxed color cut .
e.g. , using @xmath212 and @xmath213 , the average selection probability of these unusual quasars will be higher than 40% for bright quasars at @xmath214 .
the @xmath52 color of @xmath215 quasars is mainly determined by the ly@xmath11 absorption .
we have very little idea of how the ly@xmath11 forest evolves with redshift at @xmath42 . however , as shown in 4.3 , there is tentative evidence that the evolution of average absorption is stronger than if we assume a simple extrapolation from lower redshift .
if that is the case , then at a given redshift , we will overestimate the @xmath0 flux in our models , while the @xmath34 flux remains almost unchanged , as it is dominated by the quasar intrinsic spectrum ( continuum + emission lines ) redward of ly@xmath11 .
thus the true @xmath52 color of quasars is redder than the model we assume here , while the @xmath73 color changes little .
this means that our @xmath216 color cut would be sensitive to quasars at redshifts slightly below 5.8 ; the selection probability at @xmath43 would remain unchanged .
the presence of foreground dust in the quasar environment would redden the quasar colors .
in figure 2 , we indicate the reddening vector for @xmath217 for a quasar at @xmath17 . as discussed in fan et al .
( 2001b ) , the number density of high - redshift quasars could be significantly affected by the uncertain amount of dust extinction . with four quasars in the sample
, we can only derive the total spatial density of the quasars in the redshift and luminosity range the survey covers .
we calculate this quantity using the @xmath218 method , following the discussion in fan et al .
( 2001a ) .
for each quasar , the volume over which it would have been observed in our survey is @xmath219 where the integral extends over the redshift range @xmath220 .
the total spatial density and its statistical uncertainty can be estimated as : @xmath221^{1/2}.\ ] ] using the selection function presented in 5.1 , we find that at the average redshift of @xmath222 , @xmath223 mpc@xmath23 in the @xmath48 model , and @xmath224 mpc@xmath23 in the @xmath44-model .
schmidt , schneider & gunn ( 1995 , hereafter ssg ) derive the high - redshift quasar luminosity function in the range @xmath225 using 90 quasars at @xmath226 .
fan et al .
( 2001b ) calculate the evolution of the quasar luminosity function over the range @xmath227 and @xmath228 , using a sample of 39 quasars ; the two luminosity functions are in good agreement . in figure 9
, we show the density of quasars at @xmath18 found in this paper , along with the results from ssg , fan et al .
( 2001b ) , and the 2df survey ( boyle et al .
2000 ) at @xmath229 .
the quasar density at @xmath42 found in this paper is consistent with extrapolating the best - fit quasar luminosity functions in the range @xmath230 from ssg and fan et al .
( 2001b ) . both ssg and fan et al . ( 2001b )
assume a single power law luminosity function at the bright end and that the number density of quasars declines exponentially as a function of redshift .
for example , in the @xmath48 model , the maximum likelihood results of fan et al .
( 2001b ) find that at @xmath231 , @xmath232 mpc@xmath23 .
extrapolating the result to @xmath17 , we predict @xmath233 mpc@xmath23 , within 1-@xmath57 of the spatial density estimated in this paper . from @xmath231 to @xmath17 ,
the density of luminous quasars drops by a factor of @xmath234 .
this drop is consistent with a decline of a factor of 3 per unit redshift found in fan et al .
( 2001b ) and 2.7 per unit redshift found in ssg .
at @xmath235 , the luminous quasars described in this paper are likely to reside in the most massive systems at high redshift .
we estimate the masses of the central black holes in these quasars following the assumptions of fan et al .
( 2000c ) : the quasars are emitting at the eddington luminosity with sed modeled by elvis et al .
( 1994 ) , and the fluxes of the observed quasars are not significantly magnified by gravitational lensing or beaming .
we find that in the @xmath44-model , the estimated black hole masses are @xmath236 , @xmath237 , and @xmath238 for sdss 0836 + 0054 , sdss
1306 + 0356 and sdss 1030 + 0524 , respectively .
under the same assumptions , @xmath239 for sdss 10440125 . in the @xmath48 model ,
the estimated black hole masses are about 20% lower .
note that the elvis et al .
( 1994 ) quasar sed , which we use to estimate the ( substantial ) bolometric correction from the flux at rest wavelength of 1450 , is based on quasars at @xmath240 .
the bolometric correction of quasars at @xmath42 could in fact be quite different , but to determine this will require observations of these high - redshift quasars over a large range of wavelengths .
there is no established relation between black hole mass and the mass of the galactic bulge in which it resides , at high redshift . if we make the rather large assumption that the magorrian et al .
( 1998 ) relation , @xmath241 , determined at @xmath242 , also holds at high redshift , we find that in the @xmath44-model , @xmath243 for the quasars in this paper .
assuming further that the ratio of the bulge to dark matter halo mass is of order the ratio of @xmath244 to @xmath245 , we find that these quasars reside in dark matter halos with mass @xmath246 for @xmath247 .
however , kauffmann & haehnelt ( 2000 ) argue that the ratio of black hole mass to bulge mass could be smaller at high - redshift ( see also @xcite ) . alternatively , using the gebhardt et al .
( 2000 ) @xmath242 relation between black hole mass and bulge velocity dispersion @xmath248 , we find @xmath249 .
note that we assume that these four high redshift quasars are not magnified by gravitational lensing . in a @xmath21 universe ,
the lensing probability rises steeply with redshift up to @xmath250 , then becomes rather flat . in a @xmath44-dominated model ,
the increase in lensing probability stays steep to somewhat higher redshift , but flattens above the redshift at which the universe is matter - dominated . using the equations of turner ( 1990 )
, we find the raw lensing probability to be @xmath251 at @xmath42 for reasonable assumptions for the lensing optical depth
. the probability of magnification bias , however , could be much larger since the sample in this paper is a flux - limited sample , and the observed quasars are on the tail of a steep luminosity distribution ( blandford & narayan 1992 ) .
a detailed calculation of magnification bias is beyond the scope of this paper .
following turner ( 1991 ) and haiman & loeb ( 2001 ) , we consider the growth of supermassive black holes in high - redshift quasars .
the @xmath252-folding timescale for black hole growth is @xmath253 yr , where @xmath254 is the ratio of the quasar bolometric luminosity to the eddington luminosity , and @xmath255 is the radiation efficiency . for sdss 0836 + 0054 , the most luminous quasar in our sample , it takes 20@xmath90 @xmath252-folding times , or @xmath256 gyr , to grow from a 10 @xmath257 stellar black hole , if @xmath258 .
this timescale is very close to the age of the universe at @xmath42 in the @xmath44-model .
if @xmath259 , it is actually _ longer _ than the age of the universe in the @xmath48 model .
note that the black hole could grow much faster if the radiation efficiency is very low .
alternatively , the black hole could have formed from a much more massive seed black hole , such as a very massive object in population iii ( vmos ; bond , arnett & carr 1984 ) , or from mergers of smaller black holes ( e.g. kauffmann & haehnelt 1999 ; menou , haiman , & narayanan 2001 ) .
the quasar luminosity function at high redshift is a sensitive test of cosmological parameters and models of quasar evolution .
the observed luminous quasars most probably represent rare peaks in the density field at @xmath215 , and hence probe the exponential , high - mass tail of the underlying dark matter halo distribution .
the slope of the luminosity function is determined by both the slope of the halo mass function and the relation between black hole mass ( proportional to quasar luminosity if the quasar is radiating at the eddington limit ) and dark halo mass .
the @xmath27 quasars presented in this paper are extremely rare objects , representing many-@xmath57 peaks in the density field .
one would therefore expect their luminosity function to be very steep .
we use the luminosity distribution of the four quasars in our sample to constrain the slope of the bright end of the quasar luminosity function , and compare it with theoretical expectations , assuming that the fluxes of the four observed quasars are not significantly magnified by gravitational lensing . among the four quasars in the complete sample in table 5 , two have @xmath206 ( @xmath48 model ) .
we use this as a constraint to calculate the expected number of fainter quasars in the survey area , assuming a power law luminosity function , @xmath260 , and taking into account the survey selection function presented in figure 8 .
we then compare this expected number with the actual number of lower - luminosity quasars observed ( two at @xmath261 ) . for @xmath262 , we expect to find less than six quasars at @xmath263 in our survey . in this case , the probability of observing fewer than three lower luminosity quasars is larger than 5% ( assuming poisson statistics ) .
therefore , @xmath262 is consistent with the observed distribution at greater than the 2-@xmath57 level .
note that this constraint encompasses the values of the bright - end slope of the quasar luminosity function from fan et al .
( 2001b ) , who found @xmath264 using a sample of quasars at @xmath209 , and boyle et al . ( 2000 ) , who found @xmath265 at @xmath266 .
however , the observed luminosity distribution is not consistent with an even steeper luminosity function .
the probability drops to 0.01 for @xmath267 , and to 0.001 for @xmath268 .
in other words , a steep quasar luminosity function with @xmath269 is ruled out at the @xmath270-@xmath57 level by the current observations .
this constraint simply reflects the fact that even after taking into account the lower selection completeness of the survey at the faint end , there are not many lower luminosity quasars in our survey at @xmath27 ( only two are observed ) , given that we have discovered two bright ones ( @xmath206 , @xmath271 ) .
we note that this constraint on the slope of the bright end of the luminosity function is quite weak , as it is based on only four objects , and depends strongly on our model of the survey selection function and the assumption that the observed quasars are not significantly magnified by gravitational lensing . in order to obtain a stronger constraint on the slope of the luminosity function , a large , and more importantly ,
a deeper sample of @xmath272 quasars is needed .
is this constraint on the bright end slope of the quasar luminosity function consistent with a simple model of quasar evolution ?
we can answer this question in the context of specific models of quasar formation .
following haiman & loeb ( 1998 ) , we assume that there is a monotonic relation between quasar luminosity at high redshift and the mass of the dark matter halo in which it resides , and assume further that every black hole shines as a quasar for a given fraction of the age of the universe to that redshift ( the `` lifetime '' or `` duty cycle '' of the quasar ) .
we can then match the observed quasar number density at @xmath17 to the number density of dark matter halos found in the hubble - volume simulations ( jenkins et al .
2000 ) , if we assume that the quasars in our sample are associated with dark matter halos more massive than @xmath273m@xmath274 . in this case
, the dark halo mass function at this mass scale has a logarithmic slope between @xmath275 and @xmath276 , depending on the cosmology and quasar life time ( @xmath277 years ) .
note that this minimum halo mass is comparable to that estimated using the @xmath278 relation of magorrian et al .
( 1998 ) , assuming that the quasars radiate at the eddington limit .
these luminous quasars represent about 5-@xmath57 peaks in the density field for a @xmath44-cdm power spectrum , and the mass function is extremely steep at this mass .
the slope of the dark halo mass function is steeper than that of the quasar luminosity function @xmath279 .
this difference in the slopes of the bright end of the quasar luminosity function and the dark halo mass function at the corresponding mass scale implies that for high - redshift quasars , the luminosity @xmath280 can not simply scale linearly with @xmath281 ; rather , high - mass black holes radiate more efficiently than do lower mass ones .
alternatively , the quasar luminosity function can be shallower than the mass function if there is large scatter in the @xmath282 relation , or if the fluxes of the observed quasars have been significantly amplified by gravitational lensing . clearly , the above results are oversimplified and are based on a number of assumptions regarding both quasar activity and the correspondence between quasar luminosity and the mass of the dark matter halo in which it resides .
however , they illustrate the importance of the measurement of the shape of the quasar luminosity function in constraining models of quasar formation and evolution .
the absence of the gunn - peterson trough in sdss 1306 + 0356 ( @xmath8 ) indicates that the universe is still highly ionized at @xmath283 , presumably by the ionizing photons from quasars and star - forming galaxies .
the photon emissivity per unit comoving volume required to keep the universe ionized at @xmath17 in the @xmath21 model is ( @xcite ) , @xmath284 where @xmath285 is the clumping factor of the igm .
based on the extrapolation of the quasar luminosity function from @xmath286 , madau et al .
( 1999 ) conclude that quasars alone do not provide enough ionizing photons to keep the universe ionized at @xmath129 , and that the major contribution to the cosmic ionizing background has to be produced by star - forming galaxies ( see also @xcite ) . here
we calculate the total contribution of the quasar population to the ionizing background at @xmath27 , using the quasar density derived in 5.2 and the constraint on the bright end slope of the quasar luminosity function in 6.1 .
we calculate the total emissivity per unit comoving volume of ionizing photons from the quasar population : @xmath287 where @xmath288 is the total quasar emissivity at 1450 , with units of erg s@xmath20 hz@xmath20 mpc@xmath23 , assuming a luminosity function @xmath289 @xmath290 and @xmath291 is the total number of ionizing photons for a source with a luminosity of 1 erg s@xmath20 hz@xmath20 at 1450 .
we assume the quasar sed following madau , haardt , & rees ( 1999 ) . @xmath292 and integrate over the energy range from 1 to 4 rydberg , with a cutoff at 4 rydberg because of heii absorption .
we assume that the luminosity function of quasars at @xmath27 is described by a double power law , and write it as a function of absolute magnitude @xmath192 : @xmath293 where @xmath294 is the characteristic luminosity .
none of the current quasar surveys at @xmath143 constrains the faint end slope of the quasar luminosity function .
for example , fan et al .
( 2001b ) only compute the luminosity function at @xmath295 .
therefore , the characteristic luminosity @xmath294 and the faint end slope @xmath296 are both unknown at high redshift , and we simply _ assume _ the faint end slope of the luminosity function to be the same as determined by boyle et al .
( 2000 ) at @xmath266 , @xmath297 .
thus , for a given bright end slope @xmath298 , we can derive @xmath299 by requiring @xmath300 mpc@xmath23 , as determined in 5.2 . in 6.1
, we demonstrated that the bright end slope @xmath298 of the quasar luminosity function at @xmath27 is consistent with that measured at lower redshift ( @xmath301 , fan et al .
2001b , and @xmath302 , boyle et al .
2000 ) , and that a much steeper slope can be ruled out ( @xmath303 is ruled out at the 99% confidence level , and @xmath304 with 99.9% confidence ) . in figure 10 , we calculate the quasar ionizing photon density integrated over the full luminosity function as a function of @xmath294 for a series of values of @xmath298 , assuming @xmath297 , and normalizing to the quasar density at @xmath18 .
the quasar ionizing photon density is compared with what is required to keep the universe ionized at @xmath27 ( heavy solid line ) . from figure 10 ,
it is evident that with @xmath305 , quasars would not produce enough photons to keep the universe ionized at @xmath27 , unless @xmath306 , i.e. , only if the ionizing background is dominated by low luminosity agns can the quasar luminosity function provide enough ionizing photons .
this would imply a large negative luminosity evolution of the luminosity function .
the universe could be ionized with luminous quasars only if the bright end slope of the luminosity function were much steeper than that measured at low redshift ; but such a steeper slope is not consistent with our sample .
this calculation underlines the importance of searching for faint quasars and studying the shape of the high - redshift quasar luminosity function at the low luminosity end .
while it is unlikely that quasars provide the majority of photons that ionized the universe at @xmath307 , it still remains an open question .
in this paper , we present three new quasars at @xmath43 , selected from @xmath1 deg@xmath2 of multicolor imaging data from the sloan digital sky survey .
the quasar candidates are selected as @xmath0-dropout objects , with red @xmath216 colors .
they are separated from red cool dwarfs using follow - up @xmath6-band photometry .
the three new quasars , with redshifts of 5.82 , 5.99 and 6.28 , are the highest - redshift objects with spectral confirmation known to date .
the spectra of the new quasars show very strong ly@xmath11 absorption blueward of the ly@xmath11 emission , with more than 90% of the flux absorbed . the amount of absorption increases rapidly toward @xmath308 . for
the @xmath9 quasar sdss 1030 + 0524 , the flux is consistent with zero in a region of 300 immediately blueward of ly@xmath11 emission .
it suggests a tentative detection of a complete gunn - peterson trough , indicating that at @xmath42 the universe is close to the reionization epoch .
the discovery spectra show tentative detection of the nv emission line , suggesting the existence of large amount of heavy metals in the gas around the quasars .
a @xmath6-band spectrum of sdss 1030 + 0524 shows strong civ emission .
the limits on the nv / civ and nv / heii line ratios are consistent with the gas around the quasar having super - solar metallicity .
the three new quasars described in this paper , plus the @xmath15 quasar sdss 10440125 ( fan et al .
2000c ) , form a complete color - selected , flux - limited sample at @xmath42 .
we calculate the color selection completeness of this sample , and derive the spatial density of luminous quasars at @xmath42 , @xmath309 mpc@xmath23 , for @xmath310 and @xmath19 km s@xmath20 mpc@xmath20 .
this density is about a factor of two lower than that at @xmath311 , and is consistent with an extrapolation of the observed redshift evolution of quasars at @xmath312 . using the luminosity distribution of the sample
, we show that the luminosity function is shallower than @xmath313 at the bright end .
a larger and deeper sample is needed to better constrain the shape of the quasar luminosity function .
we confirm that the quasar population is unlikely to provide enough photons to ionize the universe at @xmath27 , unless the luminosity function is much steeper than @xmath314 at the faint end .
the black hole masses of these quasars are probably several times @xmath28 .
the quasars are likely to reside in very massive systems , with the minimum mass of host dark halos @xmath315 .
these massive dark halos represent rare peaks in the density field at high redshift , and are in the steep tail of the mass function , with a slope that is appreciably steeper than that of the quasar luminosity function .
follow - up observations of the luminous high - redshift quasars described in this paper will provide excellent probes of galaxy formation and igm evolution at high redshift . in paper ii , we present high - resolution , high - s / n optical spectroscopy of these quasars . these observations are compared with detailed cosmological simulations to constrain models of reionization in fan et al .
( 2001c ) .
x - ray observations ( e.g. , brandt et al . 2001 ) will reveal the status of the environment very close to the quasar central engine ; follow - up ir spectroscopy can provide more diagnostics on the metallicity of the quasars and reveal the possible existence of strong associated absorption ( e.g. , @xcite ) ; and detections of the sources at sub - millimeter wavelengths will shed light on the possible connections between quasar activity and starbursts ( e.g. , carilli et al . 2001 ) .
the total area of the sdss survey is 10,000 deg@xmath2 . assuming the quasar luminosity function of fan et al .
( 2001b ) and normalizing it to have the spatial density at @xmath42 found in this paper , we expect to detect _
one _ @xmath316 quasar with @xmath59 ( @xmath317 ) in the entire survey .
a quasar at this redshift has @xmath318 and @xmath73 colors similar to those of early t dwarfs .
this is the highest redshift of objects we expect to find using solely sdss imaging data .
the sloan digital sky survey ( sdss ) is a joint project of the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , princeton university , the united states naval observatory , and the university of washington .
apache point observatory , site of the sdss telescopes , is operated by the astrophysical research consortium ( arc ) .
funding for the project has been provided by the alfred p. sloan foundation , the sdss member institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is http://www.sdss.org .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and nsf .
we thank the staff at apo , keck , calar alto , and ukirt , for their expert help , gemini staff scientist marianne takamiya for carrying out the germini / nirspec service observation .
we thank ed turner , peng oh , julian krolik , kristian finlator , john bahcall and an anonymous referee for helpful discussions and comments .
we acknowledge support from nsf grant phy00 - 70928 and a frank and peggy taplin fellowship ( xf ) , nsf grant ast-0071091 ( mas ) , and nsf grant ast-9900703 ( dps ) .
mclean , i. s. , becklin , e. e. , bendiksen , o. , brims , g. , canfield , j. , figer , d. f. , graham , j. r. , hare , j. , lacayanga , f. , larkin , j. , larson , s. b. , levenson , n. a. , magnone , n. , teplitz , h. , & wong , w. 1998 , spie , 3354 , 566 rrrrcc 745 & 10 n & 51.1 & 65.0 & 1.23 & 1.25 + 752 & 10 s & @xmath319 & 51.0 & 1.43 & 1.50 + 756 & 10 n & @xmath320 & 51.0 & 1.34 & 1.38 + 1140 & 9 n & @xmath321 & 17.1 & 1.44 & 1.47 + 1231 & 9 s & @xmath322 & 14.7 & 1.51 & 1.53 + 1239 & 10 s & @xmath320 & @xmath323 & 1.41 & 1.46 + 1241 & 9 s & @xmath324 & @xmath325 & 1.49 & 1.57 + 1241 & 9 s & @xmath326 & @xmath327 & 1.74 & 1.76 + 1331 & 36 s & @xmath328 & @xmath329 & 1.18 & 1.18 + 1331 & 36 s & @xmath330 & @xmath331 & 1.38 & 1.36 + 1332 & 36 s & @xmath332 & @xmath333 & 1.52 & 1.46 + 1336 & 42 n & 23.4 & 36.1 & 1.51 & 1.50 + 1339 & 42 s & 23.3 & 36.0 & 1.41 & 1.38 + 1345 & 36 n & @xmath334 & @xmath335 & 1.42 & 1.42 + 1345 & 36 n & 1.3 & 73.9 & 1.23 & 1.26 + 1350 & 37 s & @xmath336 & 14.5 & 1.54 & 1.54 + 1356 & 43 n & 22.8 & 36.5 & 1.64 & 1.62 + 1359 & 43 s & 27.7 & 37.2 & 1.61 & 1.57 + 1402 & 37 n & @xmath337 & @xmath338 & 1.24 & 1.22 + 1412 & 37 n & @xmath339 & 33.2 & 1.28 & 1.28 + 1449 & 1 n & 2.9 & 13.4 & 1.46 & 1.51 + 1450 & 37 n & @xmath340 & @xmath341 & 1.53 & 1.54 + 1462 & 11
s & @xmath342 & 40.6 & 1.27 & 1.35 + 1468 & 32 n & 10.4 & 29.1 & 1.37 & 1.35 + 1469 & 32 s & 7.5 & 29.4 & 1.49 & 1.48 + 1478 & 12 s & 25.6 & 55.8 & 1.43 & 1.49 + 2074 & 35 s & @xmath343 & @xmath344 & 1.51 & 1.56 + 2076 & 35 n & @xmath345 & @xmath346 & 1.33 & 1.35 + 2078 & 36 s & @xmath347 & 8.5 & 1.47 & 1.49 + 2125 & 12 s & @xmath348 & @xmath349 & 1.53 & 1.58 + 2126 & 12 n & @xmath350 & 0.2 & 1.30 & 1.36 + 2131 & 34 s & @xmath351 & @xmath352 & 1.41 & 1.50 + 2134 & 36 s & 8.4 & 23.5 & 1.48 & 1.41 + 2137 & 34 n & @xmath353 & @xmath354 & 1.37 & 1.43 + 2138 & 34 n & @xmath355 & @xmath356 & 1.51 & 1.55 + 2140 & 34 s & @xmath357 & @xmath358 & 1.69 & 1.76 + 2141 & 10 s & @xmath359 & @xmath360 & 1.42 & 1.51 + 2190 & 12
n & 2.3 & 28.2 & 1.47 & 1.54 + 2190 & 12 n & 32.9 & 51.7 & 1.48 & 1.53 rrr @xmath43 quasars & 4 & 3.3% + t dwarfs & 11 & 9.1% + m / l dwarfs by spectroscopy & 10 & 8.3% + m / l dwarfs by photometry & 60 & 49.6 % + bal quasar & 1 & 0.8% + false detection & 35 & 28.9% + total & 121 & 100% cccccc j083643.85 + 005453.3 & 5.82 @xmath152 0.02 & 21.04 @xmath152 0.08 & 18.74 @xmath152 0.05 & 17.89 @xmath152 0.05 & 1239 + j103027.10 + 052455.0 & 6.28 @xmath152 0.03 & 23.23
@xmath152 0.43 & 20.05 @xmath152 0.10 & 18.87 @xmath152 0.10 & 2125 + j130608.26 + 035626.3 & 5.99 @xmath152 0.03 & 22.58 @xmath152 0.26 & 19.47 @xmath152 0.05 & 18.77 @xmath152 0.10 & 2190 ccccccc j104433.04012502.2 & 5.80 & 0.91 @xmath152 0.02 & 0.95 @xmath152 0.02 & 5.5 & 0.088 @xmath152 0.004 + j083643.85 + 005453.3 & 5.82 & 0.90 @xmath152 0.02 & 0.91 @xmath152 0.02 & 5.5 & 0.099 @xmath152 0.014 + j130608.26 + 035626.3 & 5.99 & 0.92 @xmath152 0.02 & 0.95 @xmath152 0.02 & 5.7 & 0.069 @xmath152 0.014 + j103027.10 + 052455.0 & 6.28 & 0.93 @xmath152 0.02 & 0.99 @xmath152 0.01 & 6.0 & 0.003 @xmath152 0.020 ccccccc j083643.85 + 005453.3 & 5.82 & 18.88 & 18.81 & 27.62 & 27.88 & 0.050 + j103027.10 + 052455.0 & 6.28 & 19.73 & 19.66 & 26.89 & 27.15 & 0.023 + j104433.04012502.2 & 5.80 & 19.28 & 19.21 & 27.15 & 27.50 & 0.054 + j130608.26 + 035626.3 & 5.99 & 19.61 & 19.55 & 26.93 & 27.19 & 0.028 figure 1 .
@xmath216 vs. @xmath39 color - magnitude diagram for the @xmath0-dropout sample .
colors and magnitudes are plotted as asinh magnitudes measured by sdss imaging .
the symbols represent the classification in table 2 : circles : t dwarfs ; triangles : l dwarfs ; stars : @xmath43 quasars ; square : bal quasar . filled symbols are objects with s / n in the @xmath0 band higher than 4 ; open symbols are objects not detected in the @xmath0 band at the 4-@xmath57 level .
the median track of simulated @xmath216 color and @xmath39 magnitude for quasars with @xmath53 is also shown as a function of redshift , with plus signs every 0.1 in redshift .
for comparison , the data for a random sample of 50,000 high - latitude stars are also shown as dots .
the dashed line shows the cut @xmath54 that we use to select high - redshift quasar candidates .
@xmath216 vs. @xmath78 color - color diagram for the @xmath0-dropout sample .
the symbols are the same as in figure 1 .
the median track of simulated quasar colors is shown as a function of redshift .
the survey selection criteria are illustrated by the shaded area . for comparison ,
colors of sdss-2mass stars in a 50 deg@xmath2 area at high latitude are also shown .
the arrow at the lower - right corner indicates the reddening vector for quasars at @xmath17 with @xmath361 .
projection of the survey area in j2000 equatorial coordinates .
note that for some sdss stripes , only one of the two scans is covered by the current survey .
the stripes are each @xmath362 wide .
@xmath34 finding charts of the three new @xmath43 quasars . for sdss 0836 + 0054 , the sdss @xmath34 image is shown .
for sdss 1030 + 0524 and sdss 1306 + 0356 , the arc 3.5 m @xmath34 images taken with spicam are shown ( 180 seconds exposure ) .
the size of the finding chart is 160 .
north is up and east is left .
the discovery spectra of the three new @xmath43 quasars , taken with the arc 3.5 m telescope and dis spectrograph .
the exposure time is 3600 seconds for each quasar and the spectral resolution is about 20 .
the spectra have been normalized to the photometry in the @xmath34 filter .
figure 6 . combined optical + near ir spectrum of sdss 1030 + 0524 .
the optical spectrum is a 3600 second exposure taken with arc 3.5 m telescope ( same as in figure 5 ) .
the near ir spectrum is a 3000 second exposure taken with keck / nirspec .
the resolution of the near ir spectrum is @xmath363 .
evolution of the transmitted flux ratio in the ly@xmath11 forest region , averaged over the wavelength range @xmath364 @xmath365 .
the vertical error bars represent the photon noise .
note that for sdss 1030 + 0524 , the flux is consistent with zero in the ly@xmath11 region immediately blueward of the ly@xmath11 emission .
the selection probability of @xmath43 quasars as a function of redshift and luminosity , for the @xmath48 and the @xmath44 models .
the probability contours of 5% , 20% , 40% , 60% and 80% are shown .
the large dots represent the locations of the 4 quasars in the sample .
the evolution of quasar comoving spatial density at @xmath18 in the @xmath48 model .
the large dot represents the result from this survey .
the dashed and dotted lines are the best - fit models from fan et al .
( 2001b ) and schmidt et al .
( 1995 ) , respectively .
the solid line is the best - fit model from the 2df survey ( boyle et al .
2000 ) at @xmath229 .
figure 10 .
comoving emission density rate of hydrogen lyman continuum photons from quasars integrated over the full luminosity function , compared with the minimum rate ( heavy solid line ) that is needed to fully ionize a universe at @xmath17 , in the @xmath48 model , with a clumping factor of 30 , and the baryon fraction @xmath366 . the recombination time is assumed to be faster than the hubble time .
the quasar emission rate is calculated for different bright end slopes of the quasar luminosity function ( @xmath298 ) and turnover luminosity @xmath294 , assuming the same faint end slope @xmath297 and the total density of luminous quasars matched to that found in this paper , @xmath367 mpc@xmath23 . | we present the results from a survey of @xmath0-dropout objects selected from @xmath1 deg@xmath2 of multicolor imaging data from the sloan digital sky survey ( sdss ) , to search for luminous quasars at @xmath3 .
objects with @xmath4 and @xmath5 are selected , and follow - up @xmath6 band photometry is used to separate l and t type cool dwarfs from high - redshift quasars .
we describe the discovery of three new quasars , sdssp j083643.85 + 005453.3 ( @xmath7 ) , sdssp j130608.26
+ 035626.3 ( @xmath8 ) and sdssp j103027.10 + 052455.0 ( @xmath9 ) . the quasar sdssp j083643.85 +
005453.3 is a radio source with flux of 1.1@xmath10mjy at 20@xmath10 cm .
the spectra of all three quasars show strong and broad ly@xmath11+nv emission lines , and very strong ly@xmath11 forest absorption , with a mean continuum decrement @xmath12 .
the arc 3.5 m spectrum of sdssp j103027.10 + 052455.0 shows that over a range of @xmath13 immediately blueward of the ly@xmath11 emission , the average transmitted flux is only @xmath14 times that of the continuum level , consistent with zero flux over a @xmath13 range of the ly@xmath11 forest region , and suggesting a tentative detection of the complete gunn - peterson trough .
the existence of strong metal lines in the quasar spectra suggests early metal enrichment in the quasar environment .
the three new objects , together with the previously published @xmath15 quasar sdssp j104433.04 - 012502.2 , form a complete color - selected flux - limited sample at @xmath16 .
we estimate the selection function of this sample , taking into account the estimated variations in the quasar spectral energy distribution as well as observational photometric errors .
we find that at @xmath17 , the comoving density of luminous quasars at @xmath18 ( @xmath19 km s@xmath20 mpc@xmath20 , @xmath21 ) is @xmath22 mpc@xmath23 .
this is a factor of @xmath24 lower than that at @xmath25 , and is consistent with an extrapolation of the observed quasar evolution at @xmath26 . using the current sample , we discuss the constraint on the shape of the quasar luminosity function and the implications for the contribution of quasars to the ionizing background at @xmath27 .
the luminous quasars discussed in the paper have central black hole masses of several times @xmath28 by the eddington argument , with likely dark halo masses of order @xmath29 .
their observed space density provides a sensitive test of models of quasar and galaxy formation at high redshift . |
despite the great progress in the grb field with the launch of _ cgro _ and the discovery of their afterglows by _ bepposax _ @xcite ( a fact that allowed the determination of their redshifts and provided an unequivocal measure of their distance and luminosity ) , a detailed understanding of the physics involved is still lacking . several major issues concerning the dynamics and radiative processes of these events remain , to say the least , opaque . while the models of the grb afterglow emission are reasonably successful in providing the time evolution of their afterglow flux @xcite , there is an entire host of issues which are tacitly ignored while discussing and modeling the details of grb emission .
some of these issues have been with us since the inception of the cosmological grb models , while others are rather new , the outcome of the wealth of new observations , most prominently those by _ swift _ and _
we provide below a partial list of the most important ( to our view ) such open grb issues . * the nature of the grb inner engine " .
it is widely believed that this has its origin to the collapse of a stellar core or the merging of two neutron stars ( we will not be concerned with it in the rest of the paper ) . *
the dissipation of the rbw kinetic energy .
this is converted efficiently into relativistic protons ( 1/2000th of it into electrons ) of lorentz factor ( lf ) @xmath5 ( @xmath6 is the lf of the blast wave ) .
however , the efficient conversion of the latter into radiation , needed to account for the observed grb variability , is one of the least discussed or explored issues . * the energy distribution of the prompt grb phase , in particular the presence of a peak in the @xmath7 spectra at @xmath1@xmath8 @xcite , @xcite . * the differentiation between prompt and afterglow phases .
what determines , besides detector sensitivity , the separation in and the transition from the prompt to the afterglow stages in a grb ?
* is there a unification between grb , xrr and xrf ?
the discovery of transients with timing properties similar to those of grb but with smaller @xmath1 values ( xrr , xrf ) and smaller isotropic luminosities , @xmath9 , raised the question of their relation to grb and the a possible unification of all these phenomena .
the model outlined in the present note has been conceived to resolve the second of the issues enumerated above , i.e. that of the dissipation of the energy stored in the form of relativistic protons within the rbw of a grb .
the name of the model is of relevance : its basic physics is a radiative - type instability which can convert the free energy of the relativistic proton plasma into relativistic @xmath10 pairs on time scales @xmath11 ( @xmath12 is typical size of the plasma ) , under certain criticality conditions@xcite,@xcite .
the nickname of the model derives from the fact that the criticality conditions are identical to those of a supercritical nuclear pile , where the free energy of nuclear binding can be explosively released once these conditions are fulfilled .
the instability of a relativistic proton plasma , in distinction with that of a nuclear pile , involves a two step process : ( a ) the production of @xmath13pairs by the @xmath0 reaction .
( b ) the production of additional photons , through the synchrotron process , which can then serve as targets for the above reaction to produce more pairs etc .
, offering the possibility of a runaway .
item ( a ) entails a _
threshold , while the true criticality is related to item ( b ) which in turn entails a _ dynamic _ threshold .
consider a spherical volume of size @xmath12 , containing a relativistic proton plasma of _ integral _ spectrum @xmath14 ( being the proton lorentz factor ) , along with an ( infinitesimal ) number of photons of energy @xmath15 ( in units of @xmath16 ) ; these photons can produce pairs via the @xmath17 reaction , provided that the proton population extends to lorentz factors @xmath18 such that @xmath19 . in the presence of a magnetic field @xmath20 , the pairs ( of lorentz factor also equal to @xmath21 ) produce synchrotron photons of energy @xmath22 where @xmath23 is the magnetic field in units of the critical one @xmath24 g. for the reaction network to be _ self - contained _ the energies of the seed and synchrotron photons should be equal , yielding the _ kinematic _ threshold of the process i.e. @xmath25 the reaction network will be _ self - sustained _ if _ at least one _ of the synchrotron photons pair produces before escaping the volume of the plasma .
since the number of photons produced by an electron of energy @xmath26 is @xmath27 , the _ dynamic _ threshold implies that the column density of the protons at energy @xmath21 should be greater than @xmath28 or @xmath29 where @xmath30is the cross section of the @xmath31 reaction . in the case of the rbw of a grb , on which the majority of particles are relativistic with mean energy @xmath32
, we can eschew the presence of an accelerated non - thermal relativistic population and consider only the thermal " relativistic protons present behind the forward shock of the rbw .
these can be considered as monoenergetic of energy @xmath33 ( or of a maxwellian of similar mean energy ) where @xmath6 is the lorentz factor of the rbw .
the linear dimension of the plasma in this case should be considered to be the ( comoving ) width of the rbw @xmath34 , while its density the comoving proton density @xmath35 . however , since @xmath36 and @xmath37 , @xmath38 ; hence we can express the _ dynamic _ threshold in terms of the shock radius @xmath12 and ambiend density @xmath39 as @xmath40 with the kinetic threshold used in the last step above .
it was noticed in @xcite that the constraints on the thresholds given above can be alleviated if a fraction of the synchrotron photons ( which are only at a distance @xmath41 ahead of the rbw ) , can scatter in a mirror " .
scattering allows the rbw to catch - up with these photons , whose energy upon reinterception by the rbw is now increased by a factor @xmath42 , so that now the synchrotron photons have an energy @xmath43 .
the kinematic threshold then becomes @xmath44 or @xmath45 with the last steps assuming that the protons and electrons of the relativistic post - shock plasma have lorentz factors @xmath46 .
the dynamic threshold condition remains the same , however when simplified using eq .
( [ kinematic ] ) leads to @xmath47 a condition a lot easier to fulfill than that of eq .
( [ dynamic ] ) .
it is of interest to note that the fundamental radiative process of this model , namely the @xmath48 reaction , involves a threshold electron energy [ eqs .
( [ kinematic0 ] ) or ( [ kinematic ] ) ] .
this is an important fact because it implies the absence of electron injection at energies lower than @xmath49 and leads naturally to a peak in the @xmath50 grb spectra at an ( unspecified as yet ) energy @xmath1 .
the presence of this peak energy has been one of the landmark features of grb spectra @xcite , @xcite and the issue of why @xmath51 has vexed scientists over the years . the more recent discovery of transients with values of @xmath1 smaller than those found by _
batse _ suggested that this may in fact be a selection effect with ( perhaps ) little importance .
however , the relation between @xmath1 and the isotropic emitted energy @xmath9 discovered by amati et al .
@xcite suggests that the value of @xmath1 is not random but it is tied to the physics of the burst and ( as we content ) to the conversion of its kinetic energy to radiation .
the specificity of the above model allows the qualitative computation of the resulting specta in a rather straightforward fashion : the basic process for photon production is synchrotron radiation with corresponding energy at @xmath52 .
the photons that are scattered on the mirror " , they have , upon their re - interception by the rbw , energy ( on the rbw frame ) @xmath53 .
these photons will then scatter : ( a ) elastically by cold " electrons ( @xmath54 ) of the rbw to preserve their energy at @xmath55 , ( b ) inelastically by hot " ( @xmath56 ) electrons to boost their energy to @xmath57 ( or rather to min@xmath58 as the last scattering likely takes place in the klein nishina regime ) .
these energies will appear in the lab frame blue - shifted by a factor @xmath59 to occur correspondingly at energies @xmath60 .
if the process operates close to its kimematic threshold , then , by virtue of eq .
( [ kinematic ] ) , the energy of the second ( bulk comptonized ) component will be at @xmath61 , thereby resolving the issue raised by the _ batse _
systematics @xcite .
in addition to this aspect , this model also predicts the presence of additional peaks in the grb spectra at energies @xmath62 and @xmath63 ( all energies in units of @xmath16 ) . for typical values of @xmath6 ( @xmath64 ) , these are respectively @xmath65 ev and @xmath66 gev .
the relative normalization of these three distinct components depends on the details of the processes involved @xcite ( also @xcite ) and imply the presence of prompt grb emission in the optical and the _ glast _ energy bands .
the properties of the model outlined above have been explored in detail through numerical simulations . to this end
the distribution functions of the protons , electrons and photons have been followed both in energy and time ( or correspondingly in space ) , assuming a spherical shape for the emitting plasma . this approximation is justified by considering that the radial width of the rbw only @xmath67 in the lab frame , has a comoving frame size @xmath68 ; since an external observer sees " only a fraction of the angular extent of the rbw of transverse spacial extent @xmath69 ( of the same size as the photon horizon ) , the radiating plasma on the comoving frame , can be considered as spherical of radius @xmath70 .
the calculations are thus preformed in the comoving frame and the radiation intensity is transformed to the lab frame at the very end .
the treatment ( given in detail in @xcite ) follows closely that of bttcher & dermer @xcite agn variability study . one important feature of this treatment is the fact that the radiation reflected " in the mirror " contributes to the proton and electron losses only after the rbw photons have scattered on the mirror " and re - intercepted by the rbw .
therefore there is a causality constraint associated with the corresponding terms in these equations .
assuming the mirror " to be extended , covering the range between @xmath71 and @xmath72 , these terms become effective only after time @xmath73 where @xmath74 is the velocity of the rbw ( normalized to @xmath75 ) . as a function of the magnetic field strength @xmath20 ( both quantities measured in the comoving frame ) .
the lorentz factor of the rbw is @xmath76 .
solid and dashed lines represent the cases with and without relativistic electrons respectively .
the lower horizontal dotted line is the total energy stored in electrons while the upper one is the energy stored in protons .
the vertical dotted line is the equipartition magnetic field.[ub ] ] the calculations at this first stage were performed assuming a constant value for the rbw lorentz factor @xmath6 .
these were performed under two different assumptions concerning the presence or not of relativistic electrons in the rbw , a feature crucial for the time development of the system : ( a ) no relativistic electrons initially present within the rbw .
all necessary photons are produced ( initially ) from infinitesimal ( numerical ) fluctuations in the photon field .
however , these suffice to produce pairs which in turn produce the required photons as discussed above ( see fig . 5 of @xcite ) .
the increase in the ambient photons is exponential with characteristic time scale of the order of the light crossing time of the shock ; as shown in the said figure , the characteristic time scale gets shorter the farther the parameters of system beyond their threshold values ( i.e. the larger the value of the magnetic field @xmath20 ) , with the process eventually saturating when enough pairs have accumulated to shield the protons from photon scattering .
( b ) relativistic electrons are present in the rbw same in number and lf as the protons .
these cool very quickly to produce the photons necessary for the production of more electrons etc .
, as discussed above .
the evolution in this case is much faster and the proton energy can be radiated away far more efficiently than in case ( a ) and for parameter values below those of the threshold .
these arguments are exemplified in figure [ ub ] where we plot the total energy radiated by the protons during the plasma traversal of a mirror " of thompson depth @xmath77 , @xmath78 cm and @xmath79 cm .
the presence of relativistic electrons facilitates the transfer of energy from the protons to the photons for a given value of @xmath20 . for sufficiently large values of @xmath20 , however , all the proton energy is eventually radiated away irrespective of the presence of the original relativistic electrons because the secondary pairs from the photon - proton collisions dominate the effects of the primary electrons . at the same time , because of the presence of the additional photons due to the electron cooling , the maximum available luminosity can be achieved with parameter values ( say that of the magnetic field @xmath20 , below that necessitated by the threshold conditions .
@xmath80 and @xmath76 and @xmath81 g at maximum flux .
the three components discussed in the text are apparent with the middle one having a peak near 1 mev.[spectrum ] ] in figure [ spectrum ] we also plot the resulting spectrum at flux maximum for a rbw with @xmath82 , @xmath83 g and @xmath84 @xmath80 ( the values of magnetic field and the density are measured on the rbw frame ) .
the spectrum exhibits clearly the three components discussed in 2.2 , with the middle one at an energy @xmath1 of a few mev .
a decrease in the value of the magnetic field will result in a lower value of @xmath1 , while a decrease in the proton density @xmath85 ( more correctly in the product @xmath86 ) will lead to a decrease in the lumisosity of the bulk comptonized component relative to those of synchrotron and inverse compton ones .
following their discovery @xcite , afterglows have become an integral part of the the grb study and , in fact , they presently consume the largest fraction of effort to understand the grb phenomenon .
based on the sparsely sampled early afterglow light curves and the ensuing theoretical models of their spectro - temporal behavior @xcite , their light curves were expected to be smooth with rather well specified breaks in their time profiles and spectra .
the launch of _ swift _ and the possibility of continuously following a burst from the prompt to its afterglow stages , presented a number of surprises in variance with the earlier models ; a partial list includes ( a ) the steep decline in x - ray flux shortly after ( or perhaps in the transition to the afterglow from ) the prompt grb phase ( b ) an unexpectedly slow decline following this steep early phase and ( c ) the delayed large flares in the x - ray band @xcite thousands of seconds since the beginning of the burst .
it appears , however , that in the excitement of the _ swift _ results the issue of precise definitions for the prompt and afterglow grb stages and the transition from one to the other has been neglected . in this respect ,
the supercritical pile " model , precisely because it is founded on well defined and specific physical processes , does offer an answer to the previous question .
the remaining issue , then , is to check whether this answer is in fact consistent with existing ( or future ) observations . the kinematic threshold associated with the model ( eq . [ kinematic ] ) , in conjunction with the slowing down of the rbw as it accumulates mass from the surrounding medium ,
imply that the injection of electrons ( and hence the grb ) should terminate as soon as @xmath6 drops below the value that satisfies the kinematic threshold condition .
however , in the presence of an accelerated population of relativistic protons that extends to energies much higher than @xmath87 , i.e. the energy associated with the ( quasi-)thermal postshock population , the injection of relativistic electrons will continue even for @xmath88 ; this is because the the non - thermal tail " in the proton distribution guarantees the presence of protons with sufficiently high energies @xmath89 which can fulfill the more general threshold condition @xmath90 . under these conditions
the model predicts that electron injection will continue at a well defined energy but given in this case by the more general kinematic threshold of eq .
( [ kinematic0 ] ) rather than at @xmath91 .
herein we put forward the _
( detector sensitivity independent ) _ proposal that _ the transition from the prompt to the afterglow stage takes place when the rbw lorentz factor drops below that of eq .
( [ kinematic ] ) or equivalently when @xmath92 .
_ using the arguments presented in 2.1 , one can now estimate the energies at which the spectral peaks discussed there will occur under these new circumstances . because most of the grb phenomenology covers the peak associated with the middle , bulk - comptonized component
we restrict ourselves to this component . following the arguments presented earlier , this peak will now be ( in the lab frame ) at an energy @xmath93 < 1 \label{epeak2}\ ] ] + where the threhold condition @xmath94 has been used in the last step .
according to eq .
( [ epeak2 ] ) above , the value of the peak energy of the @xmath50 spectra is time dependent and presumably decreases with decreasing @xmath95 .
the decrease in time can be determined once a relation between @xmath96 and @xmath6 is found .
such a relation can be obtained from the threshold condition ( eq . [ kinematic0 ] ) @xmath97 and it is @xmath98^{1/3}$ ] . for magnetic field in equipartition with the postshock plasma @xmath99 , yielding @xmath100 $ ] , where @xmath39 is the ambient ( preshock ) density in @xmath80 .
finally , the relation between @xmath1 and @xmath95 reads @xmath101 ^ 2 ~. \label{epeak3}\ ] ] therefore , one of the consequences of the electron injection process and spectrum formation within the present model is that the value of the energy of peak emission @xmath1 should decrease with time as the grb passes from the prompt emission to its afterglow stage .
( [ epeak3 ] ) provides a specific dependence of this energy on @xmath6 and therefore on time . for the standard " time dependence of @xmath6 for rbw propagation through a medium of constant density ( @xmath102 ) and for magnetic field in equipartition , @xmath103 .
clearly the late time evolution favors observations by detectors which are more sensitive to lower energies , as indeed seems to be the case with the _ swift _ observations . .
besides a relativistic maxwellian of the form @xmath104 with @xmath82 , there is also a power law ` tail ' to the spectrum with normalizations ( relative to the peak of the maxwellian ) 0.1 ( solid line ) , 0.01 ( dashed line ) and 0.001 ( dot - dashed line ) .
[ protdf ] ] one of the surprising features of the grb light curves obtained by _
swift _ have been the steep decay of their x - ray flux , following the end of their prompt emission ( or more specifically their becoming too faint for the bat but not for the xrt ) .
generally , the models predicted flux decay @xmath105 @xcite,@xcite , consistent with the observations of the earlier afterglows ; steeper decrease in flux , attributed to emission from angles ( between the fluid motion and the observer s line of sight ) @xmath106 @xcite were also considered leading to x - ray flux profiles @xmath107 .
however , the new observations exhibited decay rates as steep as @xmath108 @xcite , in clear disagreement with the theory .
the model we have presented here incorporates the possibility of producing such steeply falling light curves .
the essence behind such behavior within the present model lies in the form of the proton distribution function behind the shock .
crucial for the extension of the model into the afterglow is the presence of a power - law tail " in addition to the quasi - maxwellian distribution of the bulk of the protons .
figure [ protdf ] shows the type of distribution we have been advocating in this section .
we also assume that as the shock slows down the shape of the distribution function ( i.e. the relative normalization of the power law and the maxwellian ) remains invariant and only shifts to lower energies .
as the rbw slows down , the critical value of the proton energy necessary to fulfill the pair production threshold increases and as it sweeps past the maximum of the maxwellian and into the power law section , the sharp drop in the proton density manifests as a sharp drop in the resulting photon flux . the corresponding change in the flux due to this effect
is given in figure [ flux ] , with the different curves representing the flux produced by the corresponding proton distribution of figure [ protdf ] .
it is apparent that the model has sufficient freedom to account for this sort of observation . as a function of time assuming @xmath109 constant for @xmath110 seconds and @xmath111 for @xmath112 seconds .
the different curves correspond to the proton distributions with the power law normalizations given in fig .
[ protdf ] .
there is an apparent very steep drop of emission with time beginning at @xmath113 sec .
the discovery of transients with @xmath1 at energies smaller than those of the classic grb @xcite put in question the characteristic " value of @xmath1 indicated by @xcite raised the issue of their nature and their relation to the classic " grb .
this discovery and the restriction of the value of @xmath1 within the supercritical pile " model to range near the electron rest mass seemed to effectively rule out the specific model ( and for that matter to any model that would produce a limited range in the value of @xmath114 ) .
a resolution to this issue was suggested in @xcite , who proposed that the reduced value of @xmath1 and the reduced peak flux of xrfs is related to the angle of the observer relative to that of the velocity of the rbw , with the xrfs being observed at large such angles .
this ( and any similar ) interpretation , however , tacitly assumes the presence of a unique " value for @xmath115 for a rbw moving along the observer s line of sight , a feature that is not accounted for by the typical grb model .
however , this particular account of the xrf - grb unification has been challenged by the data of xrf 050416a @xcite .
the authors of this work pointed out that an orientation unification between the classic grbs and xrfs implies that the source flux should increase as the rbw slows down since a larger fraction of the emitted luminosity would now be emitted into the observer s line of sight .
however , this particular event showed no such tendency , even though it was observed sufficiently early on so that this feature would not have been missed .
furthermore , it was shown that with an @xmath116 kev , it nicely filled the gap between grb and xrf in the amati relation @xcite .
the extension of the supercritical pile " model with the additional consideration of the high energy proton tail " suggested in the previous section can satisfy both these above constraints : for an initial rbw lorentz factor smaller than that implied by the threshold condition of eq . ( [ kinematic ] ) , the production of pairs will take place by the interaction of the photons not with the protons of the maxwellian part of the distribution but with the relativistic protons of its power law section . as indicated above , in this case
, the value of @xmath1 will be smaller than the electron rest mass the lower @xmath95 lies below the threshold value necessary to pair produce with the protons of lorentz factor @xmath6 ( i.e. those at the peak of the maxwellian distribution ) .
in fact , for field in equipartition with the protons , this value should be @xmath117 , so that for @xmath118 this value will be in the range 20 - 40 kev , even for a rbw moving along the observer s line of sight .
also , because of the smaller value of the rbw lorentz factor @xmath6 the total luminosity and radiated energy will be generally smaller , in agreement with the xrf observations .
one of the main aspects of the model described herein is that of the mirror " whose purpose is to scatter the synchrotron photons so that they can be re - intercepted ( blue - shifted now by @xmath119 ) by the rbw . to this point
it has been assumed that the albedo of this mirror is very close to 1 , this being the reason that it does not figure in the expressions of the thresholds ( eqs .
[ kinematic],[dynamic ] ) .
it is easy to see that a smaller value for the mirror s albedo will not change the kinematic threshold which is only concerned with the photon energies following the scattering by the mirror and the rbw .
however , the dynamic threshold cares about the number of photons available to be scattered . assuming that the fraction of the photons that scatter is proportional to the value of the mirror s " albedo @xmath120 , then the corresponding expression for the dynamic threshold becomes @xmath121 the issue of the nature of the mirror " and the effect of its kinematic state on the thresholds has been discussed in detail in @xcite .
one can easily see that an outlfowing mirror " will lead to an increase in the value of the lf of the rbw @xmath6 necessary to fulfill the kinematic threshold ; however , the energy of the bulk comptonized component of the resulting spectra will still appear , in the lab frame , at @xmath122 ( assuming that the process operates near threshold ) , thus preserving this feature of the model .
the simplest assumption concerning the mirror " is that it is simply due to thompson scattering on the electrons of the circumburst matter ( it appears unlikely that any matter would not be fully ionized in the intense grb photon field ) . in this case ,
@xmath123 , modifying the dynamic threshold to @xmath124 , which for @xmath125 @xmath80 and @xmath126 cm yields @xmath127 .
these values of the density and radius are consistent with those encountered in a spherical wind of mass loss @xmath128 m@xmath129 yr@xmath130 and velocity @xmath131 cm / s , parameters consistent with those associated with the wolf - rayet stars whose collapse presumably leads to a grb .
the process responsible for producing the radiation observed in the _ batse _ and _ swift _ energy bands within this model , i.e. the bulk comptonization of the synchrotron radiation reflected in the mirror " , provides the possibility of narrow ` spikes ' in the grb light curves , the presence of lags between hard and soft photons in these spikes and allows for rough estimates of these timing features : the synchrotron photons emitted by the rbw are at a distance @xmath132 ahead of it and their scattering in the mirror " produces ( in the infinitely thin mirror approximation ) a photon shell of width @xmath133 ; the bulk comptonized component is produced as these photons are swept by the rbw ; this sweeping " will take place on a time scale of order @xmath134 sec ( and proportionally longer by @xmath135 if the mirror " thickness @xmath136 is larger than @xmath133 ) .
therefore , the model can produce fine structure in the grb light curves by a process which could be considered as a variant of that of ` internal shocks ' , while in reality involving only external ones .
in addition , photons arriving to the observer at slightly different paths after these repeated scatterings will have slightly different energies resulting to lags of the same order of magnitude , in general agreement with observations @xcite .
we have presented above the outline ( and some details ) of a model that provides a well defined procedure for converting the kinetic energy of the rbw of grb into relativistic electrons and then into photons . to the best of our knowledge ,
this constitutes the first attempt to provide a detailed model of the dissipation ( i.e. the conversion to radiation ) of the kinetic energy of the grb blast waves . in this respect
the highly relativistic state of these flows is instrumental in effecting the dissipation of proton energy .
furthermore , the kinematic threshold of the reactions involved translates to a minimum energy in the injected electron distribution , which in turn leads naturally to a peak in the @xmath50 grb spectra that lies ( because of the details of the same kinematics ) in the vicinity of the electron rest mass , in agreement with observation .
the model provides also for a well defined demarkation between the prompt and the afterglow emissions based on the section of the proton distribution function which contributes to the electron injection : the prompt emission terminates when the injection shifts ( because of the decrease in the rgw lorentz factor ) from the ( relativistic ) maxwellian to the power law section of that distribution .
the kinematics of the pair production and spectral formation then imply that the energy of peak emission @xmath1 shifts to lower energies , also in general agreement with observation . while our calculations todate
have been limited to the @xmath137 reaction , the same photons could also produce pions , whose radiation can be added to the calculations , which become somewhat more complicated but without fundamental changes of the main point of the model . however , the possibility of photo - pion production can lead to a qualitatively new process , namely the production of neutrinos of energies that can be easily calculated : a rough estimate of the neutrino energy is @xmath138 the enery of the proton ; given that the protons of the rbw frame have energies @xmath139 and in the lab frame @xmath140 , the lab neutrino energy will be @xmath141 tev @xmath142 , a value of interest for the present and upcoming neutrino telescopes .
one should note that this estimate does not consider the possibility of the power - law tail in the spectrum considered above .
should that be present , then correspondingly higher energies are possible .
finally , this model can produce light curves with peaks of very narrow duration , despite the large size of the emission region , provided that the scattering medium has inhomogeneities of sufficiently short longitudinal dimensions .
robust as it is , this model can not ( as yet at least ) address all questions pertaining to the physics of grb , however , the grb phenomenology is sufficiently diverse that addressing a fraction only of the open issues constitutes concrete progress .
this work has been supported by the program pythagoras at the univsrsity of athens and an integral go grant at gsfc and umbc . | we present the general notions and observational consequences of the supercritical pile " grb model ; the fundamental feature of this model is a detailed process for the conversion of the energy stored in relativistic protons in the grb relativistic blast waves ( rbw ) into relativistic electrons and then into radiation . the conversion is effected through the @xmath0 reaction , whose kinematic threshold is imprinted on the grb spectra to provide a peak of their emitted luminosity at energy @xmath1@xmath2 mev in the lab frame .
we extend this model to include , in addition to the ( quasi)thermal relativistic post - shock protons an accelerated component of power law form .
this component guarantees the production of @xmath3pairs even after the rbw has slowed down to the point that its ( quasi)thermal protons can not fulfill the threshold of the above reaction . we suggest that this last condition marks the transition from the prompt to the afterglow grb phase .
we also discuss conditions under which this transition is accompanied by a significant drop in the flux and could thus account for several puzzling , recent observations .
finally , we indicate that the same mechanism applied to the late stages of the grb evolution leads to a decrease in @xmath1@xmath4 , a feature amenable to future observational tests .
[ 2001/04/25 1.1 ( pwd ) ] |
this project was in part supported by national basic research program of china under grant no .
2003cb716300 and by nnsfc under grant no .
90403032 . 99 a.g .
riess et al . , astron .
j. * 116 * ( 1998 ) 1009 ; s. perlmutter et al .
, astrophys .
j. * 517 * ( 1999 ) 565 .
spergel et al . , astrophys .
* 148 * ( 2003 ) 175 .
m. tegmark et al .
, phys . rev .
* d69 * ( 2004 ) 103501 . b. ratra and p.j.e .
peebles , phys . rev .
* d37 * ( 1988 ) 3406 ; c. wetterich , nucl . phys . * b302 * ( 1988 ) 668 ; r. caldwell , r. dave and p.j .
steinhardt , phys .
* 80 * ( 1998 ) 1582 .
i. zlatev , l. wang and p.j .
steinhardt , phys .
* 82 * ( 1999 ) 896 ; p.j .
steinhardt , l. wang and i. zlatev , phys . rev .
* d59 * ( 1999 ) 123504 .
ferreira and m. joyce , phys . rev .
* d58 * ( 1998 ) 023503 ; p. binetruy , phys . rev .
* d60 * ( 1999 ) 063502 ; p. brax and j. martin , phys .
* b468 * ( 1999 ) 40 ; p. binetruy , int .
* 39 * ( 2000 ) 1859 ; p. brax , j. martin and a. riazuelo , phys .
* d62 * ( 2000 ) 103505 ; s.m .
carroll , phys.rev.lett .
81 ( 1998 ) 3067 ; c. kolda and d.h .
lyth , phys .
b458 * ( 1999 ) 197 . v. sahni and a. starobinsky , int
d9 * ( 2000 ) 373 ; j.e .
peebles and b. ratra , rev .
mod . phys .
* 75 * ( 2003 ) 559 ; t. padmanabhan , phys . rept . * 380 * ( 2003 ) 235 ; e.j .
copeland , m. sami and s. tsujikawa , hep - th/0603057 .
caldwell , phys .
b545 * ( 2002 ) 23 . s. nojiri and s.d .
odintsov , phys . lett .
* b562 * ( 2003 ) 147 ; s. nojiri and s.d .
odintsov , phys . lett .
* b565 * ( 2003 ) 1 ; p. singh , m. sami and n. dadhich , phys.rev .
d68 ( 2003 ) 023522 ; r.g .
cai and a. wang , jcap * 0503 * ( 2005 ) 002 ; y.s .
piao and e. zhou , phys . rev .
* d68 * ( 2003 ) 083515 ; y.s .
piao and y.z .
zhang , phys . rev .
* d70 * ( 2004 ) 063513 ; f.s.n .
lobo , phys . rev . *
d71 * ( 2005 ) 084011 ; f.s.n .
lobo , phys . rev .
* d71 * ( 2005 ) 124022 ; h.m .
sadjadi , phys . rev . *
d73 * ( 2006 ) 063525 ; j.m .
aguirregabiria , l.p .
chimento and r. lazkoz , phys . rev .
* d70 * ( 2004 ) 023509 ; l.p .
chimento and r. lazkoz , int .
d14 * ( 2005 ) 587 ; l.a .
urena - lopez , jcap * 0509 * ( 2005 ) 013 ; l. amendola , s. tsujikawa and m. sami , phys .
b632 * ( 2006 ) 155 ; a.a .
coley , s. hervik and w.c .
lim , gr - qc/0512085 ; a.a .
coley , s. hervik , j. latta , astro - ph/0503169 ; m. kaplinghat and a. rajaraman , astro - ph/0601517 ; f.s.n .
lobo , gr - qc/0603091 ; c.j .
gao and s.n .
zhang , hep - th/0604114 ; v.a .
rubakov , hep - th/0604153 ; z.g .
huang , h.q .
lu and w. fang , hep - th/0604160 ; a.k .
sanyal , astro - ph/0605388 .
p.h.frampton , phys.lett .
b555 ( 2003 ) 139 .
nilles , phys . rep . * 110 * ( 1984 ) 1 .
onemli and r.p .
woodard , class .
* 19 * ( 2002 ) 4607 ; v.k .
onemli and r.p .
woodard , phys .
* d70 * ( 2004 ) 107301 .
pollock , phys . lett .
* b215 * ( 1988 ) 635 .
d.f.torres , phys.rev .
d66 ( 2002 ) 043522 ; v. faraoni , phys . rev .
* d68 * ( 2003 ) 063508 ; l. perivolaropoulos , jcap * 0510 * ( 2005 ) 001 ; s.m .
carroll et al . , phys . rev .
* d71 * ( 2005 ) 063513 ; j. martin , c. schimd and j. uzan , phys .
rev . lett .
* 96 * ( 2006 ) 061303 .
s. das , p.s .
corasaniti and j. khoury , phys .
* d73 * ( 2006 ) 083509 .
guo , y.s .
piao and y.z .
zhang , phys .
b594 * ( 2004 ) 247 .
caldwell , m. kamionkowski and n.n .
weinberg , phys .
* 91 * ( 2003 ) 071301 ; v. faraoni and w. israel , phys . rev .
* d71 * ( 2005 ) 064017 ; v. faraoni , class .
* 22 * ( 2005 ) 3235 ; m. bouhmadi - lopez and j.a.j .
madrid , jcap * 0505 * ( 2005 ) 005 ; s. nojiri , s.d .
odintsov and s. tsujikawa , phys .
rev . * d71 * ( 2005 ) 063004 ; s. nojiri and s.d .
odintsov , phys . rev . * d72 * ( 2005 ) 023003 ; m. sami , a. toporensky , p.v .
tretjakov and s. tsujikawa , phys . lett . *
b619 * ( 2005 ) 193 ; p. wu and h. yu , nucl . phys .
* b727 * ( 2005 ) 355 ; s. nesseris , l. perivolaropoulos , phys . rev .
* d70 * ( 2004 ) 123529 ; m.p .
dabrowski and t. stachowiak , annals phys .
* 321 * ( 2006 ) 771 ; e.m .
barboza jr .
lemos , gr - qc/0606084 ; y.h .
wei , astro - ph/0607359 .
a. vikman , phys . rev .
* d71 * ( 2005 ) 023515 . b. feng , x.l .
wang and x. zhang , phys .
* b607 * ( 2005 ) 35 .
guo , y.s .
piao , x. zhang and y.z .
zhang , phys .
* b608 * ( 2005 ) 177 .
m. li , b. feng and x. zhang , jcap * 0512 * ( 2005 ) 002 .
simon , phys . rev . * d41 * ( 1990 ) 3720 ; a. elizer and r.p .
woodard , nucl .
b325 * ( 1989 ) 389 ; t.g .
erler and d.j .
gross , hep - th/0406199 .
arefeva , a.s . koshelev and s.yu .
vernov , phys . rev . *
d72 * ( 2005 ) 064017 . b. feng , m. li , y.s .
piao and x. zhang , phys .
* b634 * ( 2006 ) 101 ; j.q .
xia , b. feng and x. zhang , mod .
phys . lett . *
a20 * ( 2005 ) 2409 .
v. sahni and y. shtanov , jcap * 0311 * ( 2003 ) 014 ; h. stefancic , phys . rev . * d71 * ( 2005 ) 124036 ; j. sola and h. stefancic , phys . lett .
* b624 * ( 2005 ) 147 ; x.f .
zhang , h. li , y.s .
piao and x. zhang , astro - ph/0501652 .
andrianov , f. cannata and a.y .
kamenshchik , phys .
* d72 * ( 2005 ) 043531 ; b. wang , y. gong and e. abdalla , phys.lett .
* b624 * ( 2005 ) 141 ; x. zhang and f.q .
wu , phys .
* d72 * ( 2005 ) 043524 ; z. chang , f.q . wu and x. zhang , phys
b633 * ( 2006 ) 14 ; p. wu and h. yu , int . j. mod
* d14 * ( 2005 ) 1873 ; h. zhang and z.h .
zhu , phys . rev .
* d73 * ( 2006 ) 043518 ; w. zhao and y. zhang , class .
* 23 * ( 2006 ) 3405 ; w. zhao and y. zhang , phys .
* d73 * ( 2006 ) 123509 ; r. lazkoz , g. leon , astro - ph/0602590 ; f. cannata and a.y .
kamenshchik , gr - qc/0603129 ; l.p .
chimento and r. lazkoz , astro - ph/0604090 ; m. alimohammadi and h.m .
sadjadi , gr - qc/0608016 .
h. wei , r.g .
cai and d.f .
zeng , class .
* 22 * ( 2005 ) 3189 ; h. wei and r.g .
cai , phys .
rev . * d72 * ( 2005 ) 123507 ; h. wei and r.g .
cai , phys . lett .
* b634 * ( 2006 ) 9 ; h. wei and r.g .
cai , phys . rev .
* d73 * ( 2006 ) 083002 ; m. alimohammadi and h.m .
sadjadi , phys . rev .
* d73 * ( 2006 ) 083527 .
g. kofinas , g. panotopoulos and t.n .
tomaras , jhep * 0601 * ( 2006 ) 107 ; r.g .
cai , y. gong and b. wang , jcap * 0603 * ( 2006 ) 006 ; p.s . apostolopoulos and n. tetradis ,
phys . rev .
* d74 * ( 2006 ) 064021 .
caldwell and e.v .
linder , phys .
* 95 * ( 2005 ) 141301 .
scherrer , phys . rev . *
d73 * ( 2006 ) 043502 .
t. chiba , phys . rev .
* d73 * ( 2006 ) 063501 .
m. sami and a. toporensky , mod .
* a19 * ( 2004 ) 1509 ; l. perivolaropoulos , phys . rev . *
d71 * ( 2005 ) 063503 ; h.m .
sadjadi and m. alimohammadi , phys . rev .
* d74 * ( 2006 ) 043506 ; j. kujat , r.j .
scherrer and a.a . sen ,
phys . rev .
* d74 * ( 2006 ) 083501 . | the @xmath0 plane , defined by the equation of state parameter for the dark energy and its derivative with respect to the logarithm of the scale factor , is useful to the study of classifying the dynamical dark energy models . in this note , we examine the evolving behavior of the two - field quintom models with @xmath1 crossing the @xmath2 barrier in the @xmath0 plane .
we find that these models can be divided into two categories , type a quintom in which @xmath1 changes from @xmath3 to @xmath4 and type b quintom in which @xmath1 changes from @xmath4 to @xmath3 as the universe expands .
recent observations of type ia supernovae suggest that the expansion of the universe is accelerating and that two - thirds of the total energy density exists in a dark energy component with negative pressure @xcite .
in addition , measurements of the cosmic microwave background @xcite and the galaxy power spectrum @xcite also indicate the existence of the dark energy .
the simplest candidate for the dark energy is a cosmological constant @xmath5 , which has pressure @xmath6 .
specifically , a reliable model should explain why the present amount of the dark energy is so small compared with the fundamental scale ( fine - tuning problem ) and why it is comparable with the critical density today ( coincidence problem ) . the cosmological constant suffers from both these problems .
one possible approach to constructing a viable model for dark energy is to associate it with a slowly evolving and spatially homogeneous scalar field @xmath7 , called `` quintessence '' @xcite .
such a model for a broad class of potentials can give the energy density converging to its present value for a wide set of initial conditions in the past and possess tracker behavior ( see , e.g. , @xcite for a review ) .
quintessence models describe the dark energy with a time - varying equation of state parameter , @xmath1 , the ratio of its pressure to energy density and @xmath8 . recently
, such models have been extended to phantom dark energy with @xmath9 @xcite ( see for example @xcite and references therein ) .
the physical sources for phantom fields with strongly negative pressure may be looked for in string theory @xcite and supergravity @xcite .
such fields may arise from quantum effects on a locally de sitter background @xcite .
they may also be present in modified gravity theories , such as higher order theories @xcite and scalar - tensor theories @xcite . coupled quintessence with dark matter
may also lead to @xmath9 @xcite .
phantom field with a negative kinetic term may be a simplest implementing , in which the weak energy condition is violated .
it has been shown that such models possess the attractor behavior similar to quintessence models @xcite . if @xmath10 in an expanding universe , the energy density of the dark energy increases with time , which leads to unwanted future singularity called `` big rip '' @xcite .
thus from this point of view the transition from @xmath11 to @xmath10 or vice versa would be desirable for the history of the universe . on the other hand , the analysis on the properties of dark energy from the recent observations mildly favors models with @xmath1 crossing @xmath12 in the near past . from the theoretical viewpoint ,
it is necessary to explore possibilities for dark energy with @xmath1 crossing @xmath12 .
however , neither quintessence nor phantom can fulfill this transition .
the similar conclusion has also been obtained for the k - essence models @xcite .
quintom models easily provide a way to realize this transition @xcite .
the quintom fields may be associated with some higher derivatives terms @xcite derived from fundamental theories , for instance due to the quantum corrections or the non - local physics in the string theory @xcite .
they may also arise from a slowly decaying d3-brane in a local effective approximation @xcite .
interestingly , the quintom models differ from the quintessence or phantom in evolution and the determination of the fate of universe @xcite .
there exist lots of interests in the literature presently in building of quintom - like models @xcite , such as hessence models @xcite and brane models @xcite .
recently , caldwell and linder examined the evolving behavior of quintessence models of dark energy in the @xmath0 phase plane , where @xmath13 is the time derivative of @xmath1 with respect to the logarithm of the scale factor @xmath14 , and showed that these models occupy the thawing and freezing regions in the phase plane @xcite .
more recently , these results were extended to a more general class of quintessence models with a monotonic potential @xcite and phantom dark energy @xcite . in this note
we extend these studies with single - field quintessence or phantom models to two - field quintom models .
our results show that there exist two types of quintom models according to the evolving behavior around @xmath2 .
moreover , we plot the trajectories numerically for the two types in the @xmath0 plane .
let us consider the following model which contains a negative - kinetic scalar field @xmath7 ( phantom ) and a normal scalar field @xmath15 ( quintessence ) : @xmath16,\end{aligned}\ ] ] where @xmath17 the gravitational coupling , @xmath18 the scalar potential and @xmath19 the lagrangian density of matter fields . in a flat frw cosmology
the evolutions of the fields are governed by @xmath20 where @xmath21 and @xmath22 . in what follows
we assume that there is no direct coupling between the phantom field and the normal scalar field , i.e. , @xmath23 .
then the effective equation of state @xmath1 is given by @xmath24 where @xmath25 is the density parameter of dark energy , @xmath26 and @xmath27 .
for a model with a normal scalar field , the equation of state @xmath28 .
the toy model of a phantom energy component with a negative kinetic term possesses an equation of state @xmath10 . in our model , eq .
( [ es ] ) implies @xmath28 when the quintessence component @xmath15 dominates the density of the universe and @xmath9 when the phantom component @xmath7 dominates . by using the equations of motion ( [ em1 ] ) and ( [ em2 ] ) , the derivative of ( [ es2 ] ) with respect to @xmath29 can be rewritten as
@xcite @xmath30 by using the following relations @xmath31 the expression for @xmath13 becomes @xmath32\,.\end{aligned}\ ] ] for a phantom field @xmath7 climbing up its potential , the @xmath33 sign before @xmath34 depends on whether @xmath35 or @xmath36 , respectively . for a quintessence field
@xmath15 rolling down its potential , the @xmath33 sign before @xmath37 corresponds to @xmath38 or @xmath39 , respectively . according to the evolving behavior of @xmath1 around @xmath12
, the two - field quintom models of dynamical dark energy are classified into the following two types : type a quintom characterized by @xmath1 from @xmath8 to @xmath9 and type b quintom characterized by @xmath1 from @xmath9 to @xmath8 .
* type a quintom models : * in such models , the equation of state changes from @xmath8 to @xmath9 , i.e. , the universe evolves from a quintessence - dominated phase to a phantom - dominated phase .
therefore , the properties of the late - time attractor solution are determined by the phantom potential @xmath40 .
the cosmological evolution of the type a quintom model with @xmath41 was investigated in detail in ref .
@xcite . when the phantom component is dominated at late times , eq .
( [ weq ] ) reduces to @xmath42,\ ] ] which has three critical points @xmath43 , @xmath2 and @xmath44 .
it is easily shown that the scaling solution @xmath44 is the stable attractor of this type of models , i.e. , the ratio of kinetic to potential energy of the phantom field becomes a constant .
the top panels of fig . [ fig : f1 ] shows the evolving behavior of the models in the @xmath0 phase plane .
we shall next consider a general case in which the effective equation of state @xmath1 tends to @xmath12 at late times . in this case
, the ratio of kinetic to potential energy tends to zero @xcite .
the features of the behavior are virtually independent of the precise shape of the quintessence potential since the contribution of the quintessence component becomes negligible at late times compared to the phantom component .
for example we consider a positive power - law potential , which have been previously investigated in refs .
@xcite . for the following potential @xmath45 in the @xmath46 limit ,
we have @xmath47 . the critical point with @xmath2
is the late - time attractor , i.e. , the quintom field becomes ultimately frozen , as shown in the bottom panels of fig .
[ fig : f1 ] .
* type b quintom models : * in such models , the equation of state changes from @xmath9 to @xmath8 . in principle , for the two - field system consisting of quintessence and phantom , if the phantom initially dominates the universe , it will still dominate up to future , and thus the universe can not exit the phantom phase ( @xmath9 ) forever .
we take a simple potential @xmath48 for a illustration .
the relevant figure is plotted in the bottom panels of fig .
[ fig : f2 ] . in general the phantom field will climb up its potential , which makes its energy density increasing constantly during its evolution .
thus in order to implement the change of equation of state from @xmath9 to @xmath8 , a naive choice is taking the potential of phantom field zero , for example , considering the following potential @xmath49 whose evolving behavior is plotted in the top panels of fig . [
fig : f2 ] . in this case ,
the initial kinetic energy of phantom field is not zero , which naturally leads to the state equation @xmath9 of quintom system . but
with the expansion of the universe , the kinetic energy of phantom field will become negligible gradually while the energy of quintessence field will begin to dominate the universe .
thus we have @xmath8 .
however , it should be noted that here the energy of phantom field is negative , but the total energy of quintom system is still positive .
we do not want to deeply discuss the relevant physics with this case , and we here only emphasize that to obtain such a transition from @xmath9 to @xmath8 in two - field system ( quintessence+phantom ) , the above condition seems to be necessarily satisfied . in conclusion
, it is well - known that the two - field quintom models give a simplest realization to the dynamical dark energy with the equation of state parameter crossing the @xmath2 barrier , thus it is very interesting to explore the full evolving behavior of various two - field quintom models . in this note
, we have examined the evolving behavior of the two - field quintom models in the @xmath0 plane .
we find that these models can generally be divided into two categories , type a quintom models in which @xmath1 changes from @xmath3 to @xmath4 and type b quintom models in which @xmath1 changes from @xmath4 to @xmath3 , which has not been analyzed in detail before .
compared to the latter , the former is easily constructed since the energy density of the phantom field increases as the universe expands and ultimately dominates the universe .
the latter requires the phantom field with a flat potential or a potential with a maximum . |
the large hadron collider ( lhc ) will permit the exploration of physics at unprecedented energy scales toward the end of this decade , but it will also produce @xmath0 quarks more copiously than any other accelerator . if the hadrons containing these quarks can be identified , many questions we now face can be addressed , while undoubtedly others will arise . in this talk
i would like to give some examples of _ current _ questions in @xmath0 physics to which we would like answers .
others may well be more timely in the lhc era .
in section [ sec : ckm ] we review information on weak quark transitions as encoded in the cabibbo - kobayashi - maskawa ( ckm ) matrix .
we then discuss cp asymmetries in @xmath1 decays to cp eigenstates ( section [ sec : eig ] ) and to self - tagging modes ( `` direct asymmetries , '' section [ sec : dir ] ) .
rare radiative @xmath1 decays , mentioned briefly in section [ sec : rad ] , provide useful information on possible new physics .
hadron colliders such as the lhc are the tool of choice for the study of strange @xmath1 ( @xmath2 ) properties and decays ( section [ sec : bs ] ) .
excited states of @xmath1 and @xmath2 mesons , for which there have been interesting parallel developments in the charm sector , are discussed in section [ sec : exc ] .
the search for heavier quarks which which could mix with the @xmath0 quark is noted in section [ sec : exo ] , while section [ sec : sum ] concludes .
the relative strengths of charge - changing weak quark transitions are illustrated in fig .
[ fig : trans ] .
this pattern is one of the central mysteries of particle physics , along with the values of the quark masses .
we shall not address its deeper origin here , but will seek better knowledge of strengths and phases of the transitions , to see whether all weak phenomena including cp violation can be described satisfactorily via this pattern . .
breadths of horizontal lines denote estimated errors for masses .
[ fig : trans],height=288 ] the interactions in fig .
[ fig : trans ] may be parametrized by a unitary @xmath3 matrix , the cabibbo - kobayashi - maskawa ( ckm ) matrix . a convenient form @xcite , unitary to sufficiently high order in a small quantity @xmath4 , is [ eqn : wp ] v_ckm = , where @xmath5 and @xmath6 .
the columns refer to @xmath7 and the rows to @xmath8 .
the parameter @xmath9 @xcite is @xmath10 , where @xmath11 is the cabibbo angle .
the value @xmath12 , obtained from @xmath13 decays , indicates @xmath14 , while @xmath15 , obtained from @xmath16 decays , implies @xmath17 .
we shall generally use the ckm parameters quoted in ref .
@xcite .
the unitarity of the ckm matrix implies that the scalar product of any column with the complex conjugate of any other column is zero ; for example , @xmath18 .
if one divides by @xmath19 , this relation becomes equivalent to a triangle in the complex @xmath20 plane , with vertices at ( 0,0 ) ( angle @xmath21 ) , ( 1,0 ) ( angle @xmath22 ) , and @xmath23 ( angle @xmath24 ) .
the triangle has unit base and its other two sides are @xmath25 ( opposite @xmath22 ) and @xmath26 ( opposite @xmath27 ) .
the result is shown in fig.[fig : ut ] .
, height=144 ] in addition to the direct measurements of ckm parameters mentioned above , flavor - changing loop diagrams provide a number of indirect constraints .
cp - violating @xmath28@xmath29 mixing is dominated by the second - order - weak virtual transition @xmath30 with virtual @xmath31 and @xmath32 intermediate states , and thus constrains im@xmath33 , leading to a hyperbola in the @xmath34 plane .
@xmath35@xmath36 mixing is similarly dominated by @xmath31 and @xmath32 in the loop diagram for @xmath37 , and thus constrains @xmath38 and hence @xmath39 . by comparing @xmath2@xmath40 and @xmath35@xmath36 mixing , one can reduce dependence on unknown matrix elements and learn a lower limit on @xmath41 or an upper limit on @xmath39 . the range of parameters allowed at 95% c.l .
@xcite is @xmath42 , @xmath43 ( but see , e.g. , @xcite for more a more optimistic view of our present knowledge ) .
one can learn ckm phases from decays of neutral @xmath1 mesons to cp eigenstates @xmath44 , where @xmath45 , @xmath46 . as a result of @xmath35@xmath47 mixing ,
a state which is @xmath35 at proper time @xmath48 will evolve into one , denoted @xmath49 , which is a mixture of @xmath35 and @xmath47 .
thus there will be one pathway to the final state @xmath44 from @xmath35 through the amplitude @xmath50 and another from @xmath47 through the amplitude @xmath51 , which acquires an additional phase @xmath52 through the mixing .
the interference of these two amplitudes can differ in the decays @xmath53 and @xmath54 , leading to a time - integrated rate asymmetry _ cp as well as to time - dependent rates \ { c + } ~e^- t [ 1 _ f m t _ f m t ] , where _ f , s_f , e^-2 i , where @xmath55 .
more details may be found in refs .
i now note some specific cases . for this decay
one has @xmath57 .
one finds that the time - integrated asymmetry @xmath58 is proportional to @xmath59 .
using this and related decays involving the same @xmath60 subprocess , babar @xcite finds @xmath61 while belle @xcite finds @xmath62 .
the two values agree with each other ; the world average @xcite is @xmath63 , consistent with other determinations @xcite . here
two amplitudes contribute to the decay : a `` tree '' @xmath65 and a `` penguin '' @xmath66 .
the decay amplitudes are a = - ( |t|e^i + |p| e^i ) , |a = - ( |t|e^-i + |p| e^i ) , where @xmath67 is the relative @xmath68 strong phase .
the asymmetry @xmath58 would be proportional to @xmath69 if the penguin amplitude could be neglected .
however , one must account for its contribution .
an isospin analysis @xcite of @xmath1 decays to @xmath70 , @xmath71 , and @xmath72 separates the contributions of decays involving @xmath73 and @xmath74 final states .
information can then be obtained on both strong and weak phases . since the branching ratio of @xmath35 to @xmath72 may be very small , of order @xmath75
, i shall discuss instead methods @xcite in which flavor su(3 ) symmetry is used to estimate the penguin contribution @xcite .
the tree amplitude for @xmath76 involves @xmath77 , with the spectator @xmath78 quark combining with @xmath79 to form a @xmath80 .
its magnitude is @xmath81 ; its weak phase is arg(@xmath82 ; by convention its strong phase is 0 .
the penguin amplitude involves the flavor structure @xmath83 , with the final @xmath84 pair fragmenting into @xmath70 .
its magnitude is @xmath85 .
the dominant @xmath86 contribution in the loop diagram for @xmath87 can be integrated out and the unitarity relation @xmath88 used .
the @xmath89 contribution can be absorbed into a redefinition of the tree amplitude , after which the weak phase of the penguin amplitude is 0 ( mod @xmath90 ) . by definition ,
its strong phase is @xmath67 .
the time - dependent asymmetries @xmath91 and @xmath92 specify both @xmath93 ( or @xmath94 ) and @xmath67 , if one has an independent estimate of @xmath95 .
one may obtain @xmath85 from @xmath96 using flavor su(3 ) @xcite and @xmath81 from @xmath97 using factorization @xcite .
an alternative method @xcite uses the measured ratio of the @xmath98 and @xmath64 branching ratios to constrain @xmath95 .
i shall discuss the first method .
in addition to @xmath91 and @xmath92 , a useful quantity is the ratio of the @xmath64 branching ratio @xmath99 ( averaged over @xmath35 and @xmath47 ) to that due to the tree amplitude alone : r _ = 1 + 2 | - 2 |p / t| .
the value of @xmath100 is specified to within a few degrees ; we shall take it to have its central value of @xmath101 .
the value of @xmath95 ( updating @xcite ) is @xmath102 . taking the central value
, one can plot trajectories in the ( @xmath103 ) plane as @xmath67 is allowed to vary from @xmath104 to @xmath90 .
the experimental situation regarding the time - dependent asymmetries is not yet settled .
as shown in table [ tab : sa ] , babar @xcite and belle @xcite obtain very different values , especially for @xmath105 . even if this conflict were to be resolved
, however , there is a possibility of a discrete ambiguity , since curves for different values of @xmath106 intersect one another .
the discrete ambiguity may be resolved with the help of @xmath107 , but the error is still too large to be helpful . at present values of @xmath108 are favored , but with large uncertainty .
it is not yet settled whether @xmath109 , corresponding to `` direct '' cp violation .
.values of @xmath91 and @xmath92 quoted by babar and belle and their averages .
here we have applied scale factors @xmath110 to the errors for @xmath91 and @xmath92 , respectively . [ tab : sa ] [ cols="^,^,^,^",options="header " , ] if the amplitude @xmath50 for a process receives two contributions with differing strong and weak phases , one can write a = a_1 + a_2 e^i e^i , |a = a_1 + a_2 e^-i e^i .
the cp - averaged decay rate is proportional to @xmath111 , while the cp asymmetry is a_cp = - . in the case of @xmath112 the rates and cp asymmetry suggest that @xmath113 .
details of this pattern and its implications for other processes are described in ref .
it is predicted there that @xmath114 ( below current upper bounds ) and that @xmath115 .
a number of processes in which a @xmath1 or @xmath2 decays to final states with photons or lepton pairs are particularly sensitive to non - standard physics .
an example is @xmath116 , for which the standard model predicts @xmath117 @xcite .
charged higgs boson exchanges or other effects could enhance this branching ratio significantly while respecting the constraint associated with the branching ratio for @xmath118 , which appears compatible with standard model predictions .
for a good discussion of this process and of @xmath119 see ref .
@xcite , as well as several presentations at the present conference @xcite . in the latter decay
the forward - backward asymmetries exhibit interesting behavior as a function of @xmath120 , with signs and a characteristic zero in the standard model which can be different in variant theories .
the ratio of the @xmath2@xmath121 mixing amplitude @xmath122 to the @xmath35@xmath36 mixing amplitude @xmath123 ( @xmath124 ) is given by = | while @xmath125 are numbers of order 1 expressing the degree to which the mixing amplitude can be calculated by saturating with vacuum intermediate states .
the latest lattice estimate of the ratio @xmath126 is @xmath127 @xcite .
we have taken a generous range |v_td| = a ^3|1 - |- i || = ( 0.8 0.2 ) a ^3 with @xmath128 and @xmath129 . with @xcite @xmath130 ps@xmath131 one then predicts m_s = 24 ps^-1 2 ^ 1
. the lower portion of this range is already excluded by the bound @xcite m_s > 14.4 ps^-1 ( 95% c.l . ) .
when @xmath122 is measured it is likely to be known fairly well immediately , and will constrain @xmath132 significantly . since the weak phase in @xmath134 is expected to be zero while that of @xmath2@xmath40 mixing is expected to be very small [ in the parametrization of eq .
( 1 ) an imaginary part im(@xmath135 was not written explicitly ] , one expects cp asymmetries to be only a few percent in the standard model for those @xmath2 decays dominated by this quark subprocess .
the @xmath136 final state is not a cp eigenstate but the even and odd cp components can be separated using an angular analysis .
the final states of @xmath137 and @xmath138 are cp - even so no such analysis is needed .
a comparison of time - dependent asymmetries in @xmath139 and @xmath140 @xcite allows one to separate out strong and weak phases and relative tree and penguin contributions . in @xmath139
the @xmath141 penguin amplitude is dominant , while the strangeness - changing tree amplitude @xmath142 is subsidiary . in @xmath143
it is the other way around : the @xmath144 tree amplitude dominates , while the @xmath83 penguin is cabibbo - suppressed .
the u - spin subgroup of su(3 ) , which interchanges @xmath145 and @xmath78 quarks , relates each amplitude in one process to that in the other aside from the ckm factors .
a potential problem with @xmath139 and @xmath64 is that the mass peaks will overlap with one another if analyzed in terms of the same final state ( e.g. , @xmath70 ) @xcite .
thus , in the absence of good particle identification , a variant on this scheme employing the decays @xmath147 and @xmath148 ( also related to one another by u - spin ) may be useful @xcite . for these final states
, kinematic separation may be easier .
a further variant is to study the time - dependence of @xmath139 while normalizing the penguin amplitude using @xmath149 @xcite .
the u - spin subgroup of su(3 ) allows one to relate many other @xmath2 decays besides those mentioned above to corresponding @xmath150 decays @xcite .
particularly useful are relations between cp - violating rate _
differences_. one thus will have the opportunity to perform many tests of flavor su(3 ) and to learn a great deal more about final - state phase patterns when a variety of @xmath2 decays can be studied .
one promising method for tagging the flavor of a neutral @xmath1 meson is to study the charge of the leading light hadron accompanying the fragmentation of the heavy quark .
this method was initially proposed by ali and barreiro @xcite to identify the flavor of a @xmath2 via the charge of the accompanying kaon .
it was utilized in refs .
@xcite to distinguish @xmath35 s from @xmath36 s .
an initial @xmath0 will fragment into a @xmath36 by `` dressing '' itself with a @xmath151 .
the accompanying @xmath78 , if incorporated into a charged pion , will end up in a @xmath80 . thus a @xmath80 is more likely to be `` near '' a @xmath36 than to a @xmath35 in phase space .
this correlation between @xmath80 and @xmath36 ( and the corresponding correlation between @xmath152 and @xmath35 ) is also what one would expect on the basis of non - exotic resonance formation .
thus the study of the resonance spectrum of the excited @xmath1 mesons which can decay to @xmath153 or @xmath154 is of special interest @xcite .
the lowest such mesons are the p - wave levels of a @xmath155 antiquark and a light ( @xmath156 or @xmath78 ) quark .
a new sensation has been reported by the babar collaboration @xcite and confirmed by cleo @xcite .
partial information on the p - wave levels of a charmed quark @xmath159 and an antistrange @xmath160 consists of candidates for @xmath161 and @xmath162 states at 2535 and 2572 mev @xcite .
these levels have narrow widths and are behaving as would be expected if the spin of the @xmath160 and the orbital angular momentum were coupled up to @xmath163 .
( one expects @xmath164-@xmath164 rather than @xmath165-@xmath166 coupling in a light - heavy system @xcite . )
if the @xmath167 states were fairly close to these in mass one would then expect another @xmath161 state and a @xmath168 state somewhere above 2500 mev . instead , a candidate for a @xmath168 @xmath169 state has been found around 2317 mev , with the second @xmath161 level around 2463 mev .
both are narrow , since they are too light to decay respectively to @xmath170 or @xmath171 .
they decay instead via the isospin - violating transitions @xmath172 and @xmath173 .
they are either candidates for @xmath174 molecules @xcite , or indications of a broken chiral symmetry which places them as positive - parity partners of the @xmath157 and @xmath175 negative - parity @xmath169 ground states @xcite .
indeed , the mass splittings between the parity partners appear to be exactly as predicted ten years ago @xcite .
potential - based quarkonium models have a hard time accommodating such low masses @xcite , there should exist _ non - strange _
@xmath167 @xmath176 and @xmath177 states , lower in mass than the @xmath178 states at 2422 and 2459 mev @xcite but quite broad since their respective @xmath179 and @xmath180 channels will be open . the study of such states will be of great interest since the properties of the corresponding @xmath1-flavored states will be useful in tagging the flavor of neutral @xmath1 mesons , as noted in the previous subsection .
if a strange antiquark can bind to a charmed quark in both negative- and positive - parity states , the same must be true for a strange antiquark and a @xmath0 quark .
one should then expect to see narrow @xmath182 and @xmath177 states with the quantum numbers of @xmath183 and @xmath184 but below those respective thresholds .
they should decay to @xmath185 and @xmath186 , respectively . to see such decays
one will need a multi - purpose detector with good charged particle and @xmath187 identification !
such detectors are envisioned for both the tevatron @xcite and the lhc @xcite .
might there be heavier quarks visible at hadron colliders ? at present we have evidence for three families of quarks and leptons belonging to 16-dimensional multiplets of the grand unified group so(10 ) ( counting right - handed neutrinos as a reasonable explanation of the observed oscillations between different flavors of neutrinos ) .
now , just as so(10 ) was pieced together from multiplets of su(5 ) with dimensions 1 , 5 , and 10 , we can imagine a still larger grand unified group whose smallest representation contains the 16-dimensional so(10 ) spinor .
such a group is the exceptional group e@xmath189 @xcite .
its smallest representation , of dimension 27 , contains a 16-dimensional spinor , a 10-dimensional vector , and a singlet of so(10 ) .
the 10-dimensional vector contains vector - like isosinglet quarks `` @xmath190 '' and antiquarks @xmath191 of charge @xmath192 and isodoublet leptons .
the so(10 ) singlets are candidates for sterile neutrinos , one for each family .
the new exotic @xmath190 quarks can mix with the @xmath0 quark and push its mass down with respect to the top quark @xcite .
troy andre and i are currently looking at signatures of @xmath193 production in hadron colliders , with an eye to either setting lower mass limits or seeing such quarks through their decays to @xmath194 , @xmath195 , and possibly @xmath196 .
the @xmath197 , for example , would be identified by its decays to @xmath198 , @xmath199 , or jet @xmath200 jet , while the higgs boson would show up through its @xmath201 decay if it were far enough below @xmath32 threshold .
the process @xmath56 has provided spectacular confirmation of the kobayashi - maskawa theory of cp violation , measuring @xmath100 to a few degrees .
now one is entering the territory of more difficult measurements .
the decay @xmath64 has great potential for giving useful information on @xmath106 .
one needs either a measurement of @xmath202 @xcite , probably at the @xmath75 level ( present limits @xcite are several times that ) , or a better estimate of the tree amplitude from @xmath203 @xcite .
indeed , such an estimate has been presented recently @xcite .
the babar and belle experimental cp asymmetries @xcite will eventually converge to one another , as did the initial measurements of @xmath204 using @xmath56 .
the @xmath205 decay can display new physics via special @xmath206 operators or effects on the @xmath207 penguin .
some features of any new amplitude can be extracted from the data in a model - independent way if one uses both rate and asymmetry information @xcite . while the effective value of @xmath204 in @xmath208 seems to differ from its expected value by more than @xmath209
, cp asymmetries in @xmath210 do not seem anomalous .
the rate for @xmath211 is not a problem for the standard model if one allows for a modest flavor - singlet penguin contribution in addition to the standard penguin amplitude .
the cp asymmetries for this process are in accord with the expectations of the standard model at the @xmath212 level or better .
effects of the singlet penguin amplitude may also be visible elsewhere , for example in @xmath213 .
various ratios of @xmath214 rates , when combined with information on cp asymmetries , show promise for constraining phases in the ckm matrix .
these tests have shown a steady improvement in accuracy since the asymmetric @xmath1 factories have been operating .
one expects further progress as instantaneous and accumulated @xmath215 luminosities increase , and as hadron colliders begin to provide important contributions .
the decays @xmath216 and @xmath217 show promise for displaying large cp asymmetries @xcite since they involve contributions of different amplitudes with comparable magnitudes .
rare decays of nonstrange and strange @xmath1 s involving photons or lepton pairs are beginning to be studied in detail , and the lhc will be able to look for the rare and interesting @xmath116 decay which can greatly exceed its standard model value in some theories . in the near term the prospects for learning about the @xmath2@xmath40 mixing amplitude are good .
one hopes that this will be an early prize of run ii at the tevatron .
the study of cp violation and branching ratios in @xmath2 decays will be an almost exclusive province of hadron colliders , whose potentialities will be limited only by the versatility of detectors .
surprises in spectroscopy , as illustrated by the low - lying positive - parity @xmath169 candidiates , still can occur , and one is sure to find more surprises at the tevatron and the lhc .
finally , one can search for objects related to the properties of @xmath0 quarks , such as the exotic isosinglet quarks @xmath190 , with improved sensitivity in run ii of the tevatron and with greatly expanded reach at the lhc .
i wish to thank my collaborators on some of the topics mentioned here : troy andre , cheng - wei chiang , michael gronau , zumin luo , and denis suprun .
michael gronau and hassan jawahery also made helpful comments on the manuscript .
this work was supported in part by the united states department of energy under grant no .
de fg02 90er40560 .
# 1#2#3am .
j. phys . *
# 1 * , # 2 ( # 3 ) # 1#2#3ann .
( n.y . ) * # 1 * , # 2 ( # 3 ) # 1#2#3acta phys .
polonica * # 1 * , # 2 ( # 3 ) # 1#2#3ann .
nucl . part .
sci . * # 1 * , # 2 ( # 3 ) # 1#2#3comments on nucl . part .
# 1 * ( # 3 ) # 2 89_cp violation , _ edited by c. jarlskog ( world scientific , singapore , 1989 ) # 1#2#3electronic conference proceedings * # 1 * , # 2 ( # 3 ) # 1#2#3eur . phys .
j. c * # 1 * ( # 3 ) # 2 79_proceedings of the 1979 international symposium on lepton and photon interactions at high energies , _ fermilab , august 23 - 29 , 1979 , ed . by t. b. w. kirk and h. d. i. abarbanel ( fermi national accelerator laboratory , batavia , il , 1979 87_proceeding of the 1987 international symposium on lepton and photon interactions at high energies , _ hamburg , 1987 , ed . by w. bartel and r. rckl ( nucl .
b , proc .
3 ) ( north - holland , amsterdam , 1988 ) # 1#2#3 * # 1 * ( # 3 ) # 2 72_proceedings of the xvi international conference on high energy physics _ ,
chicago and batavia , illinois , sept .
6 13 , 1972 , edited by j. d. jackson , a. roberts , and r. donaldson ( fermilab , batavia , il , 1972 ) # 1#2#3int .
j. mod .
phys . a * # 1 * ( # 3 ) # 2 # 1#2#3jhep * # 1 * ( # 3 ) # 2 # 1#2#3j .
b * # 1 * , # 2 ( # 3 ) 87_selected topics in electroweak interactions _ ( proceedings of the second lake louise institute on new frontiers in particle physics , 15 21 february , 1987 ) , edited by j. m. cameron ( world scientific , singapore , 1987 ) # 1#2#3kong .
danske vid .
selsk . , matt - fys
. medd . *
# 1 * , no.#2 ( # 3 ) # 1#2#3mod . phys .
lett . a * # 1 * ( # 3 ) # 2 # 1#2#3nature * # 1 * , # 2 ( # 3 ) # 1#2#3nuovo cim .
* # 1 * ( # 3 ) # 2 # 1#2#3nucl .
meth . a * # 1 * , # 2 ( # 3 ) # 1#2#3nucl .
* # 1 * ( # 3 ) # 2 # 1#2#3nucl .
* # 1 * ( # 3 ) # 2 # 1#2#3nucl .
suppl . *
# 1 * ( # 3 ) # 2 # 1#2#3#4pisma zh . eksp . teor .
fiz . * # 1 * , # 2 ( # 3 ) [ jetp lett . *
# 1 * , # 4 ( # 3 ) ] # 1#2#3phys .
lett . * # 1 * ( # 3 ) # 2 # 1#2#3phys .
lett . a * # 1 * , # 2 ( # 3 ) # 1#2#3phys .
b * # 1 * ( # 3 ) # 2 # 1#2#3phys .
# 1 * ( # 3 ) # 2 # 1#2#3phys .
c * # 1 * ( # 3 ) # 2 # 1#2#3phys .
d * # 1 * ( # 3 ) # 2 # 1#2#3phys . rev .
# 1 * ( # 3 ) # 2 # 1#2#3phys
. rep .
* # 1 * ( # 3 ) # 2 # 1#2#3prog .
phys . * # 1 * ( # 3 ) # 2 # 1#2#3rev .
# 1 * ( # 3 ) # 2 # 1 rp # 1 9025th international conference on high energy physics , singapore , aug . 2 - 8 , 1990 87_proceedings of the salt lake city meeting _ ( division of particles and fields , american physical society , salt lake city , utah , 1987 ) , ed .
by c. detar and j. s. ball ( world scientific , singapore , 1987 ) 89_proceedings of the xivth international symposium on lepton and photon interactions , _ stanford , california , 1989 , edited by m. riordan ( world scientific , singapore , 1990 ) 82_proceedings of the 1982 dpf summer study on elementary particle physics and future facilities _ , snowmass , colorado , edited by r. donaldson , r. gustafson , and f. paige ( world scientific , singapore , 1982 ) 90_research directions for the decade _ ( proceedings of the 1990 summer study on high energy physics , june 25july 13 , snowmass , colorado ) , edited by e. l. berger ( world scientific , singapore , 1992 ) # 1#2#3#4yad . fiz . * # 1 * , # 2 ( # 3 ) [ sov .
j. nucl .
# 4 ( # 3 ) ] # 1#2#3#4#5#6zh .
* # 1 * , # 2 ( # 3 ) [ sov.phys
. - jetp * # 4 * , # 5 ( # 6 ) ] # 1#2#3zeit .
c * # 1 * , # 2 ( # 3 ) # 1#2#3zeit .
d * # 1 * , # 2 ( # 3 ) g. hiller , ; m. ciuchini and l. silvestrini , ; a. datta , ; m. raidal , ; b. dutta , c. s. kim , and s. oh , ; s. khalil and e. kou , and hep - ph/0303214 ; g. l. kane , p. ko , h. wang , c. kolda , j. h. park and l. t. wang , ; s. baek , ; a. kundu and t. mitra , ; k. agashe and c. d. carone , hep - ph/0304229 .
w. a. bardeen and c. t. hill , .
chiral partners of the ground states of heavy mesons were independently predicted by m. a. nowak , m. rho , and i. zahed , .
see also d. ebert , t. feldmann , r. friedrich , and h. reinhardt , ; d. ebert , t. feldmann , and h. reinhardt , . | examples are given of some current questions in @xmath0 physics to which lhc experiments may provide answers .
these include ( i ) the precise determination of parameters of the cabibbo - kobayashi - maskawa ( ckm ) matrix ; ( ii ) measurements of ckm phases using @xmath1 decays to cp eigenstates ; ( iii ) the search for direct cp asymmetries in @xmath1 decays ; ( iv ) rare radiative @xmath1 decays ; ( v ) the study of @xmath2 properties and decays , ( vi ) excited states of @xmath1 and @xmath2 mesons , and ( vii ) the search for heavier quarks which could mix with the @xmath0 quark . |
geometric langlands duality predicts existence of an automorphic @xmath1-module @xmath2 on @xmath3 attached to a ( de rham ) @xmath4-local system @xmath5 .
here @xmath6 , @xmath4 are reductive groups dual in the sense of langlands , @xmath3 is the moduli stack of @xmath6-bundles on a complete smooth irreducible curve @xmath7 and the local system @xmath5 on @xmath7 with structure group @xmath4 is assumed to be irreducible ( i.e. , it does not admit a reduction to a proper parabolic subgroup ) . in their celebrated unpublished work
@xcite beilinson and drinfeld explain that geometric langlands duality can be thought of as a quantization of a natural duality for the hitchin integrable systems associated to two langlands dual groups @xmath6 , @xmath4 .
furthermore , they present a construction of @xmath2 for a local system @xmath5 which carries an additional structure of an _ oper _ , see @xcite for an introduction to this notion .
their construction uses local to global arguments , it relies heavily on representation theory of affine lie algebras at the critical level . in this note
we describe ( for @xmath8 ) a much shorter construction bypassing affine lie algebra representations , relying on reduction to positive characteristic .
more precisely , we use the ( easy ) construction of automorphic @xmath1-modules @xmath2 for a generic local system @xmath5 on a curve over a field of positive characteristic @xcite .
we verify that under the generic geometric langlands equivalence in positive characteristic of @xcite the free critically twisted @xmath1-module corresponds to ( the pushforward of ) the structure sheaf of opers , this is deduced from a general observation relating the image of the free @xmath1-module under the hecke functor to opers presented in section [ secfil ] ( this part of the argument is closely related to ideas of @xcite ) .
this allows us to show that global sections of the sheaf of critically twisted @xmath1-modules on @xmath9 is a flat deformation of the ring of functions on the hitchin base , by first doing it in positive characteristic and then deducing the general case by a standard argument .
the construction of automorphic @xmath1-modules corresponding to opers follows from this in view of the observation from section [ secfil ] .
let us mention that the theme started in @xcite , where geometric langlands duality was established for @xmath10 local systems with a smooth spectral curve has been developed in @xcite , where the case of not necessarily smooth spectral curves has been treated and @xcite , @xcite dealing with @xmath6 local systems for @xmath11 .
however , the present note is the first work ( to the authors knowledge ) where this type of result is connected to the original setting of a characteristic zero base field .
we finish the introduction with a technical remark .
below we use a general construction , the derived category of asymptotic @xmath1-modules on stacks . here by an asymptotic @xmath1-module on a smooth algebraic variety @xmath12 we mean a sheaf of modules over the sheaf of rings @xmath13 , the sheaf of rees algebras corresponding to the sheaf of filtered algebras @xmath14 . thus @xmath13 is a flat sheaf of rings over polynomials in @xmath15 , such that @xmath16 , @xmath17 .
we refer to @xcite for a discussion of standard functors on the derived category of @xmath18-modules .
the proof of proposition [ filtr ] relies on an extension of this theory to smooth algebraic stacks over a field of an arbitrary characteristic which does not seem to be documented in the literature .
the proof of lemma [ otc ] uses rudimentary theory of dg - stacks ( the only dg - stacks appearing here are derived fiber products of ordinary stacks ) .
* acknowledgements .
* the first author would like to thank g. laumon who several years ago has asked him if the result of @xcite can be applied to the characteristic zero setting .
r.b . was supported by an nsf grant and a simons foundation fellowship during preparation of this article .
we mostly work over a field @xmath19 of characteristic different from 2 , we fix a complete curve @xmath7 over @xmath19 .
let @xmath8 , @xmath20 will denote the moduli stack of rank @xmath21 vector bundles over @xmath7 ; let @xmath22 denote the component of parametrizing bundles of degree @xmath23 . recall the stack @xmath24 with a map @xmath25 which is a @xmath26-gerbe , @xcite .
the categories of coherent sheaves and @xmath1-modules on @xmath24 and @xmath20 are closely related , while @xmath24 has the advantage of being _ good _ in the sense of @xcite ; we let @xmath27 be the image of @xmath22 .
let @xmath28 be the category of twisted @xmath1-modules on @xmath20 , where the class of the twisting equals half of the class corresponding to the canonical line bundle on @xmath20 and similarly for @xmath24 .
notice that a square root of the canonical line bundle on @xmath20 is known to exist , thus this category is equivalent to the category of @xmath1-modules on @xmath20 , respectively @xmath24 .
let @xmath29 denote the sheaf of twisted differential operators with the same twisting .
( notice that @xmath30 is _ not _ a sheaf of rings on @xmath20 , see ( * ? ? ?
1.1.3 ) . )
let also @xmath31 be the derived category of @xmath1-modules on @xmath20 .
thus @xmath28 is the heart of the natural @xmath32-structure on @xmath31 .
we also use similar notations with @xmath20 replaced by @xmath24 .
let @xmath33 denote the space of _ marked _ opers , see @xcite for a general introduction to this notion , see also the definition ( in the version we use ) below before corollary [ filtr_cor ] .
the main result of this note is the following [ main ] 1 . for every @xmath34 we have a canonical isomorphism @xmath35 .
2 . for a point
@xmath36 corresponding to a local system @xmath37 the @xmath1-module @xmath38 is a hecke eigenmodule with respect to the local system @xmath37 .
our strategy is to first establish the result when @xmath19 has prime characteristic using the result of @xcite and then formally deduce the characteristic zero case .
in this section we introduce a filtration on the image of the free @xmath1-module under the hecke functor .
this is done uniformly in all characteristics by a direct argument independent of @xcite ; the idea is close in spirit to @xcite .
we now recall the definition of the hecke functor corresponding to the tautological representation of @xmath10 .
let @xmath39 be the stack parametrizing inclusions of vector bundles of rank @xmath21 , @xmath40 , whose cokernel is of length one .
we define @xmath41 and @xmath42 by @xmath43 and @xmath44 where @xmath45 is determined by the short exact sequence @xmath46 we also let @xmath47 and let @xmath48 denote the restriction of @xmath49 , @xmath50 to @xmath51 . notice that both @xmath52 and @xmath53 are @xmath54 bundles .
the following statement is standard .
[ rel_tan ] the relative tangent bundles @xmath55 , @xmath56 for the maps @xmath57 admit a nondegenerate pairing @xmath58 .
we let @xmath59 denote the sheaf of twisted differential operators on @xmath7 corresponding to the line bundle @xmath60 ; here for odd @xmath21 we use the choice of a square root of the canonical bundle @xmath61 .
we let @xmath62 , @xmath63 denote the corresponding derived categories of twisted @xmath1-modules . using the lemma it is easy to see that the sum of pull - backs under @xmath49 , @xmath50 and @xmath64 of the above twisting classes equals the class of a line bundle ( we will use a more precise information about this class below ) ; thus we can define the _ hecke functor _ @xmath65 .
notice that we used smoothness of the hecke stack to define the hecke functor uniformly in all characteristics , a direct analogue of this definition for arbitrary reductive group @xmath6 works for hecke functors corresponding to minuscule coweights only .
let @xmath66 denote the term of the standard filtration by the order of a differential operator .
[ filtr ] the object @xmath67 lies in the abelian category @xmath68 . furthermore , the @xmath69 module @xmath67 admits a canonical map @xmath70 , such that the restriction of @xmath71 to @xmath72 is an isomorphism of quasicoherent sheaves .
the proof uses the category of `` asymptotic '' @xmath1-modules @xmath73 ( cf .
recall that for a smooth variety @xmath12 the sheaf of rings @xmath74 is obtained from the filtered sheaf of rings @xmath75 ( differential operators on @xmath12 ) by the rees construction .
thus @xmath73 is a sheaf of graded rings on @xmath12 with a central section @xmath76 , such that @xmath77 is isomorphic to the sheaf @xmath78 , while the localization @xmath79 is isomorphic to @xmath80 $ ] , one then considers the ( derived ) category of sheaves of graded modules .
a similar construction applies to twisted differential operators on stacks in the sense of @xcite .
notice that the subcategory of @xmath76-torsion free coherent asymptotic @xmath1-modules is equivalent to the category of coherent @xmath1-modules equipped with a good filtration .
the push - forward and pull - back functors are defined for `` asymptotic '' ( twisted ) @xmath1-modules in a way compatible with the natural ( derived ) functor from the category of @xmath81 modules to that of @xmath1-modules ( quotient by @xmath82 ) , see @xcite . moreover , as shown in _
the pull - back functor is compatible under the specialization at @xmath83 with the functor between the derived categories of coherent sheaves on the cotangent bundles given by the natural correspondence ; while the push - forward functor under a proper morphism @xmath84 is compatible with the functor given by the natural correspondence up to twist by the line bundle @xmath85 .
this theory can be generalized to smooth algebraic stacks .
notice that in the stack case the cotangent bundles and/or the relevant fiber products may have to be taken in the category of derived stacks .
we now proceed to spell this out in the present case .
let @xmath86 . recall that @xmath87 is the stack parametrizing _
higgs fields _ ,
i.e. pairs @xmath88 where @xmath89 is a rank @xmath21 bundle on @xmath7 and @xmath90 .
let @xmath91 denote the _ hitchin hecke stack _ parametrizing triples @xmath92 where @xmath93 and @xmath94 satisfies @xmath95 .
we have @xmath96 , @xmath97 and @xmath98 sending @xmath99 to @xmath100 where @xmath45 and @xmath101 are determined by @xmath102 , @xmath103 .
the free rank one @xmath30 module equipped with the standard filtration determines an object in the category of asymptotic @xmath30-modules on @xmath20 which we denote by @xmath104 . applying the hecke functor @xmath105 to this object ( where @xmath106 denotes the hecke functor on the category of asymptotic @xmath1-modules ) we get an object @xmath107 . using the above compatibility of pull - back and
push - forward functors with the specialization at @xmath83 and lemma [ rel_tan ] , one checks that the corresponding coherent sheaf @xmath108 } k$ ] is given by : @xmath109 } k\cong ( pr_2\times pr_c ) _ * pr_1^*(\o_{{\mathrm{hitch}}}).\ ] ] let @xmath110 be the _
universal spectral curve _ , i.e. @xmath111 parametrizes the data of @xmath112 where @xmath113 , @xmath114 such that @xmath115 .
[ otc ] we have @xmath116 .
consider subvarieties @xmath117 and @xmath118 given by : @xmath119 , @xmath120 .
let @xmath121 be the natural projection .
it is a standard fact that @xmath122 we have a natural `` evaluation '' map @xmath123 ( where @xmath10 acts on @xmath124 by the adjoint action and @xmath26 acts on @xmath125 by dialtions ) and we have natural isomorphisms : @xmath126 @xmath127 here both fiber products are understood in the derived stacks , thus both formulas are isomorphisms of dg-stacks.s between the structure sheaves of the two factors over the structure sheaf of the base are nonzero .
if this is not the case the isomorphisms can be understood as isomorphisms of ordinary stacks . ]
thus base change isomorphism applies , so lemma follows from .
using koszul resolution one can write down an explicit sheaf of dg - algebras on the cartesian product of the two factors in the displayed formulas above , such that its derived category of sheaves of modules is identified with the derived coherent sheaves category of the derived fiber product . thus one can work with these categories without invoking the general theory of dg - stacks .
we are now ready to finish the proof of the proposition .
recall that an object @xmath128 in the derived category of asymptotic @xmath1-modules on a stack @xmath12 such that the induced object @xmath129}k\in d^b(coh^{\gm}(t^*(x))$ ] lies in homological degree zero amounts to a @xmath1-module with a good filtration whose associated graded is isomorphic to @xmath130 .
thus comparing with lemma [ otc ] we see that @xmath67 is a @xmath1-module with a good filtration whose associated graded is isomorphic to @xmath131 .
since the latter coherent sheaf is cyclic , we see that @xmath132 is a cyclic @xmath1-module with a canonical generator .
since the sheaf of regular functions on @xmath133 which have degree less than @xmath21 along the fibers of projection @xmath134 maps isomorphically to @xmath131 , the proposition follows .
recall the ring of twisted differential operators @xmath135 introduced after lemma [ rel_tan ] . by an _
oper _ we will understand an @xmath136-coherent @xmath135 module @xmath136 of rank @xmath21 which has a good filtration whose associated graded is isomorphic @xmath137 . choosing a theta - characteristic ( i.e. a square root of the cotangent bundle ) we can identify the category of @xmath135-modules with the category of @xmath1-modules and connect this with the standard definition of a ( _ marked _ ) oper .
it is standard that opers in this sense are parametrized by a variety which we will denote @xmath33 .
[ filtr_cor ] assume that @xmath138 satisfies the hecke eigenproperty with respect to a local system @xmath139 .
if @xmath140 , then @xmath141 , and @xmath142 is determined by the character by which @xmath26 acts on the quasicoherent sheaf underlying @xmath130 ( where we use that @xmath143 is a @xmath26 gerbe over @xmath24 ) .
it is not hard to see that @xmath144 automatically unless @xmath145 . ] that @xmath146 has degree @xmath147 and
does not admit an oper structure .
then @xmath148 the proof will proceed by contradiction . let @xmath146 be the corresponding local system and set @xmath149 , thus @xmath150 by assumption .
consider @xmath151 , the sheaf direct image of the @xmath1-module @xmath152 under the projection @xmath153 .
then the hecke eigen - property of @xmath130 shows that @xmath154 on the other hand , proposition [ filtr ] shows that the object in the derived category of quasicoherent sheaves @xmath155 satisfies : @xmath156 comparing the two displayed isomorphisms we see that @xmath157 admits an injective map into a quasicoherent sheaf with a filtration whose associated graded is a sum @xmath158 ( @xmath159 ) . since an injective map between coherent sheaves on a curve having the same degree and the same generic rank has to be an isomorphism
, we see that @xmath146 has an oper structure .
it is easy to deduce the assertion of the theorem for @xmath19 from the assertion for the algebraic closure of @xmath19 , so we assume for simplicity that @xmath19 is algebraically closed .
recall that @xmath86 and @xmath111 is the universal spectral curve .
let @xmath161 be the hitchin map and @xmath162 be the projection .
let @xmath163 be the open subsets in the hitchin base @xmath164 parametrizing the points @xmath165 such that the fiber @xmath166 is reduced , respectively , smooth .
in this section we assume that the base field @xmath19 has prime characteristic @xmath167
. then @xmath30 can be thought of as a sheaf over @xmath168 , where the superscript denotes the frobenius twist .
let @xmath169 denote the moduli stack of @xmath59-modules which are locally free of rank @xmath21 as an @xmath136-module .
recall @xcite that we have the _ frobenius - hitchin _
map @xmath170 ; for example , for @xmath171 the fiber of @xmath76 over @xmath45 is the abelian algebraic group @xmath172 , while the fiber of @xmath173 over @xmath174 is the torsor over the abelian algebraic group @xmath175 ( here @xmath176 denotes the image of @xmath45 under frobenius ) .
we will need the following result proven in the appendix .
[ finite_flat ] the composition @xmath177 is a flat finite map of degree @xmath178 , where @xmath179 .
set @xmath180 , @xmath181 , @xmath182 , @xmath183 , where in the last expression we use the same notation @xmath30 for the object in @xmath31 and the corresponding sheaf on @xmath168 .
recall that the main result of @xcite is an equivalence and ordinary @xmath1-modules are considered instead of @xmath24 and twisted @xmath1-modules .
it is not hard to deduce that version from the result of _ loc .
_ ] @xmath184 where @xmath185 stands for the category of coherent sheaves of modules .
we let @xmath186 denote that equivalence .
the first step in the proof of the theorem is the following [ ops ] we have @xmath187 .
it is easy to see that the tautological map from the space @xmath33 of marked opers to @xmath169 is a composition of the map @xmath188 and a closed embedding @xmath189 , where we use the trivial action of @xmath26 on @xmath33 .
thus the direct image of @xmath190 to @xmath169 decomposes as a direct sum indexed by characters of @xmath26 ; we claim that the summand corresponding to the character @xmath191 is canonically isomorphic to @xmath192 .
it follows from corollary [ filtr_cor ] that the complex @xmath193 is supported on @xmath194 . from proposition [ finite_flat ]
we see that its support is finite over the base @xmath195 .
since fourier - mukai transform is exact on sheaves with finite support , sending such a sheaf of length @xmath196 into a vector bundle of rank @xmath196 , we see that @xmath197 is concentrated in homological degree zero ; moreover , its pull - back to @xmath198 is flat of rank @xmath178 as a module over @xmath199 .
we claim that @xmath197 is scheme theoretically supported on @xmath194 .
first of all , @xmath197 is torsion free as an @xmath200 module .
thus it suffices to check this claim over the generic point of @xmath201 . to see this notice
that a coherent sheaf on @xmath169 which is generically set theoretically but not scheme theoretically supported on @xmath202 would need to have length greater than one at the generic point of @xmath33 .
then its direct image to @xmath195 would have generic rank greater than @xmath178 , which contradicts the second paragraph of the proof .
now , flatness of @xmath197 over @xmath203 implies it s a cohen - macaulay module over @xmath204 .
since @xmath198 is smooth , it is actually a locally free module , since its degree over @xmath199 equals that of @xmath204 , we conclude that the pull - back of @xmath197 to @xmath198 is a line bundle on @xmath198 .
since @xmath198 is an open subvariety in @xmath33 which is isomorphic to the affine space , every line bundle on @xmath198 is trivial .
since @xmath26 acts by the character @xmath191 on the @xmath205 for @xmath206 supported on @xmath207 , we get the statement .
the proposition implies that for all @xmath208 @xmath209 is a commutative algebra .
proposition [ filtr ] allows one to construct a family of opers on @xmath7 parametrized by @xmath210 .
the family can be described as a @xmath59-module @xmath211 with an @xmath212 action and a filtration which is flat over @xmath213 and such that @xmath214 is an oper for every @xmath215 and is constructed as follows .
the sheaf @xmath211 is the ( sheaf theoretic ) direct image to the second factor of the @xmath216-module @xmath217 , equipped with the natural @xmath212 action and the filtration coming from proposition [ filtr ] .
the @xmath212-action is defined by presenting @xmath211 as the pushforward to @xmath7 of the local hom from @xmath218 to @xmath217 and using @xmath212-action on the first argument .
actually the definition of this action can be generalized to a functor @xmath219 given by pushforward to @xmath7 , and @xmath220 . applying the functor @xmath221 to the map @xmath71 from proposition [ filtr ]
, we get a map @xmath222 which restricts to an isomorphism of @xmath223-modules @xmath224 . from this
it is straightforward to see that @xmath211 defines an @xmath212-family of opers .
thus we get a map @xmath225
. we will show that it is an isomorphism .
it is easy to deduce from the hecke eigen - property for the equivalence @xmath186 that base - change of @xmath226 from @xmath195 to @xmath227 coincides with the ( dual of ) isomorphism . from this
we see that the composition @xmath228 is the natural inclusion .
thus @xmath212 is isomorphic to subalgebra of @xmath229 containing @xmath230 .
since @xmath33 is normal ( it is isomorphic to an affine space ) , it would suffice to show that @xmath212 is finitely generated as a module over @xmath230 .
we will in fact prove finite generation over a smaller algebra .
[ finite ] @xmath212 is a finitely generated torsion free module over @xmath231 .
consider the filtration on @xmath30 and the induced one on @xmath212 by degree of differential operator .
then we have @xmath232 .
the induced filtration on @xmath233 coincides with the one coming from the grading on @xmath231 multiplied by @xmath167 , and the associated graded map to this embedding is dual to @xmath234 .
it is known that all global functions on @xmath235 are pullbacks from the hitchin base , so we have @xmath236 . on the other hand
, there is an inclusion @xmath237 .
thus @xmath238 identifies with an @xmath231-submodule in @xmath239 , therefore it is finitely generated and torsion - free over @xmath231 .
but then so is @xmath212 , as desired .
as explained above , the lemma shows that the map @xmath226 is an isomorphism , so that for any @xmath23 we have a canonical isomorphism @xmath240 it also follows that @xmath211 are identified for all , so we write @xmath241 for the sheaf isomorphic to all of them .
[ c_comm_a ] the map @xmath242 from proposition [ filtr ] intertwines the two @xmath243-actions coming from the @xmath243-action on @xmath30 .
since both the source and the target are torsion - free as modules over @xmath244 ( where @xmath245 stands for the center of a category , i.e. the ring of endomorphisms of the identity functor ) , it is enough to show that the statement of the lemma holds after tensoring by @xmath229 over @xmath230 .
the localized map @xmath246 is a morphism in @xmath247 , and we can apply @xmath186 in the first factor .
then proposition [ ops ] and hecke eigen - property of @xmath186 imply that @xmath248 is a map of @xmath249-modules scheme - theoretically supported on @xmath250 , and that the elements of @xmath251 act by multiplication by the same functions on this support .
thus we proved the first part of the theorem .
the second statement of the theorem follows from the construction of the morphism @xmath226 .
indeed , we need to construct an isomorphism @xmath252 where we used the following notation : if @xmath243 is a @xmath19-algebra , @xmath253 are @xmath19-schemes ( or stacks ) and @xmath206 , respectively @xmath254 , are quasi - coherent sheaves on @xmath12 , resp .
@xmath255 , with a right , resp .
left , @xmath243-actions , then we define a quasi - coherent sheaf on @xmath256 : @xmath257 , where the rightmost symbol @xmath243 refers to the regular @xmath243-bimodule
. we will construct the isomorphism as above for each component @xmath22 of @xmath20 , so fix . consider the functor @xmath258 , which is the left adjoint to the `` @xmath23th component '' of the functor @xmath221 used above .
so we have a counit map @xmath259 for any @xmath260 .
we need to check that it is an isomorphism for @xmath261 .
it is clear that if we apply the forgetful functor @xmath262 to @xmath263 then we will get the counit morphism for the similar adjunction between @xmath264 and @xmath265 .
now since locally over @xmath7 , @xmath217 is isomorphic to @xmath266 as a relative @xmath1-module on @xmath267 over @xmath7 , and our statement is local in @xmath7 , it suffices to check for @xmath268 where it is clear .
now , to deduce hecke eigen - property , we have to show that the constructed isomorphism commutes with the @xmath243-action , where the action on @xmath67 comes by transport of structure from the @xmath243-action on @xmath30 and the action on @xmath269 comes from either of the @xmath243-actions contracted by the tensor product . for this
we consider the commutative diagram @xmath270|-{\dis=\!=}\ & \ { { { \mathcal d}_{\bun^{d}}}}\bx_a\upsilon({{{\mathcal d}_{\bun^{d}\x c } } } ) \ar[rr]^-{a_{{{{\mathcal d}_{\bun^{d}\x c}}}}}_-\sim \ar[d]_{{{{\mathcal d}_{\bun^{d}}}}\bx_a\upsilon(c)}\ & & \ { { { \mathcal d}_{\bun^{d}\x c}}}\ar[d]_c \\ { { { \mathcal d}_{\bun^{d}}}}\bx_a\funiv\ar@{}[r]|-{\dis=\!= } \ & \ { { { \mathcal d}_{\bun^{d}}}}\bx_a\upsilon(h({{{\mathcal d}_{\bun^{d-1 } } } } ) ) \ar[rr]^-{a_{h({{{\mathcal d}_{\bun^{d-1}}}})}}_-\sim \ & & \
h({{{\mathcal d}_{\bun^{d-1}}}}).\ } \ ] ] by lemma [ c_comm_a ] , the vertical arrows in this diagram commute with the @xmath243-action .
the top arrow commutes with the @xmath243-action because of the naturality of @xmath263 .
hence the bottom arrow does , too , which is what we need .
what we just proved implies the following in - families version of hecke eigen - property .
the @xmath243-action on the @xmath1-module @xmath30 allows to present it as a pushforward of a @xmath271-family @xmath128 of @xmath1-modules .
applying to @xmath128 the relative version of the hecke functor @xmath272 , we get a relative ( twisted ) @xmath1-module on @xmath273 over @xmath33 which we denote by @xmath274 .
then it follows from what we proved that @xmath275 where @xmath276 is the @xmath33-family of opers constructed from @xmath241 . taking pullback to a closed point @xmath45 of @xmath33
, we see that the derived specialization of the family @xmath128 , @xmath277 is a hecke eigen-@xmath1-module .
we want to show that @xmath278 is actually in the heart of @xmath279 . in other words , we want to prove that @xmath30 is flat over @xmath243 . since the @xmath243-module @xmath30 is a ( flat ) deformation of @xmath280 viewed as a module over @xmath239 , this follows from flatness of the hitchin map .
considering the deformation of @xmath282 to @xmath216 we get a spectral sequence @xmath283 choose a finitely generated ring @xmath284 with a homomorphism @xmath285 and a complete curve @xmath286 over @xmath284 whose base change to @xmath19 is isomorphic to @xmath7 .
we can form a similar spectral sequence starting from the moduli stack @xmath287 for vector bundles over @xmath286 .
its base change to a field of positive characteristic degenerates at @xmath288 by the result of the previous section .
hence the spectral sequence itself , as well as its base change to @xmath19 , degenerates at @xmath288 .
this implies that @xmath212 is commutative for any @xmath23 , since its base change to any residue field of @xmath284 of almost every prime characteristic is commutative .
now the construction of the previous section yields a family of opers parametrized by @xmath289 , given by @xmath290 as before .
thus we get a map @xmath291 as explained in the previous section . since the base change of @xmath226 to a field of almost any prime characteristic is an isomorphism
, we see that @xmath226 is an isomorphism .
this proves the first part of theorem [ main ] .
one then proves an analogue of lemma [ c_comm_a ] in characteristic @xmath292 by observing that it is enough to prove the statement for the reductions to finite residue fields of @xmath284 .
after that , the second part follows by the argument of the previous section .
[ thm]corollary [ thm]lemma [ thm]proposition [ thm]conjecture [ thm]theorem - definition [ thm]definition [ thm]remark [ thm]example ad aff der diag f higgs i d ind opop res tr gsp gu sl su so locsys_j^p( ) locsys_j^p( ) ic
this appendix is devoted to the proof of the following statement ( proposition [ finite_flat ] of the main text ) : [ main ] let @xmath293 be the scheme of @xmath6-opers with marking ( see [ oper ] ) .
then the composition @xmath294 is finite and faithfully flat of degree @xmath295 . here
@xmath296 is the @xmath167-hitchin map .
let @xmath19 be an algebraically closed field of characteristic @xmath167 .
let @xmath7 be a complete smooth curve over @xmath19 .
let @xmath6 be a connected reductive algebraic group over @xmath19 of rank @xmath298 .
we denote by @xmath299 the lie algebras of @xmath6 .
we fix a borel subgroup @xmath300 , and let @xmath301 be its unipotent radical and @xmath302 .
let @xmath303 be the center of @xmath6 .
we denote by @xmath304 , @xmath305 and @xmath306 .
we denote the corresponding lie algebras by @xmath307 , @xmath308 and @xmath309 .
we assume that @xmath167 does not divide the order of the weyl group @xmath310 of @xmath6 .
let @xmath311 $ ] and @xmath312 $ ] be the algebra of polynomial function on @xmath299 and @xmath309 .
by chevalley s theorem , we have an isomorphism @xmath311^{g}{\simeq}k[{{\mathfrak t}}]^{{{\mathrm w}}}$ ] .
moreover , @xmath312^{{\mathrm w}}$ ] is isomorphic to a polynomial ring of @xmath298 variables @xmath313 and each @xmath314 is homogeneous in degree @xmath315
. let @xmath316^w)$ ] .
let @xmath317 be the map induced by @xmath318{\simeq}k[{{\mathfrak g}}]^g\hookrightarrow k[{{\mathfrak g}}]$ ] .
this is @xmath319-equivariant map where @xmath6 acts trivially on @xmath320 , and @xmath321 acts on @xmath320 through the gradings on @xmath312^{{{\mathrm w}}}$ ] .
let @xmath322 be an invertible sheaf on @xmath7 and @xmath323 be the corresponding @xmath321-torsor .
let @xmath324 and @xmath325 be the @xmath321-twist of @xmath299 and @xmath320 with respect to the natural @xmath321-action .
let @xmath326)$ ] be the stack of section of @xmath327 $ ] over @xmath7 , i.e. , for each @xmath19-scheme @xmath328 the groupoid @xmath329 consists of maps over @xmath7 : @xmath330.\ ] ] equivalently , @xmath329 consists of a pair @xmath331 ( called a higgs bundle ) , where @xmath332 is a @xmath6-torsor over @xmath333 and @xmath334 is an element in @xmath335 .
if the group @xmath6 is clear from the content , we simply write @xmath336 for @xmath337 .
let @xmath338 be the scheme of sections of @xmath339 over @xmath7 , i.e. , for each @xmath19-scheme @xmath328 , @xmath340 is the set of sections over @xmath7 @xmath341 this is called the hitchin base of @xmath6 .
the natural @xmath6-invariant projection @xmath342 induces a map
@xmath343:[{{\mathfrak g}}_{{\mathcal{l}}}/g]{\rightarrow}{{\mathfrak c}}_{{\mathcal{l}}},\ ] ] which in turn induces a natural map @xmath344){\rightarrow}{\operatorname}{sect}(c,{{\mathfrak c}}_{{\mathcal{l}}})=b_{{\mathcal{l}}}.\ ] ] we call @xmath345 the hitchin map associated to @xmath322 .
we are mostly interested in the case @xmath346 . for simplicity , from now on we denote @xmath347 , @xmath348 and @xmath349 , etc .
we sometimes also write @xmath350 for @xmath351 to emphasize the group @xmath6 .
we fix a square root @xmath352 ( called a theta characteristic of @xmath7 ) . recall that
in this case , there is a section @xmath353 of @xmath354 , induced by the kostant section @xmath355 .
sometimes , we also call @xmath356 the kostant section of the hitchin fibration .
let @xmath357 be the stack of @xmath6-local system on @xmath7 , i.e. for every scheme @xmath328 over @xmath19 , @xmath358 is the groupoid of all @xmath6-torsors @xmath332 on @xmath333 together with a connection @xmath359 , here @xmath360 is the lie algebroid of infinitesimal symmetry of @xmath332 .
recall the notion of @xmath167-curvature of a @xmath6-local system following @xcite : for any @xmath361 the @xmath167-curvature of @xmath362 is defined as @xmath363 we regard @xmath364 as an element @xmath365 and call such a pair an @xmath366-higgs field .
the assignment @xmath367 defines a map @xmath368 . combining this map with @xmath369
, we get a morphism from @xmath357 to @xmath370 : @xmath371 observe that the pullback along @xmath372 induces a natural map @xmath373 , where the superscript denotes the frobenius twist . by (
* theorem 3.1 ) , the @xmath167-curvature morphism @xmath374 factors through a unique morphism @xmath375 we called this map the @xmath167-hitchin map . the construction of @xmath167-hitchin map can be generalized to @xmath376-connection .
recall that for any @xmath377 , a @xmath376-connection on a @xmath6-torsor @xmath332 is an @xmath378-linear map @xmath379 such that the composition @xmath380 is equal to @xmath381 ( where @xmath382 is the natural projection ) .
we denote by @xmath383 the stack of @xmath6-bundles on @xmath7 with @xmath376-connections .
then @xmath384 let @xmath385 .
the @xmath167-curvature of @xmath386 is defined as @xmath387 the map @xmath388 factors through a unique map @xmath389 called the @xmath167-hitchin map for @xmath376-connections .
it is clear that @xmath390 and @xmath391 , where @xmath392 is the usual hitchin map and @xmath393 is the relative frobenius of @xmath164 .
from this perspective , the @xmath167-hitchin map can be regarded as a deformation of the usual hitchin map . in this subsection
we recall the definition of opers with _ marking _ following @xcite .
there is a canonical decreasing lie algebra filtration @xmath394 of @xmath299 @xmath395 such that @xmath396 , @xmath397 and for any @xmath398 ( resp .
@xmath399 ) weights of the action of @xmath400 on @xmath401 are sums of @xmath19 simple positive ( resp .
negative ) roots .
in particular , we have @xmath402 , where @xmath403 is a simple negative root and @xmath404 is the corresponding root space .
let @xmath332 be a @xmath405-torsor on @xmath7 and @xmath406 ( resp .
@xmath407 ) be the induced @xmath6-torsor ( resp .
@xmath408-torsor ) on @xmath7 . in this subsection , we denote by @xmath409 and @xmath410 be the associated adjoint bundles ( rather than @xmath411 ) .
let @xmath412 and @xmath413 be the lie algebroids of infinitesimal symmetries of @xmath332 and @xmath406 .
there is a natural embedding @xmath414 and we have a canonical isomorphism @xmath415 for any connection @xmath362 on @xmath406 , we denote by @xmath416 the composition
@xmath417 we fix a square root @xmath352 of the canonical bundle @xmath418 . a @xmath6-oper on @xmath7 with marking
is triple @xmath419 where @xmath332 is a @xmath405-torsor on @xmath7 , @xmath362 is a connection on @xmath406 , and @xmath420 is an isomorphism of @xmath408-torsor ( we call @xmath334 the _ marking _ ) , such that 1 . the image of @xmath416 lands in @xmath421 .
the composition @xmath422 is an isomorphism for every simple negative root @xmath403 .
here @xmath423 is the natural projection .
the condition ( 2 ) implies @xmath362 induces an isomorphism @xmath424 we require the marking @xmath334 is compatible with @xmath425 . notice that if we drop condition ( 3 ) in above definition
, then we obtain the definition of @xmath6-opers in @xcite . as shown in _ loc .
_ , a @xmath6-oper has @xmath303 as its automorphism group , the additional condition ( 3 ) eliminate these automorphisms ( cf .
* proposition 2.1 ) ) .
we denote by @xmath426 the scheme of @xmath6-opers with marking on @xmath7 .
when @xmath6 is of adjoint type , there exits a unique marking @xmath334 compatible with @xmath425 .
thus , in this case , the condition ( 3 ) is automatic . in general
, the conditions ( 1 ) and ( 2 ) do not imply the existence of @xmath334 , hence we are limiting our collection of opers compared to @xcite .
consider the case @xmath8 .
then an oper with marking can be described in terms of vector bundles as follows : it consists of the data @xmath427 where @xmath332 is a rank @xmath21 vector bundle on @xmath7 , @xmath428 is a complete flag , @xmath362 is a connection on @xmath332 , and @xmath429 is an isomorphism , such that one defines a @xmath433-oper with marking as before by replacing connection @xmath362 by @xmath376-connection @xmath386 .
we denote by @xmath434 the scheme of @xmath433-opers with marking .
clearly , @xmath435 and @xmath436 .
all @xmath433-opers with marking form a scheme , flat over @xmath437 , @xmath438 , such that the fiber of @xmath439 over @xmath440 is @xmath434 .
moreover , there a @xmath321-action on @xmath439 , given by @xmath441 and the morphism @xmath438 is @xmath321-equivariant .
the @xmath167-hitchin map for @xmath376-connections gives @xmath445 the map @xmath446 is @xmath321-equivariant where @xmath321 acts diagonally on @xmath447 .
we denote by @xmath448 the base change of @xmath446 to @xmath449 .
when @xmath450 , we get a map @xmath451 and ( [ kostant map ] ) implies @xmath452 is the relative frobenius morphism @xmath453 .
let @xmath463 be the map in ( [ pi_p ] ) .
we first show that @xmath173 is finite and surjective .
i.e. , we need to show that @xmath464 is injective and @xmath457 is finitely generated as an @xmath458-module . since both rings @xmath457 and @xmath458 are filtered and @xmath461 is compatible with the filtrations , it is enough to show that the associated graded map @xmath465 is injective and @xmath466 is a finitely generated @xmath467-module . but
this is clear since by the lemma above @xmath468 is the frobenius map .
now @xmath173 is a finite map between @xmath293 and @xmath195 , which are smooth of the same dimension , and therefore it is flat .
in addition , as the relative frobenius map @xmath469 is of degree @xmath295 , so is @xmath173 .
lemma [ gr ] shows that the map @xmath470 is a deformation of the frobenius morphism @xmath471 . in the special case
when @xmath472 it is not hard to see that one can identify @xmath473 with @xmath474 so that the morphism @xmath173 is identified with @xmath475 where @xmath476 is the map induced by cartier isomorphism . in particular , @xmath173 is purely inseparable if and only if @xmath7 is supersingular .
it would be interesting to obtain a similar explicit description of the map @xmath173 for nonabelian @xmath6 .
r. bezrukavnikov , a. braverman , _ geometric langlands correspondence for d - modules in prime characteristic : the gl(n ) case , _ pure appl .
q. 3 ( 2007 ) , no .
1 , special issue : in honor of robert d. macpherson . part 3 , 153179 . | the main result of the seminal ( unpublished ) work of beilinson - drinfeld is the construction of an automorphic sheaf corresponding to a local system which carries the additional structure of an oper .
this is achieved by quantizing the hitchin intergrable system . in this note
we show ( in the case of @xmath0 ) that this result admits a short proof based on positive characteristic methods . |
hypergraphs are a natural generalization of undirected graphs in which edges " may consist of more than 2 vertices . more precisely ,
a ( finite ) _ hypergraph _ @xmath12 consists of a ( finite ) set @xmath13 and a collection @xmath14 of non - empty subsets of @xmath13 . the elements of @xmath13 are called _ vertices _ and the elements of @xmath14 are called _ hyperedges _ , or simply _ edges _ of the hypergraph .
an @xmath8-_edge _ is an edge containing exactly @xmath8 vertices .
the _ rank _ of @xmath0 , denoted by @xmath15 , is the maximum size of an edge in @xmath0 .
specially , an @xmath8-_uniform _ hypergraph @xmath0 is a hypergraph such that all edges are @xmath8-edges .
a hypergraph is called _ linear _ if any two edges of the hypergraph intersect in at most one vertex .
obviously , every ( simple ) graph is a linear 2-uniform hypergarph . throughout this paper ,
we only consider hypergraphs of rank @xmath16 without multiple edges and all edges have size at least @xmath17 .
the _ degree _ of a vertex @xmath6 in @xmath0 , denoted by @xmath18 , is the number of edges of @xmath0 containing the vertex @xmath6 .
a vertex of degree zero is called an _ isolated vertex_. a vertex of degree @xmath19
is called a _
degree_-@xmath19 vertex .
the _ minimum _ and _ maximum degree _ among the vertices of @xmath0 are denoted by @xmath20 and @xmath21 , respectively .
the _ quasidegree _ of @xmath6 in @xmath0 , denoted @xmath22 , is the maximum number of edges of @xmath0 whose pairwise intersection is only @xmath6 .
two vertices @xmath5 and @xmath6 in @xmath0 are _ adjacent _ if there is a edge @xmath23 of @xmath0 such that @xmath24 .
the _ open neighborhood _ of a vertex @xmath6 in @xmath0 , denoted @xmath25 , is the set of all vertices different from @xmath6 that are adjacent to @xmath6 .
if @xmath0 is clear from the context , we denote @xmath18 , @xmath22 and @xmath25 by @xmath26 and @xmath27 , respectively .
two edges in @xmath0 are said to be _ overlapping _ if they intersect in at least two vertices .
a _ partial hypergraph _
@xmath28 of @xmath12 , denoted by @xmath29 , is a hypergraph such that @xmath30 and @xmath31 . in the class of graphs ,
partial hypergraphs are called _
subgraphs_. in particular , if @xmath32 , @xmath33 is called a _ spanning partial hypergraph _ of @xmath0 . for a hypergraph @xmath0 and @xmath34
, @xmath35 denotes the hypergraph obtained by removing the vertices @xmath36 from @xmath0 and removing all edges that intersect @xmath36 . for a subset @xmath37 of edges in @xmath0
, we define @xmath38 to be the hypergraph obtained from @xmath0 by deleting the edges in @xmath39 and resulting isolated vertices , if any . if @xmath40 , then we write @xmath38 simply as @xmath41 . for @xmath42 and @xmath43 , if we remove the vertex @xmath6 from the edge @xmath23 , we say that the resulting edge is obtained by @xmath6-_shrinking _ the edge @xmath23 .
a _ dominating set _ in a hypergraph @xmath0 is a subset @xmath2 of vertices of @xmath0 such that for every vertex @xmath44 there exists an edge @xmath42 for which @xmath43 and @xmath45 .
equivalently , every vertex @xmath44 is adjacent to a vertex in @xmath2 .
the minimum cardinality of a dominating set in @xmath0 is called its _ domination number _ , denoted by @xmath7 .
a _ matching _ in @xmath0 is a set of disjoint edges .
the maximum cardinality of a matching in @xmath0 is called the _ matching number _ ,
denoted by @xmath1 .
a subset @xmath46 of vertices in @xmath0 is a _ transversal _
( also called _ cover _ ) if @xmath46 has a nonempty intersection with each edge of @xmath0 . the _ transversal number _ ,
@xmath47 , is the minimum size of a transversal of @xmath0 .
transversals and matchings in hypergraphs are well studied in the literature ( see e.g. @xcite ) and elsewhere .
domination in hypergraphs , was introduced by acharya @xcite and studied further in @xcite . for a hypergraph @xmath0 of rank @xmath8 ,
when @xmath48 , @xmath0 is a graph , haynes et al .
@xcite observed that @xmath49 . when @xmath50 , by definitions , clearly @xmath51 and @xmath52 still hold .
the extremal graphs , i.e. , linear @xmath17-uniform hypergraphs achieving @xmath53 were studied in @xcite recently , arumugam et al .
@xcite investigated the hypergraphs of rank @xmath50 satisfying @xmath54 , and proved that their recognition problem is np - hard on the class of linear hypergraphs of rank 3 .
in @xcite we observed that the inequality @xmath55 does not hold for a hypergraph @xmath0 of rank @xmath50 , and the difference @xmath56 can be arbitrarily large .
further , we obtained the following inequality .
( @xcite)[thm0 ] if @xmath0 is a hypergraph of rank @xmath57 without isolated vertex , then @xmath58 and this bound is sharp . in particular , if @xmath48 in theorem [ thm0 ] , as haynes et al .
@xcite observed , @xmath55 . for extremal hypergraphs of rank
@xmath8 satisfying @xmath59 , randerath et al .
@xcite gave a characterization of graphs ( hypergraphs of rank @xmath48 ) with minimum degree two .
in 2010 , kano et al .
@xcite provided a complete characterization of graphs with minimum degree one .
for the case when rank @xmath60 , we give a complete characterization of hypergraphs @xmath0 in @xcite .
for the case when @xmath61 , a constructive characterization of hypergraphs with @xmath59 seems difficult to obtain .
thus we restrict our attention to intersecting hypergraphs .
a hypergraph is _ intersecting _ if any two edges have nonempty intersection .
clearly , @xmath0 is intersecting if and only if @xmath62 .
intersecting hypergraphs are well studied in the literature ( see , for example , @xcite ) . for an intersecting hypergraph @xmath0 of rank @xmath8
, we immediately have @xmath9 . in this paper
we first give some structural properties on the intersecting hypergraphs of rank @xmath8 achieving the equality @xmath10 . by applying the properties and fano plane
, we provides a complete characterization of linear intersecting hypergraphs @xmath0 of rank @xmath11 satisfying @xmath10 .
in this section we give some structural properties on intersecting hypergraphs of rank @xmath8 satisfying @xmath10 .
the properties play an important role in the characterization of intersecting hypergraphs of rank @xmath11 with @xmath10 .
let @xmath63 be a family of intersecting hypergraphs of rank @xmath8 in which each hypergraph @xmath0 satisfies @xmath64 .
[ pro2.1 ] for every @xmath65 , there exists an @xmath8-uniform spanning partial hypergraph @xmath66 of @xmath0 such that every edge in @xmath66 contains exactly one degree-@xmath67 vertex .
* let @xmath68 .
we define recursively the hypergraph @xmath69 by @xmath70 .
if there exists an edge @xmath71 such that @xmath72 for each vertex @xmath6 in @xmath73 , then set @xmath74 for @xmath75 . by repeating this process until every edge which remains contains at least one degree-@xmath67 vertex , we obtain a spanning partial hypergraph of @xmath0 .
assume that the above process stops when @xmath76 .
let @xmath77 .
then @xmath66 is a spanning partial hypergraph of @xmath0 .
clearly , every edge in @xmath66 contains at least one degree-1 vertex and @xmath66 is still intersecting .
we claim that each edge in @xmath66 contains exactly one degree-@xmath67 vertex .
suppose not .
then there exists an edge @xmath23 containing at least two degree-@xmath67 vertices .
let @xmath78 . then @xmath79 .
since @xmath66 is intersecting , @xmath2 is a transversal of @xmath66 , so @xmath80 .
since @xmath66 is a spanning partial hypergraph of @xmath0 , we have @xmath81 , contradicting the assumption that @xmath10 .
further , we show that @xmath66 is @xmath8-uniform .
suppose not .
let @xmath82 be an edge of @xmath66 such that @xmath83 and @xmath5 the unique degree-@xmath67 vertex of @xmath82 . since @xmath66 is intersecting , @xmath84 is a dominating set of @xmath66 .
thus @xmath81 , contradicting @xmath10 again . for each @xmath65 ,
let @xmath66 be the @xmath8-uniform spanning partial hypergraph of @xmath0 in lemma [ pro2.1 ] .
further , let @xmath33 be the hypergraph obtained from @xmath66 by shrinking every edge to @xmath85-edge by removing the degree-1 vertex from each edge of @xmath66 and deleting multiple edges , if any .
obviously , @xmath33 is an @xmath85-uniform intersecting hypergraph .
[ pro2.2 ] for every @xmath65 , @xmath86 . * proof .
* let @xmath87 and @xmath5 be the unique degree-1 vertex in @xmath23 . since @xmath66 is intersecting , @xmath88 is a transversal of @xmath66 , so @xmath89 . on the other hand , note that @xmath90 .
hence @xmath91 . by the construction of @xmath33 , clearly any transversal of @xmath33 is a transversal of @xmath66 and @xmath92 is also a transversal of @xmath33 .
hence @xmath93 .
the equality chain follow .
[ pro2.3 ] for @xmath50 and every vertex @xmath6 in @xmath33 , @xmath94 . * proof . *
suppose , to the contrary , that there exists a vertex @xmath95 such that @xmath96 or @xmath97 .
suppose that @xmath96 .
note that every vertex @xmath33 has degree at least 2 , so @xmath98 .
hence @xmath99 .
let @xmath23 is an edge containing @xmath6 .
since @xmath33 is intersecting , @xmath100 for any @xmath101 .
in particular , if @xmath102 , then @xmath103 since @xmath104 and @xmath99 .
thus @xmath105 would be a transversal of @xmath33 , contracting the fact in lemma [ pro2.2 ] .
suppose that @xmath106 .
let @xmath107 be the edges whose pairwise intersection is only @xmath6 .
by lemma [ pro2.2 ] and @xmath50 , we have @xmath108 .
this implies that there exists an edge @xmath109 such that @xmath110 .
since @xmath33 is intersecting , @xmath111 for each @xmath112 .
but then @xmath113 , contracting the fact that @xmath33 is @xmath85-uniform .
[ pro2.4 ] let @xmath65 ( @xmath50 ) .
if @xmath0 is linear , then every edge of @xmath33 has at most one degree-2 vertex and @xmath114 .
* proof . *
if @xmath0 is linear , so is @xmath33 .
first , we show that every edge of @xmath33 has at most one degree-2 vertex .
suppose not , and let @xmath115 such that @xmath116 .
then there exists two distinct edges @xmath117 such that @xmath118 for @xmath119 .
since @xmath33 is a linear intersecting hypergraph , we have @xmath120 , so there exists a vertex @xmath121 such that @xmath122 .
then @xmath123 is a transversal of @xmath33 .
consequently , @xmath124 , a contradiction to @xmath125 .
next we show that @xmath114 .
suppose not , let @xmath126 where @xmath127 . as we have seen , @xmath33 is a linear intersecting @xmath85-uniform hypergraph with @xmath125 . then @xmath128 . let @xmath95 such that @xmath129
. then @xmath130 .
note that @xmath131 is a transversal of @xmath33 . but @xmath132 .
thus @xmath133 , contradicting that @xmath125 .
[ pro2.5 ] let @xmath65 ( @xmath50 ) .
if @xmath0 is linear , then @xmath134 , @xmath135 , and so @xmath136 . * proof . * since @xmath33 is a linear intersecting @xmath85-uniform hypergraph , @xmath137 for any edge @xmath138 . by lemma [ pro2.4 ]
, we immediately have @xmath139 .
we now show that @xmath135 .
let @xmath95 such that @xmath140 . then @xmath141 .
suppose that @xmath142 .
then there exists @xmath143 such that @xmath144 . by lemma
[ pro2.3 ] , @xmath145 , so there exist two edges @xmath146 and @xmath147 such that @xmath148 .
clearly , @xmath149 and @xmath150 . since @xmath33 is linear intersecting , @xmath146 intersects each one of the edges that contains @xmath6 , implying that @xmath151 .
this contradicts that @xmath33 is an @xmath85-uniform hypergraph .
therefore , @xmath135 , that is , @xmath152 .
this implies that @xmath153 is a dominating set of @xmath33 , so @xmath136 .
in the section we give a complete characterization of linear intersecting hypergraphs @xmath0 of rank @xmath11 with @xmath10 . for this purpose ,
let @xmath155 be the fano plane and let @xmath156 be the hypergraph obtained from @xmath155 by deleting any edge of @xmath155 .
the two hypergraphs @xmath155 and @xmath157 are shown in fig . 1 . and the hypergraph @xmath156,height=158 ] [ thm4.1 ] let @xmath158 and @xmath33 be the hypergraph as defined in the above section . if @xmath0 is linear , then @xmath159 or @xmath160 .
* by lemma [ pro2.5 ] , we have @xmath161 and @xmath162 for @xmath163 . note that @xmath33 is a linear intersecting 3-uniform hypergraph .
if @xmath164 , then @xmath33 must be the fano plane @xmath155 .
if @xmath165 , then @xmath33 is the hypergraph @xmath157 ( see fig .
1 ) . to complete our characterization , we let @xmath166 ( @xmath167 ) be the hypergraph obtained from @xmath155 ( @xmath157 ) by adding a new vertex to each edge of @xmath155 ( @xmath157 ) , respectively .
let @xmath168 be the hypergraph obtained from @xmath166 by shrinking one edge to @xmath169-edge by removing the degree-1 vertex in the edge .
let @xmath170 be the hypergraph obtained from @xmath167 by adding a new edge @xmath171 where @xmath172 and @xmath173 lie in distinct edges of @xmath167 and @xmath174 , @xmath175 .
we define @xmath176 ( see fig .
2 ) . clearly , every hypergraph in @xmath177 is a linear intersecting hypergraph of rank 4 .
[ fig2 ] and @xmath170,title="fig:",height=158 ] [ thm4.2 ] for a linear intersecting hypergraph @xmath0 of rank @xmath11 , @xmath154 if and only if @xmath178 . * proof . * first , suppose that @xmath178 , and let @xmath23 be an arbitrary edge of @xmath0 containing four vertices and @xmath6 the degree-1 vertex . then it is easy to check that @xmath179 is a minimum dominating set of @xmath0
. thus @xmath154 .
conversely , suppose that @xmath154 , we show that @xmath178 .
let @xmath66 and @xmath33 be the hypergraphs corresponding to @xmath0 as defined in above section . by lemma [ thm4.1 ] , @xmath180 or @xmath181 , so @xmath182 or @xmath183 .
_ case _ 1 .
@xmath184 . in this case
, we claim that @xmath185 .
it suffices to show that @xmath186 .
suppose not .
let @xmath187 .
then @xmath188 for any @xmath189 . by the construction of @xmath166 , @xmath190 where @xmath191 consists of seven degree-1 vertices in @xmath166 .
note that any two vertices of @xmath155 lie in exactly one common edge of @xmath166 , so @xmath192 .
this implies that @xmath193 , since @xmath0 is linear and intersecting .
this contradicts that @xmath194 . _ case _ 2 . @xmath183 . in this case , we show that @xmath195 .
it suffices to show that @xmath196
if @xmath197 and @xmath198 . let @xmath199 where @xmath200 is the set of degree-@xmath201 vertices in @xmath202 . then @xmath203 and @xmath204 .
suppose now that @xmath197 and @xmath198 .
then @xmath205 . let @xmath206 .
suppose that @xmath207 . since @xmath0 is linear and intersecting , @xmath188 for any @xmath208 .
note that any two vertices of @xmath209 lie in exactly one common edge of @xmath202 , so @xmath210 .
let @xmath211 .
this implies that @xmath212 .
but then @xmath213 is a dominating set of @xmath0 , contradicting that @xmath154 .
hence @xmath214 , and thus @xmath215 .
suppose that @xmath216
. then @xmath217 .
hence @xmath193 , a contradiction .
so @xmath218 since @xmath198
. it immediately follows that @xmath219 . therefore , @xmath196 .
in this paper we present the propositions of the intersecting hypergraphs that achieve the equality @xmath10 . especially , we provide a complete characterization of the linear intersecting hypergraphs with rank @xmath163 satisfying @xmath154 .
one is interested in characterizing the extremal intersecting hypergraphs with rank @xmath163 satisfying @xmath154 .
r. laskar , h.b .
valikar , on domination related concepts in graph theory , in : s.b .
rao ( ed . ) , combinatorics and graph theory , lecture notes in mathematics , vol .
885 , springer , berlin , 1981 .
pp . 308320 . | a matching in a hypergraph @xmath0 is a set of pairwise disjoint hyperedges .
the matching number @xmath1 of @xmath0 is the size of a maximum matching in @xmath0 .
a subset @xmath2 of vertices of @xmath0 is a dominating set of @xmath0 if for every @xmath3 there exists @xmath4 such that @xmath5 and @xmath6 lie in an hyperedge of @xmath0 .
the cardinality of a minimum dominating set of @xmath0 is called the domination number of @xmath0 , denoted by @xmath7 .
it is known that for a intersecting hypergraph @xmath0 with rank @xmath8 , @xmath9 . in this paper
we present structural properties on intersecting hypergraphs with rank @xmath8 satisfying the equality @xmath10 . by applying the properties we show that all linear intersecting hypergraphs @xmath0 with rank @xmath11 satisfying @xmath10 can be constructed by the well - known fano plane .
* keywords : * hypergraph ; intersecting hypergraph ; domination ; matching ; linear hypergraph * ams ( 2000 ) subject classification : * 05c65 , 05c69 , 05c70 |
consider the elliptic self - adjoint operator @xmath0 where @xmath1 are smooth functions in @xmath2 , a smooth bounded open subset of @xmath3 , satisfying @xmath4 for some @xmath5 .
it is well - known that the minimum value @xmath6 in the rayleigh - ritz variational formula @xmath7 is attained at some function @xmath8 satisfying @xmath9 the number @xmath6 is usually referred to as the principal eigenvalue of @xmath10 in @xmath11 and @xmath8 is the corresponding principal eigenfunction . for operators of the form and also more general linear operator in divergence form there is a vast literature on computational methods for the principal eigenvalue , see for example @xcite , @xcite , @xcite , @xcite .
general non - divergence type elliptic operators , namely @xmath12 are not self - adjoint and the spectral theory is then much more involved : in particular , the rayleigh - ritz variational formula is not available anymore . in the seminal paper @xcite by m.d .
donsker and s.r.s .
varadhan , a min - max formula for the principal eigenvalue of a class of elliptic operators @xmath10 including ( [ lnonself ] ) was proved , namely @xmath13 in that papers other representation formulas for @xmath14 were also proposed in terms of large deviations and of the average long run time behavior of the positive semigroup generated by @xmath10 .
a further crucial step in that direction is the paper @xcite by h. berestycki , l. nirenberg and s.r.s .
varadhan , where the validity of formula ( [ pe2intro ] ) is proved under mild smoothness assumptions ( @xmath11 a bounded open set and @xmath15 , @xmath16 , @xmath17 ) .
moreover it is proved that is equivalent to @xmath18 following this path of ideas , notions of principal eigenvalue for fully nonlinear uniformly elliptic operators of the form @xmath19= f(x , u(x ) , du(x ) , d^2 u(x))\ ] ] have been introduced and analyzed in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . a by now established definition of principal eigenvalue is given by @xmath20+{\lambda}{\varphi}\le 0\quad\text{in $ { \omega}$}\}\,\ ] ] where the inequality in is intended in viscosity sense .
it is possible to prove under appropriate assumptions , see - , that there exists a viscosity solution @xmath8 of @xmath21 + { \lambda}_1 w_1(x)=0 \qquad & x\in { \omega } , \\ w_1(x)=0 & x\in \partial{\omega}. \end{array } \right.\ ] ] moreover the characterization still holds in this nonlinear setting .
as it is well - known , the principal eigenvalue plays a key role in several respects , both in the existence theory and in the qualitative analysis of elliptic partial differential equations as well in applications to large deviations @xcite , @xcite , bifurcation issues @xcite , ergodic and long run average cost problems in stochastic control @xcite . for linear non self - adjoint operators and , a fortiori , for nonlinear ones
the principal eigenvalue can be explicitly computed only in very special cases , see e.g. @xcite , hence the importance to devise numerical algorithms for the problem .
but , apart some specific case ( see @xcite for the @xmath22-laplace operator ) , approximation schemes and computational methods are not available in the literature , at least at our present knowledge .
the aim of this paper is to define a numerical scheme for the principal eigenvalue of nonlinear uniformly elliptic operators via a finite difference approximation of formula .
more precisely , denoting by @xmath23 the orthogonal lattice in @xmath3 where @xmath24 is a discretization parameter , we consider a discrete operator @xmath25 acting on functions defined on a discrete subset @xmath26 of @xmath11 and the corresponding approximated version of , namely @xmath27(x)}{{\varphi}(x)}.\ ] ] as for the approximating operators @xmath25 , we consider a specific class of finite difference schemes introduced in @xcite , @xcite since they satisfy some useful properties for the convergence analysis . we prove that if @xmath28 is uniformly elliptic and satisfies in addition some quite natural further conditions , then it is possible to define a finite difference scheme @xmath25 such that the discrete principal eigenvalues @xmath29 and the associated discrete eigenfunctions @xmath30 converge uniformly in @xmath31 , as the mesh step @xmath32 is sent to @xmath33 , respectively to the principal eigenvalue @xmath34 and to the corresponding eigenfunction @xmath8 for the original problem ( [ pe ] ) .
it is worth pointing out that the proof of our main convergence result , theorem [ main ] , can not rely on standard stability results for fully nonlinear partial differential equations , see @xcite , since the limit problem does not satisfy a comparison principle ( see remark [ convergence ] for details ) .
we mention that our approach is partially inspired by the paper @xcite where a similar approximation scheme is proposed for the computation of effective hamiltonians occurring in the homogenization of hamilton - jacobi equations which can be characterized by a formula somewhat similar to . in section [ sect2 ]
we introduce the main assumptions and we investigate some issues related to the maximum principle for discrete operators .
in section [ sect3 ] we study the approximation method for a class of finite difference schemes and we prove the convergence of the scheme . in section [ sect4 ] we show that under some additional structural assumptions on @xmath25 the inf - sup problem can be transformed into a convex optimization problem on the nodes of the grid and we discuss its implementation .
a few tests which show the efficiency of our method on some simple examples are reported in section [ sect4 ] as well .
we start by fixing some notations and the assumptions on the operator @xmath28 .
set @xmath35 , where @xmath36 denotes the linear space of real , symmetric @xmath37 matrices .
the function @xmath38 is assumed to be continuous on @xmath39 and locally uniformly lipschitz continuous with respect to @xmath40 for each fixed @xmath41 .
we will also suppose that the partial derivatives @xmath42 , @xmath43 , @xmath44 satisfy the following structure conditions : @xmath45 for some constants @xmath46 , @xmath47 , @xmath48 , @xmath49 .
a further condition is the positive homogeneity of degree @xmath50 , that is @xmath51 the principal eigenvalue of problem is defined by @xmath52+{\lambda}{\varphi}\le 0\ \mbox{in}\ { \omega}\},\ ] ] where the differential inequality @xmath53+{\lambda}{\varphi}\le 0 $ ] is meant in the viscosity sense . under assumptions - ,
there exists a viscosity solution of and the characterization of @xmath34 holds ( see @xcite , @xcite ) .
[ negativeigenv ] it is possible to define @xmath54+{\lambda}{\varphi}\ge 0\ \mbox{in}\ { \omega}\}.\ ] ] when @xmath28 is not odd in its dependence on the hessian , then in general @xmath55 .
of course it is possible to see @xmath56 as @xmath34 of some other operator .
hence we will only consider in this paper @xmath34 .
for example , for the extremal pucci operators @xmath57 and @xmath58 , since @xmath59 , the following holds @xmath60 [ cpositive ] the assumption @xmath61 , i.e. the monotonicity of the differential operator in the zero - order term , could be removed .
indeed @xmath62 , with @xmath63 large , satisfies this assumption , moreover @xmath64 and @xmath28 have the same principal eigenfunction and the eigenvalues differ by @xmath63 .
we now describe the discrete setting that we shall consider . given @xmath24 , let @xmath23 denote the orthogonal lattice in @xmath3 .
let @xmath25 be a discrete operator acting on functions defined in @xmath26 .
we shall consider an approximation of ( which can be seen also as an eigenvalue problem for the discrete operator @xmath25 ) .
we look for a number @xmath65 and a positive function @xmath66 such that @xmath67_x)+{\lambda}w(x ) = 0 \qquad & x\in { \omega}_h,\\ w(x)=0 & x\in \partial{\omega}_h , \end{array } \right.\ ] ] where * @xmath24 is the discretization parameter ( @xmath32 is meant to tend to @xmath33 ) , * @xmath68 is the point where is approximated , * @xmath66 is a real valued mesh function in @xmath69 meant to approximate the viscosity solution of , * @xmath70_x$ ] represents the stencil of the scheme , i.e. the points in @xmath71 where the value of @xmath72 is computed for writing the scheme at the point @xmath73 ( we assume that @xmath74_x$ ] is independent of @xmath75 for @xmath76 for some fixed @xmath77 ) .
we denote by @xmath78 the space of the mesh functions defined on @xmath79 and we introduce some basic assumptions for the scheme @xmath25 ( see @xcite , @xcite ) .
* the operator @xmath25 is of positive type , i.e. for all @xmath80 , @xmath81 , @xmath82 satisfying @xmath83 for each @xmath84 , then @xmath85_x)\ge f_h(x , z,[u]_x)\ge f_h(x , z+\tau , [ u+\eta]_x)\ ] ] * the operator @xmath25 is positively homogeneous , i.e. for all @xmath80 , @xmath86 , @xmath87 and @xmath88 , then @xmath89_x)= t f_h(x , z , [ u]_x).\ ] ] * the family of operators @xmath90 , where @xmath91 is a positive constant , is consistent with the operator @xmath28 on the domain @xmath92 , i.e. for each @xmath93 @xmath94_x)\right|\to 0\quad\text{as $ h\to 0$,}\ ] ] uniformly on compact subset of @xmath31 .
we study below some properties related to the maximum principle and a comparison result for the operator @xmath25 .
let us start by the following definitions : a function @xmath87 is a subsolution ( respectively @xmath95 is a supersolution ) of @xmath96_x ) = f(x ) \qquad x\in{\omega}_h\ ] ] if @xmath97_x ) \ge
f(x),\qquad x\in{\omega}_h\\ & \left(\text{respectively , } f_h(x , v ( x),[v]_x ) \le f(x),\qquad x\in{\omega}_h\right ) .
\end{aligned}\ ] ] [ def_mp ] the maximum principle holds for the operator @xmath25 in @xmath98 if @xmath99_x)\ge 0\qquad & \text{in $ { \omega}_h$,}\\ u\le 0&\text{on $ \partial { \omega}_h$ , } \end{array } \right.\ ] ] implies @xmath100 in @xmath98 .
[ prop_mp ] assume that @xmath25 is of positive type and positive homogeneous and satisfies either @xmath101_x}\eta(y)>0$,}\\ \text{then } \quad f_h(x , z,[u+\eta]_x)>f_h(x , z,[u]_x)\qquad \end{split}\ ] ] or @xmath102_x)\ge f_h(x , z+\tau , [ u+\eta]_x)+c_0\tau\qquad \end{split}\ ] ] for some positive constants @xmath63 .
then the maximum principle holds for the operator @xmath25 in @xmath98 .
assume by contradiction that @xmath72 satisfies and @xmath103 .
let @xmath104 be such that @xmath105 . since @xmath106 on @xmath107 , it is not restrictive to assume that there exists @xmath108 such that @xmath109 . hence @xmath110_{\bar x})\le f_h(\bar x , u(\bar x)-m , [ u - m]_{\bar x } ) \\ & < f_h(\bar x,0,[0]_{\bar x})= 0,\end{aligned}\ ] ] a contradiction .
a similar proof can be done with the assumption . the assumptions and correspond to the uniform ellipticity and , respectively , to the strict monotonicity of the operator @xmath28 with respect to the zero - order term .
the following proposition shows that , as it is known in the continuous case ( see for example @xcite ) , the validity of the maximum principle for subsolutions of the operator @xmath25 is equivalent to the positivity of the principal eigenvalue for @xmath25 .
[ prop_mp2 ] assume that the scheme @xmath25 is of positive type and that it is positively homogeneous .
suppose that for @xmath111 , there exists a nonnegative grid function @xmath112 with @xmath113 in @xmath98 such that @xmath114+{\lambda}{\varphi}\le 0 $ ] .
if , for @xmath115 , the function @xmath72 satisfies @xmath116_x)+\tau u\ge 0\qquad & \text{in $ { \omega}_h$}\\ u\le 0&\text{on $ \partial { \omega}_h$ , } \end{array } \right.\ ] ] then @xmath100 in @xmath117 , i.e. @xmath118=f_h[\cdot]+\tau \cdot$ ] satisfies the maximum principle .
suppose by contradiction that @xmath119 .
let @xmath112 as in the statement and set @xmath120 ( note that the maximum is taken only with respect to the internal points ) .
then @xmath121 is continuous , decreasing , @xmath122 and @xmath123 for @xmath124 . hence there exists @xmath125 such that @xmath126 .
moreover , since @xmath127 on @xmath128 , we also have @xmath129 .
let @xmath130 be such that @xmath131 and set @xmath132 .
then @xmath133+{\lambda}\psi\le0 $ ] and @xmath134 for some @xmath135 . hence @xmath136 and @xmath137 . since @xmath25 is of positive type
, it follows that @xmath138_{\bar x})&\ge f_h((x,\psi(\bar x)+m,[\psi+m]_{\bar x})=f_h(x , u(\bar x),[\psi+m]_{\bar x})\\ & \ge f_h(x , u(\bar x),[u]_{\bar x}).\end{aligned}\ ] ] then @xmath139(\bar x)\ge -f_h[\psi](\bar x)\ge { \lambda}\psi(\bar x ) = { \lambda}{{\gamma}}{\varphi}(\bar x)\ge{\lambda}\frac{{{\gamma}}}{{{\gamma}}'}u(\bar x)\end{aligned}\ ] ] and therefore a contradiction to .
the following result gives a comparison principle for .
[ prop_comp ] assume that @xmath25 is of positive type and it satisfies either or .
let @xmath72 and @xmath140 be a subsolution and respectively a supersolution of such that @xmath141 on @xmath142 . then @xmath141 in @xmath143 .
suppose by contradiction that @xmath144 and let @xmath104 be such that @xmath145 . hence @xmath146 in @xmath98 and it is not restrictive to assume that @xmath147_{\bar x}}(v+m - u)>0 $ ] .
it follows that @xmath148_{\bar x})&=f_h(\bar x , v(\bar x)+m , [ u]_{\bar x } ) < f_h(\bar x , v(\bar x)+m , [ v+m]_{\bar x})\\ & \le f_h(\bar x , v(\bar x ) , [ v]_{\bar x})\le f(\bar x)\end{aligned}\ ] ] and therefore a contradiction . a similar proof can be carried on under assumption .
in this section we consider a specific class of finite difference schemes introduced in @xcite .
these schemes satisfy certain pointwise estimates which are the discrete analogues of those valid for a general class of fully nonlinear , uniformly elliptic equations .
+ we assume that for all @xmath149 , the stencil @xmath70_x$ ] of the scheme is given by @xmath150 where @xmath151 is a finite set containing all the vectors of the canonical basis of @xmath3 . then we consider a discrete operator @xmath25 in given by a finite difference scheme written in the form @xmath152={{{\cal f } } } ( x , u , { { \delta}}_h u,{{\delta}}^2_h u),\ ] ] where @xmath153 and for @xmath154 , @xmath87 @xmath155 set @xmath156 and denote by @xmath157 the generic points in @xmath158 . the operator @xmath25 given by is of positive type if @xmath159 & \frac{\partial{{\cal f}}}{\partial z}\le 0 , \label{p2_scheme}\end{aligned}\ ] ] and positively homogeneous if @xmath160 moreover if @xmath28 in satisfies the assumptions , then it is always possible to find a scheme of type which is consistent with @xmath28 and which , besides - , satisfies for all @xmath154 , the bounds @xmath161 where @xmath162 , @xmath163 , @xmath164 are constants depending on @xmath46 , @xmath47 , @xmath48 , @xmath49 in ( see @xcite , @xcite ) .
note that in particular implies .
+ we recall some important properties of the previous scheme ( for the proof we refer to @xcite ) [ prop_wellposed ] assume - and let @xmath165 , @xmath166 be two given mesh functions .
then for every @xmath24 sufficiently small there exists a unique solution @xmath167 to the dirichlet problem @xmath168_x)=f\quad & x\in { \omega}_h , \\ u = g & x\in { { \partial}}{\omega}_h .
\end{array } \right.\ ] ] [ prop_abp ] assume - and let @xmath169 be a subsolution of .
then @xmath170 where the constant @xmath171 is independent of @xmath32 .
moreover if @xmath169 is a solution of , then for any @xmath172 @xmath173 where @xmath174 , @xmath175 , @xmath176 and @xmath171 are positive constants independent of @xmath32 .
we give an example of a scheme of the form .
consider the hamilton - jacobi - bellman operator @xmath177 where @xmath178 it is always possible to rewrite the operator @xmath179 in in the following form ( see @xcite ) @xmath180 where @xmath181 and @xmath151 is a finite set containing all the vectors of the canonical basis in @xmath3 .
moreover the coefficients @xmath182 , @xmath183 and @xmath184 satisfy the same properties of @xmath185 , @xmath186 and @xmath187 .
then we consider @xmath188(x):=\sup_{{{\alpha}}\in a}\inf_{{{\beta}}\in b } l_h^{{{\alpha}}{{\beta}}}u(x)\ ] ] where @xmath189 for @xmath190 with @xmath191 the previous scheme reads as @xmath192 in this part we assume that the operator @xmath28 in is linear , i.e. @xmath193=lu$ ] with @xmath194 and we consider a scheme defined as in
, obviously without the dependence on @xmath195 , @xmath196 . [ kr ] under the assumption the eigenvalue problem has a simple eigenvalue @xmath197 which corresponds to a positive eigenfunction .
the other eigenvalues correspond to sign changing eigenfunctions .
choose @xmath198 large enough so that @xmath199 and set @xmath200_x)=l_h ( x , t,[u]_x)-\xi t.\ ] ] let @xmath201 be the positive cone of the nonnegative grid functions in @xmath78 .
for a given grid function @xmath165 , by proposition [ prop_wellposed ] and proposition [ prop_comp ] there exists a unique solution @xmath87 to @xmath202_y)+f=0\qquad & \text{in $ { \omega}_h$,}\\ u= 0&\text{on $ \partial { \omega}_h$. } \end{array } \right.\ ] ] since @xmath78 is a finite dimensional space it follows that @xmath203 defined by @xmath204 is a compact linear operator .
moreover , if @xmath205 , then by proposition [ prop_mp ] @xmath206 and if @xmath207 , @xmath208 .
+ therefore , by the krein - rutman theorem @xcite , @xmath209 the spectral radius of @xmath210 is a simple real eigenvalue @xmath211 with a positive eigenfunction @xmath72 such that @xmath212 .
hence for @xmath213 , @xmath214 satisfies @xmath215_x ) + { \lambda}_{1,h}w_1=0\qquad & \text{in $ { \omega}_h$,}\\ w_1= 0&\text{on $ \partial { \omega}_h$}. \end{array } \right.\ ] ] the following characterization of @xmath29 is a simple consequence of proposition [ prop_mp2 ] .
we have @xmath216+{\lambda}{\varphi}\le 0\ \mbox{in}\ { \omega}$}\right\},\ ] ] or , equivalently , @xmath217(x)}{{\varphi}(x)}\right\}.\ ] ] denote by @xmath218 the right hand side of . clearly @xmath219 .
if @xmath220 then there exist @xmath221 and @xmath113 such that @xmath222+\mu { \varphi}\le 0 $ ] .
a contradiction follows immediately by proposition [ prop_mp2 ] since the eigenfunction corresponding to @xmath29 is positive .
hence we have .
+ let @xmath113 such that @xmath222(x)+{\lambda}{\varphi}(x)\le 0 $ ] for @xmath80 .
hence @xmath223}{{\varphi}}\right\}= -\sup_{{\omega}_h}\left\{\frac{l_h[{\varphi}]}{{\varphi}}\right\}.\ ] ] consequently @xmath224(x)}{{\varphi}(x)}\right\}\right)=-\inf_{{\varphi}>0}\sup_{x\in{\omega}_h}\left\{\frac{l_h[{\varphi}](x)}{{\varphi}(x)}\right\}.\ ] ] we give next an upper bound for @xmath29 ( compare with the corresponding estimate for @xmath34 in @xcite , lemma 1.1 ) .
let @xmath225 and assume that @xmath226 lies in @xmath31 with @xmath227 .
then @xmath228 given the linear operator @xmath229 let @xmath230 , @xmath231 , @xmath232 be positive constants such that @xmath233 and @xmath234 in @xmath31 .
let @xmath235 and assume for simplicity that @xmath236 for some @xmath237 .
set @xmath238 and consider the grid function @xmath239 then for @xmath240 we have @xmath241 denote by @xmath242 , @xmath16 and @xmath243 the coefficients of the linear operator computed at the point @xmath244 . since @xmath245 it follows that @xmath246(ih)}{4{{\sigma}}_i}&\le \frac { a_i}{(r^2-|ih|^2)}-\frac{a_i |hi|^2}{(r^2-|ih|^2)^2 } + |b_i|\frac{2r}{(r^2-|ih|^2)}+\frac{c_i}{4}\\ & \le \frac { { { \gamma}}_0+br}{(r^2-|ih|^2)}- \frac{{{\gamma}}_0|hi|^2}{(r^2-|ih|^2)^2}+\frac{b}{4}. \end{split}\ ] ] if @xmath247 then the second term in dominates the first one and therefore @xmath248(ih)}{4{{\sigma}}_i}\le \frac{b}{4}.\ ] ] in the remaining part of @xmath249 , @xmath250(ih)}{4{{\sigma}}_i}\le \frac { { { \gamma}}_0+br}{(r^2-|ih|^2)}+\frac{b}{4}\le \frac{b}{4}+\frac{1}{{{\gamma}}_0r^2}({{\gamma}}_0+br)({{\gamma}}_0+{{\gamma}}_0+br).\ ] ] by and , we get @xmath251(ih)}{{{\sigma}}_i}\right)\le\frac{c}{r^2}\quad\text{for $ i =- n+1,\dots , n-1$.}\ ] ] to conclude the proof , we show that if for some positive function @xmath112 and @xmath111 , @xmath222+{\lambda}{\varphi}\le 0 $ ] , then @xmath252/{{\sigma}})$ ] . for this purpose , assume that @xmath253/{{\sigma}}):=\tau$ ] ; then @xmath254+\tau { { \sigma}}\ge 0 $ ] in @xmath249 and @xmath255 on @xmath256 , while @xmath222+\tau { \varphi}\le 0 $ ] .
hence by proposition [ prop_mp2 ] , it follows @xmath257 in @xmath249 , a contradiction , and therefore @xmath258 .
we consider now a general discrete operator @xmath25 given by and we study the corresponding eigenvalue problem . in analogy with formula , we define @xmath259+{\lambda}{\varphi}\le 0$}\right\}\ ] ] we prove for each @xmath32 the existence of a pair @xmath260 satisfying with @xmath261 in @xmath98 . [
ped : prop_exlower ] assume that @xmath25 satisfies , @xmath262 and @xmath263 .
then there exists a nonnegative solution to @xmath264_x)+ { \lambda}u(x ) = f(x ) \qquad & x\in { \omega}_h,\\ u(x)=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] we can assume @xmath265 , since for @xmath266 , @xmath267+{\lambda}u$ ] satisfies and therefore by propositions [ prop_wellposed ] and there exists a unique solution to problem .
+ let us define by induction a sequence @xmath268 by setting @xmath269 and , for @xmath270 we consider the equation : @xmath271_x ) = f(x)-{\lambda}u_{n},\qquad & x\in { \omega}_h,\\ u_{n+1}(x)=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] + for any @xmath272 there exists a non negative solution @xmath273 to . for @xmath225 ,
existence follows by proposition [ prop_wellposed ] .
moreover since @xmath269 is a subsolution to , by proposition [ prop_comp ] we get @xmath274 .
the existence of a non negative solution at the @xmath275-step is proved in a similar way ; moreover the solution is non negative since @xmath276 .
+ we claim now that , for any @xmath277 , @xmath278 . for @xmath225 the claim is trivially true since @xmath274 .
assume then by induction that @xmath279 .
since @xmath280 it follows that @xmath268 is a subsolution of . by proposition [ prop_comp ] , we get that @xmath281 . + let us show now that the sequence @xmath268 is bounded
. assume by contradiction that it is false and set @xmath282 .
then , by positive homogeneity , @xmath283 is a solution of @xmath284_x ) = \frac{f(x)}{|u_{n+1}|_\infty}-{\lambda}\frac{u_{n}}{|u_{n+1}|_\infty},\qquad x\in { \omega}_h.\ ] ] since the sequence @xmath283 is bounded , then up to a subsequence it converges to a function @xmath285 , while @xmath286 converges to @xmath287 where @xmath288 .
hence @xmath289 , @xmath290 , @xmath291 on @xmath142 and @xmath292_x ) + k{\lambda}\overline u = 0,\qquad x\in { \omega}_h.\ ] ] since @xmath293 and using the fact that for @xmath294 there exists by definition @xmath113 such that @xmath114+{\lambda}{\varphi}\ge 0 $ ] in @xmath98 , we get a contradiction to proposition [ prop_mp2 ] .
hence the sequence @xmath268 is bounded , and being in addition monotone , it converges pointwise to a function @xmath72 which solves .
the next result shows that @xmath29 is indeed an eigenvalue for the approximated operator @xmath25 .
[ ped : prop_ex ] assume that @xmath25 satisfies .
then there exists @xmath261 in @xmath98 satisfying @xmath295_x)+ { \lambda}_{1,h}w_{1,h}(x ) = 0 \qquad & x\in { \omega}_h,\\ w_{1,h}=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] moreover that the characterization is still valid for the nonlinear operator @xmath25 .
let @xmath296 be an increasing sequence converging to @xmath29 .
by proposition [ ped : prop_exlower ] there exists a positive solution @xmath268 of @xmath297_x)+{\lambda}_n u_{n } = -1,\qquad & x\in { \omega}_h,\\ u_{n}(x)=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] we claim that @xmath268 is not bounded .
assume by contradiction that @xmath268 is bounded so that , up to a subsequence , @xmath268 converges to a function @xmath298 which solves @xmath299_x)+{\lambda}_{1,h } u = -1,\qquad & x\in { \omega}_h,\\ u(x)=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] then , for @xmath300 small enough , @xmath72 satisfies @xmath301_x)+({\lambda}_{1,h}+{\varepsilon } ) u = -1+{\varepsilon}u\le 0\ ] ] which gives a contradiction to the definition .
hence @xmath302 .
+ define now @xmath303 that solves @xmath304_x)+{\lambda}_n w_{n } = -\frac{1}{|u_n|_\infty},\qquad & x\in { \omega}_h,\\ w_{n}(x)=0 & x\in \partial{\omega}_h .
\end{array } \right.\ ] ] then , up to a subsequence , @xmath305 converges to a bounded function @xmath306 which has norm 1 and which satisfies , so that @xmath261 . + it is immediate that is still valid for @xmath25 .
[ convergence ] there is a huge literature about the approximation of viscosity solutions of first and second order pdes . in this framework a well established technique to prove the convergence of a numerical scheme is the barles - souganidismethod @xcite : besides some natural properties of the scheme ( stability , consistency , monotonicity ) , a key ingredient for this technique is a _
strong comparison result _ for the continuous problem which allows to show that a subsolution is always lower than or equal to a supersolution .
the comparison principle implies in particular that there is at most one viscosity solution of the problem .
but it is immediate that can not satisfy a comparison principle since @xmath307 and the principal eigenfunction @xmath8 are two distinct solutions of the problem , hence the convergence proof can not rely on the barles - souganidismethod and it needs a different argument .
we now discuss the convergence of the discrete principal eigenvalue @xmath29 to the continuous one defined by .
we recall the definition of weak limits in viscosity sense ( see @xcite ) @xmath308 [ main ] assume - , - and that @xmath25 is consistent with @xmath28 .
let @xmath260 be the sequence of the discrete eigenvalues and of the corresponding eigenfunctions , solutions of .
then @xmath309 and @xmath310 uniformly in @xmath311 as @xmath312 , where @xmath34 and @xmath8 are respectively the principal eigenvalue and a corresponding eigenfunction associated to @xmath28 . + by the positive homogeneity of the scheme , it is not restrictive to assume that @xmath313 , hence the sequence @xmath306 is bounded .
we first prove that @xmath314 assume by contradiction that @xmath315 for some some @xmath316 .
consider a subsequence , still denoted by @xmath29 , such that @xmath317 .
set @xmath318 . by standard stability results in viscosity solution theory ,
see @xcite , @xmath319 satisfies in viscosity sense @xmath320+\tau \overline w\ge 0\qquad \text{in $ { \omega}_h$,}\ ] ] and @xmath321 let @xmath322 be such that for @xmath32 sufficiently small , @xmath323 . hence @xmath324=-{\lambda}_{1,h}w_{1,h}\ge -\tau-\eta , \qquad x\in { \omega}_h.\end{aligned}\ ] ] set @xmath325 , @xmath326 and let @xmath169 be the corresponding solution of , while @xmath72 is the solution of @xmath327 then by proposition [ prop_comp ] and the consistency of the scheme for @xmath32 sufficiently small @xmath328 and therefore @xmath329 by , and we get a contradiction to the maximum principle for the operator @xmath28 ( see @xcite , @xcite ) and therefore .
+ we now prove that @xmath330 assume by contradiction that there exists @xmath331 such that @xmath332 we consider a subsequence , still denoted by @xmath29 , such that @xmath333 and we set @xmath334 . by standard stability results
@xmath335 satisfies @xmath336 and @xmath337+({\lambda}_1+\eta ) \underline w\le 0\qquad & \mbox { in } { \omega},\\ \underline w= 0&\mbox { on } \
\partial { \omega } , \end{array } \right.\ ] ] in viscosity sense .
let @xmath338 be a sequence such that @xmath339 and @xmath340 for all @xmath24 . by , @xmath341 .
we claim that @xmath342 assume by contradiction that @xmath343 , hence there exists a sequence @xmath344 such that @xmath345 . by with @xmath346 and @xmath347
we get @xmath348 since @xmath349 we get a contradiction for @xmath32 sufficiently small and therefore . + we are in a position to apply the maximum principle for the continuous problem ( see @xcite ) , and so we obtain that @xmath350 .
but and the positivity of @xmath335 give a contradiction to the definition of @xmath34 . by
and we get @xmath351 . + by and a local boundary estimate for @xmath306 , see ( * ? ? ?
5.1 ) and ( * ? ? ?
* thm.3 . )
, we get the equi - continuity of the family @xmath352 and therefore the uniform convergence , up to a subsequence , of @xmath306 to @xmath8 with @xmath353 .
the simplicity of the eigenfunction associated to the principal eigenvalue @xmath34 gives the uniform convergence of all the sequence @xmath306 to @xmath8 .
in this section we discuss an algorithm for the computation of the principal eigenvalue based on the inf - sup formula .
in fact we show that this formula results in a finite dimensional nonlinear optimization problem .
we first present the scheme in one dimension .
note that since the eigenfunction corresponding to the principal eigenvalue vanishes on the boundary of @xmath98 and it is strictly positive inside , then the minimization in can be restricted to the internal points . by the formula and the homogeneity of @xmath354 , we have @xmath355(x_i)}{u(x_i)}={{{\cal f}}}\left(x_i , 1 , \frac{u(x_i+h)-u(x_i - h)}{2hu(x_i)},\frac{u(x_i+h)+u(x_i+h)}{h^2u(x_i)}-\frac{2}{h^2}\right).\ ] ] we identify the function @xmath356 with the values @xmath357 , @xmath358 , at the points of the grid ( with @xmath359 ) .
assume that @xmath360 is linear or more generally convex in @xmath361 .
then the functions @xmath362 , defined by @xmath363 for @xmath364 , is either linear or respectively convex in @xmath365 , @xmath366 .
moreover , since @xmath367 , @xmath368 is also convex in @xmath357 .
taking the maximum of the functions @xmath369 over the internal nodes of the grid gives a convex function @xmath370 defined by @xmath371 hence the computation of @xmath29 is equivalent to the minimization of the convex function @xmath372 of @xmath373 variables : this problem can be solved by means of standard algorithms in convex optimization .
note also that the minimum is unique and the map is sparse , in the sense that the value of @xmath372 at @xmath357 depends only on the values at @xmath366 and @xmath365 .
+ in general , if @xmath360 is not convex , the computation of the principal eigenvalue is equivalent to the solution of a min - max problem in @xmath374 .
+ to solve min - max problem we use the routine ` fminmax ` available in the optimization toolbox of matlab .
this routine is implemented on a laptop and therefore the number of variables is modest
. a better implementation of the minimization procedure which takes advantage of the sparse structure of the problem would allow to solve larger problems .
* example 1 . * to validate the algorithm we begin by studying the eigenvalue problem : @xmath375 in this case the eigenvalue and the corresponding eigenfunction are given by @xmath376 note that since the eigenfunctions are defined up to multiplicative constant , we normalize the value by taking @xmath377 ( the constraint for @xmath306 is included in the routine ` fminmax ` ) .
given a discretization step @xmath32 and the corresponding grid points @xmath378 , @xmath379 , the minimization problem is @xmath380\ ] ] ( with @xmath359 ) . in table 4.1
, we compare the exact solution with the approximate one obtained by the scheme .
we report the approximation error for @xmath34 and @xmath8 ( in @xmath381-norm and @xmath382-norm ) and the order of convergence for @xmath34 .
we can observe an order of convergence close to @xmath383 for @xmath34 and therefore equivalent to one obtained by discretization of the rayleigh quotient via finite elements ( see @xcite ) .
.space step ( first column ) , eigenvalue error ( second column ) , convergence order ( third column ) , eigenfunction error in @xmath381 ( fourth column ) , eigenfunction error in @xmath382 ( last column ) [ cols="^,^,^,^,^",options="header " , ] | we present a finite difference method to compute the principal eigenvalue and the corresponding eigenfunction for a large class of second order elliptic operators including notably linear operators in nondivergence form and fully nonlinear operators .
+ the principal eigenvalue is computed by solving a finite - dimensional nonlinear min - max optimization problem .
we prove the convergence of the method and we discuss its implementation .
some examples where the exact solution is explicitly known show the effectiveness of the method .
* msc 2000 * : : : 35j60 , 35p30 , 65m06 . *
keywords * : : : principal eigenvalue , nonlinear elliptic operators , finite difference schemes , convergence . |
a large number of theoretical models have been constructed over the years since it was first observed that inclusively produced hyperons are polarized .
these models have met with varying degrees of success , but it is clear that more data and theoretical work are needed to clarify the picture .
a review of the current status is found in @xcite .
@xmath5 polarization is well studied with a proton beam , both in inclusive @xcite and exclusive @xcite reactions . for @xmath3 beam ,
only the wa89 experiment at cern reports polarization measurements for @xmath0 and @xmath1 as function of @xmath7 for one average value of @xmath6 @xcite , and for @xmath0 as function of both @xmath7 and @xmath6 @xcite . in the latter case
the polarization is negative for small @xmath6 , incrementing to positive values with @xmath6 , but with a non - monotonic behavior as a function of @xmath7 for different values of @xmath6 . in this study
we exploit the capabilities of the selex apparatus to measure the polarization of @xmath0 and @xmath1 inclusively produced by @xmath3 and proton beams .
the goal is to measure the polarizations as functions of @xmath6 and @xmath7 with higher beam momenta as the earlier wa89 measurement , using a different analysis method presenting different systematics , and extending the @xmath1 polarization measurements to several values in @xmath6 ; in addition , we add as a cross check the measurements with proton beam within the same experiment .
the selex ( e781 ) experiment at fermilab is a fixed target experiment designed primarily for high statistics studies of charmed baryons produced by a charged hyperon beam incident on a segmented copper / carbon target .
however , the versatility of the apparatus allowed for the study of other reactions @xcite in the same target . in this work
we study the polarization of inclusively produced @xmath0 and @xmath1 .
more specifically we determine the polarization and its @xmath6 and @xmath7 dependence .
we follow the basel convention @xcite for the sign of the polarization .
the charged hyperon beam , which we use as a primary beam for @xmath5 production , was obtained in the fermilab proton center beamline by steering @xmath11 protons from the tevatron onto a @xmath12 beryllium target located at the entrance of a @xmath13 , @xmath14 hyperon magnet @xcite .
a curved channel in this magnet selected a beam of negative ( positive ) particles with a mean momentum of @xmath2 ( @xmath15 ) and a spread of @xmath16 hwhm .
the negative beam consisted of approximately equal parts @xmath3 and @xmath17 with a small admixture of @xmath18 and @xmath19 , while the positive beam contained about @xmath20 protons , in a total of @xmath21 beam particles per second .
the targeting angle was set to zero degrees , thereby insuring an unpolarized incident beam . the portion of the experimental setup relevant to this analysis , which is shown in fig .
[ figapp ] , consisted of a beam spectrometer , a vertexing region and two downstream spectrometers ( m1 and m2 ) .
the beam spectrometer included the hyperon magnet , beam transition radiation detectors ( btrd ) and beam silicon strip detectors ( bssd ) .
the vertex region consisted of five segmented targets ( 2 cu , 3 c each separated by @xmath22 - altogether @xmath23 of an interaction length ) and 20 planes of vertex silicon strip detectors ( vssd ) , which resolved the interaction vertex and secondary vertices .
for vssd tracks , the transverse position resolution was @xmath24 at @xmath25 .
the m1 spectrometer was a wide angle spectrometer , designed to analyze particles with momenta between @xmath26 and @xmath27 .
this spectrometer contained the m1 magnet ( @xmath28 ) , large angle silicon strip detectors ( lasd ) , and proportional multi - wire chambers ( pwc ) . the second spectrometer ( m2 ) , downstream of m1 , analyzed particles with momenta @xmath29 .
its components were the m2 magnet ( @xmath30 ) , lasd , pwc , and a ring - imaging cherenkov ( rich ) detector @xcite used for particle identification , which provided @xmath31 separation up to @xmath32 .
a hardware trigger and software filter were used in selex to select the secondary interactions .
the hardware trigger consisted of the beam , veto and hodoscope scintillation counters .
the first level synchronized the trigger to beam particles .
the second level selected events where the btrd determined the beam particle was most likely a heavy particles ( @xmath3 or proton ) , with an additional refined definition in the analysis stage .
the beam interacted in one of the five targets , and more than 3 charged tracks were observed downstream of the target region of which at least 2 were positive tracks with momentum greater than @xmath27 .
triggered events accounted for @xmath33 of interactions .
a software filter looked for evidence of a secondary vertex and a reconstructed primary vertex using high momentum ( @xmath29 ) tracks .
the software filter further reduced the data set by a factor of 8 .
a sample of unfiltered data was always recorded for filter performance and systematic error analysis .
the @xmath34 ( @xmath35 ) candidates were selected by requiring oppositely charge track pairs to form a vertex at least @xmath36 downstream of the primary vertex , where @xmath37 is the combined error of the @xmath38-coordinates of the primary and secondary vertex , and upstream of the first vssd plane .
the positive ( negative ) track was required to be identified by the rich as a proton candidate . in fig .
[ signals ] we show the invariant mass distributions of these @xmath39 and @xmath40 candidates , and in table [ tab : yields ] we summarize the available statistics .
invariant mass distributions for @xmath39 ( left ) and @xmath40 ( right ) for @xmath3 ( top ) and proton ( bottom ) beams .
the signal and sideband regions are indicated.,scaledwidth=45.0% ] .[tab : yields ] number of @xmath0 , @xmath1 and @xmath41 for the different beams . [ cols="^,>,>,>,>",options="header " , ]
for the inclusive production of @xmath0 by a @xmath3 beam , our results confirm the wa89 @xcite measurements ( obtained via the bias canceling method ) of a generally positive polarization , increasing with @xmath6 , as shown in fig .
[ fig : lamsigxf ] and [ fig : lamsigpt ] . at small @xmath7 and
@xmath42 the polarization is almost @xmath10 , and in general the dependence on @xmath7 is non - monotonic for different bins of @xmath6 , increasing and decreasing after reaching some maximum value . comparing to the earlier , lower - statistics wa89
@xcite result with the dependence on @xmath7 for @xmath43 ( fig .
[ fig : others ] , upper left and table [ tab : otherspt ] ) , we observe small polarization values confirming the previous measurement .
the authors of @xcite attribute the non - monotonic behavior of the polarization as function of @xmath7 and @xmath6 to different production mechanism in different kinematic regimes .
we also examined the target dependence , but no difference was found . for the polarization for @xmath0 produced by a proton beam as a function of @xmath7 ,
our results coincide with the higher statistics , but lower beam momentum , results from e008 @xcite . in fig .
[ fig : others ] ( upper right ) and table [ tab : othersxf ] we also present polarization results as a function of @xmath6 and compare them to our @xmath3 data .
the @xmath1 polarization is compatible with @xmath10 , for @xmath3 and proton beams for all @xmath7 values measured , as obtained in @xcite .
in addition we present also the polarization as function of @xmath6 ( fig .
[ fig : others ] ( lower right ) and table [ tab : othersxf ] ) . in conclusion , we measured the polarization of inclusively produced @xmath0 with a @xmath3 beam , and confirming the wa89 results with a different analysis method and higher beam momentum .
we do not observe any polarization of the @xmath1 at @xmath44 and @xmath45 .
the @xmath0 polarization with a proton beam is compatible with earlier results .
the authors are indebted to the staff of fermi national accelerator laboratory and for invaluable technical support from the staffs of collaborating institutions .
this project was supported in part by bundesministerium fr bildung , wissenschaft , forschung und technologie , consejo nacional de ciencia y tecnologa ( conacyt ) , the secretara de educacin pblica ( mexico ) ( grant number 2003 - 24 - 001 - 026 ) , fondo de apoyo a la investigacin ( uaslp ) , conselho nacional de desenvolvimento cientfico e tecnolgico , fundao de amparo pesquisa do estado de so paulo ( fapesp ) , the israel science foundation founded by the israel academy of sciences and humanities , istituto nazionale di fisica nucleare ( infn ) , the international science foundation ( isf ) , the national science foundation ( phy # 9602178 ) , nato ( grant cr6.941058 - 1360/94 ) , the russian academy of science , the russian ministry of science and technology , the russian foundation for basic research ( rfbr grant 05 - 02 - 17869 ) , the turkish scientific and technological research board ( tbitak ) , and the u.s .
department of energy ( doe grant de - fg02 - 91er40664 and doe contract number de - ac02 - 76cho3000 ) .
+ j. lach , nucl .
suppl . * 50 * , 216 ( 1996 ) .
j. felix , mod .
a * 12 * , 363 ( 1997 ) .
j. felix , mod .
a * 14 * , 827 ( 1999 ) .
j. felix , mod .
a * 16 * , 1741 ( 2001 ) .
k. j. heller _ et al .
_ , phys . rev
. lett . * 41 * , 607 ( 1978 ) [ erratum - ibid . *
45 * , 1043 ( 1980 ) ] .
j. felix _ et al .
_ , phys .
lett . * 76 * , 22 ( 1996 ) .
j. felix _ et al .
_ , phys .
* 82 * , 5213 ( 1999 ) .
j. felix _ et al .
_ [ e690 collaboration ] , phys . rev .
* 88 * , 061801 ( 2002 ) .
m. i. adamovich _ et al . _
[ wa89 collaboration ] , z. phys .
a * 350 * , 379 ( 1995 ) [ arxiv : hep - ex/9409001 ] .
m. i. adamovich _ et al . _
[ wa89 collaboration ] , eur .
j. c * 32 * , 221 ( 2004 ) .
p. pogodin _ et al .
_ [ selex collaboration ] , phys . rev .
d * 70 * , 112005 ( 2004 ) .
hughes _ et al . _ , ``
last page '' , proceedings of the international symposium on polarization phenomena of nucleons , basel , switzerland , july 1960 .
j. lach , `` e781 hyperon beam and targeting system , '' fermilab
tm2129 ( 2000 ) .
j. engelfried _ et al .
_ , nucl .
instrum . meth .
a * 431 * , 53 ( 1999 ) [ arxiv : hep - ex/9811001 ] .
w. m. yao _ et al .
_ [ particle data group ] , j. phys .
g * 33 * , 1 ( 2006 ) .
j. l. sanchez - lopez , master thesis , instituto de fsica , universidad autnoma de san luis potos , 2006 .
fermilab - masters-2006 - 06
. k. d. nelson , ph.d .
thesis , university of iowa , 1999 .
fermilab - thesis-1999 - 55 .
j. engelfried , e. a. blanco , j. l. sanchez , selex internal note h-866 ( 2006 ) . | we have measured the polarization of @xmath0 and @xmath1 inclusively produced by @xmath2 @xmath3 and @xmath4 proton beams in the experiment selex during the 1996/7 fixed target run at fermilab .
the polarization was measured as a function of the @xmath5 longitudinal momentum fraction @xmath6 and transverse momentum @xmath7 . for
the @xmath0 produced by @xmath3 the polarization is increasing with @xmath6 , from slightly negative at @xmath8 to about @xmath9 at large @xmath6 ; it shows a non - monotonic behavior as a function of @xmath7 . for the proton beam ,
the @xmath0 polarization is negative and decreasing as a function of @xmath6 and @xmath7 .
the @xmath1 polarization is compatible with @xmath10 for both beam particles over the full kinematic range .
the target dependence was examined but no statistically significant difference was found . |
graphene is a two - dimensional material consisting of a monolayer of carbon atom forming hexagonal lattices @xcite .
much has been learned since its first isolation from graphite and deposition onto a dielectric substrate @xcite . due to its linear and massless band structure near the dirac points
, graphene has been shown to posses astonishing linear electronic and optical properties .
electron mobility in graphene reaches @xmath0 @xmath1/vs , much higher than normal materials such as silicon @xcite .
graphene is also the only known material exhibiting anomalous quantum hall effect under room temperature @xcite .
the optical absorption per layer of graphene is related to the fine - structure constant by @xmath2 over a broad range of terahertz and optical wavelengths @xcite .
potential applications include ultrabroadband fast detectors , and the replacement of transparent conductive oxides with graphene . also of interest are potential coherent electronics applications that would use quantum interference of electrons to provide new device functionality . besides linear electronic and optical properties , the third - order nonlinear response of materials is useful for certain applications such as wavelength conversion @xcite and quantum - entanglement generation @xcite .
the most important physical parameter for nonlinear - optical experiments is the nonlinear coefficient , typically the nonlinear susceptibility , of the material used .
however , in practice , nonlinear - optical experiments are limited by other factors , in particular , the phase - matching conditions arising from wavelength - dependent refractive indices .
the accumulation of phase - mismatch as the interacting waves propagate through the nonlinear medium sets a limit on the bandwidths of the nonlinear applications . because the graphene thickness is much less than an optical wavelength , non - critical
phase - matching occurs simply , though at the cost of some optical absorption .
nonlinear properties of graphene have been investigated theoretically in @xcite .
theoretical results predict high nonlinearity of graphene at the terahertz - frequency @xcite and optical - frequency @xcite ranges .
more recently , the nonlinearity of graphene in the visible - optical range was experimentally measured by four - wave mixing @xcite . when taking into account the nonlinear response per unit thickness , the theory adopted in ref .
@xcite shows that graphene s effective third - order nonlinear susceptibility @xmath3 is of @xmath4 orders higher than glass .
@xcite estimated the @xmath5 of graphene by evaluating the nonlinear surface - current density induced by an external electromagnetic field and estimating the dynamic nonlinear response of electrons coupled with an external field is characterized by the electron - photon interaction hamiltonian . instead of the dipole interaction hamiltonian utilized by @xcite ,
a more general form of the interaction hamiltonian can be obtained by the minimal substitution @xmath6 , giving a more accurate description of the dynamics .
as we will see in this paper , the dipole interaction hamiltonian is not adequate to describe the quantum dynamics of electrons in graphene .
additionally , the quantum dynamics of electrons in graphene is also affected by ultrafast many - body interactions such as electron - electron and electron - phonon scattering @xcite , which in turn may depend on carrier densities , conduction band energies , and temperature in non - trivial ways , resulting in ultrafast quantum dephasing .
other quantum dephasing mechanisms may exist .
a complete physical model needs to include these effects , which have not yet been included in previous theoretical calculations of the nonlinearity of graphene . in this paper , we present a quantum - dynamical model for investigating the quantum dynamics of electrons in graphene .
the electron - photon interaction hamiltonian is obtained by the minimal substitution on the free - electron hamiltonian to relate the quantum dynamics of electrons with the vector potential of the electromagnetic field , as in @xcite .
this paper differs in that it includes electron - electron and electron - phonon scattering by introducing two phenomenological decay rates .
this paper also focuses on nonlinear mixing rather than on the linear decay - free , dephasing - free model . in a quasi - continuous - wave pump experiment ,
in which the electron - photon coupling time ( order of ps ) is much longer than the carrier - relaxation time ( order of fs ) , these phenomenological decay rates allow us to avoid dealing with complicated microscopic quantum - mechanical calculations and to obtain analytical solutions for quantum dynamics of electrons in graphene by exploiting quantum - perturbation method .
the proposed model is appropriate for excitation power well below the saturation threshold of graphene ( 4 gw/@xmath1 ) @xcite , in which case the population inversion for each quantum state in the brillioun zone does not significantly change .
once the quantum dynamics of electrons in graphene is obtained , both linear and nonlinear optical conductivity produced by different optical processes will be derived .
this paper is organized as follows . in sec .
2 , we discuss the quantum dynamics of electrons in graphene .
our discussion in this section covers the electron - photon interaction and ultrafast quantum dephasing caused by carrier - relaxation processes in graphene . in sec .
3 , we derive the surface - current density in graphene .
we calculate both the linear and nonlinear surface - current densities that are caused by the driving optical frequency fields .
we also discuss the linear complex optical conductivity and connect it to ultrafast quantum dephasing in graphene . in sec . 4
, we apply the nonlinear surface - current density obtained in sec . 3 to evaluate four - wave mixing in graphene , and fit the theory to the recently reported experimental data @xcite . in sec . 5
, we discuss the validity of the proposed quautum - dynamical model under certain physical circumstances .
conclusions are reviewed in sec .
in this section , the quantum dynamics of electrons in graphene will be reviewed , starting from first principles .
we begin with the free - electron hamiltonian , obtained for the 2d tight binding model of nearest neighbour interaction approximation .
the electron - photon interaction hamiltonian is derived later by minimal substitution .
we then discuss the ultrafast carrier relaxation , which plays an important role in the quantum dynamics of electrons in graphene .
the resulted ultrafast quantum dephasing in graphene is included by addition of phenomenological decay rates into the model . in the 2d tight - binding model , or nearest neighbour interaction approximation ,
the motion of electrons is limited by assuming that they can only hop to their nearest neighbours .
the hamiltonian of electrons in graphene under this assumption is written in the second - quantized language @xcite : @xmath7 where @xmath8 is the hopping energy and the product of the creation and annihilation operators @xmath9 illustrates a process in which an electron with spin @xmath10 is annihilated on site @xmath11 of sublattice a and a new electron of the same spin is created on its neighbour site @xmath12 of sublattice b. the fourier transform of the hamiltonian in eq .
[ eqhamiltonianspace ] allows the derivation of the free - electron hamiltonian in the momentum space , which is @xmath13 where @xmath14 is the first - quantized free - electron hamiltonian , characterizing the quantum dynamics of a single electron with momentum @xmath15 .
the summation in eq .
[ eqhamiltonian_k ] is performed over the entire brillioun zone . in the vicinities of two dirac points
@xmath16 in the brillioun zone , we have the following linear dispersion relation for @xmath17 : @xmath18 where @xmath19 m / s is the fermi velocity of electrons in graphene .
energy eigenstates and corresponding eigen - energies can be found by solving the 2d dirac equation @xcite : @xmath20 given the electron momentum @xmath21 , one finds two energy eigenstates with opposite eigenvalues @xmath22 , corresponding to the conduction - band and valence - band respectively : @xmath23 , related to the electric field @xmath24 and the magnetic field @xmath25 by @xmath26 suppose at time @xmath27 , we turn on a homogeneous a.c . electromagnetic field
having a vector potential with two polarization components : @xmath28 where @xmath29 and @xmath30 are two orthogonal unit vectors of the cartesian coordinate .
each polarization component @xmath31 , @xmath32 , is the summation of different frequency modes on a polarization : @xmath33 both positive and negative frequencies are allowed .
the complex conjugate relation @xmath34 holds to guarantee a real field . by using eq .
[ eqvectorpotentialtofields ] , the electric field on polarization @xmath35 is related to the vector potential by @xmath36 the electric - field amplitude and the vector - potential amplitude have the following relation : @xmath37 the electron - photon interaction hamiltonian can be obtained by the minimal substitution @xmath38 on the free - electron hamiltonian @xmath39 @xcite .
the taylor expansion of the electron - photon interaction hamiltonian around @xmath40 reads @xmath41 due to the linear band structure of graphene , higher - order terms in the taylor expansion vanish in the vicinities of two dirac points .
let @xmath42 , so that we have @xmath43 where @xmath44 . in the representation of energy eigenstates
@xmath45 and @xmath46 , @xmath47 is organized into the following matrix form : @xmath48 around the two dirac points @xmath16 , by use of eq .
[ eqhplinear ] , one obtains @xmath49 in general , the quantum dynamics of an electron can be obtained by solving the time - dependent schrdinger equation : @xmath50,\ ] ] with @xmath51 the density operator of the electron .
[ eqschrodingermotion ] describes the quantum dynamics of an electron with momentum @xmath21 that is subject to an external electromagnetic field characterized by its vector potential @xmath52 .
a valence - band electron can be excited , and become a conduction - band electron by absorbing a photon .
the transition happens with a particular probability that depends on the field intensity and photon energy .
the excited conduction - band electron could be stimulated into the valence band and emit a photon .
this whole process is the rabi oscillation .
however , in a complete physical picture , electrons in the conduction band undergo ultrafast carrier relaxation processes caused by the faster electron - electron scattering followed by the slower electron - phonon scattering @xcite .
these scattering processes destroy the quantum coherence of electrons and result in quantum dephasing . in a strong excitation regime ,
a strong pump pulse depletes the population in the valence band and excites electrons into the conduction band .
thereafter , the hot electrons in the conduction band undergo an ultrafast electron - electron scattering process ( as do the holes in the valence band ) that results in quasi - thermalized statistics for electrons in the conduction band and holes in the valence band . the quasi - thermalized distributions
establish new state - occupation probabilities within the two bands . as a result , a strong pump not only changes the state - occupation probability of the state having the corresponding energy , but also affects the state - occupation probabilities over the whole band . a complete treatment that would include the electron - photon , electron - electron , and electron - phonon interactions requires sophisticated quantum - mechanical calculations @xcite .
there may be additional quantum dephasing processes .
however , if we limit ourselves to low - excitation quasi - continuous - wave regime , in which the state - occupation probabilities over the whole bands do not change significantly by the pump nor in time , the situation is much simpler . to model the ultrafast quantum dephasing in a low excitation quasi - continuous - wave regime
, we simply introduce two phenomenological decay rates @xmath53 and @xmath54 , describing the population and quantum - coherence damping that phenomenologically represent the overall effect of the ultrafast scattering processes . with the two phenomenological decay rates ,
one arrives at the bloch equations for graphene under low excitation power : @xmath55 \notag\\ \dot{\rho}_{vc}&=&\left(-i\omega_{vc}-\gamma_2\right)\rho_{vc}-\frac{ei}{\hbar}\sum_{u , j } \widetilde{a}_{u , j}(t)[2v_{vv}^u\rho_{vc}+v_{vc}^u\rho ] \notag\\ \dot{\rho}_{cv}&=&\left(-i\omega_{cv}-\gamma_2\right)\rho_{cv}-\frac{ei}{\hbar}\sum_{u , j } \widetilde{a}_{u , j}(t)[-2v_{vv}^u\rho_{cv}+v_{vc}^u\rho],\notag\\\end{aligned}\ ] ] where @xmath56 is the population inversion , @xmath57 , and @xmath58 is the population inversion in thermal equilibrium .
compared to the bloch equations for atom vapours and normal semiconductors @xcite , both diagonal and off - diagonal driving terms exist for the off - diagonal elements of the density matrix .
general solutions of eq .
[ eqschrodingertwolevel ] are difficult to find .
instead , we seek perturbative solutions , which are are good approximations when the pump power is well below the saturation threshold , which is measured to be @xmath59 gw/@xmath60 @xcite .
the perturbative method gives the solution in the following cascaded form in the frequency domain , with each order denoted by a positive index @xmath61 : @xmath62 \notag\\ \rho^{(k+1)}_{vc}(\omega)&=&- \frac{e}{\hbar}\sum_{u , j } \frac { a_{u , j}}{(\omega+\omega_{vc}-i\gamma_2)}\left[2v_{vv}^u\rho^{(k)}_{vc}(\omega+\omega_{u , j})+v_{vc}^u\varrho^{(k)}(\omega+\omega_{u , j})\right ] \notag\\ \rho^{(k+1)}_{cv}(\omega)&=&-\frac{e}{\hbar}\sum_{u , j}\frac { a_{u , j}}{(\omega+\omega_{cv}-i\gamma_2)}\left[-2v_{vv}^u\rho^{(k)}_{cv}(\omega+\omega_{u , j})+v_{vc}^u\varrho^{(k)}(\omega+\omega_{u , j})\right],\notag\\\end{aligned}\ ] ] where @xmath63 with @xmath64 in order for the laplace transform to converge .
in this section , the quantum dynamics of electrons will be derived perturbatively .
we first derive the first - order solution , which determines the linear optical response of graphene .
we then compare our theoretical result with the linear complex conductivity measured by @xcite . the cascaded relation in eq .
[ eqcascadedrhodecayfreq ] allows the derivation of higher - order density matrix terms , which determine the nonlinear quantum dynamics of electrons in graphene .
we continue to calculate up to the third - order terms , which allows the prediction of nonlinear optical responses such as four - wave mixing .
we finally fit the theory to recently reported four - wave mixing experimental data @xcite .
the zeroth - order solution to the density matrix is simply the thermal - equilibrium values of the population inversion and quantum coherence . at temperature @xmath65
, we have @xmath66 , @xmath67 . inserting zeroth - order solution into eq .
[ eqcascadedrhodecayfreq ] leads to the first - order solution : @xmath68 where @xmath69 is the laplace transform of the time - dependent vector potential of one frequency mode indexed by @xmath70 on polarization @xmath71 . the expected velocity on @xmath35 polarization of a field - coupled electron
is given by @xmath72,\ ] ] where the velocity operator @xmath73 is defined as @xmath74 to simplify our discussion , we only consider an external field polarized in the @xmath75 direction with vector potential @xmath76 .
the linear electron velocity is contributed by both the diagonal elements from @xmath77 and the off - diagonal elements from @xmath78 @xcite : @xmath79.\ ] ] the linear surface - current density @xmath80 can be obtained by an integral over the whole brillouin zone . since the integrand @xmath81 is only appreciable in the vicinity of the input optical frequency @xmath82 , instead of an integral limited in the brillioun zone , we can extend the integral to infinity without significantly affecting the result .
the integral is written as @xmath83 by using the frequency - domain relation @xmath84 , the linear optical conductivity is derived as @xmath85 where @xmath86 is the theoretical prediction of the universal optical conductivity of graphene @xcite . a factor @xmath59
is added due to valley and spin degeneracies . if we compare eq .
[ eqsigmalinear ] with the experimental measurements of the complex optical conductivity of graphene at @xmath87 nm in @xcite , we find that @xmath54 ranges from @xmath88 s@xmath89 to @xmath90 s@xmath89 by normalizing the real parts to @xmath86 .
in @xcite , the authors ascribe the imaginary part of the conductivity of graphene to the virtual transition of electrons at m and @xmath91 points in the brillioun zone
. however , our quantum mechanical calculations based on the nearest - neighbour - interaction approximation find that we have @xmath92 at the @xmath91 point , leading to a zero interaction hamiltonian , i.e. , @xmath93 . at the m point ,
the interaction hamiltonian gives pure real diagonal elements , and pure imaginary off - diagonal elements in eq .
[ eqvtmatrix ] .
thus , the conductivity at the m point is still real as long as we set @xmath94 .
thus it seems that the relatively large imaginary part of the conductivity arises from the ultrafast quantum dephasing in graphene .
the fact that this time constant is approximately 1 fs indicates both that optical measurements can be used to characterize quantum dephasing in particular samples and that dipole quantum coherence would seem to be a poor candidate for coherent electron devices in graphene .
to evaluate the nonlinear interactions in graphene , we need to solve for higher - order density - matrix terms @xmath95 and @xmath96 . having obtained @xmath95 and @xmath96 , the expected nonlinear velocity on polarization @xmath97 for an electron with momentum @xmath40 can be calculated as @xmath98,\ ] ] according to eq .
[ eqvelocityoperator ] , both @xmath95 and @xmath96 contribute to @xmath99 . to derive the nonlinear surface - current density , we need to integrate over the entire brillouin zone . @xmath95 and
@xmath96 both are appreciable only in the vicinities of their resonant frequencies .
the contribution from @xmath95 can be neglected since the second derivative of @xmath17 gives zero due to the linear band structure of graphene near the dirac points .
thus , in the nonlinear surface - current - density calculation , we only keep the @xmath100 term and drop the @xmath101 term in eq .
[ eqvelocityoperator ] .
the resulting nonlinear velocity is written as @xmath102=\sum_{u , l}\sum_{v , m}\sum_{w , n } \langle v_{u , l;v , m;w , n}^q\rangle,\ ] ] where the expected velocity of one frequency component on polarization @xmath97 is defined as @xmath103 . since the main contribution of @xmath104 comes from the vicinities of each resonant frequency , we can extend the integration in the brillouin zone to infinity and without significantly affecting the result . by assuming the dirac cone goes to infinity , we obtain analytical solutions to the third - order surface - current density of at one frequency component having polarization @xmath35 to be @xmath105 \times \notag \\ & & \widetilde{a}_{w , n}(\omega+\omega_{u , l}+\omega_{v , m}),\notag\\\end{aligned}\ ] ] where @xmath106 and @xmath107 are @xmath108 in the time domain
, the steady - state solution to the nonlinear surface - current density reads @xmath109e^{-i(\omega_{u , l}+\omega_{v , m}+\omega_{w , n})t},\notag\\\end{aligned}\ ] ] where a coefficient @xmath59 is added to include valley and spin degeneracies .
we have substituted the vector - potential amplitude with the electric - field amplitude by use of eq .
[ eqearelation ] . having obtained the surface - current density of one frequency component @xmath110 on polarization @xmath35 , the total surface - current density on this polarization is the summation over all frequency components , @xmath111 eq . [ eqjtotal ] is a general expression for the total third - order nonlinear surface - current density , composed of different frequency modes , each of which is produced by a particular nonlinear - optical process .
for example , let @xmath112 be the frequencies of the pump modes and @xmath113 be the frequency of the signal mode , the frequency @xmath114 results from third - harmonic generation of the pump at frequency @xmath115 while @xmath116 is produced by four - wave mixing .
we mark one frequency mode with frequency @xmath117 on polarization @xmath97 as @xmath118 .
@xmath119 is the summation over all frequency components @xmath110 satisfying @xmath120 : @xmath121
of particular interest for this paper is the study of four - wave mixing , a nonlinear - optical process involving four modes .
two of these serve as the pump modes , one as the signal mode , and one as the idler mode .
the optical frequencies @xmath122 and @xmath123 correspond to the two pump modes , @xmath113 to the signal mode , and @xmath124 to the idler mode .
energy conservation gives @xmath125 .
microscopically , four - wave mixing is a process in which two pump photons are annihilated while one signal and one idler photon are created . in the special case where the two pump modes share the same optical frequency , i.e. , @xmath126 , the four - wave mixing is pump degenerate . since
four - wave mixing is a third - order nonlinear process , its strength in the electric field is proportional to the nonlinear susceptibility @xmath3 .
if the frequency of the optical modes are far away from the resonant frequency of the nonlinear medium , the results is a non - resonant susceptibility @xmath127 . in the quantum - mechanical picture ,
@xmath127 results from the fact that the photon energies are far from the bandgap of the nonlinear medium .
a typical value for the non - resonant susceptibility in optical fibers is @xmath128 esu @xcite .
however , a much higher @xmath3 @xcite is present if the bandgap of the nonlinear medium is close to the interacting - photon energies .
the resonance - enhanced nonlinear susceptibility @xmath129 is typically of the order of @xmath130 esu .
graphene is a zero - bandgap semiconductor with linear band structure near the dirac points , making it `` more resonant '' than typical resonant media .
the zero bandgap of graphene results in the fact that any frequency within the range from dc to optical frequencies is resonant .
this is the physical origin of the high and uniform absorption of graphene . while absorption is a resonance - enhanced linear process , nonlinear processes can also be enhanced by resonance
. in particular , four - wave mixing in graphene can be enhanced by the five resonance - enhanced processes plotted in fig .
[ fig : fwmdiagram ] . five resonance - enhanced four - wave mixing processes in graphene .
( a ) two - photon absorption enhanced four - wave mixing .
( b ) one - photon absorption enhanced four - wave mixing .
( c ) idler resonance - enhanced four - wave mixing .
( d ) signal resonance - enhanced four - wave mixing .
( e ) detuning - enhanced four - wave mixing . ]
different resonance - enhanced four - wave mixing processes exist for electrons with different energies in graphene .
we next discuss these resonance - enhanced four - wave mixing processes .
\(a ) two - photon absorption enhanced four - wave mixing : two - photon absorption happens when the bandgap of a material is twice the photon energy of the incident light .
in graphene , this corresponds to @xmath131 .
\(b ) one - photon pump absorption enhanced four - wave mixing : unlike two - photon absorption , in which one electron is excited by two incoming photons simultaneously , an electron is excited by a single photon in a one - photon absorption process . in graphene
, the condition for one - photon absorption is @xmath132 .
\(c ) idler - enhanced four - wave mixing : in four - wave mixing , if the emitted light frequency , i.e. , @xmath124 , is close the the bandgap of the nonlinear medium , we also have resonance - enhanced effect . in graphene
, this happens at the bandgap @xmath133 .
+ ( d ) signal - enhanced four - wave mixing : if the signal - photon energy is close to the bandgap of the nonlinear medium , we also have enhanced effect for four - wave mixing . in graphene , signal - enhanced four - wave mixing happens at bandgap @xmath134 .
+ ( e ) detuning - enhanced four - wave mixing : detuning - enhanced four - wave mixing in graphene is similar to the four - wave mixing enhancement by coherent anti - stokes raman scattering @xcite , when the frequency detuning between the pump mode and the stokes mode is close to the molecule - vibration energy . in graphene
, this is the case when the bandgap is close to the frequency detuning between the pump and the signal , i.e. , @xmath135 .
+ in the co - polarized case , we let all frequency modes be @xmath136 polarized .
we define the conductivity of the co - polarized non - degenerate four - wave mixing to be @xmath137 the frequency components contributing to @xmath138 are listed in tab .
[ tabcnfwm ] .
.[tabcnfwm]frequency components for co - polarized non - degenerate four - wave mixing [ cols="^,^,^,^,^",options="header " , ] by comparing table [ tabxnfwm ] with table [ tabcnfwm ] , we find that @xmath139 .
thus , @xmath140 and @xmath138 differ only by a factor of @xmath141 as in typical nonlinear media .
we can adopt the same approach as in @xcite to compare the effective third - order susceptibility of graphene with the third - order susceptibility of silica .
the effective third - order susceptibility of graphene is given in the mks units by @xmath142 where @xmath143 is the effective thickness of a monolayer of graphene .
typical value for non - resonant third - order susceptibility of silica such as glass is @xmath144 m@xmath145/@xmath146 @xcite .
we plot the ratio between the two susceptibilities in fig .
[ figchi3ratio ] . in each figure
, three curves correspond to center pump wavelengths at @xmath147 nm , @xmath148 nm and @xmath149 nm respectively .
for each center pump wavelength , we choose three detuning frequencies at @xmath150thz , @xmath151thz and @xmath152thz between the two pump modes .
+ + although ref .
@xcite predicts that the nonlinear susceptibility of graphene is approximately 8 orders of magnitude greater than that of insulators , our more complete model shows that the ratio between the two susceptibilities can vary from 5 to 9 orders of magnitude , depending on the pump wavelengths , the signal detuning , and the decay rates @xmath53 and @xmath54 . a faster quantum dephasing results in a lower nonlinearity due to the fact that the coherence of electrons is quickly destroyed by the environment via scattering processes .
our theory also predicts that ss the pump - signal detuning becomes smaller , the strength of four - wave mixing increases due to resonance - enhanced effect .
another interesting phenomenon is that as the quantum dephasing times approach @xmath1531 fs , the typical symmetry in the nonlinear susceptibility breaks down .
the optical conductivity for degenerate four - wave mixing can be obtained by setting @xmath154 in the derivation of the non - degenerate four - wave mixing conductivity and dividing the result by @xmath155 due to degeneracy . in experiment ,
the observed idler photon - current intensity @xmath156 relates to the degenerate four - wave mixing conductivity @xmath157 by @xmath158 with @xmath157 as a function of @xmath159 , @xmath160 , @xmath53 and @xmath54 . in fig .
[ figcompareprl ] we compare the idler photon - current intensity predicted by eq .
[ eqiidler ] with that is predicted by eq . 4 in @xcite .
depending on the decay rates @xmath53 and @xmath54 , the ratio between the two current intensities varies .
the result shows that faster quantum - dephasing rates lead to a weaker four - wave mixing . for decay rates at @xmath161 s@xmath89 ,
the four - wave - mixing photon - current intensity is roughly equal to what was predicted in @xcite .
we next fit our theory to experimental data obtained by @xcite .
we use three values for @xmath162 found in sec .
[ seclinearconductivity ] by investigating the imaginary part of the linear complex optical conductivity measured by experiments .
thus , the only parameter to determine is @xmath53 .
fits to experimental four - wave - mixing data yields @xmath163 ranging from @xmath164 s@xmath165 to @xmath166 s@xmath89 .
the fits are shown in fig .
[ figfit ] . to summarize , the decoherence time is taken from linear refractive index measurement , without fitting .
because the amplitude of the nonlinear response is taken directly from the theory at the decoherence time , only one parameter is used to make the fit .
the two decay rates were both obtained from experimental data , which provides evidence for ultrafast quantum dephasing in graphene at the 1 fs timescale .
our fully quantum - mechanical model results in faster decay rates than what was reported in @xcite based on semi - classical calculations .
we note here that the decay rates we found are different than carrier - relaxation time @xcite in two ways .
first , carrier - relaxation time was obtained by differential transmission ( dt ) experiments where the graphene was excited by an ultrafast strong pump and then probed by a weak pulse . in the probing phase ,
only electron - electron and electron - phonon interactions exist .
however , in a continuous - wave or quasi - continuous - wave experiment , the relatively strong pump ( still weaker than the saturation threshold ) interact with the graphene through the whole process .
thus , we expect the decay rates to be faster since electron - photon , electron - electron and electron - phonon interactions all happen simultaneously .
second , dt experiments measure the population inversion at the given time .
however , our decay rates also include the damping of quantum coherence of electrons .
in addition , pure dephasing processes may exist in graphene .
finding a correct hamiltonian for a quantum system is of course the key step in derive its dynamics .
as we have discussed , we derive the hamiltonian of electrons in graphene by the minimal substitution @xmath167 .
the hamiltonian obtained by minimal substitution is general and it produces the correct electron - photon interactions .
however , sometimes the dipole interaction hamiltonian @xmath168 was used for semiconductors @xcite and atom vapours @xcite .
the dipole interaction hamiltonian can be derived from the original electron hamiltonian by the minimal substitution to get @xmath169 followed by replacement of @xmath21 by means of the commutation relation @xcite @xmath170 = \frac{i\hbar\textbf{p}}{m}.\ ] ] we note here that eq .
[ eqsubptor ] only holds for parabolic energy - dispersion relations .
however , the energy dispersion is linear near the dirac points in graphene .
thus , the dipole interaction hamiltonian does not apply to graphene in the area near the dirac points . near the m point and @xmath91 point in graphene
, we do have parabolic energy - dispersion relations . if we compare the graphene hamiltonian in eq .
[ eqgraphenehamiltonian ] with the typical hamiltonian in eq .
[ eqhamiltoniannormal ] for normal materials , we find that the hamiltonian of graphene commutes with the momentum operator , and therefore we have energy - momentum degeneracy in graphene . at the m point and the @xmath91 point , the transition matrix element @xmath171 .
thus , as previously discussed , electrons near the @xmath91 point do not interact with the field in the tight - binding model within the nearest neighbour interaction approximation .
electrons near the m point only contribute to the real part of the conductivity if the quantum - dephasing is not present .
the perturbative approach is a good approximation when the saturation effect can be neglected .
when saturation effect starts to become significant , the population inversion @xmath172 becomes so large that our perturbative solution can fail @xcite .
however , as we have shown , there exist five different resonance - enhanced four - wave mixing processes in graphene .
we find that the population inversion @xmath173 approaches to its maximal value around @xmath132 , induced by the one - photon absorption .
thus , the one - photon absorption enhanced four - wave mixing is the most likely to saturate .
if we inspect the purturbative solution @xmath174 , we find that population inversion induced by one - photon absorption is a second - order process . in comparison ,
the population inversions induced by two - photon absorption , pump - signal detuning , or idler simulation are third - order processes with much weaker magnitudes .
although the signal mode can also induce second - order population inversion , it can be neglected when the strong pump approximation is valid , since the signal intensity is much less than the pump intensity . under high excitation power levels that can completely change the state - occupation probabilities over the valence and the conduction band ,
a full - quantum - mechanical many - body theory is required to derive the quantum dynamics of graphene @xcite .
the quantum many - body theory can also be used to calculate the instantaneous response of electrons in graphene . for steady - state electrons under high excitation power ,
a quasi - thermalized carrier statistics is established within both the valence and the conduction band .
knowing this quasi - thermalized carrier statistics , one could plug this into @xmath175 in eq .
[ eqschrodingertwolevel ] and derive the steady - state quantum - dynamics of electrons under high excitation power . in our calculations
, we made a linear approximation of the band structure of graphene near the dirac points .
this approximation is valid up to @xmath59ev of bandgap given the hopping energy @xmath176 ev . for photon wavelengths below @xmath177 nm ,
the approximated linear band structure starts to be inaccurate and the optical conductivity diverges from the universal value @xmath86 . a full - band calculation @xcite would be needed for this situation .
we have used a quantum - mechanical model to investigate the quantum dynamics of electrons in graphene . the tight - biding model with nearest neighbour approximation and linear band structure near the dirac points
are assumed in order to obtain analytical results .
we have also included ultrafast decay and quantum - dephasing effects by adding two phenomenological decay constants into the model . matches to linear complex optical conductivity and fits to nonlinear four - wave mixing experiments establish quantum - dephasing rates of approximately one femtosecond .
this provides evidence for ultrafast dipole quantum - dephasing in graphene .
our approach shows that continuous wave or quasi - continuous - wave optical measurements can yield information about ultrafast scattering processes in graphene .
this work also has relevance for proposed coherent electronics applications .
we gratefully acknowledge support from nsf grant dmr-0820382 , the embassy of france puf .
we also acknowledge useful discussions with prof .
b. rosenstein and prof .
d. citrin . | we use a quantum - dynamical model to investigate the optical response of graphene under low excitation power .
ultrafast carrier relaxation processes , which play an important role for understanding the optical response of graphene , are included phenomenologically into the model .
we obtain analytical solutions for the linear and third - order nonlinear optical response of graphene , and four - wave mixing in particular .
this theory shows agreement with recently reported experimental data on linear complex optical conductivity and four - wave mixing , providing evidence for ultrafast quantum - dephasing times of approximately 1 fs
. 99 a.h .
castro neto , f. guinea , n.m.r .
peres , k.s .
novoselov , and a.k .
geim , `` the electronic properties of graphene , '' * 81 * , 109 - 162 ( 2009 ) .
k.s .
novoselov , a.k .
geim , s.v .
morozov , d. jiang , y. zhang , s.v .
dubonos , i.v .
grigorieva , and a.a .
firsov , `` electric field effect in atomically thin carbon films , '' science * 306 * , 666 - 669 ( 2004 ) .
x. du , i. skachko , a. barker , and e.y .
andrei , `` approaching ballistic transport in suspended graphene , '' nature nanotech .
* 3 * , 491 - 495 ( 2008 ) .
v.p .
gusynin and s.g .
sharapov , `` transport of dirac quasiparticles in graphene : hall and optical conductivies , '' * 73 * , 245411 ( 2006 ) .
r.r .
nair , p. blake , a.n .
grigorenko , k.s .
novoselov , t.j .
booth , t. stauber , n.m.r .
peres , and a.k .
geim , `` fine structure constant defines visual transparency of graphene , '' science * 320 * , 1308 ( 2008 ) .
t. stauber , n.m.r .
peres , and a.k .
geim , `` optical conductivity of graphene in the visible region of the spectrum , '' * 78 * , 085432 ( 2008 ) .
m. mecklenburg , j. woo , and b.c .
regan , `` tree - level electron - photon interactions in graphene , '' * 81 * , 245401 ( 2010 ) .
d.j . harter and r.w .
boyd , `` nearly degenerate four - wave mixing enhanced by the ac stark effect , '' * 16 * , 1126 - 1131 ( 1980 ) .
x. li , p.l .
voss , j.e .
sharping , and p. kumar , `` optical - fiber source of polarization - entangled photons in the 1550 nm telecom band , '' * 94 * , 053601 ( 2005 ) .
s.a .
mikhailov , `` non - linear electromagnetic response of graphene , '' euro .
phys .
lett . * 79 * , 27002 ( 2007 ) .
s.a . mikhailov and k. ziegler , `` nonlinear electromagnetic response of graphene : frequency multiplication and the self - consistent - field effects , '' j. phys . : condens .
matter * 20 * , 384204 ( 2008 ) .
b. rosenstein , m. lewkowicz , h.c .
kao , and y. korniyenko , `` ballistic transport in graphene beyond linear response , '' * 81 * , 041416(r ) ( 2010 ) .
b. dra , and r. moessner , `` nonlinear electric transport in graphene : quantum quench dynamics and the schwinger mechanism , '' * 81 * , 165431 ( 2010 ) .
k.l .
ishikawa , `` nonlinear optical response of graphene in time domain , '' * 82 * , 201402(r ) ( 2010 ) .
e. hendry , p.j .
hale , j. moger , a.k .
savchenko , and s.a .
mikhailov , `` coherent nonlinear optical response of graphene , '' * 105 * , 097401 ( 2010 ) .
d. sun , z - k .
wu , c. divin , x. li , c. berger , w.a . de heer , p.n .
first , and t.b norris , `` ultrafast relaxation of excited dirac fermions in epitaxial graphene using optical differential transmission spectroscopy , '' * 101 * , 157402 ( 2008 ) . h. haug and s.w .
koch , `` quantum theory of the optical and electronic properties of semiconductors , '' world scientific ( 2004 ) .
t. winzer , a. knorr , and e. malic , `` carrier multiplication in graphene , '' nano lett .
* 10 * , 4839 - 4843(2010 ) s. piscanec , m. lazzeri , f. mauri , a.c .
ferrari , and j. robertson , `` kohn anomalies and electron - phonon interactions in graphene , '' * 93 * , 185503 ( 2004 ) .
g. xing , h. guo , x. zhang , t.c .
sum , and c.h.a .
huan , `` the physics of ultrafast saturable absorption in graphene , '' * 18 * , 4564 - 4573 ( 2010 ) .
h.c .
kao , m. lewkowicz , and b. rosenstein , `` ballistic transport , chiral anomaly , and emergence of the neutral electron - hole plasma in graphene , '' * 82 * , 035406 ( 2010 ) .
r. shankar , `` principles of quantum mechanics , '' springer ( 1994 ) .
m. kira and s.w .
koch , `` many - body correlations and excitonic effects in semiconductor spectroscopy , '' prog .
quantum electron . * 30 * , 155 - 296 ( 2006 ) .
r.w .
boyd , `` nonlinear optics , '' academic press , new york ( 1992 ) .
m. bruna and s. borini,``optical constants of graphene layers in the visible range , '' appl .
phys .
lett . * 94 * , 031901 ( 2009 ) .
x. wang , y.p .
chen and d.d .
nolte , `` stronganomalous optical dispersion of graphene : complex refractive index measured by picometrology , '' opt .
express * 16 * , 2210522112 ( 2008 ) .
z.h .
ni , h.m .
wang , j. kasim , h.m .
fan , t. yu , y.h .
wu , y.p .
feng , and z.x .
shen , `` graphene thickness determination using reflection and contrast spectroscopy , '' nano lett.*7 * , 27582763 ( 2007 ) .
h.s .
skulason , p.e .
gaskell and t. szkopek , `` optical reflection and transmission properties of exfoliated graphite from a graphene monolayer to several hundred graphene layers , '' nanotech .
* 21 * , 295709 ( 2010 ) .
m. lewkowicz and b. rosenstein , `` dynamics of particle - hole pair creation in graphene , '' * 102 * , 106802 ( 2009 ) .
r.w .
boyd , m.g .
raymer , p. narum , and d.j .
harter , `` four - wave parametric interaction in a strongly driven two - level system , '' * 24 * , 411 - 423 ( 1981 ) .
j. nilsen and a. yariv , `` nondegenerate four - wave mixing in a doppler - broadened resonant medium , '' j. opt .
. am . *
71 * , 180 - 183 ( 1981 ) .
a. zumbusch , g.r .
holtom , and x.s .
xie , `` three - dimensional vibrational imaging by coherent anti - stokes raman scattering , '' * 82 * , 4142 - 4145 ( 1999 ) .
c.w .
freudiger , w. min , b.g .
saar , s. lu , g.r .
holtom , c. he , j.c .
tsai , j.x .
kang , and x.s .
xie , `` label - free biomedical imaging with high sensitivity by stimulated raman scattering microscopy , '' science * 322 * , 1857 - 1861 ( 2008 ) .
t.a . reichardt and r.p .
lucht , `` degenerate four - wave mixing spectroscopy with short - pulse lasers : theoretical analysis , '' * 13 * , 2807 - 2816 ( 1996 ) .
m. klintenberg , s. lebgue , c. ortiz , b. sanyal , j. fransson , and o. eriksson , `` evolving properties of two - dimensional materials : from graphene to graphite , '' j. phys . : condens .
matter * 21 * , 335502 ( 2009 ) . |
while the field of iron - based superconductors has focused primarily on selenides , tellurides , and arsenides,@xcite recent developments show that sulfides are a possible new avenue for high-@xmath9 superconductors .
the first iron - sulfide superconductor , bafe@xmath10s@xmath11 , has been reported to have a superconducting critical temperature ( @xmath12 ) = 14 k at 11 gpa.@xcite an even simpler sulfide , h@xmath13s , under high pressure ( 90 gpa ) , has been found to exhibit superconductivity as high as 203 k , which is the highest reported @xmath9 thus far.@xcite sulfides in general therefore merit closer inspection for exploring high temperature superconductivity , and iron sulfides in particular could point the way towards new superconducting compounds .
recently , lai _
_ et al . _
_ found that the simple binary compound , fes , in its tetragonal polymorph known as mackinawite is a superconductor with a @xmath14 k.@xcite similar to the superconducting @xmath15-form of iron selenide , mackinawite also adopts the anti - pbo structure where fes@xmath16 tetrahedra edge - share to form two - dimensional ( 2d ) layers ( figure [ fgr_xrd]b inset).@xcite unlike its heavier analogues , fese and fete , however , mackinawite is metastable and therefore can not be synthesized from their respective elements using solid state methods , unless it is alloyed with significant amounts of co , ni or cu.@xcite due to the thermodynamic limitations in its preparation , single crystal growth of mackinawite is a challenge .
growing single crystals of fes is imperative , however , towards understanding its true physical properties .
fe@xmath2s@xmath3 template s body - centered tetragonal structure ( @xmath17 ) .
( b ) refinement of the fes product s structure with a primitive tetragonal model ( @xmath18 ) .
fe ( orange ) ions are tetrahedrally coordinated to s ( yellow ) anions , and the k ( purple ) cations are located between two fes layers .
tick marks corresponding to their respective phase are shown below the difference curve . ] before the report by lai _ _ et al . _ _ on superconductivity , several studies had found fes to be a ferrimagnetic semiconductor .
@xcite the conflicting reports on the properties of polycrystalline fes by different groups may be due to impurities not observed through powder x - ray diffraction , especially since iron provides a high background from fluorescence with cu k-@xmath19 radiation .
powder fes samples prepared through aqueous methods may form small crystallites as indicated by the broad bragg reflections in the diffraction patterns of past studies.@xcite the small particle size and polycrystalline nature of these samples impede accurate electrical resistivity and magnetization measurements due to grain boundary effects and the facile oxidation of surfaces of small particles.@xcite despite their ground - breaking work on polycrystalline fes , lai _ _ et al . _ _ also called for high quality single crystal data for definitive determination of the physical properties of fes .
we report a method for the preparation of high quality single crystals of mackinawite fes . since fes is metastable,@xcite single crystal growth through slow cooling of a melt is not possible . in the case of fese@xmath20te@xmath21@xcite and fe@xmath22te,@xcite large single crystals
were grown through bridgeman techniques allowing detailed transport and spectroscopic experiments . for fese , which has limited window of phase stability , chemical vapor transport methods at elevated temperatures
is the only technique that has been reported.@xcite we present a general technique for the de - intercalation of the ternary phase k@xmath1fe@xmath2s@xmath3 figure [ fgr_xrd]a inset ) , which melts congruently and can therefore be prepared in single crystal form.@xcite we link how studying the materials chemistry of layered iron sulfides is key to discovering the underlying physics in new superconductors such as mackinawite fes .
in this work , superconducting fes single crystals were prepared by de - intercalation of potassium cations from k@xmath1fe@xmath2s@xmath3 ( @xmath23 , @xmath24 ) single crystals under hydrothermal condition .
the growth of k@xmath1fe@xmath2s@xmath3 single crystals was modified by the method described by lei _ _ et al . _ _
@xcite for a typical reaction , 1.00 g ( 11.4 mmol ) of hexagonal fes powder ( alfa aesar , 99.9% ) was mixed with 0.18 g ( 4.5 mmol ) of potassium metal ( alfa aesar , 99% ) to match the nominal composition of k@xmath25fe@xmath26s@xmath3 .
the mixture was loaded in a quartz ampoule inside an argon - filled glovebox , and the ampoule was flame sealed under vacuum ( 10@xmath27 torr ) . in order to avoid oxidation of the sample due to the potassium - induced corrosion of quartz ,
the sample containing ampoule was sealed in a larger ampoule under vacuum ( 10@xmath27 torr ) . for crystal growth of k@xmath1fe@xmath2s@xmath3 ,
the mixture was heated to 1000 @xmath28c over 10 hours and held at 1000 @xmath28c for 3 hours to form a homogeneous melt .
subsequently , the melt was slowly cooled at a rate of 6 @xmath28c / hour to 650 @xmath28c to allow crystal growth .
after cooling to room temperature , k@xmath1fe@xmath2s@xmath3 single crystals approximately 3 mm 8 mm in diameter and approximately 0.1 mm in thickness were recovered . for the preparation of fes single crystals , the k@xmath1fe@xmath2s@xmath3 precursor (
0.2 g - 0.4 g ) , 0.28 g ( 5 mmol ) fe powder ( alfa aesar , 99.9% ) , 0.84 g ( 5 mmol ) na@xmath10s @xmath29 @xmath30h@xmath10o ( dried from na@xmath10s @xmath29 @xmath31h@xmath10o , sigma - aldrich , 98% ) and 0.20 g ( 5 mmol ) naoh ( sigma - aldrich , 98% ) were added to 10 ml water .
the mixture was placed in a teflon - lined stainless steel autoclave at 120 @xmath28c for 3 - 4 days .
silver colored fes single crystals were recovered by washing away excess powder with water and drying under vacuum overnight .
samples prepared in the absence of excess iron powder were not superconducting , which could be due to either oxidation of the iron or vacancy formation in the fes layer . in the crystallographic studies of layered iron selenide analogues such as fese@xcite and ( li@xmath32fe@xmath33oh)fese,@xcite iron vacancy formation is implicated in the loss of superconducting properties .
initial powder x - ray diffraction ( xrd ) data were collected using a bruker d8 x - ray diffractometer with cu k@xmath34 radiation , @xmath35 = 1.5418 (step size = 0.025@xmath28 , with 2@xmath36 ranging from 7@xmath28 - 90@xmath28 ) .
temperature dependent x - ray diffraction on ground single crystals was performed using a bruker c2 diffractometer with a vantec500 2d detector , @xmath35 = 1.5418 ( step size = 0.05@xmath28 , with 2@xmath36 ranging from 11@xmath28 - 80@xmath28 ) .
the sample was heated using an anton paar dhs 1100 graphite - dome hot stage .
rietveld refinements were carried out using topas software .
differential scanning calorimetry ( dsc ) was conducted on a mettler - toledo tga / dsc 3 + thermogravimetric analyzer with high temperature furnace .
samples were heated from room temperature to 800 @xmath28c .
magnetic susceptibility measurements were performed using a quantum design magnetic properties measurement system ( mpms ) .
both field - cooled ( fc ) and zero field - cooled ( zfc ) measurements were taken from 2 k to 300 k in direct current mode with an applied magnetic field of 10 oe 30 oe .
hysteresis measurements were carried out at 2 k with @xmath377 t. magnetic susceptibility measurements under hydrostatic pressure were performed using a becu piston - cylinder clamp cell employing n - pentane : isoamyl alcohol as a pressure - transmitting medium .
pressures produced on the single crystal sample at low temperatures were calibrated by measuring the meissner effect of a small piece of pb , placed in the pressure cell .
the known pressure dependences of the superconducting transition temperature of pb @xcite were used for this purpose .
electrical transport measurements were performed on a 14 t quantum design dynacool physical properties measurement system ( ppms ) .
single crystal samples were mounted on a rotator ac transport sample board and measured using the electrical transport option , applying currents between 0.1 - 0.5 ma and frequencies near 10 hz .
heat capacity measurements were performed in a 14 t quantum design dynacool ppms system .
the single - crystal sample of mass 2.9 mg was measured using the relaxation method with field applied perpendicular to the basal plane .
our strategy for preparing single crystals of a metastable phase can be summarized as crystal - to - crystal conversion from a thermodynamically stable phase . during the preparation of our fes samples , we found that maintaining a reducing and basic hydrothermal environment was crucial to observing superconductivity in fes .
the de - intercalation of potassium cations from k@xmath1fe@xmath2s@xmath3 resulted in the shift of alternating planes of fes along the @xmath38 direction of the unit cell to form the primitive layered fes ( figure [ fgr_xrd ] ) .
note that lei _
_ had found k@xmath1fe@xmath2s@xmath3 to be non - superconducting,@xcite so our reductive de - intercalation technique tunes this spin glassy material into a superconductor .
a similar structural transformation from a body - centered tetragonal structure to a primitive tetragonal structure has also been previously observed in the selenide analogue , k@xmath1fe@xmath2se@xmath3.@xcite when exposed to air or moisture , oxidation of iron and formation of iron vacancies was suggested to be the driving force for the structural transition . after the structural change induced by oxidation in water , the superconducting k@xmath1fe@xmath2se@xmath3 became non - superconducting.@xcite in contrast ,
our reductive de - intercalation was driven by preference of potassium cations to solvate into solution under strongly basic conditions , which consequently alters the non - superconducting k@xmath1fe@xmath2s@xmath3 , figure s1 in supplementary materials ( sm ) , into superconducting fes . also , the reducing environment in the autoclave maintained by the presence of fe metal as a reagent prevented oxidation of fe@xmath39 to fe@xmath40 or the formation of iron vacancies . a more drastic structural change could be possible under stronger oxidizing conditions .
neilson and mcqueen@xcite reported that kni@xmath26se@xmath3 , a ni analogue of the k@xmath1fe@xmath2se@xmath3 , forms hexagonal nias - type , k@xmath41fe@xmath42se@xmath3 , by oxidative de - intercalation of k@xmath43 by cui@xmath3 in acetonitrile .
this caused a complete structural reconstruction from edge - sharing layered nise@xmath44 tetrahedra to corner - sharing nise@xmath45 octahedra .
such a reconstruction was not seen in our de - intercalation reaction of k@xmath1fe@xmath2s@xmath3 since we did not utilize strong oxidizing environment but rather maintained reducing conditions .
we similarly found this strategy in achieving the highest @xmath9 s for the ( li@xmath33fe@xmath32oh)fese and ( li@xmath33fe@xmath32od)fese single crystals in their single crystal - to - single crystal conversion also utilizing k@xmath1fe@xmath2se@xmath3 as the template.@xcite a similar method was used for ion exchange in the single - crystal conversion of the selenide analogues k@xmath1fe@xmath2se@xmath3 to ( li@xmath32fe@xmath33oh)fese,@xcite which demonstrates how powerful this technique is for exploring new layered iron chalcogenides . c. under air ( blue curve ) , the loss of ( 001 ) peak is more abrupt and the peak is absent above 200 @xmath0c .
( b ) dsc results , plotted as heat flow as a function of temperature , for single crystal fes . the sudden change in heat flow at 300 @xmath0c is associated with an endothermic reaction . ]
the xrd powder pattern of ground single crystals of k@xmath1fe@xmath2s@xmath3 , presented in figure [ fgr_xrd]a , shows pure crystalline product before the de - intercalation reactions . the pattern for k@xmath1fe@xmath2s@xmath3 was fit with a body - centered tetragonal structural model with space group @xmath17 and
lattice parameters @xmath38 = 3.745(1 ) and @xmath46 = 13.627(9 ) (table [ crystal_data ] , figure [ fgr_xrd ] ) .
full structural parameters from the fits are presented in table [ crystal_data ] and are in good agreement with those presented in an earlier study.@xcite recently , pachmayer _ et al .
_ found that fes powders prepared by hydrothermal methods remain tetragonal down to low temperatures;@xcite while the heavier congeners , fese@xcite and fete,@xcite are known to have a crystallographic phase transitions .
after hydrothermal de - intercalation of potassium cations , the xrd pattern of the newly formed superconducting fes crystals were fit to a primitive unit cell with space group @xmath47 and lattice parameters @xmath48 = 3.6286(5 ) and @xmath49 = 5.03440(9 ) .
these values were consistent with values previously reported for tetragonal fes.@xcite due to the layered nature of the samples , the xrd powder patterns for k@xmath1fe@xmath2s@xmath3 and fes were refined with preferred orientation along the [ 002 ] and [ 001 ] directions , respectively .
table [ crystal_data ] presents the parameters of our structural refinements for ground single crystals of k@xmath1fe@xmath2s@xmath3 and fes as well as the powder samples of fes prepared as a side reaction during the single - crystal - to - single - crystal conversion .
this powder consisted primarily of the product from the reaction of the iron powder in the presence of sodium sulfide and naoh during the hydrothermal preparation ( figure s2 in sm ) .
for comparison , we have also prepared a powder sample of fes through a modified method employed by lai _
et al._,@xcite and the results from our diffraction measurements of a powder sample with @xmath50 k are presented in the sm ( figures s3s5 and table s1 in sm ) .
[ crystal_data ] to test the thermal stability of our new fes single crystals , samples were heated under inert argon atmosphere in steps ranging from 25 @xmath28c , 50 @xmath28c , and 100 @xmath28c .
the 001 peak is visible up to 250 @xmath28c ( figure s6 in sm ) , and its integrated intensity versus temperature under an argon atmosphere is presented in figure [ fgr_tga]a along with a plot of the dsc .
the decomposition of mackinawite fes as determined by the integrated intensity of the ( 001 ) peak begin above 100 @xmath28c and disappeared completely above 250 @xmath28c . due to the geometry of the xrd experiment ,
the ( 00@xmath51 ) reflections in the single crystal sample were observed while other reflections were not .
therefore , it is likely that if greigite were to form above @xmath52 = 100 @xmath0c , it would not have been detected in our experiment .
dsc measurements of fes in argon up to 600 @xmath0c , shown in figure [ fgr_tga]b , give some clues on the thermal behavior during the decomposition of mackinawite .
the dip in the heat flow around 300 @xmath28c indicates an endothermic reaction that could be associated with the crystallization of a phase such as pyrrhotite not seen in our temperature dependent diffraction studies .
the appearance of this transition in the dsc after the disappearance of the ( 001 ) reflection in the xrd , indicates that the two are related .
xrd analysis on the residue from the dsc experiment indicated formation of hexagonal pyrrhotite ( figure s7 in sm )
. the higher than expected thermal stability of the mackinawite compared to past studies could be due to the single crystalline nature of our samples , which have larger surface areas and are therefore less reactive than a polycrystalline product with small particle sizes . from their high - resolution x - ray diffraction study , lennie _ _ et al . _ _ reported that mackinawite begins to decompose to greigite ( fe@xmath53s@xmath16 ) above 100 @xmath0c and that all fes reflections disappear above @xmath52 = 200 @xmath0c under a he atmosphere.@xcite above 260 @xmath0c , greigite decomposes and hexagonal pyrrhotite begins to emerge.@xcite lennie _ _ et al . _ _ also reported that mackinawite - fes rapidly oxidizes under air.@xcite . to test the air stability of our single crystals , we heated samples under ambient atmosphere in steps ranging from 25 @xmath28c , 50 @xmath28c , and 100 @xmath28c .
as presented in fig [ fgr_tga]b , the ( 001 ) peak is visible up to 200 @xmath28c . as this level of air stability
has not been reported for mackinawite before , it could imply that there may be some alkali metal incorporation that could passivate the surface and prevent oxidation of fes .
eds mapping on the surface of fes single crystals shows up to 9% total alkali ( k and na ) on the surface of the fes crystals ( figure s8 in sm ) . due to the similarity of the @xmath46-parameter to those previously reported fes , it is unlikely that large cations such as sodium or potassium intercalate between layers .
of an fes crystal with a @xmath54 shows pauli paramagnetic behavior in the normal state and transitions to the superconducting at @xmath9 = 3.5 k. ( b ) susceptibility for @xmath55 with an increased diamagnetic response with a relative volume fraction increase of 30% ( c ) magnetization @xmath56 as a function of applied field at 2 k. the diamagnetic response weakens for fields greater than 4 mt ( @xmath57 ) and 5 mt ( @xmath55 ) . ]
the temperature - dependent fc and zfc magnetic susceptibilities of fes crystals measured in a constant field of 1 mt are presented in figure [ fgr_mag ] , for fields applied both parallel and perpendicular to the crystallographic @xmath46-axis .
the volume susceptibility @xmath58 under zfc conditions exhibits an onset superconducting transition at @xmath59 k and a shielding fraction of @xmath6060 - 90% ( without geometric factors taken into account ) .
the significant superconducting volume fractions indicate that fes is a bulk superconductor . in both cases of the field orientation ,
the zfc and fc curves in the normal state above @xmath9 are largely temperature independent , indicative of pauli paramagnetism and therefore metallicity in fes .
figure [ fgr_mag]c presents magnetization ( @xmath56 ) as a function of applied field ( @xmath61 ) along two different directions for the applied field .
the @xmath62 isotherms indicate the values of the lower critical field @xmath63 to be 4 mt and 5 mt at 1.8 k for @xmath64 and @xmath65 , respectively .
one difference between our single crystal results and those of lai _ et al .
_ is the maximum critical temperature observed .
et al . _ reported the superconducting powder samples of fes to have a @xmath9 = 4.5 k,@xcite which is approximately 1 k greater than found for our single crystals .
magnetic susceptibility of our own prepared powder samples show @xmath66 = 4 k ( figures s4 and s5 in sm ) .
heat capacity was measured on a large single crystal in both the superconducting ( 0 t ) and normal ( 3 t ) states . as shown in figure [ fgr_hc ] , a 3 t field is large enough to suppress the superconducting state in the crystal , making for a good comparison with the 0 t curve . in zero
applied field , a clear signature of the superconducting transition develops at @xmath9=3.9 k , consistent with magnetic susceptibility and resistivity ( below ) measurements , confirming bulk superconductivity in single crystal fes .
fitting the 3 t data to a standard electron and phonon contribution specific heat model , @xmath67 , yields a normal state sommerfield coefficient to be @xmath68=5.1 mj / mol - k@xmath5 and phonon term @xmath15=0.23 mj / mol - k@xmath69 , the latter corresponding to a debye temperature @xmath70= 257 k. unlike reports for fese where the specific heat was fit to @xmath71 , @xcite , for fes a plot for @xmath72 vs @xmath73 is linear in the normal state .
fes does share some similarities with fese , however , as @xmath68 was estimated to be 5.4(3 ) mj / mol - k@xmath5,@xcite , which is within error to the value we found for @xmath68 in fes .
= 3.9 k. ] temperature dependent electrical resistivity of single - crystal fes is presented in figure [ fgr_res]a .
the resistivity exhibits metallic character down to the superconducting state with @xmath66 = 3.5 k and @xmath74 = 2.4 k. the residual resistivity of fes was determined to be @xmath75 cm based on an average of the values measured for several samples ( figure s9 in sm ) , all of which exhibit a room temperature to residual resistivity ratio ( rrr ) of approximately 10 , indicative of the high quality of our crystalline samples and the low uncertainty in geometric factors that may vary widely due to the micaceous nature of the crystals .
= 3.5 k. the geometry of the resistivity measurement for the single crystal also shown as inset .
( b ) resistivity as a function of applied magnetic field for both @xmath57 and @xmath55 orientations ( always transverse to current direction ) . ] , defined as the deflection from @xmath46-axis direction .
angular dependence of ( a ) longitudinal ( @xmath76 ) and ( b ) transverse ( @xmath77 ) mr taken at 1.8 k are presented .
insets in each figure display a zoom of the superconducting @xmath7 transition . ( c ) comparison of angular dependence of transverse and longitudinal mr at 1.8 k and 14 t. ] figure [ fgr_res]b presents the normalized magnetoresistance ( mr ) as a function of applied magnetic field at 1.8 k. as shown , a significant anisotropy appears in both the normal state high - field mr as well as the @xmath7 transition , with the latter ranging from 0.16 t for @xmath78 to 1.6 t for @xmath79 .
the full angular dependence of these features are presented in figure [ fgr_rtheta ] .
panels ( a ) and ( b ) present the angular variation of mr for both longitudinal ( @xmath76 ) and transverse ( @xmath77 ) orientations , respectively . indeed ,
as shown in figure [ fgr_rtheta]c , the mr angular variation is well represented by a cosine - like dependence for both longitudinal and transverse orientation angles .
a very large anisotropy is also evident in the upper critical field @xmath7 as the field angle is rotated away from the @xmath46-axis . in both longitudinal and transverse orientations , @xmath7 is observed to diminish strongly as the field rotates toward the basal plane , as shown in the insets of figure [ fgr_rtheta]a - b .
taking the two extremes , one can define an @xmath7 anisotropy @xmath80 , which is a value of 10 at 1.8 k. a more complete evaluation of the full @xmath81 dependence allows for an extrapolation of @xmath82 to zero temperature .
as shown in figure [ fgr_hc2]a - d , extracting the @xmath81 values from the resistive transitions at several angles ( all transverse to current direction , with @xmath9 values chosen at the 50% resistance midpoint ) leads to a full @xmath81 plot given in figure [ fgr_hc2]e .
for all field directions , _
h@xmath83(0 ) _ was estimated using the werthamer - helfand - hohenberg ( whh ) formula ( @xmath84|_{t_c}t_c$]).@xcite fitting results give @xmath85 = 2.75 t and @xmath86 t , yielding nearly the same anisotropy value @xmath87=10 as for 1.8 k. the coherence lengths calculated from the estimated @xmath88 values ( @xmath89 where @xmath90 is the flux quantum ) are calculated to be @xmath91 = 343 and @xmath92= 104 . with respect to the crystallographic @xmath46-axis ( transverse to current direction ) .
panel e ) presents the compiled @xmath93 temperature dependences for all angles measured .
@xmath9 values were determined by the resistance transition midpoint .
solid lines represent the whh orbital pair - breaking expectation for @xmath81 in each case ( see text for details ) . ] these large changes in @xmath7 with field angle and the concomitant coherence length anisotropy are in line with the strong anisotropy observed in the normal state mr as discussed above . to determine whether the large @xmath7 anisotropy is indicative of a truly two - dimensional and not a strongly anisotropic three - dimensional superconducting system
, we performed detailed measurements of the angular dependence of @xmath7 at 1.8 k. figure [ fgr_hc2theta ] presents the angle dependence of @xmath7(1.8 k ) as determined from midpoints of field sweep resistive transitions .
( using different criterion to define @xmath7 results in slight variation in absolute anisotropy , but the shape of the @xmath93 curve remains constant ) .
the shape of the @xmath93 curve , especially near the @xmath79 ( @xmath94 ) orientation , is indicative of the true dimensionality of the superconductor with respect to the coherence length .
tinkham s model for thin - film superconductors incorporates the effect of reduced dimensionality , @xcite yielding an angular dependence given by @xmath95 whereas ginzburg landau ( gl ) theory @xcite can be used to determine the effect of an anisotropic effective mass @xmath96 on the angular dependence as @xmath97 as shown in the inset of figure [ fgr_hc2theta ] , the @xmath93 data is much better represented by the anisotropic gl theory , suggesting a highly anisotropic 3d environment for the superconductivity in fes .
this can be quantified by using the calculated anisotropy for this sample @xmath98 to extract the effective mass ratio @xmath99=@xmath100=164 .
this is believed to be the largest upper critical field anisotropy observed in any fe based superconductor reported so far . at 1.8 k. black diamonds represented
measured transition fields ( defined as the resistive transition midpoint ) , and lines represent fits to the theoretical expectation for the angular dependence : solid line represents the ginzburg landau expectation for a 3d system with anisotropic effective mass , and dashed line represents the tinkham s model expectation for 2d superconductors .
inset displays zoomed data near @xmath101 ( @xmath79 ) .
all data were collected with magnetic field direction always transverse to the current direction . ]
the previous report for powder samples of fes found @xmath88 to be 0.4 t,@xcite which is much lower than that of fese and other iron - based superconductors .
@xmath7 for fese has been reported to be 16.3 t in powder samples.@xcite this difference between the upper critical fields in fese and fes has significant effects on their coherence lengths as well .
coherence lengths calculated from @xmath88 for fes powders@xcite and fese powders@xcite are 287 and 45.0 , respectively . we confirm lai s report of a much lower @xmath88 and higher coherence length in fes compared to other iron - based superconductors , but also demonstrate that these properties are highly anisotropic .
as important as the comparatively smaller critical fields in fes , the anisotropy also appears to be much larger in this system .
we find an anisotropy ratio of @xmath102 , and to our knowledge this is the largest reported @xmath82 yet for an iron - based superconductor .
for fete@xmath20s@xmath21 single crystals , the field dependence on @xmath9 is mostly isotropic with a reported @xmath103 = 18 t / 19 t = 0.95.@xcite .
recent studies on fe(se@xmath33s@xmath104 ) single crystals has shown sulfur to increase @xmath9 from 8.5 k for @xmath105 = 0 to 10.7 k for @xmath105 = 0.11 , and the anisotropy is also more pronounced in crystals with higher sulfur content as @xmath106 for @xmath107 and 3.5 for @xmath108.@xcite surprisingly , in our studies of angular dependence of mr , both longitudinal and transverse rotation studies show a diminishment of mr as the field is rotated toward the crystallographic basal plane , irrespective of whether the field direction is rotated parallel or perpendicular to the current direction ( figure [ fgr_rtheta]a , b ) .
this is consistent with either a projection - like orbital mr of a very thin specimen ( _ i.e. _ , with a large mr when @xmath61 is perpendicular to the plane where orbital motion is allowed and zero mr when orbital motion of charge carriers is prohibited by geometric confinement ) , or with a very strong electronic anisotropy as found in other materials with reduced electronic dimensionality .
given the micaceous nature of fes single crystals , the anisotropic behavior of the mr may arise due to a microscopic physical separation of crystalline layers resulting in effectively two - dimensional layers that would act much as in a thin film .
such a description of our sample s behavior would imply that it contains a slab thickness that is less than the characteristic magnetic length scale .
our studies of @xmath7 anisotropy and its angular variation ( figure [ fgr_hc2theta ] ) suggest that the measured superconducting state of fes is in fact inhabiting a three - dimensional environment with strong anisotropy , given the lack of a cusp in @xmath93 near the @xmath101 field alignment ( figure [ fgr_hc2theta ] ) .
the result for our case is in good agreement with gl theory .
therefore , the appropriate length scale to consider is the superconducting coherence length which is 104 for @xmath91 . in other words
, our single - crystal samples must entail crystalline slabs of at least 104 thickness in order to exhibit the gl - type behavior of @xmath7 that follows from eq .
an estimate of the mean free path of quasiparticles @xcite yields @xmath109 , which is much smaller than 104 , suggesting the scattering length is at least much smaller than the known slab thickness . at the very least
, the fact that the effective thickness must be at least @xmath11020 unit cells suggests quasiparticles are not artificially confined , and that the the observed two - dimensional behavior in mr may be intrinsic to the electronic structure .
the tetragonal fes system was originally predicted to be semiconductor in nature by bertaut _ _
_ et al.___@xcite this claim was recently supported by resistivity measurements performed by denholme _ _ _ et al.___@xcite , which showed that their samples were non - superconducting with ferrimagnetic - like behavior .
similarly , samples prepared by sines _ _ _ et al.___@xcite were also exhibited semiconducting and ferrimagnetic behavior .
contrary to experimental evidence published before the work of lai _
et al._,@xcite several other groups had predicted tetragonal fes to be metallic.@xcite vaughan and ridout@xcite proposed that the bonding in the tetragonal fes was metallic in nature due to delocalized _ d _ electrons in iron sublattice .
recent density functional theory ( dft ) calculations also supported metallicity , in tetragonal fes.@xcite geochemists studying mackinawite have suggested that the ferrimagnetic - like behavior from earlier magnetization data might have risen from the well - known thiospinel ferrimagnetic impurity , fe@xmath53s@xmath16 , considering the ease of conversion of mackinawite fes to fe@xmath53s@xmath16.@xcite several of our powder fes samples prepared through the synthesis detailed by lennie _ _
et al.__@xcite form with an fe@xmath53s@xmath16 impurity as revealed by combined magnetization measurements and neutron powder diffraction ( figures s10 - 12 in sm ) .
even denholme _ et al .
_ acknowledged that the semiconductor behavior of fes could be attributed to the surface oxide layers of fes , as suggested by bertaut _
_ @xcite indeed , similar oxidation has been observed in the fese system , as greenfield _ _
et al.__@xcite reported that amorphous surface oxide layers of fese particles suppressed the superconductivity in fese .
our single crystal results definitively support a metallicity in the normal state properties and superconductivity in the ground state . compared to tetragonal fese
, mackinawite fes contains more regular tetrahedral @xmath111fe@xmath111 bond angles where @xmath112 chalcogenide . in fese , the se
fe se out - of - plane bond angle is 112.32(6)@xmath28 and the se
fe se in - plane bond angle is 103.91(7)@xmath28.@xcite the respective bond angles for our fes powder and single crystal samples were calculated to be close to 108.1(3)@xmath28 and 110.2(2)@xmath28 ( table [ crystal_data ] ) .
several studies have suggested that higher @xmath9 could be achieved from more regular bond angles,@xcite as is with iron pnictide superconductors.@xcite however , this structural parameter does not seem to be as important an indicator in the iron chalcogenides since fese exhibits a higher @xmath9 ( 8 k ) than fes ( @xmath9 = 4 k ) even though it is comprised of more distorted tetrahedra .
this suggests that structural factors controlling @xmath9 in iron pnictides may not be identical to those of the iron chalcogenides .
anion height has also been implicated as a reliable predictor for @xmath9 in iron - based superconductors.@xcite for iron pnictides , @xmath9 increases with increasing anion height as fep - based superconductors have lower anion height and lower @xmath9 than feas - based superconductors .
however , @xmath9 begins to drop off for anion heights greater than 1.38 , which suggests there is an optimal anion height for maximizing @xmath9 .
for fese with @xmath9 = 8 k , the se height is 1.45 , and upon application of physical pressure , the se height decreases to 1.425 , which leads to an increase in @xmath9 up to 37 k ( 8 gpa).@xcite for larger anions , i.e. fete , the anion height is larger than that of fese and while fete is not superconducting at ambient pressure isovalent anionic substitution as in fete@xmath113s@xmath114 induces superconductivity ( anion height = 1.75 , @xmath9 = 10 k).@xcite from this anion height principle , we should expect the smaller anionic radius of sulfide to lead to a larger @xmath9 .
however , the anion height in fes was found in the range from 1.27(1 ) to 1.34(1 ) (table [ crystal_data ] ) , which is below the optimal height of 1.38 .
this result for fes could therefore explain why the @xmath9 is remains low and between 3.5 and 5 k despite having more regular tetrahedra than fese or fete . as a preliminary study on modifying the anion height in fes to affect @xmath9 , we have performed magnetization measurements as a function applied pressure .
as shown in figure [ fgr_pressure ] , measurements of magnetic susceptibility in a clamp - cell setup show that the transition temperature decreases with increasing pressure , at least up to 10 kbar . while it is known that @xmath9 in the related superconductor fese undergoes a dramatic enhancement under pressure , the increase in @xmath9 for fese occurs at much higher pressures than currently reached in the present experiment for fes ( on the order of 10 gpa )
. further work to study the relation between @xmath115 and the crystallographic parameters as a function of applied pressure will shed more light on the relation between structure and superconductivity in fes .
in conclusion , we have synthesized superconducting single crystals of fes and characterized their thermal , magnetic , and electrical properties . the synthesis of fes single crystals was accomplished through the novel method of reductive de - intercalation of k@xmath1fe@xmath2s@xmath3 single crystals under hydrothermal conditions .
the fes crystals are stable up to 250 @xmath28c in argon and 200 @xmath28c in air . at 4
k the fes crystals transition from a metallic , pauli paramagnetic state to the superconducting state . in both the normal state and superconducting states ,
we observe a large anisotropy in the properties of fes .
the upper critical field expresses a large anisotropy with a @xmath116 , the largest reported for any iron - based superconductor thus far .
magnetoresistance measurements for the normal state performed as a function of applied field angle reveal a remarkable two - dimensional behavior in fes .
overall , the physical property results indicate that the fermi surface of fes may be highly two - dimensional , and perhaps even more so than other closely - related iron - based superconductors .
since the metastable system , mackinawite - type fes , is now confirmed as a superconductor and not a magnetic semiconductor , this system could be a template for the preparation of new sulfide - based superconductors that exhibit strong anisotropic behavior .
research at the university of maryland was supported by the nsf career dmr-1455118 , afosr grant no
. fa9550 - 14 - 10332 , and the gordon and betty moore foundation grant no .
we acknowledge the support of the national institute of standards and technology , u. s. department of commerce , in providing the neutron research facilities used in this work .
j. paglione and r. l. greene , _ nat .
phys . _ , 2010 , * 6 * , 645658 d. c. johnston , _ adv . phys . _ , 2010 , * 59 * , 8031061 a. ivanovskii , _ physica c _ , 2011 , * 471 * , 409427 h. takahashi , a. sugimoto , y. nambu , t. yamauchi , y. hirata , t. kawakami , m. avdeev , k. matsubayashi , f. du , c. kawashima , h. soeda , s. nakano , y. uwatoko , y. ueda , t. j. sato and k. ohgushi , _ nat .
_ , 2015 , * 14 * , 10081012 a. p. drozdov , m. i. eremets , i. a. troyan , v. ksenofontov and s. i. shylin , _ nature _ , 2015 , * 525 * , 7376 x. lai , h. zhang , y. wang , x. wang , x. zhang , j. lin and f. huang , _
soc . _ , 2015 , * 137 * , 1014810151 o. kouvo , j. long and y. vuorelainen , _ am .
_ , 1963 , * 48 * , 511 e. bertaut , p. burlet and j. chappert , _ solid state commun . _ , 1965 , * 3 * , 335 338
h. t. evans jr , c. milton , e. chao , i. adler , c. mead , b. ingram and r. a. berner , _ us geol .
survey prof . paper _
, 1964 , * 475 * , 1312
1318 a. lennie , s. redfern , p. schofield and d. vaughan , _ mineral .
_ , 1995 , * 59 * , 677684 d. rickard and g. w. luther , _ chem . rev . _ , 2007 , * 107 * , 514562 s. denholme , s. demura , h. okazaki , h. hara , k. deguchi , m. fujioka , t. ozaki , t. yamaguchi , h. takeya and y. takano , _ mater
phys . _ , 2014 , * 147 * , 50 56 i. t. sines , d. d. vaughn ii , r. misra , e. j. popczun and r. e. schaak , _ j. solid state chem . _ , 2012 , * 196 * , 1720 a. k. dutta , s. k. maji , d. n. srivastava , a. mondal , p. biswas , p. paul and b. adhikary , _ acs appl . mater . & interfaces _ , 2012 , * 4 * , 19191927 a. r. lennie , k. e. england and d. j. vaughan , _ am . mineral .
_ , 1995 , * 80 * , 960967 d. csakberenyi - malasics , j. d. rodriguez - blanco , v. k. kis , a. recnik , l. g. benning and m. posfai , _ chem
. geol . _ , 2012 , * 294 - 295 * , 249
258 w. bao , y. qui , q. huang , m. a. g. a. p. zajdel , m. r. fitzsimmons , m. zhernenkov , s. chang , m. fang , b. qian , e. k. vehstedt , j. yang , h. m. pham , l. spinu and z. q. mao , _ phys .
_ , 2009 , * 102 * , 247001 t. j. liu , j. hu , b. qian , d. fobes , z. q. mao , w. bao , m. reehuis , s. a. j. kimber , k. prokes , s. matas , d. n. argyriou , a. hiess , a. rotaru , h. pham , l. spinu , y. qiu , v. thampy , a. t. savici , j. a. rodriguez and c. broholm , _ nat .
_ , 2010 , * 9 * , 716720 v. bhatia , e. e. rodriguez , n. p. butch , j. paglione and m. a. green , _ chem .
comm . _ , 2011 , * 47 * , 11297 e. e. rodriguez , c. stock , p. zajdel , k. l. krycka , c. f. majkrzak , p. zavalij and m. a. green , _ phys .
b _ , 2011 , * 84 * , 064403 c. stock , e. e. rodriguez , m. green , p. zavalij and j. rodriguez - rivera , _ phys .
b _ , 2011 , * 84 * , 045124 y. hara , k. takase , a. yamasaki , h. sato , n. miyakawa , n. umeyama and s. ikeda , _ physica c _ , 2010 , * 470 * , s313
s314 a. e. bhmer , t. arai , f. hardy , t. hattori , t. iye , t. wolf , h. v. lhneysen , k. ishida and c. meingast , _ phys .
lett . _ , 2015 , * 114 * , 027001 r. hu , k. cho , h. kim , h. hodovanets , w. e. straszheim , m. a. tanatar , r. prozorov , s. l. budko and p. c. canfield , _ supercond .
_ , 2011 , * 24 * , 065006 x. g. luo , x. f. wang , j. j. ying , y. j. yan , z. y. li , m. zhang , a. f. wang , p. cheng , z. j. xiang , g. j. ye , r. h. liu and x. h. chen , _ new j. phys . _ , 2011 , * 13 * , 053011 h. lei , m. abeykoon , e. s. bozin and c. petrovic , _ phys . rev .
b _ , 2011 ,
* 83 * , 180503 t. m. mcqueen , q. huang , v. ksenofontov , c. felser , q. xu , h. zandbergen , y. s. hor , j. allred , a. j. williams , d. qu , j. checkelsky , n. p. ong and r. j. cava , _ phys .
b _ , 2009 , * 79 * , 014522 h. sun , d. n. woodruff , s. j. cassidy , g. m. allcroft , s. j. sedlmaier , a. l. thompson , p. a. bingham , s. d. forder , s. cartenet , n. mary , s. ramos , f. r. foronda , b. h. williams , x. li , s. j. blundell and s. j. clarke , _ inorganic chemistry _ , 2015 ,
* 54 * , 19581964 t. f. smith and c. w. chu , _ phys .
_ , 1967 , * 159 * , 353358 h. lei , m. abeykoon , e. s. bozin , k. wang , j. b. warren and c. petrovic , _ phys . rev
. lett . _ , 2011 , * 107 * , 137002 d. p. shoemaker , d. y. chung , h. claus , m. c. francisco , s. avci , a. llobet and m. g. kanatzidis , _ phys .
b _ , 2012 , * 86 * , 184511 j. r. neilson and t. m. mcqueen , _ j. amer
. chem . soc .
_ , 2012 , * 134 * , 77507757 x. zhou , c. k. h. borg , j. w. lynn , s. r. saha , j. paglione and e. e. rodriguez , _ http://arxiv.org/abs/1512.03399 _ x. dong , k. jin , d. yuan , h. zhou , j. yuan , y. huang , w. hua , j. sun , p. zheng , w. hu , y. mao , m. ma , g. zhang , f. zhou and z. zhao , _ phys . rev .
b _ , 2015 , * 92 * , 064515 u. pachmayr , n. fehn and d. johrendt , _ chem
. commun . _ , 2016 , s. margadonna , y. takabayashi , y. ohishi , y. mizuguchi , y. takano , t. kagayama , t. nakagawa , m. takata and k. prassides , _ phys . rev .
b _ , 2009 , * 80 * , 064506 e. e. rodriguez , d. a. sokolov , c. stock , m. a. green , o. sobolev , j. a. rodriguez - rivera , h. cao and a. daoud - aladine , _ physical review b _ , 2013 , * 88 * , 165110 a. r. lennie , s. a. t. redfern , p.
e. champness , c. p. stoddart , p. f. schofield and d. j. vaughan , _ am .
_ , 1997 , * 82 * , 302309 n. r. werthamer , e. helfand and p. c. hohenberg , _ phys
_ , 1966 , * 147 * , 295302 m. tinkham , _ phys
_ , 1963 , * 129 * , 24132422 j. ketterson and s. song , _ superconductivity _ , cambridge university press , 40 west 20th street , new york , ny 10011 - 4211 , usa , 1999 , pp .
4546 f .- c .
hsu , j .- y .
luo , k .- w .
yeh , t .- k .
chen , t .- w .
huang , p. m. wu , y .- c .
lee , y .- l .
huang , y .- y .
_ , _ p. natl
usa _ , 2008 , * 105 * , 1426214264 r. hu , e. s. bozin , j. b. warren and c. petrovic , _ phys .
b _ , 2009 ,
* 80 * , 214514 m. abdel - hafiez , y .- y .
zhang , z .- y .
cao , c .- g .
duan , g. karapetrov , v. m. pudalov , v. a. vlasenko , a. v. sadakov , d. a. knyazev , t. a. romanova , d. a. chareev , o. s. volkova , a. n. vasiliev and x .- j .
chen , _ phys .
b _ , 2015 , * 91 * , 165109 a. j. millis , s. sachdev and c. m. varma , _ phys . rev . b. _ , 1988 , * 37 * , 4795 d. vaughan and m. ridout , _ j. inorg
_ , 1971 , * 33 * , 741 746 a. subedi , l. zhang , d. j. singh and m. h. du , _ phys . rev .
b _ , 2008 , * 78 * , 134514 k. d. kwon , k. refson , s. bone , r. qiao , w .- l .
yang , z. liu and g. sposito , _ phys .
b _ , 2011 , * 83 * , 064402 a. j. devey , r. grau - crespo and n. h. de leeuw , _ j. phys . chem .
c _ , 2008 , * 112 * , 1096010967 j. brgoch and g. j. miller , _ j. phys .
a _ , 2012 , * 116 * , 22342243 j. t. greenfield , s. kamali , k. lee and k. kovnir , _ chem .
_ , 2015 , * 27 * , 588596 c. lee , k. kihou , a. iyo , h. kito , p. shirage and h. eisaki , _ solid state commun . _ , 2012 ,
* 152 * , 644 648 y. mizuguchi , y. hara , k. deguchi , s. tsuda , t. yamaguchi , k. takeda , h. kotegawa , h. tou and y. takano , _ supercond .
sci . tech .
_ , 2010 , * 23 * , 054013 t. imai , k. ahilan , f. l. ning , t. m. mcqueen and r. j. cava , _ phys . rev
. lett . _ , 2009 , * 102 * , 177005 y. mizuguchi , f. tomioka , s. tsuda , t. yamaguchi and y. takano , _ appl .
lett . _ , 2009 , * 94 * ,
p. zajdel , p .- y .
hsieh , e. e. rodriguez , n. p. butch , j. d. magill , j. paglione , p. zavalij , m. r. suchomel and m. a. green , _
soc . _ , 2010 , * 132 * , 1300013007 | we report the novel preparation of single crystals of tetragonal iron sulfide , fes , which exhibits a nearly ideal tetrahedral geometry with s fe s bond angles of 110.2(2 ) @xmath0 and 108.1(2 ) @xmath0 .
grown via hydrothermal de - intercalation of k@xmath1fe@xmath2s@xmath3 crystals under basic and reducing conditions , the silver , plate - like crystals of fes remain stable up to 200 @xmath0c under air and 250 @xmath0c under inert conditions , even though the mineral `` mackinawite '' ( fes ) is known to be metastable .
fes single crystals exhibit a superconducting state below @xmath4 k as determined by electrical resistivity , magnetic susceptibility , and heat capacity measurements , confirming the presence of a bulk superconducting state .
normal state measurements yield an electronic specific heat of 5 mj / mol - k@xmath5 , and paramagnetic , metallic behavior with a low residual resistivity of 250 @xmath6 cm .
magnetoresistance measurements performed as a function of magnetic field angle tilted toward both transverse and longitudinal orientations with respect to the applied current reveal remarkable two - dimensional behavior .
this is paralleled in the superconducting state , which exhibits the largest known upper critical field @xmath7 anisotropy of all iron - based superconductors , with @xmath8(2.75 t)/(0.275 t)=10 .
comparisons to theoretical models for 2d and anisotropic-3d superconductors , however , suggest that fes is the latter case with a large effective mass anisotropy .
we place fes in context to other closely related iron - based superconductors and discuss the role of structural parameters such as anion height on superconductivity . |
we compare the theoretical anisotropic upper critical field @xmath0 of a quasi - two - dimensional d - wave superconductor with recent @xmath1 data for the layered organic superconductor @xmath2 .
we find agreement both with regard to the angular and the temperature dependence of @xmath3 .
this supports the suggestion that the fulde - ferrell - larkin - ovchinnikov state ( fflo state ) exists in this material for exactly plane - parallel orientation of the magnetic field .
indications of precursor states , occurring for small deviations from the plane - parallel field direction , are also pointed out and further measurements for confirming the existence of the fflo state are proposed .
if the superconducting state in high magnetic fields is limited by paramagnetic pair breaking alone , the transition from the homogeneous superconducting state to the normal conducting state may proceed either directly , at the pauli limiting field @xmath4 , or via an interposed inhomogeneous superconducting state predicted in 1964 by fulde and ferrell [ 1 ] and by larkin and ovchinnikov [ 2 ] ( fflo state ) .
the latter state , which stabilizes as a consequence of spin polarization , has attracted considerable interest over the years , but has , despite of this , not yet been definitely verified experimentally .
favourable conditions for observing the fflo state are found in clean superconductors with orbital critical fields much larger than @xmath4 . in practice
, it seems always necessary to reduce the orbital pair breaking effect by using layered superconductors with nearly decoupled planes or extremely thin films ( quasi - two - dimensional superconductors ) and applying the magnetic field in a direction parallel to the conducting planes .
several classes of superconducting materials with favourable conditions for observing the fflo state do exist .
these include the `` classical '' intercalated transition metal dichalcogenides as well as more exotic materials like high-@xmath5 compounds and organic superconductors .
in this letter we refer to a recent measurement of the upper critical field @xmath1 in the organic superconductor @xmath2 [ 3 ] .
this layered material shows strong anisotropy of the superconducting properties with regard to out - of - plane directions .
in addition , a number of experiments listed in reference [ 3 ] are interpreted in terms of a strong in - plane anisotropy , of d - wave type , of the gap parameter .
the coherence length @xmath6 perpendicular to the layers of this clean material is smaller than the interlayer spacing @xmath7 and one expects an extreme reduction of orbital pair breaking for plane - parallel applied field .
the angular dependence of @xmath1 was measured in ref [ 3 ] both with respect to the angle @xmath8 between applied magnetic field and the direction normal to the conducting planes and with respect to the azimuthal angle @xmath9 , which denotes the direction of magnetic field lying within the plane .
the results showed , as expected , a strong variation of @xmath1 with @xmath8 . on the other hand ,
no dependence on @xmath9 was observed .
the maximal value of @xmath1 at the plane - parallel position @xmath10 was of the order of , but 50% higher than , the pauli paramagnetic limit @xmath4 .
these facts led the authors of reference [ 3 ] to propose that their in - plane critical field is the phase boundary between the normal - conducting state and the fflo state of a d - wave superconductor .
we examined this question by calculating the angular and temperature dependence of the theoretical phase boundary @xmath11 between the normal - conducting state and the superconducting states of a quasi - two - dimensional d - wave superconductor .
comparison with the data of reference [ 3 ] showed good agreement , supporting the hypothesis that in this experiment the phase boundary of a fflo state ( for d - wave superconductors ) has been observed for the first time .
we assume that the coupling between the conducting planes of @xmath2 can be neglected , making our problem effectively two - dimensional .
then , if the field has both a perpendicular and a parallel component , the superconducting state is limited by both orbital and paramagnetic pair breaking . for exactly plane - parallel magnetic field
the fflo phase should be realized below a reduced temperature ( @xmath12 ) of @xmath13 [ 4 ] .
such a situation , with competition between both pair breaking effects , has been studied first by bulaevskii [ 5 ] .
his treatment was later generalized to arbitrary temperatures and d - wave superconductors by shimahara and rainer [ 6 ] .
below the stability limit of the normal - conducting state , which will be referred to as @xmath11 , a series of different inhomogeneous superconducting states , depending on @xmath8 , appear .
for s - wave superconductors each one of these states belongs to a particular value of landau s quantum number @xmath14 , which takes integer values @xmath15 .
these states are ( for s - wave ) the following : * the vortex state for small @xmath8 belongs to @xmath16 . * a series of inhomogeneous states for @xmath8 near @xmath17 , each one characterized by a single value @xmath18 , with @xmath14 increasing with increasing @xmath8 . * the fflo state for @xmath10 , which may be characterized by @xmath19 .
the structure of the higher landau level states , for @xmath18 , has been calculated recently for s - wave superconductors by minimizing the quasiclassical free energy [ 7 ] . for d - wave superconductors [ 6 ]
a state below @xmath11 is no longer characterized by a single value of @xmath14 but rather by an infinite subset @xmath20 .
however , the dominant contribution may still be characterized by a single number @xmath14 , which increases again with increasing @xmath8 and approaches infinity in the fflo limit .
thus , basically the above classification scheme remains valid for d - wave symmetry .
the phase boundary of the ` pure ' fflo state for d - wave superconductors [ the curve @xmath21 has been calculated by maki and won [ 4 ] . the linearized gap equation to be solved
is given by [ 5,6 ] @xmath22 \right][1 \nonumber\\
- \cos[s\{\mu_0h - \frac{1}{2 } \vec{v}_f^\prime \vec{\pi}\ } ] ] \delta(\vec{r } ) \mbox{. } \end{aligned}\ ] ] we consider a cylindrical fermi surface , appropriate for the present two - dimensional problem , with the fermi velocity @xmath23 .
the gap parameter is given by @xmath24 , where @xmath25 for s - wave , and @xmath26 for d - wave pairing .
the canonical momentum is defined by @xmath27 .
the magnetic field @xmath28 is assumed to lie in the @xmath29-plane , with @xmath30 and @xmath31 .
we use the following gauge for the vector potential : @xmath32 .
paramagnetic pair breaking enters via the term @xmath33 in ( 1 ) ; the electron s magnetic moment is @xmath34 , with the lande factor @xmath35 and bohrs magneton @xmath36 . the method of reducing equation ( 1 ) to a set of algebraic equations follows exactly reference [ 6 ] and need not be repeated here . as a first check of our numerical method we compared our results with reference [ 6 ] and found complete agreement whenever the same set of input parameters was used .
let us proceed to a comparison of the solutions @xmath3 of equation ( 1 ) with the @xmath1 data reported in reference [ 3 ] .
the first point to address is the independence of @xmath1 on the azimuthal angle @xmath9 .
such a dependence may easily be incorporated in the present model by replacing @xmath37 in ( 1 ) by @xmath38 $ ] .
as long as orbital pair - breaking is present for a parallel field component , the symmetry - breaking term @xmath39 has to be kept in ( 1 ) .
if , on the other hand , complete decoupling of the ( infinitely thin ) conducting planes can be assumed , the term @xmath40 can be dropped and @xmath3 becomes independent of @xmath9 , as can be explicitely confirmed numerically . in this context
, we recall that a three - dimensional d - wave superconductor shows anisotropy of the upper critical field [ 8 ] .
thus , the observed independence of @xmath1 on @xmath9 is in agreement with the present model and , in particular , with the assumption of quasi - two - dimensional superconductivity . if @xmath41 is measured in units of @xmath5 and @xmath42 in units of @xmath43 ( where @xmath44 is the bcs - gap at @xmath45 ) , the present model requires only one single parameter @xmath46 to be fitted .
this parameter is proportional to the `` bulk '' ratio of the orbital and spin critical fields @xmath47 and @xmath48 respectively
. in the present anisotropic model the actual ratio of spin and orbital pair - breaking depends on @xmath8 and may be written in the form @xmath49 .
the best fit to the @xmath8-dependence of @xmath1 at @xmath50 ( see figure 4 of reference [ 3 ] ) has been obtained for @xmath51 .
this value of @xmath46 is consistent with a critical temperature @xmath52 of the sample studied in reference [ 3 ] , and a fermi energy of the order of @xmath53 as estimated from several experiments [ 9 ] . using this value we found very good agreement between @xmath54 and the data of reference
[ 3 ] , as shown in figure 1 .
the theoretical curves in figure 1 have been calculated assuming d - wave symmetry ; the difference between d - wave and s - wave was found to be small except very close to @xmath55 and @xmath56 .
, as a function of @xmath8 .
full squares : data of reference [ 3 ] at @xmath57 .
full line : theoretical result for the bulaevskii - shimahara - rainer phase boundary for @xmath58 . dashed line : theoretical result for @xmath59.,width=491 ] the @xmath1 data show a small but clearly visible kink near the plane - parallel orientation , at @xmath60 .
a similar feature is also found in the theoretical phase boundary @xmath61 , as shown in figure 1 ( where the square of the critical fields has been plotted in order to make this discontinuous change in slope better visible ) .
this kink indicates the transition from the vortex state , with @xmath16 , to the first of the above mentioned fflo - precursor states , with @xmath62 .
still closer to @xmath63 equation ( 1 ) yields additional transitions corresponding to @xmath64 , which are still visible in figure 1 in the theoretical curve but not in the data points .
the @xmath54-curves describing the @xmath65 transitions , for the same material but higher @xmath41 , are shown on a larger scale in figure 2 .
the order parameter structure of the precursor phases , where pairing takes place in landau levels @xmath18 , has been investigated recently for s - wave superconductors [ 7 ] .
two types of such precursor states have been found : ( i ) quasi - one - dimensional states , which may be considered as a mixture of rows of vortices and one - dimensional fflo - type oscillations , and ( ii ) two - dimensional lattices with several zeros of the order parameter with different vorticity [ 7 ] .
such unusual states [ of type ( ii ) ] have been predicted to occur in the extremely high field region where quantization of single electron levels becomes important [ 10 ] .
the present arrangement might provide a relatively feasible way of observing such vortex structures .
it should be mentioned , however , that for d - wave superconductors neither the equilibrium structure of the fflo state nor that of the @xmath18 states has been calculated so far . of the upper critical field , as calculated from equation ( 1 ) ,
are plotted in more detail , using the same value of @xmath66 as in figure 1 but a higher temperature @xmath67.,width=491 ] -dependence of the upper critical field of @xmath68 , as reported in reference [ 11 ] , with the solutions of equation ( 1 ) for s - wave superconductivity at @xmath69 and @xmath70.,width=491 ] the shape of the @xmath61 curve depends distinctively on temperature , as shown by the plot ( dashed line ) of @xmath71 at @xmath59 in figure 1 .
the reason for this enhancement at higher @xmath41 is , of course , that paramagnetic pair breaking becomes less effective at higher temperature .
data at higher @xmath41 have not been reported in reference [ 3 ] but would be useful in order to check the present interpretation . looking for similar measurements we found old data by morris and coleman
[ 11 ] for intercalated transition metal dichalcogenide @xmath68 samples . in this material , which represents a nearly perfect realization of two - dimensional superconductivity , an unexplained anomaly with regard to the behavior of the upper critical field at different @xmath41 has been reported ( figure 10 of reference [ 11 ] ) .
we find excellent agreement ( see figure 3 ) comparing the data of reference [ 11 ] with the solutions of equation ( 1 ) for an s - wave superconductor .
again , a single parameter has been adjusted ( @xmath72 ) to obtain both of the theoretical curves shown in figure 3 .
the resistance data reported in reference [ 11 ] show a non - monotonic behavior near the plane - parallel field orientation ( see figure 3 of reference [ 11 ] ) , which may be due to transitions to the @xmath18 states .
the latter states are discussed in more detail in reference [ 7 ] . with the theoretical result reported in reference [ 4].,width=491 ] finally ,
let us compare the measured temperature - dependence of @xmath1 for the plane - parallel field orientation ( figure 3 of reference [ 3 ] ) with @xmath73 .
according to our interpretation of these data , the states below @xmath74 should be a d - wave version of the fflo state for @xmath75 , and the homogeneous superconducting state for @xmath76 .
this phase boundary has been calculated first by maki and won [ 4 ] .
the shimahara rainer d - wave phase boundary must agree with reference [ 4 ] in the limit @xmath77 .
we found agreement , except for the steep rise of @xmath73 below @xmath78 reported in reference [ 4 ] ( a possible reason for this discrepancy may be slow convergence of our numerical method for low @xmath41 and high @xmath14 ) .
the comparison between theory [ 4 ] and experiment [ 3 ] depicted in figure 4 shows again fairly good agreement .
a characteristic difference in temperature variation above and below @xmath79 is visible in the data points , although it is less pronounced than in the theoretical curve .
the difference in critical fields between s - wave and d - wave is again rather small ; at @xmath80 , the lowest temperature , where measurements for @xmath2 have been reported [ 3 ] , both are approximately given by the standard @xmath81-result [ 12 ] for the fflo state , @xmath82 , which exceeds the pauli limiting field by @xmath83 .
thus , these numbers do also fit well into the fflo interpretation of the phase boundary for plane - parallel field orientation .
summarizing , the proposal of nam et al . [ 3 ]
that upper critical field data for a plane - parallel field orientation in the layered organic superconductor @xmath2 should be interpreted in terms of a fflo state , has been supported by our calculations .
the data agree with the predictions of a model of a quasi - two - dimensional superconductor both with regard to the angular and the temperature dependence of the critical field .
further confirmation of this interpretation could be obtained by means of measurements at higher temperatures , where paramagnetic pair - breaking is strongly reduced .
if this interpretation is correct , precursor states with interesting properties should appear for applied fields close to the plane - parallel orientation . |
|
the absence of any experimental signal , so far , in the searches for supersymmetry ( susy ) at the lhc @xcite and the discovery of a new higgs - like particle by atlas @xcite and cms @xcite with a mass @xmath18 125 - 126 gev , are pushing the susy mass parameters above the 1-tev range . on one hand ,
the present lower mass bounds for squarks of the first and second generations and for gluinos are already above 1 tev , and on the other hand , if the observed higgs boson is identified with the lightest higgs boson @xmath6 of the minimal supersymmetric standard model ( mssm ) , a radiatively corrected mass @xmath19 125 - 126 gev also implies rather heavy squark masses of the third generation ( mainly stop masses ) at or larger than 1 tev . in principle , to place the susy masses at the multi - tev range
seems discouraging , both from an experimental point of view due to the inability to detect susy directly , and from a theoretical point of view , in regard to the naturalness of the theory , which contrarily suggests a soft susy - breaking scale , @xmath20 , at or below the tev scale .
however , leaving the naturalness issue aside , the mssm scenarios with very heavy susy masses can have other interesting aspects @xcite .
in particular , this is the case of specific higgs boson observables , like certain higgs partial decay widths , which present a non - decoupling behavior with @xmath20 , as shown , for instance , in @xcite , opening a new window to the indirect detection of heavy susy .
as it is well known , the decoupling of susy radiative corrections in the asymptotic large susy mass limit is valid for susy theories with an exact gauge symmetry , in agreement with the general decoupling behavior of heavy states in quantum field theory as stated in the famous appelquist - carazzone theorem @xcite .
nevertheless , it is also known that this theorem does not apply to theories with spontaneously broken gauge symmetries , nor with chiral fermions , which is the case of the mssm .
furthermore , in order to have decoupling , the dimensionless couplings should not grow with the heavy masses .
otherwise , the mass suppression induced by the heavy - particle propagators can be compensated by the mass enhancement provided by the interaction vertices , with an overall non - decoupling effect , which is exactly what happens in some mssm higgs boson decays to fermions .
for instance , it was studied in @xcite how non decoupling appears for large susy masses in the @xmath21 decay through one - loop susy - qcd corrections , when the involved susy particle masses grow simultaneously with a generic soft susy - breaking scale @xmath20 ( see also @xcite ) .
a similar non - decoupling behavior was obtained for flavor changing neutral higgs boson decays into quarks of the second and third generations through both susy - qcd corrections @xcite and susy - ew corrections @xcite .
other interesting non - decoupling susy - ew effects have also been seen in lepton flavor violating ( lfv ) higgs boson decays within the context of the mssm - seesaw model @xcite ; and in higgs - mediated lfv processes like : @xmath22 decays @xcite , some semileptonic @xmath23 decays @xcite and in @xmath24 conversion in heavy nuclei @xcite ; all of them within the mssm - seesaw model too .
other non - decoupling effects from heavy susy particles have also been noticed within the context of the susy inverse - seesaw model @xcite .
some studies of susy non decoupling within the mssm have also been performed in the effective field theory approach where the effective higgs - fermion - fermion vertices are induced from one - loop corrections of heavy susy particles @xcite and in @xmath25 production @xcite . in the present work , motivated by the recent discovery of the sm - like higgs boson
, we will focus on the study of the lfv higgs boson decays and we will work within the context of the mssm at one - loop order with the hypothesis of general slepton flavor mixing .
we will extend the study to the three neutral higgs bosons of the mssm @xmath6 , @xmath7 and @xmath26 , considering the new higgs - like particle to be the lightest higgs boson @xmath6 .
in particular , we will study the lfv higgs decays @xmath13 , @xmath27 and @xmath8 .
this kind of processes provide an unique window into new physics due to the strong suppression of flavor violation in the standard model ( sm ) , where the flavor mixing is induced exclusively by the yukawa couplings of the corresponding fermion sector .
this is specially interesting in the lepton sector in which the flavor mixing will be hugely suppressed because of the smallness of the lepton yukawa couplings .
therefore , the discovery of any process which violates the lepton flavor number would be a clear signal of physics beyond the sm .
lfv is nowadays a very active field which is being studied in different models , through several channels ( for a review , see for instance @xcite ) : radiative decays @xmath28 , @xmath29 and @xmath30 ; 3-body decays @xmath22 , @xmath31 and @xmath32 ; @xmath24 conversion in nuclei ; semileptonic @xmath23 decays ; among others . the specific case of lfv higgs
decays within supersymmetric models has also deserved special attention in the literature @xcite . also encouraging results for the reach of lfv higgs
decays at the lhc have been recently obtained in @xcite within the context of a general effective lagrangian approach .
our purpose here is to take advantage of the mentioned non - decoupling behavior with @xmath0 in the lfv higgs decay widths into charged leptons of different generations , @xmath33 , with @xmath34 and @xmath35 , in order to look for sizeable branching ratios which can give rise to detectable signals at the lhc .
here and from now on , @xmath36 with @xmath34 in the final state of the lfv decays refers to either @xmath37 or @xmath38 . in general , the radiatively corrected lfv higgs couplings to leptons are proportional to the heaviest lepton mass involved , being this the reason why we select the @xmath39 channel as the most promising one .
in addition , the @xmath40 channel is extremely restricted by the associated radiative decay , @xmath30 @xcite , leaving us almost no room to move in the allowed parameter space of slepton flavor mixing , and driving us to extremely low rates , not measurable in any lhc detector . the @xmath41 channel , on the other hand , gives us very similar results to the @xmath39 channel , and from an experimental point of view , the lhc sensitivity to the former should be equivalent to the latter @xcite .
therefore the results obtained along this work for the lfv higgs decays into @xmath39 are straightforwardly translated into the @xmath41 channels . for the present study we will focus then on
the @xmath42 decays within the mssm at one loop with general slepton flavor mixing between the second and the third generations , without assuming any particular source of flavor mixing .
consequently , this general slepton mixing will be parametrized in a model - independent way by four dimensionless parameters , @xmath43 , @xmath44 , @xmath45 and @xmath46 .
it is important to remark that our calculations will be made in the mass eigenstate basis for all mssm particles , including both charged sleptons and sneutrinos with the corresponding exact diagonalization of the full charged slepton and sneutrino mass matrices , and therefore we will not use here the mass insertion approximation ( mia ) .
the most important constraints to these four slepton mixings come from the @xmath28 searches @xcite .
a recent study on updated constraints to all these general slepton flavor mixings @xmath47 in @xcite indicates that , indeed , the present bounds to br@xmath48 lead to constraints on the 23 mixings which for 500 gev @xmath49 1500 gev and 5 @xmath50 60 are at @xmath51 , and @xmath52 , and for heavier susy masses lead to weaker bounds . we will check that by raising the susy mass scale into the multi - tev range will turn into the needed relaxation of these bounds , since these lfv radiative decays present a decoupling behavior with the susy scale .
this will allow us to explore a high susy mass scale region , where we will find very promising values for the lfv higgs rates .
our paper is structured as follows : in section [ theory ] we describe the mssm frame which we will use in the rest of the paper , introducing the flavor mixing and the scenarios we will work with .
section [ lfvrates ] is devoted to the lfv higgs decay rates in the heavy susy particle mass limit , showing the non - decoupling behavior of these rates with @xmath0 .
we dedicate section [ lfvxsections ] to the relevant numerical results of the lfv higgs event rates at the lhc , considering the effects induced by each type of delta , @xmath53 .
this will be done taking into account the conditions of the present and future phases of the lhc , in order to estimate the number of events one could expect for each lfv higgs channel .
we will finally close the paper with the conclusions , contained in section [ conclusions ] .
in this paper we explore susy scenarios with general flavor mixing in the slepton sector , which have the same particle content as the mssm . within these scenarios ,
the most general hypothesis of flavor mixing assumes a non - diagonal mass matrix in flavor space , for both charged slepton and sneutrino sectors .
the charged slepton mass matrix is a @xmath54 matrix due to the six electroweak interaction eigenstates , @xmath55 with @xmath56 .
the inclusion , within the mssm , of only three sneutrino eigenstates , @xmath57 with @xmath58 , reduces the sneutrino mass matrix to a @xmath59 matrix . in the case in which slepton and sneutrino mass matrices were diagonal
, we would still have a tiny flavor violation induced by the pmns matrix of the neutrino sector and transmitted by the tiny neutrino yukawa couplings , but we neglect it here . the non - diagonal entries in the @xmath54 slepton matrix can be described in a model - independent way in terms of a set of dimensionless parameters @xmath60 ( @xmath61 ; @xmath62 ) , where @xmath63 stand for the possible chiralities of the lepton partners and @xmath64 indexes run over the three generations . these scenarios with general sfermion flavor mixing lead generally to larger lfv rates than in the so - called minimal flavor violation ( mfv ) scenarios , where the mixing is induced exclusively by the yukawa coupling of the corresponding fermion sector .
this statement is true for both squarks and sleptons but it is obviously of special interest in the slepton case due to the extremely small size of the lepton yukawa couplings .
hence , in the present case of slepton mixing , the @xmath60 s will provide the unique origin of lfv processes with potentially measurable rates .
one usually starts with the non - diagonal @xmath54 slepton squared mass matrix referred to the electroweak interaction basis , which we order here as ( @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 ) , and write this matrix in terms of left- and right - handed blocks @xmath71 ( @xmath61 ) , which are non - diagonal @xmath72 matrices , @xmath73 where @xmath74 with flavor indexes @xmath62 , corresponding to the first , second and third generations , respectively .
@xmath75 is the weak angle , @xmath76 is the @xmath77 gauge boson mass , @xmath78 are the lepton masses , @xmath79 with @xmath80 and @xmath81 , the two vacuum expectation values of the corresponding neutral higgs boson in the higgs @xmath82 doublets , @xmath83 and @xmath84 , and @xmath85 is the usual higgsino mass term .
it should be noted that the non diagonality in flavor comes exclusively from the soft susy - breaking parameters , which could be non vanishing for @xmath34 , namely : the masses @xmath86 for the slepton @xmath82 doublets @xmath87 , the masses @xmath88 for the slepton @xmath82 singlets @xmath89 , and the trilinear couplings , @xmath90 . in the sneutrino sector
there is , correspondingly , a one - block @xmath72 mass matrix , respect to the @xmath91 electroweak interaction basis , @xmath92 where @xmath93 it should be also noted that , due to @xmath94 gauge invariance , the same soft masses @xmath86 enter in both the slepton and sneutrino @xmath2 mass matrices . besides , in the previous equations we have neglected the neutrino mass terms , which due to their extremely small value are totally irrelevant for the present computation .
the general slepton flavor mixing is then introduced via the non - diagonal terms in the soft - breaking slepton mass matrices and trilinear coupling matrices , which are defined here as @xmath95
@xmath96 @xmath97 in all this work , for simplicity , we are assuming that all @xmath60 parameters are real .
therefore , hermiticity of @xmath98 and @xmath99 implies @xmath100 . besides , in order to avoid extremely large off - diagonal matrix entries we restrict ourselves to @xmath101 .
it is worth to have in mind for the rest of this work that our parametrization of the off - diagonal in flavor space entries in the above mass matrices is purely phenomenological and does not rely on any specific assumption about the origin of the mssm soft - mass parameters . in particular
, it should be noted that our parametrization for the @xmath3 and @xmath4 squared mass entries connecting different generations ( i.e. for @xmath34 ) assumes a similar generic form to the @xmath2 and @xmath5 entries .
for instance , @xmath102 .
this implies that our hypothesis for the trilinear off - diagonal couplings @xmath103 with @xmath34 ( as derived from eq.([v1al ] ) ) is one among other possible definitions considered in the literature . in particular , it is related to the usual assumption @xmath104 , by setting @xmath105 . here , @xmath106 , @xmath107 is the charged gauge boson mass , @xmath108 is the @xmath94 gauge coupling and @xmath20 is a typical susy mass scale .
it should be also noted that the diagonal entries in eq.([v1al ] ) have been normalized as it is usual in the literature , namely , by factorizing out the corresponding lepton yukawa coupling : @xmath109 , with @xmath110 , @xmath111 and @xmath112 .
finally , it should be mentioned that our choice in eqs.([mll ] ) , ( [ v1al ] ) and ( [ mrr ] ) is to normalize the non - diagonal in flavor entries with respect to the geometric mean of the corresponding diagonal squared soft masses .
for instance , @xmath113 the next step is to rotate the sleptons and sneutrinos from the electroweak interaction basis to the physical mass eigenstate basis , @xmath114 with @xmath115 and @xmath116 being the respective @xmath117 and @xmath72 unitary rotating matrices which yield the diagonal mass - squared matrices as follows , @xmath118 in the numerical computations of the present study we will restrict ourselves to the case where there are flavor mixings exclusively between the second and third generations of sleptons , thus we set all @xmath60 s to zero except for @xmath119 . on one hand ,
the lfv one - loop corrected higgs couplings are proportional to the heaviest lepton mass involved @xcite and , therefore , the higgs decay rates into @xmath40 are suppressed by a factor @xmath120 with respect to the @xmath121 decay rates . on the other hand ,
the related lfv radiative decay @xmath30 has a much more restrictive upper bound @xcite than @xmath29 and @xmath28 decays @xcite , and the present allowed values of the @xmath122 s would not drive us to any measurable @xmath123 rates . in order to simplify our analysis , and to reduce further the number of independent parameters
, we will focus on the following numerical study on simplified susy scenarios , where the relevant soft - mass parameters are related to a single susy mass scale , @xmath0 .
in particular we choose the following setting for the relevant mass parameters : @xmath124 where @xmath125 is a constant coefficient that we will fix in the next sections to different values , namely , @xmath126 , @xmath127 , 1 .
we also set an approximate gut relation for the gaugino masses : @xmath128 here and in the following we use a short notation for the common soft masses , namely , @xmath129 for @xmath130 , etc .
for simplicity , we have also assumed vanishing soft - trilinear couplings for the first and second generations in the slepton sector , i.e. , @xmath131 . we have checked that other choices with non - vanishing values for any of these two couplings do not alter the conclusions of this work .
the trilinear coupling for the third generation has been fixed here to the generic susy mass scale , @xmath132 . regarding the non - diagonal trilinear couplings we have also assumed a rather simple but realistic setting by relating them with the single soft susy - breaking mass scale , @xmath0 . specifically , we assume the following linear relation : @xmath133 where the new dimensionless parameters @xmath134 and @xmath135 are trivially related to the previously introduced ones @xmath136 and @xmath137 of eq.([deltas23defs ] ) by : @xmath138 it is clear from eq.([lrversuslrtilde ] ) that @xmath136 and @xmath137 scale with @xmath0 as @xmath139 . therefore ,
in the forthcoming analysis of the lfv observables , whenever the decoupling behavior of these observables with large @xmath0 be explored we will use instead the more suited parameters @xmath134 and @xmath135 , which can be kept fixed to a constant value while taking large @xmath0 values . concerning the size of the flavor violating trilinear couplings that are of relevance here , @xmath140 and @xmath141
, there are well - known theoretical upper bounds that arise from vacuum stability .
if any of these trilinear couplings is too large , the mssm scalar potential develops a charge and/or color breaking ( ccb ) minimum deeper than the standard - model - like ( sml ) local minimum or an unbounded from below ( ufb ) direction in the field space @xcite .
then the requirement of the absence of these dangerous ccb minima or ufb directions implies specific upper bounds on the size of the non - diagonal trilinear couplings , and consequently on the size of the flavor changing deltas . for the case of interest here the upper bound from stability can be written simply as @xcite : @xmath142 and similarly for @xmath141 . here
, @xmath143 is the yukawa coupling of the tau lepton , and the squared soft mass @xmath144 can be written as : @xmath145 in our simplified scenarios for the slepton , gaugino and higgs sectors , with just three mssm input parameters , @xmath0 , @xmath146 , and @xmath147 , and by considering eqs.([trilinear23 ] ) and ( [ lrversuslrtilde ] ) , the previous bound implies in turn the following bound on @xmath136 , and correspondingly on @xmath134 ( and similar bounds for @xmath148 ) : @xmath149 for example , if we take @xmath150 tev we get upper bounds for @xmath151 of @xmath152 in the low @xmath146 region close to 5 , and of @xmath153 in the large @xmath146 region close to 50 .
these correspond to an upper bound on @xmath154 of @xmath155 that is nearly independent on @xmath146 and it gets weaker for larger @xmath0 values due to the scaling factor ( @xmath156 ) in eq.([delta23bounds ] ) .
however , the reliability of these bounds have been questioned in the literature because of the fact that the existence of deeper minima than the sml local minimum does not necessarily imply a problem whenever the lifetime of this false sml vacuum is sufficiently long . in this later case ,
other theoretical upper bounds based on metastability then apply . indeed , by demanding that the lifetime of the whole observable universe staying at the sml vacuum be longer than the age of the universe the constraints on the flavor changing deltas
get much more relaxed @xcite . according to @xcite the upper bounds on @xmath136 from metastability ,
in contrast to the limits from stability , turn out to be independent on the yukawa coupling , they do not decouple for asymptotically large @xmath157 and they are dependent on @xmath146 .
for instance , for @xmath158 and @xmath159 tev the metastability limit on @xmath136 gets weaker than its stability bound by a factor of 40 , whereas for @xmath160 and @xmath159 tev it is weaker by a smaller factor of 4 , leading to approximate upper bounds of @xmath161 and @xmath162 respectively .
this translates into an upper bound of about @xmath163 for @xmath164 and @xmath165 tev .
the numerical estimates of these metastability bounds are done considering each delta separately ( i.e setting the other deltas to zero value ) and for a specific assumption of the relevant euclidean action providing the decay probability via tunneling of the metastable vacuum into the global minimum .
changing the input value for the euclidean action can increase notably the maximum allowed value up to almost doubling it , leading to roughly @xmath166 .
the effects on these bounds from switching on more than one delta at the same time have not been considered yet in the literature but they could relevantly modify these bounds . for the present work , and given the uncertainty in all these estimates of the upper bounds from stability and metastability arguments , we will consider a rather generous interval when performing the numerical estimates of branching ratios and event rates . concretely
we will choose several examples for @xmath134 of very different size that will be taken within the wide interval @xmath167 .
this corresponds to @xmath168 for the particular values of @xmath169 5 tev and @xmath170 40 . on the other hand ,
the values of the soft masses of the squark sector are irrelevant for lfv processes , except in the present case of lfv mssm higgs bosons decays where these parameters enter in the prediction of the radiatively corrected higgs boson masses . since we want to identify the discovered boson with the lightest mssm higgs boson , we will set these parameters to values which give a prediction of @xmath171 compatible with the lhc data . specifically , we fix them to the particular values @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 5 tev , which we have checked provide a value for @xmath171 that lies within the lhc - favored range [ 121 gev , 127 gev ] for all the mssm parameter space considered here . in summary ,
the input parameters of our simplified susy scenarios are : the mass of the pseudoscalar higgs boson , @xmath147 , the ratio of the two higgs vacuum expectation values , @xmath146 , the generic susy mass scale , @xmath0 , and the four delta parameters , @xmath9 , @xmath12 , @xmath10 and @xmath11 ( or @xmath134 and @xmath177 alternatively to the two latter ) , which we vary within the following intervals : * 200 gev @xmath178 1000 gev , * 1 @xmath179 60 , * 0.5 tev @xmath180 10 tev , * @xmath181 @xmath9 , @xmath12 @xmath182 , * @xmath183 @xmath184 , @xmath185 @xmath186 , + ( or , equivalently , @xmath187 for @xmath169 5 tev and @xmath170 40 ) .
in this section we study the behavior of the radiative corrections from susy loops to the lfv neutral higgs bosons decays in the heavy susy particle mass limit and compare them with the case of lfv radiative lepton decays . for the forthcoming estimates of the lfv higgs decay rates
, we use the complete one - loop formulas and the full set of diagrams contributing to the @xmath188 and @xmath189 partial decay widths , with @xmath34 , within the mssm , which are written in terms of the mass eigenvalues for all the involved mssm sparticles , including the physical slepton and sneutrino masses , @xmath190 and @xmath191 , and the rotation matrices @xmath115 and @xmath116 of eqs .
( [ sleptons ] ) and ( [ sneutrinos ] ) .
we take these general one - loop formulas from @xcite and emphasize that they are valid for the general slepton mixing case considered here , with all the mixing effects from the @xmath47 s being transmitted to the lfv higgs decay rates via the physical slepton and sneutrino masses and their corresponding rotations . the off - diagonal trilinear couplings in eq.([v1al ] ) also enter into this computation of the lfv higgs decay rates . for a more detailed discussion about these analytical results
we refer the reader to @xcite . it should be also noted that , since we are assuming real @xmath47 s , the predictions for br@xmath192 and the cp - conjugate br@xmath193 are the same .
thus , we will perform our estimates for just one of them and will denote this rate generically by br@xmath194 . obviously , in the case that these two channels can not be differentiated experimentally one should then add the two contributions to the total final number of events .
however , for the present computation we do not sum them and report instead the results for @xmath195 , meaning that they are valid for any of the two cases .
as said above , we focus only on @xmath13 , @xmath27 and @xmath8 decay channels and consider the constraints imposed over the parameter space by the current upper bound on the related lfv radiative decay @xmath28 @xcite .
the susy mass spectra are computed numerically with the code spheno @xcite .
the slepton and sneutrino spectra are computed from an additional subroutine that we have implemented into spheno in order to include our parametrization of slepton mixing given by the @xmath47 s .
the lfv decay rates are computed with our private fortran code in which we have implemented the complete one - loop formulas for the lfv partial higgs decay widths of @xcite and the complete one - loop formulas for the lfv radiative @xmath23 decay widths which we take from @xcite .
note that these latter formulas for the @xmath28 decays are also written in terms of the physical sparticle eigenvalues and eigenstates and do not neglect any of the lepton masses .
the mass spectrum of the mssm higgs sector , with two - loop corrections included , and their corresponding total widths are calculated by means of the code feynhiggs @xcite . [ cols="^,^ " , ] finally , the results for the @xmath13 and @xmath8 channels in the double @xmath2 and @xmath3 mixing case are summarized in figures [ neventsh - matanb - deltalldeltalr ] and [ neventsa - matanb - deltalldeltalr ] , respectively . the predictions for the contour lines of @xmath13 event rates ( figure [ neventsh - matanb - deltalldeltalr ] ) show a clear different pattern than in the previous cases of single deltas . in both lhc phases ,
we achieve an increase in the number of events respect to the single @xmath2 and @xmath3 ( and/or @xmath4 ) mixing cases .
the most interesting numbers are at the lower part of these plots with @xmath196 .
we find as large as 75 lfv events in the present phase of the lhc for very low values of @xmath197 2 and @xmath198 600 gev . in the future lhc phase we predict up to 750 events , for @xmath199 2 and @xmath198 450 gev .
it should be also noted that these conclusions apply to both choices for the @xmath200 mixings of -5 and + 5 , as can be seen in figure [ neventsh - matanb - deltalldeltalr ] .
the only notable differences that we find between the results of these two cases are in the slight different patterns of the contour lines , signalling a small different sensitivity to @xmath147 and/or @xmath146 . also one can appreciate in these plots that one gets a bit lower rates for + 5 than for -5 , in agreement with our previous findings reported in figure [ brs - deltatildelrrl ] the number of events for the @xmath8 channel in the double @xmath2 and @xmath3 mixing case are displayed in figure [ neventsa - matanb - deltalldeltalr ] .
the @xmath2 mixing is set here to 0.9 and the @xmath201 mixings are fixed to -5 .
as expected , the predicted event rates increase as @xmath146 grows and are reduced as @xmath147 gets bigger , due to the suppression in the production cross section of a heavy pseudoscalar higgs boson .
however , for the present lhc phase one can not say much about this channel , since all the @xmath202 regions which could produce a relevant number of lfv events are excluded at present by cms searches
. we find at the most one event in the tiny low left corner of the allowed region in this @xmath203 plot .
in contrast , the predictions in the future phase of the lhc are more promising .
we predict up to about 5 lfv events for large values of @xmath147 and @xmath146 and up to 10 in the low @xmath146 region with @xmath204 200 gev .
similar conclusions are found for the case of positive @xmath3=+5 mixing ( not shown ) .
the shape of the contour lines in this case are slightly modified at low @xmath146 but with no relevant implications in terms of event rates .
the discovery of a new higgs - like particle at the lhc is concentrating a lot of efforts in studying its properties , couplings and decays , in order to investigate if there is new physics behind it . in that sense ,
lepton flavor violating higgs decays are of special interest , since they would clearly imply the existence of physics beyond the sm .
we have discussed in this paper the possibility of obtaining sizeable lfv higgs rates , induced by heavy susy , and detectable at the lhc . in particular , we have studied in detail the most interesting lfv decays of the three neutral mssm higgs bosons : @xmath13 , @xmath27 and @xmath8 .
we have shown that these three channels present a non - decoupling behavior with @xmath0 .
on one hand , independently if @xmath9 , @xmath12 , @xmath10 or @xmath11 are the responsible for the intergenerational mixing in the slepton sector , these lfv higgs decay rates remain constant with @xmath0 at large @xmath205 2 tev . on the other hand , the related lfv radiative decay ,
@xmath28 , manifests a fast decoupling behavior with @xmath0 and its rates are very suppressed for large values of the susy scale . in this work ,
we have taken advantage of these two remarkable different behaviors with @xmath0 in order to reach sizeable lfv higgs branching ratios which are yet allowed by the present @xmath28 upper bound . from our detailed analysis of the lfv
higgs decays , we have learned that the single @xmath206 mixing of @xmath5 type in the slepton sector can not by itself provide sufficiently large rates which could be measurable at the lhc .
the situation ameliorates slightly if we consider the single @xmath2 mixing case , with the largest rates at the large @xmath146 region and amounting up to about 5 lfv events for the @xmath207 channels at the future phase of the lhc .
we find that the mixing parameter @xmath10 ( and @xmath11 ) is the most relevant one and even when acting as single mixing parameter already gives rise to sizeable and allowed by data lfv higgs - mediated event rates for sufficiently large values of @xmath208 tev , where the lfv radiative @xmath28 rates are suppressed below its present experimental upper bound . in this single @xmath3-mixing case , we also find that the most promising channel is by far the @xmath13 decay for which we predict up to 50 events for low @xmath209 in the present phase of the lhc . regarding the @xmath207 channels ,
no events are expected in the present lhc phase if there is single @xmath3 ( and/or @xmath4 ) mixing .
the situation improves noticeably if one considers the future phase of the lhc . in this case , with a total integrated luminosity of 100 fb@xmath210 , we predict hundreds of lfv events from the lightest higgs boson decay into @xmath39 , as much as @xmath211 events for very low values of @xmath146 . for the lfv heavy higgs bosons
decays the expectations are lower , but also increase in comparison with the present lhc phase , and a few events could be obtained depending on the values of @xmath147 , @xmath146 and @xmath0 .
finally , by considering double @xmath2 and @xmath3 mixings we obtain the most interesting situation , since on one hand the rates are slightly increased with respect to the single @xmath3 mixing case and , on the other hand , the slight change in the sensitivity to @xmath146 makes that larger values of this parameter than in the single @xmath3 mixing case give rise also to sizeable event rates .
for instance , one can get in this double mixing case a few events at the present stage of the lhc even for moderate @xmath212 and large @xmath213 500 gev . in the future lhc phase the reach to larger @xmath209 values increases and one gets some event even at very large @xmath214 and @xmath215 800 gev .
the largest rates found are in any case for @xmath13 and are clearly localized at the low @xmath146 region where we predict for the present lhc phase up to about 75 events , and up to about 750 lfv events for the future lhc phase . in the future lhc phase , we get about 10 events at the most for the @xmath207 channels in the low @xmath209 region and 5 events at the most in the high @xmath209 region . as a final comment ,
it is worth recalling that all the rates presented in this work are doubled if one adds the decay events for the two possible final states , @xmath216 and @xmath217 .
the encouraging results presented along this work strongly suggest that a dedicated search for the proposed lfv higgs decays is extremely worthwhile and we believe that it should be further studied by the experiments at the lhc
. if the susy mass scale is too heavy , as the present experiments are pointing out , and the susy particles can not be directly reachable at the present or next future lhc energies , our proposal for lfv higgs decays could provide an unique window to explore new physics and to find some hint of very heavy susy at the lhc .
e.a . thanks alejandro szynkman for fruitful discussions and a careful reading of the manuscript .
m.j.h thanks andreas crivellin for a critical reading of a preliminary version of this paper and for some interesting comments and discussions .
m.j.h also thanks alberto casas for his clarifying comments on the general bounds from vacuum instabilities , their domain of applicability and the way out from these bounds .
the authors also thank sven heinemeyer for helpful information about the code feynhiggs .
is financially supported by a micinn postdoctoral fellowship ( spain ) , under grant no .
fi-2010 - 0041 , and partially by anpcyt ( argentina ) under grant no .
pict - prh 2009 - 0054 and by conicet ( argentina ) pip-2011 .
e.a . thanks iflp - conicet for hospitality and support . the work of m.h . and
was partially supported by cicyt ( grant fpa2009 - 09017 ) , the comunidad de madrid project hephacos , s2009/esp-1473 , the european union fp7 itn invisibles ( marie curie actions , pitn - ga-2011 - 289442 ) and also by the spanish consolider - ingenio 2010 programme cpan ( csd2007 - 00042 ) .
m. t. dova , _ susy searches at lhc _ , ix latin american symposium on high energy physics , silafae 2012 ; j. marrouche , _
search for 3rd generation and gluino induced susy production _ ,
xlviiith rencontres de moriond ew 2013 ; m. verducci , _ search for ewkino production and long - lived particles at the lhc _ , xlviiith rencontres de moriond ew 2013 ; j. maurer , _
search for supersymmetry in events with same - sign dileptons , jets and missing transverse momentum with the atlas detector _ , xlviiith rencontres de moriond ew 2013 ; s. folgueras , _ search for new physics using events with two same - sign isolated leptons in the final state _ , xlviiith rencontres de moriond ew 2013 ; a. favareto , _ third generation susy searches at the lhc _ , xlviiith rencontres de moriond qcd 2013
; a. mann , _ susy searches for inclusive squark and gluino production at the lhc _ , xlviiith rencontres de moriond qcd 2013 ; p. everaerts , _ susy searches for ewk production of gauginos and sleptons at the lhc _ , xlviiith rencontres de moriond qcd 2013
. g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] . for recent works oriented to heavy susy searches at the lhc ,
see for instance : l. j. hall , y. nomura and s. shirai , jhep * 1301 * ( 2013 ) 036 [ arxiv:1210.2395 [ hep - ph ] ] ; j. unwin , phys .
d * 86 * ( 2012 ) 095002 [ arxiv:1210.4936 [ hep - ph ] ] ; e. arganda , j. l. diaz - cruz and a. szynkman , eur . phys .
j. c * 73 * ( 2013 ) 2384 [ arxiv:1211.0163 [ hep - ph ] ] ; n. arkani - hamed , a. gupta , d. e. kaplan , n. weiner and t. zorawski , arxiv:1212.6971 [ hep - ph ] ; e. arganda , j. l. diaz - cruz and a. szynkman , phys . lett .
b * 722 * ( 2013 ) 100 [ arxiv:1301.0708 [ hep - ph ] ] ; d. mckeen , m. pospelov and a. ritz , arxiv:1303.1172 [ hep - ph ] ; a. djouadi and j. quevillon , arxiv:1304.1787 [ hep - ph ] . h. e. haber , m. j. herrero , h. e. logan , s. penaranda , s. rigolin and d. temes , phys .
d * 63 * ( 2001 ) 055004 [ hep - ph/0007006 ] .
a. dobado , m. j. herrero and d. temes , phys . rev .
d * 65 * ( 2002 ) 075023 [ hep - ph/0107147 ] .
a. m. curiel , m. j. herrero and d. temes , phys .
d * 67 * ( 2003 ) 075008 [ hep - ph/0210335 ] .
a. m. curiel , m. j. herrero , w. hollik , f. merz and s. penaranda , phys .
d * 69 * ( 2004 ) 075009 [ hep - ph/0312135 ] .
e. arganda , a. m. curiel , m. j. herrero and d. temes , phys .
d * 71 * ( 2005 ) 035011 [ hep - ph/0407302 ] .
t. appelquist and j. carazzone , phys .
d * 11 * ( 1975 ) 2856 .
m. j. herrero , j. portoles and a. m. rodriguez - sanchez , phys .
d * 80 * ( 2009 ) 015023 [ arxiv:0903.5151 [ hep - ph ] ] .
e. arganda , m. j. herrero and a. m. teixeira , jhep * 0710 * ( 2007 ) 104 [ arxiv:0707.2955 [ hep - ph ] ] .
a. abada , d. das and c. weiland , jhep * 1203 * ( 2012 ) 100 [ arxiv:1111.5836 [ hep - ph ] ] ; a. abada , d. das , a. vicente and c. weiland , jhep * 1209 * ( 2012 ) 015 [ arxiv:1206.6497 [ hep - ph ] ] .
a. crivellin , phys .
d * 83 * ( 2011 ) 056001 [ arxiv:1012.4840 [ hep - ph ] ] ; a. crivellin , l. hofer and j. rosiek , jhep * 1107 * ( 2011 ) 017 [ arxiv:1103.4272 [ hep - ph ] ] .
n. liu , l. wu , p. w. wu and j. m. yang , jhep * 1301 * ( 2013 ) 161 [ arxiv:1208.3413 [ hep - ph ] ] .
m. raidal , a. van der schaaf , i. bigi , m. l. mangano , y. k. semertzidis , s. abel , s. albino and s. antusch _ et al .
_ , eur .
j. c * 57 * ( 2008 ) 13 [ arxiv:0801.1826 [ hep - ph ] ] .
j. l. diaz - cruz and j. j. toscano , phys .
d * 62 * ( 2000 ) 116005 [ hep - ph/9910233 ] ; j. l. diaz - cruz , jhep * 0305 * ( 2003 ) 036 [ hep - ph/0207030 ] ; a. brignole and a. rossi , phys . lett .
b * 566 * ( 2003 ) 217 [ hep - ph/0304081 ] ; a. brignole and a. rossi , nucl .
b * 701 * ( 2004 ) 3 [ hep - ph/0404211 ] ; s. kanemura , k. matsuda , t. ota , t. shindou , e. takasugi , k. tsumura and , phys .
b * 599 * ( 2004 ) 83 [ hep - ph/0406316 ] ; j. k. parry , nucl .
b * 760 * ( 2007 ) 38 [ hep - ph/0510305 ] ; j. l. diaz - cruz , d. k. ghosh and s. moretti , phys .
b * 679 * ( 2009 ) 376 [ arxiv:0809.5158 [ hep - ph ] ] ; p.
t. giang , l. t. hue , d. t. huong and h. n. long , nucl .
b * 864 * ( 2012 ) 85 [ arxiv:1204.2902 [ hep - ph ] ] ; a. arhrib , y. cheng and o. c. w. kong , europhys .
* 101 * ( 2013 ) 31003 [ arxiv:1208.4669 [ hep - ph ] ] ; a. arhrib , y. cheng and o. c. w. kong , phys .
d * 87 * ( 2013 ) 015025 [ arxiv:1210.8241 [ hep - ph ] ] .
g. blankenburg , j. ellis and g. isidori , phys .
b * 712 * ( 2012 ) 386 [ arxiv:1202.5704 [ hep - ph ] ] .
j. adam _ et al . _
[ meg collaboration ] , arxiv:1303.0754 [ hep - ex ] .
r. harnik , j. kopp and j. zupan , jhep * 1303 * ( 2013 ) 026 [ arxiv:1209.1397 [ hep - ph ] ] ; s. davidson and p. verdier , phys .
d * 86 * ( 2012 ) 111701 [ arxiv:1211.1248 [ hep - ph ] ] .
b. aubert _ et al . _
[ babar collaboration ] , phys .
* 104 * ( 2010 )
021802 [ arxiv:0908.2381 [ hep - ex ] ] . m. arana - catania , s. heinemeyer and m. j. herrero , phys .
d * 88 * ( 2013 ) 015026 [ arxiv:1304.2783 [ hep - ph ] ] .
j. a. casas and s. dimopoulos , phys .
b * 387 * ( 1996 ) 107 [ hep - ph/9606237 ] .
j. -h .
park , phys .
d * 83 * ( 2011 ) 055015 [ arxiv:1011.4939 [ hep - ph ] ] .
w. porod , comput .
commun .
* 153 * ( 2003 ) 275 [ hep - ph/0301101 ] ; w. porod and f. staub , comput .
commun .
* 183 * ( 2012 ) 2458 [ arxiv:1104.1573 [ hep - ph ] ] .
s. heinemeyer , w. hollik and g. weiglein , comput .
* 124 * ( 2000 ) 76 [ hep - ph/9812320 ] ; s. heinemeyer , w. hollik and g. weiglein , eur .
j. c * 9 * ( 1999 ) 343 [ hep - ph/9812472 ] ; g. degrassi , s. heinemeyer , w. hollik , p. slavich and g. weiglein , eur .
j. c * 28 * , ( 2003)133 [ hep - ph/0212020 ] ; m. frank , t. hahn , s. heinemeyer , w. hollik , h. rzehak and g. weiglein , jhep * 0702 * ( 2007 ) 047 [ hep - ph/0611326 ] .
s. chatrchyan _ et al .
_ [ cms collaboration ] , cms - pas - hig-12 - 050 . | the recent discovery of a sm - like higgs boson at the lhc , with a mass around 125 - 126 gev , together with the absence of results in the direct searches for supersymmetry , is pushing the susy scale ( @xmath0 ) into the multi - tev range .
this discouraging situation from a low - energy susy point of view has its counterpart in indirect susy observables which present a non - decoupling behavior with @xmath0 .
this is the case of the one - loop lepton flavor violating higgs decay rates induced by susy , which are shown here to remain constant as @xmath0 grows , for large @xmath1 2 tev values and for all classes of intergenerational mixing in the slepton sector , @xmath2 , @xmath3 , @xmath4 and @xmath5 . in this work
we focus on the lfv decays of the three neutral mssm higgs bosons @xmath6 ,
@xmath7 , @xmath8 , considering the four types of slepton mixing ( @xmath9 , @xmath10 , @xmath11 , @xmath12 ) , and show that all the three channels could be measurable at the lhc , being @xmath13 the most promising one , with up to about hundred of events expected with the current lhc center - of - mass energy and luminosity .
the most promising predictions for the future lhc stage are also included .
ift - uam / csic-13 - 024 + m. arana - catania@xmath14 , e. arganda@xmath15 , m.j .
herrero@xmath14 _ @xmath16departamento de fsica terica and instituto de fsica terica , ift - uam / csic + universidad autnoma de madrid , cantoblanco , madrid , spain _
@xmath17iflp , conicet - dpto .
de fsica , universidad nacional de la plata , + c.c .
67 , 1900 la plata , argentina |
in double - lobed radio sources with single - sided jets more rapid depolarization has been observed with increasing wavelength in the lobe on the side with no jet ( the laing - garrington effect ) .
important in the explanation of this effect is the question of whether the jet one - sidedness is _ intrinsic _ or due to _
relativistic beaming .
_ if it is the latter , and both lobes are still being actively fed , no significant internal differences would be expected in the lobes and the depolarization asymmetry must then originate in an external faraday screen . furthermore , in this model the lobe on the side with no jet will be receding and its radiation will be expected to be depolarized to a greater degree if it is viewed through more of a faraday screen associated with a halo surrounding the parent galaxy @xcite .
this model works best if the jet viewing angles are small so that the path length through the screen to the more distant , more highly depolarized lobe , will be a maximum .
small jet viewing angles can also be a natural result if relativistic beaming plays a significant role .
_ _ _ _ on the other hand , if the jet one - sidedness is intrinsic there is an equal chance that the lobe on the side containing the visible jet is moving away from us , which means that the intervening faraday screen model no longer works and the depolarization asymmetry must then originate inside the lobe itself , with the outer regions of the lobe acting as a faraday screen .
the fact that no jet is seen on one side implies that the jet is either much weaker or turned off completely . in this case
high energy particles are no longer being pumped into its associated radio lobe at the same rate and the conditions inside that lobe will be expected to be different than in the lobe that is still being actively fed , which would be required if the depolarization is internal .
evidence that the lobe dissipates with time when it is no longer being fed is easily seen in the trailing edges of the lobes in many of the double - lobed sources , such as 3c66b or 3c 296 @xcite .
it might also be expected that if one of the lobes is no longer being fed its outward motion will slow down , with the result that it will be located closer to the central compact object as is often seen . however , this observation can also be explained in the beaming model by differences in light travel time . before proceeding further
it is important to note that there is a third type of faraday screen that can add some confusion to this analysis .
it has been demonstrated that there can sometimes be weak faraday screens present that cover the entire source @xcite and these are most easily seen in those sources with two jets ( * ? ? ?
* see for example ) where both lobes are still being fed equally and the asymmetric depolarizing component we are discussing here is either very low or not present .
these screens can be due either to extended material in the inner regions of clusters , material in our own galaxy , or intergalactic material , but they should not be confused with the asymmetric faraday screen proposed in the l - g sources which covers only one lobe .
because it was assumed by @xcite that the jet one - sidedness was likely due to relativistic beaming it was concluded that the laing - garrington effect is more likely to originate in an intervening faraday screen that covers the receding lobe and this explanation has endured .
in fact , in most investigations since then it has also been assumed without proof that the one - sidedness is due to relativistic beaming . as an example , @xcite set out to see if the l - g effect was present in low - luminosity radio galaxies .
it was thought that the jets in these objects would be less likely to be relativistic and they argued that a positive detection for the l - g effect would prove that they were .
however , that argument is only valid if the relativistic beaming model is the correct explanation for the jet one - sidedness , or if it could somehow rule out the internal explanation .
their analysis included two samples ; one with single - sided jets and one with two - sided jets .
only the sample with one - sided jets showed a significant l - g effect .
however , although their result is consistent with the l - g explanation if the one - sidedness is due to beaming , it is also consistent with the internal depolarization explanation if the one - sidedness is intrinsic . in the two - sided case
both lobes are still being fed so no differences in the depolarization would be expected in either the beaming or intrinsic models . in the single - sided case
the depolarization asymmetry can be explained in the intrinsic case by the fact that only one lobe is being fed .
these authors also found that the radio galaxies with strong radio cores showed a more pronounced asymmetry in depolarization .
this is also what would be expected in the internal depolarization model where the depolarization is tied directly to the rate at which the lobe is being fed .
it is thus apparent that , although their results may be consistent with the l - g explanation , they are also not inconsistent with the internal depolarization explanation .
if the jets are intrinsically one - sided , the lobes are being excited in different ways so different internal faraday depths might be expected @xcite .
these authors found differences in the spectra between the two lobes and suggested that a possible explanation might be due to differences in the physical conditions in the lobes .
they suggested that possible differences might arise in the rates of deposition of energy from the nucleus , and this would certainly be the case if one jet had turned off .
they concluded that there must be differences in the physical parameters of the two lobes which influence both spectrum and depolarization .
expected for random orientations , where @xmath0 is the angle between the jet and the line - of - sight as shown .
see text for an explanation of r@xmath1 and r@xmath2 .
[ fig1],width=340 ] the laing - garrington effect was also looked at closely by @xcite .
they point out that if this effect is due to an intervening faraday screen of the nature proposed by @xcite , the viability of the explanation demands that the faraday screen somehow maintains a size that is within a factor of two of the total extent of the expanding radio source , which may be difficult to explain . in fig 1 no difference in depolarization
will be expected if the radius of the faraday screen is r @xmath3 r@xmath1 .
the biggest effect will occur for r = r@xmath2 , but even then the two path lengths will not differ significantly .
the situation gets worse again for r @xmath4 r@xmath2 . for very large r , as would be the case for screens that cover the entire central regions of clusters the difference in path length to the two lobes would be negligible , as would be the effect .
it seems unlikely that the halo would just happen to be the optimum size r = r@xmath2 .
furthermore , there is no solid evidence that the proposed faraday screen ( galactic halo ) even extends out as far as the lobes , which can be up to 1500 kpc in projected linear size .
there is clearly no evidence that it does in the map of 3c296 @xcite where the central elliptical galaxy , and its optically bright halo , are superimposed on top of the 20 cm radio map . here
both lobes lie well outside the optical halo of the galaxy with neither being viewed through it . because of these problems @xcite suggested that the superdisks seen in some radio galaxies
might be an alternative to the faraday screen envisioned by @xcite that would remove the need to postulate a magnetoionic corona around every high - z quasar whose dimensions somehow remain closely matched to the steadily growing radio source . however , the superdisks they discuss have so far only been seen in radio galaxies and whether or not they even exist in quasars is unclear .
they do not appear to be common even in radio galaxies .
although the flow in the jet may start out relativistically at parsec scales , there is much evidence from radio maps that the flow slows down considerably on kiloparsec ( kpc ) scales @xcite .
it is these large kpc - scale jets that the l - g investigations have used to conclude that beaming is present . unlike the pc - scale jet motion near the core that can easily be measured with the vlba to milliarsec ( mas ) accuracy
, no direct method has been found to measure the motions of material in the kpc - scale jets to mas accuracy .
this is because in this case the vla must be used and detecting mas motions with an arcsec beamwidth is next to impossible .
although @xcite have claimed to have made such a measurement in the kpc jet of 3c120 , this measurement has not been confirmed .
but it is well known that radio galaxies , especially the closer ones , do not show the highly relativistic motions in their pc - scale jets that are seen in many quasars @xcite(see their fig 6 ) .
the l - g explanation works best if the jet viewing angles are small so that the path difference to the two lobes is greatest .
but there are several things that need to be kept in mind when dealing with sources with giant radio lobes . for these sources the dominant component of the flux density in the early surveys , such as the 3c , 4c , or parkes surveys , that were carried out at low frequencies , more than 90 percent of the flux comes from the lobes and their hotspots @xcite . because the lobes are separating from the core at relatively slow speeds from 0.01c to 0.03c @xcite , their radiation can not be doppler boosted .
_ by definition , the l - g sources all fall into the lobe - dominant category . _
the percentage of un - beamed radiation is even higher if the core radiation is not beamed as has recently been claimed @xcite .
when the dominant component of the radiation is not beamed there can be no selection effect that will preferentially pick up those sources with small jet viewing angles @xcite .
it was also pointed out by @xcite that using low frequencies in source surveys , as was the case for the early source - finding surveys that found most of the radio - loud sources , provides a unique way of obtaining an _ orientation unbiased _ sample of agns . when there are no selection effects to preferentially pick up sources with specific viewing angles , _ all _ sources above the detection limit in this category ( with giant lobes ) will be detected . in this case
the number distribution of their orientations will be random ( sin@xmath0 ) as shown in fig 2 of @xcite . here , and throughout this paper , @xmath0 is the angle between the jet and the line - of - sight . for a random distribution
, @xmath5 will have viewing angles above @xmath6 .
less than 1 percent will have viewing angles below @xmath7 .
this means that very few of the detected sources will have the small viewing angles required for the l - g model to work effectively .
_ _ _ _ _ _ the situation may be different for core - dominant quasars .
most astronomers who study superluminal motion argue that in searches , doppler boosting will preferentially pick up those sources with their jets pointing towards us @xcite , although some have claimed otherwise @xcite . however , whether or not this is the case for the core - dominant sources ( quasars ) , these are not being examined here . here
we are looking at l - g sources .
these are sources that by definition all have giant radio lobes and have fanaroff - riley classifications fri and frii @xcite . for the l - g radio galaxies being considered here , those sources with their jets pointed in our direction will not have been @xmath8 selected in the original surveys .
what is also important to realize here is that when giant lobes are present the jet one - sidedness can not be an indication of relativistic beaming since the jet component is too small to be able to change the distribution from a random one .
this means that , _ essentially the same viewing angle distribution would be obtained regardless of whether the jet strength was intrinsic or beamed .
this , in lobe - dominant radio galaxies , makes it very difficult for the l - g model to work even if the kpc jet is relativistic . _
several recent investigations have been carried out to determine how the flow decelerates in the jet as it moves outward @xcite .
it was concluded @xcite that jets in fri radio galaxies , although initially relativistic with velocities from 0.8c to 0.9c at parsec scales , decelerate rapidly in the first few kpc to non - relativistic velocities between 0.4c and 0.1c beyond the flaring region . since in most cases only about the first ten percent of the kpc jet was studied by these investigators , and it is obvious from the radio maps that most continue to decelerate , these values are likely to be upper limits .
furthermore , @xcite also found that beyond a few kpc the jet flow decelerates to 0.1c .
it is therefore assumed here that , at least for fri sources , when the entire kpc jet is considered the average outward speed ( beyond @xmath9 kpc ) is likely to be no more than 0.1c .
this would be completely consistent with the outward motions of the lobes which have been shown to be close to @xmath10 in both fri and frii sources @xcite .
for these reasons we have assumed in this study that at least for the low luminosity , fri galaxies studied here that the kpc flow speed is likely to be close to 0.1c in the outer jet .
values by beaming effects plotted as a function of jet viewing angle _ i. _ the horizontal line at @xmath11 = 0.1 represents the flow in the kpc jet found for fri sources .
clearly at this flow rate even an asymmetry as low as s@xmath12 = 2 can not be completely explained by beaming even for small viewing angles .
see text for further discussion .
[ fig2]__,width=340 ]
for a given jet asymmetry ratio ( s@xmath12 = s@xmath13/s@xmath14 ) it is possible to calculate how the jet flow speed must increase as a function of viewing angle @xmath0 from the relation s@xmath13/s@xmath14 = [ ( 1+@xmath15)/(1-@xmath15)]@xmath16 here s@xmath13 and s@xmath14 are the fluxes in the jet and counterjet respectively , @xmath11 is the jet velocity in units of c , and @xmath17 is the spectral index @xcite . in fig 2 , @xmath11 is plotted against the viewing angle @xmath0 assuming several values for s@xmath12 .
if the jet flow is assumed to be 0.1c as shown by the solid line in fig 2 , the maximum jet asymmetry that can be produced by beaming effects is less than s@xmath18 , even for very small jet viewing angles .
although a kpc jet speed of 0.1c in these sources has not been proven conclusively , it has been reported by several investigators to slow to close to this value in the outer jet @xcite . for a jet flow speed of 0.9c , an asymmetry of s@xmath12 = 24
can not be produced by beaming effects if the jet viewing angle is greater than @xmath19 .
but as discussed above , although the flow may start out close to 0.9c when it first spews out of the core , it slows down significantly in the outer ( kpc ) jet where for fri sources it may have mean values closer to 0.1c .
this slows further to speeds near 0.02c @xmath20 in the lobes @xcite . , versus depolarization ratio for low - luminosity sources from @xcite .
[ fig3],width=340 ] variations in the spectral index of @xmath210.3 produced small shifts in the dashed curves in fig 2 but these were nowhere near big enough to affect these conclusions .
since the outer - jet flow value of 0.1c that we have assumed here has been obtained from low - luminosity fri sources it is necessary to use these sources for this investigation . in fig 3 the kpc jet asymmetry ratio , found by @xcite for the single - sided low - luminosity fri sources studied by @xcite ,
is plotted versus their corresponding jet depolarization ratio .
the up arrows indicate those sources with lower limits .
sources that show the l - g effect must lie to the right of the vertical dashed line and as reported by @xcite these one - sided sources all appear to show this effect . however , if the outer jet flow in low - luminosity sources is assumed to slow to @xmath220.1c beyond a few kpc , as has been shown to be the case by several investigators , from fig 2 relativistic beaming can only explain those jet asymmetries that fall below the solid horizontal line in fig 3 .
all of the asymmetry that lies above this line must then be intrinsic . from fig 3
it can be seen that most of the jet asymmetry in these low - luminosity sources must be intrinsic if our assumption of an outer jet speed near 0.1c is correct .
these results argue strongly that when the asymmetry in the kpc - scale jet is large it must be mostly intrinsic and not due to relativistic beaming .
this in turn implies that the l - g effect in these sources originates inside the lobe and not in an external faraday screen .
this result also allows for much larger viewing angles , which is more consistent with that predicted for a random distribution .
is the angle between the jet and the line - of - sight .
[ fig4],width=321 ] although the outflow speed in the jets of frii sources is not as accurately known it seems unlikely that the l - g explanation would differ between the two types , and this would require that it be similar to that of the fri sources .
however , there are other reasons why this is likely to be the case .
first , it has already been found that the outward motion of the lobes in frii sources is similar to that seen in fri sources @xcite , second , it has already been found that the viewing angles of frii galaxies may be even larger than those of fri galaxies .
recently @xcite have estimated the jet inclination angles @xmath0 in several frii radio galaxies selected from a larger sample to cover a redshift range from 1 to 5.2 .
they estimate the inclination angle using the core dominance factor r , which measures the relative strengths of the core and inner jet flux to the extended flux component .
this should work well for the lobe - dominant sources where the core flux can be referenced against a much stronger extended component .
whether it works as well for core - dominant sources ( quasars ) is questionable since the extended flux component is much weaker and may vary significantly from source to source .
their results show that most of the sources in their sample that are classified as frii sources by ned have jet inclinations relative to the line - of - sight near 85 degrees .
this is as large , or larger , than expected for fri viewing angles , but it should not be unexpected since these sources , like the l - g sources being discussed here , all have giant lobes containing most of the flux .
fig 4 shows the jet viewing angle @xmath0 distribution for the galaxies as measured by @xcite and listed in their table 5 .
most of the sources have jet viewing angles near 80 degrees , which means that their giant lobes will also be close to the plane of the sky .
these viewing angles are much too large for the l - g model to work effectively , which is completely consistent with what we found above for fri galaxies .
although an intrinsic explanation for one - sidedness in jets might have seemed unlikely 20 years ago , it has recently been shown to be the case in m87 where a special attempt was made to show that the jet asymmetry observed could be explained by relativistic motion @xcite .
it was found that at 15 ghz the material producing the asymmetric radiation was not moving relativistically , and it must be concluded that here , too , the asymmetry is likely intrinsic .
it is unfortunate that direct measurements of the motions in kpc jets are almost impossible to make since they require milliarcsec accuracy but must work with the arcsec resolution provided by the vla .
although the speed in the outer jet is difficult to measure we have assumed here that the flow in the low - luminosity fri sources we studied slows to near 0.1c in the outer jet , as was reported previously by several other investigators .
the main conclusions obtained here can then be summarized as follows .
for the low - luminosity sources with highly asymmetric jets and which show the l - g effect , if their kpc - scale jet flow decelerates as quickly as has been claimed by several previous investigators the jet asymmetry can not be produced by beaming .
the asymmetry must then be intrinsic .
if the jet one - sidedness is intrinsic there is an equal chance that the strongest jet is directed away from us .
this means that the currently accepted l - g model can not work because the proposed faraday screen would then lie behind the more rapidly depolarized lobe associated with the weaker jet .
this means that the depolarization must occur inside the relevant lobe with the outer regions of the lobe acting as a faraday screen , and the currently accepted model is then unlikely to be correct .
although the jet flow in frii sources is less accurately known , it seems likely that the explanation for the depolarization in these sources would be the same .
furthermore , if the viewing angles of frii sources are very large , as recent results seem to show is the case , the l - g model can not work because the lobes of these sources would be close to the plane of the sky .
it may be too soon then to completely rule out internal depolarization in the lobes as the true explanation for the l - g effect .
alexander , p. , and leahy , j.p .
1987 , , 225 , 1 barthel , p.d .
1989 , , 336 , 606 bell , m.b .
, international j. of astron . and astrophys .
2012 , vol .
1 , pp 52 - 61 , doi : 10.4236/ijaa.2012.21008 bell , m.b . , and comeau , s.p .
2010 , , 325 , 31 best , p.n . ,
carilli , c.l . ,
garrington , s.t . ,
longair , m.s . , and r@xmath23ttgering , h.j.a .
1998 , , 299 , 357 bicknell , g.v .
1994 , , 422 , 542 bridle , a.h .
2006 , ( radio maps located at the following web address ) http://www.cv.nrao.edu/ abridle / images.htm bridle , a.h . and perley , r.a .
1984 , , 22 , 319 cleary , k. , lawrence , c.r . , marshall , j.a .
, hao , l. , and meier , d. 2007 , , 660 , 117 dreher , j.w . and carilli , c.l .
1987 , , 316 , 611 drouart , d.g et al .
2012 , arxiv:1209.5399 , astron . and astrophys .
( in press ) fanaroff , b.l . and riley , j.m .
1974 , , 167 , 31 garrington , s.t . ,
leahy , j.p . ,
conway , r.g . , and laing , r.a .
1988 , nature , 331 , 149 garrington , s.t . and conway , r.g .
1991 , , 250 , 198 goodlet , j.a . ,
kaiser , c.r .
, best , p.n . , and dennett - thorpe , a. 2004 , , 347 , 508 gopal - krishna and wiita , p.j .
2000 , , 529 , 189 hocuk , s. , and barthel , p.d .
2010 , , 523 , a9 kellermann , k.i .
, et al . 2004 , , 609 , 539 kovalev , y.y . , lister , m.l .
, homan , d.c . , and kellermann , k.i .
2007 , , 668 , l27 laing , r.a . 2988 , nature , 331 , 149 laing , r.a . and bridle , a.h .
1987 , , 228 , 557 laing , r.a . , and bridle , a.h .
2002a , , 336 , 328 laing , r.a . , and bridle , a.h .
2002b , , 336 , 1161 laing , r.a . , and
bridle , a.h .
2007 , arxiv:0801.0147 laing , r.a . ,
parma , p. , de ruiter , h.r . , and fanti , r. 1999 , , 306 , 513 laing , r.a . , canvin , j.r . , bridle , a.h . , and hardcastle , m.j .
2006 , , 372 , 510 lister , m.l . , and marscher , a. 1997 , , 476 , 572 liu , r. and pooley , g. 1991 , , 249 , 343 lpez - corredoira , m. and perucho , m. 2012 , , 544 , a56 morganti , r. , et al .
1997 , , 326 , 919 odea , c.p . , and owen , f.n .
1986 , , 301 , 841 odea , c.p . ,
daly , r.a .
, kharb , p. , freeman , k.a . , and baum , s.a .
2009 , , 494 , 471 perley , r.a . , bridle , a.h . and willis , a.g .
1984 , , 54 , 292 urry , c.m . , and padovani , p. 1995
, , 107 , 803 saripalli , l. 2012 , , 144 , 85 vermeulen , r.c . and cohen , m.h .
1984 , , 430 , 467 walker , r.c . ,
walker , m.a . , and benson , j.m .
1988 , , 335 , 668 | the depolarization asymmetry seen in double - lobed radio sources , referred to as the laing - garrington ( l - g ) effect where more rapid depolarization is seen in the lobe with no visible jet as the wavelength increases , can be explained either by internal differences between the two lobes , or by an external faraday screen that lies in front of only the depolarized lobe .
if the jet one - sidedness is due to relativistic beaming the depolarization asymmetry must be due to an intervening faraday screen . if it is intrinsic the depolarization asymmetry must be related to internal differences in the lobes . for a random viewing angle distribution ,
which must be the case here where un - beamed lobe radiation dominates , jet one - sidedness is unrelated to viewing angle and therefore can not be used either to estimate the viewing angle or to imply beaming . the outflow speed in the kpc jet
is notoriously difficult to determine .
however , although it has not yet been proven conclusively , we assume in this paper that the speed in the outer jet of several fanaroff - riley class 1 ( fri ) sources exhibiting the l - g effect is close to the 0.1c reported by several other investigators . for these sources
we find that the jet one - sidedness can not be explained by beaming and therefore must be intrinsic . in these fri sources
the l - g effect must be due to differences that originate inside the lobes themselves , with the outer regions of the relevant lobe acting as a faraday screen .
although it is not known if the flow in the outer jets of frii sources also slows to this speed it is suggested that the explanation of the l - g effect is likely to be the same in both types .
this argument is strengthened by the recent evidence that frii galaxies have very large viewing angles , which in turn implies that the l - g model can not work regardless of the jet velocity .
it may therefore be too soon to completely rule out internal depolarization in the lobes as the true explanation for the l - g effect . |
spin glass models defined on finite - connectivity random graphs have two control parameters , one is the temperature @xmath4 and the other is the interaction ( or constraint ) density , defined as the number @xmath5 of interactions ( constraints ) versus the number @xmath6 of vertices , @xmath7 .
theoretical work of the last ten years @xcite have established the understanding that , a random spin glass system with many - body interactions will experience a clustering transition if either the temperature @xmath4 is lowered below certain threshold value or the constraint density @xmath8 is increased beyond certain threshold value . at the clustering transition ,
the configuration space of the system splits into an exponential number of gibbs pure states and ergodicity is broken .
this transition is followed by another phase transition of the configuration space , the condensation transition , as @xmath4 is further lowered or @xmath8 further increased . at the condensation transition ,
a sub - exponential number of large gibbs states start to dominate the configuration space and hence the equilibrium property of the system @xcite .
these properties were first observed in fully - connected mean - field spin glass models @xcite .
whether they are also valid in @xmath9-dimensional real - world spin glass systems ( @xmath10 or @xmath11 ) is still a debated open issue .
although the configuration space geometric properties of random finite - connectivity spin glass models at or after the clustering transition have been well characterized by statistical physics methods , much less is known about the configuration space structure just before the clustering transition .
why do an exponential number of gibbs states suddenly appear at the clustering transition ? are they preceded by precursor structures in the ergodic phase of the configuration space ? if yes , when do the gibbs - state precursors start to form and how to describe their evolution ? what are the impacts of these precursor structures to the equilibrium dynamical properties of the system ?
these questions are important for a full understanding of the structural evolution of the configuration space . in this paper , as a continuation of our recent efforts @xcite , we study the fine structure of the configuration space of random finite - connectivity spin glass models in the vicinity of the clustering transition .
we focus on the evolution of the ground - state configuration space ( corresponding to @xmath12 ) using the constraint density @xmath8 as the control parameter .
as demonstrated through two case studies , a heterogeneity transition occurs to the ground - state configuration space at certain critical value of @xmath8 . at the transition point
, exponentially many communities of configurations emerge from the ground - state configuration space .
each configuration community is a collection of relatively similar configurations and it forms a stable thermodynamic phase in the presence of a suitable external field ; they are the precursors for the gibbs states at the clustering transition .
the entropy density @xmath2 of configuration - pairs as a function of configuration - pair overlap @xmath3 is calculated by the replica - symmetric ( rs ) and the first - step replica - symmetry - broken ( 1rsb ) cavity methods .
extensive numerical simulations are performed to confirm that , the entropic trapping effect of ground - state configuration communities leads to strong dynamical heterogeneity of diffusion processes within the configuration space .
dynamical heterogeneity in the fully - connected @xmath0-spin interaction spherical model ( with continuous spins ) @xcite was quantitatively investigated by donati and co - workers @xcite within the framework of the franz - parisi effective potential theory @xcite .
the results of the present paper for finite - connectivity systems with discrete spins are qualitatively similar to the results of @xcite
. some aspects of the heterogeneity transition in random finite - connectivity systems are also shared by real - world @xmath9-dimensional spin glass and structural glass systems , such as hard - sphere poly - disperse colloidal systems , where the particle volume fraction plays the role of the constraint density @xmath8 @xcite . in real - world glass systems and supercooled liquid systems
, dynamic heterogeneity occurs in real space . the present work is also related to the work of krzakala and zdeborova on the adiabatic evolution of a single gibbs state of a finite - connectivity spin glass system as a function of temperature @xmath4 @xcite .
two model systems are studied in the paper .
section [ sec : kxorsat ] concerns with the @xmath0-body spin glass system , which is equivalent to the random @xmath1-xorsat ( exclusive - or - satisfiability ) problem of computer science .
replica - symmetric and 1rsb mean - field calculations are carried out to obtain the entropy density of ground - state configuration - pairs .
section [ sec : ksat ] focuses on the heterogeneity transition and the dynamical heterogeneity of the random @xmath1-satisfiability ( @xmath1-sat ) problem .
results obtained by rs calculations and numerical simulations are reported in this section .
we make further discussions in section [ sec : discuss ] .
the random @xmath0-body spin glass model is defined by the energy function @xmath13 where @xmath14 denotes a spin configuration for the @xmath6 vertices @xmath15 $ ] , with each spin variable @xmath16 ; the index @xmath17 $ ] denotes one of the @xmath5 interactions of the system and @xmath18 is the quenched random coupling constant , whose value is fixed to @xmath19 or @xmath20 with equal probability ; the set @xmath21 includes all the vertices that participate in the interaction @xmath22 , its size is fixed to @xmath0 , and each of its @xmath0 different elements is randomly and uniformly chosen from the whole set of @xmath6 vertices .
similar to @xmath21 , we denote by @xmath23 the set of interactions that involve vertex @xmath24 . while each interaction
@xmath22 in the @xmath0-body spin glass model affects the same number @xmath0 of vertices , the sets @xmath23 may have different sizes for different vertices @xmath24 . actually , when @xmath6 is large enough , the probability that a randomly chosen vertex participates in @xmath25 interactions is governed by the poisson distribution @xmath26 , with mean value @xmath27 . in the @xmath0-body spin glass model
, each interaction contributes either a positive energy @xmath28 or a negative energy @xmath29 to the total energy . if we set @xmath30 , this model is equivalent to the random @xmath1-xorsat problem with the energy function @xmath31 for the random @xmath1-xorsat problem , each interaction @xmath22 is also referred to as a constraint , whose energy @xmath32 is either zero ( constraint being satisfied ) or unity ( constraint violated ) .
the energy @xmath33 counts the total number of violated constraints by the spin configuration @xmath34 .
the constraint density @xmath8 of the system is by @xmath35 .
the random @xmath1-xorsat problem ( or equivalently , the random @xmath0-body spin glass model ) has been well studied in the statistical physics community .
it serves as an interesting system for understanding the low - temperature equilibrium property of finite - connectivity spin glasses @xcite , and for understanding the dynamical property of glassy systems @xcite .
this model is also closely related to error - correcting code systems of information science , such as the sourlas code @xcite .
the ground - state configuration space structure of the random @xmath1-xorsat problem has been investigated in great detail @xcite and was found to depend only on @xmath1 and the constraint density @xmath8 in the limit of @xmath36 . for a given value of @xmath37
there is a satisfiability threshold @xmath38 .
when the constraint density @xmath8 is below @xmath38 , the ground - state energy of model is zero ( the system is in the sat phase ) , but it becomes positive when @xmath39 ( the unsat phase ) .
a zero - energy spin configuration of model is referred to as a solution , and all the solutions form the solution space @xmath40 of this system .
the solution space of a large random @xmath1-xorsat problem is non - empty only if its constraint density is in the range of @xmath41 .
we have @xmath42 , @xmath43 , and @xmath44 @xcite .
before the solution space of the random @xmath1-xorsat ( @xmath45 ) becomes empty at @xmath39 , it experiences an ergodicity - breaking ( clustering ) transition at the threshold value @xmath46 , with @xmath47 and @xmath48 @xcite . for @xmath49 ,
the whole solution space forms a single gibbs state ( the meaning of a gibbs state is explained geometrically in the following subsection ) . on the other hand , for @xmath50 ,
the solution space is no longer ergodic but is composed of exponentially many solution clusters ( gibbs states ) , each of which containing exactly the same number of solutions . for @xmath51 ,
solutions from different solution clusters are separated by energy barriers that are proportional to the vertex number @xmath6 @xcite .
the similarity of two solutions @xmath52 , @xmath53 in the non - empty solution space @xmath40 of a random @xmath1-xorsat problem can be measured by the overlap value @xmath54 we denote by @xmath55 the total number of solution - pairs in the solution space @xmath40 with an overlap value @xmath3 .
this number is exponential in @xmath6 in the sat phase @xmath41 .
therefore in the limit of large @xmath6 , an entropy density @xmath2 is defined as @xmath56 the entropy density function @xmath2 contains rich information about the structure of the solution space @xmath40 ( see figure [ fig : community ] ) .
the shape of @xmath2 has several qualitative changes as the constraint density @xmath8 increases .
the first qualitative change occurs at @xmath57 , where @xmath2 becomes non - concave .
this concavity change corresponds to the formation of ( exponentially ) many solution communities in the solution space ( the large - scale homogeneity of the solution space is then broken ) @xcite .
we refer to @xmath58 as the heterogeneity transition point .
non - concavity of the entropy function @xmath2 as defined by and the large scale organization of the solution space of a random constraint satisfaction problem .
@xmath2 is a concave function of the solution - pair overlap @xmath3 when the constraint density @xmath8 is low enough ( left - most panel ) ; at @xmath59 , @xmath2 starts to be non - concave , marking the formation of many solution communities in the solution space ( second panel from left ) ; then as @xmath8 increases to @xmath60 , @xmath2 becomes non - monotonic , the solution communities separate into different solution clusters and ergodicity of the solution space is broken ; the solution clusters shrink in size as @xmath8 further increases , and eventually the number of solution - pairs with intermediate overlap values @xmath61 $ ] will be zero and @xmath2 is only positive for @xmath62 and @xmath63 .
, scaledwidth=70.0% ] we introduce the following partition function @xmath64 for the solution space @xmath40 @xcite @xmath65 \ , \ ] ] where @xmath66 is a coupling field between the solutions .
when @xmath66 is positive , solution - pairs with larger overlap values have larger weights in the partition sum . under the field
@xmath66 , the mean solution - pair overlap value is @xmath67 . for @xmath68 , we see from that , @xmath69 . at @xmath70 ,
@xmath71 is equal to @xmath72 , the most probable solution - pair overlap value of the solution space .
if the entropy density @xmath2 is concave in @xmath73 $ ] , then @xmath74 increases continuously with @xmath66 for @xmath75 .
however if @xmath2 is non - concave , because there exists a line of slope @xmath76 which touches @xmath2 at two different points @xmath77 and @xmath78 ( ) , then the value of @xmath74 jumps from @xmath79 to @xmath80 at @xmath81 @xcite .
this discontinuity of @xmath74 reveals the existence of a field - induced first - order phase transition at @xmath81 .
the solution - pairs exhibit two different levels of similarity . for @xmath82 ,
the partition function @xmath64 are contributed overwhelmingly by ( intra - community ) solution - pairs with overlap values @xmath83 ( overlap - favored phase ) , while for @xmath84 , @xmath64 is dominated by ( inter - community ) solution - pairs with overlap values @xmath85 ( entropy - favored phase ) .
the difference @xmath86 increases from zero as the constraint density @xmath8 exceeds the critical value @xmath58 . at @xmath87 ,
the solution space is in a critical state , where the boundaries between different solution communities are elusive , and the field - induced phase transition is second - order .
the hamming distance between two solutions @xmath52 and @xmath53 is defined as @xmath88 , which is related to the solution - pair overlap by @xmath89 .
the solution space @xmath40 can be represented by a graph of nodes and edges .
each node of this solution graph denotes a solution , and an edge is linked between two nodes of the graph if the corresponding two solutions has a hamming distance not exceeding a specified value @xmath90 . for the random @xmath1-xorsat problem at @xmath91 , there exists a minimum value of @xmath92 such that all the nodes of the solution graph are in a single connected component @xcite .
we take this minimum value as our edge linking criterion .
( for @xmath93 , if @xmath90 is not of the same order as @xmath6 , the solution graph will be a collection of exponentially many disjointed components . ) in the ergodic phase of @xmath49 , we may introduce two particles to the solution graph .
initially these two particles are residing on the same node , say @xmath94 , of the graph . in case
the particles are uncoupled
, then each particle performs a random diffusion in the solution graph independent of the other : suppose at time @xmath95 the particle is at node @xmath34 , then at the next time step it will , with probability @xmath96 , make a move to a randomly chosen nearest - neighbor of this node , where @xmath97 is the number of attached edges ( the degree ) of node @xmath34 , and @xmath98 is the maximal node degree in the solution graph .
in the case the particles are coupled by a field @xmath66 , however , the particle diffusions are mutually influenced , and the visited node - pairs ( say @xmath52 and @xmath53 ) are no longer uniformly distributed but are favored to more similar pairs by a factor of @xmath99 . in the thermodynamic limit @xmath36 , when the coupling field @xmath66 is larger than @xmath100 , then even at time approaching infinity , the two particles will still be diffusing in the neighborhood of each other and in the neighborhood of the initial node @xmath94 .
such a strong memory effect at field value @xmath101 is a dynamical manifest of the existence of communities in the solution space .
if one knows a solution @xmath102 for the @xmath1-xorsat system , it is convenient to perform a gauge transform @xmath103 to the spin value of each vertex @xmath24 . under this transform , is simplified to @xmath104 in this transformed system , all the coupling constants are positive ( @xmath19 ) , and the overlap of the transformed solution @xmath34 with the reference solution @xmath52 is @xmath105 .
is independent of the reference solution @xmath52 .
this is a well - known property of the @xmath1-xorsat problem , namely its solution space @xmath40 has the same local and global structure when viewed from any of its solutions .
because of this nice property , instead of calculating the solution - pair number @xmath55 , we calculate the number @xmath106 of solutions which have an overlap value @xmath3 with a reference solution @xmath52 .
we denote the corresponding entropy density also as @xmath2 , i.e. , @xmath107 .
( with this slight abuse of notation , the solution - pair entropy density as defined by is @xmath108 , where @xmath109 is the entropy density of the whole solution space at constraint density @xmath8 . )
0.82 cm the entropy density @xmath2 of solutions at an overlap level @xmath3 to a reference solution for the random @xmath1-xorsat problem with @xmath110 and @xmath111 , @xmath112 , @xmath113 , @xmath114 ( from top to bottom ) .
black solid lines and red dashed lines are results obtained by the annealed approximation , and blue dots are results of the replica - symmetric cavity method ( dotted lines just for guiding the eye ) .
when is valid , expression gives the exact mean value of @xmath2 , but if it is violated , is only an upper - bound for the mean value of @xmath2 . , title="fig:",scaledwidth=60.0% ]
the average value of @xmath106 over the ensemble of random @xmath1-xorsat problems with fixed vertex number @xmath6 and constraint density @xmath8 is @xmath115 for @xmath36 we obtain the following annealed approximation for @xmath2 as @xmath116 for @xmath110 , the function @xmath117 is concave in @xmath3 only for @xmath118 . for @xmath119 ,
@xmath117 is non - concave in @xmath3 but is still monotonic in the range @xmath120 $ ] ; the monotonicity of @xmath117 in @xmath120 $ ] is lost when @xmath8 exceeds @xmath121 ( ) . the same qualitative results are obtained for the random @xmath1-xorsat problem with @xmath122 . the annealed approximation @xmath117 is an upper bound of the entropy density @xmath2 for a typical random @xmath1-xorsat problem .
these two quantities are identical only when @xmath106 has the self - averaging property , i.e. , the value distribution of @xmath106 among the ensemble of random @xmath1-xorsat problems of constraint density @xmath8 approaches a delta function in the limit of @xmath36 . to check whether this self - averaging property is valid
, we need to calculate the mean value of @xmath123 .
after some combinatorial analysis , we obtain @xmath124 where @xmath125/4 $ ] . in the @xmath36 limit ,
we therefore have @xmath126 with @xmath127 being expressed as @xmath128 self - averaging of @xmath106 requires that @xmath129 numerical calculations reveal that is satisfied only for values of @xmath3 very close to @xmath130 or very close to @xmath131 . in
, @xmath117 is plotted as a black solid line if @xmath132 and as a red dashed line if @xmath133 .
as @xmath117 is only an upper bound to @xmath2 , the fact that @xmath117 becomes non - concave at @xmath49 can not be used as a proof that @xmath2 also becomes non - concave at @xmath91 . we proceed to calculate @xmath2 by the cavity method of statistical physics . for the gauge - transformed @xmath1-xorsat system ,
if vertex @xmath24 is involved in constraint @xmath22 , we define a cavity probability @xmath134 as the probability that @xmath135 takes the value @xmath136 when the constraint @xmath22 is absent . for each constraint @xmath22
we exploit the bethe - peierls approximation @xcite and assume that , the spin states of the vertices @xmath137 are mutually independent in the absence of @xmath22 . under this approximation , we obtain that if the vertex @xmath138 takes the spin value @xmath135 , the probability that constraint @xmath22 being satisfied is equal to @xmath139/2 $ ] , where @xmath140 means the subset of @xmath21 that is missing element @xmath24 .
then the following belief - propagation equation can be written down for each vertex - constraint association @xmath141 : @xmath142 } { e^{x } \prod\limits_{b\in \partial i \backslash a } \left [ \frac{1 + \prod\limits_{j\in \partial b\backslash i } ( 2 p_{j\rightarrow b}^+-1)}{2 } \right ] + e^{-x } \prod\limits_{b\in \partial i\backslash a } \left [ \frac{1- \prod\limits_{j\in \partial b\backslash i } ( 2 p_{j\rightarrow b}^+-1)}{2 } \right ] } \ .
\label{eq : bp - xorsat}\ ] ] for a given energy function , there are @xmath143 iteration equations , which form the replica - symmetric cavity theory @xcite .
0.82 cm the mean overlap @xmath74 for the random @xmath144-xorsat problem at @xmath145 , @xmath146 , @xmath147 , @xmath148 ( from right to left ) . for @xmath149
, @xmath74 is a continuous function of @xmath66 . at @xmath150 ,
the slope of @xmath74 diverges at @xmath151 . for @xmath152
, @xmath74 is discontinuous and has a hysteresis loop .
, title="fig:",scaledwidth=60.0% ] after this set of belief - propagation iteration equations has reached a fixed point , the mean overlap with the reference solution is calculated as @xmath153 -e^{-x } \prod\limits_{a\in \partial i } \left [ \frac{1 - \prod\limits_{j\in \partial a\backslash i } ( 2 p_{j\rightarrow a}^+-1)}{2 } \right ] } { e^{x } \prod\limits_{a\in \partial i } \left [ \frac{1 + \prod\limits_{j\in \partial a\backslash i } ( 2 p_{j\rightarrow a}^+-1)}{2 } \right ] + e^{-x } \prod\limits_{a\in \partial i } \left [ \frac{1- \prod\limits_{j\in \partial a\backslash i } ( 2 p_{j\rightarrow a}^+-1)}{2 } \right ] } \ .
\label{eq : q - xorsat - rs}\ ] ] and the entropy density as a function of @xmath66 is @xmath154 + e^{-x } \prod\limits_{a\in \partial i } \left
[ \frac{1-\prod\limits_{j\in \partial a\backslash i } ( 2 p_{j\rightarrow a}^+ -1 ) } { 2 } \right ] \right ) \nonumber \\ - \frac{1}{m}\sum\limits_{a=1}^{m } ( k-1 ) \ln \left ( \frac { 1 + \prod\limits_{j \in \partial a } ( 2 p_{j\rightarrow a}^+ -1 ) } { 2 } \right ) - x \overline{q}(x ) \ . \label{eq : s - xorsat - rs}\end{aligned}\ ] ] the mean value of @xmath155 as averaged over the ensemble of random @xmath1-xorsat problems ( at fixed value of @xmath8 ) can be calculated using the population dynamics technique @xcite . by eliminating @xmath66 from @xmath155 and @xmath74
we obtain the entropy density function @xmath2 .
the numerical results are shown in ( blue dots ) for @xmath110 at different values of @xmath8 .
the mean overlap function @xmath74 is shown in for @xmath110 and @xmath156 .
similar results are obtained for the cases of @xmath122 . for the random @xmath144-xorsat problem
, @xmath74 is a continuous and smooth function of field @xmath66 when @xmath157 .
as @xmath8 approaches @xmath58 from below , however , the maximal slope of @xmath74 is proportional to @xmath158 and diverges at @xmath159 .
this divergence is a consequence of the fact that the entropy density @xmath2 starts to be non - concave at @xmath160 . for @xmath152 , @xmath74
as calculated by the rs cavity theory shows discontinuity and hysteresis behavior when @xmath66 is close to certain threshold value @xmath100 , indicating the existence of two distinct phases of the solution space as viewed from the reference solution ( @xmath52 ) .
one of the phases contains solution @xmath52 and the other similar solutions , whose overlap with @xmath52 is larger than certain characteristic value @xmath161 , see .
we regard this phase as the solution community of solution @xmath52 .
if we choose another solution outside the solution community of @xmath52 , we will find that this new reference solution is also associated with a different solution community . for @xmath162 , the solution space of the random @xmath1-xorsat problem is therefore formed by exponentially many solution communities .
as the solution space is very heterogeneous at this range of @xmath8 , the replica - symmetric mean - field theory probably is not sufficient to describe its statistical property .
we now proceed to study the solution space heterogeneity using the 1rsb cavity theory . for the gauge - transformed model under the coupling field
@xmath66 , to apply the 1rsb mean - field theory , the solution space is first divided into an exponential number of gibbs states @xmath163 @xcite .
each gibbs state @xmath163 represents a subspace @xmath164 , and its partition function is defined as @xmath165 we can then define a @xmath166free energy density @xmath167 as @xmath168 .
this free energy density is the sum of two parts , @xmath169 , where @xmath170 is the entropy density of gibbs state @xmath163 and @xmath171 is the mean overlap level of solutions in @xmath172 to the reference solution @xmath52 .
results of the @xmath173 1rsb mean - field theory for the random @xmath144-xorsat problem with @xmath174 .
the complexity , the entropy density of solution communities , and the total entropy density are shown in the upper , middle , and lower panel , respectively , as a function of the overlap @xmath3 .
two sets of 1rsb mean - field results are obtained in the population dynamics when different types of initial values are used for the population . at each value of @xmath3 , the result with the larger value of complexity and the smaller value of community entropy density should be considered .
the two vertical dashed lines mark the boundary between the replica - symmetric ( rs ) and the replica - symmetry - broken ( rsb ) region .
, scaledwidth=60.0% ] the total 1rsb partition function of the system is @xmath175 \right ) \ .\ ] ] in the above equation , @xmath176 is the parisi parameter , and @xmath177 is the complexity , which measures the @xmath166entropy density of gibbs states at the free energy density level @xmath178 @xcite . if we set @xmath173 in , each gibbs state contributes a term @xmath179 to @xmath180 , and then @xmath180 is the total partition function of the system , provided that the complexity @xmath177 calculated at @xmath173 is non - negative .
the existence of a nontrivial 1rsb solution at @xmath173 is also a signature of the instability of the rs theory of the previous subsection @xcite .
we therefore set @xmath173 in our calculations .
the details of the 1rsb cavity mean field theory are presented in [ sec : appendix - a ] , and here we discuss some of the numerical results obtained by population dynamics on the random @xmath144-xorsat problem .
same as , but for @xmath181 .
the complexity @xmath182 is positive for @xmath183 .
, scaledwidth=60.0% ] at @xmath184 , no nontrivial 1rsb mean field results are obtained .
the fixed point of the 1rsb population dynamics reduces to that of the rs theory for all the overlap values @xmath185 . as @xmath8 exceeds @xmath159 ,
the replica - symmetric mean field theory becomes unstable for intermediate values of overlap @xmath3 , and nontrivial fixed points of the 1rsb mean - field population dynamics are observed .
for example , at @xmath174 we find that the complexity @xmath182 is exactly zero for overlap value @xmath186 and @xmath187 , and @xmath182 is negative for @xmath188 ( see ) .
the fact that @xmath189 for @xmath190 suggests that , the solutions with overlap levels @xmath190 to @xmath52 are in a single solution community . in the corresponding solution subgraph of this solution community , any two solutions ( nodes ) with the same hamming distance @xmath9 to @xmath52 are connected by at least one path that involves only other solutions with the same hamming distance @xmath9 to @xmath52 ( i.e. , the subspace of solutions having the same overlap @xmath3 ( @xmath191 ) with @xmath52 is ergodic within itself ) .
the second message we get , from the fact that @xmath192 for @xmath188 , is that the solution subgraph formed by all the solutions at the same overlap level @xmath3 [ @xmath193 to @xmath52 is not ergodic within itself but is divided into greatly many disjointed connected sub - components , with a sub - exponential number of dominating ones . in this overlap range ,
the community entropy density and the total entropy density as obtained by the @xmath173 1rsb mean - field theory can only be regarded as upper bounds for the true values .
one needs to work with @xmath194 to obtain better estimates for the community entropy density and the total entropy density .
the third message we get from @xmath189 for @xmath195 is that , the solutions with overlap levels to @xmath52 less than @xmath196 form an ergodic subspace within itself .
more precisely , all the solutions at each overlap level @xmath3 ( @xmath197 ) to @xmath52 are ergodic within themselves .
the solutions in such a subspace of fixed overlap @xmath3 come from different solution communities , but they are connected with each other in the solution graph even when only the edges inside the subgraph are remained .
same as , but for @xmath198 and @xmath199 . at @xmath198 ,
the complexity @xmath182 is positive for @xmath200 and is negative ( @xmath201 ) for @xmath202 . at @xmath203 ,
the complexity @xmath182 is positive even at @xmath204 .
, scaledwidth=60.0% ] the theoretical results as shown in and for the random @xmath144-xorsat problem at @xmath181 and @xmath198 , respectively , are similar with the results of . at @xmath181 ,
the complexity @xmath182 calculated at @xmath173 is positive for @xmath183 , indicating an exponential number of solution communities are equally contributing to the solution subspace at each overlap level @xmath205 $ ] .
for an overlap level @xmath206 , however , the solution subspace is again dominated by a sub - exponential number of large solution communities .
the situation at @xmath198 is similar , but the range of @xmath3 values for which @xmath207 is further enlarged .
when the constraint density @xmath8 exceeds the clustering transition point @xmath208 , the complexity @xmath182 at @xmath173 becomes positive at @xmath204 ( see the exemplar case of @xmath199 in ) .
this result means that the largest subspace of solutions ( with overlap level @xmath204 to @xmath52 ) becomes non - ergodic at @xmath209 , as expected .
the calculations of this section demonstrated that the solution space of the random @xmath1-xorsat problem experiences a heterogeneity transition as the constraint density @xmath8 approaches @xmath210 , where an exponential number of solution communities start to form in the solution space .
these solution communities separate into different solution clusters at a larger constraint density value @xmath211 , where an ergodicity - breaking transition occurs . for @xmath162 , although the solution space as a whole is ergodic , the subspaces of solutions with intermediate overlap levels to a reference solution are non - ergodic within themselves .
our theoretical results , combined with the results of @xcite for @xmath50 , give a complete picture on the structural evolution of the solution space of the random @xmath1-xorsat problem .
the random @xmath1-sat problem is a famous model system for the study of typical - case computational complexity of np - complete combinatorial satisfaction problems @xcite .
its energy function @xmath212 , like the random @xmath1-xorsat problem , is defined as a sum over @xmath213 constraints @xmath22 : @xmath214 in , each constraint @xmath22 affects a set @xmath21 of @xmath1 randomly chosen vertices from the vertex set @xmath215 ; @xmath216 is the preferred spin state of constraint @xmath22 on the vertex @xmath217 , it takes the quenched value @xmath28 or @xmath29 with equal probability . if at least one of the vertices @xmath217 takes the spin value @xmath218 , the energy of the constraint @xmath22 is zero , otherwise its energy is unity . the solution space @xmath40 of model is formed by all the spin configurations @xmath34 of zero total energy ( i.e. , satisfying all the constraints ) . in model ,
each vertex @xmath24 is constrained by a set ( denoted as @xmath23 ) of constraints . after the experimental demonstration of a satisfiability phase - transition in the random @xmath144-sat problem by kirkpatrick and selman @xcite , studies on the solution space structure of the random @xmath1-sat problem have been carried out through rigorous mathematical methods ( see , e.g. , @xcite ) and through statistical physics methods ( see , e.g. , @xcite ) .
the threshold constraint density @xmath38 for the solution space to be empty is calculated to be @xmath219 for the random @xmath144-sat problem @xcite and its values for @xmath220 are also predicted by the 1rsb zero - temperature energetic cavity method @xcite ; a lower bound on @xmath38 is calculated by the zero - temperature long - range frustration theory @xcite .
ergodicity of the solution space is broken at the clustering transition point @xmath221 @xcite . at @xmath222 ,
the solution space contains an exponential number of gibbs states .
the value of @xmath211 is calculated by the 1rsb zero - temperature entropic cavity method in @xcite , reporting @xmath223 and @xmath224 .
we are interested in the heterogeneity of the ergodic solution space at @xmath49 . in the following subsection
we calculate the solution - pair mean overlap as defined by using the replica - symmetric cavity method .
the heterogeneity transition point @xmath210 is determined by this rs mean - field theory .
the partition function is a weighted sum over all the solution - pairs @xmath225 .
each vertex is then associated with a spin vector - state @xmath226 . under the coupling field @xmath66
, we define the cavity probability @xmath227 as the probability that constraint @xmath22 is satisfied if the vertex @xmath217 takes the spin vector - state @xmath228 .
similarly , we define the cavity probability @xmath229 as the probability that vertex @xmath24 is in the spin vector - state @xmath228 in the absence of the constraint @xmath230 . if we exploit the bethe - peierls approximation as mentioned in section [ subsec : kxorsat - rs ] , the following rs cavity equations can be written down for @xmath231 and @xmath232 @xmath233 -\delta_{\sigma^\prime}^{-j_a^i } \prod_{j\in \partial a \backslash i } \left[\sum_{\sigma_j } p_{j\rightarrow a}(\sigma_j , -j_a^j)\right ] \nonumber \\
+ \delta_{\sigma}^{-j_a^i } \delta_{\sigma^\prime}^{-j_a^i } \prod_{j\in \partial a\backslash i } p_{j\rightarrow a}(-j_a^j , -j_a^j ) \ , \label{eq : p_a_to_i } \\
p_{i\rightarrow a}(\sigma , \sigma^\prime ) = \frac{e^{x
\sigma \sigma^\prime } \prod_{b\in \partial i\backslash a } \hat{p}_{b\rightarrow i}(\sigma , \sigma^\prime ) } { \sum\limits_{\sigma_i } \sum\limits_{\sigma_i^\prime } e^{x \sigma_i \sigma_i^\prime } \prod_{b\in \partial i\backslash a } \hat{p}_{b\rightarrow i}(\sigma_i , \sigma_i^\prime ) } \ , \label{eq : p_i_to_a}\end{aligned}\ ] ] where @xmath234 is the kronecker symbol ( @xmath235 if @xmath236 and @xmath237 if @xmath238 ) .
there is a symmetry requirement for @xmath239 , namely that @xmath240 .
this condition is a result of the fact that , the solution - pair @xmath225 has the same contribution to the partition function as the solution - pair @xmath241 . after a fixed - point has been reached for the rs iterative equations , the mean value @xmath74 of the solution - pair overlap and the solution - pair entropy density @xmath2 as defined by
can then be calculated .
for example , the mean solution - pair overlap is expressed as @xmath242 where @xmath243 means averaging under the coupling field @xmath66 .
the predicted mean overlap function @xmath74 for the random @xmath144-sat problem is shown in ( a ) for @xmath8 in the vicinity of @xmath244 .
similar to the results of the random @xmath144-xorsat problem shown in , the continuity of @xmath74 changes as @xmath8 exceeds a critical value @xmath245 .
the jumping and hysteresis behavior of the mean overlap @xmath74 for @xmath246 indicates that solution communities start to emerge in the solution space of the random @xmath144-sat problem at @xmath247 .
0.82 cm mean - field predictions on the mean solution - pair overlap @xmath74 ( a ) and the overlap susceptibility @xmath248 ( b ) for the random @xmath144-sat problem . in ( a ) the constraint density
@xmath8 increases from @xmath249 to @xmath250 ( right to left ) with step size @xmath251 .
the inset of ( b ) shows how the inverse of the peak value of @xmath248 approaches zero as @xmath8 increases to @xmath252 from below . at @xmath57 ,
@xmath248 becomes infinite at @xmath253 . , title="fig:",scaledwidth=60.0% ] 1.22 cm mean - field predictions on the mean solution - pair overlap @xmath74 ( a ) and the overlap susceptibility @xmath248 ( b ) for the random @xmath144-sat problem . in ( a ) the constraint density
@xmath8 increases from @xmath249 to @xmath250 ( right to left ) with step size @xmath251 .
the inset of ( b ) shows how the inverse of the peak value of @xmath248 approaches zero as @xmath8 increases to @xmath252 from below . at @xmath57 ,
@xmath248 becomes infinite at @xmath253 .
, title="fig:",scaledwidth=60.0% ] the overlap susceptibility @xmath248 measures the sensitivity of the mean solution - pair overlap with the coupling field @xmath66 , it is defined by @xmath254 . the susceptibility @xmath248 is related to the overlap fluctuation by @xmath255 \ .
\label{eq : ksat - susceptibility}\ ] ] ( b ) demonstrates that , as @xmath8 approaches @xmath244 from below , the peak value of @xmath248 becomes more and more pronounced and finally diverges . from the divergence of the peak value of @xmath248
, we obtain that @xmath252 for the random @xmath144-sat problem .
this value is much below the value of @xmath256 .
similar mean - field calculations are performed for the random @xmath257-sat problem and we obtain that @xmath258 . this value is again much below the clustering transition point @xmath259 .
the solution space heterogeneity of the random @xmath1-sat problem influences the dynamics of random walking diffusion processes @xcite .
similar to the random @xmath1-xorsat problem , we can represent the solution space of the random @xmath1-sat problem as a solution graph of nodes and edges , with the nodes denoting individual solutions and the edges connecting pairs of solutions of unit hamming distance . when the constraint density @xmath8 of the random @xmath1-sat problem is less than @xmath211 , this solution graph has a single giant connected component that includes almost all the solutions of the solution space . in this ergodic phase of @xmath49 ,
the structure of this huge solution graph become heterogeneous when the solutions aggregate into many different communities .
there are many domains of high edge density in the solution graph , corresponding to the different solution communities .
the nodes of different domains are also connected by many edges , but the density of inter - domain edges is much lower than the density of intra - domain edges .
as was demonstrated in @xcite , this heterogeneity of edge density causes an entropic trapping effect to diffusive particles on the solution graph .
the dynamics of a diffusive particle can be decomposed into a trapping mode ( the particle diffuses within a relatively dense - connected domain of the solution graph ) and a transition mode ( the particle escapes from one domain of the solution graph , wonders for a while , and then enters into another domain of the solution graph ) . as the trajectory of the diffusive particle oscillates between the trapping mode and the transition mode , if a clustering analysis is performed on a set of solutions sampled from this trajectory at equal time interval , a clear community structure can be observed among the sampled solutions @xcite .
the solution space diffusion process can be turned into a stochastic search algorithm .
one such algorithms , the seqsat of @xcite , constructs a solution for a random @xmath1-sat problem in a sequential manner .
constraints of the problem are added one after another in a random order , and as a new constraint is added , a random walk process of single - spin flips is performed in the solution space of the satisfied sub - problem to reach a solution that also satisfies the new constraint .
it was observed @xcite that , when the constraint density @xmath8 of the satisfied sub - problem exceeds @xmath210 , the seqsat search process becomes viscous , the mean waiting time needed to satisfy a new constraints starts to increases rapidly with @xmath8 , and the sequence of waiting times starts to have large fluctuations .
this dynamical behavior is easily understood in terms of the heterogeneity of the underlying solution space .
solution space structural heterogeneity results in dynamical heterogeneity of diffusion processes . to demonstrate this point more clearly and to estimate a typical relaxation time
, we study in this subsection a simple solution space glauber dynamics . for a large random @xmath1-sat problem with @xmath6 variables and @xmath213 constraints ( @xmath49 ) ,
first we construct through seqsat a spin configuration that satisfies all the @xmath5 constraints .
( of cause we can also use other heuristic algorithms to generate an initial solution .
the only requirement is that this solution should be a typical solution , or in other words , it should belong to the single largest solution cluster of the solution space .
this requirement is satisfied in our simulation studies . )
then we set the initial time as @xmath260 and denote the initial solution as @xmath261 .
the spin configuration is then updated by single - spin flips at each elementary time step @xmath262 .
let us suppose at time @xmath95 the spin configuration is @xmath263 .
then a vertex @xmath24 is chosen uniformly randomly from the whole vertex set @xmath264 ; a candidate spin configuration @xmath265 is constructed , with @xmath266 if @xmath267 and @xmath268 . if @xmath265 is not a solution of the @xmath1-sat problem , then at time @xmath269 , the old spin configuration is kept , i.e. , @xmath270 . however , if @xmath265 is also a solution , then with probability one - half @xmath271 and with the remaining probability one - half @xmath272 .
a unit time of the above - mentioned glauber dynamics corresponds to @xmath6 spin - flip attempts .
the actual number of accepted spin flips in a unit time is about @xmath273 for the problem instances of and .
if this random walk diffusion process is simulated for an extremely long period of time , every solution in the connected component of the solution graph which the initial solution @xmath274 belongs will have the same frequency of being visited .
the time @xmath275 is set to be large enough ( e.g. , @xmath276 ) to ensure that the diffusion process has completely forget the initial solution @xmath274 at time @xmath277 .
0.22 cm glauber dynamics simulation results for a single random @xmath1-sat ( @xmath110 ) problem instance with @xmath278 vertices . at a given value of @xmath8 ,
the first @xmath213 constraints of this single random instance are used in the simulation .
( upper panel ) mean value of the overlap @xmath279 between two solutions separated by time @xmath280 along the trajectory of configuration evolution .
( lower panel ) the variance @xmath281 of the solution - pair overlap @xmath279 .
, title="fig:",scaledwidth=60.0% ] for the trajectory of spin configurations at @xmath277 , the quantity @xmath282 measures the overlap between two spin configurations @xmath263 and @xmath283 that are separated by a time @xmath280 .
@xmath279 is a random variable , its value fluctuates with different choices of the time @xmath95 .
we are mainly interested in the mean value @xmath284 and the variance @xmath281 of the overlap @xmath279 .
the mean overlap @xmath284 is calculated by @xmath285 where @xmath286 means averaging over different starting times @xmath277 along the diffusion trajectory . the variance @xmath281 is expressed as @xmath287 \ .
\nonumber \\ \ ; \label{eq : ksat - fluctuation}\end{aligned}\ ] ] comparing with , a dynamical susceptibility @xmath288 can be defined as @xmath289 ( such a quantity was introduced in @xcite , see also the review @xcite ) . same as , but for @xmath290 .
, scaledwidth=60.0% ] the upper panel of shows the relaxation behavior of @xmath284 for a large random @xmath144-sat formula with @xmath278 vertices . to study how the shape of @xmath284 changes with constraint density @xmath8
, the first @xmath213 constraints of this same formula is used for glauber dynamics simulation at each value of @xmath8 .
we focus on @xmath8 values in the vicinity of @xmath159 . for @xmath291 , the mean overlap @xmath284 reaches its plateau value of @xmath292 at @xmath293 . for @xmath294 , the plateau of @xmath284 is reached at @xmath295 , and its value increases to @xmath296 . for @xmath297 , the plateau of @xmath284 is reached at @xmath298 , with a value of about @xmath299 . these and our other unshown simulation results clearly confirm that , as @xmath8 exceeds @xmath159 , the relaxation of the solution space diffusion process is slowed down greatly .
the lower panel of shows the variance @xmath281 of the solution - pair overlap values . at @xmath291 ,
the variance @xmath281 increases with @xmath280 and reaches its plateau value at @xmath293 . at @xmath300
, @xmath281 starts to show a peak at @xmath301 .
the peak of @xmath281 becomes more and more pronounced as @xmath8 further increases , and the time @xmath302 , corresponding to the peak value of @xmath281 , shifts to larger values with @xmath8 . at @xmath297 , @xmath303 .
the peak time @xmath302 of the variance @xmath281 gives a measure of the typical time scale of dynamical heterogeneity of the glauber dynamics .
for @xmath304 , two solutions @xmath263 and @xmath283 have a large chance of being in the same solution community , therefore the fluctuation of overlap values @xmath279 is small ; a large fraction of vertices are inactive in this relatively short time window of @xmath280 , as their spin values are flipped only with low frequencies . on the other hand , for @xmath305 , the compared two solutions @xmath263 and @xmath283 have a large chance of belonging to two different solution communities , and therefore the fluctuation of their overlap values is again small ; the spin values of most of the vertices have been flipped many times during such a large time window , and hence dynamical heterogeneity is destroyed .
when the time window @xmath280 is comparable to @xmath302 , the solutions @xmath263 and @xmath283 have comparable probabilities of being in the same solution community and being in two different communities ; this causes relatively large fluctuation of the overlap values . as @xmath8 increases from @xmath159 further , it takes more time for the diffusion process to escape from a solution community , and the difference between intra- and inter - community overlap values become larger , these two facts make the peak value @xmath281 and the peak time @xmath302 both to increase rapidly . reports the simulation results for a random @xmath257-sat problem instance of @xmath278 vertices and @xmath306 constraints .
similar to the case of random @xmath144-sat , we find that the viscosity of the glauber dynamics increases rapidly with constraint density @xmath8 . for @xmath307 , the mean
overlap @xmath279 reaches its plateau value only at a very large value of @xmath308 .
as demonstrated by the variance @xmath281 of solution - pair overlaps , dynamical heterogeneity become to manifest itself at @xmath309 and it becomes more and more pronounced as @xmath8 further increases . comparing and
we can also notice an important dynamical difference between the random @xmath144-sat and the random @xmath257-sat case .
for the random @xmath257-sat system , we observe that the relaxation curve of @xmath284 is @xmath166z form in shape at @xmath310 ( i.e. , @xmath311 of @xmath312 ) , with a plateau value of @xmath313 at intermediate time intervals @xmath280 before it finally decay to a much lower @xmath314 at @xmath315 .
the peak of the overlap variance @xmath281 corresponds to the time interval @xmath280 at which @xmath316 starts to decay from the larger plateau .
however , for the random @xmath144-sat system , we find that the curve of @xmath284 has a @xmath166l shape even at @xmath297 ( @xmath317 of @xmath208 ) , and at the time when @xmath281 reaches its peak value , the mean overlap @xmath284 is already close to its asymptotic value of @xmath318 .
we suggest that this dynamical difference is a strong reflect of the difference between the community structures of the random @xmath144-sat and the general random @xmath1-sat problems with @xmath220 . for the random @xmath144-sat problem , in the heterogeneity phase of @xmath319 , the solution space probably is dominated by a sub - exponential number of largest solution communities . because of this predominance , the mean intra - community solution - pair overlap of each of these dominating communities is only slightly higher than the mean solution - pair overlap of the whole solution space .
then at typical time @xmath302 when the diffusive particle jumps between different dominating solution communities , one will not observe much drop in the mean overlap value @xmath316 .
such a picture is consistent with the prediction that , at @xmath320 , only a sub - exponential number of solution gibbs states dominate the solution space @xcite .
for the random @xmath1-sat problem with @xmath220 , however , in the range of @xmath321 the solution space probably is contributed mainly by a exponential number of median - sized solution communities .
the sub - exponential number of the largest solution communities has only a negligible contribution to the whole solution space and therefore barely influence the dynamics of the diffusive particle .
for such a community structure , then the mean intra - community solution - pair overlap will be much larger than the mean solution - pair overlap of the whole solution space , resulting in a change of trend of @xmath284 at @xmath322 .
such a picture for @xmath220 is again consistent with the prediction that , at @xmath93 the solution space is dominated by an exponential number of median - sized gibbs states @xcite . in this section
, we confirmed that the solution space of the random @xmath1-sat problem becomes heterogeneous at @xmath323 and determined by the replica - symmetric cavity method that @xmath324 and
@xmath325 . we demonstrated that the existence of many solution communities in the solution space caused heterogeneous behavior in the dynamics of a solution space diffusion process .
the typical time scale of dynamical heterogeneity of this diffusion process is determined by computer simulations .
a heterogeneity transition was found to occur in the ground - state configuration spaces of two multiple - spin interaction systems , the random @xmath1-xorsat problem and the random @xmath1-sat problem .
we expect that this transition is a general phenomenon that occurs in many other spin glass systems before the ergodicity - breaking transition of the ground - state configuration space .
such a heterogeneity transition is unlikely to be special to the ground - state configuration space but should also be observed as the energy level ( or equivalently the temperature @xmath4 ) of the configuration space is lowered to certain critical level .
if the configuration space of a spin glass system is ergodic but highly heterogeneous at certain temperature @xmath4 , this structural heterogeneity probably will manifest itself through heterogeneous behaviors in various spin relaxation dynamical processes of the system .
more deep understanding on the relationship between the phenomenon of dynamical heterogeneity and the structural heterogeneity of the configuration space is very desirable .
such efforts may bring new ways of probing configuration space heterogeneity from observing features of dynamical heterogeneity .
the solution space heterogeneity of the random @xmath1-sat problem has been studied analytically only through the replica - symmetric cavity method . however , from the experiences gained on the random @xmath1-xorsat problem , we believe the replica - symmetric cavity theory is not sufficient for a heterogeneous solution space . a complete study on the heterogeneity transition of the random @xmath1-sat using the 1rsb mean - field cavity theory will be reported in a later paper .
in the case of the random @xmath1-sat problem , we have not yet performed a systematic investigation on the scaling behaviors of the peak value of overlap variance ( equivalently , the overlap susceptibility @xmath326 ) and the typical time @xmath302 of dynamical heterogeneity .
probably both the peak value of @xmath326 and the characteristic time @xmath302 diverge at the clustering transition point @xmath327 . to get unambiguous results , we need a more efficient protocol of simulating the diffusion dynamics on the solution space .
simulations results of @xcite and @xcite indicated that , for the random @xmath144-sat and random @xmath257-sat problem , the single solution space gibbs states at @xmath93 are themselves very heterogeneous in internal structure .
to study analytically the heterogeneity of single solution clusters in the ergodicity - breaking phase , however , appears to be a very challenging task .
such kind of investigations may be very valuable for us to understand how the largest solution clusters of the random @xmath1-sat problem evolve with constraint density @xmath8 .
hz thanks dr . lenka zdeborova for helpful correspondences ; cw thanks professor taiyu zheng for her kind support . this work was partially supported by the national science foundation of china ( grant numbers 10774150 and 10834014 ) and the china 973-program ( grant number 2007cb935903 ) .
here we list the technical details of the 1rsb mean - field cavity theory for the random @xmath1-xorsat problem as discussed in section [ sec:1rsb - xorsat ] . for a comprehensive description of the 1rsb mean - field theory ,
the reader is referred to @xcite . when there are an exponential number of gibbs states , the probability @xmath328 as defined in section [ subsec : kxorsat - rs ] will be different for different gibbs states .
the distribution of @xmath328 among all the gibbs states is denoted by @xmath329 .
we define @xmath330 as the average value of @xmath328 , i.e. , @xmath331 .
we also define two auxiliary distributions as @xcite @xmath332 @xmath333 is the conditional probability of @xmath328 given that @xmath334 .
similarly , @xmath335 is the conditional probability of @xmath328 given that @xmath336 .
at the special case of @xmath173 , it can be shown that the iteration equation for @xmath330 has the same form as , with the only difference that all the @xmath337 values are replaced by their corresponding mean @xmath338 , i.e. , @xmath339 @xcite .
the iteration equations for the conditional probabilities @xmath340 and @xmath341 at @xmath173 have the following expression @xcite @xmath342 \delta ( p_{i\rightarrow a}^+ - \hat{p } ( \{p_{j\rightarrow b}^+ \ } ) ) \ , \nonumber \\
\ \end{aligned}\ ] ] where @xmath343 denotes a spin configuration for the vertex set @xmath344 of constraint @xmath345 , and @xmath346 is the probability of a satisfying spin assignment @xmath347 for constraint @xmath345 given the spin value @xmath135 of vertex @xmath24 ( the probability @xmath348 is the mean probability of vertex @xmath349 taking spin value @xmath350 in the absence of constraint @xmath345 ) .
the cavity iterative equations for @xmath330 and @xmath351 can be solved by the population dynamics technique @xcite .
we are interested in the solution space property at a fixed value of the overlap value @xmath3 , so in the population dynamics simulation the magnitude of the coupling field @xmath66 is adjusted from time to time to ensure that the mean overlap value as expressed by with @xmath352 replaced by @xmath353 is equal to the specified value @xmath3 ( see , e.g. , @xcite ) .
the grand free - energy density @xmath354 of the model system , as defined by @xmath355 , has the following simplified expression at @xmath173 : @xmath356 the mean free energy density @xmath357 of a gibbs state is expressed as @xmath358 where @xmath359 and @xmath360 are , respectively , the free energy increase caused by vertex @xmath24 and constraint @xmath22 .
these two free energy increases are expressed by the following expressions at @xmath173 : @xmath361 \times \nonumber \\ & \ln\left ( \sum\limits_{\sigma_i } e^{x \sigma_i } \prod\limits_{a\in \partial i } \frac{1 + \sigma_i \prod_{j\in \partial a\backslash i } ( 2 p_{j\rightarrow a}^+ -1)}{2 } \right ) \ , \label{eq : d - f - i}\end{aligned}\ ] ] and @xmath362 \ln\left(\frac{1+\prod_{i\in \partial a } ( 2 p_{i\rightarrow a}^+ -1)}{2}\right ) \ .
\label{eq : d - f - a}\ ] ] in ) , @xmath363 is expressed as @xmath364 } { e^{x } \prod\limits_{a\in \partial i } \left [ \frac{1 + \prod\limits_{j\in \partial a \backslash i } ( 2 \overline{p}_{j\rightarrow a}^+-1)}{2 } \right ] + e^{-x } \prod\limits_{a\in \partial i } \left [ \frac{1- \prod\limits_{j\in \partial a\backslash i } ( 2 \overline{p}_{j\rightarrow a}^+ -1)}{2 } \right ] } \ , \ ] ] and @xmath365 .
the probability @xmath366 in is expressed as @xmath367 at @xmath173 , the complexity @xmath368 , the total free entropy density @xmath369 , and the mean entropy density of a solution community , @xmath370 , are calculated to be @xmath371 the equality @xmath372 holds at @xmath173 .
it appears that at some overlap values @xmath3 , more than one mean - field 1rsb solutions can be produced by the population dynamics simulation at @xmath173 .
these different mean - field solutions have the same value of @xmath369 but different values of @xmath368 and @xmath370 . in such a case
we choose the mean - field solution with the largest value of @xmath368 ( similar situations of multiple mean - field solutions were observed earlier in the random @xmath1-sat problem at @xmath194 @xcite ) . | we demonstrate through two case studies , one on the @xmath0-spin interaction model and the other on the random @xmath1-satisfiability problem , that a heterogeneity transition occurs to the ground - state configuration space of a random finite - connectivity spin glass system at certain critical value of the constraint density . at the transition point , exponentially many configuration communities emerge from the ground - state configuration space , making the entropy density @xmath2 of configuration - pairs a non - concave function of configuration - pair overlap @xmath3 .
each configuration community is a collection of relatively similar configurations and it forms a stable thermodynamic phase in the presence of a suitable external field .
we calculate @xmath2 by the replica - symmetric and the first - step replica - symmetry - broken cavity methods , and show by simulations that the configuration space heterogeneity leads to dynamical heterogeneity of particle diffusion processes because of the entropic trapping effect of configuration communities .
this work clarifies the fine structure of the ground - state configuration space of random spin glass models , it also sheds light on the glassy behavior of hard - sphere colloidal systems at relatively high particle volume fraction . |
with the advent of the five - year wmap data @xcite the inflationary paradigm @xcite ( henceforth called _ canonical inflation _ ) has been confirmed as the most successful candidate for explaining the physics of the very early universe @xcite .
the very rapid acceleration period generated by canonical inflation has solved some of the puzzles of the standard cosmological model , such as the horizon , flatness and entropy problems . however , existing models for inflation are phenomenological in character , and a fundamental explanation of inflation is still missing . also , the scalar field responsible for the inflationary expansion - the _ inflaton _ - is generically highly fine - tuned . over the past few years developments in string theory have shed new light on these two problems of canonical inflation .
string theory predicts a broad class of scalar fields associated with the compactification of extra dimensions and the configuration of lower - dimensional branes moving in a higher - dimensional bulk space .
this fact gave rise to some phenomenologically viable inflation models , such as the kklmmt scenario @xcite , racetrack inflation @xcite , roulette inflation @xcite , and the dirac - born - infeld ( dbi ) scenario @xcite .
in particular , from a phenomenological point of view , the dbi scenario has a very interesting and far - reaching feature : being a special case of a larger class of inflationary models with non - canonical lagrangians called _ k - inflation _
@xcite , the dbi model possesses a varying speed of sound .
this is a far - reaching feature because , in this case , slow - roll can be achieved via a low sound speed instead of from dynamical friction due to expansion which , in turn , leads to substantial non - gaussianity @xcite .
also , dbi inflation admits several exact solutions to the flow equations ( first introduced in @xcite for the canonical case , and generalized in @xcite for dbi inflation ) , as discussed in @xcite . in this paper
we derive a family of non - canonical models characterized by a power - law in the inflaton field @xmath4 for the speed of sound @xmath5 and the flow parameter @xmath6 .
we then derive the forms of the associated inflationary potentials , and group them according to the classification scheme introduced in ref .
@xcite , in order to have a better physical picture of the solutions obtained .
we later show that some particular cases possess spectral indices in agreement with the current observational values , and establish the limits for the speed of sound in order that their tensor - to - scalar ratio correspond to the observed values .
the outline of the paper is as follows : in section [ sec : dbiinflation ] we review the dbi inflation and the flow formalism for non - canonical inflation with time - varying speed of sound . in section [ sec : powerspec ] we review the tools needed to calculate the scalar and tensor spectral indices for slow - roll . in section [ sec : powerlawmodel ] we introduce the main features of our model and derive general solutions , postponing the discussion of their physical properties to section [ sec : inflpotentials ] , where we extend the concept of large - field , small - field , hybrid and exponential potentials of canonical inflation to the non - canonical case . in section [ sec : application ] we compare the observational predictions of a set of non - canonical large - field models with wmap5 data , and show that there is a generic correlation between small sound speed ( and therefore significant non - gaussianity ) and a low tensor / scalar ratio .
we present conclusions in [ sec : conclusions ] .
in warped d - brane inflation ( see @xcite and @xcite for a review ) , inflation is regarded as the motion of a d3-brane in a six - dimensional throat " characterized by the metric @xcite @xmath7 where @xmath8 is the warp factor , @xmath9 is a sasaki - einstein five - manifold which forms the base of the cone , and @xmath10 is the radial coordinate along the throat . in this case , the inflaton field @xmath4 is identified with @xmath10 as @xmath11 , where @xmath12 is the brane tension .
the dynamics of the d3-brane in the warped background ( [ eq : dbimetric ] ) is then dictated by the dbi lagrangian @xmath13 where @xmath14 is the background spacetime metric , @xmath15 is the inverse brane tension , @xmath16 is an arbitrary potential , and @xmath17 is the kinetic term .
we assume that the background cosmological model is described by the flat friedmann - robertson - walker ( frw ) metric , @xmath18 .
dbi inflation is a special case of _ k - inflation _ , characterized by a varying speed of sound @xmath19 , whose expression is given by @xmath20 where the subscript @xmath21 " indicates a derivative with respect to the kinetic term .
it is straightforward to see that in the model ( [ dbilagrangian ] ) the speed of sound is given by @xmath22 in those models it is convenient to introduce an analog of the lorentz factor , related to @xmath23 by : @xmath24 we next introduce the generalization of the inflationary flow hierarchy for k - inflation models @xcite .
the two fundamental flow parameters are @xmath25 where the prime indicates a derivative with respect to @xmath4 .
an infinite hierarchy of additional flow parameters can be generated by differentiation , and are defined as follows : @xmath26 where @xmath27 is an integer index .
it is convenient to express the flow parameters ( [ eq : flowparams ] ) in terms of the number of e - folds before the end of inflation , which is defined in the following way : @xmath28 where @xmath29 indicates the value of the inflaton field at the end of inflation .
the definition above indicates that the number of e - folds increases as we go backward in time , so that it is zero at the end of inflation .
we can also determine the scale factor , @xmath30 , in terms of @xmath31 in a straightforward way : since @xmath32 , we simply have @xmath33 in terms of @xmath31 , it is easy to see that the key flow parameters @xmath6 and @xmath34 , as defined in ( [ defepsilon1 ] ) and ( [ defs1 ] ) , assume the following equivalent form @xmath35 so that we may derive a set of first - order differential equations @xmath36 + { } ^{\ell + 1}\lambda , \cr \frac{d s}{d
n } & = & -s \left(2 s + \epsilon - \eta\right ) + \epsilon \rho,\cr \frac{d \rho}{d n } & = & -2 \rho s + { } ^2 \alpha,\cr & \vdots & \cr \frac{d { } ^\ell \alpha}{d n } & = & -{}^\ell \alpha \left[\left(\ell + 1\right ) s + \left(\ell - 1\right ) ( \epsilon-\eta)\right ] + { } ^{\ell + 1 } \alpha.\end{aligned}\ ] ] the dynamics of the inflaton field can be completely described by this hierarchy of equations , which are equivalent to the hamilton - jacobi equations @xcite
@xmath37 where @xmath38 is the reduced planck mass , and @xmath39 ^ 2}.\ ] ] in terms of the flow parameter @xmath6 , the potential , @xmath16 , and the inverse brane tension , @xmath40 , can be written as @xmath41 and @xmath42 respectively .
our strategy in this paper is to generate inflationary solutions by _ ansatz _ for the flow parameters @xmath6 and @xmath34 , for which the forms of the functions @xmath43 and @xmath44 are determined . in the next section ,
we briefly review the generation of perturbations in non - canonical inflation models .
scalar and tensor perturbations in k - inflation were first studied in ref .
@xcite , where it was shown that the quantum mode function @xmath45 obeys the equation of motion @xmath46 where @xmath47 is the comoving wave number , a prime denotes a derivative with respect to conformal time @xmath48 , and @xmath49 is a variable given by @xmath50 it is convenient to change the conformal time , @xmath51 , to @xmath52 so that @xmath53,\ ] ] where @xmath54 we can also show that @xmath55 where @xcite @xmath56 substituting ( [ dery ] ) and ( [ zf ] ) into ( [ modeeq ] ) , we find @xmath57 u_k = 0,\ ] ] which is an exact equation , without any assumption of slow - roll . we now proceed to solve equation ( [ exactmode ] ) to first - order in slow - roll . in this case , this equation takes the form @xmath58 u_k = 0,\end{aligned}\ ] ] whose solution is @xcite @xmath59 where @xmath60 is a hankel function , and @xmath61 the power spectra for scalar and tensor perturbations are given respectively by @xmath62 so that , for the solution ( [ uksolsr ] ) , @xmath63\left.\frac{h^2}{2\pi \dot\phi}\right|_{k=\gamma a h}\end{aligned}\ ] ] where @xmath64 is euler s constant . from the definitions of the scalar and tensor spectral indices ,
@xmath65 we find @xmath66 valid to first - order in the slow - roll limit .
the tensor - to - scalar ratio is given by @xcite @xmath67 which , to first - order in slow - roll , gives the well - known result @xmath68 since in this paper we will work only to the first - order in slow - roll approximation , we make use of expression ( [ stratio ] ) to compute the tensor / scalar ratio .
it is immediately clear from this expression that small @xmath19 has the generic effect of suppressing the tensor / scalar ratio .
the usual approach to the construction of a model of inflation normally starts with a choice of the inflaton potential , @xmath16 ; then , all the flow parameters are derived , and the dynamical analysis is performed . in this work we adopt the reverse procedure : we first look for the solutions to the differential equation satisfied by the hubble parameter @xmath69 , @xmath70 and only afterwards derive the form of the potential
. equation ( [ eqhubble ] ) can be easily derived from the definition of the flow parameter @xmath6 , given by ( [ defepsilon1 ] ) , and the sign ambiguity indicates in which direction the field is rolling .
notice that in order to solve equation ( [ eqhubble ] ) we must know the form of the functions @xmath71 and @xmath72 ; we choose them to be power - law functions of the inflaton field , @xmath73 where @xmath74 is the value of the lorentz factor at the end of inflation , and @xmath2 and @xmath3 are constants .
another case worth studying appears when @xmath6 is _ constant _ , so that @xmath75 with the same parametrization for @xmath76 .
we have kept @xmath29 for the following reason : in the @xmath77 dbi model the inflaton field rolls down from the tip of the throat toward the bulk of the manifold with increasing speed of sound ; then , when the field enters the bulk @xmath19 becomes equal to @xmath78 , and then inflation ends " . in the @xmath79 case the behavior is the opposite , that is , the field evolves away from the bulk and reaches the tip when @xmath80 . to reproduce both cases
we could have set @xmath81 from the onset , so that @xmath82 , as required ; also , @xmath83 in the canonical limit , that is , when @xmath84 .
however , by taking @xmath74 arbitrary , we also reproduce the non - canonical models with constant speed of sound introduced by spalinski @xcite .
it is clear that in the latter case @xmath74 does not refer necessarily to the end of inflation , so that if we take @xmath85 it does not mean a superluminal propagation ( as would be the case if @xmath86 ) .
bearing this distinction in mind we can use the same notation unambiguously .
when @xmath87 , substituting ( [ powerepsilon ] ) and ( [ powercs ] ) into ( [ eqhubble ] ) , we see that the hubble parameter takes the form @xmath88,\end{aligned}\ ] ] where we have defined @xmath89 and @xmath90 accounts for the sign ambiguity appearing in ( [ eqhubble ] ) .
when @xmath6 is constant , the solution to ( [ eqhubble ] ) reads @xmath91,\end{aligned}\ ] ] where the integral @xmath92 is the same as in ( [ eqhubble2 ] ) .
it is clear that the integral ( [ eqhubble2 ] ) admits two distinct solutions : a logarithmic one when @xmath93 , and power - law for @xmath94 .
these two solutions will give rise to different classes of inflationary potentials , which we shall address in the next subsections . to conclude this section we derive the general formula for the number of e - folds . for @xmath87
this expression can be determined by equations ( [ eq : numefolds ] ) , ( [ powerepsilon ] ) , and ( [ powercs ] ) , so that @xmath95 where @xmath96 if @xmath97 we must use the parametrization ( [ powerepsilonrep ] ) , so that expression ( [ efolds1 ] ) changes to @xmath98 where @xmath99 in the next section , we discuss particular cases of this general class of solutions . for this class of solutions ,
the parameters @xmath2 and @xmath3 are related by @xmath100 notice that the case @xmath97 implies @xmath101 , which corresponds to the case where the flow parameters @xmath6 and @xmath34 are constant @xcite .
evaluating the integral ( [ eqhubble2 ] ) and substituting it into ( [ eqhubble1 ] ) , we find @xmath102 where the exponent @xmath103 is determined by @xmath104 let us now analyze the sign ambiguity appearing in ( [ sol1defp ] ) .
we first write the expression ( [ sol1hubpar ] ) as @xmath105 then , for @xmath106 , we have , from ( [ powerepsilon ] ) , in the slow - roll limit , @xmath107 which implies @xmath108 and @xmath109 for @xmath110 ( _ large - field limit _ ) , so that @xmath111 . since the weak - energy condition implies that @xmath112 , we have @xmath113 , and then @xmath114 . therefore , from ( [ sol1defp1 ] ) , @xmath106 implies @xmath115 . from definition ( [ sol1defp ] )
we see that @xmath115 implies @xmath116 if @xmath117 , and @xmath118 if @xmath110 .
conversely , if @xmath119 , we have , from ( [ powerepsilon ] ) , in the slow - roll limit , @xmath120 which implies @xmath121 and @xmath122 for @xmath110 ( _ small - field limit _ ) , so that @xmath123 .
again , @xmath114 , so that , from ( [ sol1defp1 ] ) , @xmath119 implies @xmath124 . from definition ( [ sol1defp ] )
we see that @xmath124 implies @xmath116 if @xmath110 , and @xmath118 if @xmath117 .
therefore , what really distinguishes the models is the sign of the exponent of @xmath2 ; the sign of @xmath90 simply dictates which direction the field is rolling in , exactly as in canonical inflation .
we are left with two distinct models , whose properties are summarized in table [ t1 ] .
.the sign rule for models with @xmath93 and @xmath110 . [ cols="^,^,^,^,^",options="header " , ]
in this section we study some applications of the large - field models derived in section [ subsec : largefield ] .
we choose this class of non - canonical potentials because the expressions for the scalar spectral index , the tensor / scalar ratio and the level of non - gaussianity are particularly simple , depending on two parameters solely , @xmath103 and @xmath3 .
let us first derive an expression for the flow parameter @xmath6 in terms of @xmath31 . from ( [ lfieldeps ] ) and ( [ lfieldefolds ] ) we find @xmath125 the other two flow parameters @xmath34 and @xmath126
are given by @xmath127 @xmath128 where we have used ( [ defs1 ] ) , the first expression of ( [ eq : flowparams ] ) , plus ( [ powercs ] ) , ( [ lfieldhub ] ) and ( [ aplfieldepsn ] ) .
inserting ( [ aplfieldepsn ] ) , ( [ aplfields ] ) and ( [ aplfieldeta ] ) into ( [ scalarspecindex ] ) , we find , in the slow - roll limit , @xmath129 the expression for the speed of sound in terms of @xmath31 can be calculated in the same way : we use ( [ powerepsilon ] ) , ( [ powercs ] ) and ( [ lfieldefolds ] ) , so that @xmath130^{\beta/(\beta+2)}.\ ] ] through the use of ( [ stratio ] ) , ( [ aplfieldepsn ] ) and ( [ aplfieldcsn ] ) we can derive a general expression for the tensor / scalar ratio , which is given by @xmath131^{2(\beta+1)/(\beta+2)}.\ ] ] the expression for the level of non - gaussianity @xmath132 is given by @xcite @xmath133 which can be easily evaluated by using expression ( [ aplfieldcsn ] ) .
therefore , the tensor / scalar ratio will have a power - law dependence as well , with exponent @xmath134 , which means that , for a given value of @xmath103 , a larger @xmath3 corresponds to a smaller @xmath10 . since @xmath135 for non - canonical large - field models , we have , from ( [ gammafac ] ) and ( [ powercs ] ) , that @xmath136 ; then , fields rolling with _ slower _ speed of sound would produce _ lower _ tensor / scalar ratios .
however , from ( [ aplfieldfnl ] ) , we see that @xmath132 depends on @xmath137 , and then a low speed of sound would produce a _ larger _ level of non - gaussianity ; then , for large - field models low-@xmath10 tensor modes are strongly correlated with the amplitude of non - gaussianity , as has been discussed in the reference @xcite for isokinetic inflation .
then , the suppression of tensor modes by a large amount of non - gaussianity is a feature shared by all non - canonical models with large - field polynomial potentials .
let us next make some predictions on the values of @xmath138 , @xmath10 and @xmath132 through expressions ( [ aplfieldns ] ) , ( [ aplfieldr ] ) and ( [ aplfieldfnl ] ) respectively , when the modes cross the horizon @xmath139 or @xmath140 e - folds before the end of inflation .
as we have discussed in section [ subsec : gensetting ] , we have set @xmath81 , which characterizes the end of inflation ; the results are depicted in figures [ fig : nsrfnln46 ] and [ fig : nsrfnln60 ] . in both figures , the top left plots refer to the variation of the scalar index in terms of @xmath3 for each value of @xmath103 .
the top right plots in figs . [ fig : nsrfnln46 ] and [ fig : nsrfnln60 ] show the corresponding tensor / scalar ratio . in these plots
we see that for larger values of @xmath103 and small @xmath3 the modes have large values of @xmath10 ( the observable lower bound is @xmath141 ) ; then , as @xmath3 increases , the speed of sound gets lower and , in consequence , the tensor / scalar ratio as well .
however , as shown in ( [ aplfieldfnl ] ) , a field rolling very slowly produces a large amount of non - gaussianity , as can be seen in the bottom left plots of figures [ fig : nsrfnln46 ] and [ fig : nsrfnln60 ] .
therefore , as was first discussed in the particular case of isokinetic inflation @xcite , the production of large non - gaussianity is strictly correlated with low tensor amplitudes , and this is a feature common to all large - field polynomial potentials .
this behavior is shown in the bottom right plots of figures [ fig : nsrfnln46 ] and [ fig : nsrfnln60 ] .
we next compare the results obtained with the current wmap5 data @xcite , @xcite .
the results are depicted in fig .
[ fig : nrplot ] for different values of @xmath103 .
straight lines indicate the different values of @xmath3 , with the left ( right ) extremity indicating the value of @xmath142 evaluated at @xmath143 ( @xmath144 ) .
as shown in @xcite , all canonical models with @xmath145 are ruled out by wmap5 data alone ; however , in the non - canonical case , figure [ fig : nrplot ] shows that the models with @xmath146 are also consistent with the observable data .
a field evolving with slow - varying speed of sound produces low - amplitude tensors , then pushing the values @xmath142 inwards the observable region .
however , a large amount of non - gaussianity is produced , which is a distinct signature of non - canonical large - field polynomial models and can be a powerful observable to discriminate among inflationary models .
in this paper we propose a general dbi model characterized by a power - law flow - parameter power - law flow parameter @xmath0 and speed of sound @xmath1 , where @xmath2 and @xmath3 are constants .
we show that this general model has distinct classes of solutions depending on the relation between @xmath2 and @xmath3 , and on the time evolution of the inflaton field .
these classes of solutions are summarized in tables [ t1 ] , [ t2 ] and [ t3 ] . in particular , we show that in the slow - roll limit the four well - known canonical potentials arise naturally in this general dbi model , having similar properties to their canonical counterparts , except that the speed of sound in general varies with time .
we also show that this general dbi model encompasses not only all the canonical models with the mentioned potentials , but other d - brane scenarios as well : the dbi model with constant speed of sound @xcite , with constant flow parameters @xcite , and isokinetic inflation @xcite .
the four non - canonical models are summarized in table [ t4 ] .
we also derive the expressions for the spectral index , tensor / scalar ratio and the amplitude of non - gaussianity for large - field potentials in the slow - roll limit .
we show that a low speed of sound suppresses the tensor / scalar ratio @xmath10 and produces a large amount of non - gaussianity , a feature already explored in the case of isokinetic inflation , and shown to be a general property of all large - field dbi models with polynomial potentials . unlike canonical inflation , where all polynomial models with @xmath145 are ruled out , the suppression of tensor modes in the non - canonical version allows a larger class of polynomial potentials to lie within the observable range ; also , the production of large amount of non - gaussianity is a distinct signature of these dbi large - field models , which can be a powerful observable to discriminate among inflationary models .
a. h. guth , phys .
d * 23 * , 347 ( 1981 ) .
a. d. linde , phys .
b * 108 * , 389 ( 1982 ) .
a. albrecht and p. j. steinhardt , phys .
* 48 * , 1220 ( 1982 ) .
w. h. kinney , e. w. kolb , a. melchiorri , & a. riotto , phys .
d * 78 * , 087302 ( 2008 ) [ arxiv:0805.2966 ] .
s. kachru , r. kallosh , a. linde , j. m. maldacena , l. mcallister and s. p. trivedi , jcap * 0310 * ( 2003 ) 013 [ arxiv : hep - th/0308055 ] . j. j. blanco - pillado _ et al .
_ , jhep * 0411 * , 063 ( 2004 ) [ arxiv : hep - th/0406230 ] . j. r. bond , l. kofman , s. prokushkin and p. m. vaudrevange , phys . rev .
d * 75 * , 123511 ( 2007 ) [ arxiv : hep - th/0612197 ] .
e. silverstein and d. tong , phys .
d * 70 * , 103505 ( 2004 ) [ arxiv : hep - th/0310221 ] . c. armendariz - picon , t. damour and v. f. mukhanov , phys .
b * 458 * , 209 ( 1999 ) [ arxiv : hep - th/9904075 ] .
m. alishahiha , e. silverstein and d. tong , phys .
d * 70 * , 123505 ( 2004 ) [ arxiv : hep - th/0404084 ] .
x. chen , m. x. huang , s. kachru and g. shiu , jcap * 0701 * , 002 ( 2007 ) [ arxiv : hep - th/0605045 ]
. m. spalinski , phys .
b * 650 * , 313 ( 2007 ) [ arxiv : hep - th/0703248 ] .
r. bean , x. chen , h. v. peiris and j. xu , phys .
d , * 77 * , 023527 ( 2008 ) arxiv:0710.1812 [ hep - th ] .
m. loverde , a. miller , s. shandera and l. verde , jcap , * 4 * , 14 ( 2008 ) arxiv:0711.4126 [ astro - ph ] . w. h. kinney , phys . rev .
d * 66 * , 083508 ( 2002 ) [ arxiv : astro - ph/0206032 ] .
h. v. peiris , d. baumann , b. friedman and a. cooray , phys .
d * 76 * , 103517 ( 2007 ) arxiv:0706.1240 [ astro - ph ] . l. p. chimento and r. lazkoz , gen .
rel . and grav .
* 40 * , 2543 ( 2008 ) arxiv:0711.0712 [ hep - th ] .
m. spalinski , jcap * 4 * , 2 ( 2008 ) [ arxiv:0711.4326 ] .
k. tzirakis and w. h. kinney , phys .
d * 77 * , 010 ( 2008 ) [ arxiv:0712.2043 ] . l. mcallister and e. silverstein , gen
. rel .
grav . * 40 * , 565 ( 2008 ) [ arxiv:0710.2951 [ hep - th ] ] . j. m. cline , arxiv : hep - th/0612129 . i. r. klebanov and m. j. strassler , jhep * 0008 * , 052 ( 2000 ) [ arxiv : hep - th/0007191 ]
. m. spalinski , jcap * 0704 * , 018 ( 2007 ) [ arxiv : hep - th/0702118 ] .
j. garriga and v. f. mukhanov , phys .
b * 458 * , 219 ( 1999 ) [ arxiv : hep - th/9904176 ] .
s. e. shandera and s. h. tye , jcap * 0605 * ( 2006 ) 007 [ arxiv : hep - th/0601099 ] .
b. a. powell , k. tzirakis and w. h. kinney , jcap * 0904 * , 019 ( 2009 ) [ arxiv:0812.1797 [ astro - ph ] ] .
w. h. kinney and k. t. mahanthappa , phys .
d * 53 * , 5455 ( 1996 ) [ arxiv : hep - ph/9512241 ] .
k. freese , j. a. frieman and a. v. olinto , phys .
* 65 * ( 1990 ) 3233 . | we study dbi inflation based upon a general model characterized by a power - law flow parameter @xmath0 and speed of sound @xmath1 , where @xmath2 and @xmath3 are constants .
we show that in the slow - roll limit this general model gives rise to distinct inflationary classes according to the relation between @xmath2 and @xmath3 and to the time evolution of the inflaton field , each one corresponding to a specific potential ; in particular , we find that the well - known canonical polynomial ( large- and small - field ) , hybrid and exponential potentials also arise in this non - canonical model .
we find that these non - canonical classes have the same physical features as their canonical analogs , except for the fact that the inflaton field evolves with varying speed of sound ; also , we show that a broad class of canonical and d - brane inflation models are particular cases of this general non - canonical model .
next , we compare the predictions of large - field polynomial models with the current observational data , showing that models with low speed of sound have red - tilted scalar spectrum with low tensor - to - scalar ratio , in good agreement with the observed values .
these models also show a correlation between large non - gaussianity with low tensor amplitudes , which is a distinct signature of dbi inflation with large - field polynomial potentials . |
one of the advantages of superspace is that it makes supersymmetry manifest .
any action which is invariant under arbitrary diffeomorphisms of curved superspace automatically yields a locally supersymmetric component action .
in addition , it offers the possibility to construct general couplings from lagrangians of ( essentially ) arbitrary functional form . for these reasons , higher derivative supersymmetric actions , especially those coupled to supergravity , can be quite efficiently constructed in superspace .
such actions have been of interest recently @xcite .
we review below a new class of higher derivative actions that was constructed directly in superspace in our recent paper @xcite .
the supergeometry which enables this construction is an off - shell formulation for general @xmath0 supergravity - matter couplings in four dimensions @xcite , which allows a curved - space extension of @xmath0 projective superspace @xcite . for our purposes ,
the relevant detail is that the required superspace is @xmath1 .
the curved supermanifold @xmath2 is parametrized by local coordinates @xmath3 , @xmath4 .
its geometry is described by covariant derivatives @xmath5 where @xmath6 is the supervielbein ; @xmath7 and @xmath8 are the lorentz generators and superconnections respectively ; and @xmath9 and @xmath10 are respectively the @xmath11 generators and superconnections
. the auxiliary manifold @xmath12 is parametrized by an isotwistor @xmath13 defined modulo @xmath14 for @xmath15 .
all superfields and operators are required to have fixed homogeneity in @xmath16 . for a more extensive discussion of the supergeometry
, we refer the reader to the original references @xcite .
our higher derivative construction is based on a duality between two basic off - shell representations of @xmath0 supersymmetry the tensor multiplet and the vector multiplet . in curved superspace
, the tensor multiplet is described by its field strength @xmath17 , which is a real isotriplet superfield , @xmath18 , constrained to obey @xmath19 .
these conditions are solved in terms of an unconstrained chiral prepotential @xmath20 , @xmath21 , as @xmath22 the superfield @xmath23 and its conjugate @xmath24 are components of the superspace torsion tensor .
the superspace torsion and the algebra of covariant derivatives are given in @xcite .
the vector multiplet , on the other hand , is described by its field strength @xmath25 , which is a chiral superfield , @xmath26 , obeying the bianchi identity @xmath27 such a superfield is called reduced chiral . comparing this equation to ,
it is obvious that the superfield @xmath28 is a _
tensor multiplet .
moreover , it is clear that the tensor prepotential @xmath20 is defined only up to shifts @xmath29 where @xmath30 is a reduced chiral superfield .
this is just the superfield version of the component gauge transformation @xmath31}$ ] which leaves the three - form field strength @xmath32}$ ] invariant . within the projective superspace formulation of conformal supergravity @xcite ,
both of these multiplets find a natural realization .
the tensor multiplet is described in terms of a real @xmath33 multiplet @xmath34 obeying @xmath35 where @xmath36 and @xmath37 .
the vector multiplet can be described by a tropical prepotential @xmath38 , which is a real projective multiplet , @xmath39 , of weight zero , @xmath40 .
the reduced chiral superfield @xmath25 is related to @xmath38 by @xmath41 the fixed isotwistor @xmath42 introduced here is subject only to the condition that @xmath43 along the contour @xmath44 in @xmath12 ; the construction proves to be independent of the choice of @xmath42 .
the complementary nature of these properties allows us to generate tensor multiplets from vector multiplets and vice - versa .
consider a system of @xmath45 abelian vector multiplets with field strengths @xmath46 , @xmath47 . given a holomorphic homogeneous function @xmath48 of degree one ( to guarantee the correct super - weyl transformation of @xmath49 )
, we can define a composite tensor multiplet @xmath50 this is a standard construction and a trivial application of using the chiral function @xmath51 as a composite tensor prepotential .
conversely , we may take a system of @xmath52 tensor multiplets described by their field strengths @xmath53 , with @xmath54 , with @xmath55 the corresponding @xmath33 multiplets .
we may similarly construct a composite vector prepotential from any function @xmath56 that is real and homogeneous of degree zero . applying leads immediately to a composite field strength @xmath57 , @xmath58 these two observations together enable us to take any action involving tensor and/or vector multiplets and convert it to a higher derivative action by identifying one or both multiplets as composite .
iterating the procedure leads to increasingly complex higher derivative interactions .
we first demonstrate how this procedure allows the construction of general two - derivative interactions of tensor multiplets ; then we extend to two - derivative interactions of real @xmath59 multiplets ; and then finally we extend the argument to higher derivative theories .
consider the simplest interaction between vector and tensor multiplets : the supersymmetric @xmath60 coupling .
this action can be written either as a chiral superspace integral involving the tensor prepotential @xmath20 and vector field strength @xmath25 , @xmath61 or as a projective superspace integral involving the tensor field strength @xmath62 and the vector prepotential @xmath38 , @xmath63 this action is topological and involves no propagating degrees of freedom ; however , by identifying one or both multiplets as composite , we may construct nontrivial actions .
for example , the unique superconformal action for a single tensor multiplet , known as the improved tensor multiplet action @xcite , can be written in either form . in projective superspace , it is given by where @xmath38 is replaced by the composite vector prepotential @xmath64 .
the arctic multiplet @xmath65 appearing here can be shown to be a pure gauge degree of freedom .
equivalently , the action is given by with the corresponding replacement @xmath66 .
this form is simplest to use in principle since the techniques to evaluate the superspace integral are well known : we simply require @xmath57 . to construct it
, we must evaluate the contour integral in .
the way this is usually done is to make a choice for @xmath42 and then to evaluate the contour explicitly , _
e.g. _ by representing @xmath67 .
this breaks manifest @xmath11 covariance .
it is possible , however , to keep @xmath11 covariance along the way @xcite .
one does this by first evaluating the spinor derivatives on their argument and exploiting the contour integral to rewrite the integrand as @xmath68 where @xmath69 and @xmath70 .
this has the advantage that all dependence on the auxiliary isotwistor @xmath42 has vanished .
the remaining contour integral can be done in an @xmath11 covariant way @xcite , with the result recast as @xmath71 it is remarkable that this expression is chiral and obeys .
( an equivalent but less compact expression for @xmath57 , obtained by brute force , was given in @xcite . )
superconformal actions involving several tensor multiplets can be written in an analogous fashion . in projective superspace , they are given by a general real weight - two projective lagrangian @xmath72 which is homogeneous of degree - one .
such a lagrangian can always be rewritten ( though not uniquely ) as @xmath73 the corresponding projective superspace action can then be converted into a chiral action analogous to , @xmath74 with @xmath75 given by @xmath76 where @xmath77 and @xmath78 are functions of @xmath79 given by @xmath80 this construction appeared originally in flat superspace @xcite .
its component form coupled to conformal supergravity appeared in @xcite .
we may extend the above procedure by considering @xmath59 multiplets . as an example , consider a more general projective lagrangian of the form @xmath81 where @xmath82 is a real @xmath59 multiplet obeying @xmath83 the composite vector multiplet we construct from the above expression has the prepotential @xmath84 .
as before , it is possible to evaluate the spinor derivatives in such a way as to completely eliminate the auxiliary isotwistor @xmath42 . the contour integral can then be evaluated in an @xmath11 covariant way and the result recast as @xmath85 where @xmath86 an interesting application of this result is for the case @xmath87 with the lagrangian @xmath88 . this is a curved - superspace version of that proposed in @xcite to describe the classical universal hypermultiplet @xcite . applying our general formula leads to @xmath89 the construction of the component lagrangian
can then be carried out by conventional means .
we now turn to our main goal : the construction of higher derivative actions .
this can be done quite straightforwardly by iterating the above procedure . begin with a set of tensor multiplets @xmath79 and abelian vector multiplets @xmath46 .
we construct a set of degree - zero functions @xmath90 of the tensor multiplets @xmath91 , which lead to the composite vector multiplet field strengths @xmath92 using both sets of vector multiplets , we introduce a set of degree - one holomorphic functions @xmath93 which can be used to construct composite tensor multiplet field strengths @xmath94 then we may introduce these composite tensor multiplets into the functions @xmath90 in , leading to @xmath95 this new vector multiplet can be used to construct new composite tensor multiplets and so on and so forth , with each iteration adding two spinor derivatives ( or one vector derivative ) to the interaction .
this method of constructing higher derivative actions should be contrasted with the more traditional way of generating higher derivative structures using the chiral projection operator @xmath96 , which is a curved space generalization of @xmath97 . given any scalar superfield @xmath98 which is inert under super - weyl transformations , its descendant @xmath99 is chiral and super - weyl weight two . given a vector multiplet @xmath25 that is nowhere vanishing
, we can then define the chiral scalar @xmath100 which is invariant under the super - weyl transformations .
we can then construct @xmath101 , and so on and so forth .. ] using these composite chiral operators , one may construct higher derivative actions involving chiral superspace actions . however , it is usually possible to convert the chiral superspace action into an integral over the whole superspace by eliminating one of the chiral projection operators .
schematically , if @xmath102 for some chiral superfield @xmath103 and a well - defined local and gauge - invariant operator @xmath104 , then @xmath105 thus , higher derivative actions of this type are invariably most naturally written as integrals over the entire superspace and are not intrinsically chiral .
this has important ramifications for perturbative calculations , where non - renormalization theorems place strong restrictions on intrinsic chiral lagrangians .
the constructions we are considering are interesting partly because they include higher derivative terms which _ can not _ be written as full superspace integrals without introducing prepotentials .
we take the example of a projective lagrangian @xmath106 of several tensor multiplets @xmath91 and one composite tensor multiplet @xmath107 the action can be rewritten @xmath108 which is a special case of a more general action @xmath109 the complex integrand @xmath110 is required to be annihilated by only 1/4 of the spinor derivatives .
such an action is the locally supersymmetric version of @xmath111 in other words , it is an integral over 3/4 of the grassmann coordinates of superspace .
supersymmetric yang - mills effective actions , which are given as an integral over 3/4 of superspace , have been discussed in @xcite . ] because this class of higher derivative action can not be written as an integral over the full superspace without introducing gauge - dependent prepotentials , an @xmath0 extension of the well known @xmath112 non - renormalization theorems likely applies . in order for the action to be super - weyl invariant ,
@xmath110 must be degree 1 in @xmath46 and degree zero in @xmath91 .
the simplest example is @xmath113 where @xmath114 .
( in this case , the arctic multiplets are again pure gauge degrees of freedom . )
a straightforward argument shows that the action can be rewritten @xmath115 which is the supersymmetric generalization of @xmath116 thus the actions describe three - derivative couplings in the bosonic sector .
needless to say , the integrand @xmath110 can be generalized to include composite vector and tensor multiplets , which leads to four - derivative and higher interactions . for simplicity
we have restricted our higher - derivative discussion to vector and tensor multiplets . however , as emphasized in @xcite , the composite vector multiplets may be built out of not only tensor multiplets , but also more general @xmath59 multiplets and even polar multiplets .
the corresponding chiral superspace action can then be used to give a new and very general class of higher - derivative supergravity - matter couplings .
10 i. antoniadis , s. hohenegger , k. s. narain and t. r. taylor , nucl .
b * 838 * , 253 ( 2010 ) [ arxiv:1003.2832 [ hep - th ] ] .
b. de wit , s. katmadas and m. van zalk , jhep * 1101 * , 007 ( 2011 ) [ arxiv:1010.2150 [ hep - th ] ] .
d. butter and s. m. kuzenko , jhep * 1103 * , 047 ( 2011 ) [ arxiv:1012.5153 [ hep - th ] ]
. s. m. kuzenko , u. lindstrm , m. roek and g. tartaglino - mazzucchelli , jhep * 0809 * , 051 ( 2008 ) [ arxiv:0805.4683 ] ; s. m. kuzenko , nucl .
b * 810 * , 135 ( 2009 ) [ arxiv:0807.3381 [ hep - th ] ] ; s. m. kuzenko , u. lindstrm , m. roek and g. tartaglino - mazzucchelli , jhep * 0908 * , 023 ( 2009 ) [ arxiv:0905.0063 [ hep - th ] ] .
a. karlhede , u. lindstrm and m. roek , phys .
b * 147 * , 297 ( 1984 ) ; u. lindstrm and m. roek , commun . math .
phys . * 115 * , 21 ( 1988 ) ; u. lindstrm and m. roek , commun .
* 128 * , 191 ( 1990 ) .
b. de wit and f. saueressig , jhep * 0609 * , 062 ( 2006 ) [ arxiv : hep - th/0606148 ] .
n. berkovits and w. siegel , nucl .
b * 462 * , 213 ( 1996 ) [ arxiv : hep - th/9510106 ] ; b. de wit , m. roek and s. vandoren , jhep * 0102 * , 039 ( 2001 ) [ arxiv : hep - th/0101161 ] .
i. l. buchbinder , s. m. kuzenko and a. a. tseytlin , phys .
d * 62 * , 045001 ( 2000 ) [ arxiv : hep - th/9911221 ] .
p. c. argyres , a. m. awad , g. a. braun and f. p. esposito , jhep * 0307 * , 060 ( 2003 ) [ arxiv : hep - th/0306118 ] . | using a recently developed off - shell formulation for general 4d @xmath0 supergravity - matter systems , we propose a construction to generate higher derivative couplings .
we address here mainly the interactions of tensor and vector multiplets , but the construction is quite general . for a certain subclass of terms , the action is naturally written as an integral over 3/4 of the grassmann coordinates of superspace . |
quantum imaging , which is based on the control of images in quantum systems , has been a topic of growing interest in recent years @xcite . for an ideal laser source used in a classical imaging system with a mean number of photons @xmath0 ,
the uncertainty in the number of photons is @xmath1 and the heisenberg uncertainty principle dictates that the uncertainty in phase varies as @xmath2 .
this uncertainty corresponds to the photon shot noise limit ( snl ) , a fundamental result of the poissonian statistics of coherent light sources . with recent advances in the demonstration of highly multi - spatial - mode quantum light sources
@xcite , it is now possible to take advantage of quantum correlations between individual coherence areas in order to perform imaging with sensitivity or resolution @xcite beyond classical limits and to implement parallel quantum information protocols @xcite .
in particular , quantum noise reduction ( qnr ) - or squeezed light - can be exploited in low - light - intensity imaging applications in order to achieve increased sensitivity and contrast beyond the snl by reducing the uncertainty in photon number at the expense of the uncertainty in phase .
other authors have recently demonstrated quantum imaging of individual images with sensitivity below the snl with both discrete photon sources @xcite and continuous variable sources @xcite .
however , rather than examining the excess noise properties present in twin beams as a way to estimate the geometry of a quantum image @xcite , we demonstrate that direct analysis of squeezing can be used to the same end . because any attenuation of twin beams results in a degradation of quantum correlations @xcite , previous research demonstrating imaging with sub - shot - noise level sensitivity with ccd cameras relied on expensive high quantum - efficiency ccd arrays @xcite . the ability to perform sub - shot - noise imaging with a more economical
, high quantum efficiency , single pixel detector with a dramatically reduced overall integration time will facilitate real - time high - sensitivity imaging in low - light situations . in this manuscript
, we demonstrate real - time control of the spatial modes in a squeezed light source by incorporating a spatial light modulator ( slm ) into the seed beampath prior to the generation of multi - spatial - mode squeezed light by four - wave mixing ( 4wm ) in @xmath3rb vapor .
this ` quantum movie projector ' exhibits real - time quantum noise reduction as a continuous series of images are flashed across the slm , thereby allowing for the analysis of the quantum noise reduction present in arbitrary spatial modes in real - time .
there is a limited literature that applies compressive sampling algorithms to pseudothermal @xcite and quantum @xcite ghost imaging , but there is currently no literature examining the high sensitivity imaging possible with the application of compressive sampling to squeezed light sources . with the addition of two slms after the rb vapor cell
, we measure the quantum noise reduction between macropixels within the twin beams , a significant step towards real - time compressive quantum imaging with a single pixel camera .
in addition , we demonstrate compressive sampling of the beam profiles using a customized sampling matrix suited to twin beam detection .
the squeezed light source used in this experiment was generated by a 4wm process in @xmath3rb vapor based on a double-@xmath4 system between the hyperfine ground states and the d1 excited states .
the hot @xmath3rb vapor in a 12.5 mm thick antireflection coated glass cell absorbs two photons from the pump beam ( weakly focused to a waist of 1 mm and denoted by ` p ' in fig .
1(b ) ) thereby generating a coherence between the two ground states .
the probe beam ( focused to a waist of 450 @xmath5 m and denoted by ` pr ' in fig .
1(b ) ) is red - shifted from the pump frequency by roughly 3 ghz via a double - pass acousto - optic - modulator in order to stimulate the re - emission of photons into the probe frequency with the simultaneous emission of photons into the conjugate beam ( ` c ' in fig .
1(b ) ) that is blue - shifted from the pump by 3 ghz . because of the strong amplitude correlations between these twin beams , the amplitude difference noise measured with a balanced photodiode and shown in fig .
1(c ) is below the photon shot noise limit ( snl ) for most sideband frequencies between 50 khz and 5 mhz .
all squeezing values reported in this manuscript were measured at a sideband frequency of 500 khz with 20 khz resolution bandwidth and 3 khz video bandwidth . in each case , the average of the 401 datapoints in each spectrum was used as the reported noise level .
the combined systematic and statistical uncertainty in each data point is 0.1 db . because of the presence of a buffer gas in the rb vapor cell used for this experiment , increased doppler broadening limited the maximum squeezing to 4.5 db below the snl .
vapor cells with no buffer gas demonstrate 15 - 20% less absorption of the probe field within the cell , yielding quantum noise reduction of greater than 9 db inferred with no losses . a digital micromirror device ( dmd ) comprising a 1024 x 768 array of square micromirrors with a pitch of 13.68 @xmath5 m provided intensity slm functionality , thereby yielding real - time control over what spatial modes undergo four - wave mixing in the vapor cell .
because the quantum noise reduction observed in individual coherence areas within an image is highly dependent on the pump - probe overlap , the observed squeezing is expected to be smaller for images with higher order spatial frequencies that would yield poor overlap at the fourier plane in the vapor cell .
this is a similar result to that seen when masks were used to imprint various images on the probe beam prior to four - wave mixing @xcite . in order to examine the quantum noise reduction in arbitrary images ,
two experimental designs were used .
first , the probe and conjugate were imaged with ccd cameras placed in the respective image planes , and flip mirrors were used as shown in fig .
1(a ) to measure the quantum noise reduction of each image with a balanced photodiode .
subsequently , two additional dmds replaced the cameras in the probe and conjugate image planes as shown in fig .
1(a ) , and the dmds were used to selectively pass macropixels of each beam to the photodiode in order to examine the feasibility of single pixel quantum imaging .
as an initial demonstration of the dependence of quantum noise reduction on the spatial modes present in the probe , various images were placed on the slm illustrated schematically in fig .
1(a ) and imprinted on the probe seed beam , and the image quality and quantum noise reduction of the twin beams generated by 4wm were recorded . for each image ,
the optics illustrated in fig .
1 were unchanged so that while the probe waist remained centered in the vapor cell , the probe diameter in the cell varied with the structure of the fourier image . compared with gaussian twin beams demonstrating 4.5 db of quantum noise reduction , the happy face twin beams illustrated in fig .
2 demonstrated 3.8 db of quantum noise reduction below the snl .
the spatial overlap between the pump and probe therefore effectively functions as a low - pass filter that removes higher frequency spatial modes from the fourier image , resulting in reduced squeezing in images containing significant high - spatial - frequency components .
this is made more clear in fig .
3 , where the quantum noise reduction associated with several other images is presented : checkerboard patterns possessing significant high frequency spatial modes demonstrated the least squeezing . in order to transition from the examination of quantum noise reduction in static images to real - time quantum imaging ,
a simple video of a spinning cross was implemented on the dmd prior to the rb vapor cell , with the center of the cross coincident with the center of the gaussian probe beam .
the quantum noise reduction observed in real - time on a spectrum analyzer was unchanged from the quantum noise reduction of a static cross image .
ringing in individual micromirrors resulted in the observation of noise peaks at isolated frequencies below 5 mhz within the qnr spectrum . in order to eliminate any noise due to micromirror ringing effects
, a 10 ms delay was introduced between the introduction of each image to the dmd and each qnr measurement .
a video of the probe and conjugate beam profiles recorded in the probe image plane38 cm from the vapor cell is shown in fig .
4(a)with the corresponding real - time quantum noise reduction plotted as a function of cross angle in fig .
the probe and conjugate image planes are separated in the propagation direction due to cross phase modulation with the pump near resonance ( the probe is near resonance while the conjugate is 6 ghz away ) .
the buffer gas in the vapor cell caused pressure broadening of the absorption resonances , which increased the cross phase modulation between the probe and pump fields compared to a pure rb cell . as a result
, the probe and conjugate image planes were separated by 483 cm . as can be seen in fig .
4(a ) the conjugate beam profile at the probe image plane is the fourier transform of a spinning cross , while the video of the conjugate beam profile recorded in the conjugate image plane and shown in fig .
4(c ) is clearly that of a spinning cross .
while the videos in fig . 4 are evidence of real - time quantum imaging , they were recorded asynchronously from the quantum noise reduction measurements , with the flip mirrors illustrated in fig .
1(a ) used to separately facilitate the recording of images or squeezing spectra . because the dmd shown in fig .
1(a ) provided us complete control over the images present in the quantum video , it was straightforward to demonstrate the reproducibility of these measurements , and it was clear which squeezing spectra corresponded to which images . in the case of more arbitrary real - time
quantum imaging applications , it will be necessary to synchronously measure both the quantum noise reduction and the spatial properties of the twin beams .
the use of slms that selectively pass either individual correlated coherence areas or macropixels of many coherence areas onto a balanced photodiode will allow for real - time quantum imaging with a single pixel camera , enabling myriad applications in low light , sub - shot - noise imaging .
the two dmds in the probe and conjugate image planes illustrated in fig .
1(a ) facilitated the measurement of quantum noise reduction in macropixels of individual coherence areas a significant step toward real - time single pixel quantum imaging with macroscopic photon numbers .
as a first demonstration of this capability , the probe beam profile was recorded by using the probe dmd and a photodiode as a rastered single pixel camera .
the left - most image in fig .
5 is the thresholded image of the probe beam profile that was recorded in this manner . when the thresholded image was placed on the probe dmd , and the conjugate dmd
was set to pass all spatial modes , with a variable neutral density filter introduced to proportionately attenuate the conjugate essentially acting as a bucket detector1.67 db of qnr was recorded .
the significant loss in squeezing compared with the 4.0 db of qnr recorded with mirrors in place of the two dmds is a result of the losses associated with the dmds @xcite : when the dmds reflect all spatial modes , 40% attenuation is observed in the zero order diffraction spot .
[ fig5 ] the predicted squeezing values in fig .
5 were calculated using the input - output relations for a _ single spatial mode _ phase - insensitive amplifier @xcite : @xmath6 , where @xmath7 is the total gain of the 4wm process ( approximately 4 in this experiment ) , and @xmath8 corresponds to the transmission of the vapor cell - dmd - detector system . while the dmd generally exhibits 60% diffraction efficiency into the zero order mode , additional attenuation occurs with each image placed on its surface , since some incident light falls onto pixels which are turned `` off '' ( set to reflect light away from the detector ) . as seen in the bottom of fig .
5 , slight variations in the image present on the probe dmd resulted in dramatic changes to the observed qnr . for each image
, the conjugate dmd passed all spatial modes and the conjugate power was proportionately attenuated in order to eliminate excess noise and optimize squeezing .
in all cases the conjugate attenuation that resulted in optimal squeezing was equal to the probe attenuation within 1% . because the predicted squeezing was provided for a single spatial mode , the losses are applied to the entire beam equally rather than to individual coherence areas . for certain images
the measured squeezing is much lower because non correlated coherence areas between the probe and conjugate are being detected .
this provides a novel technique for quantum imaging analogous to a recently reported technique that relied on measurements of excess noise made by homodyne detection to provide rudimentary quantum imaging without the use of a ccd camera @xcite .
in contrast to that manuscript , this technique allows us to approximate an object s size and shape via the direct examination of quantum noise reduction as a function of input image in the slm .
the strong dependence of the squeezing on the geometry present on the probe dmd is a result of the multi - spatial - mode nature of the quantum correlations present in the twin beams . in order to progress to true single pixel quantum imaging
, a simple solution is to selectively pass quantum correlated coherence areas with each dmd , resulting in a rastered single pixel quantum image . such techniques are slow , and acquiring the images with only one super pixel at a time would likely introduce sufficient attenuation to make the subsequent observation of squeezing very difficult .
compressive imaging resolves this issue with the utilization of sampling matrices containing many pixels . as a first step toward single pixel quantum imaging ,
preliminary rastered quantum imaging results were obtained by rastering a line instead of individual pixels . the line illustrated in fig .
6(a ) was centered on the location of a gaussian conjugate beam on the conjugate dmd in order to pass only those conjugate photons that were spatially coincident with the line . a line scaled in width by the ratio of the probe and conjugate image sizes at their respective image planes
was rastered across the probe dmd as shown in fig .
excess noise was observed for all line positions until they sufficiently overlapped corresponding areas in the probe and conjugate beams to make squeezing observable . the quantum noise reduction spectrum shown in fig .
6(c ) , recorded when both lines were centered on the twin beams , shows [email protected] db squeezing at 500 khz , indicating that this approach should be sufficient to allow for rastered quantum imaging .
correspondingly , these results indicate the feasibility of compressive quantum imaging when a sampling matrix having a comparable number and arrangement of pixels is selected . in order to demonstrate the effect of a reduction in macropixel size on the measured squeezing , this experiment was repeated with lines having half the width of the lines illustrated in figs .
6(a ) and 6(b ) ( corresponding to approximately one fourth of the @xmath10 beam diameter at the dmd ) .
in this instance , a line was centered on the probe beam , and a separate line was rastered across the conjugate dmd . as illustrated in fig .
6(d ) , excess noise was observed for all beam displacements except where the line was centered on the conjugate beam . the goal in this experiment was to show that images could be placed on the probe and conjugate dmds simultaneously in correlated locations such that squeezing was still observable .
this is important to demonstrate because the types of sampling matrices used in our compressive imaging algorithm correspond to shapes almost exactly like those shown in the bar image .
figure 7 shows an example of one such sampling matrix .
we have also perfomed initial compressive imaging with sampling matrices of this form .
an important question is the relation between super pixel size and coherence area . in simple terms ,
a coherence area acts as a single coherent beam .
if a single pair of correlated coherence areas in the two beams were isolated and studied , one would find that the quantum noise reduction as a function of transmission is effectively independent of the spatial structure of the attenuator .
that is , uniformly attenuating each beam would yield the same results as attenuating them with sharp apertures as long as each beam consisted of only one coherence area .
on the other hand , if the beams contain multiple coherence areas , then the noise reduction as a function of transmission would depend strongly on the spatial structure of the attenuation .
this effect is illustrated in fig .
_ @xcite found that experimentally maximizing the ratio of super pixel size to coherence area leads to reduced noise , which agrees with the theoretical findings of brambilla _ et al .
_ @xcite the coherence area size is estimated from the conjugate size inside the vapor cell and the beam size in the far field to be 1.4 mrad ( about 17% larger than in reference @xcite ) , while the angular beam diameter of the probe and conjugate was measured to be 2.4 mrad .
therefore approximately three to four coherence areas were present within the @xmath11 diameters of the probe and conjugate beams .
note that this is the number of coherence areas within the total angular bandwidth of the 4wm process that the nonzero intensity regions of the probe and conjugate overlap with , and the low intensity tails of the probe and conjugate contain yet more coherence areas beyond those located within the gaussain beam width .
the actual number of supported modes within the angular bandwidth of the four wave mixing process was found to be approximately 70 using a previously reported method@xcite .
the bar sizes were chosen to be approximately equal to half the angular diameter ( @xmath11 radius ) of each beam , which suggests that the bars used as macropixels in fig .
6(a ) and 6(b ) likely contained only individual coherence areas in the narrow dimension .
thus , the bar size chosen is likely close to the smallest size allowable to observe squeezing . however , the bar size can be smaller and still exhibit quantum noise reduction as a result of subsampling individual , corresponding coherence areas in each beam ( this case would look like simple attenuation for a single mode , and noise reduction would reduce to zero with complete attenuation ) .
the exact distribution of the coherence areas within the beams likely does not correspond to a striped shape ( but rather would correspond to the shape of the schmidt modes ) .
thus , the dramatic reduction in squeezing resulting from the choice of narrow bar sizes and shown in fig .
6(d ) is a result of subsampling individual coherence areas while introducing excess noise from uncorrelated coherence areas .
we note that determining the exact shape of the coherence areas may be possible using the dmds employed in this experiment by selectively rastering each beam with specific shapes and sizes for superpixels , but such a study is beyond the scope of the present work .
we can also determine the average power per coherence area illuminated by the probe and conjugate by multiplying the total beam power by the ratio of the coherence area size to total beam size .
the probe in fig .
6 had approximately 1.2 mw total power , corresponding to approximately 400 @xmath5w per coherence area .
the actual number is slightly less , because the nonzero intensity size of the beams on the dmd extended to approximately three to four times the beam width , meaning the total power was distributed among slightly more coherence areas .
nonetheless , this power per coherence area is in stark contrast to quantum imaging experiments with spontaneous parametric downconversion which rely on photon counting , and easily satisfies the condition of a large number of photons per coherence area needed for quantum imaging .
primary issues with performing rastered single pixel imaging include long acquisition times and low signal to noise ratios associated with small pixel sizes . both of these issues can be addressed with the increasingly popular technique of compressive imaging ( ci ) .
ci deals with the acquisition of images by recovering the coefficients in a sparse representation basis .
an image containing n pixels requires only m@xmath12n coefficients to provide a good approximation of the image x under the transform : @xmath13 where @xmath14 is the sparsifying basis and @xmath15 is the coefficient vector comprising no more than k significant nonzero values .
ci theory provides the framework to acquire such an image via the linear projection : @xmath16 where y is an m - dimensional measurement vector , a is an m - by - n sensing matrix , x is the n dimensional signal of interest , and @xmath17 is the acquisition noise .
ci is attractive for single pixel imaging applications because it allows for image reconstruction with only @xmath18 measurements , and the use of sparse arrays instead of individual pixels significantly increase the signal to noise ratio in applications where the dominant noise source is the shot noise . in addition
, real - time compressive imaging is now a viable technology @xcite .
optimization of ci for a given application requires two separate thrusts : the choice of appropriate sparse sampling matrices , and the choice of an appropriate reconstruction algorithm to facilitate the reconstruction of @xmath19 from @xmath20 . after comparing various permutations of random orthonormal matrices and hadamard matrices
it was determined that individual rows randomly sampled from block hadamard matrices @xcite modified using semilocal randomizers yielded an efficient and high quality image reconstruction @xcite .
reconstruction techniques including total variation with equality constraints from l1 magic @xcite , gradient projection @xcite , and total variation with augmented lagrangian @xcite were investigated for this application . ultimately , the total variation with equality constraints reconstruction algorithm was found to reliably reconstruct beam profiles with optimal signal to noise ratios .
figure 7 illustrates the 32x32 reconstructions of the probe and conjugate beam profiles generated from m=300 measurements by the total variation algorithm for a test image of the letter e. with no probe or conjugate dmd , the e twin beams demonstrated [email protected] db of squeezing . by choosing sampling matrices with appropriate pixel sizes in comparison to the probe and
conjugate coherence areas , similar to the demonstration in fig .
6 , it is possible to perform differential compressive sampling in order to acheive real - time single pixel compressive quantum imaging .
it is interesting to compare compressive imaging with other imaging methods such as rastering with ccd cameras or direct imaging using local oscillators .
a big advantage for compressive imaging can be seen in the use of a large number of pixels to obtain each data point , ratehr than a single pixel per measurement .
7 shows one type of sampling matrix which sends light to the detector whenever light is incident on the `` white '' part of the dmd .
a single pixel would correspond to a square approximately sixteen times smaller than the sampling matrix shown here .
further , the sampling matrices used in this experiment are scalable in size by selecting a block size for the block diagonal sparse array that generates the sampling matrices .
this allows the signal to noise ratio of each sample to be scaled as needed to overcome detector dark noise , for instance .
thus , compressive imaging can offer in principle better signal to noise ratios compared to ccd cameras for a given quantum efficiency and dark current noise .
compared to the homodyne detection method of imaging , compressive imaging offers in principle the same resolution and signal to noise ratio for a given transmission , but offers no phase information .
this is because a direct intensity measurement as performed in our experiments is analagous to a perfectly aligned homodyne detector with local oscillator phase locked to detect the amplitude quadrature .
however , the ease of alignement using dmd s is a potential benefit over the difficulty of aligning a homodyne detector for maximum visbility , especially in more comon enironments which are not necessarily suited to interferometry . deriving a local oscillator can in itself
be difficult task as well . while compressive imaging certainly has drawbacks , its advtantages make it a good candidate for quantum imaging .
in this manuscript we have demonstrated all the necessary ingredients to achieve real time compressive quantum imaging .
this will facilitate the development of high sensitivity low - light - intensity imaging systems that will prove valuable in applications ranging from covert imaging to the imaging of photosensitive biological samples .
the examination of qnr as the mask on a dmd was varied provides a novel technique for quantum imaging in addition to motivating the study of single pixel quantum imaging .
the rastered quantum imaging results of fig .
6 make it clear that the appropriate choice of sampling matrices , in conjunction with the use of high - throughput slms now commercially available with throughput greater than 95%will be sufficient to achieve real - time quantum compressive imaging with squeezed light for the first time .
this work was performed at oak ridge national laboratory , operated by ut - battelle for the u.s .
department of energy under contract no .
de - ac05 - 00or22725 .
b.j.l was supported by a fellowship from the ic postdoctoral research program .
r.c.p . acknowledges partial support from the wigner fellowship . | we present a workbench for the study of real - time quantum imaging by measuring the frame - by - frame quantum noise reduction of multi - spatial - mode twin beams generated by four wave mixing in rb vapor . exploiting the multiple spatial modes of this squeezed light source , we utilize spatial light modulators to selectively pass macropixels of quantum correlated modes from each of the twin beams to a high quantum efficiency balanced detector . in low - light - level imaging applications ,
the ability to measure the quantum correlations between individual spatial modes and macropixels of spatial modes with a single pixel camera will facilitate compressive quantum imaging with sensitivity below the photon shot noise limit .
10 v. boyer , a. m. marino , r. c. pooser , and p. d. lett , `` entangled images from four - wave mixing , '' science * 321 * , 544547 ( 2008 ) .
q. glorieux , j. b. clark , a. m. marino , z. zhou , and p. d. lett , `` temporally multiplexed storage of images in a gradient echo memory , '' opt .
express * 20 * , 1235012358 ( 2012 ) .
d .- s .
ding , z .- y .
zhou , b .- s .
shi , x .- b .
zou , and g .- c .
guo , `` image transfer through two sequential four - wave - mixing processes in hot atomic vapor , '' phys .
rev .
a * 85 * , 053815 ( 2012 ) .
j. b. clark , z. zhou , q. glorieux , a. m. marino , and p. d. lett , `` imaging using quantum noise properties of light , '' opt .
express * 20 * , 1705017058 ( 2012 ) .
e. brambilla , l. caspani , o. jedrkiewicz , l. lugiato , and a. gatti , `` high - sensitivity imaging with multi - mode twin beams , '' phys .
rev .
a * 77 * , 053807 ( 2008 ) .
v. boyer , a. m. marino , and p. d. lett , `` generation of spatially broadband twin beams for quantum imaging , '' phys .
rev .
lett . * 100 * , 143601 ( 2008 ) .
n. v. corzo , a. m. marino , k. m. jones , and p. d. lett , `` noiseless optical amplifier operating on hundreds of spatial modes , '' phys .
rev .
lett . * 109 * , 043602 ( 2012 ) .
m. i. kolobov and c. fabre , `` quantum limits on optical resolution , '' phys .
rev .
lett . * 85 * , 37893792 ( 2000 ) .
m. lassen , v. delaubert , j. janousek , k. wagner , h .- a .
bachor , p. k. lam , n. treps , p. buchhave , c. fabre , and c. c. harb , `` tools for multimode quantum information : modulation , detection , and spatial quantum correlations , '' phys . rev .
lett . * 98 * , 083602 ( 2007 ) .
g. brida , m. genovese , and i. berchera , `` experimental realization of sub - shot - noise quantum imaging , '' nature photon . * 4 * , 227230 ( 2010 ) .
o. aytr and p. kumar , `` squeezed - light generation with a mode - locked q - switched laser and detection by using a matched local oscillator , '' opt .
lett .
* 17 * , 529531 ( 1992 ) .
d. smithey , m. beck , m. belsley , and m. raymer , `` sub - shot - noise correlation of total photon number using macroscopic twin pulses of light , '' phys .
rev .
lett . * 69 * , 26502653 ( 1992 ) .
l. jiying , z. jubo , l. chuan , and h. shisheng , `` high - quality quantum - imaging algorithm and experiment based on compressive sensing , '' opt . lett . * 35 * , 12061208 ( 2010 ) .
o. katz , y. bromberg , and y. silberberg , `` compressive ghost imaging , '' app .
. lett . * 95 * , 131110 ( 2009 ) .
p. zerom , k. w. c. chan , j. c. howell , and r. w. boyd , `` entangled - photon compressive ghost imaging , '' phys .
rev .
a * 84 * , 061804 ( 2011 ) .
t. do , l. gan , n. nguyen , and t. tran , `` fast and efficient compressive sensing using structurally random matrices , '' ieee trans . signal process . * 60 * , 139154 ( 2012 ) .
d. smith , j. gore , t. yankeelov , and e. welch , `` real - time compressive sensing mri reconstruction using gpu computing and split bregman methods , '' int . j. of biomed .
imag . *
2012 * , 864827 ( 2012 ) .
r. pooser , b. lawrie , and w. grice , `` compressive sampling low light beam profiler using structural random matrices with a semilocal randomizer , '' in preparation .
e. cands , j. romberg , and t. tao , `` robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , '' ieee trans .
inf .
theory * 52 * , 489509 ( 2006 ) .
m. figueiredo , r. nowak , and s. wright , `` gradient projection for sparse reconstruction : application to compressed sensing and other inverse problems , '' ieee j. sel .
top .
signa .
* 1 * , 586 597 ( 2007 ) . c. li , w. yin , h. jiang , and y. zhang , `` an efficient augmented lagrangian method with applications to total variation minimization , '' tech .
rep .
tr12 - 13 , , computational and applied mathematics , rice university , houston , tx , ( 2012 ) . |
two - dimensional ( 2d ) antiferromagnets under doping with non - magnetic impurities have emerged as interesting systems with rich possibilities to explore various disorder - driven phase transitions belonging to different universality classes.@xcite non - magnetic impurities ( vacancies ) enhance quantum fluctuation by reducing the connectivity of the spins .
many earlier calculations @xcite for the 2d @xmath0 heisenberg model had indicated that the quantum fluctuation can become strong enough to destroy the antiferromagnetic long - range order at a vacancy concentration @xmath20 less than the classical percolation threshold @xmath21whence @xmath20 would be a quantum critical point . however , more recent quantum monte carlo ( qmc ) simulations of the diluted quantum heisenberg model,@xcite studies of effective classical systems , @xcite as well as experiments on la@xmath22cu@xmath23zn@xmath24o@xmath25 ( with non - magnetic zn substituting @xmath0 cu ions ) @xcite all suggest that long range order actually survives all the way up to the percolation point @xmath21 , i.e. , @xmath26 for the single 2d layer . the percolating cluster at @xmath21 is ordered,@xcite which implies that the static properties at the dilution - driven transition in the quantum heisenberg model scale as in the classical ( percolation ) problem . however , quantum fluctuations lead to changes in the low - energy spin dynamics . the critical exponents
therefore in general depend on classical percolation exponents as well as the dynamic exponent @xmath10 of the quantum spin clusters.@xcite the dynamic exponent of the percolating cluster is therefore important , and the focus of this paper .
the dynamic exponent governs the scaling of the gap @xmath6 between the ground state and the lowest excited state of a finite cluster . with @xmath4 denoting the cluster length ( defined in some suitable way for a random cluster with irregular shape ) , the gap scales , on average , as @xmath27 . for a clean @xmath28-dimensional antiferromagnetic system on a bipartite lattice with @xmath29 ( even ) sites ,
every spin can be paired up with a nearest - neighbor spin on the opposite sublattice to effectively form a `` quantum rotor '' with angular momentum @xmath30 states . in the mapping to a quantum rotor model,@xcite these local degrees of freedom
are replaced with angular momenta @xmath31 taking all integer values , with the high @xmath31 states suppressed due to their energy being @xmath32 .
the ground state of the coupled quantum rotor system is a singlet . if the system is long - range ordered ( but the global rotational symmetry has not been broken by any external perturbation ) , then the low - energy excitations of the coupled rotors ( and the nel ordered spin system@xcite ) are those of a single quantum rotor with mass @xmath33 .
thus @xmath34 , i.e. , @xmath35 .
according to one recently proposed scenario for randomly diluted antiferromagnets,@xcite the quantum rotor states remain the lowest - energy excitations even at @xmath21 , where the dimensionality @xmath36 of the percolating cluster is fractal ; @xmath37.@xcite following the discussion above , this would seem to require that each spin can be paired up into a dimer with one of its nearest neighbors to effectively form a quantum rotor with @xmath38 ground state .
this situation can be realized in the special case of the dimer - diluted bilayer,@xcite in which two coupled layers are diluted exactly in the same way by removing inter - layer spin dimers .
all the remaining spins can then be paired with spins on the opposite layer . at sufficiently weak inter - layer coupling ,
the ground state of the largest connected cluster of spins in this system is long - range ordered for @xmath39,@xcite and , thus , the ground state should fall into the class of quantum rotor states with gap @xmath40 .
however , in the case of a single diluted layer ( or a bilayer with inter - layer coupling @xmath41 ) , there are in general some `` dangling spins '' ( or more generally , regions with local sublattice imbalance ) in which not all spins can be simultaneously paired up into nearest - neighbor dimers .
one may still be able to pair spins over longer distances ( which would also imply longer - range interactions between the rotors in the effective model ) , but at some point , when very long distances are required , the mapping to simple quantum rotors should break down .
our assertion is that , at the percolation point , there are regions of spins that effectively form isolated magnetic moments , which can not be described within an effective model containing only coupled rotors .
the spatial distribution of these moment regions , and weak effective interactions between them ( mediated by the magnetically inert parts of the percolating cluster ) , lead low - energy excitations which are dramatically different from those of the quantum rotor system .
we introduced this scenario and presented supporting numerical evidence in a recent paper.@xcite using finite - size scaling , we found a considerably larger dynamic exponent than the quantum - rotor value ; @xmath42 instead of @xmath43 . here
we provide more details of this work , and also expand significantly on the previous calculations .
we use several different methods to indirectly and directly examine the low - energy excitations of different types of clusters , both at and away from the percolation point .
the conclusion that @xmath42 for clusters at the percolation point is based largely on quantum monte carlo ( qmc ) calculations of an upper - bound @xmath5 to the lowest excitation gap @xmath6 for finite clusters with singlet ( @xmath8 ) ground states .
the bound is defined using standard sum rules , discussed in detail in sec .
[ ssemethod ] [ and summarized as eqs .
( [ gapbound ] ) , ( [ sumrule1 ] ) , and ( [ sumrule2 ] ) ] .
the bound is exact , @xmath44 , for a spectrum with a single mode , and is known to scale with the system size in the same way as @xmath6 more generally , e.g. , in the clean heisenberg model.@xcite it can be evaluated for large clusters using qmc calculations , in contrast to the exact gap , which is difficult to evaluate directly ( because it is dominated by statistical errors if the gap is small ) .
we also found that the probability distribution of local gaps @xmath45 ( also defined using a sum rule ) scales with the system size.@xcite defining @xmath46 ( where the exponent is determined from simulation data and is @xmath47 for @xmath8 clusters ) , the distribution @xmath48 is size - independent .
moreover , the low - energy tail of this distribution is well described by a power - law , @xmath49 , with @xmath50 . analyzing the local gaps using extremal - value statistics
, we found that the dynamic exponent should be related to the parameters of the local gap distribution according to @xmath51 .
our simulation results satisfy this exponent relation remarkably well .
the applicability of the exponent relation supports the notion that the low - energy excitations involve a number @xmath52 finite regions ( containing the effective moments ) , while an exponent @xmath53 shows that individual excitations are not localized ( since for localized excitations the energy should be independent of @xmath4 for large @xmath4 ) .
the effective moments should be located in regions of imbalance in the number of spins on the two sublattices , and many moments can be involved in an excitation .
the value of the exponent @xmath54 reflects the way in which the weak interactions between the effective moments involved in a particular excitation decrease with increasing system size , as these moments become further separated from each other . in this paper
, we report scaling results for larger clusters than previously and also compare results for clusters constructed in different ways . on the bipartite square lattice ,
we denote the number of sites on sublattice @xmath55 and @xmath56 as @xmath57 and @xmath58 , respectively .
the ground state has spin @xmath59 .
we analyze in detail both @xmath8 and @xmath12 clusters at the percolation point @xmath21 .
we use the gap upper - bound @xmath5 from sum rules , as well as lanczos exact diagonalization results for the excitation spectrum . for clusters with ground - state spin @xmath12 , we point out that the spectral weight entering in the sum - rule approach is dominated by @xmath15 excitations , whereas the lowest - energy excitations typically correspond to @xmath14 .
the quantity @xmath5 in this case describes only excitations where @xmath15 , for which we find @xmath60 based on finite - size scaling .
however , the lower @xmath14 excitations most likely follow the same @xmath42 scaling as the @xmath61 excitations of @xmath62 clusters .
we also discuss results for the dimer - diluted bilayer at @xmath21 , as well as the single layer at @xmath18 . for these systems , we observe behavior consistent with quantum rotor excitations ( although other scenarios , e.g. , fractons,@xcite are also possible ) . to explain the existence of localized moments in the percolating cluster , we also introduce a classical dimer - monomer aggregation model to study the purely geometrical local sublattice imbalance , which we believe is at the heart of this problem .
the dimers correspond to nearest - neighbor sites that can form minimal local quantum rotors , and the monomers lead to `` dangling '' spins that are , due to local sublattice imbalance , left over after the maximum number of dimers has formed .
the monomers , individual ones or groups of several of them , can lead to effective magnetic moments in the spin system .
we find that the classical monomers indeed are confined within regions of finite size , both at and away from the percolation point .
the anomalous dynamics with @xmath42 in the single - layer quantum spin system at @xmath21 should therefore be a consequence of localized quasi - free magnetic moments interacting very weakly because of the vanishing spin stiffness of the percolating cluster.@xcite away from the percolation point , the moments can lock to the global nel order of the cluster ( as a single magnetic impurity in two dimensional is known to do @xcite ) and do not form an effective independent low - energy system . to further investigate the nature of the excitations of the quantum spins and their relationship to the classical monomers
, we have also applied a projector qmc method in the valence bond basis @xcite to directly study the triplet excitations of clusters with singlet ground states . in the valence bond basis
, a triplet state can be described by a lone triplet bond ,
the location of which fluctuates among the background singlet bonds .
we find that the triplet bond is indeed predominantly localized at a subset of the classical monomer regions .
the total size of the excitation ( i.e. , the number of spins involved in it ) is not finite , however , but grows with the cluster size according to a non - trivial power law .
the outline of the rest of the paper is as follows . after defining the spin models and describing several computational methods in sec .
[ perco - model ] , we present results of both lanczos exact diagonalization and sum - rule qmc calculations for single - layer clusters at @xmath63 in sec . [ moredata ] . in sec .
[ spectralsection ] we discuss the distribution of spectral weight in the dynamic structure factor originating from excited states of different total spin , using lanczos exact diagonalization as well as an approximate analysis based on valence bond states .
we discuss scaling results for percolating bilayer clusters in sec .
[ bilayerp ] , and for single - layer clusters away from the percolation point in sec .
[ belowpc ] .
the classical dimer - monomer aggregation model is discussed in sec .
[ classicalmodel ] , and results of the valence - bond projector qmc simulations of triplet excitations in sec .
[ rvbqmc ] .
we conclude in sec .
[ perco - summary ] with a summary and discussion .
the heisenberg hamiltonian on a single site - diluted layer is given by @xmath64 where @xmath65 denotes nearest neighbors on a 2d square lattice and @xmath66 ( vacancy ) and @xmath67 ( magnetic site ) with probability @xmath1 and @xmath68 , respectively .
we study clusters with two types of boundary conditions . in open - boundary @xmath69 systems , we start with all magnetic sites and introduce vacancies with probability @xmath1 .
we study the largest cluster of connected magnetic sites .
the number of spins @xmath70 in such clusters fluctuates and scales as @xmath71 , with @xmath9 .
we also study clusters grown on an infinite lattice . starting from a single magnetic site , we add more sites to the cluster with probability @xmath68 by transversing along the boundary sites , leaving sites unfilled with probability @xmath1 , but flagging each site as visited ( so that sites assigned as vacancies are not visited again ) .
this procedure terminates at random at some stage where all neighbors of the cluster have been assigned as vacancies .
we only keep clusters of some desired target size @xmath70 .
these clusters have a characteristic average length @xmath72 ( defined , e.g. , as their radius of gyration ) such that @xmath73 .
the two types of clusters will be referred to as @xmath69 and fixed-@xmath70 , respectively . in ref . , we only studied fixed-@xmath70 clusters . here
we also consider the @xmath69 variant to check whether the finite - size scaling properties depend on the boundary conditions in the cluster construction . for @xmath74 , we consider only the @xmath69 clusters , because the fixed-@xmath70 construction rarely terminates at reasonably small @xmath70 in this case . under each type of boundary condition , we further consider two different ensembles of sublattice occupations ; @xmath62 , in which case all clusters have ground state spin @xmath8 , as well as arbitrary @xmath75 ( with the distribution give by the cluster construction ) , corresponding to ground state spin @xmath59 .
the latter ensemble includes also the @xmath8 clusters .
a bilayer cluster is constructed by coupling two identical single - layer clusters with an interlayer coupling constant @xmath76 .
the hamiltonian is thus @xmath77 where the subscripts @xmath78 refer to the two layers .
also in this case we can study @xmath69 or fixed-@xmath70 clusters , but , in contrast to the single layer , the ground state of a bilayer cluster is always a singlet because each spin can be paired with its neighbor in the opposite layer .
we consider small coupling ratios @xmath79 , for which the ground state has long - range order.@xcite here our main interest is in the the energy gap @xmath6 between the ground state and the first excited state , which in the case of an @xmath62 cluster is a singlet - triplet gap . for clusters with general @xmath80 such that @xmath81 , the lowest excitation can have total spin @xmath82 , or @xmath83 .
in addition to the gap , the distribution of the spin @xmath84 of the lowest - energy excitation is also interesting
. we will also study the localization properties of the excitations very explicitly , by formulating the problem in the valence bond basis and carrying out unbiased quantum monte carlo calculations of @xmath85 excitations of clusters with @xmath8 ground states . to calculate the gaps
, we use both direct and indirect ( approximate , through sum - rules ) estimates , using the methods discussed in secs .
[ ed ] and [ ssemethod ] . in sec .
[ ipr ] we will introduce the valence bond qmc scheme for directly imaging the spatial distribution of triplet excitations . the most straight - forward approach is to diagonalize the hamiltonian numerically in sectors of different magnetization , @xmath86 using the lanczos method .
however , for irregular clusters ( without lattice symmetries to exploit for block - diagonalization ) , this can be done in practice only for up to @xmath87 spins , due to the rapid growth of the matrix sizes with @xmath70 ( considering also that we have to average over a large number typically thousands of random cluster realizations ) .
nevertheless , such calculations are very useful and give some important insights into the role of `` dangling '' spins in low - energy excitations .
in addition to studying the level spectrum , focusing on a few low - lying states and calculating their total spin to classify the excitations , we also compute the full dynamic spin structure factor ( in the standard way with the lanczos method , as described , e.g. , in ref . ) ; @xmath88 where @xmath89 is the fourier transform of the spin operators ; @xmath90 in a clean heisenberg antiferromagnet on a bipartite lattice , the lowest excitation is a triplet at @xmath91 .
we can use this wave - vector also for the diluted system , although the momentum is no longer conserved , i.e. , the energy eigenstates @xmath92 in ( [ dynamicss ] ) are not classified by the quantum number @xmath93 , but the spin operators @xmath89 are still completely well defined .
we expect @xmath94 to exhibit the largest spectral weight for the low - energy excitations , since these should involve out - of - phase fluctuations of neighboring spins . as we will see in sec .
[ generals ] , the dynamic structure factor is of great utility in judging the validity of our sum - rule based approach for an upper - bound of the energy gap , which we discuss next .
we use the stochastic series expansion ( sse ) qmc method @xcite to calculate quantities which are closely related to the gap .
an upper - bound @xmath5 to the ground state energy gap @xmath6 can be obtained using the static spin structure factor @xmath95 and susceptibility @xmath96 at the staggered wave - vector @xmath97 ; @xmath98 this bound follows from the well - known sum - rules ; @xmath99 which , in the way written here , are valid at temperature @xmath100 . in a system with a sole triplet mode ( a hypothetical situation ) with energy @xmath101
, we get @xmath102 .
any spectral weight above this lowest mode will render the ratio larger than @xmath101 . for a clean system ,
the lowest quantum rotor state is at @xmath97 ( whereas at other wave - vectors spin - waves are the lowest excitations ) . as we discussed above
, we expect @xmath97 to be the best choice for examining low - energy excitations also in the diluted system , and we here focus exclusively on this case .
the staggered structure factor and susceptibility can be efficiently calculated with the sse method with `` operator - loop '' updates.@xcite using the standard definitions , for a given cluster of @xmath70 sites the static staggered structure factor is @xmath103 and the corresponding susceptibility is given by @xmath104 where @xmath105 .
disorder averages are subsequently calculated for the ratio in ( [ gapbound ] ) ( where , it should be stressed , we first evaluate the ratio separately for each cluster , in order to obtain the gap bound specifically for each of them , and then take the average ) using , typically , thousands of random realizations of either the largest cluster on @xmath69 lattices or fixed-@xmath70 clusters . for a clean heisenberg antiferromagnet , @xmath5 is known@xcite to scale with the system size as the true gap ; @xmath106 .
this is because the dominant spectral weight is at the very lowest excitation energy
the spectral function in the thermodynamic limit has a delta - function at the lowest energy , followed by a continuum at higher energies .
we expect similar spectral features in the percolating cluster and suspect that @xmath5 should scale as @xmath6 ( and will show supporting numerical results in the next section ) . at the very least , _
if the true power - law behavior is @xmath107 , then the value @xmath108 extracted from finite - size scaling of @xmath5 must be a lower bound to the true dynamic exponent_. actually , in sec .
[ moredata ] we will use lanczos results for the dynamic structure factor on small clusters to show that the finite - size scaling of @xmath5 _ does not _ reflect the true lowest - energy excitations in the case of @xmath12 clusters , but all indications are that the sum - rule approach is valid for @xmath8 clusters . we will also study an effective local ( site - dependent ) excitation gap @xmath109 which is analogous to the gap bound ( [ gapbound ] ) but here the `` local structure factor '' is just a constant ; @xmath110 .
the local susceptibility @xmath111 is defined as @xmath112 although the imaginary - time dependent correlation function @xmath113 is asymptotically , for @xmath114 , dominated by the lowest excitation , in practice the integral will be dominated by the excitation(s ) which predominantly affects the given site @xmath115 . for a disordered system
, different sites can be affected by different excitations , and @xmath45 then represents a typical energy scale of excitations affecting spin @xmath115 .
we should note that for clusters with ground state spin @xmath12 , the grand - canonical sse method samples over all magnetization sectors @xmath116.@xcite we therefore have to subtract the static @xmath117 contributions in eqs .
( [ sfactor ] ) , ( [ staggerchi ] ) , and ( [ localx ] ) arising from a non - zero @xmath118 , i.e. , in eq .
( [ localx ] ) we subtract @xmath119 computed in the different @xmath118 sectors and averaged over all @xmath118 .
the sse method operates at @xmath120 , but we can achieve the @xmath121 limit by choosing @xmath122 sufficiently low for all quantities of interest to converge .
we use a `` @xmath123 doubling '' procedure,@xcite in which the inverse temperature is successively doubled until there is no longer any detectable dependence of calculated quantities on @xmath123 . since the dynamic exponent is large , the temperature @xmath124 has to be very low indeed for large clusters . as an example of the ultra - low temperatures required , the largest @xmath123 we use for @xmath125 clusters with @xmath8 is @xmath126 .
since the simulation ( cpu ) time and memory usage scale essentially linearly in both @xmath123 and @xmath70 , these calculations are quite demanding .
fortunately , the sse code for the isotropic heisenberg model can be effectively parallelized,@xcite and we have run most of the simulations on a massively parallel computer very well suited for these calculations.@xcite when studying disorder averaged static properties , the sse runs for each individual cluster can be rather short .
as long as each run is properly equilibrated ( for which the @xmath123-doubling procedure also helps @xcite ) , the average over many realizations will give an unbiased estimate to any simple average , e.g. , a spin correlation function .
however , when computing nonlinear functions involving several quantities , such as the ratio ( [ gapbound ] ) , the statistical errors introduce a bias .
it is therefore important to collect sufficient statistics for the individual clusters .
we have compared results of runs of different lengths in order to make sure that the results presented here do not suffer from significant bias effects . to study the nature of the lowest triplet excitation of clusters with @xmath8 ground states , we apply a valence bond projector monte carlo method.@xcite this method has been described in detail in recent papers @xcite and we here only review the elements necessary to understand the way we can access the triplet excitations and study their spatial distribution on the clusters .
first , consider the singlet ground state @xmath127 , which we want to project out from a singlet `` trial '' state @xmath128 .
the latter has an expansion in all singlet energy eigenstates ; @xmath129 in the standard way , if the ground state energy is the eigenvalue of the hamiltonian which is the largest in magnitude , which can always be assured by subtracting a constant from @xmath130 ( which we assume has been done , if necessary ) , the ground state can be projected out by applying a high power of @xmath130 to the trial state ; @xmath131 where we include the minus sign because normally @xmath132 . for large @xmath133 all the excited states
are filtered out because the ratios @xmath134 .
valence - bond basis states are products of @xmath135 singlets , @xmath136 where we consider the first , @xmath115 , and second , @xmath137 , spins to always be on sublattice @xmath55 and @xmath56 , respectively .
the trial state is thus expressed in this over - complete basis as @xmath138 where @xmath139 labels all the different tilings of the cluster into valence bonds , of which there are @xmath140 , and we have introduced the short - hand notation @xmath141 for a valence - bond basis state . for a trial state in a ground state projector calculation , it is convenient to use an amplitude product state,@xcite where the expansion coefficients are given by @xmath142 where @xmath143 and @xmath144 is the number of bonds of size @xmath145 in the configuration , i.e. , the length of the bond is @xmath146 . note that it is not necessary to normalize the trial state . for the clean 2d system ,
the optimal amplitudes are translationally invariant and decay as @xmath147.@xcite for random clusters , the optimal amplitudes are naturally not translationally invariant . while the average bond probabilities ( which are related to the amplitudes ) decay with @xmath148 , for any given cluster there are typically some regions spanned by long bonds ( a feature intimately connected with the low - energy physics , as we will discuss in secs . [ spectralsection ] and [ rvbqmc ] ) .
one could in principle optimize all the @xmath149 different amplitudes for each specific cluster .
however , the effort involved in individual optimizations for hundreds or thousands of clusters does not necessarily pay off , compared to just projecting the trial state with a somewhat larger power @xmath133 of @xmath130 .
we here use a very simple trial state with all @xmath150 . to carry out the projection using monte carlo sampling , we write the @xmath0 heisenberg hamiltonian in terms of singlet projection operators on all the pairs @xmath151 of nearest - neighbor sites @xmath152 ; @xmath153 and
write the projection operator in ( [ projhp ] ) as @xmath154 where @xmath155 denotes the possible strings , @xmath156 , of the singlet projectors .
when a singlet projector @xmath157 acts on a state with a valence bond on the two sites @xmath158 , the state remains unchanged with a matrix element of unity ; we call this a diagonal projection
. if the operator acts on a state with no valence bond on the two sites , then the two bonds @xmath159 and @xmath160 connected to @xmath158 are broken , and new singlets @xmath161 and @xmath162 are formed .
this process has matrix element @xmath163 , and we call it an off - diagonal projection
. thus the projection rules are ; @xmath164 acting on a component @xmath165 of the trial state , a string @xmath166 effects a number of rearrangements ( [ hijoff ] ) of pairs of valence bonds , resulting in another valence bond basis state which we call @xmath167 ; @xmath168 here the `` projection weight '' @xmath169 for a combination of operator string @xmath166 and state @xmath165 is given by the number @xmath170 of off - diagonal operations ( [ hijoff ] ) in the course of the projection ; @xmath171 the expectation value of an operator @xmath55 can be written @xmath172 where @xmath173 and @xmath174 are the weights computed according to ( [ wamp ] ) for the bonds in the states @xmath165 and @xmath175 in the expansion ( [ psisinglet ] ) of the trial ket @xmath128 and bra @xmath176 states .
the sampling weight to be used in monte carlo calculations of ( [ vbaexpvalue ] ) is @xmath177 where the overlap of the two projected states is given by @xmath178 where @xmath179 is the number of loops formed when the bond configurations of the states @xmath167 and @xmath180 are superimposed ( forming the transposition graph @xcite ) .
simple sampling procedures for the operator strings and trial state bonds are described in refs . .
more efficient sampling methods have been developed recently,@xcite which we use but do not discuss here .
for the purpose of the present paper , the most interesting aspect of the valence bond projector scheme is the fact that we can easily extend the scheme to also study a triplet state .
a trial wave function in the triplet sector can be expressed in the overcomplete basis of a lone triplet bond among @xmath181 singlets .
we denote a zero - magnetization triplet by square brackets ; @xmath182=(\uparrow_i\downarrow_j+\downarrow_i\uparrow_j)/\sqrt{2},\ ] ] and expand the triplet trial state as @xmath183\ldots(i^v_{n/2},j^v_{n/2})\rangle \nonumber \\ & = & \sum_v w_v \sum_{m=1}^{n/2 } |v_m\rangle , \label{psitriplet}\end{aligned}\ ] ] where the normalization is again irrelevant .
here we use the same expansion coefficients the amplitude products ( [ wamp])as in the singlet trial state . note that for a clean system , the singlet state ( [ psisinglet ] ) has momentum @xmath184 , whereas the triplet ( [ psitriplet ] ) has @xmath185 .
these are the known momenta of the lowest states in the two spin sectors ( with the triplet being the lowest member of anderson s tower of quantum rotor states@xcite ) .
the wave - function signs corresponding to ( [ psisinglet ] ) and ( [ psitriplet ] ) should be correct for the lowest singlet and triplet states also for a diluted system , since all conditions for marshall s sign rule [ which corresponds to all positive expansion coefficients in eq ( [ psitriplet ] ) ] @xcite remain valid .
when acting on a triplet bond , the singlet projector @xmath157 destroys the state , while the action between a singlet and a triplet bond are very similar to the pure singlet rules ( [ hijoff ] ) .
the two triplet rules are @xmath186 ...
\rangle = 0 , \label{thijdia } \\ & & h_{ij}| ... [i , k] ...
(l , j) ...
\rangle = \hbox{$\frac{1}{2}$}| ... (i , j) ... [l , k] ... \rangle .
\label{thijoff}\end{aligned}\ ] ] in the projector method , it is straight forward to convert one bond of the singlet trial wavefunction into a triplet and trace its evolution .
the triplet states that survive after all @xmath133 operations [ i.e. , that are not destroyed by a diagonal operation ( [ thijdia ] ) ] are used to measure properties in the triplet sector . to measure triplet expectation values
, we have to project triplets like this both in the bra and ket in the triplet version of ( [ vbaexpvalue ] ) .
we also have to check the overlap ( [ vboverlap ] ) of the surviving triplet states .
one can show that the two triplet bonds have to be in the same transposition - graph loop in order for the overlap to be non - zero , and it is then equal to the singlet overlap ( [ vboverlapvalue ] ) . for surviving pairs of triplets ,
the weight of the triplet configuration is the same as that of the original singlet one .
one can therefore sample the configurations in the singlet sector , and carry out measurements with all the surviving triplets _ without reweighting_. this is one of the strengths of the valence bond projector method .
there can still be problems with this approach , because the number of surviving triplets decreases with the projection power @xmath133 [ because the probability of a triplet to be destroyed by a diagonal triplet operation ( [ thijdia ] ) increases ) .
it helps considerably that the starting trial state can have the triplet at @xmath135 different locations , in both the bra and the ket state , and as long as one pair out of the total of @xmath187 combinations survives ( and gives non - zero overlap ) , we can collect statistics .
one can carry out the summation over triplet locations @xmath188 in ( [ psitriplet ] ) efficiently , without introducing any additional factor @xmath135 in the computational effort , in a single traversal of the operator sequence .
in some cases it can still happen that the triplet quantities of interest have not converged well to the @xmath189 limit before the triplet survival probability becomes too low to be useful .
this is not a serious problem in the present application , although an extrapolation to infinite @xmath133 based on several calculations with reasonable triplet survival probability is necessary to ensure that the results represent the lowest triplet .
an exponential asymptotic convergence can be expected based on eq .
( [ projhp ] ) .
( left ) and valence bond projection calculation of the triplet density @xmath190 ( center ) . to the right ,
the clusters are shown covered with dimers ( pairs of spins enclosed by ovals ) and left - over monomers ( black circles ) .
the black circles inside ovals indicate other possible locations of the monomers , corresponding to alternative maximal dimer coverings ( which here always corresponds to two left - over monomers).,width=255 ] -3 mm we will discuss the spatial distribution of the triplets .
the surviving triplet states have the triplet bond located at two particular sites ( which can be different in the ket and the bra , and we do the measurements in both of these states ) . in a random system
, the average triplet density will not be uniform and provides a very concrete measure of the localization properties of the lowest triplet excitation .
note that the distribution of the @xmath191 triplet bond is equivalent to the magnetization distribution in a state with @xmath192 , which could also be studied using the sse method at low temperatures ( e.g. , by including a weak magnetic field @xcite ) .
however , the valence bond states also contain other relevant information , e.g. , the statistics of the length of the triplet bond , which can only be accessed in the valence bond basis and which will be useful for analyzing the nature of the excitations ( as we will do in sec .
[ rvbqmc ] ) . having introduced the technical aspects of all the methods , we now present illustrative results for two small clusters .
this will help to clarify the subsequent analysis and discussion of results for larger clusters .
the local gaps @xmath45 and the triplet density @xmath190 are visualized for two different clusters in fig .
[ clusterdemo ] . here
the color scales were created separately for the two clusters , with the minimum and maximum values for each quantity on a particular cluster corresponding to the extrema of the scales shown ( and , thus , the plots should only be used to examine the variations within the clusters , not comparing the values for the two clusters ) .
the two clusters differ qualitatively in a way which is directly related to our arguments pertaining to a low energy scale .
the lower cluster can be completely subdivided into pairs of nearest - neighbor sites ( dimers , represented by ovals ) , whereas the upper one has two `` dangling spins '' left ( monomers , shown as black circles outside ovals ) after the sites have been paired up as much as possible .
the pairing into dimers is not unique the black circles inside ovals show all other possible monomer locations for this cluster . in all cases there
are two monomers in two separate regions . the classical dimer - monomer aggregation model discussed in sec .
[ classicalmodel ] contains the statistics of the distribution of the monomers .
our main argument is that the presence of monomer sites leads to small gaps , i.e. , a large dynamic exponent .
for the two clusters shown , the exact gaps are @xmath193 and @xmath194 , respectively , for the cluster with and without monomer sites .
the gap upper bounds @xmath5 are @xmath195 and @xmath196 . while in particular the former is quite far from the exact result ( in a relative measure ) , the difference between the two clusters is still large .
large clusters are likely to have dangling spins , and the top cluster in fig .
[ clusterdemo ] is therefore the more interesting case .
one can clearly see a strong correspondence between small local gaps and large triplet density , and they both coincide very well with sites where monomers can be located .
although this is in accord with the notion of monomers leading to finite regions of spins affected by the excitation , these clusters are clearly too small to give any meaningful quantitative insights into the localization properties of the triplet . in the following four sections we will carry out quantitative scaling analyses of the gaps in different types of clusters , while further discussion of the monomer and triplet distributions will be postponed to secs . [ classicalmodel ] and [ rvbqmc ] , respectively .
we here discuss the distribution of exact gaps obtained with the lanczos method , as well as sse qmc results for the gap upper bound and local gaps .
first we consider @xmath8 clusters ( @xmath62 ) , and then arbitrary @xmath197 . and its upper bound qmc estimate @xmath5 for @xmath198 clusters ( a ) , and the @xmath5 distribution for @xmath199 ( b ) and @xmath200 ( c).,width=207 ] -3 mm fig . [ lanczos - sse](a ) shows the probability distribution of the logarithm of the exact gap @xmath6 of @xmath198 clusters obtained using @xmath201 samples .
we also show results for the upper - bound @xmath5 for clusters of the same size , obtained from sse calculations for @xmath202 different clusters .
we presented these results in ref . and here re - graph them in a different way for added clarity .
the @xmath5 curve is visibly shifted up in energy relative to the @xmath6 distribution ( with the average @xmath203 ) , but the shapes of the two curves are remarkably similar . the two - peak structure is related to the `` dangling spins '' discussed in sec .
[ sec_examples ] .
the large - gap peak originates almost exclusively from clusters that can be completely partitioned into nearest - neighbor dimers , whereas the low - gap peak corresponds to clusters with dangling spins ( monomers ) . clearly ,
as the cluster size grows , it will be less and less likely to find clusters with no monomers , and the weight of the high - energy peak should therefore gradually diminish and be absent for large clusters .
the relative size of the large-@xmath5 peak is indeed much smaller in the @xmath199 distribution graphed in fig .
[ lanczos - sse](b ) . in the @xmath204 histogram , shown in panel ( c ) , only a single peak can be discerned ( with only a weak tail suggesting some remaining contributions from no - monomer clusters ) .
[ z - scale ] shows the size dependence of the disorder averaged @xmath16 on log - log scales for both fixed-@xmath70 ( top panel ) and @xmath69 ( bottom panel ) clusters .
we also show the typical values @xmath205 , obtained by averaging @xmath206 for the individual clusters . while the typical and average values do not exactly coincide , for large systems they scale in the same way .
linear fits to the @xmath205 data on the log - log scales gives @xmath207 for both types of clusters . here
the estimated error reflects the purely statistical errors of the line fits in combination with small variations depending on what range of system sizes are included . and typical @xmath205 gap upper - bound for @xmath8 ( @xmath62 ) clusters .
the top and bottom panels show results for fixed-@xmath70 and @xmath69 clusters , respectively .
the lines correspond to the scaling expected with dynamic exponent @xmath208 ( i.e. , the size dependence is @xmath209 and @xmath210 , respectively , for the two types of clusters).,width=245 ] -3 mm as shown in figs .
[ gapscalingn](a ) and [ gapscalingl](a ) , for fixed-@xmath70 and @xmath69 clusters , respectively , not only do the averages and typical values of @xmath5 scale with the system size , but the entire distribution can be collapsed onto a common size - independent curve , by scaling the gap estimates with the cluster size .
we define the scaled gap upper - bounds for the two types of clusters according to @xmath211 as can be seen in the figures , the small - gap side of the distribution of @xmath212 is very well described by a power law ;
@xmath213 \propto \epsilon^{\omega + 1}$ ] , with @xmath214 .
this distribution of the logarithm of @xmath215 corresponds to a probability distribution @xmath216 for the scaled gap @xmath215 itself [ since the differential @xmath217 .
( a ) and local gap bound @xmath218 ( b ) for fixed-@xmath70 clusters with ground state spin @xmath8 . the exponents are indicated in the panels .
the solid lines correspond to small - gap exponent @xmath214 and the curve in ( a ) is a frechet form.,width=245 ] -3 mm ( a ) and local gap bound @xmath219 ( b ) for @xmath69 clusters with ground state spin @xmath8 .
the solid lines correspond to @xmath214 and the curve in ( a ) is a frechet form.,width=245 ] -3 mm we next consider the local gap estimate @xmath45 , i.e. , the inverse local susceptibility ( [ localx ] ) . measuring this quantity at each site , we define size - scaled local gaps ; @xmath220 the probability distributions of @xmath221 for different cluster sizes , based on several hundred clusters of each size , collapse onto each other for a suitably chosen @xmath47 , as shown in figs .
[ gapscalingn](b ) and [ gapscalingl](b ) for the two types of clusters .
the small - gap tails of the distributions are again very well described by a power law ; @xmath222 , with the same @xmath214 as for the scaled `` global '' gap bound @xmath5 . in ref .
we used extremal value statistics @xcite ( in a way generalizing a treatment of localized excitations by lin et al.@xcite ) and found a relationship between the exponents @xmath223 , and @xmath224 . for completeness , we repeat and further clarify our arguments here .
our hypothesis is that , for a large cluster of size @xmath70 , there is a number @xmath52 of regions of sublattice imbalance .
these regions act as localized magnetic moments , which interact weakly with each other through the magnetically inert parts of the percolating cluster .
the excitations of this effective low - energy system of coupled moments are not localized because several distant moments can be involved .
it is then natural to expect some size dependence of the local gaps , due to the dependence of the effective interactions on the distance between the moments involved in a low - energy excitation , combined with the increasing distance ( on average ) between these moments with increasing cluster size .
we posit that this size dependence can be captured by the single exponent @xmath54 in eq .
( [ epiloni ] ) .
the actual finite - size gap @xmath6 for a given cluster should correspond to the smallest of the local gaps @xmath45 for that cluster , for which we use the notation @xmath225 . of course , the local gaps that we measure are only approximations ; one can not unambiguously define a local gap in an interacting system . nevertheless , @xmath45 reflects the local distribution of spectral weight , and there should be some site @xmath115 within the regions affected by the lowest excitation for which @xmath226 ( and @xmath227 ) . in our numerical analysis
, @xmath6 is approximated by the bound @xmath5 , and we expect @xmath228 .
examining the numerical data , we indeed find a very strong correlation between the two quantities , as shown in fig .
[ gapcorr ] for fixed-@xmath70 clusters . here
it can be seen that @xmath225 is typically @xmath229 times larger than @xmath5 , which reflects larger spectral weight above the true lowest excitation energy in the local dynamic structure factor @xmath230 than in @xmath94 .
it should be noted that @xmath231 is allowed within the sum - rule approach , although @xmath232 has to hold strictly .
we now assume that there is a number @xmath233 of different local scaled gaps @xmath234 and investigate the consequences of this in light of the scaling behavior found above .
we assume a probability distribution @xmath235 for some window of small @xmath215 ( where @xmath55 is a constant and we consider a more general case than just @xmath214 extracted from the finite - size scaling of the data ) .
we derive the probability distribution for the smallest scaled local gap @xmath236 for large @xmath237 using extremal - value statistics .
we should clarify why we assume @xmath233 for the number of local gaps , instead of just @xmath238 , which is the actual number of different numerical values @xmath45 that we compute for a given cluster .
the distinction will not matter in the analysis , but it has an important physical significance . in our scenario ,
a cluster consists of regions with localized moments , which participate in the low - energy excitations , as well as inert parts which have only high - energy excitations .
the form of the probability distribution @xmath235 should only hold for sites @xmath115 within the moment regions .
it is then important in our analysis that also the number of such sites scales as @xmath70 ( although one could also generalize to @xmath239 with @xmath240 , but the consistency of our analysis with @xmath233 will show that this is not necessary ) .
later , we will provide more concrete proof that the moment regions are finite and the total number of spins belonging to moments grows linearly with @xmath70 . and the gap upper bound @xmath5 for clusters of size @xmath200 and @xmath241 .
each data point corresponds to monte carlo results for a randomly generated @xmath62 cluster .
the line shows the ideal ( single - mode ) case of complete equivalence of the two estimates of the finite - size gap.,width=245 ] -3 mm note also that in reality the distribution of local gaps must be cut off ( equals zero exactly ) below some very small value for a given finite cluster size .
however , this should not affect the results of the analysis to follow , because also the assumed power - law probability is very small below such a threshold .
we thus expect the results derived below to be valid within some significant window of scaled gaps @xmath215 .
we denote the probability of finding a local gap at an arbitrary chosen site ( within one of the moment regions ) smaller than some value @xmath242 by @xmath243 .
it is given by @xmath244 if one of the scaled gaps @xmath245 is the smallest and has the value @xmath215 , then all the other ( @xmath246 different ) values @xmath234 , @xmath247 must be larger than @xmath215 . the probability of these @xmath246 values being smaller than @xmath215 is @xmath248^{m-1}$ ] .
since any of the @xmath249 values could be the smallest one , we get a factor of @xmath249 , and finally the distribution of the @xmath245 value is given by @xmath250 .
thus , the distribution of the smallest scaled local gap is @xmath251^{m-1},\ ] ] which for small @xmath215 also can be expressed as ; @xmath252 using eq . ( [ localgap - distribution ] ) here gives the frechet distribution,@xcite @xmath253 , \label{u - distribution}\ ] ] where @xmath254 .
thus , the probability distribution of the scaled global gap should be governed by the same exponent @xmath224 as the scaled local gaps .
the frechet form can indeed be fitted to the @xmath5 data in figs .
[ gapscalingn](a ) and [ gapscalingl](a ) , with the same exponent @xmath214 as in the local - gap ( b ) panels , but only in the small - gap region .
one can not expect the frechet distribution to work for large gaps , since the local gap distribution we started from is linear only in the small @xmath234 region ( and , as discussed above , we expect the large - gap part of the distribution to be dominated by excitations of the magnetically inert cluster regions without moments ) . the fitted forms in figs .
[ gapscalingn](a ) and [ gapscalingl](a ) are therefore also not normalized .
nevertheless , it is encouraging that the data is in agreement with the result that both the local and global gaps should scale with the same exponent , which here is @xmath214 . clusters with no restriction on the sublattice occupation numbers @xmath57 and @xmath58 .
the solid line shows the asymptotic small - gap behavior expected with @xmath214.,width=245 ] -3 mm let us now use the distribution ( [ localgap - distribution ] ) in a different way .
since we assume that there are @xmath233 local gaps , the typical smallest gap should correspond to @xmath243 for which @xmath255 , i.e. , @xmath256 .
this gives @xmath257 .
since @xmath225 should equal @xmath6 , and , by definition , @xmath258 , we arrive at the following relationship between the three exponents ; @xmath259 this generalizes the relation @xmath260 used as a criterion for a localized excitation by lin et al .
@xcite to excitations originating from two or more finite entangled regions distributed over the cluster . with our numerical values from the finite - size scaling above , @xmath261 and @xmath214 ( the latter of which
is not based on a fit , but is a value consistent with all our results ) , we obtain @xmath262 , in very reasonable agreement with the value @xmath47 obtained in figs . [ gapscalingn](b ) and [ gapscalingl](b ) from the scaling of the @xmath234 data for the fixed-@xmath70 and @xmath69 clusters .
the applicability of the exponent relation ( [ zrelation ] ) provides strong support to our hypothesis of `` globally entangled local moment excitations '' .
we now turn to clusters with no restriction on the sublattice occupations @xmath57 and @xmath58 in the generated ensemble . scaling results for the global and local gaps of @xmath69 clusters
are shown in fig .
the finite - size scaling of the average and typical values of @xmath5 are shown for both fixed-@xmath70 and @xmath69 clusters in fig .
[ z - scale-2 ] .
we obtain @xmath263 and @xmath264 for both cluster types .
these exponents differ significantly from the ones obtained previously for the ensemble including @xmath8 clusters only .
in particular , @xmath13 , whereas the @xmath8 clusters gave @xmath42 .
the exponent relationship ( [ zrelation ] ) still holds approximately , albeit with somewhat larger deviations than in the @xmath8 case .
the small - gap behavior remains consistent with the exponent @xmath214 in all cases .
we believe that the much smaller exponents @xmath223 are due to a failure of the sum rule approach to capture the true low - energy states for @xmath12 .
to demonstrate this , we next investigate the dynamic structure factor ( [ dynamicss ] ) .
for the ensemble with unrestricted @xmath57 and @xmath58 .
the upper and lower panels show results for fixed-@xmath70 and @xmath69 clusters , respectively .
lines corresponding to a dynamic exponent @xmath265 are shown with all the data sets.,width=245 ] -3 mm
we will investigate how the spectral weight of the dynamic structure factor is distributed among different spin sectors of the excited states in ( [ dynamicss ] ) .
acting on the ground state with the @xmath97 spin operator ( [ sqop ] ) , or the corresponding @xmath242 or @xmath266 components , on one of the @xmath267 degenerate ground states of spin @xmath197 results in states with spin @xmath197 and @xmath268 .
this well known selection rule for the dynamic structure factor ( [ dynamicss ] ) can be easily demonstrated in the valence bond basis .
here we do this as a prelude to discussing the distribution of the spectral weight among the three sectors of final spin for clusters with ground state @xmath12 .
we consider an extended valence bond basis with an arbitrary number of @xmath191 triplet bonds ( [ psitriplet ] ) in addition to singlet bonds , for a state with total @xmath191 ( hence the number of spins , @xmath29 , is even ) .
later , we will consider also @xmath269 .
the standard valence bond basis for @xmath62 is restricted to bipartite bonds only.@xcite here , for @xmath270 and @xmath191 , we require a maximal number of bipartite bonds , i.e. , if the total sublattice imbalance is defined as @xmath271 , there will be @xmath272 bipartite bonds and @xmath273 bonds connecting sites on the same sublattice ( with all such pairs either on the a or b sublattice , depending on which sublattice has the larger number of sites ) .
this basis is clearly overcomplete . a state with two triplet bonds and one non
- bipartite bond is illustrated in fig .
[ vbconfigs](a ) .
first , let us discuss the relationship between the total spin @xmath197 and the number of triplet bonds .
a state with @xmath274 triplets does not have fixed spin when @xmath275 ( while for @xmath276 and @xmath277 , the state has fixed @xmath8 and @xmath277 , respectively ) . according to the rules for addition of angular momenta
, one might at first sight suspect that @xmath274 triplets could be used to form states with @xmath278 .
however , consider the operator @xmath279 which inverts all the spins ; @xmath280 which is a special case of a rotation in spin space .
since the total magnetization @xmath191 , a state with fixed @xmath197 is also an eigenstate of this operator , with eigenvalue @xmath281 . since a triplet pair ( bond ) is even under @xmath279 while a singlet pair is odd , the eigenvalue @xmath10 of a state with a fixed number @xmath274 of triplet bonds is @xmath282 .
thus , in order to construct a state with fixed @xmath197 ( fixed @xmath10 ) , one can not mix valence bond states with even and odd number of triplets . since the minimum number of triplets required to construct a state with fixed spin is @xmath283 , we conclude that the triplet numbers that can be mixed are @xmath284 .
this , in turn , implies that a state with fixed number of triplets is a linear combination of states with @xmath285 , where the lower limit @xmath286 or @xmath277 applies for even and odd @xmath197 , respectively .
( requiring one non - bipartite bond here the top one ) .
open and solid circles indicate the two sublattices .
singlet and triplet bonds are shown as solid and dashed lines , respectively .
( a ) shows a state with two triplet bonds and @xmath191 . in ( b ) , there are two unpaired up spins and @xmath192.,width=236 ] -3 mm next , we let the @xmath91 spin operator ( [ sqop ] ) act on a given valence bond state with @xmath274 triplets . we can write the operator in a way tailored specifically for the state under consideration ; @xmath287 . \label{szbipartite}\ ] ] here the subscripts @xmath288 and @xmath289 refer to two sites connected by a bipartite valence bond @xmath151 , and @xmath290 denotes a pair of sites on the same sublattice , connected by a non - pipartite bond @xmath291 .
the bonds can be singlets or triplets , and the possible outcomes when operating with one of the terms are ; @xmath292 ... \rangle , \nonumber \\ & & ( s^z_{i(b)}-s^z_{j(b)})| ... [i_b , j_b ] ...
\rangle = |
... (i_b , j_b ) ...
\rangle , \nonumber \\ & & ( s^z_{i(c)}+s^z_{j(c)})| ...
(i_c , j_c ) ...
\rangle = 0,\\ & & ( s^z_{i(c)}+s^z_{j(c)})| ... [i_c , j_c ] ...
\rangle = 0.\nonumber\end{aligned}\ ] ] thus , operating with the full @xmath293 , we obtain a linear combination of states with @xmath294 and @xmath295 triplets . extending this result to the case of a fixed-@xmath197 state @xmath296 , which is a linear combination of states with different @xmath274 ( all even or all odd ) , we can think of the triplet bond created or destroyed in each term [ with the operator ( [ szbipartite ] ) written in the appropriate way for operation on each term ] as adding or subtracting a spin @xmath277 to or from a spin @xmath197 . then
, considering also that even and odd @xmath197 corresponds to mixtures of even and odd @xmath274 , respectively , we conclude that the state @xmath297 is a mixture of only @xmath268 states ( which is also consistent with the fact that for @xmath8 ground states , the spectral weight is exclusively due to @xmath298 excitations ) . in order to respect the spin - rotational invariance when using the @xmath10-component operator @xmath293 in the dynamic structure factor for @xmath12 , we also have to consider non - zero @xmath118 .
some of the spins are then not paired up into valence bonds . in a minimal basis mixing valence bonds and spins , there are @xmath299 unpaired up or down spins for @xmath300 and @xmath301 , respectively .
the unpaired spins can not be restricted to the same sublattice , so now the basis consists of the unpaired spins at arbitrary locations , a maximal number of bipartite bonds on the remaining locations , and the rest of the sites covered by non - bipartite bonds .
an example of such a state is illustrated in fig .
[ vbconfigs](a ) . in eq .
( [ szbipartite ] ) @xmath302 and @xmath303 are the number of bipartite and non - bipartite bonds in a given basis state and @xmath304 .
we now also have to add a sum over the @xmath299 unpaired spins .
it is then clear that @xmath297 will contain also a spin-@xmath197 component , arising from this added sum , in addition to the @xmath268 components ( which can be argued for in analogy with the @xmath191 case ) .
some , but not all , of the corresponding spectral weight in the spin @xmath197 sector is at @xmath305 , as discussed in sec .
[ ssemethod ] .
averaging over all @xmath306 , it is also clear that the amount of @xmath307 spectral weight should increase with @xmath197 , as it is zero for @xmath8 and the relative weight of the operations on unpaired spins increases with @xmath118 .
we now turn to numerical results for the dynamic structure factor . investigating small clusters with the lanczos method
, we have found that the lowest excitation of a cluster with ground state spin @xmath197 almost always has spin @xmath308 , whereas the dominant spectral weight arises from a state with @xmath83 .
an example of this behavior is shown in fig .
[ spectrum](a ) for a cluster with ground state spin @xmath309 .
the spectrum is dominated by a large contribution from an @xmath310 state at @xmath311 . however , there are numerous very small contributions from @xmath312 states below this peak , including the lowest excitation at @xmath313 . in this case
the sum rules give a bound @xmath5 very close to the energy of the lowest @xmath310 state , and , thus , differs from the true gap @xmath6 by a factor of @xmath314 .
in contrast , fig .
[ spectrum](b ) shows results for a cluster with @xmath8 ground state .
here there are of course no excitations with @xmath308 , and all the spectral weight is in the @xmath315 channel .
moreover , the dominant weight originates from the lowest excitation .
the sum rule approach here gives a bound reasonably close to the true gap .
-site clusters with ground state spin @xmath309 ( a ) and @xmath8 ( b ) .
the cluster shapes are drawn in the panels .
the delta - functions in eq .
( [ dynamicss ] ) are represented by vertical lines of length equaling the spectral weight .
the symbols on top of the line indicate the spin of the corresponding excited states relative to the ground state @xmath197 . in ( b ) , there is no @xmath307 spectral weight ; the circles only indicate the locations of such states.,width=245 ] -3 mm in @xmath316 clusters with ground state spin @xmath197 .
results for @xmath317 are shown in panels ( a),(b),(c ) . in all panels ,
results for the lowest triplet excitation of @xmath8 clusters is shown for comparison.,width=226 ] -3 mm we further examine the statistics of the gaps corresponding to excitations with @xmath268 for @xmath316 clusters with ground state @xmath197 . in fig .
[ lanc - histo ] we show histograms based on several hundred clusters with @xmath317 , along with results for @xmath8 clusters for comparison .
we can see that the distribution of the @xmath83 excitations is peaked at higher energies than the @xmath308 ones , and the distance between the two distributions grows with @xmath197 . as we discussed in sec .
[ moredata_a ] , the distribution of singlet - triplet excitation gaps is double - peaked for small systems , with the upper peak diminishing as a function of the cluster size . in fig .
[ lanc - histo ] the lower part of the @xmath61 distribution is located below the @xmath308 distributions for @xmath12 clusters .
it appears plausible from these results that the @xmath14 and @xmath15 gaps can have different scaling properties .
it is also clear from these calculations that the sum rule approach for @xmath12 clusters does not reflect the true smallest gaps , which are due to @xmath308 excitations , but instead reflect the distribution of spectral weight of @xmath83 excitations .
the quantity @xmath5 therefore has a different meaning , which can still be physically relevant because many experimental techniques probe @xmath318 directly , e.g. , neutron scattering and nuclear magnetic resonance .
these experiments should observe low - energy dynamics corresponding to @xmath13 , according to our results in the previous section .
the most plausible scenario is that the lowest @xmath308 excitation energies , for large clusters and typical ground state spin ( which is of the order @xmath319 ) , scale with the same dynamic exponent @xmath320 as the triplet excitations of @xmath8 clusters ( which we will argue further also in the next section ) . while their low spectral weights would make them difficult to observe in measurements sensitive to @xmath318 , they are of course still relevant for thermodynamic properties such as the specific heat . .
solid and dashed bonds correspond to singlets and triplets , respectively .
( a ) is the ground state , ( b ) the lowest @xmath8 excitation , and ( c ) is obtained from ( a ) by acting on it with @xmath293 ( which is a good approximation to the lowest @xmath312 excitation).,width=226 ] -3 mm we now address the important issue of why the @xmath14 contribution to the spectral weight is so small .
we will argue that this is , in fact , consistent with our scenario of the low - energy excitations being due to effectively isolated magnetic moments .
to illustrate this point , fig .
[ cluster6 ] shows valence bond states for a 6-site cluster with ground state spin @xmath298 .
the state in ( a ) is constructed as an approximate ground state based on the notion that triplet bonds should be predominantly located in regions of sublattice imbalance .
this cluster has two `` dangling '' spins , which we take at maximum separation .
for the two singlet bonds , we construct a symmetric combination ( which corresponds to the true ground state of the heisenberg model on the four sites in isolation ) .
it is now natural to assume that the lowest excitation corresponds to converting the triplet bond into a singlet , as shown in ( b ) . diagonalizing the hamiltonian exactly
, we find that these simple states indeed are good approximations to the eigenstates ; the overlap of ( a ) with the true ground state is @xmath321 , while the overlap of ( b ) with the lowest @xmath8 state ( which is the lowest excited state ) is even larger , at @xmath322 . on the other hand ,
if we act with @xmath293 on state ( a ) , as explained above with the spin operator written in the form ( [ szbipartite ] ) , we obtain the state shown in fig .
[ cluster6](c ) .
this state mixes @xmath8 and @xmath312 states , and its overlap with the actual lowest @xmath312 state is @xmath323 .
the overlap of ( c ) with the approximate @xmath8 state ( b ) is exactly @xmath286 , and the overlap with the exact lowest singlet also vanishes . in the case of the ( b),(c ) overlap , it is immediately clear that it is zero because of their different states of the long bond .
the states also differ in the quantum number related to a @xmath324 rotation of the cluster ; ( b ) is odd and ( c ) even under this symmetry transformation .
the true ground state is also odd , which explains why the overlap with state ( c ) is exactly @xmath286 .
this latter property is of course particular to this symmetric 6-site cluster . in general , for a less symmetric larger cluster with two dangling spins
, we would expect some small overlap between @xmath297 and the lowest @xmath308 state , because the triplet will not be exactly localized at only two sites based on the above example , we can understand that , in general , @xmath12 ground states contain some triplet bonds connecting non - bipartite sites .
the lowest excitation should normally have spin @xmath308 and closely correspond to converting one triplet bond into a singlet . on the other hand ,
acting with the spin operator one obtains a linear combination of @xmath308 and @xmath83 states with an additional triplet bond , and the overlap of the @xmath308 component with the low - energy states with this spin is low ( because of the differing singlet / triplet state of one non - bipartite bond ) .
the lowest @xmath308 excitation should thus be very similar to the excitations we have argued for in the case of the @xmath298 excitations of a singlet ground state , which essentially corresponds to promoting a long singlet ( between two moments , which can be located far away from each other ) into a triplet . for an @xmath12 cluster
we instead demote a long triplet bond into a singlet .
this similarity also suggests that the true dynamic exponent ( giving the scaling of the lowest energy , not the dominant spectral weight ) in the case of @xmath12 clusters should be the same @xmath320 that we have found for the @xmath8 clusters .
our hypothesis for the low - energy excitations is that they are due to effectively unpairable spins on the percolating cluster . to test this hypothesis further
, we consider a case where there are no such spins ; the bilayer heisenberg antiferromagnet with `` dimer dilution '' , i.e. , two identical clusters coupled through a nearest - neighbor inter - layer coupling @xmath325 .
the hamiltonian for this system was already written down in eq .
( [ hambilayer ] ) .
its static properties were studied in refs . .
the percolating cluster remains ordered at @xmath100 when the coupling ratio @xmath326 , whereas for larger inter - layer couplings the cluster is quantum disordered .
here we consider @xmath327 ; well inside the ordered regime .
one might then expect the quantum rotor picture to be valid , as has been argued also based on field theoretical considerations,@xcite and , thus , the dynamic exponent should be @xmath328 .
scenarios , involving `` fractons '' are also possible.@xcite for dimer - diluted bilayer systems at inter - layer coupling @xmath327 at the percolation point .
( a ) shows the scaling of the full probability distribution with @xmath329 , while ( b ) shows the size dependence of the average and typical values , along with lines corresponding to the asymptotic behavior with @xmath329.,width=264 ] -3 mm fig .
[ bilayerscaling ] shows scaling results of the kind we previously discussed for the single layer .
the peak of the probability distribution of the bound @xmath5 for different cluster sizes @xmath4 ( the largest cluster of diluted @xmath69 lattices ) coincides when scaled with @xmath330 and @xmath331 .
this exponent is slightly smaller than @xmath36 , but considering statistical uncertainties of several percent and effects of subleading size corrections , @xmath43 is plausible , in contrast to @xmath320 in the single layer .
note that the data in fig .
[ bilayerscaling](a ) do not collapse onto a single curve as clearly as in the single - layer plots [ gapscalingn ] and [ gapscalingl ] .
the scaled gap distributions instead appear to become narrower with increasing @xmath4 .
this may be due to self - averaging following from the global nature of quantum rotor excitations . the local gap distribution ( not shown here ) also does not scale well with @xmath4 .
an interesting question is whether the small energy scale of the single - layer clusters at @xmath21 survives also away from the percolation point .
we here examine @xmath69 systems diluted at @xmath74 , again studying the largest cluster for each dilution realization ( which now is two - dimensional ; @xmath332 ) .
we only consider clusters with ground state spin @xmath8 .
[ singlelayerp03 ] shows results for the gap upper - bound at @xmath333 . for the largest few sizes the data are consistent with power - law scaling corresponding to @xmath334 ( with a statistical error of @xmath335 ) ; very different from the behavior at @xmath21 .
given our scenario for the excitations exactly at @xmath21 , the much smaller @xmath10 away from @xmath21 is either an indication of the moment regions not existing , or their mutual effective couplings ( or their couplings to the rest of the cluster ) being much stronger , thereby invalidating the picture of an effective low - energy subsystem .
we still expect regions of sublattice imbalance away from @xmath21 , as we will discuss further in the next section .
it may not be surprising , however , that the moments associated with these are not weakly coupled , because for any @xmath74 the largest cluster has a finite spin stiffness ( also in the thermodynamic limit ) , whereas exactly at @xmath21 the stiffness vanishes ( although the cluster is still ordered).@xcite the order is thus much more robust , and as a consequence all the effective moments at @xmath74 may be locked to the global nel vector and can not be regarded as weakly coupled semi - independent degrees of freedom .
we will discuss this further in sec .
[ perco - summary ] . of single - layer clusters at dilution fraction @xmath333 . for @xmath336
, the distribution can be collapsed with dynamic exponent @xmath337 as shown in ( a ) . for smaller sizes ,
there is a cross - over behavior , as shown in ( b ) for the average and typical @xmath5 .
the two lines correspond to @xmath338 ( a possible asymptotic value ) and @xmath339 ( a pseudo - scaling exponent in a cross - over regime).,width=245 ] -3 mm for systems very close to the percolation point , @xmath340 , one can not expect to detect differences from the behavior exactly at @xmath21 . for the smaller cluster sizes at @xmath333 we can observe in fig .
[ singlelayerp03 ] what is likely a cross - over behavior from the behavior at @xmath21 to the asymptotic scaling behavior at @xmath333 .
the effective exponent below sizes @xmath341 is smaller than the value we found at @xmath21 , but , on the other hand , @xmath333 is already quite far away from @xmath21 and it is not surprising that a different behavior obtains here .
closer to @xmath21 we expect data for small sizes to scale with @xmath320 , but to observe clearly this scaling , followed by a cross - over to @xmath334 , would require larger clusters than we can access currently .
it should be noted that the results discussed here ( and those for the bilayer in the previous section ) do neither prove that the mapping to quantum rotors holds for @xmath74 ( and in the bilayer at @xmath21 ) , nor that the dynamic exponent exactly equals @xmath342 ( or @xmath36 ) .
for fracton excitations , one would expect @xmath343 ( but close to @xmath36).@xcite it would therefore be useful to determine @xmath10 for the single layer at @xmath74 and the bilayer at @xmath21 to higher precision , which , however , is a very demanding task that we leave for future studies .
in the mapping of a quantum antiferromagnet onto a quantum rotor model,@xcite one assumes that there is local antiferromagnetic order on some length scale @xmath344 .
a subsystem @xmath115 of the system , of length @xmath344 , is then replaced by a quantum rotor @xmath345 , which can reproduce the `` anderson tower '' of low - energy states of different total spin @xmath197 ( which the subsystem would exhibit in isolation ) .
the rotors for all the subsystems are then coupled in a way consistent with the expected dominant fluctuations and symmetries of the system .
for such a mapping to produce the correct physics , the subsystems should consist of an even number of spins , arranged in such a way that their ground state , in isolation , is a singlet .
if the system geometry does not allow for such a decomposition , the situation will be more complicated .
the question is then ; how can one decompose the system into quantum rotors and `` left - over '' spins in a well defined way , which maintains the salient features of the disordered clusters ?
the smallest unit for which a local quantum rotor can be considered in a hypothetical mapping is a dimer consisting of two nearest - neighbor spins , which in isolation has a singlet @xmath346 ground state and a triplet ( @xmath347 ) excited state .
this corresponds to a quite severely truncated rotor tower , but the local cut - off should not matter for the low - energy physics of the coupled system .
we have already discussed the fact that a disordered cluster can not normally be fully decomposed into such dimers , as there would in most cases be some `` dangling spins '' ( or , more generally , regions of imbalance in the sublattice occupation numbers ) left over after the cluster has been maximally covered with close - packed dimers . if we consider larger subsystems , there will be similar problems , i.e. , not all subsystems will have singlet ground states in isolation .
we will here proceed to investigate the geometric decomposition of the system into nearest - neighbor dimers and left - over monomers . in a standard classical dimer model,@xcite
a dimer corresponds to two connected nearest - neighbor sites , here on a square lattice .
the statistical mechanics problems corresponds to counting all the dimer coverings . in a dimer - monomer model,@xcite
there are also some unpaired sites present , and the counting now includes all possible dimer and monomer configurations ; normally at a fixed density of monomers . in the case at hand here
, we investigate disordered clusters , and we want to maximally cover the clusters with dimers .
there will then typically be some left over monomers that can not be paired . for a given cluster , we want to sample dimer - monomer configurations with the smallest possible number of monomers .
we are interested in the spatial distribution of monomers , which provides us with a concrete quantitative measure of `` sublattice imbalance '' .
we want to identify the regions of sublattice imbalance and investigate the size distribution of these regions . here
we consider clusters constructed on @xmath69 lattices , with , as before , only the largest cluster found in each realization included in the statistics . in monte carlo sampling of the dimer - monomer configurations on these clusters ,
we start with an arbitrary configuration , e.g. , one containing only monomers .
the updating scheme is illustrated in fig .
[ dimermcmoves ] . in ( a ) ,
when two monomers are located next to each other , they are annihilated and form a dimer . dimers and monomers
can be updated together according to the two moves shown in ( b ) .
we can also break a dimer into two monomers , as in ( c ) .
one of the monomers is then moved , together with dimers as in ( b ) , until it encounters a monomer ( which can , but does not have to be , the same as the one it was originally paired with ) , together with which it again can be combined to form a dimer as in ( a ) .
this is an efficient way to update parts of the cluster where there are no monomers ( other than the two introduced for the purpose of the update ) .
this simulation process will eventually converge to a state with the minimum number of monomers for a given cluster , because the monomer annihilation process ( a ) is always carried out when possible .
whenever two monomers are created , they will eventually be annihilated .
our model is therefore also an aggregation model for dimers .
-3 mm for a given cluster generated on an @xmath69 lattice , after a long equilibration to make sure that the minimum monomer number has been reached , we collect statistics .
one quantity of interest is the average monomer density for each site .
most of the sites never have any monomers .
we define a `` moment '' as a region consisting of @xmath348 sites to which one or several monomers are confined . by definition , a monomer inside such a moment can not move to a different moment through the monte carlo processes .
in addition , the moments consist only of sites on the same sublattice , because an individual monomer only moves on a given sublattice , as in fig .
[ dimermcmoves](b ) .
two moments on different sublattices can not have any sites that are nearest neighbors , because then two monomers in these different regions could become adjacent and annihilate each other .
( top panel ) and at @xmath333 ( bottom panel ) graphed on a log - log scale .
hundreds of realizations of the largest cluster on @xmath69 lattices were used . for large @xmath4 the distributions collapse onto a single curve , reflecting a finite typical moment size .
the average moment size is @xmath349 at @xmath21 and @xmath350 at @xmath333 .
the asymptotic form of the probability distribution for large clusters is @xmath351 , as shown in the insets using a semi - logarithmic scale.,width=274 ] -3 mm keeping track of the moment regions and their sizes involves straight - forward book - keeping , and we just proceed to discuss results .
[ momentsize ] shows the size distribution of the moments both at the percolation point and away from it , at @xmath333 . for small moment sizes
@xmath348 , the distribution is close to a power - law , especially at @xmath333 , but there is a cross - over to a clearly exponential decay for large @xmath348 .
the distribution can be fitted well with the form @xmath352 , with @xmath353 and @xmath354 at @xmath21 and @xmath333 , respectively .
the average moment size @xmath355 computed as a sum over all the sizes is smaller ; @xmath356 at @xmath21 and @xmath357 at @xmath333 . in the figure ,
the largest cluster sizes are much larger than @xmath355 and @xmath358 , and the curves for the largest @xmath4 overlap almost completely .
these calculations prove that the notion of local sublattice imbalance is well defined and quantifiable .
finite moment regions exist both at and away from the percolation point . in the next section , we will present result from valence bond quantum monte carlo simulations in the triplet sector .
we will there also look at the spatial distribution of the monomers in the classical dimer - monomer model , and compare it with the distribution of the triplet in the lowest excitation of the actual quantum spin model .
as discussed in sec . [ ipr ] , the valence - bond projector qmc method offers us the possibility to examine the lowest triplet state in a unique way . in a disordered system ,
the spatial distribution of the triplet bond gives a very direct measure of the extent to which different parts of the system are affected when exciting a cluster with @xmath8 ground state to its lowest @xmath298 state .
an example of the triplet density for a very small cluster was already presented in fig .
[ clusterdemo ] .
it should be noted that the statistics of the singlet bonds is also affected by the presence of a triplet bond , and , thus , just examining the properties of the triplet bond does not give a complete picture of the excitation .
however , if a large region of the system has no ( or very low ) average triplet density , then the singlets of that region should also not be much affected .
the spatial distribution of the triplet should therefore provide a valid measure of the tendency ( if any ) of the triplet excitation to localize . .
the clusters were generated on @xmath359 lattices .
the dashed curves are of the form @xmath360 , with @xmath361 adjusted to fit the last few ( large @xmath362 ) points.,width=264 ] -3 mm sites ; the classical monomer density ( a ) , the triplet density ( b ) , and the local gap estimate @xmath45 ( c ) .
all quantities are shown on a color scale ranging from the smallest ( @xmath286 in the case of @xmath363 and @xmath364 ) to the largest value .
the absolute values are irrelevant for the purpose of the discussion here . in ( a ) , the numbers inside the squares label the different classical moment regions .
the dots indicate empty sites.,width=217 ] -3 mm we study the site dependent triplet density @xmath365 , where the triplet occupation number @xmath366 is defined such that if a triplet bond connects sites @xmath115 and @xmath137 , then @xmath367 and @xmath368 , while @xmath369 for all other sites @xmath370 .
in addition to visually examining the triplet density for representative individual clusters , it is also useful to have a quantitative measure of localization . for this purpose
, we use the inverse participation ratio ( ipr ) corresponding to the triplet density ; @xmath371 this quantity characterizes the number of sites involved in a triplet excitation , and can be averaged over cluster realizations .
two extreme cases can help to clarify the meaning of @xmath372 ; if the triplet is completely localized on only two sites , then @xmath373 , while if it is equally spread out over all the sites of an @xmath70-site cluster , then @xmath374 . we will study the dependence of @xmath375 on the cluster size . ) and the classical monomers ( @xmath376 ) .
the lines correspond to scaling @xmath377 , with @xmath378 ( based on a line fit ) for @xmath372 and @xmath379 for @xmath376.,width=245 ] -3 mm in a projector method based on a power @xmath380 , one converges to the lowest state in a given symmetry sector when the power @xmath133 of the hamiltonian is sufficiently high . for an @xmath70-site cluster
, one would expect that the @xmath133 required for convergence scales as @xmath70 or worse .
this can be seen if we compare with an alternative projection method the imaginary - time evolution @xmath381 of the trial state . here
@xmath123 is analogous to an inverse temperature ; starting from a state at some `` temperature '' , we `` cool it '' by increasing @xmath123 .
a better trial state corresponds to a lower initial temperature . for large @xmath123 ,
the dominant power @xmath382 in a taylor expansion of @xmath383 is @xmath384 , where @xmath385 is the ground state energy , which is proportional to the cluster size @xmath70 .
thus , if we project with just a fixed power @xmath133 of @xmath130 , we would get essentially the same result if @xmath386 . the energy scale of the excitations decrease with increasing @xmath70 ( very quickly so in the problem under consideration here , because the dynamic exponent is large ) , and we should therefore expect to need larger @xmath123 for larger @xmath70 .
thus , in the fixed - power scheme , the @xmath133 required for convergence should increase as some power ( larger than one ) of @xmath70 .
we show examples of the convergence of the ipr for three different clusters in fig .
[ vbsimulation ] .
the large fluctuations in the finite - size gap , discussed in sec .
[ moredata ] , naturally also imply large variations in the convergence rate of the triplet ipr ( which is governed by the gap between the first and second triplet , which also exhibits large fluctuations ) . as explained in sec .
[ ipr ] , we are restricted to @xmath133 for which the triplet survival probability in the projection is reasonably large . in order to ensure that the results truly reflect the lowest excitation for each cluster , we carry out extrapolations to infinite @xmath133 using a simple exponential form , as explained in the caption of fig .
[ vbsimulation ] .
the fluctuations of the disorder averaged @xmath375 are completely dominated by the cluster - to - cluster variations , and we believe that any remaining errors related to the convergence are smaller than the final error bars ( based on a few hundred clusters of each size ) . of the singlet ( top panel ) and triplet ( bottom panel ) bonds for the cluster in fig .
[ 3clusters ] .
the singlet distribution is strongly peaked at short bonds , and we have therefore cut off the corresponding peak in the low left corner of the histogram . the the remaining weight is @xmath387 of the total . , width=264 ] -3 mm we first examine the spatial distribution of the triplets . the triplet density for each site of a typical cluster
is shown using a color scale in fig .
[ 3clusters](b ) .
here we compare the triplet density with two other calculations the classical monomer density in ( a ) and the local gap @xmath45 in ( c ) .
it is apparent that the triplet is concentrated to a relatively small fraction of all the sites of the cluster . at the same time , the affected sites form groups that are spread out over the cluster .
this is exactly in agreement with our hypothesis of low - energy excitations involving a number of localized moments .
it is also clear from fig .
[ 3clusters ] that the classical monomer density is high wherever the triplet probability is significant .
this proves that our measure of sublattice imbalance in terms of classical monomers indeed corresponds very closely to the actual locations affected by excitations .
note also that some sites with high monomer density do not have a high triplet density .
this is also expected , because the _ lowest _ triplet excitation should not necessarily involve all of the classical monomer regions . higher triplet states may involve other subsets of moments . finally , there is a very good correspondence between regions of low local gaps @xmath45 and hight triplet density .
next , we discuss the ipr of the triplet .
it is interesting to compare this with the total number of spins in the classical dimer - monomer model .
we therefore also define a classical ipr , as in eq .
( [ iprdefinition ] ) but with the triplet density @xmath190 replaced by the classical monomer density .
both these iprs , averaged over several hundred clusters , are shown versus the cluster length @xmath4 on a log - log scale in fig .
they both scale according to power laws .
the classical ipr is consistent with the form @xmath388 . in combination with the fact that the individual moment regions are finite ,
as we showed in the previous section , this is in agreement with our extremal - value analysis in sec .
[ extremal ] , which relied on the number of effective moments being proportional to @xmath70 . however , the triplet ipr scales with a smaller power ; @xmath389 with @xmath378 .
thus , not all the effective moments are involved in the lowest excitation , but since the size of the excitation still grows with @xmath4 these are not localized excitations .
another important aspect of the valence - bond calculation is that the bond lengths also contain information directly pertaining to the nature of the excitations . in fig .
[ stlengths ] we show the distributions of both the singlet and triplet bonds lengths for the cluster shown in fig . [ 3clusters ] . while the triplet bond is typically long , the singlet distribution is strongly peaked for the shortest bonds .
we have therefore cut off more than @xmath390 of the weight in the singlet histogram in order to be able to show the more interesting distribution of long bonds .
every peak in the triplet distribution can be perfectly matched to the distance between two regions ( on different sublattices ) with a high concentration of triplets / monomers in fig .
[ 3clusters ] .
this again supports the notion of excitations of weakly interacting effective moments . in the singlet distribution , there are also features corresponding to the same lengths as in the triplet case .
this is also what one would expect if the triplet state essentially corresponds to promoting a long singlet to a triplet in a superposition , as discussed in sec .
[ vbtheory ] .
the average length of the triplet bond also scales with the cluster size according to a power - law , as shown in fig .
[ ltscaling ] .
this power law , in combination with that for the triplet ipr ( fig .
[ rtrc ] ) and classical percolation exponents , should be related to the dynamic exponent @xmath261 .
exactly how is not presently clear , however .
the line corresponds to a power - law divergence @xmath391 , with @xmath392.,width=245 ] -3 mm
to summarize , we have discussed several calculations aimed at elucidating the quantum dynamics of the @xmath0 antiferromagnetic heisenberg model on randomly diluted clusters .
quantum monte carlo simulations in combination with sum rules show that the low - energy excitations at the percolation point are described by an unexpectedly large dynamic exponent ; @xmath393 , where @xmath9 is the fractal dimension of 2d percolation . using extremal - value statistics
, we were able to relate @xmath10 and two exponents characterizing the probability distribution of local gaps ( an exponent @xmath54 governing the size dependence and @xmath224 describing the distribution of small gaps ) , according to eq .
( [ zrelation ] ) .
this kind of scaling indicates that the excitations involve effective localized finite magnetic moments , which interact through the remaining , magnetically inert parts of the cluster .
this is also confirmed directly by imaging the spatial distribution of the triplet excitations in the valence bond basis , where the triplet state can be described in terms of a pair of spins forming a triplet in a `` singlet soup '' of valance bonds .
the triplet bond fluctuates between several isolated regions , and its average length scales as a power of the cluster size .
the average number of spins affected by the excitation also grows as a power @xmath394 of the system size , with @xmath395 .
the excitations are thus localized at multiple moment regions , which are spread out over the cluster .
all these results lead to a picture of an effective low - energy system consisting of a network of globally entangled local moments , where the moments correspond to regions of sublattice imbalance .
we have introduced a quantitative measure of sublattice imbalance , in terms of a classical dimer - monomer aggregation model .
monte carlo simulations of this model show that the monomers form isolated finite regions , and the number of such regions scales linearly in the cluster size @xmath70 . sites with a high triplet concentration coincide very well with high monomer density , confirming directly that sublattice imbalance in the heisenberg model is associated with the formation of weakly interacting effective moments .
we have also shown that when two identical clusters are coupled in a bilayer , with a small inter - layer coupling @xmath396 ( smaller than the value at which the long - range order vanishes @xcite ) , the low - energy excitations change dramatically . a finite - size scaling analysis for clusters with @xmath397
show a much smaller dynamic exponent , @xmath17 , than for the single - layer clusters .
there is no sublattice imbalance in this `` dimer diluted '' bilayer model , and the results therefore provide additional evidence for the important role played by effective moments at imbalanced regions in the single - layer clusters .
the result @xmath261 was obtained by studying clusters in which the sublattices are balanced globally , i.e. , the ground state spin is @xmath8 .
we have also pointed out that for clusters with global imbalance in the sublattice occupation ( leading to @xmath12 ) , the dynamic structure factor has spectral weight predominantly arising from the @xmath398 channel , while there is very little weight in the @xmath307 and @xmath399 channels ( apart from the elastic @xmath307 weight ) .
experiments directly probing the inelastic spectral weight , e.g. , neutron scattering and nuclear magnetic resonance , should be dominated by the @xmath398 channel , and @xmath197 should be typically large ( @xmath400 ) for random clusters .
for this situation our sum rule method gives a smaller effective dynamic exponent , @xmath401 , than for the @xmath61 excitations of globally balanced clusters .
the lowest - energy excitations , which are in the @xmath399 channel , are not accessible with the sum rule approach .
we have argued , based on an analysis of approximate ( variational ) valence bond states , that their energy should scale with the same @xmath402 as in the case of @xmath403 excitations .
there have been attempts previously to determine the dynamic exponent of the diluted heisenberg model based on quantum monte carlo calculations .
_ studied the temperature dependence of the correlation length and concluded that it scales in a way consistent with @xmath43 at the percolation point.@xcite however , the scaling assumption was one corresponding to quantum - criticality , which may not apply because the percolating cluster at @xmath100 does not have quantum critical fluctuations in the sense of power - law decaying correlation functions .
a similar treatment of the clean 2d heisenberg model would fail to give @xmath35 ( which is not a quantum - critical exponent but one characteristic of the quantum - rotor excitations of the nel state ) , because the correlation length diverges exponentially as @xmath121.@xcite we have here focused exclusively on the dynamics of the percolating cluster . in order to relate the results quantitatively to specific experiments , one should include the contributions from all clusters .
the cluster distribution is given by classical percolation theory,@xcite which can be combined with the finite - size scaling properties that we have found here for the distributions of the local and smallest gaps .
we have also not discussed the consequences of our @xmath100 results for the @xmath120 behavior .
this is of course also an important experimental issue , and will be interesting to consider in future studies . moving away from the percolation point @xmath21 ,
our calculations show that the dynamic exponent of the single layer is @xmath404
. however , the classical dimer - monomer model has finite localized monomer regions also away from the percolation point .
this suggests that the change in the spin dynamics upon moving away from @xmath21 is related to the effective interactions between the moment regions , not the disappearance of the moments .
such a qualitative change in the interaction aspects of the moments is not completely unexpected , since the spin stiffness of the percolating cluster at @xmath21 is strictly zero in the thermodynamic limit @xcite ( although the cluster is ordered ) , whereas it becomes finite ( according to a power - law ) away from the percolation point .
the more robust cluster order for @xmath74 should qualitatively change the effective interactions between distant monomer regions , likely locking all of them to the global nel vector ( which is the case for a single moment in a 2d system @xcite ) .
the effectively independent nature of the magnetic moments exactly at the percolation point ( and the very weak interactions between them , are thus intimately related to the fractal structure and related vanishing spin stiffness of the network connecting the moment regions .
we presented some results showing the cross - over from scaling controlled by the percolation point to 2d behavior . in spite of the close agreement with the dynamic exponent expected based on the quantum rotor mechanism @xcite for the single layer away from @xmath21 and the bilayer at @xmath21 ,
it is still not certain that the lowest - energy excitations in these systems are quantum rotor states . over the years
there have been considerable efforts to understand the dynamics of various randomly diluted systems close to and at the percolation point .
the `` fracton '' has been introduced as a generic excitation which develops out of plane waves ( e.g. , spin waves for an antiferromagnet ) for a translationally invariant system upon dilution.@xcite numerical calculations based on spinwave theory show that the dynamic exponents for fractons in the 2d percolating antiferromagnet is very close to @xmath36.@xcite this calculation does not properly account for the vanishing spin stiffness of the percolating cluster at @xmath21 and the existence of localized moments , but it may still be valid close to the percolation point ( where a finite stiffness develops ) .
it is possible that the dynamic exponents @xmath405 and @xmath17 that we have obtained here for the single layer with @xmath18 and the bilayer at @xmath21 , respectively , are due to fractons , not quantum rotors .
however , exactly at the percolation point , the physics of the globally entangled local moments that we have discussed here is clearly different from fractons ( which can exist in systems that do not have any objects corresponding to localized moments ) and the value of the dynamic exponent is twice that expected based on fractons.@xcite it is thus possible that several types of excitations co - exist at and close to the percolation point ; quantum rotors , fractons , and the globally entangled moment excitations .
one may still be able to describe the low - energy physics of the system away from the percolation point as a network of weakly interacting moments , but now in the presence of a staggered field mimicking the coupling to a common nel vector ( i.e. , the sign of the field depends on which sublattice a moment is associated with on the original lattice ) .
the strength of the effective staggered field ( which for a finite cluster should be allowed to have a fluctuating direction as well @xcite ) should increase upon moving away from the percolation point ( being @xmath286 at @xmath21 , due to the vanishing spin stiffness the energy scale of twisting the nel order globally ) .
most likely , even an infinitesimal field will asymptotically ( for large clusters ) change the dynamic exponent .
the role of effectively isolated spins in the formation of long - range order @xcite on the percolating cluster has been pointed out by bray - ali et al .
@xcite although some spins can be very weakly coupled ( effectively ) to the rest of the cluster , correlations between them can be stronger than within the backbone of the cluster .
arbitrarily weakly coupled moments formed by groups of spins can also correlate over long distances , and hence even a `` floppy '' fractal cluster ( one with vanishing spin - stiffness ) can order at @xmath100 .
this picture seemingly contains some of the ingredients of our entangled moments picture .
however , the same ordering mechanism was argued to apply both to antiferromagnets and systems of coupled quantum rotors , whereas we have shown here that the bilayer ( which should correspond more closely the coupled rotor system , since there is no sublattice imbalance ) and the single layer behave dramatically different .
since the excitations of the single layer cluster are also much lower in energy than the quantum rotor states considered in previous discussions of the dynamic exponent,@xcite a theoretical treatment within a quantum rotor picture is clearly not adequate . in refs . and
the excitations of the diluted system were analyzed using spin - wave theory , but this method also does not capture the significance of almost isolated moments and their long - range global entanglement , and no unusually low energy scale was discussed . in field - theory language ,
the `` dangling '' spins , or regions of sublattice imbalance , that we have discussed here correspond to uncompensated berry phases.@xcite although it is quite clear that these should exist in diluted quantum antiferromagnets , how to properly take them into account in analytical calculations for these systems is not well understood . to our knowledge ,
the resulting globally entangled moments excitations that we have argued for here have not been discussed previously in the literature .
the effective low - energy system is similar to the random antiferromagnet considered by bhatt and lee , @xcite and also by sachdev and ye.@xcite however , there is an important difference in that the nearest - neighbor interactions in our system are not frustrated .
an effective low - energy hamiltonian should then also not be frustrated .
the bhatt - lee calculation @xcite was focused on the thermodynamic properties and did not address the dynamic exponent .
the method applied was a generalization of the strong disorder renormalization ( singlet decimation ) scheme by ma , dasgupta , and hu,@xcite which has been applied to numerous random antiferromagnetic heisenberg systems.@xcite it would be interesting to apply this method also to the diluted clusters .
however , there is a technical problem in doing this directly , since the decimation scheme is based on random couplings ( successively eliminating the strongest coupled spin pair and including their remaining effects as modified couplings calculated perturbatively ) , whereas in the diluted system all couplings are the same
. it may be possible to carry out a decimation procedure by eliminating strongly _ correlated _ spins , instead of strongly _ coupled _ ones .
the correlations could be computed perturbatively based on regions of a small number of spins , or using quantum monte carlo simulations .
this way , one could study the renormalization flows of the correlations and how they relate to the sublattice imbalance that we have quantified here in terms of the classical dimer - monomer systems .
the final stages of the decimation procedure should lead to bonds ( entanglement ) between the sites on which our projector qmc calculations give a high triplet probability .
however , we have shown that there are large fluctuations in the long bonds ( singlet as well as triplet ) and it is therefore clear that the scheme can not asymptotically give the correct ground state and low - energy excitation in terms of a single bond configurations . in one dimension
, the final `` random singlet state '' is known to be asymptotically exact,@xcite in the sense that a single bond configuration is a good representation of a superposition including fluctuations around this reference state.@xcite with large fluctuations of long valence bonds among many moments in the percolating clusters , it seems unlikely that a single reference configuration would be a good approximation in this case .
it would still be interesting to investigate the flow of the renormalized coupling distribution . to go further in developing an understanding of the excitations of the weakly interacting effective moments , instead of working with the full percolating clusters it may be better to explicitly construct the effective low - energy hamiltonian we have discussed here . while the geometrical locations of the moments could be obtained using the classical dimer - monomer model , the effective couplings are more challenging .
one approach would be to just study a bipartite network of spins with some suitable form of the interactions ( which should be non - frustrated , with antiferromagnetic couplings between sublattices and ferromagnetic intra - sublattice couplings ) . in principle
the spins should have mixed @xmath197 . in one dimension
such a system is known to have different properties than the random @xmath0 chain with only antiferromagnetic couplings.@xcite the effective system could be studied with the methods used here , as well as with the strong - disorder decimation scheme . comparing results for the moment network with the dynamic exponents we have extracted here for the full cluster system ( and investigating the robustness of the exponents to variations in the couplings ) could shed further light on this challenging problem .
we would like to thank yu - cheng lin and antonio castro neto for useful discussions .
this work was supported by the nsf under grant
most of the numerical computations were carried out at the center for computational science at boston university .
c. c. wan , a. b. harris , and j. adler , j. appl .
phys . * 69 * , 5191 ( 1991 ) ; c. yasuda and a. oguchi , j. phys . soc.jpn . * 66 * , 2836 ( 1997 ) ; ibid . *
68 * , 2773 ( 1999 ) ; y .- c .
chen and a. h. castro neto , phys .
b * 61 * , r3772 ( 2000 ) . | we study the low - energy dynamics of @xmath0 antiferromagnetic heisenberg clusters constructed by diluting a square lattice at vacancy concentration @xmath1 at and below the percolation threshold @xmath2 .
the finite - size scaling behavior of the average excitation gap , @xmath3 , where @xmath4 is the cluster length , is obtained using quantum monte carlo results for an upper bound @xmath5 to @xmath6 , derived from sum rules . at the percolation threshold ,
we obtain a dynamic exponent @xmath7 for clusters with singlet ( @xmath8 ) ground state . here
@xmath9 is the fractal dimensionality of the percolating cluster .
we argue that this large dynamic exponent roughly twice that expected for quantum - rotor excitations is a consequence of weakly interacting localized effective magnetic moments , which form due to local sublattice imbalance .
this picture is supported by an extremal - value analysis of local spectral gaps , which delivers an exponent relation ( between @xmath10 and two exponents characterizing the local gap distribution ) reproduced by our simulation data .
however , the average @xmath11 over all clusters , which have mostly ground state spin @xmath12 , scales with a smaller exponent than for the @xmath8 clusters alone ; @xmath13 .
lanczos exact diagonalization for small clusters show that typically , @xmath14 in the lowest - energy excitations , while the dominant spectral weight originates from @xmath15 excitations .
thus , the scaling of @xmath16 for clusters with ground state @xmath12 does not reflect the lowest - energy excitations , but the higher @xmath15 excitations .
this result can be understood within a valence - bond picture . to further explore the scenario of localized moments ,
we introduce a classical dimer - monomer aggregation model to study the distribution of nearest - neighbor sites forming dimers ( which are the objects used in mapping to the quantum - rotor model ) and unpaired spins ( monomers ) .
the monomers are localized , and , thus , effective magnetic moments should form in the spin system .
we also study the lowest triplet excitation of @xmath8 clusters using quantum monte carlo calculations in the valence bond basis .
the triplet is concentrated at some of the classical monomer regions , confirming the mechanism of moment formation .
the number of spins ( and moment regions ) affected by the excitation scales as a non - trivial power of the cluster size . for a dimer - diluted bilayer heisenberg model with weak inter - layer coupling ( where the system remains nel ordered ) , there is no sublattice imbalance . in this case
we find @xmath17 , consistent with quantum rotor excitations .
for a single layer at @xmath18 we find @xmath19 , which indicates that the weakly interacting localized moment mechanism is valid only exactly at the percolation point .
there is a cross - over behavior close to the percolation point . |
the scattering amplitude for one - gluon exchange in an @xmath0 gauge theory is proportional to @xmath1 where the @xmath2 is the generator of the gauge group , and the @xmath3 and @xmath4 are the color indexes of the two quarks in the incoming and outgoing channels respectively . for @xmath5 ,
the negative sign in front of the antisymmetric antitriplet indicates the interaction is attractive , while the positive sign in front of the symmetric sextet indicates the interaction is repulsive @xcite .
the attractive interaction favors the formation of the diquark states in the color antitriplet , and the most stable diquark states maybe exist in the color antitriplet @xmath6 , flavor antitriplet @xmath7 and spin singlet @xmath8 channels due to fermi - dirac statistics @xcite . on the other hand
, the study based on the random instanton liquid model indicates that the instanton induced quark - quark interactions are weakly repulsive in the vector and axial - vector channels , strongly repulsive in the pseudoscalar channel , and strongly attractive in the scalar and tensor channels @xcite .
the conception of diquarks has many phenomenological applications @xcite , for example , the quark - diquark bound states scenario of the ground state baryons @xcite , the @xmath9 rule in the non - leptonic weak decays @xcite , the @xmath10 ratio of the proton and neutron deep inelastic structure functions in the limit @xmath11 @xcite , the color superconductivity in the cold dense quark matters @xcite . taking the diquarks as basic constituents , we can obtain a new spectroscopy for the mesons and baryons @xcite . the numerous candidates with @xmath12 below @xmath13 can not be accommodated in one @xmath14 nonet , some are supposed to be glueballs , molecular states and tetraquark states @xcite .
the @xmath15 and @xmath16 are good candidates for the @xmath17 molecular states @xcite , however , their cousins @xmath18 and @xmath19 lie considerably higher than the corresponding thresholds , it is difficult to identify them as the @xmath20 and @xmath21 molecular states respectively .
there maybe different dynamics dominating the @xmath22 mesons below and above @xmath23 which result in two scalar nonets below @xmath24 .
the strong attractions between the diquark states @xmath25 and @xmath26 in @xmath27-wave may result in a nonet tetraquark states manifested below @xmath23 .
the conventional @xmath28 @xmath29 nonet would have masses about @xmath30 , and the well established @xmath31 and @xmath32 @xmath29 nonets with @xmath33 and @xmath34 respectively lie in the same region .
furthermore , there are enough candidates for the @xmath28 @xmath29 nonet mesons , @xmath35 , @xmath36 , @xmath37 , @xmath38 and @xmath39 @xcite . in the tetraquark models ,
the structures of the nonet scalar mesons in the ideal mixing limit can be symbolically written as @xmath40 the four light isospin-@xmath41 @xmath42 resonances near @xmath43 , known as the @xmath19 mesons , have not been firmly established yet , there are still controversy about their existence due to the large width and nearby @xmath42 threshold @xcite .
the e791 collaboration observed a low - mass scalar @xmath42 resonance with the breit - wigner mass @xmath44 and width @xmath45 in the decay @xmath46 @xcite , and the bes collaboration observed a clear low mass enhancement in the invariant @xmath42 mass distribution in the decay @xmath47 with the breit - wigner mass @xmath48 and width @xmath49 @xcite .
recently , the bes collaboration reported the charged @xmath19 in the decay @xmath50 with the breit - wigner mass @xmath51 and width @xmath52 @xcite . in general
, we may expect constructing the tetraquark currents and studying the nonet scalar mesons below @xmath23 as the tetraquark states with the qcd sum rules approach @xcite , which is powerful tool in studying the ground state hadrons .
for the conventional mesons and baryons , the `` single - pole + continuum states '' model works well in representing the phenomenological spectral densities , the continuum states are usually approximated by the contributions from the asymptotic quarks and gluons , the borel windows are rather large and reliable qcd sum rules can be obtained . however , for the light flavor tetraquark states ( and pentaquark states , for example , the @xmath53 ) , we can not obtain a borel window to satisfy the two criteria ( pole dominance and convergence of the operator product expansion ) of the qcd sum rules @xcite .
for the heavy tetraquark states and molecular states , the two criteria can be satisfied , but the borel windows are rather small @xcite .
we can take the colored diquarks as point particles and describe them as the scalar , pseudoscalar , vector , axial - vector and tensor fields respectively to overcome the embarrassment @xcite . in ref.@xcite
, we construct the color singlet tetraquark currents with the scalar diquark fields , take the diquark masses as the basic parameters , parameterize the nonperturbative effects with the new vacuum condensates besides the gluon condensate , and perform the standard procedure of the qcd sum rules to study the nonet scalar mesons below @xmath23 , the numerical results are satisfactory . in recent years , the babar , belle , cleo , d0 , cdf and focus collaborations have discovered ( or confirmed ) a large number of charmonium - like states , and revitalized the interest in the spectroscopy of the charmonium states @xcite .
for example , maiani et al identify the @xmath54 observed by the belle collaboration @xcite as a @xmath55 state with the diquark - antidiquark structure @xmath56_{s=1}[{\bar c}{\bar q}]_{s=0}+[cq]_{s=0}[{\bar c}{\bar q}]_{s=1}$ ] @xcite .
the @xmath57 and @xmath58 , @xmath59 observed in the decay modes @xmath60 and @xmath61 respectively by the belle collaboration are the most interesting subjects @xcite .
they ca nt be pure @xmath62 states due to the positive charge , and may be @xmath63 tetraquark states ( irrespective of the molecule type and the diquark - antidiquark type ) . in refs.@xcite
, the mass spectrum of the scalar light diquark states are studied using the qcd sum rules .
so it is interesting to study the scalar and axial - vector heavy diquark states with the qcd sum rules .
once reasonable values of the heavy diquark masses are obtained , we can take them as basic parameters and study the new charmonium - like states as the tetraquark states .
there have been several theoretical approaches to deal with the heavy diquark masses , such as the relativistic quark model based on a quasipotential approach in qcd @xcite , the bethe - salpeter equation @xcite , the constituent diquark model @xcite , etc .
the article is arranged as follows : we derive the qcd sum rules for the scalar and axial - vector heavy diquark states in sect.2 ; in sect.3 , we present the numerical results and discussions ; and sect.4 is reserved for our conclusions .
in the following , we write down the interpolating currents for the scalar and axial - vector heavy diquark states , @xmath64 the @xmath65 are color indexes , @xmath66 , and the @xmath67 is the charge conjugation matrix .
the two - point correlation functions @xmath68 and @xmath69 can be written as @xmath70|0\rangle \ , , \nonumber\\ \pi_{\mu\nu}(p)&=&i\int d^4x ~e^{ip.x}\langle 0|t[j_\mu(x)j_\nu^\dagger(0)]|0\rangle \ , , \end{aligned}\ ] ] where the currents @xmath71 and @xmath72 denote the @xmath73 , @xmath74 and @xmath75 , @xmath76 respectively .
the one - gluon exchange results in strong attractions in the color antitriplet channel @xmath77 , the quark - quark system maybe form quasibound states ( diquark states ) which are characterized by the correlation length @xmath78 . at the distance @xmath79 , the @xmath77 diquark state combines with the one quark or one @xmath80 antidiquark to form a baryon state or a tetraquark state , while at the distance @xmath81 , the @xmath77 diquark states dissociate into asymptotic quarks and gluons gradually .
we carry out the operator product expansion at not so deep euclidean space where the approximation of the correlation functions by the perturbative contributions and the vacuum condensates makes sense .
if we take the diquark state as an effective colored hadron and the diquark mass as an effective quantity , @xmath82 , the correlation function can be continued to the physical region , where the quark - quark correlations exist , although the diquarks are not asymptotic states , there are significant differences between the diquark states and conventional hadrons .
the correlation functions are approximated by a pole term plus a perturbative continuum .
we can insert a complete set of intermediate `` hadronic '' states ( effective hadron states ) with the same quantum numbers as the current operators into the correlation functions to obtain the `` hadronic '' representation @xcite . isolating the ground state contributions from the pole terms of the scalar and axial - vector heavy diquarks
, we get the results , @xmath83 where the pole residues @xmath84 and @xmath85 ( to be more precise , the current - diquark coupling strengths ) are defined as @xmath86 the @xmath87 is the polarization vector , the @xmath88 is the threshold parameter , the @xmath89 is the continuum threshold parameter , and the @xmath90 denotes the perturbative contributions in the operator product expansion . in the following ,
we briefly outline the operator product expansion for the correlation functions in perturbative qcd .
the calculations are performed at large space - like momentum region @xmath91 , which corresponds to small distance @xmath92 required by validity of operator product expansion .
we write down the `` full '' propagators @xmath93 and @xmath94 of a massive quark in the presence of the vacuum condensates firstly @xcite , @xmath95 where @xmath96 and @xmath97 , then contract the quark fields in the correlation functions with wick theorem , and obtain the result : @xmath98 substitute the full @xmath99 , @xmath100 and @xmath101 quark propagators into above correlation functions and complete the integral in coordinate space , then integrate over the variable @xmath102 , we can obtain the correlation functions at the level of quark - gluon degrees of freedom . once the analytical expressions are obtained , then we can take the dualities below the thresholds @xmath89 ( the continuum states above the thresholds @xmath89 are asymptotic quarks and the spectral densities on both sides coincide ) and perform the borel transform with respect to the variable @xmath103 , finally we obtain the following sum rules for the heavy diquark states contain one @xmath99 quark , @xmath104 where the subscript ( or superscript ) @xmath102 denotes the scalar and axial - vector channels , i.e. @xmath105 , @xmath106 , @xmath107 - 2m_q\langle \bar{s}s\rangle\delta(s - m_q^2)\nonumber\\ & & + m_s\langle\bar{s}s\rangle\left ( 1+\frac{m_q^2}{m^2}\right)\delta(s - m_q^2)+\frac{m_q^3\langle\bar{s}g_s\sigma gs\rangle}{2m^4}\delta(s - m_q^2)\nonumber\\ & & -\frac{m_sm_q^4\langle\bar{s}g_s\sigma gs\rangle}{6m^6}\delta(s - m_q^2)-\frac{m_q^4}{12m^4}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\frac{1-x}{x^3 } \delta(s-\widetilde{m}_q^2 ) \nonumber \\ & & + \frac{m_sm_q}{6m^2}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\frac{1}{x^2}\left(1-\frac{m_q^2}{2m^2}\right ) \delta(s-\widetilde{m}_q^2 ) \nonumber \\ & & + \frac{1}{8}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\left(1+\frac{s}{m^2}\right ) \delta(s-\widetilde{m}_q^2 ) \ , , \end{aligned}\ ] ] @xmath108 - 2m_q\langle \bar{s}s\rangle\delta(s - m_q^2)\nonumber\\ & & + \frac{m_sm_q^2\langle\bar{s}s\rangle } { m^2}\delta(s - m_q^2)+\frac{m_q^3\langle\bar{s}g_s\sigma gs\rangle}{2m^4}\delta(s - m_q^2)\nonumber\\ & & + \frac{m_s\langle\bar{s}g_s\sigma gs\rangle}{6m^2}\left(1+\frac{m_q^2}{m^2}-\frac{m_q^4}{m^4}\right)\delta(s - m_q^2)\nonumber\\ & & + \frac{m_q^2}{12m^2}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\frac{1-x}{x^2}\left(1-\frac{s}{m^2 } \right)\delta(s-\widetilde{m}_q^2 ) \nonumber \\ & & + \frac{m_sm_q}{6m^2}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\frac{1}{x^2}\left(1-\frac{m_q^2}{2m^2}\right ) \delta(s-\widetilde{m}_q^2 ) \nonumber \\ & & -\frac{1}{12}\langle\frac{\alpha_sgg}{\pi}\rangle\int_0 ^ 1dx\left(1+\frac{s}{2m^2}\right ) \delta(s-\widetilde{m}_q^2 ) \
, , \end{aligned}\ ] ] @xmath109 , @xmath110 , @xmath111 , and the @xmath112 is the borel parameter . in eqs.(9 - 11 ) , we use the dispersion relation to write the spectral densities in a compact form , the integrals of the types @xmath113 and @xmath114 should be carried out formally , i.e. @xmath115 and @xmath116 despite the values @xmath117 and @xmath118 or @xmath119 , where the @xmath120 denotes the spectral densities concerning the vacuum condensates . with a simple replacement @xmath121 , @xmath122 , @xmath123 , we can obtain the corresponding sum rules for the heavy diquark states contain one @xmath124 quark .
differentiate eq.(9 ) with respect to @xmath125 , then eliminate the pole residues @xmath126 , we can obtain the sum rules for the diquark masses , @xmath127
the input parameters are taken to be the standard values @xmath128 , @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 and @xmath135 at the energy scale @xmath136 @xcite .
the @xmath137-quark masses appearing in the perturbative terms are usually taken to be the pole masses in the qcd sum rules , while the choice of the @xmath138 in the leading - order coefficients of the higher - dimensional terms is arbitrary @xcite .
the @xmath139 mass @xmath140 relates with the pole mass @xmath141 through the relation @xmath142^{-1 } $ ] . in this article
, we take the approximation @xmath143 without the @xmath144 corrections for consistency .
the value listed in the review of particle physics is @xmath145 @xcite , it is reasonable to take @xmath146 . for the @xmath101 quark ,
the @xmath139 mass is @xmath147 @xcite , the gap between the energy scale @xmath148 and @xmath23 is rather large , the approximation @xmath149 seems rather crude .
it would be better to understand the quark masses @xmath150 and @xmath151 we take at the energy scale @xmath152 as the effective quark masses ( or just the mass parameters ) .
our previous works on the mass spectrum of the heavy and doubly heavy baryon states indicate such parameters can lead to satisfactory results @xcite . in the conventional qcd sum rules
@xcite , there are two criteria ( pole dominance and convergence of the operator product expansion ) for choosing the borel parameter @xmath112 and threshold parameter @xmath89 . in practice
, we usually consult the experimental data in choosing those parameters . here
we take a short digression to illustrate the two criteria of the qcd sum rules .
the pole contributions ( or the ratios of the pole contributions ) @xmath153 are defined by @xmath154 for a definite channel @xmath102 at the hadronic representation , and the pole dominance requires @xmath155 , i.e. the pole contributions dominate over the continuum contributions . at the level of quark - gluon degrees of freedom ,
the convergence of the operator product expansion requires the operators of increasing dimension of mass ( for example , @xmath156 , @xmath157 , @xmath158 , @xmath159 , @xmath160 ) should have smaller contributions .
if we multiply the bethe - salpeter amplitudes of the diquark states by a charge conjunction matrix @xmath67 , the scalar and axial - vector diquark states have the same bethe - salpeter equation as the pseudoscalar and vector mesons respectively , except for the interacting kernels have an additional factor @xmath41 @xcite , the scalar and axial - vector diquark states maybe have slightly larger ( or equal ) masses than ( or of ) that of the corresponding pseudoscalar and vector mesons respectively . in the qcd sum rules for the conventional mesons and baryons , we usually take the energy gap between the ground states and the first radial excited states to be @xmath161 .
we take the approximation @xmath162 and @xmath163 , and determine the central values of the threshold parameters tentatively , @xmath164 and @xmath165 , where the @xmath27 and @xmath106 denote the scalar and axial - vector diquark states respectively , the @xmath166 and @xmath167 denote the corresponding pseudoscalar and vector mesons respectively . in calculation , we take analogous pole contributions and uniform borel windows , i.e. @xmath168 and @xmath169 in the charmed and bottom channels respectively , where the platforms are rather flat .
the revelent parameters are shown explicitly in table 1 , from the table , we can see that the pole contributions are in the range @xmath170 , the contributions from the different terms in the operator product expansion have the hierarchy : perturbative - term @xmath171 , the two criteria ( pole dominance and convergence of the operator product expansion ) are well satisfied . taking into account all uncertainties of the relevant parameters , we can obtain the values of the masses and pole resides of the scalar and axial - vector heavy diquark states , which are shown in tables 2 - 3 . from table 2 , we can see that the scalar and axial - vector diquark states have almost degenerate masses with the corresponding pseudoscalar and vector mesons respectively .
we should bear in mind that those values are not necessarily the lowest masses .
in this article , we intend to obtain the possible lowest masses , which correspond to the largest correlation lengths , and impose the two criteria of the qcd sum rules on the scalar and axial - vector heavy diquark states to choose the borel parameter @xmath112 and threshold parameter @xmath89 , i.e. we take smaller threshold parameters and borel parameters ( also borel windows ) than that presented in table 1 , and adjust them to warrant the uniform pole contributions ( about @xmath172 , the smallest pole contribution presented in table 1 ) .
the preferred values are shown in fig.1 and table 4 , where we can see explicitly that the pole contributions are about @xmath173 , and the contributions from the different terms in the operator product expansion have the hierarchy : perturbative - term @xmath174 , the two criteria of the qcd sum rules are well satisfied also . on the other hand ,
the values of the masses and pole residues are rather stable with variations of the borel parameters in the borel windows .
in fact , we can take even smaller threshold parameters than that presented in table 4 , however , the borel windows are too small to make reliable predictions .
taking into account all uncertainties of the relevant parameters , finally we obtain the ( lowest ) values of the masses and pole resides of the scalar and axial - vector heavy diquark states , which are shown in figs.2 - 3 and tables 2 - 3 . in table 2
, we also present the masses from the relativistic quark model based on a quasipotential approach in qcd @xcite , the bethe - salpeter equation @xcite , and the constituent diquark model @xcite . from the table
, we can see that the values from different theoretical approaches differ from each other greatly , and one should be careful when using them . in ref.@xcite
, we introduce new qcd sum rules to study the nonet scalar mesons and take the values of the scalar diquark masses from the qcd sum rules for consistency @xcite .
the @xmath175 breaking effects for the masses of the scalar and axial - vector heavy diquark states are buried in the uncertainties .
naively , we expect the axial - vector heavy diquark states have larger masses than the corresponding scalar heavy diquark states . from table 2
, we can see that it is not the case , they have degenerate masses .
lattice qcd calculations for the light flavors indicate that the strong attraction in the scalar diquark channels favors the formation of good diquarks , the weaker attraction in the axial - vector diquark channels maybe form bad diquarks , the energy gap between the axial - vector and scalar diquarks is about @xmath176 of the @xmath177-nucleon mass splitting , i.e. @xmath178 @xcite , which is expected from the hypersplitting color - spin interaction @xmath179 , where the @xmath67 is a coefficient @xcite .
the coupled rainbow dyson - schwinger equation and ladder bethe - salpeter equation also indicate such an energy hierarchy @xcite . comparing with the light diquark states , the contribution from the hypersplitting color - spin interaction @xmath180 to the heavy diquark states
is greatly suppressed due to the large constituent quark masses , and the scalar and axial - vector heavy diquark states have almost degenerate masses . .
the borel parameters @xmath112 and threshold parameters @xmath89 for the heavy diquark states .
the `` pole '' stands for the contribution from the pole term to the spectral density .
the `` perturbative '' stands for the contribution from the perturbative term in the operator product expansion , etc , where the central value of the threshold parameter @xmath89 is taken .
the contributions from the @xmath181 and @xmath182 are not shown independently for simplicity , and @xmath183 . [ cols="^,^,^,^,^,^,^,^",options="header " , ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath190 corresponds to the central values of the threshold parameters , the energy gaps among @xmath191 , @xmath190 and @xmath192 are @xmath193 and @xmath194 for the charmed and bottom diquark states respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively .
, title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ] .
the @xmath106 , @xmath184 , @xmath67 and @xmath185 denote the channels @xmath186 , @xmath187 , @xmath188 and @xmath189 respectively .
the @xmath191 , @xmath190 and @xmath192 denote the upper bound , the central value and the lower bound respectively .
the ( i ) and ( ii ) denote the scalar and axial - vector diquark states respectively.,title="fig:",width=226 ]
in this article , we study the mass spectrum of the scalar and axial - vector heavy diquark states with the qcd sum rules in a systematic way .
the diquark masses are basic parameters in studying the tetraquark states , once reasonable values are obtained , we can study the new charmonium - like states as the tetraquark states with the new qcd sum rules developed in our previous work .
this work is supported by national natural science foundation of china , grant numbers 10775051 , 11075053 , and program for new century excellent talents in university , grant number ncet-07 - 0282 , and the fundamental research funds for the central universities .
j. weinstein and n. isgur , phys .
* 48 * ( 1982 ) 659 ; f. e. close , n. isgur and s. kumana , nucl . phys .
* b389 * ( 1993 ) 513 ; n. n. achasov , v. v. gubin and v. i. shevchenko , phys . rev .
* d56 * ( 1997 ) 203 .
r. d. matheus and s. narison , nucl .
* 152 * ( 2006 ) 236 ; z. g. wang , s. l. wan and w. m. yang , eur . phys
. j. * c45 * ( 2006 ) 201 ; z. g. wang , w. m. yang and s. l. wan , j. phys . * g31 * ( 2005 ) 971 ; h. x. chen , a. hosaka and s. l. zhu , phys . rev . * d76 * ( 2007 ) 094025 ; z. g. wang , nucl . phys .
* a791 * ( 2007 ) 106 .
z. g. wang , eur .
j. * c62 * ( 2009 ) 375 ; phys . rev .
* d79 * ( 2009 ) 094027 ; j. phys . *
g36 * ( 2009 ) 085002 ; eur .
j. * c63 * ( 2009 ) 115 ; eur .
j. * c67 * ( 2010 ) 411 ; z. g. wang and x. h. zhang , eur .
j. * c66 * ( 2010 ) 419 ; commun .
* 54 * ( 2010 ) 323 .
n. drenska , r. faccini , f. piccinini , a. polosa , f. renga and c. sabelli , arxiv:1006.2741 ; e. s. swanson , phys . rept . * 429 * ( 2006 ) 243 ; e. klempt and a. zaitsev , phys . rept . * 454 * ( 2007 ) 1 ; m. b. voloshin , prog .
* 61 * ( 2008 ) 455 ; s. godfrey and s. l. olsen , ann .
nucl . part .
* 58 * ( 2008 ) 51 . | in this article , we study the mass spectrum of the scalar and axial - vector heavy diquark states with the qcd sum rules in a systematic way . once the reasonable values are obtained , we can take them as basic parameters and study the new charmonium - like states as the tetraquark states . + zhi - gang wang + department of physics , north china electric power university , baoding 071003 , p. r. china pacs numbers : 12.38.lg ; 13.25.jx ; 14.40.cs * key words : * diquark states , qcd sum rules |
in @xcite , t. friedrich proves that on a closed connected @xmath0-dimensional riemannian spin manifold @xmath1 , the first eigenvalue of the dirac operator satisfies @xmath2 where @xmath3 is the scalar curvature of @xmath1 .
then , for @xmath4 , using the conformal behavior of the dirac operator , o. hijazi proves in @xcite that , up to a dimensional constant , @xmath5 is at least equal to the first eigenvalue of the yamabe operator . for the case of @xmath6-spheres , c. br @xcite ( see also @xcite ) shows that @xmath7 and the equality holds only for the round metric .
more recently , the spectrum of the dirac operator has been studied from an extrinsic point of view . indeed , in @xcite and @xcite , the authors prove that if @xmath8 is a @xmath0-dimensional compact hypersurface which bounds a compact domain @xmath9 in an @xmath10-dimensional riemannian spin manifold and if the scalar curvature of @xmath9 is such that @xmath11 , with @xmath12 , then the first positive eigenvalue @xmath13 of the dirac operator on @xmath8 , satisfies @xmath14 where @xmath15 is the mean curvature of @xmath8 in @xmath9 satisfying @xmath16 .
note that here the mean curvature is the trace of the second fundamental form and not the normalized trace as in @xcite and @xcite .
several geometric applications , like alexandrov type theorems or rigidity results for compact manifolds with boundary , are derived from these estimates ( see also @xcite ) .
+ in this paper , we generalize the inequality ( [ extri ] ) for untrapped surfaces in initial data sets of the einstein equation . from this estimate
, we deduce several rigidity results for the constraint equation with spherical boundary as well as uniqueness results for constant mean curvature surfaces in the minkowski spacetime . here
we introduce the setting and give the precise statement of our main result .
let @xmath17 be a spacetime satisfying the einstein field equations , that is @xmath17 is a @xmath18-dimensional time oriented lorentzian manifold such that @xmath19 where @xmath20 ( resp .
@xmath21 ) denotes the scalar curvature ( resp .
the ricci curvature ) of ( @xmath22 and @xmath23 is the energy - momentum tensor of the spacetime .
we also assume that @xmath17 satisfies the dominant energy condition that is its energy - momentum tensor @xmath23 has the property that the vector field dual to the one form @xmath24 is a future direct causal vector of @xmath25 , for every future direct causal vector @xmath26 .
let @xmath27 be an immersed spacelike hypersurface of @xmath17 with induced riemannian metric @xmath28 .
assume that @xmath29 is the future directed timelike normal vector to @xmath30 and denote by @xmath31 the associated second fundamental form defined by @xmath32 , for all @xmath33 .
here @xmath34 denotes the levi - civita connection of the spacetime .
then the gauss , codazzi and einstein equations provide _ constraint equations _ on @xmath30 given by @xmath35 where @xmath3 is the scalar curvature of @xmath36 , @xmath37 and @xmath38 denote the squared norm and the trace of @xmath31 with respect to @xmath28 and @xmath39 and @xmath40 are respectively the energy and the momentum densities of the matter fields given by some components of the energy - momentum tensor @xmath23 .
a consequence of the dominant energy condition for the spacetime is that the inequality @xmath41 holds on @xmath30 .
a triplet @xmath42 which satisfies the dominant energy condition is called an _ initial data set_. now we consider a spacelike @xmath6-surface @xmath8 in @xmath17 which bounds a compact domain @xmath9 in the initial data set @xmath42
. denote by @xmath43 the inward unit vector field normal to @xmath8 in @xmath9 so that @xmath29 and @xmath43 span the normal bundle @xmath44 of @xmath8 in @xmath45 .
the second fundamental form of @xmath8 in @xmath45 is defined by @xmath46 for all @xmath47 and where @xmath48 denotes the second fundamental form of @xmath8 in @xmath9 . here
@xmath49 denotes the levi - civita connection of the riemannian metric @xmath28 on @xmath30 .
the trace of the second fundamental form @xmath50 along the surface @xmath8 defines the mean curvature vector field @xmath51 of @xmath8 in @xmath45 and it is given by : @xmath52 its norm can be expressed as : @xmath53 where @xmath54 is the mean curvature of @xmath8 in @xmath9 and @xmath55 are its null expansions . the spacelike surfaces with @xmath56 ( or @xmath57 )
are referred to as future ( or past ) trapped surfaces . a surface with @xmath58 or @xmath59
is called an _ apparent horizon_. finally ,
if @xmath8 is such that @xmath60 and @xmath61 ( that is @xmath62 ) , it will be referred to as an untrapped surface . in particular
, an untrapped surface has a spacelike mean curvature vector .
we now give the precise statement of our main result : [ spacetime ] let @xmath9 be a compact domain with smooth boundary @xmath8 in an initial data set @xmath42 and assume that @xmath9 has no apparent horizon in its interior . if @xmath63 is an untrapped surface in @xmath17 , then the first positive eigenvalue @xmath13 of the dirac operator @xmath64 of @xmath8 satisfies @xmath65 moreover , if equality occurs then @xmath66 can be locally embedded into the minkowski flat spacetime as a spacelike hypersurface with @xmath67 as second fundamental form . using an approach proposed in @xcite , we prove ( see theorem [ mink ] ) that this estimate holds in the minkowski spacetime without any assumption on the existence of apparent horizons .
moreover , in this context , we show that equality holds in ( [ eigenspacetime ] ) only for round spheres .
this rigidity result combined with a new upper bound for @xmath13 ( see proposition [ upper ] ) applies to obtain an alexandrov type theorem ( see corollary [ genalex ] ) as well as a unified proof of the euclidean and hyperbolic versions of the alexandrov theorem for surfaces ( see corollary [ alex ] ) .
let us briefly discuss the proof of theorem [ spacetime ] .
here we use arguments developed by schoen - yau @xcite in their proof of the positive mass theorem for general initial data sets and by liu - yau @xcite in their proof of the positivity of a quasi - local mass . indeed in a first part , we prove a generalization of the inequality ( [ extri ] ) with @xmath68 for riemannian manifolds on which a non negativity condition holds on the scalar curvature .
more precisely , we have [ newextrin ] let @xmath8 be a compact hypersurface in a riemannian spin manifold @xmath30 bounding a compact domain @xmath9 .
assume that the scalar curvature @xmath3 of the domain @xmath9 and the mean curvature @xmath15 of @xmath8 satisfy @xmath69 the lowest nonnegative eigenvalue @xmath13 of the dirac operator @xmath64 satisfies @xmath70 moreover equality occurs if and only if @xmath9 carries a parallel spinor , @xmath8 has constant mean curvature , @xmath71 on @xmath9 and the eigenspace corresponding to @xmath13 consists of the restrictions of parallel spinors on the domain @xmath9 . then , starting from an untrapped surface @xmath8 which bounds a compact domain @xmath9 in an initial data set @xmath42 , we use a solution of the jang equation with dirichlet boundary condition , to get a new metric @xmath72 on @xmath9 for which theorem [ newextrin ] applies .
the conclusion follows from boundary calculations of the mean curvature of @xmath8 in @xmath73 due to yau @xcite . if @xmath8 is the boundary of a domain @xmath9 in an initial data set @xmath74 with @xmath75 , the dominant energy condition reads as @xmath76 and the untrapped condition gives @xmath77
. we can then apply theorem [ spacetime ] to get @xmath78 which is precisely the eigenvalue estimates proved in @xcite for @xmath68 and in @xcite for @xmath79 .
* acknowledgements :
* i would like to thank oussama hijazi and emmanuel humbert for their remarks and suggestions , as well as their support .
in this section , we consider a compact riemannian spin manifold @xmath80 with smooth boundary @xmath8 ( possibly with several connected components ) and denote by @xmath49 the levi - civita connection on the tangent bundle @xmath81 .
we fix a spin structure on @xmath9 and denote by @xmath82 the corresponding spinor bundle .
this complex vector bundle is naturally endowed with a left clifford multiplication @xmath83 which is a fiber preserving algebra morphism .
then @xmath82 becomes a bundle of complex left modules over the clifford bundle @xmath84 of @xmath9 .
the spinor bundle @xmath82 is equipped with a natural hermitian metric , denoted by @xmath85 , and with the spinorial levi - civita connection @xmath49 acting on spinor fields which are compatible with the clifford multiplication @xmath86 . the dirac operator @xmath87 on @xmath82 is locally given by : @xmath88 where @xmath89 is a local orthonormal frame of @xmath81 and it defines a first order elliptic differential operator . on the boundary @xmath8
, we denote also by @xmath28 the induced riemannian metric and by @xmath90 its levi - civita connection . recall that since @xmath8 is oriented , it admits a spin structure induced from the one on @xmath9 .
hence we have that the restriction @xmath91 is a left module over @xmath92 with clifford multiplication given by @xmath93 for every @xmath94 and @xmath95 . here
@xmath43 denotes the unit inward vector field normal to @xmath8 .
now the riemannian gauss formula implies that the spin connection @xmath96 defined on @xmath97 by @xmath98 for @xmath95 and the hermitian metric @xmath85 induced from that of @xmath82 are compatible with the clifford multiplication @xmath99 . here
@xmath100 denotes the weingarten map of @xmath101 in @xmath102 .
the extrinsic dirac operator of @xmath8 acting on @xmath97 is then defined by @xmath103 .
it is a well known fact that @xmath104 is a first order elliptic differential operator which is formally @xmath105-selfadjoint . by ( [ spin - gaus ] ) ,
it is straightforward to compute that for all @xmath106 : @xmath107 here @xmath108 denotes the mean curvature of @xmath8 in @xmath9 for the metric @xmath28 . in our conventions , the unit @xmath6-sphere in the @xmath109-dimensional euclidean space has constant mean curvature equal to @xmath6 .
it is important to note that since @xmath8 is spin , we can also define an intrinsic spinor bundle @xmath110 over @xmath8 and an associated dirac operator @xmath64 related with the extrinsic spinor bundle by the following identification ( see @xcite for more details ) : @xmath111 recall that in the context of riemannian spin manifold with boundary , the classical schrdinger - lichnerowicz formula asserts that : @xmath112 and it gives the integral formula : @xmath113 for all @xmath114 .
here @xmath115 denotes the @xmath105-adjoint of @xmath49 and @xmath3 is the scalar curvature of @xmath102 . from now
, we assume that there exits a smooth vector field @xmath116 such that : @xmath117 where @xmath118 and the divergence of a vector field @xmath119 is locally defined by @xmath120 we first generalize an argument of liu - yau @xcite and prove a modified version of formula ( [ lichneint ] ) appropriate to the assumption ( [ scalarcond ] ) .
more precisely , we have : [ modlich ] let @xmath80 a compact riemannian spin manifold with boundary such that there exists a smooth vector field @xmath116 satisfying ( [ scalarcond ] ) , then @xmath121 for all @xmath122 . moreover , equality occurs if and only if the spinor field @xmath123 satisfies @xmath124 for all @xmath125 .
_ proof : _ first note that since @xmath126 the stokes formula gives @xmath127 inserting this identity in the integral form ( [ lichneint ] ) of the schrdinger - lichnerowicz formula leads to : @xmath128 and using ( [ scalarcond ] ) , we conclude that : @xmath129 if we let @xmath130 , it is straightforward to compute @xmath131 we get : @xmath132 with equality if and only if @xmath133 for all @xmath125 . combining this last inequality with ( [ schrint ] ) leads to the result .
@xmath134 + now following @xcite , we use the fact that the dirac operator of the domain @xmath9 defines an isomorphism between suitable spaces to extend a spinor field on @xmath8 harmonically on @xmath9 .
more precisely , we consider the @xmath135 bag boundary condition which defines an elliptic condition for the dirac operator of @xmath9 .
it is given by the pointwise orthogonal projection @xmath136 where @xmath137 is the eigensubbundles of @xmath97 associated with the eigenvalues @xmath138 of the involution @xmath139 we then recall : [ existencemit](@xcite ) let @xmath80 be a compact riemannian spin manifold with smooth boundary @xmath8 . then for all @xmath140 , the boundary problem @xmath141 has a unique smooth solution @xmath142 . _
proof of theorem [ newextrin ] : _ let @xmath143 be a smooth eigenspinor for the dirac operator @xmath104 associated with the eigenvalue @xmath144 .
lemma [ existencemit ] ensures the existence of a smooth spinor field @xmath145 satisfying the boundary value problem : @xmath146 using formula ( [ modlich1 ] ) leads to : @xmath147 which gives the first part of the result .
the last inequality is easily derived from the boundary condition ( see @xcite or @xcite for more details ) .
moreover equality occurs in this inequality if and only if @xmath148 .
assume now that equality is achieved , then because of ( [ truc ] ) the spinor @xmath145 is a parallel spinor such that : @xmath149 on the other hand , equality is also achieved in ( [ modlich1 ] ) then : @xmath150 for all @xmath125 . since @xmath151 has no zeros , we finally have @xmath71 on @xmath9 .
the end of the proof is then similar to theorem @xmath152 in @xcite . @xmath134
theorem [ newextrin ] is a generalization of theorem @xmath152 in @xcite .
indeed , for @xmath153 we get precisely their result .
to conclude this section , it should be pointed out that the theorem [ newextrin ] can be viewed as a corollary of a more general result : [ functest ] assume that the assumptions of theorem [ newextrin ] are fulfilled .
let @xmath154 be a smooth nonnegative function on @xmath9 and @xmath155 a smooth spinor field such that @xmath156 then @xmath157 where @xmath158 is the unique harmonic extension of @xmath159 for the ( mit ) condition .
moreover , equality occurs if and only if @xmath9 carries a parallel spinor , @xmath8 has mean curvature @xmath160 , @xmath71 on @xmath9 and the solutions of ( [ diracequation ] ) consist of restrictions of parallel spinors on the domain @xmath9 .
theorem [ newextrin ] follows for @xmath161 , an eigenspinor of the dirac operator @xmath104 associated with the first nonnegative eigenvalue @xmath144 ( that is @xmath162 ) .
in this section , we recall some well - known facts on the jang equation . for more details on this subject , we refer to @xcite , @xcite or @xcite .
this equation first appears in @xcite aiming to prove the positive mass theorem using the inverse mean curvature flow but without success .
however , as shown by schoen and yau @xcite , this equation can be used to reduce the proof of the general positive mass theorem to the case of time - symmetric initial data sets ( that is @xmath163 previously obtained by the same authors in @xcite .
more recently , yau and liu @xcite defines a quasi - local mass , generalizing the brown - york quasi - local mass , and prove its positivity using the jang equation .
the problem can be stated as follow : let @xmath42 be an initial data set for the einstein equation and consider the four dimensional manifold @xmath164 equipped with the _ riemannian _ metric @xmath165 .
the problem is to find a smooth function @xmath166 such that the hypersurface @xmath167 of @xmath164 obtained by taking the graph of @xmath168 over @xmath30 satisfies the equation @xmath169 where @xmath170 denotes the mean curvature of @xmath171 in @xmath172 and @xmath173 is the trace on @xmath171 with respect to the induced metric .
this geometric problem is equivalent to solve the non - linear second order elliptic equation @xmath174 where @xmath49 ( resp .
@xmath175 ) denotes the levi - civita connection ( resp .
the hessian ) of the metric @xmath28 , @xmath176 and @xmath177 .
note that the metric induced by @xmath178 on @xmath171 is @xmath179 and can be viewed as a deformation of the metric @xmath28 on @xmath30 . in the following ,
we adopt the convention that @xmath30 and @xmath171 denote respectively the riemannian manifolds @xmath180 and @xmath181 .
analogously , if @xmath49 denotes the levi - civita connection for @xmath30 , then @xmath182 denotes the levi - civita connection on @xmath171 and so on .
assuming now that the initial data set @xmath42 comes from a spacetime satisfying the dominant energy condition , schoen and yau proved that the following relation holds on @xmath171 : @xmath183 where @xmath184 @xmath185 is the tangent part of the vector field dual to @xmath186 , @xmath187 and @xmath188 denotes the unit normal vector field to @xmath171 in @xmath164 .
all the quantities @xmath67 , @xmath39 and @xmath40 are defined on @xmath164 by parallel transport along the @xmath189-factor
. moreover , equality occurs in ( [ schoenyau ] ) if and only if @xmath190 and the second fundamental form of @xmath171 in @xmath164 is @xmath31 .
it is important to note here that in theorem [ spacetime ] , we assume that there is no apparent horizon in the interior of @xmath9 to ensure the existence of a global solution to the jang equation .
indeed , if @xmath191 is an apparent horizon then a solution of ( [ jang ] ) blows up at @xmath191 and the resulting graph @xmath171 is asymptotic to a cylinder @xmath192 . to overcome this difficulty ,
we could try to apply a trick of schoen - yau @xcite to compactify these ends , however it does not work in our situation .
indeed , the first step in schoen - yau s method is to deform the metric @xmath72 on the parts which are asymptotic to the cylinders so that these components _ coincide _ with the cylinders .
however this transformation destroys equation ( [ schoenyau ] ) and this adds a real difficulty for our purpose .
nevertheless , it seems reasonable to think that our result is true in this more general context . from the work @xcite and
since we assumed that @xmath9 has no apparent horizon in its interior , there exists a solution @xmath168 to the jang equation ( [ jang ] ) defined and smooth on @xmath9 with dirichlet boundary condition @xmath193 this boundary condition ensures that the metric @xmath72 coincides with the metric @xmath28 on the boundary @xmath8 so that the dirac operators @xmath104 acting on @xmath97 and @xmath194 on @xmath195 also coincide .
in particular , we have @xmath196 where the last equality comes from the fact that @xmath8 is a @xmath6-dimensional manifold and then @xmath197 .
moreover , from a calculation in @xcite we have : @xmath198 where @xmath199 and thus @xmath200 holds since @xmath201 . here @xmath202 denotes the unit outward normal vector field of @xmath8 in @xmath203 . on the other hand , the resulting riemannian manifold @xmath203 satisfies condition ( [ scalarcond ] ) because of ( [ schoenyau ] ) and where the vector field @xmath204 is defined by ( [ defx ] ) .
the assumptions of theorem [ newextrin ] are fulfilled and then @xmath205 combining ( [ boundblack ] ) and ( [ new1 ] ) give the estimate .
now assume that equality is achieved .
once again we apply theorem [ newextrin ] and then @xmath203 has a parallel spinor field @xmath159 .
in particular , @xmath203 is ricci flat and since it is a @xmath109-dimensional domain , it is flat . moreover , if we have equality in ( [ schoenyau ] ) , then the second fundamental form of @xmath203 in @xmath164 is @xmath67 .
so we can choose a coordinates system @xmath206 in a neighborhood @xmath207 of a point @xmath208 such that @xmath209 .
in this chart , we have : @xmath210 [ modlocmass ] this result leads to the definition of a new quasi - local mass for @xmath6-spheres with positive gauss curvature similar to liu - yau @xcite .
the idea here , following @xcite and @xcite , consists in localizing witten s proof of the positive mass theorem @xcite .
namely let @xmath8 denotes a @xmath6-sphere with positive gauss curvature which bounds a compact domain @xmath9 in an initial data set @xmath42 .
since @xmath8 has positive gauss curvature , the weyl s embedding theorem ( @xcite or @xcite ) ensures the existence of an isometric embedding ( unique up to the isometries of @xmath214 ) of @xmath215 in the flat euclidean space with positive mean curvature @xmath216 .
then , using formula ( [ dira - extr2 ] ) , the restriction of a parallel spinor field @xmath217 to @xmath8 gives a solution to the boundary dirac equation @xmath218 where @xmath224 denotes the space of constant spinors with unit norm .
form ( [ posquas ] ) , it is clear that @xmath225 and if @xmath226 , it is straightforward to see , from the proof of theorem [ spacetime ] , that @xmath8 embeds in the minkowski spacetime .
we provide here a direct application of our main estimate to get a rigidity result for the constraint equation with spherical boundary .
more precisely , we will say that a compact domain @xmath9 of an initial data set @xmath42 has a spherical boundary if the boundary of @xmath9 endowed with the induced metric is isometric to a round sphere . in the following
, we can assume without loss of generality , that the boundary sphere has radius one .
the two following domains : give examples of compact domains with spherical boundary .
it is interesting to point out that these two domains can be seen as domains in spacelike hypersurfaces of the minkowski spacetime : a totally geodesic hyperplane @xmath214 for the unit euclidean ball and a totally umbilical hyperbolic space @xmath229 with constant curvature equals to @xmath230 for the spherical caps .
moreover , it is clear that , in both cases , the mean curvature vector has constant length equal to 2 .
in fact , this comes from a more general fact : [ uniqueness ] let @xmath9 be a compact domain with spherical boundary in an initial data @xmath42 and assume that @xmath9 has no apparent horizon in its interior . if the mean curvature vector of @xmath8 is such that @xmath231 , then @xmath9 embeds as a spacelike hypersurface in the minkowski spacetime . _ proof : _ first , note that since we assume that the boundary is spherical we have @xmath232 and the associated eigenspinor is a real killing spinor . on the other hand
, we can apply theorem [ spacetime ] and the assumption on the norm of the mean curvature vector to conclude that @xmath233 and so equality occurs in ( [ eigenspacetime ] ) . from the equality case , we have that @xmath9 can be locally embedded in the minkowski spacetime . as seen in the proof of theorem [ spacetime ] , this embedding is constructed from a solution of the jang equation @xmath234 .
here we easily see that the manifold @xmath235 , with @xmath236 , is a compact flat riemannian manifold with spherical boundary and so it is isometric to the unit euclidean ball . in particular ,
@xmath9 is simply connected and thus @xmath102 can be globally embedded in the minkowski spacetime .
@xmath134 + if we assume that the extrinsic mean curvature ( that is @xmath38 ) is constant , we prove that the only domains with spherical boundary are the euclidean balls and the spherical caps . in particular , this statement provides an unified approach to previous results of miao @xcite , shi - tam @xcite , @xcite and the author @xcite for the rigidity of euclidean balls and spherical caps .
more precisely , we prove : [ uniqueness1 ] under the assumptions of theorem [ uniqueness ] and if the extrinsic mean curvature is constant , then @xmath9 is isometric to an euclidean ball ( if @xmath237 ) or to a spherical cap ( if @xmath238 ) . _
proof : _ from theorem [ uniqueness ] and since the extrinsic mean curvature is constant , the domain @xmath9 is a constant mean curvature spacelike hypersurface with spherical boundary in the minkowski spacetime . by a result of alias and pastor @xcite , @xmath9 is an euclidean ball or a spherical cap . @xmath134
+ using a result of alias and malacarne @xcite similar to @xcite , it is straightforward to see that the statement of corollary [ uniqueness1 ] is also true for initial data sets with constant higher order mean curvature . in particular , it holds if the scalar curvature of @xmath9 is constant . in this section
, we prove that inequality ( [ eigenspacetime ] ) holds in the case of surfaces embedded in the minkowski spacetime with spacelike mean curvature vector without any assumption on the existence of apparent horizon .
moreover , since this estimate only involves geometrical data of the surface @xmath8 and does not depend on the bounding domain @xmath9 , we focus especially on the geometrical properties of the surfaces for which ( [ eigenspacetime ] ) is an equality .
more precisely , following @xcite , we prove [ mink ] let @xmath8 be a closed , connected surface with spacelike mean curvature vector field which bounds a compact domain of a spacelike hypersurface in the minkowski spacetime @xmath213 , then inequality ( [ eigenspacetime ] ) holds .
moreover , equality occurs if and only if @xmath8 is a round sphere in @xmath213 . before giving the proof of this result , we fix some notations and recall some basic facts on spinors in the minkowski spacetime . in the following ,
the triplet @xmath239 represents the dirac bundle over @xmath240 made of the complex spinor bundle @xmath241 , the spin levi - civita connection @xmath242 acting on @xmath243 and the clifford multiplication @xmath244 acting on @xmath245 .
if @xmath9 denotes a @xmath109-dimensional spacelike hypersurface of @xmath240 with second fundamental form @xmath31 , we define the _ hypersurface _ dirac bundle @xmath246 on which the following spinorial gauss formula holds : @xmath247 since @xmath240 is four - dimensional .
to simplify our notations , we identify these two bundles and only work with the extrinsic one . similarly , the surface @xmath8 is endowed with two dirac bundles , an intrinsic one @xmath252 and an extrinsic one @xmath253 . in this situation ,
since @xmath8 is @xmath6-dimensional , we have the following identification @xmath254 moreover , we also have the following identification for the clifford multiplications : @xmath255 for all @xmath95 and @xmath256 . combining the spin gauss formula of the immersions @xmath257 and @xmath258
give that , for all @xmath95 and @xmath259 , we have @xmath260 using this identity , it is straightforward to check that the dirac operator @xmath104 satisfies @xmath261 where @xmath262 is a local @xmath28-orthonormal frame of @xmath263 .
_ proof of theorem [ mink ] : _ from lemma @xmath264 in @xcite , @xmath8 spans a compact , smoothly immersed , maximal hypersurface @xmath9 in @xmath213 .
denote by @xmath31 the second fundamental form of @xmath9 in @xmath213 ( which satisfies @xmath237 since @xmath9 is maximal ) , the gauss formula gives @xmath265 .
here @xmath3 is the scalar curvature of @xmath9 equipped with the metric induced by the minkowski spacetime .
moreover , the mean curvature vector of @xmath8 satisfies @xmath266 from lemma @xmath267 in @xcite and since @xmath8 has a spacelike mean curvature vector , we have @xmath268 . now since @xmath269 and @xmath268
, we can apply the hijazi - montiel - zhang estimate @xcite ( which is inequality ( [ ne1 ] ) with @xmath71 ) to get @xmath270 because of ( [ mcv ] ) .
assume now that equality is achieved .
from the equality case of ( [ ne1 ] ) , we deduce that @xmath9 has a parallel spinor ( in particular it is flat since ricci flat and @xmath109-dimensional ) and that the mean curvature @xmath15 is constant .
moreover since @xmath271 , the domain @xmath9 is totally geodesic in @xmath213 . in particular , it is straightforward to see that the value of the mean curvature @xmath15 is @xmath272 .
we also know that the eigenspace associated with the eigenvalue @xmath13 is given by the restriction to @xmath8 of the space of parallel spinor fields on @xmath9 .
let @xmath273 such a parallel spinor and define for @xmath274 : @xmath275 since @xmath280 has no zeros , we immediately get @xmath281 for all @xmath95 . from the gauss formula of the immersion @xmath282 , we observe that the gauss curvature of @xmath8 is constant equal to @xmath283 and so @xmath8 is isometric to a round sphere .
if now we assume that @xmath8 is a round sphere with radius @xmath284 , it embeds in a totally geodesic hyperplane @xmath214 such that @xmath285 . on the other hand ,
the first eigenvalue of the dirac operator is @xmath286 and so equality is achieved in ( [ eigenspacetime ] ) .
@xmath134 + for our geometric applications , we prove a new upper bound for this first eigenvalue in terms of geometrical data of @xmath8 stated as follow : [ upper ] let @xmath8 be a compact and connected spacelike surface isometrically immersed in the minkowski spacetime with spacelike mean curvature vector .
assume that @xmath8 is contained in a spacelike hypersurface @xmath30 of @xmath213 such that the trace of its second fundamental form @xmath31 along @xmath8 is constant .
then the first eigenvalue of the dirac operator @xmath64 satisfies @xmath287 it is straightforward to check that this result holds for codimension two compact submanifolds isometrically immersed in a spacelike hypersurface of the minkowski space @xmath289 .
in this situation , our estimate generalizes a previous result of ginoux @xcite for hypersurfaces in the hyperbolic space . for @xmath256 .
by a direct calculation , we have : @xmath291 where @xmath43 is the unit inner vector field normal to @xmath8 in @xmath30 .
moreover , since @xmath292 we compute @xmath293 and the rayleigh quotient for the first eigenvalue @xmath294 writes @xmath295 for all @xmath256 since @xmath296 is purely imaginary .
now , since @xmath30 is an immersed spacelike hypersurface of @xmath213 , we consider the restriction to @xmath30 of a constant spinor field @xmath297 which satisfies by the spinorial gauss formula ( [ spingaussformule ] ) @xmath298 for all @xmath299 and where @xmath29 denotes the unit timelike future oriented vector field normal to @xmath30 . from the identification ( [ idclif ] ) of the clifford multiplications @xmath86 and @xmath244 we have : @xmath300 for all @xmath299 .
now we compute @xmath301 using @xmath280 as a test - spinor in ( [ rayl ] ) gives immediately the estimate . if equality is achieved , it is straightforward to see that the norm of the mean curvature vector is the first eigenvalue of @xmath304 .
@xmath134 + as an application of our previous estimates , we obtain the following uniqueness result : [ genalex ] let @xmath8 be a closed , connected , spacelike surface in the minkowski spacetime . if @xmath8 bounds a compact spacelike domain @xmath9 such that @xmath15 and @xmath305 are constant , then @xmath8 is a round sphere . on the other hand , since @xmath15 and @xmath305 are constant , the norm of the mean curvature vector is also constant so that we have equality in the previous inequalities .
the rigidity statement of theorem [ mink ] allows to conclude .
@xmath134 + this result gives a unified proof of the euclidean and hyperbolic version of the alexandrov theorem for surfaces ( see @xcite and @xcite for example ) .
namely , we have : _ proof : _ it is enough to note that the euclidean space embeds as a totally geodesic spacelike hyperplane in the minkowski spacetime and in the hyperbolic space as a totally umbilical with constant mean curvature spacelike hypersurface .
it is then clear that in both cases , the surface @xmath8 satisfies the assumptions of corollary [ genalex ] so that we get immediately the result . | we establish a sharp extrinsic lower bound for the first eigenvalue of the dirac operator of an untrapped surface in initial data sets without apparent horizon in terms of the norm of its mean curvature vector .
the equality case leads to rigidity results for the constraint equations with spherical boundary as well as uniqueness results for constant mean curvature surfaces in minkowski space . |
many attempts have been made to determine the amount and distribution of dark matter in clusters , since zwicky ( 1933 , 1937 ) and smith ( 1936 ) concluded that the mass implied by the sum of the luminosities of the galaxies falls short of the total mass by as much as a factor of 10 . in recent years
, cosmological simulations have shown that dark matter halos have a universal density profile ( navarro , frenk , & white 1996 , 1997 ; nfw hereafter ) , but its precise form is still debated ( moore et al . 1999 , m99 hereafter ) , while the universality of the mass density profile has also been questioned ( jing & suto 2000 ; thomas et al .
2001 ; ricotti 2002 ) .
the issue is critical , since knowledge of the total mass ( visible and dark , baryonic and non - baryonic ) of clusters , and of its distribution ( also in relation to the light distribution ) , gives important clues about the formation process of the clusters ( e.g. crone , evrard , & richstone 1994 ; jing et al .
1995 ) , and of the galaxies in them ( e.g. mamon 2000 ) , as well as on the nature of dark matter ( see , e.g. , natarajan et al .
2002a ) .
the dark matter can be ` weighed ' and ` imaged ' in three ways .
first , the gravitational lensing of distant objects yields an estimate of the projected mass distribution in the cluster ( see e.g. tyson , wenk , & valdes 1990 ; squires et al .
the mass density profile can subsequently be derived assuming the geometry of the cluster .
lensing mostly works for clusters at intermediate and large distances since the lensing equation is unfavourable for nearby clusters , and only few nearby cluster lenses are known ( see e.g. campusano , kneib , & hardy 1998 ) . some results from gravitational lensing observations are consistent with the nfw profile ( e.g. athreya et al .
2002 ; clowe & schneider 2001 ) , while others favour flatter mass density profiles , such as the isothermal sphere ( see fischer & tyson 1997 ; sand , treu , & ellis 2002 ; shapiro & iliev 2000 ; taylor et al .
1998 ) .
a second method uses the distribution of the hot , x - ray emitting , gas in the cluster potential , and the radial variation of the temperature of the gas ( see e.g. hughes 1989 ; ettori , de grandi , & molendi 2002 ) .
recently , temperature maps have become available for a sizable number of clusters ( see e.g. markevitch et al .
1998 ; irwin & bregman 2000 ; de grandi & molendi 2002 ) . however , the uncertainties in the x - ray temperature profiles are still substantial ( see e.g. irwin , bregman , & evrard 1999 ; kikuchi et al .
1999 ) , with corresponding uncertainties in the total mass and the mass profile .
moreover , x - ray observations in general sample only the inner cluster regions , even with the new class of x - ray satellites ( see , e.g. , pratt & arnaud 2002 ) .
cooling can also play a rle , although it is probably not dominant ( suginohara & ostriker 1998 ) .
most recent results from x - ray observations indicate consistency with the nfw or the m99 profile ( e.g. allen , ettori , & fabian 2001 ; allen , schmidt , & fabian 2002 ; markevitch et al .
1999 ; pratt & arnaud 2002 ; tamura et al .
2000 ) , but a flatter mass density profile like that of the isothermal sphere seems to be preferred in some clusters ( arieli & rephaeli 2003 ; ettori et al . 2002 ) .
generally , the distribution of the the x - ray emitting gas is found to be flatter than that of the total cluster mass ( e.g. allen et al .
2001 , 2002 ; markevitch & vikhlinin 1997 ; nevalainen , markevitch , & forman 1999 , pratt & arnaud 2002 ) . the third , and most traditional way to probe the dark matter content and its distribution in clusters is through the kinematics and spatial distribution of tracer particles moving in the cluster potential .
the virial theorem applied to the galaxy population gives an estimate of the total mass but , since the data in general comprise only the central part of a cluster , projection effects and possible anisotropy of the velocity distribution must be taken into account for an accurate estimate ( see e.g. heisler , tremaine , & bahcall 1985 ; the & white 1986 ) .
in addition , the virial theorem assumes that mass follows light and if this assuption is not justified that can have a large effect on a cluster mass estimate ( see , e.g. , merritt 1987 ) .
recently , diaferio & geller ( 1997 ; see also diaferio 1999 ) used the caustics in the plane of line - of - sight velocities vs. clustercentric distances to estimate the amplitude of the velocity field in the infall region .
this yields an estimate of the escape velocity at several distances , and thus of the mass distribution .
this method has the advantage of being free of the assumption of dynamical equilibrium , and hence can be ( and has been ) used to constrain the cluster mass distribution well beyond the virialization region ( geller , diaferio , & kurtz 1999 ; reisenegger et al .
2000 ; rines et al . 2000 , 2001 , 2002
; biviano & girardi 2003 ) . however , systematic uncertainties limit the accuracy in the mass determination to @xmath5% ( diaferio 1999 ) .
results obtained with this method indicate consistency with the nfw or m99 profiles , or strongly constrain or rule out the isothermal sphere model . the most detailed form of the kinematical weighing uses the distribution and kinematics of the galaxies to estimate the mass distribution , through application of the full jeans equation of stellar dynamics ( see , e.g. , binney & tremaine 1987 , bt hereafter ) .
this method requires knowledge of the orbital characteristics of the galaxies . a first analysis along these lines
was made for the coma cluster by kent & gunn ( 1982 ) .
merritt ( 1987 ) used the same data to estimate the orbital anisotropy of the galaxies for various assumptions about the dependence of the mass - to - light ratio @xmath6 on radius .
the shape of the mass profile of a cluster would follow directly from the projected luminosity density profile of the galaxies if the mass - to - light ratio @xmath7 were constant .
evidence in favour of a constant @xmath7 comes from the analyses of carlberg , yee , & ellingson ( 1997a ) , van der marel et al .
( 2000 ) , and rines et al .
carlberg et al . (
2001 ) instead find an increasing @xmath7-profile in groups from the cnoc2 survey , while both rines et al .
( 2000 ) and biviano & girardi ( 2003 ) argue for a decreasing @xmath7-profile .
the shape of the cluster @xmath7-profile seems to depend on which class of cluster galaxies is used to measure the light profile .
this is hardly surprising , since the various types of galaxies are known to have different projected distributions .
these differences are the result of the morphology - density relation , first described by oemler ( 1974 ) and melnick & sargent ( 1977 ) , and described more fully by dressler ( 1980 ) and , lately , thomas & katgert ( 2003 , paper x of this series ) . in clusters
, this relation produces differences in the radial distribution of the various galaxy classes , and those are related to differences in the kinematics .
evidence for the relation between spatial distribution and kinematics for different cluster galaxy populations was e.g. found in the coma cluster ( colless & dunn 1996 ) , in cnoc clusters at @xmath8 by carlberg et al .
( 1997b ) , in clusters observed in the eso nearby abell cluster survey ( enacs , hereafter ; de theije & katgert 1999 , paper vi ; biviano et al .
2002 , paper xi ) , as well as in other clusters ( see , e.g. , mohr et al . 1996 ; adami , biviano , & mazure 1998a ) .
galaxies with emission lines ( elg ) provide an extreme example of the effect .
the elg are less centrally concentrated and have a higher r.m.s line - of - sight velocity than the galaxies without emission lines .
this was clearly demonstrated by biviano et al .
( 1997 , paper iii ) , using 75 enacs clusters . to study the mass profile in detail
, one must combine the data for many galaxy systems , as was done by e.g. , biviano et al .
( 1992 ) , carlberg et al .
( 1997a , 1997b , 1997c , 2001 ) , van der marel et al .
( 2000 ) , biviano & girardi ( 2003 ) , and in papers iii , vi , vii ( adami et al .
1998b ) , ix ( thomas , hartendorp , & katgert 2003 ) , x , and xi of this series . for an ` ensemble ' cluster , built from 14 clusters with redshifts between about 0.2 and 0.5 , with a total of 1150 cnoc redshifts , carlberg et al .
( 1997c ) found that the number density profile is consistent with the nfw mass density profile .
this result was confirmed with a more detailed analysis by van der marel et al .
( 2000 ) . using the enacs dataset , in paper vii we found that the number density profile of a composite cluster of 29 nearby aco clusters with smooth projected galaxy distributions did not show the central cusp of the nfw mass profile , in particular when the brightest galaxies are removed from the sample .
this could mean that @xmath7 increases towards the center . recently , biviano & girardi ( 2003 ) analysed an ensemble cluster , built from 43 nearby clusters observed in the 2df galaxy redshift survey ( colless et al .
2001 ) . from that sample , which has only three clusters ( a957 , a978 and a2734 ) in common with the present sample
, they concluded that both cuspy profiles of the nfw and m99 form , and profiles with a core are acceptable , as long as the core radius is sufficiently small .
the above results indicate the importance of a study of the mass profile and of the radial dependence of @xmath7 . in this paper
we present such a study , based on a sample of rich nearby clusters observed in the enacs .
this paper is organised as follows . in [ s - data ] we summarize the data . in
[ s - massearly ] we justify our choice to use the early - type galaxies that are not in substructures as tracers of the potential . in [ s - direct ]
we present the number - density and velocity - dispersion profiles of the early - type galaxies and we describe how we obtained a non - parametric estimate of the cluster mass profile via direct solution of the jeans equation.in [ s - model ] we compare our result with models , and we derive the best - fit models from a comparison of the observed and predicted velocity - dispersion profiles . in [ s - mlprof ]
we derive the radial dependence of the @xmath7-ratio , for all galaxies together and for the early - type galaxies . in [ s - disc ] we discuss our results , which are summarized in [ s - summ ] , where we also give our conclusions . in appendix
[ a - interp ] we describe our method of interloper rejection . in appendix
[ a - prof ] we detail the methods by which we determined the number - density , luminosity - density , and velocity - dispersion profiles . in appendix
[ a - assum ] we review and discuss the basic assumptions made in the determination of the mass profile .
our determination of the mass profile of rich clusters is based on data obtained in the context of the enacs . the multi - object fiber spectroscopy with the 3.6-m telescope at la silla
is described in katgert et al .
( 1996 , 1998 , papers i and v of this series , respectively ) . in those papers ,
the photometry of the 5634 galaxies in 107 rich , nearby ( @xmath9 ) abell clusters is also discussed .
after the spectroscopic survey was done , ccd images were obtained with the dutch 92-cm telescope at la silla for 2295 enacs galaxies .
these have yielded morphological types ( thomas 2003 , paper viii ) , which were used to refine and recalibrate the galaxy classification based on the enacs spectra , as carried out previously in paper vi .
the ccd images also yielded structural parameters , through a decomposition of the brightness profiles into bulge and disk contributions ( paper ix ) .
the enacs morphological types were supplemented with morphological types from the literature , and subsequently combined with the spectral types into a single classification scheme .
this has provided galaxy types for 4884 enacs galaxies , of which 56% are morphological , 35% are spectroscopic , 6% are a combination of morphological an spectral types , and the remaining 3% were special in that they had an early morphological type ( e or s0 ) but showed emission lines in the spectrum .
these galaxy types were used to study the morphology - radius and morphology - density relations ( paper x ) .
they also form the basis of the study of morphology and luminosity segregation which uses velocities as well as positions ( paper xi ) . in combining the data for an ensemble of many clusters ,
all projected clustercentric distances were expressed in terms of @xmath3 , as derived from the global velocity dispersion ( see e.g. carlberg et al .
this ensures that we avoid , as much as possible , mixing inner virialized cluster regions with external non - virialized cluster regions .
similarly , all galaxy velocities , relative to the mean velocity of their parent cluster , were expressed in terms of the global velocity dispersion of their parent cluster , @xmath10 .
the ensemble cluster effectively represents each of our clusters if these form a homologous set , and if we adopted the correct scaling . an indication that clusters form a homologous set comes from the existence of a fundamental plane that relates some of the cluster global properties ( schaeffer et al . 1993 ; adami et al . 1998c , paper iv of this series ) . as shown by beisbart , valdarnini , & buchert ( 2001 ) clusters with substructure tend to deviate from the fundamental plane , thereby violating homology .
the exclusion of all clusters with even a minor amount of substructure would have greatly reduced the size of the ensemble cluster . instead , by eliminating the galaxies that in their respective clusters are in substructures , we have tried to reduce the effects of substructure in our analysis as much as possible . in paper xi
we combined the data for 59 clusters with @xmath11 , each with at least 20 member galaxies with enacs redshifts , and with galaxy types for at least 80% of the members ( see table a.1 in paper xi ) .
the ensemble cluster contains 3056 member galaxies , for 2948 ( or 96% ) of which a galaxy type is known .
the rejection of interlopers was based on the method of den hartog & katgert ( 1996 ) , which is summarized in appendix [ a - interp ] . in this ensemble cluster
it was found that galaxies in substructure have different phase - space distributions from those outside substructure , and that this is true for all galaxy types .
cccc e@xmath12 & early & s@xmath13 & s@xmath14 + @xmath0 & + 34 & 1129 & 177 & 328 + [ t - compos ] the analysis of the mass profile can not be based on the sample of galaxies within substructure because the members of substructure are orbiting together , so that their kinematics does not only probe the large - scale properties of the cluster potential . for the galaxies outside substructure ,
it was found that there are 4 galaxy classes that must be distinguished because they have different phase - space distributions .
these 4 classes are : ( i ) the brightest ellipticals ( with @xmath0 ) , which we refer to as e@xmath12 in the following , ( ii ) the other ellipticals together with the s0 galaxies ( referred to as early - type galaxies , in the following ) , ( iii ) the early spirals ( sa sb ) , which we refer to as s@xmath13 in the following , and ( iv ) the late spirals and irregulars ( sbc ir ) together with the elg ( except those with early morphology ) , globally referred to as s@xmath14 in what follows . in table [ t - compos ] we show the number of cluster members outside substructure in the 59 clusters , in each of the 4 galaxy classes defined above . as explained in appendix [ a - prof ] , in building the number density profiles we need to correct for sampling incompleteness . in order to keep the correction factor small
, galaxies located in poorly - sampled regions are not used in the present analysis , and are not counted in table [ t - compos ] .
since we want to derive the mass profile of the ensemble cluster by application of the jeans equation of stellar dynamics , we must assess the suitability of each of the 4 galaxy classes to serve as tracers of the potential . to begin with
, suitable tracers should be in equilibrium with the cluster potential .
however , application of the jeans equation also requires that the orbital characteristics are known or , in other words : that we have information on the ( an-)isotropy of the velocity distribution .
for that reason , and in view of the results in papers iii and vi , it is very unlikely that the s@xmath14 can be used , since the analyses in these papers suggest that they may be on first - approach orbits towards the central regions .
this could still mean that they are in equilibrium with the potential , but as we have no _ a priori _ knowledge of their orbits , except that they are unlikely to be isotropic , these galaxies do not qualify as good tracers .
the e@xmath12 could , for all that we know , have an isotropic velocity distribution .
however , it is unlikely that , as a class , they satisfy the basic assumption underlying the jeans equation , namely that it describes a collisionless particle fluid .
the usual interpretation that the brightest ellipticals have been directed to the central regions through dynamical friction after which they have grown at the expense of other galaxies might make them unfit for estimates of the mass profile . as to the choice between the two remaining classes
, the relative statistical weights clearly point to the use of the early - type galaxies as tracers of the mass profile .
however , in addition to this pragmatic argument there are other , more fundamental reasons for this choice .
the most important one is that the stellar population in most ellipticals is quite old , and evidence is mounting that most of the ellipticals formed before they entered the cluster .
this should have allowed them to settle in the potential and become good tracers of it . as was shown in paper xi ,
the @xmath15-distribution of the s0 galaxies is very similar to that of the ellipticals . at first
, this may appear somewhat surprising .
although the stellar population of many s0 galaxies is probably as old as that of the ellipticals , it is generally believed that a significant fraction of s0 s has formed rather recently through the transformation of early spirals by impulsive encounters .
apparently , the present rate of formation of s0 s is sufficiently low that the bulk of the transformation of early spirals has taken place sufficiently long ago , so that the phase space distribution of the s0 s has relaxed to that of the older population of e s .
use of the early - type galaxies requires that we know , or can make an educated guess about , the anisotropy of their velocity distribution .
we will assume that their orbits are isotropic , and this assumption can be checked to some extent because the distribution of relative radial velocities depends on the anisotropy of the 3-d velocity distribution , as was shown by merritt ( 1987 ) and van der marel et al .
( 2000 ) . by calculating the even - order gauss - hermite coefficients for the velocity distribution of the cnoc1 survey , van der marel et al .
( 2000 ) showed that they could constrain the range of allowed values of the velocity anisotropy , from a comparison with dynamical models .
the velocity distribution of the early - type galaxies in our ensemble cluster is shown in fig .
[ f - gh ] , with the best - fitting gaussian and the best - fitting gauss - hermite polynomial superposed .
it appears that the deviations from a gaussian are very small .
this is confirmed by the values of @xmath16 and @xmath17 , which are @xmath18 and @xmath19 .
although we refrain from constructing plausible distribution functions , the projected number density and therefore also the 3-d number density is sufficiently close to that used by van der marel et al .
( 2000 ) that we can conclude from their fig .
8 that our assumption @xmath20 for the e+s0 class is very plausible ( as a matter of fact we conclude that @xmath21 , or , equivalently , @xmath22 ) .
application of the jeans equation for the determination of the cluster mass profile requires two observables : the projected number - density profile @xmath23 , and the velocity - dispersion profile @xmath24 of the early - type galaxies .
here we summarize the steps involved in the determination of these profiles and their inversion to the 3-d space .
details can be found in appendix [ a - prof ] . for the determination of both number - density and velocity - dispersion profiles we use the lowess technique ( gebhardt et al .
the number - density profile , @xmath23 , is corrected for sampling incompleteness
. the confidence levels ( c.l .
hereafter ) on @xmath23 are measured via a bootstrap procedure . the projected number - density profile @xmath23 is shown in the upper panel fig .
[ f - dprof ] .
the projected number density @xmath23 must be deprojected to yield the 3-d number density @xmath25 .
this deprojection is obtained via a straightforward integration of the abel integral , which involves no assumptions , apart from the behaviour of @xmath23 towards large radii .
we checked that the deprojected profile is essentially independent from how we extrapolate @xmath23 to large radii ( see appendix [ a - prof ] ) .
we show the deprojected profile @xmath25 in the botttom panel of fig .
[ f - dprof ] . in the upper panel of fig .
[ f - vdp ] we show the projected velocity - dispersion profile , determined with the lowess technique , together with a binned version , where the value of the velocity dispersion in each radial bin is computed using the robust bi - weight estimator ( see beers , flynn , & gebhardt 1990 ) . with the estimates of @xmath23 , @xmath25 and @xmath24 we now proceed to obtain the cluster mass profile @xmath26 .
the procedure is well - known ( see , e.g. , bt ) .
we make the usual assumptions that the cluster is in steady - state , and that net rotation is negligible . with these assumptions , and in the case of spherically symmetry ( see also appendix [ a - assum ] ) , the procedure is as follows .
first , one must deproject @xmath24 to obtain @xmath27 . in other words
, one should solve the integral relation : @xmath28 where , as before : @xmath29 and @xmath30 , @xmath31 are the mean squared components of the radial and tangential velocity .
for the special case @xmath32 the solution is : @xmath33}{dr } \frac{dr}{\sqrt{r^2-r^2}}. \label{e - abel3d}\ ] ] the @xmath30 profile and c.l . resulting from @xmath23 , @xmath24 , and the c.l . on @xmath24 ( see appendix [ a - prof ] for details ) are shown in the lower panel of fig .
[ f - vdp ] .
the mass profile then follows directly from the jeans equation ( see , e.g. , bt ) : @xmath34 where @xmath35 was assumed to be 0 ( see [ s - massearly ] ) .
the result is shown in the upper panel of fig .
[ f - massprof ] . since in our ensemble cluster galaxy velocities
are normalised by their cluster global velocity dispersion , @xmath10 , and galaxy clustercentric distances are scaled by their cluster virial radius , @xmath3 , the mass is also expressed in normalised units , and the median value of @xmath3 is 1.2 @xmath36 one mass unit corresponds to @xmath37 . also shown are the mass profiles of four popular models : viz .
those of nfw , m99 , the softened isothermal sphere ( sis , hereafter ; see , e.g. , geller et al .
1999 ) , and that of burkert ( 1995 ) .
they are all one - parameter models in which the linear scale ( in units of @xmath3 ) is variable .
the analytic expressions are : @xmath38 @xmath39}\ ] ] @xmath40 @xmath41}}\ ] ] our mass profile @xmath42 and the four model profiles can not be compared directly , because the c.l . of our mass profile
are only approximate ( as discussed in appendix [ a - prof ] ) . for a proper statistical comparison
we need to project our mass profile in the space of observables , and this is described in the next section . yet , for the purpose of illustration , we show in the lower panel of fig .
[ f - massprof ] the mass density profile as derived by differentiation of the observed mass profile @xmath42 , together with the four model density profiles . in both panels of fig .
[ f - massprof ] we have used the best - fit values @xmath43 , @xmath44 , @xmath45 , and @xmath46 ( see below ) . as a test of our assumption about the unsuitability of the late - type galaxies as tracers of the potential ( see [ s - massearly ] ) ,
we have also applied the procedure , described here for the early - type galaxies , to the early and late spirals ( again , assuming isotropy ) .
the s@xmath13 mass profile is not very different from that in fig .
[ f - massprof ] , but the s@xmath14 mass profile is considerably steeper , which supports our assumption . in a forthcoming paper ( biviano & katgert , 2004 ) , we discuss in detail the dynamical equilibrium ( including the orbital anisotropies ) of the e@xmath12 , s@xmath13and s@xmath14 classes , in the potential derived here from the data of the early - type galaxies .
in order to compare our observed mass profile with models from the literature , we work in the domain of observables , viz . in @xmath24 .
this is possible because one can solve for @xmath47 using the observed @xmath25 , an assumed mass profile @xmath1 , and an assumed @xmath48 , as follows : @xmath49 d \xi \label{e - vdm}\ ] ] this solution , given by van der marel ( 1994 ) , is a special case of a more general solution of the jeans equation which was developed by bacon , simien , & monnet ( 1983 ) for building dynamical models of elliptical galaxies . for @xmath50
the above equation reduces to : @xmath51 given a model mass profile , @xmath1 , and an observed number - density profile , @xmath25 , it is thus possible to compute model projected velocity - dispersion profiles , through eq .
[ e - vdm0 ] and the usual abel relation .
the comparison between the model and the observed velocity - dispersion profile yields a value for @xmath52 by which we measure the acceptability of the assumed mass - profile model .
a straightforward determination of the @xmath52 value requires the observed profile to have independent data - points . in this comparison
we therefore use a binned @xmath24-profile , rather than the lowess profile .
only the data points within the virial radius @xmath3 are considered for the fit , since galaxies at larger radii might not yet have relaxed to dynamical equilibrium .
the uncertainty in the observed @xmath53 , determined from the bootstrap resamplings of @xmath23 ( see appendix [ a - dprof ] ) is taken into account in the @xmath52-analysis . in practice , for each bootstrap resampling of @xmath23 we have a corresponding @xmath25 , and hence , a different model velocity dispersion profile , obtained via eq .
[ e - vdm0 ] .
this allows us to compute an r.m.s . for each point of the model velocity dispersion profile .
the model r.m.s .
is then added in quadrature to the uncertainty of the observed @xmath24 to give the full uncertainty to be used in the @xmath52-analysis .
we checked that our results are robust w.r.t .
different choices of both the binning radii for @xmath24 , and the way in which @xmath23 is extrapolated to large radii to yield @xmath25 through the abel inversion ( see appendix [ a - prof ] ) . different choices of the @xmath24 binning and of the @xmath23 extrapolation affect the best - fit values of the mass model parameters by @xmath54% and @xmath55% , respectively . we consider two cuspy model mass profiles ( nfw and m99 ) , and two profiles with a ` core ' ( sis and that of burkert 1995 ) see [ s - massearly ] .
all four model mass profiles provide acceptable fits to the observed @xmath24 , although the models of nfw and burkert provide marginally better fits than the m99 and , in particular , the sis models . more specifically , using the @xmath52 statistics
, we estimate the rejection probabilities of the best - fit nfw , m99 , sis and burkert models to be 19% , 28% , 72% and 11% , respectively .
note that the @xmath52 values were calculated for the points with @xmath56 , because inclusion of data beyond the virial radius does not seem very sensible . as a mattter of fact ,
if the outmost point is included all 4 models provide fits of similar quality . the 68% c.l .
range for the nfw - profile scale parameter is @xmath57 , with a best - fit value of @xmath58 . in terms of the concentration parameter ,
the best - fit value is @xmath59 .
the 68% c.l . range for the m99-profile scale parameter is @xmath60 with a best - fit value of @xmath61 . on the other hand , for no value of the scale parameter @xmath62
is the sis model acceptable at the 68% c.l . or better .
the best - fit is obtained for @xmath63 and values @xmath64 are rejected at @xmath65% c.l .
the 68% c.l .
range for the burkert scale parameter @xmath66 is @xmath67 with a best - fit value of @xmath68 .
values @xmath69 are rejected at @xmath65% c.l .
the allowed range for the burkert scale parameter seems larger than for the sis scale parameter .
note , however , that the two scale parameters do not have the same meaning .
if we define the core radius as the clustercentric distance where the density falls below half its central value , @xmath70 , this corresponds to @xmath62 in the sis model , and to @xmath71 in the burkert model .
we conclude that the core models ( and in particular the sis - model ) that fit our data have such small core - radii that they very much resemble the two core - less models . in fig .
[ f - vdpmodel ] we show the observed ( binned ) @xmath24 of the early - type class , together with the velocity dispersion profiles predicted from the best - fit nfw , m99 , sis and burkert @xmath1 models . for the sake of clarity , we do not show the uncertainties in the model velocity dispersion profiles related to the uncertainties in @xmath25 .
these are however much smaller than the uncertainties of the observed velocity dispersion profile . from fig .
[ f - vdpmodel ] it can be seen that the best - fit nfw and m99 models predict almost indistinguishable velocity dispersion profiles .
on the other hand , fitting the inner part of @xmath24 requires such a small core radius for the sis model , that the model velocity dispersion profile flattens already at @xmath72 , while the observed profile continues to drop .
formally , the best - fit burkert model also provides the best fit to the observations , as it reproduces the broad maximum in @xmath24 better than do the other three models
. however , we are somewhat suspicious of the very strong decrease of @xmath24 below @xmath73 , although it is only @xmath74 lower than the lowess estimate of @xmath24 .
as mentioned in the introduction , there is not yet a clear , unambiguous result for the dependence of @xmath7 on radius in the literature , and therefore we have also used our data to derive @xmath75 .
the luminosity - density profile @xmath76 was determined from the projected luminosity density profile of the galaxies with the same method used to deproject @xmath23 , except that galaxies are now weighed by their luminosities .
note that the luminosity density profile is constructed for all galaxies , both in and outside substructures .
as discussed in appendix [ a - prof ] , @xmath77 only accounts for the luminosities of the galaxies in our sample , as we did not try to recover the luminosity of unseen cluster galaxies ( e.g. , by fitting a luminosity function to the observed luminosity distribution , and extrapolating beyond the completeness limit ) .
in fact , we are not interested in estimating the absolute values of @xmath7 , but only in its radial behaviour , and there is no way we could estimate @xmath78 for the total galaxy population , without actually measuring positions and luminosities of all members . the full magnitude range of the galaxies on which @xmath77 is based is @xmath79 , with an average @xmath80 , and an r.m.s . of @xmath81 mag .
the average absolute magnitude is about one magnitude fainter than the knee of the cluster galaxy luminosity function , @xmath82 ( e.g. yagi et al .
note that our magnitudes are k - corrected , but that we did not apply any correction for evolutionary effects . for redshifts @xmath83 ,
those are very small in @xmath84-band , and the _ differential _ evolutionary correction for early- and late - type galaxies is estimated to be less than 5% in luminosity . in the upper panel of fig .
[ f - lnfw ] we show the ratio of the total mass - density profile @xmath85 of the best fitting nfw and burkert models , and @xmath77 calculated for all galaxies , i.e. the differential mass - to - light ratio as a function of @xmath86 .
the shaded region shows the 68% uncertainties in the profile ratio , as calculated from the union of the 68% c.l .
ranges of the best - fitting nfw and burkert models .
note that the relative uncertainties in @xmath76 are much smaller than the allowed ranges of @xmath85 .
the @xmath87-ratio is consistent with a constant value in the radial range @xmath88 , even though there is an apparent decrease from @xmath89 to @xmath90 .
however , within @xmath91 the @xmath92-ratio is significantly lower than at larger radii .
the reason for that could obviously be that the e@xmath12 which are much more centrally concentrated than the other galaxy classes , contribute relatively little mass for their luminosity . to check this explanation
, we also determined the @xmath7-ratio for all galaxies without the e@xmath12 ( i.e. @xmath93 .
the result is shown in the middle panel of fig .
[ f - lnfw ] .
as expected , the strong decrease of the @xmath7-ratio within @xmath89 has largely disappeared .
however , the apparent decrease of @xmath7 from @xmath89 to @xmath90 is now stronger . to enable comparison with data on high - redshift clusters , where selection of cluster members is often done using the red - galaxy sequence in the colour - magnitude diagram
, we also determined the @xmath7-ratio for the early - type galaxies only ( again excluding the e@xmath12 ) ; the result is shown in the lower panel of fig .
[ f - lnfw ] .
the apparent decrease of @xmath7 from @xmath89 to @xmath94 has disappeared , and the @xmath6-ratio for the early - type galaxies only , is consistent with a constant value . in other words : the luminosity of the early - type galaxies traces the total mass very well .
therefore , we conclude that to first order , the @xmath75-profile can be understood as a flat profile due to the early - type galaxies , which is modified by the extra luminosity of the e@xmath12 within @xmath95 , and by the luminosity of the spirals beyond @xmath96 .
using the enacs data - set complemented with recently obtained morphological data , we determined the mass and mass - to - light profile for an ensemble cluster built from 59 nearby clusters . following the results of the analysis of paper xi ,
the sample of cluster galaxies was divided into four classes , viz .
the brightest ellipticals ( e@xmath12 ) , the early - type galaxies ( the other ellipticals and the s0 galaxies ) , and the early and late spirals .
we used the early - type class to obtain a non - parametric estimate of the mass profile through the application of the jeans equation ( see [ s - direct ] ) , and we fitted mass - profile models ( see [ s - model ] ) . in these analyses , we assumed isotropic orbits for the early - type galaxies , which is supported by a gauss - hermite polinomial decomposition of their velocity distribution .
both ` cuspy ' models , like the nfw and m99 mass models , and models with a core , like the sis and that proposed by burkert ( 1995 ) , provide acceptable fits to the data .
this follows from a comparison of the observed and predicted velocity - dispersion profile of the early - type galaxies .
however , the best - fit core models have such small core - radii that they are not too different from the cuspy models . in other words
, we did not find compelling evidence for the existence of a core in the matter distribution in clusters .
this result , which applies to our ensemble cluster is consistent with several previous results based on smaller cluster samples , and obtained by various methods . among the latter are those based on an analysis of the distribution and temperature of the x - ray emitting gas ( e.g. markevitch et al .
1999 ; allen et al . 2002 ; arabadjis , bautz , & garmire 2002 ; pratt & arnaud 2002 ) , those based on an analysis of the weak - lensing distortion of background galaxies ( e.g. lombardi et al . 2000 ; clowe et al .
2000 ; clowe & schneider 2001 ; athreya et al .
2002 ; hoekstra et al .
2002 ; king et al . 2002
; dahle , hannestad , & sommer - larsen 2003 ) , and those based on an analysis of the projected phase - space distribution of the galaxies ( e.g. geller et al .
1999 ; van der marel et al .
2000 ; rines et al . 2002 ; biviano & girardi 2003 ) . in some of these analyses , not only was consistency with an nfw - type mass profile established , but an upper limit to the size of a core could also be given ( e.g. , arieli & rephaeli 2003 ; dahle et al .
yet , some clusters also show a sizeable core ( shapiro & iliev 2000 ; ettori et al . 2002 ; sand et al .
2002 ) . since most of these studies are based on small numbers of clusters , it is possible ( if not likely ) that individual clusters have different types of mass profile .
if so , the result of our analysis is that _ on average _ rich clusters are characterized by cuspy profiles , and that , if clusters with cores do exist , they must be a minority .
it is noteworthy that the two previous analyses involving ensemble clusters , drawn from the cnoc and the 2dfgrs datasets respectively ( by van der marel et al .
2000 , and by biviano & girardi 2003 ) , give a similar result . as a matter of fact ,
the best - fit value of the nfw scale parameter obtained in these studies ( @xmath97 and @xmath98 ) , is fully consistent with our value of @xmath99 . our result , that _ on average _ rich clusters have cuspy mass profiles in the central region , is in agreement with predictions of hierarchical cold dark matter ( cdm , hereafter ) models .
our data do not allow us to choose between different predictions , like the nfw and m99 models .
apart from the fact that the latter differ only slightly , the differences are limited to the central region , where the ( baryonic ) mass contribution of the cd is dominant ( when present ) . proper comparison would require the analysis of the internal velocity dispersion of the cd ( kelson et al .
2002 ) . in any case
, recent results from numerical simulations suggest that the concept of a universal inner slope of the mass profile might be inappropriate , as dark halo mass profiles do not seem to converge to a single central power - law shape ( power et al .
2003 ) . the allowed range @xmath99 , that we obtain for our ensemble cluster brackets the value predicted for a rich cluster halo in a @xmath100cdm cosmology , @xmath101 ( see fig.6 in nfw )
note however , that we determine the profile of _ total _ cluster mass , and not only of the _ dark _ mass .
our result therefore includes the contribution of baryons , i.e. of galaxies and the icm .
the latter can be a substantial fraction of the total mass ( e.g. ettori , tozzi , & rosati 2003 ) .
since the icm is less centrally concentrated than the dark mass ( see , e.g. , markevitch & vikhlinin 1997 ; pratt & arnaud 2002 ) , one expects the total mass distribution to be less concentrated than the dark mass .
that might be the reason why our best - fit nfw @xmath102 value is slightly larger than the theoretical prediction ( although not significantly so ) .
on the other hand , the baryonic mass contribution of the cd in the cluster center has the opposite effect , namely of making the total mass distribution _ more _ centrally concentrated than that of the dark mass ( kelson et al .
2002 ; dahle et al .
2003 ) . however , separation of the different components of the mass profile ( although important ) is beyond the scope of this paper .
cosmological simulations predict an increase of @xmath102 with the mass of the system ( nfw ) , and therefore a decrease of @xmath102 with redshift of the system ( bullock et al .
the predicted changes are rather small ; @xmath103 increases by only @xmath104% over a decade in cluster mass , while from @xmath105 to @xmath106 it decreases by a similar amount . comparing our low - z , 68% c.l .
range of @xmath103 with the value obtained for the cnoc clusters studied by van der marel et al .
( 2000 ) , of similar mass but at higher redshifts ( @xmath107 ) , we find no redshift dependence of @xmath103 .
comparing our 68% c.l .
@xmath103-range with the value of @xmath108 for the poor clusters studied by biviano & girardi ( 2003 ) , with masses 3 times smaller than ours , and with the value of @xmath109 for the groups studied by mahdavi et al .
( 1999 ) , with masses 13 times smaller than ours , we find a trend that is qualitatively in agreement with the theoretical prediction , but which is not statistically significant .
the upper limit that we obtain for the size of a possible core provides a constraint for the cross - section of self - interacting dark matter ( sidm ) models .
sidm models have become popular in recent years as a means to explain the rotation curves of low surface brightness galaxies ( see , e.g. , firmani et al .
we translate the constraints on the scale parameters of the sis and burkert models , by using the results of the simulations of meneghetti et al .
( 2001 , see their table 1 ) , and their definition of the core radius .
our results imply an upper limit to the scattering cross - section of dark matter particles of @xmath110 @xmath111 g@xmath112 at @xmath65% c.l .
, in good agreement with the conclusions of meneghetti et al .
( 2001 ) . using the magnitudes of the enacs galaxies , together with the best - fitting ` cusp- ' and `
core- ' mass models , we have studied the radial variation of the @xmath7-ratio . when the luminosity is based on all galaxies together ,
the @xmath7-ratio increases strongly from @xmath113 to @xmath89 , decreases somewhat from @xmath89 to @xmath94 and is constant beyond @xmath94 . the strong decrease to very low values of @xmath7 within @xmath114
is found to be largely due to the e@xmath12 , i.e. the brightest ellipticals with @xmath115 , because it disappears when the latter are excluded .
similarly , the decrease of @xmath7 from @xmath89 to @xmath90 largely disappears when the early and late spirals are exluded .
thus , the total mass density is traced remarkably well by the early - type ( elliptical and s0 ) galaxies .
although the early - type galaxies are the dominant population in present - day clusters , it does not follow automatically that their luminosity should trace the total mass so well . as a matter of fact ,
the _ virial _ mass profile , estimated assuming that the mass density is proportional to the number density of the early - type galaxies ( see e.g. girardi et al .
1998 ) , generally overestimates the mass , when compared with our solution in fig .
[ f - massprof ] ( by as much as a factor of 2 , although it is reduced , but does not disappear , when a correction is made for the pressure term , see the & white 1986 ) .
a similar virial - mass bias was found by carlberg et al .
( 1997b ) for the cnoc clusters .
several previous studies also yielded a flat or nearly - flat cluster @xmath7-profile when only early - type or red galaxies were selected ( e.g. carlberg et al .
1997a ; van der marel et al .
2000 ; rines et al . 2001
; hoekstra et al . 2002 ; biviano & girardi 2003 ; ettori & lombardi 2003 )
. however , near the cluster center , where the inclusion of the very bright galaxies in the sample lowers the cluster @xmath6-value , the deviation from the flat profile was noted before ( see , e.g. , athreya et al .
outwardly decreasing @xmath6-profiles are generally obtained if only late - type or blue galaxies are considered ( carlberg et al .
1997b ; biviano & girardi 2003 ) .
however , using the red photometric band , rines et al .
( 2000 ) found a steeply decreasing @xmath7-profile in a576 .
currently , the @xmath7-profiles of groups is still controversial , because it is difficult to derive a mass profile for those systems ( e.g. mahdavi et al .
1999 ; carlberg et al .
2001 ) .
two competing processes affect the relative distributions of mass and luminosity in a cluster : viz .
dynamical friction , and tidal stripping . according to mamon ( 2000 ) and takahashi et al .
( 2002 ) the effect of tidal stripping should dominate over the effect of dynamical friction , so that massive galaxies sink toward the cluster center but loose their halos due to tidal stripping .
the result is an @xmath116-value that increases towards the cluster center .
this effect is also observed in numerical simulations ( ghigna et al .
luminosity segregation of the very massive galaxies can result at an early epoch of the cluster formation ( governato et al .
the fact that we do not find significant segregation of the cluster ellipticals and s0s with respect to the cluster mass ( if we exclude the very bright ellipticals ) suggests that tidal stripping is able to remove the dark haloes of these galaxies , but not to affect the galaxies luminosities very much .
if tidal stripping affected the luminosities , of the early - type galaxies , we should expect a decrease of their mean luminosity towars the center , but our data do not show such a trend ( see paper xi ) .
this is also consistent with the finding of natarajan , kneib , & smail ( 2002b ) , based on a gravitational lensing analysis of six galaxy clusters , that the typical halo truncation radius of cluster galaxies is 510 times larger than the optical radius .
we have determined the mass profile of an ensemble cluster built from 59 nearby clusters from the eso nearby abell cluster survey . in this ensemble cluster , we distinguished four different classes of galaxies , viz .
the brightest ellipticals ( e@xmath12 , with @xmath117 ) , the early - type galaxies , and early and late spirals , because they have significantly different projected phase - space distributions .
galaxies in substructures were excluded from the analysis . because the early - type galaxies have a nearly isotropic velocity distribution we derived a non - parametric estimate of the cluster mass profile from their number - density and velocity - dispersion profiles , through the jeans equation of stellar dynamics .
we compared our result with models for the mass profile of dark matter halos , viz .
the nfw , m99 , sis and burkert models . from a comparison of observed and predicted velocity - dispersion profiles of the early - type galaxies , it appears that all 4 mass models provide acceptable fits to the data .
the best - fitt nfw model has a concentration parameter of @xmath118 ( 68% confidence level ) , in agreeement with results from numerical simulations .
the ( sis and burkert ) core - models are only acceptable if their core - radius is sufficiently small ( i.e. @xmath119 at the 99% c.l . ) .
the @xmath7-ratio of the early - type galaxies is remarkably independent of radial distance .
the @xmath7-ratio of all galaxies together is not constant , and within @xmath89 it shows a strong decrease towards the cluster center , which is due to the e@xmath12 .
the spirals cause a mild decrease of the @xmath116-ratio towards larger radii .
our results support cdm models for the build - up of galaxy clusters , and put an upper limit to the cross - section of any sidm that could dominate the cluster gravitational potential .
we caution that our results only apply to the total mass profile , since in our analysis we can not disentangle the baryonic and dark matter components . subtracting
the icm component would make the mass profile steeper , but subtracting the contribution of the central cd ( which is present in many of the 59 clusters ) would make it shallower .
we thank tom thomas , gary mamon , and srdjan samurovi for useful discussions .
ab acknowledges the hospitality of the leiden and marseille observatories .
pk acknowledges the hospitality of the trieste observatory .
this research was partially supported by the leids kerkhoven - bosscha fonds , leiden observatory , and the italian ministry of education , university , and research ( through miur grant cofin2001028932 `` clusters and groups of galaxies , the interplay of dark and baryonic matter '' ) .
it is obviously very important that the analysis of the mass profile is done on galaxy samples from which the non - members have been removed .
we indentify non - members ( or interlopers ) with the procedure devised by den hartog & katgert ( 1996 ) .
this procedure is based on an iterative determination of an interim mass profile , with which galaxies that have relative line - of - sight velocities that are inconsistent with their projected position and with the mass profile can be found .
the procedure is quite conservative in that it does not eliminate cluster members , while galaxies outside the turn - around radius can be recognized quite easily .
the performance of the method was checked on numerical models of galaxy clusters by van kampen ( 1994 ; see also van kampen 1995 ) which describe a two - component ensemble of galaxies and dark matter .
the procedure requires a minimum number of galaxies to start with , because the interim mass profile needs to be sufficiently well defined that the elimination of interlopers is meaningful . in practice , den hartog & katgert found that if the number of galaxies is less than about 45 , the iterative procedure quite often does not converge , while for 45 galaxies or more the convergence generally is quite fast .
we applied the method to the clusters in our sample which have at least 45 galaxies with redshifts .
for the clusters with less than 45 galaxies with redshifts , we used the interloper identification of the richer clusters in a statistical sense .
since the interloper rejection method is blind to galaxy morphologies , the statistical definition of the separation between members and interlopers does not require us to limit ourselves to clusters with sufficient morphological data .
therefore we used 79 clusters with more than 45 galaxies with redshifts , of which 32 are from enacs and 47 from the literature . in the upper panel of fig .
[ f - intlop ] we show the 671 interlopers identified with the method of den hartog & katgert ( 1996 ) in these 79 clusters , which contain a total of 8829 galaxies . at each projected radial distance @xmath84
there is a well - defined lower limit to the velocity distribution of the interlopers . in the lower panel of the same figure
we show the 8158 galaxies that were accepted as member galaxies . in both panels , we draw the same caustic that we visually defined as a good separation between the two subsets . note that the inclusion of data from the literature helps definining the caustic , as the enacs data are mostly restricted to the inner @xmath120 .
the caustic defined in this way was used for the identification of the interlopers in clusters with less than 45 galaxies . because we transfer the caustic in the -plane , the criterion for removing interlopers is not systematically different for the two sets of clusters .
we determine the projected number density profiles @xmath23 for the galaxy classes discussed in [ s - data ] as follows . because the jeans equation requires the logarithmic derivative of the deprojected number density profile , @xmath25 , a smooth representation of @xmath25
is needed .
this is most easily obtained if the estimate of @xmath23 is also smooth , i.e. not binned . in order to obtain a smooth estimate of @xmath23 we have developed a variant of the lowess technique , which was originally designed by gebhardt et al .
( 1994 ) to produce smooth velocity dispersion profiles .
the galaxies are ordered in increasing projected distance @xmath84 . at the position of each galaxy
the density @xmath23 is calculated as the inverse of the distance between the n - th neighbours on either side .
a density estimate typically involves about 10 galaxies . a smoothed version of these densities is produced with the lowess routine , through a weighted linear fit at each galaxy position , which uses between 30 and 80% of all densities , but with a weight that steeply drops at large distances from the galaxy in question .
we chose the lowess smoothing factor guided by the binned versions of @xmath23 .
in the estimation of @xmath23 , each galaxy has a weight associated .
this is because the enacs spectroscopy was done with circular plug - plates used to position the fibers .
even if a cluster was observed with a single plate , this was not necessarily centered on what , after the spectroscopy , turned out to be the cluster center .
the resulting azimuthal incompleteness , which at large projected distances can actually be quite severe ( see e.g. fig . 10 in paper
i ) , depends only on the positioning of the plate(s ) with respect to the cluster center , and can be easily calculated .
assuming azimuthal symmetry , which should be a good approximation for the ensemble cluster , we derived this so - called optopus weight , which is the inverse of the fraction of the azimuth covered in the parent cluster of the galaxy , at its projected distance .
galaxies beyond the distance where the optopus weight gets larger than 3.33 are ignored in the analysis of the mass profile .
the optopus weight was multiplied by a second weight , which takes into account the fact that in the ensemble cluster , not all clusters do contribute at large projected distances @xmath121 .
if not properly accounted for , this incompleteness effect would produce an artificial steepening at larger radii .
the correction for this effect assumes that in the radial range where a cluster does not have data , its contribution can be ` invented ' from the clusters that _ do _ have data , as first described by merrifield & kent ( 1989 ) . in our application of the method
we invent data only from measured data , and not from data that are themselves ( partly ) invented .
we check the consistency of the smoothed @xmath23 with 64 bootstrap resamplings of the data , because the latter give a good idea of the range of @xmath23 s implied by our data .
the results are shown in fig .
[ f - dprof ] ( upper panel ) .
we show the best lowess estimate ( heavy line ) , within the 68% confidence interval determined from the 64 bootstrap resamplings ( dashed lines ) .
we determine the deprojected profile , @xmath25 , for each of the 64 bootstrap resamplings of @xmath23 , through the standard abel deprojection equation ( see , e.g. , bt ) , viz .
@xmath122 via numerical integration .
the deprojection requires knowledge of the @xmath23 to arbitrarily large @xmath84 . in practice , we set the upper limit of the integral equal to @xmath123 ( it corresponds to @xmath124 mpc for a massive cluster ) , since this is sufficiently far from the last measured point .
we use the following smooth parametric function for the extrapolation of @xmath23 , @xmath125 where @xmath126 is the largest value of @xmath84 among the galaxies in the sample .
continuity of @xmath23 and its derivative at the last measured point , @xmath126 , fixes the values of the parameters @xmath127 and @xmath128 , while the shape of the extrapolation depends on the value of @xmath129 .
however , it turns out that for @xmath130 the detailed form of the extrapolation is not important , as different values of @xmath129 lead to changes in @xmath25 which , when projected , produce variations in @xmath23 which are much smaller than the intrinsic uncertainties ( as indicated by the bootstrap resamplings ) . the deprojected number density profile so obtained is shown in fig .
[ f - dprof ] ( lower panel ) for @xmath131 .
we show the best estimate ( heavy line ) within the 1-@xmath132 confidence interval determined from the 64 bootstrap resamplings ( dashed lines ) .
the same procedure is adopted for the determination of the luminosity - density profiles , @xmath78 , but with an additional weight proportional to the luminosity of each galaxy , @xmath133 , where @xmath134 is the absolute magnitude . in paper
v we have shown that the enacs galaxy samples form unbiased subsets of the ( magnitude - complete ) cosmos catalogue ( e.g. wallin et al .
1994 ) as far as position is concerned , so that there is no magnitude - dependent bias in the selection of the galaxies in the enacs spectroscopic sample .
it is important to note , however , that we do not estimate the _ total _ cluster luminosity - density , since this would require the determination of the cluster luminosity function and its extrapolation beyond the observed magnitude range . in the jeans equation
( eq . [ e - jeans ] ) knowledge of the logarithmic derivative @xmath135 is needed .
we determine it by fitting straight lines to the @xmath136 vs. @xmath137 data - points within intervals of @xmath91 width , spaced by @xmath138 .
the procedure is repeated for each of the 64 bootstrap estimates of @xmath25 , thus allowing us to estimate the r.m.s . of the values of the fitted line slopes , and hence the confidence levels on @xmath135 .
we checked for consistency that this estimate of @xmath135 is close to the direct ( smoothed ) numerical differentiation of @xmath139 . in order to determine the function @xmath30 for a galaxy class with zero velocity - anisotropy
, we need to deproject the observed line - of - sight velocity dispersion profile @xmath24 , via the abel equation ( see , e.g. , bt ) , @xmath140}{dr } \frac{dr}{\sqrt{r^2-r^2 } } , \label{e - abel2}\ ] ] which requires a smooth estimate of the product @xmath141 .
we determine @xmath24 using the lowess technique , with a smoothing length chosen to ensure consistency with the binned estimate of @xmath24 .
here , galaxies beyond @xmath3 are not used , since some of them could be out of equilibrium , which could bias the estimate of the cluster velocity dispersion .
the @xmath10-profile is shown in fig .
[ f - vdp ] .
we show the best lowess estimate ( heavy line ) , within the 68% confidence interval ( dashed lines ; determined with 1000 montecarlo simulations in the lowess program , see gebhardt et al .
1994 ) , as well as the binned estimates .
after extrapolating the observed @xmath24 to @xmath142 , we deproject @xmath143 to obtain @xmath30 through eq .
[ e - abel2 ] .
the inversion is also done for 64 of the montecarlo simulations of @xmath24 , in order to obtain the confidence intervals in @xmath30 ( shown in fig .
[ f - vdp ] ) .
we did not include the uncertainties in @xmath23 , since the uncertainties in @xmath143 are dominated by the uncertainties in @xmath144 . in the determination of @xmath145 and its confidence levels
, we followed the same procedure that we used for @xmath135 ( see appendix [ a - dprof ] ) .
the confidence levels are obtained by running the same procedure on the 64 @xmath146 profiles obtained from the montecarlo simulations .
we check for consistency that the @xmath147 obtained by this procedure is close to the smoothed estimate of the numerical differentiation of @xmath30 .
> from the @xmath135 , @xmath146 , and @xmath148 profiles , we finally determine the mass profile via the jeans equation ( eq . [ e - jeans ] ) . the c.l . on the mass profile
immediately follow from the c.l . of the individual terms of the equation , under the simplified assumption of mutually independent errors .
such an assumption is certainly too simplistic , but all we want to achieve is a rough estimate of the mass profile c.l . , to be used for illustration purposes only .
when combining several clusters together , it is important that the centers of all individual clusters are determined with sufficient accuracy and in a uniform manner .
we followed the procedure described in paper iii of this series , i.e. we used in decreasing order of preference the x - ray center , the position of the brightest cluster member near the center , the peak in the galaxy distribution , or the biweight average of the positions of all galaxies .
the positions of the adopted centers are given in table a.1 in paper xi .
in our analysis , we assume that the ensemble cluster is not rotating . such an assumption is probably valid for individual clusters .
rotational motions are expected ( and observed ) in the collisional component of clusters ( the intra - cluster gas ) as a consequence of cluster mergers ( dupke & bregman 2001 ) , but not in the collisionless components ( galaxies and dark matter ; see roettiger & flores 2000 ) .
velocity gradients have been reported in clusters but they have not been interpreted as a signature for rotation ( e.g. biviano et al .
1996 ; den hartog & katgert 1996 ) .
another assumption we make is that of steady state .
this is likely to be a valid assumption for _ nearby _ clusters .
in fact , according to ellingson et al .
( 2001 ) the infall rate of field galaxies in clusters at recent epochs is negligible ( probably less than 1% in the last 2 gyr ) .
moreover , girardi & mezzetti ( 2001 ) found no indication of evolution in the density and velocity dispersion profiles of galaxy clusters between @xmath149 and @xmath150 .
finally , in the currently favoured low-@xmath151 cosmology , merging and matter accretion from surrounding filaments is expected to have stopped long ago , leaving the time for clusters to dynamically relax ( e.g. plionis 2002 ) .
we also assume spherical symmetry . while individual clusters depart from spherical symmetry , the ensemble cluster is spherically symmetric by construction .
an analysis of cluster ellipticities is beyond the scope of this paper , but we expect them not to be very large for the following reasons .
first , our clusters are rich , and de theije , katgert , & van kampen ( 1995 ) showed ellipticity to decrease significantly with increasing richness , reaching a value of @xmath152 for abell richness @xmath153 .
second , our clusters are nearby , and both melott et al .
( 2001 ) and plionis ( 2002 ) showed cluster ellipticity to decrease with decreasing redshift , reaching @xmath154 at @xmath155 .
third , we only considered the inner , virialized parts of clusters , where the tidal elongation and the effects of accretion from the surrounding large scale structure are less conspicuous ( e.g. melott et al . 2001 ; west & bothun 1990 ) .
adami , c. , biviano , a. , & mazure , a. 1998a , a&a , 331 , 439 adami , c. , mazure , a. , biviano , a. , katgert , p. , rhee , g. 1998c , a&a , 331 , 493 ( paper iv ) adami , c. , mazure , a. , katgert , p. , biviano , a. 1998b , a&a , 336 , 63 ( paper vii ) allen , s. w. , ettori , s. , & fabian , a. c. 2001 , mnras , 324 , 877 allen , s. w. , schmidt , r. w. , & fabian , a. c. 2002 , mnras , 335 , 256 arabadjis , j. s. , bautz , m. w. , & garmire , g. p. 2002
, apj , 572 , 66 arieli , y. , & rephaeli , y. 2003 , new a , 8 , 517 athreya , r. m. , mellier , y. , van waerbeke , l. , pell , r. , fort , b. , & dantel - fort , m. 2002 , a&a , 384 , 743 bacon , r. , simien , f. , & monnet , g. 1983 , a&a , 128 , 405 beers , t. c. , flynn , k. , & gebhardt , k. 1990 , aj , 100 , 32 beisbart , c. , valdarnini , r. & buchert , t. 2001 , a&a , 379 , 412 binney , j. & tremaine , s. 1987 , galactic dynamics ( princeton : princeton university press ) ( bt ) biviano , a. , durret , f. , gerbal , d. , le fvre , o. , lobo , c. , mazure , a. , & slezak , e. 1996 , a&a , 311 , 95 biviano , a. & girardi , m. 2003 , apj , 585 , 205 biviano , a. , girardi , m. , giuricin , g. , mardirossian , f. , & mezzetti , m. 1992 , apj , 396 , 35 biviano , a. , katgert , p. 2004 , a&a ( to be submitted ) biviano , a. , katgert , p. , mazure , a. , moles , m. , den hartog , r. , perea , j. , & focardi , p. 1997 , a&a , 321 , 84 ( paper iii ) biviano , a. , katgert , p. , thomas , t. , & adami , c. 2002 , a&a , 387 , 8 ( paper xi ) bullock , j. s. , kolatt , t. s. , sigad , y. , somerville , r. s. , kravtsov , a. v. , klypin , a. a. , primack , j. r. , dekel , a. 2001 , mnras , 321 , 559 burkert , a. 1995 , apj , 447 , l25 campusano l. e. , kneib , j - p . &
hardy , e. , 1998 , apj , 496 , l79 carlberg , r. g. , yee , h. k. c. , & ellingson , e. 1997a , apj , 478 , 462 carlberg , r.g .
, et al . 1997b , apj , 476 , l7 carlberg , r.g .
, et al . , 1997c , apj 485 , l13 carlberg , r. g. , yee , h. k. c. , morris , s. l. , lin , h. , hall , p. b. , patton , d. r. , sawicki , m. , & shepherd , c. w. 2001 , apj , 552 , 427 clowe , d. , luppino , g. a. , kaiser , n. , & gioia , i. m. 2000 , apj , 539 , 540 clowe , d. , & schneider , p. 2001
, a&a , 379 , 384 colless , m. , & dunn , a.m. 1996 , apj , 458 , 435 colless , m. , et al . 2001 ,
mnras , 328 , 1039 crone m. m. , evrard a. e. , richstone d. o. , 1994 , apj , 434 , 402 dahle , h. , hannestad , s. , & sommer - larsen , j. 2003 , apj , 588 , l73 de grandi , s. , & molendi , s. 2002 , apj , 567 , 163 den hartog , r. , & katgert , p. 1996
, mnras , 279 , 349 de theije , p. a. m. , & katgert , p. 1999
, a&a , 341 , 371 ( paper vi ) de theije , p. a. m. , katgert , p. , & van kampen , e. , 1995 , mnras , 273 , 30 diaferio , a. 1999 , mnras , 309 , 610 diaferio , a. , & geller , m. j. 1997 , apj , 481 , 633 dressler , a. 1980 , apj , 236 , 351 dupke , r. a. , & bregman , j. n. 2001 , apj , 547 , 705 ellingson , e. , lin , h. , yee , h. k. c. , & carlberg , r. g. 2001 , apj , 547 , 609 ettori , s. , de grandi , s. , & molendi , s. 2002 , a&a , 391 , 841 ettori , s. , & lombardi , m. 2003 , a&a , 398 , l5 ettori , s. , tozzi , p. , & rosati , p. 2003 , a&a , 398 , 879 firmani , c. , donghia , e. , avila - reese , v. , chincarini , g. , & hernndez , x. 2000 , mnras , 315 , l29 fischer , p. , & tyson , j. a. 1997 , aj , 114 , 14 gebhardt , k. , pryor , c. , williams , t. b. , hesser , j. e. 1994 , aj , 107 , 2067 geller , m. j. , diaferio , a. , & kurtz , m. j. 1999 , apj , 517 , l23 ghigna , s. , moore , b. , governato , f. , lake , g. , quinn , t. , & stadel , j. 1998 , mnras , 300 , 146 girardi , m. , & mezzetti m. 2001 , apj , 548 , 78 governato , f. , ghigna , s. , moore , b. , quinn , t. , stadel , j. , & lake , g. 2001 , apj , 547 , 555 heisler , j. , tremaine , s. , & bahcall , j. n. 1985 , apj , 298 , 8 hoekstra , h. , franx , m. , kuijken , k. , & van dokkum , p. g. 2002 , mnras , 333 , 911 hughes j. p. , 1989 ,
apj , 337 , 21 irwin , j. a. , & bregman , j. n. 2000 , apj , 538 , 543 irwin , j. a. , bregman , j. n. , & evrard , a. e. , 1999 , apj , 519 , 518 jing , y. p. , mo , h. j. , brner , g. , fang , l. z. 1995 , mnras , 276 , 417 jing , y. & suto , y. 2000 , apj , 529 , l69 katgert , p. , mazure , a. , den hartog , r. , adami , c. , biviano , a. , & perea , j. 1998 , a&as , 129 , 399 ( paper v ) katgert , p. , et al .
1996 , a&a , 310 , 8 ( paper i ) kelson , d. d. , zabludoff , a. i. , williams , k. a. , trager , s. c. , mulchaey , j. s. , & bolte , m. 2002 , apj , 576 , 720 kent s. m. , & gunn j. e. , 1982 , aj , 87 , 945 kikuchi , k. , furusho , t. , ezawa , h. yamasaki , n. y. , ohashi , t. , fukazawa , y. , & ikebe , y. 1999 , pasj , 51 , 301 king , l. j. , clowe , d. i. , lidman , c. , schneider , p. , erben , t. , kneib , j .-
, & meylan , g. 2002 , a&a , 385 , l5 lombardi , m. , rosati , p. , nonino , m. , girardi , m. , borgani , s. , & squires , g. 2000 , a&a , 363 , 401 mahdavi , a. , geller , m. j. , bhringer h. , kurtz , m. j. , & ramella , m. , 1999 , apj , 518 , 69 mamon , g. 2000 , in asp conf . ser .
197 , galaxy dynamics : from the early universe to the present , ed .
f. combes , g.a .
mamon , & v. charmandaris ( san francisco : asp ) , 367 markevitch m. , forman w. r. , sarazin c. l. , & vikhlinin , a. , 1998 , apj 503 , 77 markevitch , m. , & vikhlinin , a. 1997 , apj , 491 , 467 markevitch , m. , vikhlinin , a. , forman , w. r. , & sarazin , c. l. 1999 , apj , 527 , 545 melnick , j. , & sargent , w. l. w. 1977 , apj , 215 , 401 melott , a. l. , chambers , s. w. , & miller , c. j. 2001 , apj , 559 , l75 meneghetti , m. , yoshida , n. , bartelmann , m. , moscardini , l. , springel , v. , tormen , g. , & white , s. d. m. 2001 , mnras , 325 , 435 merrifield , m. r. , & kent , s. m. 1989 , aj , 98 , 351 merritt , d. 1987 , apj , 313 , 121 mohr , j. j. , geller , m. j. , fabricant , d. g. , wegner , g. , thorstensen , j. , & richstone , d. o. 1996 , apj , 470 , 724 moore , b. , quinn , t. , governato , f. , stadel , j. , & lake , g. 1999 , mnras , 310 , 1147 ( m99 ) natarajan , p. , kneib , j .-
p . , & smail , i. 2002b , apj , 580 , l11 natarajan , p. , loeb , a. , kneib , j .-
p . , & smail , i. 2002a , apj , 580 , l17 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1996 , apj , 462 , 536 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , apj , 490 , 493 ( nfw ) nevalainen , j. , markevitch , m. , & forman , w. 1999 , apj , 526 , 1 oemler , a. jr .
1974 , apj , 194 , 1 plionis , m. 2002 , apj , 572 , l67 power , c. , navarro , j. f. , jenkins , a. et al . 2003 ,
mnras , 398 , 14 pratt , g. w. , & arnaud , m. 2002 , a&a , 394 , 375 reisenegger , a. , quintana , h. , carrasco , e. r. , & maze , j. 2000 , aj , 120 , 523 ricotti , m. 2002 , mnras , submitted ( astro - ph/0212146 ) rines , k. , geller , m. j. , diaferio , a. , mahdavi , a. , mohr , j. j. , & wegner , g. 2002 , aj , 124 , 1266 rines , k. , geller , m. j. , diaferio , a. , mohr , j. j. , & wegner , g. a. 2000 , aj , 120 , 2338 rines , k. , geller , m. j. , kurtz , m. j. , diaferio , a. , jarrett , t. h. , & huchra , j. p. 2001 , apj , 561 , l41 roettiger , k. & flores , r. 2000 , apj , 538 , 92 sand , d. j. , treu , t. , & ellis , r. s. 2002 , apj , 574 , l129 schaeffer , r. , maurogordato , s. , cappi , a. , & bernardeau , f. 1993 , mnras , 263 , l21 shapiro , p. r. , & iliev , i. t. 2000 , apj , 542 , l1 smith , s. 1936 , apj , 83 , 23 solanes , j.m . , & salvador - sol , e. 1990 , a&a , 234 , 93 squires , g. , kaiser , n. , fahlman , g. , babul , a. , & woods , d. 1996 , apj , 469 , 73 suginohara t. , & ostriker j.p . , 1998 , apj , 507 , 16 takahashi , k. , sensui , t. , funato , y. , & makino , j. 2002 , pasj , 54 , 5 taylor , a. n. , dye , s. , broadhurst , t. j. , benitez , n. , & van kampen , e. 1998 , apj , 501 , 539 tamura , t. , makishima , k. , fukazawa , y. , ikebe , y. , & xu , h. 2000 , apj , 535 , 602 the , l. s. , & white , s. d. m. 1986 , aj , 92 , 1248 thomas , t. 2003 , in preparation ( paper viii ) thomas , t. , hartendorp , m. , & katgert , p. 2003 , in preparation ( paper ix ) thomas , t. & katgert , p. 2003 , in preparation ( paper x ) thomas , p. a. , muanwong , o. , pearce , f. r. , couchman , h. m. p. , edge , a. c. , jenkins , a. , & onuora , l. 2001 , mnras , 324 , 450 tyson , j. a. , wenk , r. a. , & valdes , f. 1990 , apj , 349 , l1 van der marel , r. 1994 , mnras , 270 , 271 van der marel , r. , magorrian , j. , carlberg , r. , yee , h. , & ellingson , e. 2000 , aj , 119 , 2038 van kampen , e. , 1994 , phd thesis , leiden university van kampen , e. , 1995 , mnras , 273 , 295 wallin , j. f. , yentis , d. , macgillivray , h. t. , bauer , s. b. , & wong , c. s. 1994 , baas , 26 , 897 west , m. j. , & bothun , g. d. 1990 , apj , 350 , 36 yagi , m. , kashikawa , n. , sekiguchi , m. , doi , m. , yasuda , n. , shimasaku , k. , & okamura , s. 2002 , aj , 123 , 87 zwicky , f. 1933 , helv .
acta , 6 , 10 zwicky , f. 1937 , apj , 86 , 217 | we determine the mass profile of an ensemble cluster built from 3056 galaxies in 59 nearby clusters observed in the eso nearby abell cluster survey .
the mass profile is derived from the distribution and kinematics of the early - type ( elliptical and s0 ) galaxies only , with projected distances from the centers of their clusters .
these galaxies are most likely to meet the conditions for the application of the jeans equation , since they are the oldest cluster population , and are thus quite likely to be in dynamical equilibrium with the cluster potential . in addition , the assumption that the early - type galaxies have isotropic orbits is supported by the shape of their velocity distribution . for galaxies of other types ( the brightest ellipticals with @xmath0 , and the early and late spirals ) these assumptions are much less likely to be satisfied . for the determination of the mass profile
we also exclude early - type galaxies in subclusters .
application of the jeans equation yields a _ non - parametric estimate _ of the cumulative mass profile @xmath1 , which has a logarithmic slope of @xmath2 in the density profile at @xmath3 ( approximately the virial radius ) .
we compare our result with several analytical models from the literature , and we estimate their best - fit parameters from a comparison of observed and predicted velocity - dispersion profiles .
we obtain acceptable solutions for all models ( nfw , moore et al .
1999 , softened isothermal sphere , and burkert 1995 ) .
our data do not provide compelling evidence for the existence of a core ; as a matter of fact , the best - fitting core models have core - radii well below 100 @xmath4 kpc .
the upper limit we put on the size of the core - radius provides a constraint for the scattering cross - section of dark matter particles .
the total - mass density appears to be traced remarkably well by the luminosity density of the early - type galaxies . on the contrary ,
the luminosity density of the brightest ellipticals increases faster towards the center than the mass density , while the luminosity density profiles of the early and late spirals are somewhat flatter than the mass density profile . |
inflation @xcite is the leading paradiam of the very early universe cosmology .
the quantum fluctuations during inflation are stretched to super - hubble scales and return to the hubble horizon after inflation .
they provide the seeds of cmb anisotropy and the large scale structure @xcite . here
we review the cosmic perturbation theory , which calculates inflationary fluctuations .
there have been already many books , reviews and lecture notes in this area @xcite . in the present review we focus on an explicit treatment of the standard theory , including linear and non - linear perturbations . after that
, we make a broad survey of models . before those aspects
, we briefly overview the inflationary cosmology in the rest of this section .
the dynamics of the standard hot big bang cosmology is governed by the friedmann equation . by assuming the universe is homogeneous , isotropic and flat .
thus for algebraic simplicity we also assume flatness unless stated otherwise . ]
, the metric can be written as the frw form : @xmath0 here @xmath1 is the scale factor of the universe .
note that in a flat universe , the scale factor itself does not have physical meaning . because we can always rescale @xmath1 and spatial coordinate @xmath2 together . on the other hand ,
ratios of scale factors are important .
for example the hubble parameter @xmath3 and the redshift @xmath4 are key quantities in cosmology . with the frw metric
, the 00-component of the einstein equation is the friedmann equation : @xmath5 where @xmath6 is the energy density . for non - interacting energy components ,
the time evolution ( characterizing how this component dilutes with expansion of the universe ) at the homogeneous and isotropic level is determined by the continuity equation @xmath7 where @xmath8 is pressure , which can be determined by @xmath9 ( and perhaps other parameters ) from the equation of state for this form of matter .
for example , for thermal matter , @xmath10 can be calculated from the thermal partition function .
for the cosmological constant , @xmath11 follows from lorentz invariance .
our present universe is matter ( @xmath12 ) dominated . at earlier times , the universe was hotter and radiation ( @xmath13 ) dominated .
however , radiation domination can not be eternal to the past .
this is because from eq .
( [ eq : friedmann ] ) , inserting @xmath14 , we get @xmath15 and @xmath16 , where @xmath17 is an integration constant .
the moment @xmath18 is a physical singularity , where scale factor sinks to zero and physical energy density diverges .
note that @xmath19 is the proper time measured by a comoving observer .
thus the radiation dominated universe must start at some finite physical time ( 137gyr ago ) .
the start of the radiation dominated universe is known as the `` hot big bang '' .
the singularity at @xmath18 requires the hot big bang . but this singularity by itself does not imply what happens before the hot big bang , or even if anything , including spacetime , exists or not before that . nevertheless , the hot big bang cosmology is sensitive to its initial conditions .
there are strong evidences ( almost firm proof ) showing that new physics exists before the big bang .
and inflation is so far the most successful candidate for this very early epoch . in the remainder of the section
, we shall focus on those evidences and how inflation explains them .
it is a surprising fact that we find ourselves living in a huge universe about @xmath20 kilometers in the observable radius and @xmath21 years in age ( where kilometers and years are human s natural units ) .
this turns into a greater surprise when we realize that our universe is supported by fundamental physics , where the natural scales are ev , gev or even planck scale . in those units the age and size of our universe
are about @xmath22ev@xmath23 ( @xmath24gev@xmath25 ) .
those surprises can be formulated in more accurate terms , by the horizon and flatness problems of the standard hot big bang cosmology . to see those problems ,
let us assume for a time that nothing happened before the hot big bang .
in other words assume the universe is radiation dominated all the way back in the early times .
then problems arise : * _ horizon problem ( i ) : _ let us start with an example of the cosmic microwave background ( cmb ) .
+ from observations , we actually see the universe is homogeneous and isotropic in the past , because light ( or any other causal signal ) needs time to propagate to us . by optical probes
, we can observe the homogeneity and isotropy of the universe up to as early as recombination : when the cmb forms , the universe is already very homogeneous and isotropic ( up to fluctuations of order @xmath26 , which we will discuss soon ) .
+ however , from the hot big bang to the time of recombination , causal signals could only propagate a limited distance in space .
this is because those signals could at most propagate at speed of light , with null geodesic @xmath27 .
thus the physical distance of propagation ( measured by today s ruler @xmath28 ) from the big bang to recombination is at most .
nevertheless this does nt make a big difference for this estimate . ]
@xmath29 to put in numbers , the causal signals , from the hot big bang to recombination , could only propagate as long as about 1/100 of the radius of our observable universe .
thus different patches of the cmb sky were casually disconnected .
this situation is illustrated in fig .
[ fig : horizon ] .
+ one could also consider this problem using conformal time .
this is because conformal time , together with comoving spatial coordinates , straightforwardly shows the causal structure of spacetime .
the flat frw metric can be rewritten as @xmath30 where @xmath31 is defined ( up to a constant ) by @xmath32 . using conformal time , the light cones are 45 degree straight lines in the @xmath31-@xmath33 plane . thus the distance luminal signals travel equals to conformal time .
+ consider equation of state @xmath34 , where @xmath35 is a constant .
one finds @xmath36 .
thus @xmath37 for a fluid with @xmath38 ( including matter and radiation ) , a very short period of proper time @xmath39 corresponds to a very short period of conformal time @xmath40 .
thus light can not propagate far during this period of time .
+ the horizon problem of hot big bang cosmology . the space and time in this diagram should be understood as comoving and conformal respectively .
the several smaller light cones are truncated between the hot big bang and recombination ( the time of cmb formation ) .
those light cones have no overlap at the time of hot big bang , thus they are causally unrelated at recombination . but why , they at recombination ( for example a and b in the figure ) , look similar to each other ( considering we can observe many of them , as indicated by the greater light cone between recombination and the present time)?,scaledwidth=60.0% ] + note that in the cmb example of the horizon problem , @xmath41 is set by the scale of observation instead of the intrinsic nature of the problem itself . more generally , we could consider different observables ( replace @xmath41 with other time scales ) and generate a family of similar horizon problems .
for example , problems of causally disconnect patches of large scale structure .
* _ horizon problem ( ii ) : _ one could alternatively formulate the puzzle of horizon size in another way .
imagine we are observers sitting at @xmath42 , who have existed since the hot big bang , and there exists some structural information @xmath43 which are comoving in the universe ( including homogeneity , isotropy and flatness , and departures from those ideal conditions ) .
we could observe @xmath43 by luminal or sub - luminal signals emitted from the structures on @xmath43 . now at time @xmath44 ( not before or after ) , how much portion of @xmath43 could we observe ( in comoving size ) ?
+ we find there is a competition in this situation . on the one hand ,
as @xmath44 becomes larger , the past light cone is larger thus we tend to see more structures on @xmath43 . on the other hand , the expansion of the universe itself tends to hide those structures into more distant regions , where we may not be able to observe .
interestingly , those two effects depends on different order of derivatives of @xmath1 .
thus typically there is a winner between them .
the comoving size of the observable universe is characterized by the comoving horizon size : + @xmath45 + we find that as we move forward in time , we see more and more portion of @xmath43 . in other words , comoving objects are floating inwards from outside of our horizon .
this is not to be confused with the concepts of the particle horizon or the event horizon . ] . or from the horizon point of view
, the universe is not expanding `` fast '' enough , and thus the horizon keeps expanding in comoving size .
( here we have quoted `` fast '' , because the time dependence of the quantity @xmath46 is actually the acceleration of expansion , instead of the expansion speed . )
+ now comes the horizon problem : if no one could never `` see '' ( more accurately , interact with ) those structures in the early universe , how could they be causally prepared , and prepared in a consistent way ?
this situation is illustrated in fig .
[ fig : flatness ] and fig .
[ fig : horizon - return ] .
+ the horizon problem ii , or the flatness problem . at early times
, the observer could only observe a small portion of the universe thus unaware of larger scale structures ( e.g. the spherical topology in the figure ) . at later times though
radiation or matter domination , the universe expands .
however , the observer s horizon expands faster .
thus the observer could obtain more large scale structural information ( e.g. easier to measure the non - flatness of the universe , if exists ) . from current observation ,
the universe is rather flat now .
if there were non - flatness in our universe , how could causal process produce them in the early universe , considering no causal process could see this non - flatness in the early universe?,scaledwidth=60.0% ] + the comoving scales return to the horizon in radiation and matter dominated eras.,scaledwidth=95.0% ] * _ flatness problem : _ the flatness problem could be considered as an example of the version ( ii ) of the horizon problem . here
the structural information @xmath43 is the spatial curvature of the universe . for simplicity ,
we havent put spatial curvature into the metric .
nevertheless , we could imagine we live on a sphere instead of a flat plane .
+ when the universe expands , the curvature radius @xmath47 should expand together , keeping @xmath48 ( with proper normalization of @xmath1 ) . nevertheless ,
as horizon problem ( ii ) , in matter or radiation dominated universes , the universe does not expand `` fast '' enough thus we still see more and more portion of the sphere as time goes on .
+ so far , we havent seen any curvature ( @xmath49 ) .
thus at a much earlier time , @xmath47 has to be greater than @xmath50 by a much greater amount .
how could such a flat universe be prepared at the beginning , without the help of causal processes ?
this is not yet the end of the story .
some uv - completion theories of the particle physics standard model predict topological defects . from those theories ,
another question arises : * _ monopole problem _ :
phase transitions of the early universe may produce gauge theory monopoles . if produced , they are supposed to dominate the energy density the universe .
but we havent found them .
( on the other hand , global monopoles is not necessarily a problem . )
so far we have outlined a few problems for the standard picture of the hot big bang cosmology . however , it is worth noticing that sometimes it is hard to judge the seriousness of conceptional problems : on the one hand , it is because those conceptional problems are paradigm dependent .
for example , the monopole problem could be simply solved by starting with a theory without monopole solutions .
one could as well argue that the universe could start from some almost deterministic initial conditions where the universe is extremely flat and uniform .
those theories exist , such as the quantum creation of universe @xcite ( see also @xcite for a different approach with different predictions ) .
on the other hand , solution to some problems may not be complete , or may also raise other ( though arguably less serious ) problems .
we will come back later to the new problems brought by inflation .
however , there does exist physical problem of the hot big bang , without proper extension before the bang .
this is the problem of large scale correlations on the cmb ( or in the large scale structure ) . *
_ correlation problem : _ again consider fig .
[ fig : horizon ] .
actually the cmb map is not absolutely homogeneous and isotropic , but instead has small fluctuations on it , with an amplitude @xmath51 . now consider the statistical correlation of those fluctuations , in regions a and b. assuming nothing exists before the hot big bang , a and b should nt have correlations .
however , those statistical correlations were caught , mode by mode in momentum space , in the cmb experiments @xcite . with nowadays cmb map ( wmap:@xcite act:@xcite spt:@xcite ) , one can actually observe by eyes that the fluctuations are very different from a map generated from white noise . indeed , historically it is the verified predictions of inflation from the correlation in the cosmic perturbations , instead of the conceptual `` postdictions '' , that convinced the majority that inflation should be the theory of the very early universe and produce the dominant seed of cmb and structure formation . inflation is a period of time before the hot big bang , during which the universe undergoes accelerating expansion : @xmath52 for example , consider a constant equation of state @xmath34 . in this case acceleration
implies @xmath53 .
and as we shall see , inflation typically require @xmath54 ( approach -1 from the greater side ) . before going into detailed dynamics which could drive the above accelerating expansion ,
let us first have a look how inflation solves the problems in the previous subsection . * _ horizon problem ( i ) : _ from equation ( [ eq : sol - conformal - time ] )
, we find that in case of acceleration ( @xmath53 ) , @xmath55 .
thus small proper time interval @xmath56 means negative and large conformal interval @xmath57 .
+ remember that @xmath40 corresponds to the distance that light ray could travel .
@xmath57 means that light ray could travel a very long distance during the @xmath53 epoch ( although could be a small interval of proper time @xmath58 ) before reaching the time of recombination . thus fig .
[ fig : horizon ] is replaced by fig .
[ fig : horizon - inflation ] , where the initial @xmath17 hypersurface is stretched downwards in conformal distance , and light cones with tips on cmb becomes highly overlapped at @xmath17 .
+ alternatively , we can re - calculate the light propagation distance , similarly to eq . : during inflation @xmath59 this corresponds to an exponentially large distance , that light signals could have travelled during inflation .
* _ horizon problem ( ii ) and the flatness problem : _ those problems arises because @xmath60 was a decreasing function all the way since the creation of the universe ( i.e. @xmath61 ) .
now we have a period of inflation @xmath62 before the hot big bang .
thus the horizon problem ( ii ) and the flatness problem are solved provided that the @xmath62 period is long enough .
this situation is illustrated in fig .
[ fig : horizon - inflation ] , where the left and right panels plots the same physics , in spacetime diagram and in terms of change of comoving horizon scale , respectively .
+ inflation solves the horizon problem .
thanks to the accelerated expansion , the conformal time @xmath31 is no longer tightly bounded by the big bang singularity @xmath63 .
thus the past light cones on the cmb overlap in the past , which solves the horizon problem.,scaledwidth=60.0% ] + the comoving scales exits the horizon during inflation , and later returns during radiation and matter dominated eras.,scaledwidth=95.0% ] * _ monopole problem : _ if monopoles were created before inflation , they could easily be inflated away .
thus it is no wonder we can not observe them today .
( however , one should still pay attention not to produce gauge theory monopoles , domain walls or very strong cosmic strings after inflation . ) * _ correlation problem : _ how inflation solves this problem
is discussed in detail in later sections .
the key idea is that , the sub - hubble quantum fluctuations are stretched by cosmic expansions during inflation , into super - hubble fluctuations .
those fluctuations return to the horizon later after the end of inflation .
those returned modes cause large scale correlations on the cmb .
this is illustrated in fig .
[ fig : horizon - inflation ] .
inflation is clearly a very powerful tool for solving problems with standard cosmology , so let us now proceed to see how inflation can be explicitly realized .
the requirement @xmath64 is not that easy to obtain . because our familiar components in the universe have positive pressure .
for example , consider a gas of particles in a box .
the particles always `` want '' to escape the box , and hence generate positive pressure by hitting the surface of the box .
the same difficulty can be also observed from the continuity equation @xmath65 . again
consider a gas component ( but now in the whole universe ) .
both particle density and momentum of particles get diluted when the universe expands .
thus @xmath9 should dilute at least as quickly as @xmath66 .
especially , when @xmath54 , we want a component whose energy density is almost not diluted when the universe expands . to have such an acceleration , one immediate idea is to use vacuum energy .
this is because , the vacuum , after adiabatic expansion , is still vacuum .
we shall come back to this issue later in details . ] .
thus we have @xmath67 and the universe accelerates .
however , if we use the vacuum energy to accelerate our early universe , problems arise : * _ inflation must end .
_ our universe is not always in a state of acceleration .
observationally , significant parts of our directly observable history was radiation and matter dominated and the universe decelerates instead of accelerates in expansion .
theoretically , we need the comoving structures to return to the horizon . * _ the end of inflation must be controlled by a rather accurate clock_. the overlapping of light cones in fig .
[ fig : horizon - inflation ] does not guarantee ( though makes it possible ) that we observe an almost homogeneous and isotropic universe .
for example , if one significant part of the universe has much earlier end of inflation , then energy density drops quickly in that part and it will produce a void in our universe , breaking homogeneity and isotropy . the vacuum energy itself does not provide an exit mechanism for inflation . moreover ,
even if we add anything , like tunneling to another vacuum , as the end of inflation , it will not solve the problem as long as there is no clock .
for example , the tunneling to another vacuum happened spontaneously instead of being controlled by a clock .
be the decay rate of the false vacuum .
then when @xmath68 , inflation will end within one e - fold , thus do not really solve problems .
on the other hand , if @xmath69 , true vacuum bubbles can not percolate because the space between true vacuum bubbles expand too quickly . in this case
we found ourselves in an open bubble @xcite , and the flatness problem returns .
thus inflation will either end too quickly or will never end globally . ]
thus , we have to attach a clock to vacuum energy , and use this clock to end inflation .
a clock is something that takes a number at every physical point in the spacetime manifold , with the numbers increasing along an observer s world line .
thus a clock can be simply realized by a rolling scalar field .
in other words we would need an effective vacuum energy , controlled by the motion of a scalar field . the change from a vacuum energy to a rolling scalar field
can be thought of as follows : a vacuum energy plus a massless scalar can be viewed in an alternative way as a scalar field lying on a flat potential . by slightly tilting the potential , we have a slowly rolling scalar field acting as an effective vacuum energy
. this picture can be visualized by the analog of a rolling ball along a very flat surface , as shown in fig .
[ fig : inflation - potential ] .
the rolling scalar is called the `` inflaton '' .
further , inflation could be ended by adding some additional mechanisms triggered by some particular value of inflaton .
for example , as in fig .
[ fig : inflation - potential ] , after @xmath70 , the potential @xmath71 is no longer flat thus inflation ends when @xmath72 rolls to @xmath70 . in this case
, the inflaton field is the clock field of the universe .
left panel : a rolling scalar field is a modest deformation of pure vacuum energy .
right panel : the inflationary potential .
the shape of the inflationary potential is largely unconstrained , as long as it is flat enough .
inflation ends at where the potential is no longer flat and a process of reheating follows to reheat the universe.,title="fig:",scaledwidth=30.0% ] left panel : a rolling scalar field is a modest deformation of pure vacuum energy .
right panel : the inflationary potential .
the shape of the inflationary potential is largely unconstrained , as long as it is flat enough .
inflation ends at where the potential is no longer flat and a process of reheating follows to reheat the universe.,title="fig:",scaledwidth=30.0% ] to be explicit , the lagrangian for a scalar field is @xmath73~,\end{aligned}\ ] ] in the homogeneous and isotropic background level , the above equation is reduced to @xmath74 ~.\end{aligned}\ ] ] the equation of motion for @xmath72 can be obtained from varying this action with respect to @xmath72 : @xmath75 where prime denote derivative with respect to @xmath72 .
it is interesting to compare this equation with the flat space equation of motion for a homogeneous and isotropic scalar field : @xmath76 . the new term in eq .
( [ eq : eom - phi - bg ] ) is a friction term . to see this ,
consider @xmath77 . in this case
the solution of @xmath72 is @xmath78const and @xmath79 .
the @xmath78const solution means a `` ball '' which is at rest .
the @xmath79 solution means that when there is no additional driving force , the initial velocity decreases . to solve eq .
( [ eq : eom - phi - bg ] ) , we need yet to know @xmath80 .
thus we have to couple the @xmath72 field to gravity .
there are a few ( essentially the same ) ways to couple @xmath72 to gravity .
one could either vary the gravitational lagrangian together with @xmath81 with respect to @xmath82 , or vary @xmath81 in eq .
( [ eq : lag - phi ] ) with respect to @xmath82 to get the stress tensor , where @xmath83 , and @xmath9 and @xmath8 are given in and .
] and insert this stress tensor into the einstein equations . instead , here we take a simpler route : the physical energy density can be obtained by a legendre transform : @xmath84 compare the continuity equation @xmath85 with the equation of motion ( [ eq : eom - phi - bg ] ) , one concludes @xmath86 thus we have inflation when potential energy dominates . inserting @xmath9 into the friedmann equation , we get @xmath87 now we are in the position to solve this system ( with eqs .
( [ eq : eom - phi - bg ] ) and ( [ eq : eom - h - bg ] ) ) .
it is in general difficult to find solutions for a general @xmath88 .
for some special potentials , the system can be solved .
for example , one can make an ansatz @xmath89const , and this case is called `` power law inflation '' @xcite , resulting in an exponential potential of @xmath72 .
however , here we expect more general solutions .
thanks to the friction force , there is a slow - roll - approximated solution of eq .
( [ eq : eom - phi - bg ] ) when the potential is flat enough , regardless the details of the potential .
this is the solution that the fraction force balances the external force @xmath90 , such that the acceleration of @xmath72 does not change significantly .
in addition , to have inflation , we want the potential energy @xmath71 dominates over the kinetic energy @xmath91 .
thus the approximation is @xmath92 the first condition guarantees inflation because in this case @xmath93 .
the second condition requires @xmath94 to change slowly such that the first condition can be satisfied for long enough time . with the above approximations ( [ eq : sr - raw ] ) ,
given the value of @xmath72 , we can first solve @xmath80 by @xmath95 and then insert the resulting @xmath80 into @xmath96 to obtain @xmath97 after obtaining those solutions , we need to step back to verify the assumptions in eq .
( [ eq : sr - raw ] ) . inserting the solutions of @xmath80 and @xmath94
, we have @xmath98 the subscript @xmath88 means that those small `` slow roll parameters '' are defined using the inflation potential .
this is convenient because by model construction , we typically start with a potential and want to know immediately if the potential is flat enough to have inflation . in those _
small _ parameters .
it turns out quite hard to get those slow roll conditions satisfied . ]
especially , when inflation is driven by the mass of the inflaton , we have @xmath99 thus @xmath100 actually denotes the mass square of the inflaton , measured in hubble units .
the condition @xmath101 means , the dynamics of the _ inflationary background _ , has an energy scale much smaller than hubble parameter .
as we shall see later , the inflationary perturbations actually lives on the energy scale @xmath80 , with initial condition set on an even higher energy scale .
it is also convenient to define another set of slow roll parameters using the dynamics of scale factor : @xmath102 when slow roll conditions @xmath103 and @xmath104 are satisfied , those slow roll parameters are related to @xmath105 and @xmath100 by @xmath106 when slow roll conditions are not satisfied , the relations between ( @xmath107 , @xmath108 ) and ( @xmath105 , @xmath100 ) are non - linear . as we have shown , in the slow roll approximation , given a flat enough potential , there exists a slowly rolling inflationary solution .
to take this solution seriously , it is important to consider the naturalness : could this solution be reached easily without highly fine - tuned initial conditions ?
this is in general a hard problem . because on the one hand , the dynamics before inflation is yet unknown .
thus the a priori naturalness is not well defined . on the other hand , even consider the mathematical naturalness , one need to consider a measure of the phase space .
unfortunately , the phase space measure for inflation divergences and different regularization method leads to different conclusions @xcite .
nevertheless the solution of slow roll inflation is an attractor , in the sense that a general point in the phase space of inflation typically flows onto a slow roll trajectory .
an attractor solution is natural , at least in a naive way .
there is an intuitive way to see this attractor behavior : the potential energy of inflation acts as an effective vacuum energy , and thus do not significantly redshift during inflation .
however , the kinetic energy of inflaton tends to be redshifted away quickly .
as discussed in the last section , without continuous injection of energy from the potential , the kinetic energy of @xmath72 decays as @xmath109 .
thus , on the one hand , if the universe starts potential dominated , we are fine to have inflation already . on the other hand ,
if the universe starts kinetic energy dominated , with a small coherent potential energy , the kinetic energy will get diluted away and the universe will eventually become potential dominated .
thus inflation is a dynamical attractor solution .
as we already emphasized , it is important to end inflation . from local energy conservation ,
the effective vacuum energy should go somewhere .
the simplest case is that this effective vacuum energy decays into matter or radiation , which heats the universe .
this process is called reheating . after reheating ,
the universe is radiation dominated , thus @xmath110 where @xmath80 is the energy scale at reheating .
we require the reheating temperature to be compatible with bbn .
thus @xmath111mev .
this is considered as a hard lower bound for inflationary energy scale @xmath112gev .
nevertheless , we expect inflation to happen at much higher energy scales , because it is increasingly harder to build a flat potential at extremely low energy scale .
if reheating is sufficient , most of the vacuum energy goes into radiation . in this case
@xmath80 is close to the energy scale of inflation . unlike the dynamics during inflation , the dynamics of reheating
is extremely model dependent . here
we shall not come into detailed physics ( see @xcite for a review ) .
instead we only outline the generic class of models : * _ perturbative decay : _ in this case , the decay is characterized by a decay rate @xmath113 in particle physics @xcite .
a typical scenario of this kind has the inflaton oscillate around its minima of potential after inflation . and
when the duration of oscillation is longer than @xmath114 , the inflaton energy decays into other particles .
note that the duration of oscillation is approximately @xmath115 at the decay time .
thus the inflaton decay happens on a uniform hubble ( and thus uniform total energy density ) hypersurface in spacetime .
+ in this case the reheating temperature can be written as @xmath116 . in case of a long time of oscillation
, the hubble parameter at the inflaton decay can be significantly lower than the inflationary hubble parameter . * _ non - perturbative decay : _ this case is model dependent . for example , brane inflation typically end by collision of brane and anti - brane @xcite .
the decay is very fast due to the tachyon mode between the branes .
the decay happens on a uniform inflaton hypersurface in spacetime .
this situation is illustrated in fig .
[ fig : brane - inf ] . + brane inflation and reheating .
when the brane moves close to the stack of anti - branes , the open string mode with end point on both brane and anti - brane become tachyonic , which triggers the decay of the brane .
brane inflation ends on a uniform inflaton hypersurface.,scaledwidth=60.0% ] * _ parametric resonance ( preheating ) : _ when inflation ends and the inflaton is oscillating around its minima , the oscillation may trigger a resonance for other particles ( say , @xmath117 ) @xcite . in the resonance band ,
the energy density of @xmath117 grows exponentially until @xmath117 has comparable energy density with the inflaton , after which the system becomes non - linear while the decay continues .
note that during reheating , the small scale physics changes .
however those scales are so small that the same comoving scale today is not cosmological .
moreover , those sub - hubble scales shall not go super - hubble again thus there is no associated conserved quantity .
thus it is harder to probe the physics of reheating than to probe the physics during inflation .
nevertheless , there do exist a few possibilities that can constraint ( and in the future may observe ) scenarios of reheating : * _ modification of large scale fluctuations : _ as we shall show later in sec . [
sec : non - triv - rehe ] and [ sec : modulated - reheating ] , in some special models of reheating , not only the small scale perturbations but also the large scale ones can get modified . *
_ gravitational waves : _ the gravitational waves propagate freely after they are produced .
thus the primordial ones carry information of extremely early ages @xcite .
unfortunately the typical energy scale of reheating is yet smaller than the currently generation of gravitational wave detectors .
nevertheless , a low scale reheating or a table top detector may uncover the physics of reheating . we have discussed the dynamics of inflation , and its success . before going on to discuss the further success of inflation at the level of perturbations , let us pause and mention a few open questions for inflation .
* _ the @xmath100-problem _ @xcite : as we discussed , for inflation to happen , we need @xmath103 . on the other hand , for inflation to happen for enough e - folds to solve any problem , we need @xmath118 . for example , consider a free massive field as the inflaton . in this case , @xmath118 implies @xmath119 .
+ note that @xmath80 is a remarkably low energy scale : the inflationary @xmath80 is high in our today s high energy physics point of view .
however , during inflation @xmath120 is the size of the whole `` observable '' universe at that time .
thus it is nontrivial that we need an inflaton , whose compton wave length is much greater than the size of observable universe , during the era of inflation .
+ even worse , the mass of a scalar field is not protected by quantum fluctuations . like in the case of the higgs mass hierarchy problem , the quantum correction to the inflaton mass runs to the uv completion scale @xmath121 of the theory . even if the inflaton field only has gravitational self coupling ( or coupling to other fields ) , the mass correction , by power counting , .
thus there should be the @xmath122 factor to match the overall mass dimension .
] is @xmath123 when @xmath124 , the mass correction runs directly to planck scale . + as in the case of higgs physics , we may expect supersymmetry to save the theory from serve fine tuning .
however , the supersymmetry , if exists , has to be broken at the hubble scale .
thus the cancellation leaves a mass correction @xmath125 .
+ also , a locally supersymmetric theory typically get mass corrections of order @xmath125 .
for example , consider a kahler potential @xmath126 , where @xmath127 is a number of order one .
the kahler potential appears in the lagrangian as @xmath128 thus an effective mass @xmath129 is generated .
+ there are also other sources to generate @xmath130 corrections . for example , no symmetry forbids the coupling @xmath131 in the action , where @xmath132 is the ricci scalar . note that during inflation , @xmath133 .
thus coupling to ricci scalar generates a mass correction @xmath134 during inflation . + as another example , if @xmath72 couples to another field @xmath135 with @xmath136 coupling , the @xmath135 field gets a vev at the fluctuation level ( as we shall show in sec . [
sec : spectator - field - de ] ) : @xmath137 the above problems are collectively called the @xmath100-problem . to avoid this problem , one could introduce a shift symmetry to protect the mass of the inflaton .
nevertheless the shift symmetry has to be broken softly to have inflation @xcite . * _ the degeneracy problem : _ this is a problem at the opposite direction of the @xmath100-problem .
as we discussed , it is hard to get a flat enough potential , because of the @xmath100-problem .
however , if we allow some fine tuning to flatten the potential , there are immediately a huge number of inflation models .
we shall review those models in later sections . + as we shall discuss later , the number of observables of inflation is limited .
thus it is considered very hard to break the degeneracy , to select the correct model of inflation from observations . *
_ the trans - planckian problem _ @xcite : comoving scales expand exponentially during inflation . thus those scales are exponentially small in the past . especially those scales can be smaller than planck scale at the initial time .
how can we trust a theory with such small energy scales ? especially how to set the trans - planckian initial condition ? + it is argued that the vacuum state is adiabatic on sub - hubble scales , and thus does not have this scaling problem . yet
it is useful to show explicitly if there is a problem or not .
+ one should also note that at the level of free theory , the planck mass @xmath138 has no physical meaning and can be rescaled away .
only when we consider interactions , @xmath138 becomes physical as the energy scale where gravity is strongly coupled .
thus the trans - planckian problem is really a problem of interaction . and
at the interacting level ( which is by definition non - linear ) , there can be trans - planckian modes interacting with the super - hubble modes as well .
this situation is illustrated in fig .
[ fig : trans - planckian ] .
+ the trans - planckian problem , as a problem of non - linear interactions.,scaledwidth=60.0% ] * _ initial condition of inflation : _ inflation can not be the full story of the very early universe .
the inflationary frw metric only covers half of the de sitter spacetime . as a result , inflation is not geodesic complete ( a more general proof can be found in @xcite ) . we need a start for inflation .
this situation is illustrated in fig .
[ fig : desitter ] .
it is an open question how to prepare a homogeneous and isotropic hubble patch , for inflation to happen .
+ the frw patch is not geodesic complete .
it covers only half of the global de sitter space . ,
scaledwidth=30.0% ] + there are some candidates for the theories before inflation . for example eternal inflation @xcite ( which yet needs an initial condition ) , or the quantum creation of the universe @xcite .
so far we are still unable to experimentally verify those theories .
from now on , we shall move on to the cosmic perturbation theory .
this is the field where inflation has already shown its predictive power , and continues to excite the community with forthcoming experiments .
now we move on to the perturbations .
perturbations are by definition smaller than background . but for inflation ( and many other cases in cosmology ) they are by no means less important . because a perfectly homogeneous and isotropic universe do not tell us a lot ( and can not find anybody to tell ) .
to warm up , we first consider a spectator field in de sitter space ( @xmath139 ) @xcite .
`` spectator '' means gravity is not dynamical . instead , gravity is just an expanding playground and do not receive back - reaction from the spectator field .
operationally , one can consider a universe with a cosmological constant @xmath121 . then take @xmath140 while keeping @xmath141 fixed .
then a finite energy density scalar field will be a spectator field . for simplicity ,
we consider the spectator field to be a real scalar , non - interacting and massless .
the action of the spectator field is @xmath142~.\end{aligned}\ ] ] in fourier space , the above action is @xmath143~.\end{aligned}\ ] ] the @xmath135 field can be quantized as @xmath144 where @xmath145 and its hermitian conjugate @xmath146 are time - independent creation and annihilation operators : @xmath147 = 0 , \qquad [ a_\mathbf{k } , a^\dagger_\mathbf{k ' } ] = ( 2\pi)^3 \delta^3(\mathbf{k}-\mathbf{k}')~,\end{aligned}\ ] ] and @xmath148 satisfies the classical equation of motion of @xmath135 : @xmath149 to solve this equation , it is convenient to use conformal time .
the same equation of motion now becomes @xmath150 where prime denotes derivative with respect to the conformal time .
note that @xmath151 for de sitter space .
thus @xmath152 .
the above equation has simple solution @xmath153 now we have to fix the integration constants @xmath154 and @xmath155 . before a careful treatment ,
let us first consider a quicker way : note that @xmath156 is the canonically normalized field .
thus when this @xmath157-mode is far inside the horizon ( @xmath158 ) , @xmath156 effectively lives in flat space and should have the same solution as in minkowski space . by comparing with the minkowski space solution @xmath159 , we get @xmath160 note that the mode function @xmath148 only coincide with the minkowski space solution at @xmath158 . when @xmath161 , the solution of @xmath148 becomes completely different from the minkowski one .
now we shall determine the above integration constants in a more careful way .
there are two observations to determine @xmath154 and @xmath155 : * _ quantization and uncertainty .
_ the @xmath135 field is quantized in eq .
( [ eq : quant - a ] ) .
note that in quantum mechanics , quantization and uncertainty principle have close connections .
the size of quanta determines how accurate or inaccurate we could measure a system . + in our situation , the uncertainty principle is @xmath162 = i ( 2\pi)^3 \delta^3(\mathbf{k}-\mathbf{k}')~.\end{aligned}\ ] ] to compare this relation with ( [ eq : quant - a ] ) , we get a normalization condition @xmath163 note that we have rewritten the time derivative in terms of conformal time . applying the equation of motion for @xmath148 , we obtain @xmath164 * _ vacuum has lowest energy .
_ now we are to fix a `` vacuum '' state .
there are two things to do to fix the state : + first , we should relate the vacuum to annihilation operator .
a `` vacuum '' is a state satisfying @xmath165 however , this condition is actually a precise definition of @xmath145 , among all the possible definitions .
more importantly , we still need to know what _ is _ a vacuum as a physical state .
we define the vacuum as the lowest energy state at initial time @xmath158 . to make it more precise , the hamiltonian ( defined with respect to @xmath31 ) is @xmath166 = \frac{1}{2 } a^2 ( c_1c_1^*+c_2c_2^ * ) ( k^2+k^4\tau^2 ) + \frac{1}{2 } \left [ a^2 c_1c_2^*e^{-2ik\tau}(k^2 + 2ik^3\tau)+ \mathrm{c.c . } \right]~,\end{aligned}\ ] ] where @xmath167 denotes complex conjugate . at @xmath158 ,
the hamiltonian is @xmath168 we want this hamiltonian to be minimal with the constraint ( [ eq : vac - cond - from - quanta ] ) .
thus we get @xmath160 note that only the amplitude of @xmath154 is determined .
the phase of @xmath154 is still not determined .
this phase is a random phase , reflecting the quantum mechanical nature of the fluctuations .
eventually this random phase shows up , for example , on the cmb sky as the randomness of temperature fluctuations . in statistical quantities such as the power spectrum
, this random phase do not appear .
+ to collect the result of quantization , we have @xmath169 + so far we have discussed two ways to fix the vacuum state : instant minkowski vacuum , as well as the initial lowest energy state .
this vacuum state is known as the bunch - davies vacuum @xcite . the same vacuum state can also be obtained in various other ways .
for example , the bunch - davies vacuum can be obtained by the no boundary wave function of the universe .
this is because the euclidean version of initial time , @xmath170 , is a regular point on the sphere , thus should nt result in a divergence in the euclidean action .
alternatively , it is also believed that the bunch - davies vacuum is an attractor solution of inflation .
quite general ( but not all ) initial states are eventually driven to the bunch - davies state due to the cosmic expansion . before ending the discussion of vacuum , some more comments are in order here : * _ the relation between eq .
( [ eq : annihilate - vacuum ] ) and minimalizing the hamiltonian : _ at first glance , they look like the same condition .
this is because , in flat space qft , @xmath171 annihilate one particle .
so eq . ( [ eq : annihilate - vacuum ] ) means there is no particle , and thus is lowest energy state . to relate @xmath171 to the concept of `` annihilate one particle ''
] however , on time dependent background , things become more complicated . + on a time dependent background , matter energy is not globally conserved .
a `` empty '' state without particle could evolve into a state with lots of particles in it . note that we work in the heisenberg picture .
once we have defined the `` vacuum '' state , the state does not evolve with time .
however once we define the vacuum state as the lowest energy state at one time , the same state could be no longer the lowest energy state at another time , due to the time dependence of the matter hamiltonian .
here , we minimalize the energy at @xmath158 , as one essential piece of vacuum selection . * _ the vacuum is selected in the uv : _ note that @xmath158 corresponds to the @xmath157-mode has almost zero wave - length and infinite energy .
thus in this procedure , the vacuum is selected in the uv .
the @xmath172 limit can not be reached in qft because qft is considered as an effective field theory with a uv completion .
the uv completion scale is at most planck scale .
as we have discussed , this is the `` trans - planckian '' problem of inflation .
one might expect the selection of initial state at @xmath172 could be relaxed to a selection of lowest energy state at @xmath173 , as long as @xmath174 . before @xmath175 , the universe can be considered as expanding adiabatically and no particle is produced . * _ the vacuum is selected linearly mode by mode : _ we first linearized the theory , and consider one @xmath157-mode without worrying about other modes .
in other words , we implicitly set all other @xmath157-modes to zero .
in flat space qft , we can safely do that because in scattering process we indeed have set other modes to zero , at least locally near the incoming wave packet .
however , in a time dependent background , the situation is actually not clear .
the perturbations of other wave length , especially longer wave length exist and we can not just forget about them
. the inclusion of those modes will give a contribution at higher order in perturbation theory ( fortunately not at the leading order ) .
now we have determined the vacuum state , using a simple matching together with a more detailed discussion .
the next step is to interpret and understand the solution .
suppose we could measure the scalar field @xmath135 after the end of inflation , what shall we see ?
we see random fluctuations and we want to extract statistical information from it .
the way to do it is to consider the two point correlation function @xmath176 at @xmath177 ( the end of inflation ) . inserting the solution of the mode function
, we have @xmath178 where @xmath179 is called the power spectrum .
@xmath180 is the fluctuation amplitude of @xmath135 per hubble volume per hubble time .
when @xmath179 does not dependent on @xmath157 ( as the current case ) , we call the power spectrum `` scale invariant '' . the scale invariance of a power spectrum means if we zoom the fluctuations into a bigger volume , and compare with the true fluctuations in this bigger volume , statistically we can not find difference between those fluctuations .
the `` zoom '' operation shows up in the above formula as the @xmath181 factor , as a normalization of the volume .
note that in our present example , the dynamics of the perturbation is linear .
linearity implies that all the odd - point correlation functions ( e.g. @xmath182 ) vanishes , and all the even - point correlation functions ( e.g. @xmath183 ) can be reduced trivially into their disconnect parts ( e.g. @xmath184permutations ) .
thus the power spectrum carries all the statistical information about the perturbations . considering the unpredictable nature of quantum random phase ,
@xmath179 is the unique function to connect theory and observations . as we shall show in later sections , in case of a non - linear theory ,
the above is no longer true and the predictions can become much richer .
the above discussion can be generalized to a massive spectator field . in the massive case ,
the above discussions can be directly generalized .
the equation of motion takes the form @xmath185 the solution , with the same method of vacuum choice , is ( up to a phase ) @xmath186 where @xmath187 is the hankel function of the first kind . in the @xmath188
limit the solution returns to the same minkowski limit as discussed in the massless case .
finally , we heuristically apply the above discussion of @xmath135 to the fluctuation of the inflaton .
this `` application '' must be considered with a _ big caution _ that , considering that inflaton is the field that drives inflation , the inflaton fluctuation couples to gravity dynamically .
thus it is dangerous and non - rigorous to assume the fluctuation of inflaton lives on a non - dynamical gravitational background .
we will leave those issues to the next subsections .
now we postulate that @xmath135 is the fluctuation of the inflaton field : @xmath189 where @xmath190 is small and has the same solution as @xmath135 .
return to the intuitive picture of @xmath72 as a clock . during inflation , on super - hubble scales different parts of the universe temporarily lost casual contact with each other .
it is the clock field within the hubble patch , instead of the impact from other parts of the universe , that controls the reheat time of the local universe .
when @xmath72 has a fluctuation @xmath191 , @xmath191 brings an uncertainty to the clock .
the time uncertainty , measured in hubble unit , is @xmath192 .
this time difference may be conserved , considering different patches of the universe do not talk to each other .
if we consider this @xmath193 as an observable , the power spectrum of @xmath193 is @xmath194 we shall see in the next sections how the above argument could be made rigorous . in the last section , we considered a spectator scalar field . in that case
we could operationally take @xmath140 keeping @xmath195 to be small .
however , this limit can not be applied to the inflaton .
this is because from the friedmann equation , the inflaton energy density is exactly @xmath196 .
thus gravity may not decouple . to an effective cosmological constant .
it is the time variation of @xmath197 that invalids decoupling of gravity .
thus one can expect the gravitational corrections ( per e - fold ) to the inflaton are slow roll suppressed , compared with the case of a spectator field in de sitter space . as we shall show
it is indeed the case . ] to consider the inflationary perturbations , we need to perturb gravity as well .
it is not enough to promote the scale factor @xmath198 in the frw metric to be spacetime dependent .
instead , we have to consider the full metric , with the full dynamics of gravity .
the dynamics of gravity is governed by the einstein - hilbert action @xmath199 to calculate it explicitly in terms of gravitational perturbations , we have to write the metric into an explicit form . instead of writing @xmath82 in terms of @xmath200 ,
it is convenient to write the metric in the adm form @xcite , without loss of generality : @xmath201 the convenience that the adm form of the metric brings include : the adm decomposition of the metric .
the green arrows are the components of the metric , namely , lapse of time @xmath202 , a shift of hypersurface due to the time flow @xmath203 and the spatial distance @xmath204 .
the distance @xmath205 is first measured by 3-metric @xmath206 , and then together with @xmath202 , makes @xmath207 .
if @xmath208 , the time part themselves sums up to @xmath209 , along the flow direction of time.,scaledwidth=80.0% ] * _ clear physical meaning . _
the meaning of each component is demonstrated in fig .
[ fig : adm - metric ] .
* _ convenience of calculation .
_ various quantities can be calculated easily using the adm decomposition . for example , the determinant of the metric is @xmath210 and the inverse metric can be written as @xmath211 where @xmath212 , and @xmath213 is the inverse matrix of @xmath206 ( note that in general @xmath214 ) .
* _ decomposition of the ricci scalar . _
the 4-dimensional ricci scalar is much more difficult to calculate than the 3-dimensional one .
the adm decomposition provides a connection between the 3- and 4-dimensional ricci scalars .
the derivation can be found in e.g. @xcite .
here we outline the results : + one can construct normal of the hypersuface as @xmath215 the extrinsic curvature of the hypersurface can be written as @xmath216 this @xmath217 can be calculated explicitly as @xmath218 ~,\end{aligned}\ ] ] where @xmath219 is the 3-dimensional covariant derivative , defined from @xmath206 .
+ the 3- and 4-dimensional ricci scalars are related by @xmath220 note that there is a total divergence term in the rhs . for einstein gravity , this term is a boundary term . in the case
there is no spatial or temporal boundary , this term can be dropped . *
_ constraints easily solvable . _
note that in the above setup , the einstein - hilbert action has no derivative acting on @xmath221 , and no time derivative acting on @xmath222 .
thus @xmath221 and @xmath222 are constraints and easily solvable .
the disappearance of derivatives on @xmath221 and @xmath222 are not accidental .
the reason is that gravity can be viewed as a gauge theory and there are constraints to be solved .
this can be also understood from fig .
[ fig : adm - metric ] : @xmath221 and @xmath222 only tells how to define ( instead of dynamically evolve ) the spatial hypersurface at the next instant of time .
thus they should be non - dynamical .
one can still decompose the adm metric further .
note that the 3-dimensional rotational symmetry is explicitly preserved during inflation .
thus it is convenient to decompose the metric components according to the irreducible representations of the rotational group .
the relevant representations are 3d scalar , 3d vector and 3d tensor representations .
thus it is convenient to decompose the metric into those components . at linear level ( i.e. calculation of two point function ) , those components do not mix with each other , because different representations are orthogonal to each other . for this purpose ,
we decompose part defined here may be not very convenient .
see maldacena s definition . ]
@xmath223 ~.\end{aligned}\ ] ] here the vector sector contains @xmath224 and @xmath225 , satisfying @xmath226 the tensor sector contains @xmath227 , satisfying @xmath228 the scalar sector contains @xmath221 , @xmath229 , @xmath193 and @xmath230 .
the meaning of @xmath193 , @xmath230 and @xmath206 .
@xmath193 is an overall rescaling . @xmath230
rescale the points with respect to its original radius to the origin . @xmath206
rescale the points with respect to its own direction.,scaledwidth=70.0% ] from the orthogonal theorem in the theory of group representations , modes in different representations do not propagate into each other .
for example , we do need to consider the loop contribution from @xmath227 .
also , more - than - two - point functions can be mixed .
for example , @xmath231 .
] , @xmath232 .
the above scalar , vector , tensor perturbations need to be considered together with matter perturbations , for example , @xmath191 of the inflaton .
here we shall consider the scalar sector of the perturbations in detail .
the tensor sector looks similar to the scalar sector and we shall only discuss it briefly .
the vector sector decays in most inflation scenarios .
thus we shall neglect the vector sector here .
einstein s theory of gravity is invariant under general coordinate transformation .
the action ( and thus physics ) do not change under a change of coordinate @xmath233 the metric itself transform as a tensor under this coordinate transformation .
thus components in the metric , @xmath221 , @xmath229 , @xmath193 , @xmath234 , indeed change under this transformation . those transformations can be written collectively as @xmath235 scalar fields also change under the gauge transformation as @xmath236 those transformations can be grouped into two parts : the scalar part @xmath237 and @xmath238 , and the vector part @xmath239 , where @xmath240 those two parts transform the scalar sector and vector sector respectively .
the tensor sector is gauge invariant by itself .
for example , under a shift of time @xmath241 , and at linear order , @xmath193 and @xmath191 changes as @xmath242 consider a featureless spacetime with all those perturbations @xmath221 , @xmath229 , @xmath193 , @xmath234 being zero .
if we perform a gauge transformation on this spacetime , we find those perturbation quantities no longer zero .
thus there appears to be perturbations .
those apparent perturbations are called gauge artifacts , which need to be eliminated from the calculation . for example
, a gauge artifact @xmath243 brings pure background spacetime into @xmath244 there are two ways to eliminate those gauge artifacts .
one way is to use gauge - invariant formalism . in this formalism , the gauge artifacts cancels within the gauge - invariant quantities .
for example , at linear order , one can choose the gauge invariant quantity as the mukhanov - sasaki variable @xmath245 however , one need to note that free of gauge artifact do not necessarily mean the effect is observable .
we still need to identify theoretical observables with what one actually observe in experiments .
the situation is particularly dangerous for non - linear perturbations .
this is because at the non - linear level , the short wavelength fluctuations , which are accessible by local observers , can be mixed with long wavelength perturbations , which are not accessible by local observers ( but still gauge invariant ) @xcite .
a related issue with gauge invariant variable is that , at next to leading order , the gauge invariant quantities are non - local and non - unique @xcite .
here we follow a simpler way : to fix the gauge , and the fluctuations in the metric .
thus it is still convenient to consider an observer and consider the intuitive picture of the world understood by this observer . ] .
once the gauge is fixed , there are no longer available gauge transformations thus pure artifacts can not appear .
it is convenient to choose a gauge with @xmath246 .
we have yet another gauge condition to choose in the scalar sector .
this additional gauge condition fixes spatial slicing .
there are a few frequent choices .
they give the same result of observables . on the other hand
each gauge has pros and cons : * _ the @xmath191-gauge ( spatial flat gauge ) : _ we can choose to let the spatial part of the ( scalar sector ) metric to be flat : @xmath247 .
this gauge is convenient to discuss sub - hubble scale physics . because on small scales ( sub - hubble , but not as small as planck scale ) , gravity is weak .
thus it is intuitive to think about gravity as decoupled .
also , the calculation of the action is simplest in this gauge . * _ the @xmath193-gauge ( uniform inflaton gauge ) : _ we can choose to let the inflaton field have no fluctuation ) . ] : @xmath248 .
this gauge is a comoving gauge , in that the observer moves together with the cosmic fluid and measures no matter flux .
+ note that the three dimensional curvature is @xmath249 thus @xmath193 is the perturbation of @xmath250 .
for this reason , the @xmath193 variable in the @xmath193-gauge is known as the comoving curvature perturbation . +
this @xmath193-gauge is convenient to discuss the super - hubble scale physics . because on the one hand , gravity accumulates .
thus on large scales it is more intuitive to think about the perturbations as @xmath193 instead of @xmath191 . on the other hand ,
as we shall show later , @xmath193 is conserved in the @xmath193-gauge for single field inflation .
this provide a connection between the inflationary physics and current observations .
+ however , on small scales , @xmath193-gauge is not intuitive .
let s consider an analogue situation : a table has different height from the ground ( if we choose euclid coordinates ) . and in our room , gravity is not too strong .
however , from general covariance , we can choose a coordinate system where the table and the ground has the same coordinate height .
however , in this coordinate system gravity is extremely strongly coupled to cancel the same - height artifact . in the @xmath193-gauge , on small scales
the situation is similar : our intuition gets mixed up by a small physical effect plus a big gauge artifact ( this comment also applies for @xmath191-gauge on super - hubble scales ) .
+ technically , we shall indeed find that in the @xmath193-gauge , we have to do a lot of integrations by parts to cancel large numbers in the gravitational action .
* _ the newtonian gauge : _ we can choose to let the shift vector vanish : @xmath251 . in this gauge one can rewrite the scalar sector of the metric as @xmath252 here @xmath253 and @xmath254 are gravitational potentials .
the newtonian gauge allows us to think intuitively about how particle moves in the gravitational field : @xmath253 is the newtonian potential that non - relativistic particles move in newtonian mechanics .
relativistic particles could feel both @xmath253 and @xmath254 . in einstein gravity ,
constraint equations yield @xmath255 , and a departure is a smoking gun of modified gravity . for those reasons ,
the newtonian gauge is very convenient in analysis of the late time universe . * _ mixed gauges : _ one could even choose one gauge at one time , and another gauge at another time . as we discussed , the @xmath191-gauge is nice before and at the hubble - exit , while the @xmath193-gauge is nice after the hubble exit .
thus one can choose @xmath256 , and @xmath257 .
this can be accomplished by the partition of unitary .
there are infinitely differentiable functions which vanish at @xmath258 and equal to 1 at reheating , and vise versa .
for example , @xmath259 here @xmath260 when @xmath261 and @xmath262 for @xmath263 .
the metric @xmath264 takes @xmath191-gauge at hubble - crossing and takes @xmath193-gauge at reheating .
despite the complexity of gauge condition , we shall find the usage of this gauge when reviewing the @xmath265 formalism .
it is useful to note that , sometimes those intuitive gauge choices do not completely fix the gauge .
for example , in the @xmath193-gauge , there is still residue global symmetry , known as the large gauge transformation @xcite . for single field inflation , @xmath266 @xmath267 for multiple field inflation
, the change of other fields should be considered together .
one could further fix this residue gauge ( see , e.g. @xcite ) but we shall not need those further fixings at the moment .
now we are in the position to calculate the perturbations in the gravitational theory .
there are typically two ways to do this .
one way is to start from the classical einstein equation , and perturb the einstein equation to the linear order .
however , the cosmic perturbations are quantum in nature . as we observed in the spectator field case
, we need to know how to normalize the field canonically ( how large one quanta is ) .
thus it is better to start from the action . in the action approach ,
the first order action is just a variation principle of the background and thus vanishes . to get the linear equation of motion , we need the second order action , and the variation principle on top of that . we start from the @xmath191-gauge . in the @xmath191-gauge , we set @xmath247 and @xmath246 .
the perturbation variables in the scalar sector are @xmath191 , @xmath268 , @xmath229 defined as : @xmath269 the gravitational plus inflaton action is @xmath270 ~.\end{aligned}\ ] ] to calculate the ricci scalar @xmath132 by hand , one had better go to the adm formalism eq .
( [ eq : adm - r ] ) and drop the total derivative . or if a computer algebra system is used , it is technically easier to start directly from the 4-dimensional ricci scalar . in the adm case
the 2nd order action in momentum space takes the form @xmath271 ~.\end{aligned}\ ] ] in 4-dimensional ricci scalar case , the 2nd order action in momentum space takes the form @xmath272 ~.\end{aligned}\ ] ] the above two actions can be related by some integrations by parts , as desired from the boundary term of ( [ eq : adm - r ] ) . note that in the action we have not written the k - dependence of the perturbation variables explicitly .
those k - dependence should be understood as , e.g. @xmath273 .
the @xmath274 part follows from momentum conservation . from eq .
( [ eq : s2-adm ] ) , we observe that there is no time derivative acting on @xmath268 or @xmath229 .
this is as expected from the general principles we discussed when we introduce the adm formalism .
thus @xmath268 and @xmath229 can be solved : @xmath275 actually we shall not need the solution of @xmath229 when only the linear perturbations are considered . this is because eq .
( [ eq : s2-adm ] ) is linear in @xmath229 . thus inserting the solution of @xmath268 into eq .
( [ eq : s2-adm ] ) eliminates @xmath268 and @xmath229 together .
then after an integration by part to eliminate the @xmath276 term , we get the second order action @xmath277 \delta\phi^2 \right\ } ~.\end{aligned}\ ] ] a few comments are in order : * the 3rd term in eq .
( [ eq : s2-fg ] ) is a mass term for the perturbations .
the mass is @xmath278 . due to gravitational corrections
, this mass is different from the naively expected @xmath279 .
nevertheless they are of the same order . * note that this mass term is rather small . to see this , we recall that the energy scale for inflationary perturbations is @xmath80 . and @xmath280 .
* could this small mass be neglected ?
the answer is actually subtle . although @xmath281 , it may take effect on super - hubble scales . for the mode corresponding to the current cosmological scales
, the mass term takes effect for about 60 hubble times ( the observable e - folding number of inflation ) .
the same subtlety can be also observed in that , the @xmath282 term in the action would vanish exponentially on super - hubble scales , and eventually become comparable with the `` small '' mass term .
we for the time being neglect this mass term , and leave the subtleties to the next subsection .
( doubtfully ) neglecting the mass term , the discussion for the @xmath191 perturbations become completely parallel to the discussion of a massless spectator scalar field on de sitter background .
the power spectrum of the @xmath191 perturbations can be calculated as @xmath283 as we have discussed in the previous subsection , in the @xmath191-gauge , perturbations could evolve on super - hubble scales .
this invalids the naive slow roll approximation . actually the problem is more serious : inflation takes place at an extremely high energy scale . the physics from after inflation until tev scale is unknown .
how could we evolve the inflationary perturbations ( or any other observables produced by inflation ) through this unknown epoch and make predictions for inflation ?
fortunately , for single field inflation , the @xmath193-perturbation in the @xmath193-gauge is conserved on super - hubble scales .
we shall show the conservation at linear level in this subsection , and at non - linear level in the later sections .
it is the conserved perturbation on super - hubble scales , that brings connection between the very early time physics and very late time physics , after hubble - reentering . for this consideration ,
it is important to consider the perturbations in the @xmath193-gauge . here
we set @xmath248 and @xmath246 . inserting the perturbations into the gravitational plus inflaton action ( [ eq : single - field - action ] ) , we get using the adm ricci scalar : @xmath284 ~.\end{aligned}\ ] ] or from the 4-dimensional ricci scalar : @xmath285 ~.\end{aligned}\ ] ] again they are related by integration by parts .
the constraints can be solved as @xmath286 insert the constraints into the 2nd order action , and do some integration by parts to eliminate cross terms .
eventually one finds the action to be in a very simple form @xmath287 note that unlike the case of @xmath191-gauge , now there is no mass term for the @xmath193-gauge perturbations .
thus the awkward super - hubble evolution problem no longer exists in the @xmath193-gauge .
there is a decaying mode and a constant mode for @xmath193 .
after a few e - folds after hubble exit , the decaying mode goes away and @xmath193 becomes conserved .
we shall verify this conservation argument explicitly at the end of this section .
it is useful to note two facts : * no slow roll approximation is assumed in the above derivation .
thus the above action applies even @xmath288 . * in general , even if the cosmic fluid is not a scalar field , as long as @xmath289 , one can choose the @xmath193-gauge .
thus the conservation of @xmath193 should hold after inflation ( with the assumption of @xmath289 ) .
this answers the question at the beginning of this subsection : the conservation of @xmath193 makes inflation predictive .
now we are in the place to calculate the quantum perturbations of @xmath193 .
the equation of motion of @xmath193 is @xmath290 to the lowest order of slow roll approximation , the @xmath108 factor in the above equation can be ignored and approximately @xmath291 .
the solution of the above equation , as a function of conformal time , is @xmath292 in terms of the quantized fields , the above solution is equivalent to @xmath293 as in the case of a scalar field in de sitter space , the `` negative frequency '' mode is disregarded and the constant factor is determined from the consistency of quantization together with the assumption that the initial state of the k - mode has lowest energy . before closing this subsection
, we would like to verify the conservation of @xmath193 on super - hubble scales , without the reference of slow roll or inflationary background .
note that is an exact equation without making slow roll approximation . on super - hubble scales , @xmath294
thus the @xmath295 term can be neglected in .
thus the equation of motion can be approximated as @xmath296 considering the exponential growth of @xmath1 , this equation is exponentially accurate on super - hubble scales .
the solution is @xmath297h(t_2 ) dt_2 \right\ } ~.\end{aligned}\ ] ] thus there exists a constant mode @xmath154 .
the other mode is decaying as long as @xmath298 .
this inequality is satisfied by most of the cosmological conditions ( note that @xmath299 for a constant equation of state parameter @xmath35 , and @xmath300 for slow roll inflation ) .
, @xmath193 finds itself in an effectively contracting universe , although the real metric is expanding . ] similarly to the case of massless scalar field , the power spectrum of comoving curvature perturbation can be defined as @xmath301 the observational value of the power spectrum is @xmath302 .
this value is the power spectrum on large scales , spanning between the scale of the whole observable universe , and scale about @xmath303 smaller .
on the other hand , there are constraints that the power spectrum should be perturbative on small scales all the way down to the comoving hubble scale at the end of inflation , by not observing the primordial black holes ( see @xcite for a review , and the references therein ) formed by those small scale fluctuations . ]
is known as the cobe normalization @xcite .
sometimes one also use @xmath304 defined as @xmath305 .
again , the power spectrum does not depend on @xmath157 at the lowest order .
thus the inflationary perturbations are nearly scale invariant .
although the power spectrum is nearly scale invariant , it is still important to consider the small non - scale - invariant correction to the power spectrum
. this correction is called the tilt of the spectrum , or the spectral index . according to the wmap result ,
the exact scale invariant power spectrum is almost ruled out and the derivation from scale invariance should be a few percents . to describe the scale dependence , the spectral index @xmath306 is defined as @xmath307 to calculate this quantity , it is helpful to have a closer look at the equation of motion ( [ eq : zeta - eom ] ) .
there are two ways to take the scale dependence into account in ( [ eq : zeta - eom ] ) .
the most straightforward way is to consider the small derivation from @xmath291 together with no longer neglecting @xmath108 .
the equation of motion can be solved in this case and a derivation from scale invariance is observed @xcite . on the other hand
, there is an easier way to calculate the scale dependence .
weak scale dependence means that the parameters in ( [ eq : zeta - eom ] ) would vary by a small amount per hubble time . within a few e - folds near the hubble crossing , eq .
( [ eq : zeta - eom ] ) should still be applicable , with @xmath80 considered as a constant and @xmath108 neglected . a few e - folds before the hubble crossing the solution is linked with the instant minkowski solution , and a few e - folds after the hubble crossing the solution is linked with the conserved quantity .
both the sub - hubble instant minkowski solution and the super - hubble conserved quantity are not slow roll approximated but instead exact result . in other words , the prescription should be to take the hubble parameter in the equation of motion as a function of momentum @xmath157 : @xmath308 , and calculate the value of @xmath80 at hubble crossing @xmath258 . thus up to higher order
slow roll corrections , we have @xmath309 . the spectral index can be calculated as @xmath310 the observational constraint for @xmath311 is @xcite @xmath312 thus although inflation is nearly scale invariant , an exact scale invariance with @xmath313 is ruled out at over 5@xmath135 .
one can also define the running of spectral index , as @xmath314 for the above single field inflation model we are considering , we have @xmath315 thus @xmath316 is one order smaller than @xmath311 in terms of slow roll parameters in the above simple model .
the observational constraint for @xmath316 is @xcite @xmath317 as we discussed at the end of sec .
[ sec:2nd - order - grav ] , on super - hubble scales the equation of motion simplify .
actually , the @xmath318 parameter itself can be used as an expansion parameter .
one can set up perturbation theory with respect to @xmath318 , which could be exact to all orders of @xmath193 or the slow roll parameter .
this method is called the gradient expansion .
there is an intuitive way to view the gradient expansion , at least at the leading order @xmath319 . during inflation ,
the universe broken up into exponentially many `` local universes '' , which temporarily can not talk to each other .
each local universe is quite homogeneous and isotropic and thus can be locally approximated as a frw universe .
the local properties of those local frw universes have no difference between one another .
however , when we compare those local universes in spacetime , there could be a difference of expansion between them at a fixed time , or a difference of time at a fixed rate of expansion .
this difference can be produced by the quantum fluctuations .
for example , in one local universe , if there is a quantum fluctuation to drive the inflaton a little bit up on the potential , this local universe is then driven farther away from reheating .
as we shall show , this difference is actually the comoving curvature perturbation .
this situation is illustrated in fig .
[ fig : deltan ] . time difference between local frw universes .
after reheating , the universe become radiation dominated and the energy density dilutes quickly with the expansion of the universe .
the change of color illustrate the drop of energy density.,scaledwidth=70.0% ] it is useful to make this separate universe idea explicit .
there are a few ways to do this and here we follow the method of @xcite .
consider a local universe with scale factor @xmath320 where @xmath321 is almost a constant on scales comparable with the hubble scale , and only change significantly on super - hubble scales .
the local hubble expansion can be defined as @xmath322 .
the continuity equation is @xmath323 in terms of @xmath1 and @xmath193 , this equation can be written as @xmath324 so far we havent specified any gauge yet .
now we shall look at this equation in the uniform energy density gauge .
( it can be shown that on large scales this gauge is equivalent to the comoving gauge . )
if there is no entropy perturbation , i.e. @xmath289 , then in eq . only @xmath325 is a perturbation variable .
the precise definition and meaning of entropy perturbation will be discussed in sec .
[ sec : curv - isoc - pert ] .
all other variables are background variables , independent of @xmath33 .
considering a perturbation variable should not have zero mode , we have @xmath326 for yet another time we confirmed the conservation of comoving curvature perturbation on super - hubble scales . and
this time the confirmation is exact for all orders of perturbations in @xmath193 ( but assuming that @xmath193 is classical , both in external lines and in loops .
for a quantum treatment , see @xcite ) .
now let us make connection between @xmath193 and the perturbation of e - folding number .
note that this part is independent to the above discussion , in the sense that we do not assume @xmath289 .
thus the result also applies for a time dependent @xmath193 on super - hubble scales .
the connection is intuitive because @xmath193 is the perturbation of scale factor , and on the exponent . to see this connection
more precisely , we note that the e - folding number of a local frw universe can be written as @xmath327 + \zeta(t_2,\mathbf{x } ) - \zeta(t_1,\mathbf{x})~.\end{aligned}\ ] ] again this equation does not dependent on gauge choice .
now we choose a mixed gauge : at an initial time @xmath328 , @xmath329 ( flat slice ) , and at a final time @xmath330 and later times , @xmath331 ( uniform energy density slice ) . in this mixed gauge
, we have @xmath332 where @xmath265 is the e - folding number difference between the initial flat slice and the final uniform energy density slice .
this relation is known as the @xmath265-formalism @xcite . finally ,
if needed , the gauge - invariance can be restored by adding back the gauged counter - part of : @xmath333 where @xmath334 is the background energy density , which equals to the energy density on uniform energy density slice .
@xmath9 is the energy density without choosing a gauge . as a side remark ,
the above discussions could actually be made stronger .
consider there are @xmath335 components of fluids ( say , @xmath336 , @xmath337 ) in the universe .
assuming those components do not exchange energy between each other .
then the energy of those components are conserved separately . in this case
we have @xmath335 conserved quantities : one can define @xmath338 as the @xmath265 on uniform @xmath336 slices , respectively .
the above discussion holds component - wise .
the @xmath265-formalism is convenient in several aspects : * it is straightforward to calculate the inflaton fluctuation on the initial flat slice . as we have shown , the inflaton fluctuation at hubble crossing is @xmath339 plus calculable slow roll corrections . *
the comoving curvature perturbation is conserved after the final uniform energy density slice , as we have shown . * for single field inflation ,
the uniform energy density slice is the reheating surface .
thus the @xmath265 defined in the above way does coincide with the e - folding number difference during inflation . *
the @xmath265-formalism is non - linear in orders of @xmath193 .
thus the formalism is still convenient to use for non - linear perturbations .
the convenience can be observed by a sample calculation . during inflation
, the e - folding number can be calculated as @xmath340 .
a variation of this equation is @xmath341 thus @xmath342 this agrees with the power spectrum we have calculated in
. however , there is price for the simplicity .
one has to pay attention to the assumptions of the above derivation , and be careful at the situations where those assumptions break down .
for example , * _ non - conservation of @xmath193_. when there are entropy perturbations , @xmath8 is no longer a function of @xmath9 thus @xmath193 may not be conserved on super - hubble scales . * _ non - trivial reheating surface_. when there are entropy perturbations , the reheating surface may not be the uniform total energy density slice . in this case @xmath265 no longer has the physical meaning of e - folding number difference of inflation . instead , a correction term coming from the difference between uniform energy density slice and the reheating surface must be considered . * _ nontrivial initial correlation functions_. if the power spectrum or higher order perturbations of @xmath191 is modified at hubble - crossing , the modification needs to be considered as an input of the @xmath265-formalism . *
_ ignorance of the decaying mode_. the requirement @xmath289 is not the same as the requirement of single field .
this is because a scalar field has second order equation of motion and thus two modes . on super - hubble scales
one mode decays away thus @xmath289 .
however , in case both modes needs to be considered for some reason , one need to examine if @xmath289 still hold . * _ validity of local frw metric_. this is typically not a problem for inflation . because locally , non - frw energy components are diluted .
however , for alternative models to inflation , this is not necessarily true . * _ break down of classicality_. at the non - linear level , quantum fluctuations on sub - hubble scales may interact with the super - hubble fluctuation .
@xmath265-formalism is classical and do not carry this quantum nature . finally , let us make one more clarification of the gradient expansion method and the @xmath265-formalism : there exist sub - hubble fluctuations .
when considering local frw universes , why we neglect those sub - hubble fluctuations ? actually , those sub - hubble fluctuations , as high energy modes in a general qft consideration , can be integrated out . as a result ,
the background equation of motion for the inflaton acquires a stochastic source term @xcite : @xmath343 where @xmath344 is a gaussian noise obeying @xmath345 this is exactly the origin of the @xmath339 fluctuation at hubble crossing .
so far we have been talking about the observable stage of inflation .
when calculating the dynamics of inflationary perturbations , we have been using the lowest order slow roll approximation .
especially , when calculating the equation of motion for the perturbations , the mode functions are the same as that obtained in ( the frw patch of ) de sitter space ( see our toy model of spectator field ) .
however , despite of the approximations , we should be careful in thinking about the observable stage of inflation as de sitter space .
for example , one note that the power spectrum of comoving curvature perturbation is @xmath346 .
when @xmath347 with @xmath80 fixed finite , the power spectrum diverges .
this is very different from the observed power spectrum @xmath348 .
the problem not only arises as the divergence in the number .
but also , @xmath193 ( or other observables stands for density perturbation of inflation ) manifests itself only after the end of inflation and after the perturbation modes returns to the horizon .
the exponential expansion of de sitter space never end thus one can not observe the density perturbations .
even worse , in de sitter space , the `` super - hubble '' density perturbations are not only non - observable .
the whole semi - classical picture of cosmic perturbation theory may break down in the de sitter limit .
it is argued that , like the black hole complementary @xcite , de sitter space may have complementary as well @xcite .
consider two comoving observers a and b , who start near each other with distance nearer than hubble . because of cosmic expansion , at a later time , @xmath349 will exit the horizon of @xmath350 .
the horizon exit process of @xmath349 , as seen by @xmath350 , is a disaster of @xmath349 .
the observer @xmath350 measures the gibbons - hawking radiation with a temperature @xmath351 .
blue shift this temperature to the near horizon regime , @xmath350 actually finds @xmath349 destroyed by the gibbons - hawking radiation when approximating the de sitter horizon .
this picture is different from that experienced by @xmath349 , who finds nothing special because of the equivalence principle .
the different semi - classical pictures observed by @xmath350 and @xmath349 contradicts .
but this is not a problem as long as the observers no longer return to each other s horizon to compare notes .
applied to cosmic perturbations : for a local observer , the physical picture of perturbations in de sitter space should be that , the perturbations approaches the de sitter horizon , get burnt by the blue shifted gibbons - hawking radiation , and get emitted back by the horizon .
there is no other copy of information which actually `` exits '' the horizon .
thus the physical picture is different between the perturbations in a stage of observable inflation , and that in de sitter space .
the change of physical picture actually not only happens at the exact value @xmath352 , but takes place when @xmath107 becomes so small that @xmath353 .
note that the power spectrum can be also written as @xmath354 .
thus when @xmath353 we have @xmath355 the lhs is the classical motion of inflaton per hubble time , and the rhs is the quantum fluctuation per hubble time . thus the above equation can be interpreted as that , the quantum fluctuation of the inflaton dominates over the classical motion .
the classical motion of the inflaton always roll down the potential in the slow roll regime .
however , the quantum fluctuations has equal probability to fluctuate up or down .
after one hubble time , a hubble volume self - reproduces into @xmath356 copies . if in one or more of those @xmath357 new hubble volumes , the inflaton field has greater value due to the large quantum fluctuation , the dynamics in those patches are driven away from the reheating surface .
the self - reproduction continues and inflation will never globally end . this situation is known as eternal inflation @xcite . during eternal inflation , when following an observer s comoving world line , the observer finds that inflation eventually ends locally .
however , due to the lack of global end , the global structure of eternal inflation is analogue to that of de sitter space . as in the case of de sitter space ,
the semi - classical picture of cosmic perturbations break down during eternal inflation .
again this is not only because of @xmath358 , and not only because of lack of global end of inflation , but also that the whole picture can not be trusted . to see this , it is convenient to consider ( the slow roll analog of ) the static patch of de sitter space , instead of the frw patch . in the static patch , the metric can be written as @xmath359 the stretched horizon is defined as the surface with one more area unit compared with the classical horizon : @xmath360this definition is reasonable because stretched horizon @xmath361 , the gibbons - hawking temperature is the planck temperature .
the temperature has a redshift @xmath362setting @xmath363 , one get the @xmath361 in equation ( [ eq : stretched ] ) .
complementarity becomes important as soon as objects ( e.g. perturbation modes ) reach the stretched horizon .
now we want to estimate when complementarity becomes important for slow roll inflation .
roughly , we could use the result for pure de sitter space , and let @xmath80 vary with respect to time .
now the field excitation propagate towards the stretched horizon and the stretched horizon is also expanding itself .
one should compare which of the two goes faster .
the field excitation near the stretched horizon falls towards the stretched horizon at speed of light : @xmath364 measured by an observer at the center , the stretched horizon itself moves at a velocity ( in terms of @xmath365 and @xmath31 ) as @xmath366here one should note that @xmath367 is local process near the stretched horizon thus there is a huge redshift factor .
however , @xmath368 is measured at the center of static patch , thus @xmath369 when calculating @xmath368 . to compare those velocities , one note
@xmath370where @xmath371 is the power spectrum for curvature perturbation during inflation . for the observable stage of inflation , @xmath372 .
thus during the observable stage of inflation the field excitations would never reach the stretched horizon .
thus the apparent difference of the static patch and the frw patch is merely a coordinate transformation .
however , for eternal inflation , @xmath353 .
thus during eternal inflation the hubble radius is expanding so slowly that the field excitations could reach the stretched horizon .
on the other hand , even one believed the semi - classical picture of eternal inflation , the density perturbation generated during eternal inflation could never be observed even in principle .
this is because when the @xmath373 modes returns to the horizon , they form black holes .
the global structure of spacetime is not accessible by an observer in a black hole @xcite . to conclude this subsection ,
when taking the de sitter space limit , one hits eternal inflation and a shift of semi - classical physical picture .
the calculations of power spectrum and non - gaussianities are not affected ( although the de sitter approximation is used for the mode function ) , because the perturbations never reaches the stretched horizon .
the inflaton is not the unique light field during inflation .
instead , there is at least another light field
the gravitational waves @xcite ( for modern conventions , see , e.g. @xcite ) .
the calculation of gravitational waves goes similarly to the scalar sector .
one could either do a brute force calculation of the tensor part , or first calculate the ricci scalar on minkowski background and then do a conformal transformation into an inflationary background .
the result is @xmath374 note that the traceless and transverse conditions eliminates 4 components from the 6 components of a symmetric 3-tensor .
thus there are only 2 physical modes in @xmath375 .
those two modes can be decomposed by introducing the polarization tensors @xmath376~,\end{aligned}\ ] ] where the symmetric , traceless and transverse properties are carried over by @xmath377 and @xmath378 as @xmath379 further from the reality conditions we have @xmath380 and we have the freedom to normalize @xmath381 .
the normalization we take is @xmath382 finally , we need to fix the ambiguity of decomposition in eq .
( [ eq : tensor - dec ] ) . for this purpose , it is convenient to choose the two polarization tensors as the spin 2 eigenfunctions of rotation along the @xmath383-axis : @xmath384 those conditions fixes the function @xmath385 as a function of @xmath383 ( up to a phase ) .
it is helpful to construct those polarization tensors in an explicit way . to do that , we make use of the rotational symmetry of the theory to first rotate @xmath383 to be @xmath386
. then the polarization tensors are @xmath387 for a general @xmath157 , a 3-d rotation matrix is multiplied onto the above polarization tensors . with the above decomposition
, the action of the tensor mode becomes two decoupled components .
theoretically this is because they are different irreducible representations .
technically this is because @xmath388 .
the tensor mode action takes the form @xmath389~.\end{aligned}\ ] ] for each polarization mode , the action is the same as that of a massless spectator scalar field .
thus @xmath390 the tensor power spectrum is defined as @xmath391 compared with eq .
( [ eq : tensor - dec ] ) , we have @xmath392 to compare with the scalar power spectrum , a tensor to scalar ratio is defined and calculated as @xmath393 the observational constraint on @xmath365 is @xcite @xmath394 it is important to note that , unlike the scalar power spectrum ( [ eq : scalar - power ] ) , the tensor power spectrum ( [ eq : tensor - power ] ) does not depend on @xmath107 .
in other words , @xmath395 is solely determined by @xmath80 , the energy scale of the inflationary perturbations . thus if the gravitational wave power spectrum is detected , we immediately know the inflationary hubble scale . also , for the minimal single field inflation , the slow roll parameters @xmath105 and @xmath100 can be calculated as functions of observables @xmath306 and @xmath365 : @xmath396 thus measurements on @xmath306 and @xmath365 can be considered as measurements of the first two derivatives of the inflationary potential .
especially , the sign of @xmath100 determines if the inflationary potential is convex or concave .
the latter is favored by current data @xcite .
one can plot the single field inflation models ( as a function of e - folding number ) as line intervals on a two dimensional @xmath306-@xmath365 plot .
the spectral index of gravitational waves is @xmath397 note that both the very lhs and very rhs of the above equation are expressed in terms of observables .
thus @xmath398 is a consistency relation for single field inflation .
it is also interesting to note that , the tensor to scalar ratio is related to the distance that the inflaton rolled in field space during inflation .
the inflaton s rolling distance per e - fold is @xmath399 thus in 60 e - folds , @xmath72 rolls a distance @xmath400 . in words ,
large tensor to scalar ratio @xmath365 indicates a longer distance of inflaton s motion .
this relation is known as the lyth bound @xcite .
one can correspondingly classify inflation models into large field and small field models , with @xmath401 and @xmath402 respectively .
as we have seen , the construction of inflation requires two types of light fields : the inflaton field and the gravitational waves .
however , we know from qft that there are many other fields . what do other fields behave during inflation ? here we discuss several possibilities : * _ gauge fields : _ non - interacting gauge fields in 3 + 1 dimension have a conformal symmetry . the action of a non - interacting gauge field can be written as ( take u(1 ) as an example ) @xmath403 note that @xmath404 does not really contain metric or covariant derivative .
moreover , the scale factor dependence ( using conformal time ) in @xmath405 and @xmath406 cancels . thus the gauge field does not `` know '' it lives on an expanding background instead of on a minkowski background .
this is due to the conformally flat nature of the frw metric . as a result ,
the large scale fluctuations of the gauge field is suppressed by @xmath407 and thus not important or observable . + on the other hand , if for some reasons we want the gauge field to have large scale fluctuations , then we have to break the scale invariance of the field ( see , e.g. @xcite ) . * _ accidental , or other protected massless fields : _ in this case , our previous discussion of a spectator field in de sitter space applies . during each e - fold
, the field has a fluctuation @xmath408 .
thus on long time scales , the field has a random walk behavior @xmath409 .
+ however , one should note that , it is not easy to have a massless field without a symmetry .
this is because , without the protection from symmetry , the quantum corrections generates a mass for the field .
for the same reason as the higgs mass hierarchy problem , the quantum correction for a unprotected mass parameter runs to the uv completion scale of the effective field theory .
this generates a mass correction @xmath410 .
if there is supersymmetry ( which has to be broken at hubble scale ) , the mass correction is brought down to @xmath411 . * _ field with a small mass @xmath412 : _ the dynamics of such a light field depends on the field value . if the field is located at the origin of the potential , there is no much difference between the current case and the case of @xmath413 . however , as we discussed , because of the random walk behavior , the field can not naturally stay at the minimum of the potential if there were a large amount of inflation before the observable stage . to estimate the field value
, we note that there is a competition between the random walk and a classical roll - back of the field .
the classical roll - back can be estimated using the field equation @xmath414 we may approximate @xmath80 to be constant ( considering that @xmath135 is not the field that drives inflation , which modifies @xmath80 ) . in this case , @xmath135 can be solved as @xmath415 thus at large field value , the field rolls back exponentially fast and can not random walk . on the other hand ,
when the exponential factor is small , random walk dominates .
given enough e - folds of inflation ( including inflation happened at the non - observable stage ) , the balance is reached at @xmath416 during the above time ( corresponds to @xmath417 e - folds ) , the field randomly walks by a distance @xmath418 this is the typical distance for @xmath135 , away from the origin . on large scales , the very light field forms a curvaton web @xcite .
the dynamics of the curvaton web can be analyzed in detail using the langevin equation and the fokker - planck equation . if there were no enough e - folds of inflation , a cutoff of e - folding number needs also to be imposed .
+ at perturbation level , the present case share similar equations with the massless case .
however , due to the non - trivial background motion , the conversion mechanism from entropy perturbation to the comoving curvature perturbation could be very model dependent .
the definition of entropy perturbation together with the illustrations of mechanisms are discussed in sec .
[ sec : curv - isoc - pert ] . *
_ field with a small mass @xmath419 : _
+ this case is again different from the @xmath412 case .
this is because the @xmath135 field now rolls fast enough that it does affect the variation of the hubble parameter .
thus both fields contribute to the background evolution . in this case
we have multi - field inflation .
* _ field with a mass @xmath420 : _ in this case , the @xmath135 field is heavy compared with the background energy scale the inflaton mass .
however , @xmath135 is not heavier than the energy scale of perturbations the hubble scale .
thus the field stays at the origin of the potential at the level of homogeneous and isotropic background , but the perturbations could couple to the inflaton and thus observable . in this case , we have quasi - single field inflation .
+ note that if there is minimally broken supersymmetry during inflation , then the standard model fields should get a mass correction of order @xmath80 .
this is known as the `` @xmath100-problem '' for the inflaton @xcite , and opens up a variety of possibilities for other fields . * _ massive fields with @xmath421 : _ those fields decouple from the light fields at the tree level . however , considering that inflation is uv - sensitive , it is still an open question if loop corrections would make the imprint from those very massive fields observable in the sky @xcite .
as we have shown in the previous section , linear perturbations already carry interesting information about the physics during inflation
. however , the information that linear perturbation carries is quite limited .
this is because , for each field , the power spectrum is a function of @xmath157 only .
even worse , nearly scale invariance brings this function down to a few numbers with increasing difficulties to measure .
for example , consider the comoving curvature perturbation @xmath193 .
currently we have measured two numbers : the amplitude @xmath371 itself at a given @xmath157 , and the tilt @xmath311 .
if there were no features in the power spectrum , even a third number ( the running of spectral index @xmath316 ) would be very difficult to measure .
the gravitational wave sector is even worse .
even the amplitude @xmath422 is not detected yet , leaving only an upper bound .
if the above two numbers are the only things that one could get from inflation ( which actually comes directly from the nearly scale invariance of the background , not the detail dynamics ) , then it is very hard to distinguish between inflation and alternatives , or between inflation models , let alone to use inflation as an ultra high energy laboratory of physics .
one expects quite a few possibilities to overcome the above lack of prediction problem .
for example , a detection of tensor mode will be of much help .
we shall also discuss other possibilities such as features and entropy modes in later sections . here
, we consider another class of possibilities interactions , which produce non - gaussian fluctuations in the sky . to calculate the interactions , it is convenient to use the interaction picture
. let us start by reviewing how the hamiltonian formalism and the interaction picture are defined on a time dependent background @xcite .
given a theory , we have a hamiltonian @xmath423 $ ] dependence in the hamiltonian is a functional dependence , because the hamiltonian has an integration over @xmath33 . on the other hand ,
the hamiltonian inherits implicit time dependence from @xmath424 and @xmath425 . ] yet .
nevertheless , we shall mainly have the conformal time in mind thus we denote it as @xmath31 as we did in previous sections . ] , with equal time commutator @xmath426 = i \delta_{ab}\delta^3(\mathbf{x}-\mathbf{y})~ , \quad \left [ \phi_a(\mathbf{x},\tau),\phi_b(\mathbf{y},\tau ) \right ] = 0~ , \quad \left [ \pi_a(\mathbf{x},\tau),\pi_b(\mathbf{y},\tau ) \right ] = 0~,\end{aligned}\ ] ] where @xmath427 indices denote different fields or field components .
the equation of motion can be written as @xmath428~ , \qquad \pi_a ' = i \left [ h,\pi_a \right]~,\end{aligned}\ ] ] where as usual , prime denotes time derivative with respect to the time @xmath31 .
to do perturbation theory , we need to split @xmath429 and @xmath430 into background and perturbations .
the background is assumed to be c - number valued functions which satisfy the classical equation of motion . for our purpose , typically it is convenient to define the background as the homogeneous and isotropic part of the fields , which are the frw metric and the zero mode of the matter fields .
the perturbations @xmath431 and @xmath432 are define by @xmath433 the commutators for perturbations are not affected by the c - number valued background .
thus @xmath434 = i \delta_{ab}\delta^3(\mathbf{x}-\mathbf{y})~ , \quad \left [ \delta\phi_a(\mathbf{x},\tau),\delta\phi_b(\mathbf{y},\tau ) \right ] = 0~ , \quad \left [ \delta\pi_a(\mathbf{x},\tau),\delta\pi_b(\mathbf{y},\tau ) \right ] = 0~,\end{aligned}\ ] ] the hamiltonian can be expanded with respect to the perturbations as @xmath435 = h[\bar \phi , \bar \pi ] + \sum_a \int d^3x \left.\frac{\delta h[\phi,\pi]}{\delta\phi(\mathbf{x},\tau)}\right|_{\bar \phi , \bar \pi } \delta\phi(\mathbf{x},\tau ) + \sum_a \int d^3 x \left.\frac{\delta h[\phi,\pi]}{\delta\pi(\mathbf{x},\tau)}\right|_{\bar \phi , \bar \pi } \delta\pi(\mathbf{x},\tau ) + \widetilde h[\delta\phi,\delta\pi ; \tau ] ~,\end{aligned}\ ] ] where the terms which are linear in @xmath436 and @xmath437 vanish due to background equation of motion .
the term @xmath438 $ ] is the hamiltonian for perturbations , which is of second and higher orders of @xmath191 and @xmath439 .
note that @xmath438 $ ] is explicitly time dependent because now we write it as a functional of @xmath191 and @xmath439 , and the time dependence coming from @xmath440 and @xmath441 appears as an explicit time dependence .
one can verify the consistency of taking @xmath438 $ ] as the hamiltonian for perturbations , by calculating from eq .
( [ eq : eom - h - bg ] ) : @xmath442 \quad \rightarrow \quad & { \delta\phi}_a(\tau ) = u^{-1}(\tau,\tau_0 ) { \delta\phi}_a(\tau_0 ) u(\tau,\tau_0)~ , \nonumber \\ { \delta\pi}_a ' = i \left [ \widetilde h,\delta\pi_a \right ] \quad \rightarrow \quad & { \delta\pi}_a(\tau ) = u^{-1}(\tau,\tau_0 ) { \delta\pi}_a(\tau_0 ) u(\tau,\tau_0)~,\end{aligned}\ ] ] where @xmath443 is the time evolution operator , satisfying ) we can directly get @xmath444 u(\tau,\tau_0)~. $ ] and note that the canonical equations ( [ eq : time - evolution - from - h ] ) implies @xmath445 thus @xmath446 = \widetilde h[\delta\phi(\tau_0),\delta\pi(\tau_0 ) ; \tau ] $ ] .
this is how is derived .
later we shall use this trick again in . ]
@xmath447 u(\tau,\tau_0)~ , \qquad u(\tau_0,\tau_0)=1~.\end{aligned}\ ] ] the hamiltonian @xmath438 $ ] contains quadratic terms and higher order terms in @xmath191 and @xmath439 .
the quadratic terms result in linear equations of motion , which typically lead to simple solutions .
however , the non - quadratic terms in @xmath438 $ ] results in non - linear equations of motion , and in most cases can not be solved exactly .
perturbation theory is in position to solve this problem .
for this purpose , we split @xmath438 $ ] into @xmath448 = h_0[\delta\phi,\delta\pi ; \tau ] + h_i[\delta\phi,\delta\pi ; \tau]~,\end{aligned}\ ] ] where @xmath449 $ ] is the interaction hamiltonian.~ , \quad \sigma ' \equiv \pi_\sigma / a^2\end{aligned}\ ] ] we could solve @xmath438 $ ] directly , which is a straightforward generalization of @xmath450 case . or instead , when @xmath451 we could split @xmath438 $ ] into @xmath452 ~ , \quad h_i = \int d^3 x ~ \frac{1}{2 } a^2 \left [ k^2 ( c_s^2 - 1 ) \sigma^2 \right]~.\end{aligned}\ ] ] the two methods of calculation agree as they should at zeroth and 1st order of @xmath453 , with the usage of first order perturbation theory . ]
according to the split of @xmath454 and @xmath455 , we define the time evolution of the interaction picture fields @xmath456 as @xmath457 \quad \rightarrow \quad & { \delta\phi}^i_a(\tau ) = u_0^{-1}(\tau,\tau_0 ) { \delta\phi}^i_a(\tau_0 ) u_0(\tau,\tau_0)~ , \nonumber \\
{ \delta\pi}^i_a { } ' \equiv i \left [ h_0,\delta\pi^i_a \right ] \quad \rightarrow \quad & { \delta\pi}^i_a(\tau ) = u_0^{-1}(\tau,\tau_0 ) { \delta\pi}^i_a(\tau_0 ) u_0(\tau,\tau_0)~,\end{aligned}\ ] ] where @xmath458 is the time evolution operator of the free theory , which is supposed to know explicitly from a given @xmath454 via @xmath459 u_0(\tau,\tau_0)~ , \qquad u_0(\tau_0,\tau_0)= 1~.\end{aligned}\ ] ] to fix @xmath456 completely , we have to choose the initial condition as well .
formally , this can be done as in the flat space qft , by requiring that at some initial time @xmath460 , @xmath461 combining eqs . and ( which implies @xmath462 ) , and making use of , we have @xmath463 = -i u_0^{-1}(\tau,\tau_0 ) h_i[\delta\phi(\tau_0 ) , \delta\pi(\tau_0 ) ; \tau ] u(\tau,\tau_0 ) = -i u_0^{-1}(\tau,\tau_0 ) u(\tau,\tau_0 ) h_i[\delta\phi(\tau ) , \delta\pi(\tau ) ; \tau]~.\end{aligned}\ ] ] as one can directly check , the solution of this equation is @xmath464 in the last line , the time ordered integrals have been compactly written into a time ordered exponential .
however the exponential still needs to be calculated order by order in the form of the second and third line .
now an operator in heisenberg picture can be expressed via the interaction picture field as @xmath465 q^i(\tau ) \left [ t e^{-i \int_{\tau_0}^\tau h_i(\tau')d\tau'}\right ] ~,\end{aligned}\ ] ] where @xmath466 is anti - time ordering . as in flat space qft , @xmath467 is the time evolution operator of the quantum states in the interaction picture . in the aspect of quantum states ,
there is one state of particular importance : the `` vacuum '' state .
we have discussed the vacuum state in section [ sec : spectator - field - de ] .
this discussion is reused for a few times for the inflationary scalar and tensor perturbations .
however , here new ingredients come in , because of the interactions . in flat space qft
, there is already a difference between the interaction vacuum and free vacuum .
the most part of this section is parallel to flat space qft .
comments are made when the expansion of the background brings a difference .
let us use @xmath468 to denote the vacuum state of the free theory , which is defined by annihilation operators .
we use @xmath469 to denote the interaction vacuum , which is to be constructed using @xmath468 . as discussed in section [ sec : spectator - field - de ] , it is important to note that the vacuum is the lowest energy state .
we first expand @xmath468 using a complete set of energy eigenstates of the full theory . on expanding background matter energy
is not conserved and we have yet to define gravitational energy .
nevertheless , we could consider the energy eigenstates at an initial time for the perturbation modes . and
planck scale @xmath138 may not be large enough to contain all those modes safely .
] near @xmath470 , we have @xmath471 where in the last line @xmath472 denotes the states excluding the ground state .
now we add a small imaginary part to time : @xmath473 near @xmath470 , the @xmath474 factor introduces a term @xmath475 .
thus only the @xmath469 state survives from this @xmath474 factor.[multiblock footnote omitted ] now @xmath469 can be solved as @xmath476 from now on , we shall neglect the tilde on @xmath31 .
in other words the @xmath31 in later sections should be understood with a small imaginary part . now combine and , the time dependent expectation value of an operator can be written as @xmath477 q^i(\tau ) \left [ t e^{-i \int_{\tau_0}^\tau h_i(\tau')d\tau'}\right ] |0\rangle } { \left|\langle0|\omega\rangle\right|^2 } ~.\end{aligned}\ ] ]
note that in the anti - time ordering @xmath478 $ ] , the @xmath474 prescription is actually with a different sign ( i.e. @xmath479 ) from the time ordering part @xmath480 $ ] .
this is because the @xmath466 part is the hermitian conjugate of @xmath481 part , which flips the sign of @xmath474 term .
this contour of integration , specified by the above @xmath474 prescription , is known as the in - in contour ( fig .
[ fig : in - in - contour ] ) .
the left panel is the contour of integration in the in - in formalism . in the right panel , the contour is deformed to a vertical line running top - down .
the deformed contour is used numerically in @xcite and more generally in @xcite.,scaledwidth=80.0% ] the denominator deserves a more close look . by taking @xmath482(identity matrix ) , we can actually write @xmath483 as @xmath484 \left [ t e^{-i \int_{\tau_0}^\tau h_i(\tau')d\tau'}\right ] |0\rangle } { \langle\omega|\omega\rangle } = 1~.\end{aligned}\ ] ] to get the rhs , we require both the states @xmath485 and @xmath486 are normalized .
the @xmath478 $ ] part cancels the @xmath480 $ ] part because it is just the hermitian conjugate of a unitary operator .
thus finally we arrive at the master formula , known as the factorized form of in - in formalism : @xmath487 q^i(\tau ) \left [ t e^{-i \int_{\tau_0}^\tau h_i(\tau')d\tau'}\right ] |0\rangle } ~.\end{aligned}\ ] ] practically one expands the above formula order by order in perturbation theory .
for example , the first order term in the expansion is @xmath488 and the second order terms are @xmath489 however , one should note that the order in @xmath455 does not necessarily agree with the @xmath490 expansion . for example
, when calculating the four - point function of general single field inflation , both the first order expansion of in - in formalism ( with @xmath455 being the @xmath491 vertex ) and the second order expansion of in - in contributes , with comparable amount @xcite . before moving on , let us note that there are a few variants of the in - in formula : * _ the wick - rotated factorized form _
@xcite : in the right panel of fig .
[ fig : in - in - contour ] , if there are no poles or cuts between the dashed contour and the solid contour , both contours give the same result . using the solid contour , the in - in formalism can be written into a more compact form : @xmath492 |0\rangle \nonumber\\ & = \langle 0| \bar{t } \left [ q^i(\tau ) \exp \left ( - \int_{-\infty}^{\infty}h_i(\tau+i\tau_e)d\tau_e \right ) \right ] |0\rangle ~,\end{aligned}\ ] ] where in the second line a further change of variable is applied to simplify the integral .
the wick - rotated factorized form makes the ( linear ) uv convergence of in - in integral more explicitly .
* _ the commutator form _ @xcite : eq .
can be rewritten as @xmath493\cdots]]\right\rangle~.\end{aligned}\ ] ] to see this , one first observe the @xmath494 order term agrees with . then assuming the @xmath495th order term also agrees , the time derivative of the @xmath495th order terms generates the agreement of the @xmath335th order terms .
+ the advantage of is that , there are some cancellations in the ir ( i.e. @xmath496 ) from the commutators , which are not obvious in the form of .
for example , the two terms in could have cancellations between each other in the ir .
+ however , the form has also a disadvantage : the @xmath474 and @xmath497 terms are mixed in thus we may meet divergences in the uv ( i.e. @xmath158 ) , due to the difference between @xmath469 and @xmath468 .
+ similarly , the commutator form can also be written as @xmath498\cdots]]\right\rangle~. \end{aligned}\ ] ] * _ the mixed form _
@xcite : to keep both the advantage of in the uv and in the ir , one can choose an intermediate time @xmath499 ( which is typically @xmath500 ) and re - group the terms before @xmath499 using the factorized form and the terms after @xmath499 using the commutator form as @xmath501 the result is independent on the intermediate time @xmath499 , and well behaved both in the uv and in the ir .
we now use the in - in formalism to calculate the three point function for inflation . for the minimal model of inflation
, the three point function is small . on the other hand
it is dominated by pure gravitational actions , instead of @xmath502 from the explicit non - linearity in the inflationary potential . to calculate this
, we need to derive the third order gravitational action . as an example , here we only do the calculation in @xmath193-gauge and using the adm form of ricci scalar . for simplicity , here we take the planck unit : @xmath503 . for calculating the three point function
, we have to solve the constraint equations .
naively , we would expect that we have to solve the constraint equations up to third order in @xmath193 .
however , fortunately , we still only need to solve the constraints up to first order . in general ,
to derive the second order action , one needs to solve the first order constraint equations . to derive the @xmath335th order action ( @xmath504 )
, one needs to solve the @xmath505th order constraint equations @xcite . to see this , recall that there are no time derivatives acting on constraint variables .
also the spatial derivatives on constraint variables can be converted to momentum @xmath157 by going into momentum space ( to save writing , we shall not do it explicitly ) .
taking the lapse function for example , in general we have @xmath506 let us expand @xmath507 , where @xmath508 denotes the @xmath509 order terms .
the constraint equation @xmath510 can be expanded ( up to @xmath511 ) as @xmath512 @xmath513 where the subscript @xmath514 denotes the term does not contain @xmath221 , and is @xmath1th order in @xmath193 ( or the shift vector ) .
now consider the @xmath335th order action ( @xmath504 ) .
all the terms that contains @xmath508 are @xmath515 this term vanishes because of .
all the terms that contains @xmath516 are @xmath517 again this vanishes because of .
however , from @xmath518 on , the coefficients in the constraint equation no longer coincides with the coefficients in @xmath519 .
the above proof also goes through for the shift vector @xmath520 .
thus we have to solve the constraints to order @xmath521 .
now we want to calculate @xmath522 .
thus eq .
is enough for solving the constraints ( and @xmath523 at this order ) .
the second order action is already given in .
the third order action takes the form @xcite : @xmath524 note that we can set @xmath523 .
this action is simplified into @xmath525 inserting the constraint eq . into the third order action , and do a lot of integration by parts , we get @xmath526 where @xmath117 is defined by @xmath527 thus the first line of is @xmath528 in slow roll and the second line is @xmath529 . in the third line , @xmath530 is proportional to the equation of motion , thus does not contribute to the three point function to eliminate @xmath530 . in this case , all the @xmath531 terms can be eliminated without using equation of motion , and there is no boundary term contributing to the action , in terms of @xmath532 . ] .
just for keeping track , @xmath533 is @xmath534}{4a^2h } + \frac{\partial\zeta\partial\chi-\partial^{-2}[\partial_i\partial_j(\partial_i\zeta\partial_j\chi)]}{2a^2h } ~.\end{aligned}\ ] ] and @xmath535 is a collection of boundary terms on the boundary ( i.e. @xmath531 in the bulk ) , and ( b ) non - vanishing on the boundary , need to be considered . ] : @xmath536 \nonumber \\ & + \partial_i \big [ \epsilon
a^3h \zeta^2\partial_i\partial^{-2}\dot\zeta -2a \zeta^2\partial_i\zeta - \frac{2a}{h } \zeta\dot\zeta\partial_i\zeta + \frac{2a\epsilon}{h } \zeta\partial_j\zeta\partial_i\partial_j\partial^{-2}\dot\zeta + \frac{a\epsilon}{h } \partial_j\zeta\partial_j\zeta\partial_i\partial^{-2}\dot\zeta \nonumber \\ & \qquad + \frac{1}{4ah^2 } ( - 3 \partial_i\zeta\partial_j\zeta\partial_j\zeta - 2 \zeta\partial_j\zeta\partial_i\partial_j\zeta + 2 \zeta\partial_i\zeta\partial^2\zeta ) % \nonumber \\ & \qquad + \frac{1}{4ah^3 } ( 2\partial^2\zeta\partial_i\zeta\dot\zeta-2\partial_i\partial_j\zeta\partial_j\zeta\dot\zeta+\partial_j\zeta\partial_j\zeta\partial_i\dot\zeta ) \nonumber \\ & \qquad + \frac{\epsilon}{4ah^2 } ( 2\zeta\partial^2\zeta\partial_i\zeta-2\zeta\partial_i\partial_j\zeta\partial_j\zeta+\partial_i\zeta\partial_j\zeta\partial_j\zeta ) % \nonumber \\ & \qquad + \frac{a\epsilon}{2h^2 } ( \dot\zeta\partial_j\zeta\partial_i\partial_j\zeta^{-2}\dot\zeta-\zeta\partial_j\dot\zeta\partial_i\partial_j\zeta^{-2}\dot\zeta + \zeta\dot\zeta\partial_i\dot\zeta - \dot\zeta^2\partial_i\zeta ) \big]\end{aligned}\ ] ] the spatial part @xmath537 $ ] do not contribute to the three point function . because there is no spatial boundary in our inflationary universe .
however , we have to be more careful about the time total derivative .
this is because for in - in formalism , we do have a time boundary , typically chosen at a time where @xmath193 becomes conserved after that . in the @xmath538 limit , only one term in the @xmath539 $ ] survives when expanded in the bulk , and thus the variation principle is not well defined .
however , actually there is no problem because we are allowed to use linear equations of motion in the bulk .
if we choose not to use the linear equations of motion , but instead perform field redefinitions to get rid of high derivative terms in the 3rd order action , then this boundary term does not appear @xcite . ]
@xcite : @xmath540 to conclude , the third order lagrangian , which contributes to the three point function is @xmath541 where the first line dominates over the second line in slow roll approximation ( note @xmath117 is proportional to @xmath107 ) .
before to proceed , it is important to note that , the dominate contribution for @xmath542 come from gravitational interactions , instead of the inflaton s direct coupling . to see this , consider the inflaton s direct coupling @xmath543 .
the third derivative @xmath544 can be estimated as @xmath545 where @xmath546 collectively denote order of slow roll parameters .
thus @xmath547 gauge transform @xmath543 into @xmath193-gauge , using @xmath548 , we have the lagrangian for inflaton self interaction in the @xmath193-gauge : @xmath549 this belongs to the second line of , which is sub - dominated .
in other words , gravity is not the weakest force during inflation .
the inflaton self interaction is weaker ( although it is not a gauge interaction , in which case gravity is conjectured to be the weakest force @xcite ) . as in classical mechanics ,
a legendre transform @xmath550 brings lagrangian to hamiltonian .
especially , potential energy @xmath71 flips sign in the legendre transform . for the gravitational action ,
the situation is a bit non - trivial because there is derivative coupling .
fortunately , the third order interaction hamiltonian is still just flipping the sign of the third order action : @xmath551 we shall derive the above equation , together with a non - trivial relation for the 4th order relation between @xmath552 and @xmath553 @xcite .
readers who are not interested in this detail are suggested to skip the remainder of this subsection .
we consider a lagrangian up to 4th order in perturbations : @xmath554 where @xmath555 are the dynamical fields where @xmath1 runs over the number of those dynamical fields .
the coefficients @xmath556 are function of @xmath555 but do not contain @xmath557 .
the @xmath335 index in the parenthesis denotes the coefficient is @xmath335th order in @xmath268 .
first one define the conjugate momentum @xmath430 : @xmath558 where @xmath559 is the inverse of @xmath560 such that @xmath561 .
now @xmath557 can be solved in terms of @xmath430 order by order ( up to 3rd order here ) as @xmath562 @xmath563 now the hamiltonian density is @xmath564 where all the @xmath557 on the rhs are to be replaced by eqs . , . finally , define the interaction picture derivative as discussed in sec .
[ sec : inter - hamilt ] : @xmath565 inserting this definition into the hamiltonian , the hamiltonian density becomes @xmath566 we observe that at the third order , we indeed have @xmath567 even when there are derivative couplings . on the other hand , at the 4th order non - trivial terms arises in the last two lines .
now we have derived the hamiltonian , and are in the position to use in - in formalism to calculate the correlation function .
we only do it at the leading order of slow roll . inserting the mode function together with the hamiltonian ( first line of and eq . ) , and perform the time integration .
the result is @xcite @xmath568 where there is a factor of @xmath569 on the rhs .
thus the three point function is also called the bispectrum . the dimensionless shape function @xmath570 the shape functions are plotted in fig [ fig : local - shapes ] .
note that when @xmath571 , @xmath572 .
this @xmath573 limit ( or limits @xmath574 or @xmath575 ) is called the squeezed limit .
the squeezed limit is of particular importance in the study of non - gaussianities .
the panels ( a ) , ( b ) and ( c ) are the shape functions @xmath576 , @xmath577 and @xmath578 respectively .
note that only the configurations satisfying the triangle inequality @xmath579 makes sense in physics .
the part with @xmath580 is set to zero by hand and should be ignored from physical consideration .
the shape function in panel ( a ) is known as `` local shape '' , because local processes could only generate this shape .
the physical meaning of local shape is discussed extensively in sec .
[ sec : local - non - gauss ] .
the other two panels look similarly to local shape .
especially all those 3 shapes have the same scaling when @xmath573 , that @xmath572 in this limit.,title="fig:",scaledwidth=45.0% ] + the panels ( a ) , ( b ) and ( c ) are the shape functions @xmath576 , @xmath577 and @xmath578 respectively .
note that only the configurations satisfying the triangle inequality @xmath579 makes sense in physics .
the part with @xmath580 is set to zero by hand and should be ignored from physical consideration .
the shape function in panel ( a ) is known as `` local shape '' , because local processes could only generate this shape .
the physical meaning of local shape is discussed extensively in sec .
[ sec : local - non - gauss ] .
the other two panels look similarly to local shape .
especially all those 3 shapes have the same scaling when @xmath573 , that @xmath572 in this limit.,title="fig:",scaledwidth=45.0% ] the panels ( a ) , ( b ) and ( c ) are the shape functions @xmath576 , @xmath577 and @xmath578 respectively .
note that only the configurations satisfying the triangle inequality @xmath579 makes sense in physics .
the part with @xmath580 is set to zero by hand and should be ignored from physical consideration .
the shape function in panel ( a ) is known as `` local shape '' , because local processes could only generate this shape .
the physical meaning of local shape is discussed extensively in sec .
[ sec : local - non - gauss ] .
the other two panels look similarly to local shape .
especially all those 3 shapes have the same scaling when @xmath573 , that @xmath572 in this limit.,title="fig:",scaledwidth=45.0% ] we explain in this subsection the physical meaning of local non - gaussianity @xcite . the local non - gaussianity is originally introduced as a position space expansion of non - gaussian perturbations around gaussian perturbations is originally defined in terms of the gravitational potential @xmath253 .
here we use the wmap convention .
a summary of conventions can be found in @xcite . ] : @xmath582 where @xmath583 is a perturbation variable satisfying gaussian statistics .
note that this is an ansatz , instead of a general expansion .
because in general , the non - gaussian estimators @xmath584 could depend on position , and terms like @xmath585 is not necessarily expanded at the same spatial point . as a result ,
the terms like @xmath586 has to be in general written as a convolution .
the local ansatz has clear physical meaning .
this ansatz assumes non - gaussianities are generated independently at different spatial points . translating this requirement into the context of inflation ,
typically this implies the non - gaussianity is generated on super - hubble scales . especially , note
that , for positive @xmath581 , the part of contribution from positive @xmath587 gets enhanced , and the part of contribution from negative @xmath587 gets suppressed .
this provides an intuitive way to visualize the sign of @xmath581 . in position space
, this non - gaussianity can be characterized by the probability distribution function of @xmath193 .
the probability distribution is plotted in fig .
[ fig : prob - fnl ] .
note that the tail at large and positive @xmath193 indicates where structures are formed in the universe .
thus positive @xmath581 means more galaxies are formed given a fixed power spectrum .
for the cmb sky , positive @xmath581 means there are more very cold spots than very hot spots on the cmb , and more modestly hot spots than modestly cold spots .
probability distribution of @xmath193 for different @xmath581 parameters .
positive @xmath581 corresponds to the red curve in the figure .
negative @xmath581 corresponds to @xmath588 of the red curve .
this figure is only for illustration purpose .
the real world @xmath589 is much smaller than the one plotted here.,scaledwidth=70.0% ] the position space features of @xmath581 is intuitive to understand .
however , the inflationary perturbation manifests itself in momentum space .
there are much more configurations of three point function to look at , than just looking at position space probability distributions point by point .
thus the momentum space correlation functions provides a much more powerful test of non - gaussianity than the isolated position space test . in momentum space
, the corresponding expression is @xmath590 inserting the expansion into the three point function , the three point function can be calculated as @xmath591 this non - gaussianity has the same shape as the first term of .
the other two terms in eq .
has different shape , but behave quite similarly .
the observational bound of local shape @xmath581 is @xcite @xmath592 the local ansatz can be generalized to multiple species : @xmath593 note that in this generalization , the locality of non - gaussianity is not modified .
only that multi - source of fluctuations become possible . here the notation @xmath594 and @xmath595 is used , having in mind that one can interpret @xmath594 as @xmath596 . in this case
the expansion is related to the @xmath265-formalism . inserting the expansion into correlation functions , at the three point level
, we have @xmath597 while for the 4-point function , we have @xmath598 ~.\end{aligned}\ ] ] because of the @xmath599 factor , the four point function is also known as the trispectrum .
now even for the local shape , there are two estimators @xmath600 and @xmath601 . in terms of the ansatz
, we have @xmath602 note that @xmath600 depends on @xmath603 , which does not appear in the expression of three point functions . on the other hand
, @xmath601 is determined by @xmath604 and @xmath520 , the same coefficients that determines @xmath581 .
thus it is interesting to see if there are relations between @xmath601 and @xmath581 .
interestingly , such relation indeed exist . using the cauchy - schwarz inequality @xmath605
where @xmath606 and @xmath607 are real vectors , we have @xmath608 this relation is known as the suyama - yamaguchi relation @xcite . especially , when there is only one source of fluctuation , the above inequality takes the equal sign .
similar relations also exist for higher order correlation functions @xcite or loop corrections @xcite . here
we are to consider the physical meaning of the squeezed limit .
when one of the momenta is much smaller than the other two , say , @xmath609 , the @xmath610 mode is already super - hubble and become classical and conserved when the modes @xmath611 exit the horizon @xcite .
this constant mode can be considered as just a shift of background .
is for some reason not conserved on super - hubble scales , it can not be considered as a shift of background . because one can distinguish @xmath193 and a redefinition of the background scale factor by the time evolution of @xmath193 .
examples of this type include multi - field inflation , and a growing mode of @xmath193 , as discussed in secs .
[ sec : multi - field - infl ] and [ sec : gener - attr - behav ] respectively . ] . ]
thus the three point function should reduce to the background shift of a two point function . to see this explicitly
, one can expand @xmath612 with the presence of a background @xmath613 mode @xcite : @xmath614 where @xmath615 can be written as a shift of background .
recall the large gauge transformation : via a @xmath616-transformation , we have @xmath617 to get the final equation , consider a general function @xmath618 .
we have @xmath619 and @xmath620 .
thus we have @xmath621 contract this term with @xmath622 , we get @xmath623 this consistency relation should be satisfied when @xmath624 is indeed a shift of background , especially for the standard model of single field inflation . to check that , we note the shape function takes the limit @xmath625 thus @xmath626 recall that the spectral index is @xmath627 .
thus the consistency relation is indeed satisfied for the three point function that we just reviewed . comparing with the squeezed limit of eq .
@xmath628 we have for this minimal model @xmath629 the squeezed limit discussion of three point function can be straightforwardly generalized to higher point correlations . as a result , in the squeezed limit , we have @xmath630 those consistency relations can also be checked by resuming the perturbation hamiltonian with the presence of a classical mode of perturbation @xcite , or a result of unitary @xcite or operator produce expansion @xcite .
additional soft limits arises for @xmath335-point function with @xmath631 .
for example , now we can have @xmath632 without any external momentum being zero ( note the ordering of momentum is not important here .
thus this covers any partition of summation goes to zero ) .
the situation is illustrated in fig .
[ fig : sy - ineq ] . soft limit of correlation functions . in this example
, we have @xmath633 .
the left panel illustrates momentum conservation , and the right panel illustrates the soft limit is dominated by a propagation of a soft momentum . in this limit
we get @xmath634.,scaledwidth=60.0% ] note that the exchanged scalar has a divergent behavior @xmath635 when @xmath636 . thus in this limit
, this propagation should be the dominate diagram .
the @xmath335-point correlation breaks into a @xmath637-point correlation times a @xmath638-point correlation .
note that for @xmath639 and @xmath640 ( which is the requirement that @xmath383 is an internal momentum ) , the zeroth order expansion in does not vanish , the zeroth order expansion may not vanish if there are tadpole contribution to the background .
however , one can cancel this tadpole before doing the whole calculation . or
one can take the tadpole into account in the correlation function level , and the conclusion should not change . ] .
we again apply a similar argument as the squeezed limit : @xmath383 is a shift of background , thus @xmath641 . with this shift ,
the zeroth order expansion of is not affected and thus @xmath642 ^ 2 } \langle\zeta_{\mathbf{k}_1}\zeta_{\mathbf{k}_2 } \cdots \zeta_{\mathbf{k}_m}\rangle \langle\zeta_{\mathbf{k}_{m+1}}\zeta_{\mathbf{k}_{m+2 } } \cdots \zeta_{\mathbf{k}_n}\rangle\end{aligned}\ ] ] especially , in case of four point correlation function , this relation produces @xmath643 once we relax the single degree of freedom assumption , there can be more than one mode propagating as the soft mode . also the propagating mode could contribute a different shift of the background . in this case
@xmath644 this is again the suyama - yamaguchi relation .
here we derive the suyama - yamaguchi relation once more , by showing that the suyama - yamaguchi relation is actually a special example of optical theorem in cosmology @xcite .
the main purpose of this subsection is to illustrate the unitarity method , and how to cut cosmological feynman diagrams in general , despite that the same relation has already been derived twice in this review . in the in - in formalism
, an expectation value can be calculated as @xmath645 where @xmath646 is a unitary operator defined by @xmath647 and @xmath648 is the corresponding operator in the interaction picture . to be explicit , we have @xmath649 and @xmath650 for the time being , we restrict our attention to the states that created by a single field operator . because in the folded limit , a single internal propagator should be dominate considering that @xmath651 .
thus we consider the unitarity formula restricted to those states , and the limitations will be addressed in later sections . in the heisenberg picture ,
the states are time independent and thus even in a frw background we have @xmath652 where @xmath653 is the mode function of @xmath654 , defined as @xmath655 a one particle state ( well before horizon crossing ) is defined as @xmath656 this is because in the very early times with @xmath657 . thus here `` early '' means well before the horizon crossing of the comoving wave number @xmath657 . ]
, the perturbation modes live in flat spacetime . in the interaction picture
, the operators evolves with the free hamiltonian . while the states evolves with the interaction hamiltonian , with time evolution operator @xmath646 as @xmath658 .
thus we have , which contributes the sum .
but here we are only interested in the connected components in the correlation function . thus this term is ignored . ]
@xmath659 note that the rhs of this relation has time dependence .
this relation holds at any time when field content of the model do not change , or change trivially . for example
, a single field component decays into radiation can be considered as a trivial change , when the interested @xmath8 is on super - hubble scales .
this is because on super - hubble scales , a scalar field and a fluid share the same properties , and can be calculated in the same way in the in - in formalism . on the other hand , curvaton decay in the curvaton scenario
can not be considered as a trivial change of field content , if the curvaton is decayed into the same radiation as the inflaton decay product .
this is because there is a reduction of low energy degrees of freedom in this case .
one should also note that itself is not the unitarity condition .
rather it is derived from the unitarity relation from sub - hubble time scales .
thus a violation of such as a non - trivial change of field content is nothing scared .
it could be simply a violation of the derivation procedure of , instead of a violation of unitarity .
for example , there could exist monopoles , bound states or other non - local objects in minkowski space , with radius greater than the horizon size when putting the fields onto de sitter background .
those states can not exist any longer because of causal considerations , and thus may signal an apparent shortage of states in de sitter . by inserting equation ( [ eq : unitarity - prime ] ) into the calculation of in - in formalism
, we have @xmath660 where @xmath234 denote the corrections that can not be expressed by ( [ eq : unitarity - prime ] ) , ( say , non - trivial change of field content , bound states , two particle states via loop diagram , and not - so - folded contributions .
so far , we have almost restricted our attention to the minimal model of inflation . on the other hand
, there are a vast number of inflation models around .
those models brings various novel predictions .
here we introduce some of those models . in the case of single field inflation , there are already a lot of possibilities for generalization .
the dynamics of inflation is determined by its lagrangian .
thus towards generalization of the minimal inflation model , it is natural to consider how to generalize its lagrangian .
here we restrict our attention to the scalar sector an immediate question is if we can generalize the potential .
the answer is almost negative , considering we have already considered the potential as a general function of @xmath72 in the minimal model .
however , this is not the end of the story .
we have assumed @xmath71 is single valued .
one could consider multiple valued @xmath71 , for example @xmath72 moving on a monodromy potential .
this opens up interesting possibilities such as large field inflation in string theory @xcite .
one could also consider features or small oscillations on the potential .
the inflaton could either tunnel though the potential @xcite , or rolling but with slow roll conditions violated ( see sec .
[ sec : scale - depend - feat ] ) .
now let us go beyond modifying potential .
the task is to find a general scalar lagrangian , which is well behaved theoretically .
especially the lagrangian should be ghost free .
otherwise either the theory is not consistent , or we are not expanding around the correct vacuum or imposing the correct boundary conditions - symmetric quantum mechanics , the ghost may be eliminated @xcite . ] . the ghost free requirement can be expressed in more explicit terms .
consider we have some background , and want to quantize small perturbations on this background : first of all , we would like to restrict our attention to the theories with correct sign of the kinetic term : @xmath661 instead of @xmath662 .
ghost is not welcome to physics @xcite . to see this , note that the propagator is the inverse of the second order action .
thus a wrong sign kinetic term results in a wrong sign of the propagator @xmath663 we yet have the freedom to choose the sign of @xmath474 term for the ghost .
however , if we choose the plus sign , the imaginary part of has different sign from normal particle .
thus the optical theorem is violated and there is no probabilistic interpretation of the theory .
on the other hand , if the minus sign is chosen , negative energy modes propagate forward in time .
thus the ghost particle carry negative energy . in this case
the vacuum decays into ghost particles even if ghost and other particles only interact gravitationally .
even worse , the decay rate for vacuum to decay into two photon and two ghost particles is @xmath664 where @xmath121 is the uv cutoff of lorentz violation .
thus in a lorentz invariant theory , the decay rate is divergent .
the divergence is physical : in each frame there is a non - vanishing decay rate and one has to integrate over the infinite volume of lorentz group .
any small decay rate is forbidden .
secondly , we require the equation of motion for perturbations to be second order in time derivatives .
this is because , in a higher order derivative theory ( for example , 4th order ) , the propagator can be written as @xmath665~.\end{aligned}\ ] ] thus one of the two terms behaves as a ghost .
we would like to emphasize that the above requirements are imposed on the perturbations .
the background could still have higher orders derivatives as long as perturbations are well behaved .
the minimal inflation model is definitely ghost free .
however , it is not the most general ghost free theory . instead , the most general theory of a scalar , coupled to gravity and with second order equation of motion for perturbations , is @xcite @xmath666 \nonumber\\ & + g_5(\phi , x ) g_{\mu\nu}\nabla^\mu\nabla^\nu\phi - \frac{1}{6 } g_{5,x } \left [ ( \box \phi)^3 - 3\box \phi(\nabla_\mu\nabla_\nu\phi)^2 + 2(\nabla_\mu\nabla_\nu\phi)^3 \right]~.\end{aligned}\ ] ] where @xmath667 are arbitrary functions of @xmath72 and @xmath668 , and @xmath669 and @xmath670 denote derivatives with respect to @xmath671 . for perturbations ,
the high order time derivatives are removed by integration by parts .
one still need to be careful about the correct sign of the kinetic term , which cuts about half of the parameter space .
the linear perturbations are considered in @xcite and the non - gaussianities are calculated in @xcite . historically , the @xmath672 term in the is considered as k - inflation @xcite . later the @xmath673 term is added inspired by the galilean symmetry @xcite . and
finally the most general form is rediscovered in the context of cosmology .
the full theory of is complicated and we are not going into the details here .
however , it is interesting to consider the simplest non - minimal case of the theory , where @xmath674 , and all other terms vanish @xcite . in this case , by varying the action with respect to the metric , the energy density takes the form @xmath675 and one can find the pressure is @xmath676 .
the friedmann equation and the continuity equation still hold : @xmath677 for the theory of perturbations , we work in the @xmath193-gauge . the second order action can be derived as @xmath678~,\end{aligned}\ ] ] where @xmath679 note that in the minimal inflation scenario , where @xmath680 and @xmath681 .
however here this parameter is generalized , and can be understood as the sound speed of the perturbations .
this is because now the temporal and spatial directions have different weight in the action . to solve the linear perturbation equations
, one typically assumes slow variation of @xmath682 : @xmath683 . in this case , the @xmath193 field can be quantized and solved as @xmath684 note again that now the oscillation factor in the time direction is @xmath685 . on the other hand ,
the spatial fourier transform still has the same factor @xmath686 .
thus a propagating wave described by this lagrangian has a modified speed .
the power spectrum can be calculated as @xmath687 the only difference from the minimal inflation model is the factor of @xmath682 . when @xmath688 , this is an enhancement factor .
the quantum fluctuations are larger because the perturbations are frozen at a smaller horizon . for non - gaussianities , the leading order terms in the slow roll approximation is @xcite @xmath689 \frac{\dot\zeta^3}{h^3 } - \frac{3\epsilon}{c_s^4 } ( 1-c_s^2 ) \zeta\dot\zeta^2 + \frac{\epsilon}{a^2c_s^2 } ( 1-c_s^2)\zeta(\partial\zeta)^2 \right\ } ~,\end{aligned}\ ] ] where @xmath690 using some integration by parts and making use of the equation of motion , can further be simplified into - gauge and do a gauge transformation to the @xmath193-gauge . to get those leading order terms ,
the gravitational interaction in the @xmath191-gauge can be ignored . and
the gauge transformation only needs to be expanded into the leading order @xmath691 .
thus the calculation is much simplified in the @xmath191-gauge plus gauge transformation way . ]
@xmath692 using the in - in formalism , the bispectrum can be calculated as @xmath568 where @xmath693~.\end{aligned}\ ] ] in the squeezed limit @xmath694 , it is clear that the first term vanishes proportional to @xmath695 .
the second term appears divergent .
however , both the divergence and the constant term near @xmath694 actually cancel .
to see this , consider the limit @xmath696 with @xmath697 . in this limit
, @xmath698 thus @xmath699 near @xmath694 , where the linear term depends on how the squeezed limit is taken .
the shape functions are plotted in fig .
[ fig : equil - shapes ] . the left and right
panels are respectively the normalized shape functions @xmath700~.\end{aligned}\ ] ] those shapes are known as equilateral shapes .
note that those shape functions are not separable .
in other words , the shape functions can not be written in the form @xmath701 .
technically it is much harder to do data analysis of non - separable shapes than separable ones .
for this purpose , some shape templates are proposed to mimic the equilateral shapes @xcite : the equilateral template @xmath702 and the orthogonal template @xmath703 those templates are plotted in fig .
[ fig : equil - ansatz ] .
the coeffieients multiplying those shapes are constrainted as @xcite @xmath704 the left and right panels are the first and second terms of eq . .
they are equilateral shapes.,title="fig:",scaledwidth=45.0% ] the left and right panels are the first and second terms of eq . .
they are equilateral shapes.,title="fig:",scaledwidth=45.0% ] left panel : the equilateral template @xmath705 .
right panel : the orthogonal template @xmath706.,title="fig:",scaledwidth=45.0% ] left panel : the equilateral template @xmath705 .
right panel : the orthogonal template @xmath706.,title="fig:",scaledwidth=45.0% ] the trispectrum of this @xmath707 theory can also be calculated using the in - in formalism @xcite .
it is useful to note that non - trivial transformation between the lagrangian to the hamiltonian has to be considered .
also the scalar propagation diagram ( like the right panel of fig .
[ fig : sy - ineq ] ) needs to be included .
this @xmath707 class of models can also be generalized to multiple field case , where multi - field effects can appear even if the field space trajectory is a geodesic @xcite . in the multiple field case , local and equilateral non - gaussianities
may get combined @xcite . in this subsection , we take a different point of view for extending the cosmic perturbation theory . unlike the previous section , here we assume an inflationary background , and see what is the general perturbation theory living on that background .
this approach is known as the effective field theory ( eft ) approach for inflationary perturbations @xcite ( see also , @xcite ) . as in the general discussion of eft , any operator respecting
the symmetry of the theory could appear in the lagrangian .
the relative importance of those operators are determined by the dimension of the operator .
operators with higher dimensions are suppressed more by the uv cutoff scale of the eft .
now we are to write down all possible terms for perturbations . considering that there are many possibilities , it is helpful to choose a gauge to reduce the number of terms .
the gauge to choose here is the @xmath193-gauge , where @xmath248 . in this gauge ,
the most general lagrangian for single field inflation , with the foliation - preserving diff - invariance @xmath708 is @xmath709~,\end{aligned}\ ] ] where @xmath710 .
the dots stand for higher order terms ( including higher order in perturbation variable @xmath193 and higher order in derivatives ) , that can be written down straightforwardly if needed .
it can be shown that other operators as @xmath711 , @xmath712 , @xmath713 , etc can be rewritten in terms of the above operators , up to some total derivatives .
expanding around the frw background , one finds @xmath714 the form of is not yet very useful .
this is because we have to solve all the complicated gravitational interactions , which is possible but messy .
fortunately , there are a hierarchy of energy scales in the theory , among which the planck mass is the highest .
this implies the existence of a range of energy where gravitational interactions decouple . to find this range of energy , together with the simplified action with decoupled gravity
, we should go away from the @xmath193-gauge . as we discussed in previous sections ,
there are a lot of cancellations in the @xmath193-gauge , which makes decoupling not transparent at all .
the @xmath193-gauge is chosen by picking up a specific time coordinate .
thus the @xmath193-gauge can be undone by promoting the @xmath193-gauge time to a scalar field @xmath715 , such that in the @xmath193-gauge , @xmath716 ( where the subscript @xmath193 denotes quantity in the @xmath193-gauge ) .
now consider a general gauge @xmath717 @xmath715 transforms as a scale field thus @xmath718 now we are ready to make the @xmath193-gauge quantities into tensors from @xmath715 : @xmath719 @xmath720 @xmath721 @xmath722 where the left column are unitary gauge quantities and the right column are quantity without choosing a gauge . by doing the above replacements to the action , together with the replacement ,
we return from the @xmath193-gauge to a general coordinate system without choosing a gauge : @xmath723 \nonumber\\ & + m_p^2 \dot h(t+\pi ) \left [ ( 1+\dot\pi)^2g^{00 } + 2(1+\dot\pi)\partial_i\pi g^{0i}+ g^{ij}\partial_i\pi\partial_j\pi \right ] \nonumber\\ & + \frac{m_2 ^ 4(t+\pi)}{2 } \left [ ( 1+\dot\pi)^2g^{00 } + 2(1+\dot\pi)\partial_i\pi g^{0i}+ g^{ij}\partial_i\pi\partial_j\pi + 1 \right]^2 \nonumber\\ & + \frac{m_3 ^ 4(t+\pi)}{2 } \left [ ( 1+\dot\pi)^2g^{00 } + 2(1+\dot\pi)\partial_i\pi g^{0i}+ g^{ij}\partial_i\pi\partial_j\pi + 1 \right]^3 + \cdots \big\}~,\end{aligned}\ ] ] where we neglect more terms in @xmath234 for simplicity . after re - obtaining the gauge invariant action
, we are free to choose the @xmath191-gauge ( or any other gauge close to the @xmath191-gauge , like the newtonian gauge ) .
the advantage of @xmath191-gauge , or any gauge which is close to that , is that _ near the horizon crossing _ ,
gravity is weakly coupled and only provide slow roll suppressed corrections .
an example of this statement can be found in eq . .
more generally , we can compare the terms with and without dynamical gravity in the new action , and conclude : the decoupling energy range for matter perturbations from gravity is should be understood as a wave number instead of an energy scale . because we are comparing @xmath724 with @xmath725 . on super - hubble scales
there are a huge number of particles in perturbation mode @xmath157 thus the corresponding actual energy is much greater than @xmath724 . ]
@xmath726 where the dots denotes other scales ( not shown in ) that could appear , which are also suppressed by @xmath138 .
note that @xmath727 , and @xmath728 .
this result is intuitive to understand : when all scales appearing in the theory are smaller than @xmath138 , gravity should be weak .
however , when @xmath729 , things are different . because small physical wave number means large length scales .
and gravity accumulates on large length scales .
huge lump of matter makes gravity coupling non - neglectable .
as an explicit example , the @xmath528 and @xmath730 terms in the gravitational action produce non - gaussianities @xmath731 on energy scales @xmath729 and @xmath724 can be estimated as follows : eft approach ( with decoupled gravity ) predicts vanishing squeezed limit . and
on super - hubble scales , the time dependence of @xmath193 decays as @xmath732 ^ 2 $ ] .
thus the gravitational non - gaussianity , coming from @xmath729 , should be at most of order @xmath733 .
and we have checked the local @xmath581 is indeed of this order . ] .
nevertheless , if we are interested in physics of @xmath734 , or non - gaussianity with @xmath735 , it is safe to use the action with gravity ( including the constraints in the gravity sector ) decoupled : @xmath736 eq .
recovers the action , without referring to a background model explicitly .
moreover , in the @xmath234 terms we neglected here ( but can be recovered straightforwardly ) , a much wider class of models are covered , for example , ghost inflation @xcite . by construction
, eft requires in principle a uv completion .
the ultimate uv completion of inflation is not yet known .
but it is interesting to see in toy models how uv degrees of freedoms are integrated out to get the eft in cosmology .
for example , eft operators arise when a very massive scalar is integrated out @xcite .
although we have put the eft approach here as a generalization of single field inflation , the usage of eft is not limited to single field ( or field theory at all ) . instead , the eft approach can be used widely in cosmology , including applications in multi - field inflation @xcite , quasi - single field inflation @xcite and large scale structures @xcite . in the previous subsections
we have discussed how the operators of inflation could be generalized , on the full and perturbation level respectively . however
, physical observables not only depend on operators but also depend on states .
we have argued that it is natural to choose the initially lowest energy state as the vacuum state .
but yet there can be generalizations .
the simplest generalization follows from eq . .
written in terms of the mode function of comoving curvature perturbation @xmath193 , we have @xcite @xmath737 ~.\end{aligned}\ ] ] one can turn on the @xmath738 component , with @xmath739 . in this case
, we have for the power spectrum @xmath740 in case @xmath741 have some phase - sensitive @xmath157-dependence , there could be small oscillations on top of the nearly scale invariant power spectrum .
things become more interesting at the level of non - gaussianity .
we restrict our attention to a small non - bd component @xmath742 . including a factor of @xmath743 effectively flips the sign of @xmath157 ( sign of amplitude , not the vector direction of @xmath383 ) . recall that we had a shape function @xmath744 in eq . of the minimal inflation model . with sign flip of the @xmath745 s in the @xmath746 part of the mode function
is flipped .
the @xmath747 part of the mode function does not change . ] , we get @xmath748 \frac{k_1 ^ 2k_2 ^ 2+k_1 ^ 2k_3 ^ 2+k_2 ^ 2k_3 ^ 2 } { k_1k_2k_3 } \left [ \frac{1}{-k_1+k_2+k_3 } + \frac{1}{k_1-k_2+k_3 } + \frac{1}{k_1+k_2-k_3 } \right]~.\end{aligned}\ ] ] this shape function blows up at the folded triangle limit @xmath749 , or its permutations . this folded shape is plotted in fig .
[ fig : folded ] . in the squeezed limit @xmath750 , @xmath751 , this shape function behaves as @xmath752 \frac{k_2 ^ 2}{k_1 ^ 2 } \frac{2}{1-\alpha^2 } ~,\end{aligned}\ ] ] which blows up dramatically more quickly than the local shape , and depends on how the limit is taken .
however , we need to be careful in interpreting the divergence at the folded limit .
this is because this divergence originates in the uv . on the one hand , it is tricky to define the uv limit of a non - vacuum state . on the other hand , the @xmath474 prescription of selecting the interacting vacuum no longer works for an excited state .
one may put a cutoff by hand to cure those problems phenomenologically @xcite .
however , the prediction is sensitive to the detailed cutoff scheme and thus sensitive to the uv physics @xcite .
an explicit uv completion of physics above the cutoff can be found in @xcite .
a plot of the shape function .,scaledwidth=45.0% ] inflation is attractive as an attractor solution .
and we have implicitly used this nature in various places . in case that during a period of time the attractor behavior is violated , the inflationary perturbations could behave differently .
for example , typically we consider an inflaton field rolling down its potential .
however , what happens when the inflaton use its kinetic energy to climb up a potential ? note that the kinetic energy of the inflaton itself has equation of state @xmath753 .
thus even there is no potential , the kinetic energy dilutes as @xmath754 . with a potential
, the kinetic energy could only dilute faster .
thus we get @xmath755 during the climbing up stage of the potential .
recall the equation of motion for the comoving curvature perturbation : @xmath756 when @xmath757 , @xmath193 effectively live in a contracting universe ( though the actual universe is expanding ) . as a result ,
the conventionally decaying mode of @xmath193 becomes a growing mode and dominates over the constant mode .
one can show that when @xmath758 , the power spectrum is still scale invariant
. however the non - gaussianity becomes @xmath759 , which is not only order one but also breaks the consistency relation of single field inflation @xcite .
one may further enhance this local non - gaussianity by considering a small sound speed @xcite .
qualitatively , the behavior of the perturbations are similar to the case of bouncing cosmology @xcite . here
we generalize inflation towards another direction : including more light fields .
the status of perfect fluid can be described by its energy density @xmath760 , pressure @xmath761 and velocity @xmath762 . on super - hubble scales
@xmath762 decays exponentially thus the status of the perfect fluid is determined by @xmath760 and @xmath761 .
further , if the fluid has only one component , @xmath8 is determined by @xmath9 as @xmath289 . in this case
, we denote the fluctuation of the fluid as _
if it is not the case , we can parameterize @xmath8 as @xmath763 recall that when there is no entropy perturbation , the comoving curvature perturbation is conserved , following eq . , on uniform energy density slice @xmath331 :
@xmath764 thus we can define the non - adiabatic part of pressure according to the derivation from this equation .
consider a general situation without choosing a gauge .
we note : @xmath765 where @xmath234 denotes higher order terms in perturbations .
it is the second term which make @xmath193 change on super - hubble scales .
thus we can define the non - adiabatic part of @xmath766 as @xcite @xmath767 this non - adiabatic part , or entropy part of pressure has clear physical meaning : it is the difference of equation of state , between perturbing the fluid and evolving the fluid in time . when there is such a difference , perturbation in the universe can not be considered as a pure time shift between different local universes .
now we identify the adiabatic and entropy perturbations for inflation . in case
there are multiple rolling fields @xmath768 .
we can rotate the basis in the field space of those fields , such that there is only one rolling field @xcite .
this can be done by considering basis vector @xmath769 where superscript @xmath770 denotes the background level of the field , without perturbations . and the other @xmath495 basis vector are defined as the orthogonal directions .
the rotated fields , including the background and the perturbations , are defined as @xmath771 after this change of basis , the adiabatic direction , or called the inflation direction , is @xmath772 .
all other directions @xmath773 has @xmath774 . in words , those fields in the new basis
then have flat potential thus not rolling .
those non - rolling fields does not contribute energy density or pressure at leading order in field perturbations .
thus at leading order perturbation theory , the fluctuation of those field are entropy perturbations .
this decomposition is illustrated in fig .
[ fig : multi - orth ] . redefining the field directions for multi - field inflation .
the @xmath772 direction is the inflationary direction and the perturbation on @xmath772 direction is the adiabatic perturbation .
the other directions are on the hypersurface perpendicular to @xmath772 .
fluctuations on those directions are entropy perturbations.,scaledwidth=60.0% ] when the inflation trajectory is straight in field space , the basis vectors @xmath775 are time independent and the rotation is trivial .
otherwise the basis vectors @xmath775 are time dependent . for models with standard kinetic term , and at linear order in perturbations ,
@xmath772 is coupled to @xmath776 direction only through a rotation of basis . in the case of straight trajectory
those directions decouple . this is because vanishing @xmath777 and @xmath778 forbid direct coupling . on super - hubble scales ,
a time dependence of @xmath193 can be generated by this turning of trajectory : @xmath779 thus the adiabatic perturbation is sourced by the entropy perturbation along the @xmath780 direction .
note that the @xmath780 direction is perpendicular to @xmath781 .
it is sometimes convenient @xcite to take this direction as @xmath782 : @xmath783 in this case the @xmath782 direction is the source of time dependent @xmath193 on super - hubble scales .
note that those new add - ons of curvature perturbation come from different source from the quantum fluctuation of the original inflaton direction .
thus the power spectrum @xmath371 tends to increase instead of decrease , with the presence of turning of trajectory .
it is in general difficult to suppress the adiabatic perturbation , once it is generated @xcite . before going into detailed models ,
a few comments are in order here : * the adiabatic / entropy perturbations are also called curvature / isocurvature perturbations . * note that we have defined two versions for adiabatic and entropy perturbations : one for fluid and the other for field from rotating basis .
they coincide at first order in perturbation theory , and super - hubble scales . at second order perturbation theory ,
the classification of adiabatic / isocurvature perturbation becomes not unique .
one can consistently use either of them . also on sub - hubble scales ,
even a single field has non - vanishing @xmath784 . * sometimes ( e.g. the curvaton scenario ) , the fields which are rolling much more slowly than the inflaton ( though still rolling thus unlike the non - rolling @xmath785 fields ) are also considered isocurvature directions and generating entropy perturbations .
* we have shown that at linear order , entropy perturbation can convert to the comoving curvature perturbation only if the inflationary trajectory turns . however
, this is under the assumption of standard kinetic term for both fields .
otherwise , kinetic coupling can couple between adiabatic and entropy perturbations without any turning of trajectory .
* at non - linear order , there can be coupling of adiabatic and entropy fluctuations ( when defined as in the field theory way as fig .
[ fig : multi - orth ] ) even without turning of trajectory .
this is because the adiabatic direction is defined at the background level . at the perturbation level ,
the real field trajectory is perturbed slightly @xcite . in the next a few sections , we are to discuss a few models where the entropy perturbation converts to adiabatic perturbation . a summary of those models in light of the planck data can be found in @xcite . in the curvaton mechanism @xcite
, it is assumed that there is a curvaton field @xmath135 , which rolls much more slowly than the inflaton during inflation .
after inflation , the inflaton decays first .
later the curvaton energy density catches up . when the curvaton energy density becomes non - negligible , the entropy fluctuation converts into adiabatic perturbation .
this is illustrated in fig .
[ fig : curvaton ] . the curvaton mechanism.,scaledwidth=95.0% ] for simplicity here we review the simplest scenario of curvaton : the inflaton decays to radiation . and the curvaton oscillate around the minima of its potential @xmath786 before curvaton decay , which happens later than the inflaton decay .
we use the method developed in @xcite . recall the discussion in sec .
[ sec : comov - curv - pert ] , there are separately conserved quantities for non - interacting field species . the conserved quantity associated to the curvaton field is @xmath787 where we have used @xmath788 for a fast oscillating field . here and later in this subsection , we shall omit the superscript ( gi ) on @xmath789 for simplicity .
this conserved quantity provides connection between initial condition and observations .
thus we have to : ( a ) relate the conserved quantity with the horizon crossing value of the curvaton field @xmath790 ; and ( b ) relate the conserved quantity with the curvature perturbation @xmath193 . to relate @xmath789 with @xmath790 , we note that on spatial flat slice ( @xmath791 ) , eq .
can be integrated as @xmath792 the curvaton energy density is @xmath793 where @xmath135 is the amplitude of curvaton oscillation .
however , the quantum initial condition for @xmath135 perturbations are determined at hubble - crossing of the corresponding perturbation modes .
thus we have to relate @xmath135 at the oscillation stage to the value at hubble crossing mode and @xmath258 is not well defined .
however , observationally we are only interested in a small range of @xmath157 .
considering the @xmath135 perturbation is nearly scale invariant , the horizon - crossing value do not differ very much . ] : @xmath794 . at hubble
crossing , the curvaton is a spectator field , thus @xmath795 where @xmath796 is gaussian and has power spectrum @xmath797 the relation between @xmath135 and @xmath790 is model dependent . in general , we can write this dependence as @xmath798 combining , , and , we have is no longer the average energy density because there is also a contribution from @xmath799 . however , it is still consistent and convenient to define the background as @xmath800 instead of @xmath801=0 . ]
@xmath802 ^ 2\end{aligned}\ ] ] where @xmath803 and @xmath804 .
. is a nonlinear relation between the @xmath135-comoving curvature perturbation and the gaussian fluctuation @xmath796 .
we need yet a relation between @xmath193 and @xmath789 . to derive this , we make use of and its counter part for radiation , on uniform total energy density slice . as a result ,
@xmath805 @xmath806 thus we have , on uniform total energy density slice , @xmath807 in case the final adiabatic perturbation is dominated by the curvaton perturbation , we have @xmath808 . in this case , the above equation reduces to @xmath809 where @xmath810 thus finally one can solve @xmath193 from eqs . and order by order in terms of the gaussian perturbation @xmath796 .
as a result , @xmath811 @xmath812 @xmath813~,\end{aligned}\ ] ] where we have omitted the bar on @xmath814 , and @xmath815 with @xmath816 and @xmath817 evaluated at curvaton decay time .
note that in the @xmath818 limit , @xmath819 .
it is intuitive to understand why there is large non - gaussianity in the curvaton scenario : when @xmath818 , the curvaton is sub - dominate .
thus it has to fluctuate harder to have the correct @xmath371 .
the harder curvaton fluctuates , the more non - gaussian it is .
also , the inflationary slow roll parameter @xmath107 does nt enter this later time calculation thus @xmath581 is not @xmath107-suppressed .
this is because the curvaton potential near the curvaton oscillation is not slow roll suppressed .
the other trispectrum estimator is trivial , considering the single - source nature of curvaton mechanism : @xmath643 in the simplest case , if the curvaton has only quadratic potential , there is no nonlinear evolution on super - hubble scales .
thus in this case @xmath820 and @xmath821 . the above formulas simplify considerably in this limit .
otherwise the dynamics can be solved perturbatively or numerically in @xmath822 @xcite .
another way to convert entropy perturbation to adiabatic perturbation is to make use of a non - trivial reheating surface . in this section
we consider a non - dynamical reheating surface but dynamical trajectory due to entropy perturbations .
a dynamical reheating surface is discussed in the next subsection .
non - gaussianity from a non - trivial reheating surface in field space .
when the reheating surface is not perpendicular to the inflationary trajectory , entropy fluctuation converts into e - folding number difference , and thus adiabatic perturbation . ]
conventionally , we assume inflation terminates when the inflation direction @xmath72 ( or @xmath772 after a change of basis ) reach a certain ( range of ) value , where potential is no longer flat or the inflaton field suddenly decays to something else .
however , the above simple picture is not necessarily the full story . in general
, the reheating surface could be a hypersurface in field space , which is not perpendicular to the inflationary trajectory .
the reheating surface may be even curved instead of flat .
such a generalized reheating surface can convert entropy perturbation into curvature perturbation , with a possibility for large non - gaussianity .
this situation is illustrated in fig .
[ fig : reh - suf ] .
an explicit example of non - trivial reheating surface is known as multi - brid inflation @xcite , and more general analysis is performed in @xcite . as shown in the right panel of fig .
[ fig : reh - suf ] , the amount of isocurvature perturbation ( in terms of field space distance ) , converted to curvature perturbation , is characterized by the angle @xmath268 , between the reheating surface and the inflation trajectory : @xmath823 note that @xmath824 .
thus in case the @xmath825 fluctuation is conserved from hubble crossing to reheating , then the adiabatic power spectrum induced by the entropy perturbation is @xmath826 for standard inflaton sector , the total power spectrum is @xmath827 more generally , the @xmath135 field could change its value from hubble crossing to reheating . in this case
the technique in the previous section can be employed to calculate the density perturbation .
the non - gaussianity of this scenario follows from how the reheating surface is curved @xcite .
when the inflaton trajectory hits a convex reheating surface , a positive @xmath581 is created .
this fact can be observed geometrically . for small isocurvature perturbations ,
the curved reheating surface can be approximated as a circle .
consider the left panel of fig .
[ fig : reh - suf - ng ] . here
the green and red solid lines denote different inflation trajectories , with perturbation @xmath828 and @xmath825 respectively .
note that we have @xmath829 . and @xmath830 , where @xmath831 is the center of the circle .
from the construction we have angles @xmath832 ( this can be proved by rotating triangle @xmath833 by angle @xmath834 ) .
thus interval @xmath829 ] recall the observation of section [ sec : local - non - gauss ] .
@xmath836 . thus positive @xmath581 corresponds to an enhancement of positive @xmath587 .
note that the red line ( positive @xmath825 ) is positive @xmath587 ( because of greater @xmath265 ) , and @xmath829 means the second order effects enhance the @xmath265 of the red line compared with the green .
thus when the trajectory hits a convex reheating surface , a positive @xmath581 is produced .
note that if the inflation trajectory is perpendicular to the reheating surface , there is vanishing three - point function because of the symmetry of the configuration .
however , once the trajectory hits the curved reheating surface with an angle , the sign of @xmath581 only depends on how the reheating surface is curved . similarly , hitting a concave surface produces a negative @xmath581 .
a reheating surface without any curvature converts the isocurvature perturbation to curvature perturbation linearly and thus there is no non - gaussianity .
this situation is illustrated in the right panel of fig .
[ fig : reh - suf - ng ] . the curvature of reheating surface ( or any other hypersurface on which entropy perturbation is converted into adiabatic perturbation ) and the sign of non - gaussianity . , scaledwidth=90.0% ] the relation between curved reheating surface and non - gaussianity can also be generalized .
note that the curved reheating surface is nothing but a hypersurface to convert entropy perturbation to adiabatic perturbation .
if during inflation , or after reheating , such conversion happens , the blue surface in fig .
[ fig : reh - suf - ng ] can be considered as the conversion surface from entropy to adiabatic perturbations .
further , it is still possible that the reheating surface ( or other hypersurface with entropy to adiabatic conversion ) is not curved , however the field trajectories themselves are diverging or converging when they hit the surface .
this situation is illustrated in fig .
[ fig : reh - suf - gen ] . finally , it is interesting to generalize the above argument from a `` hypersurface '' of entropy to adiabatic conversion , into some continuous conversion mechanism ( see , e.g. @xcite ) . or consider originally non - gaussian @xmath825 fluctuations ( thus they are no longer @xmath587 ) .
a bundle of diverging trajectories corresponds to positive @xmath581 .
this is because the positive part of @xmath587 converts more and the negative part converts less .
similarly , a bundle of converging trajectories corresponds to negative @xmath581 .
, scaledwidth=80.0% ] in this subsection we again play with the reheating surface of multi - field inflation .
however , instead of the way to hit reheating surface , here we make the shape of the reheating surface dynamically depend on the entropy fluctuations .
this scenario is known as the modulated reheating @xcite .
this situation is illustrated in fig .
[ fig : mod - reh ] . modulated reheating .
note that despite the similarly looking of this figure with fig .
[ fig : reh - suf ] , here the `` / '' direction axis denotes spatial coordinate , instead of field value .
the other axis has the same meaning ( inflaton field value ) . ]
consider that the decay rate of inflaton depends on the entropy field : @xmath837 during reheating , the inflaton energy density @xmath197 decays into radiation @xmath838 .
this process can be described by @xmath839 @xmath840 @xmath841 the above equations are to be solved with initial condition , at a time @xmath842 before the inflaton decay , as @xmath843 and @xmath844 .
the e - folding number during this process can be calculated as @xmath845 where @xmath846 satisfies @xmath847 = \int_0^\infty dn ' \frac{e^{4n'}\gamma}{h(n ) } \frac{\rho_\phi(n')}{\rho_\phi(t_0 ) } ~.\end{aligned}\ ] ] expanding eq .
, one obtains @xmath848 we again assumed extremely slow - rolling of @xmath135 .
otherwise the dynamics of @xmath135 should be tracked similarly to the curvaton scenario .
the non - gaussianities can be calculated by expanding eq . to higher orders @xcite . for the sign of non - gaussianities , the intuition in the last section ( fig .
[ fig : mod - reh ] ) still applies .
now we work in position space instead of field space .
the distribution of @xmath135 field is gaussian .
however , the conversion @xmath849 makes the reheating surface non - gaussianly curved .
the second derivative of @xmath849 determines the sign of non - gaussianity . in this section
we discuss a mechanism similar to the modulated reheating , where the expansion history , instead of the decay rate , is modulated by an isocurvature direction @xcite formalism .
nevertheless , here we focus on the method in @xcite , reformulated following closely the principles of the @xmath265 formalism . ] .
for example , after inflation ends , the inflaton field decays to thermalized particles @xmath117 , with mass @xmath850 .
the reheating temperature is @xmath851 .
the @xmath117 particles are relatively long lived , such that it decays after the temperature of the universe drops to @xmath852 .
thus from reheating to @xmath117-decay , the universe is initially radiation dominated and later matter dominated .
after the @xmath117-decay to lighter particles the universe becomes hot again .
to track the local expansion history of this period , here we use a sudden transition approximation , neglecting the transition period between radiation domination and radiation radiation .
this should be a reasonable approximation , because after @xmath850 varies at different local universes , those different local universes undergoes the same transition ( at different times ) .
thus the error cancels when we calculate the e - folding number difference @xmath265 .
we use @xmath853 and @xmath854 to denote the temperature and hubble parameter at matter radiation equality time , for the @xmath117-particle dominated universe .
the expansion from the initial time to the equality time is @xmath855 and the expansion from the equality time to the final time is @xmath856 the total e - folds is a summation of the two stages : @xmath857 to relate the expansion to the @xmath265-formalism , we have to choose the initial slice to be the flat slice and the final slice to be the uniform energy density slice . assuming the original density fluctuation of the @xmath117 particles ( originates from inflaton fluctuation ) is negligible . in this case
the uniform @xmath455 slice is indeed the initial flat slice . on the other hand , the final uniform @xmath858 slice is the uniform energy density slice because @xmath859 from the friedmann equation .
thus when we vary @xmath860 , @xmath455 and @xmath858 should be fixed .
the only variation comes from @xmath861 . the mass , energy density and hubble parameter can be related as in the energy density for fermions , which do not affect the final result . ]
@xmath862 here we still approximately used the energy density of radiation dominated epoch .
the difference results in an error of the sudden transition approximation .
thus the comoving curvature perturbation can be calculated as @xmath863 now we need to calculate @xmath864 in the @xmath265-formalism . as a simplest case : we may directly have @xmath865 , where @xmath135 is an isocurvature field which gets fluctuations during inflation . in this case , @xmath866 . if @xmath135 is rolling extremely slowly during inflation , such that the hubble - exit value is the same as the value at equality time , the induced power spectrum for curvature perturbation is @xmath867 where @xmath868 is the horizon crossing value of hubble parameter .
similarly , the non - gaussian estimators can be calculated . on the other hand , if @xmath135 is rolling not so slowly , but follow linear evolution ( i.e. the potential is just a mass term ) , @xmath869 is a constant on super - hubble scales . in this case , when @xmath870 , we have @xmath871 and @xmath872 finally , if the @xmath135 potential is dominated by terms other than mass during some periods of time evolution , one can define a function @xmath822 such that @xmath873 , and calculate @xmath822 perturbatively , as is done in the curvaton mechanism . as the last step
, @xmath117 has to decay , preferably to radiation components and re - reheat the universe .
note that the decay rate of @xmath117 also depends on @xmath850 .
there are at least two possibilities : _ no modulation in re - reheating : _ it is possible that the @xmath135 rolls back to origin , or decays , before the decay of @xmath117 . in this case , at @xmath117 decay , the mass @xmath850 no longer depends on @xmath135 .
thus the re - reheating at @xmath117 decay does not introduce extra curvature perturbations .
_ modulated re - reheating : _ in this case , the discussion for the extra curvature perturbation transfer is in parallel to the modulated reheating . as alternative realizations
, one may either modulate the expansion rate at other stages of the universe .
for example , one can modulate the dilution rate of isocurvature direction , to generate entropy perturbation . as another example , during the oscillation of the inflaton , before reheating , the shape of the potential may get modulated . for example , near the end of inflation , the mass term of inflaton may get an anomalous dimension massless . ] : @xmath874~,\end{aligned}\ ] ] as a result , the dilution of inflaton no longer has equation of state @xmath875 , but instead @xmath876 thus the expansion history is modified as a function of the @xmath135 value in local universes .
assuming @xmath877 , @xmath878 where @xmath879 is the number of e - folds of inflaton oscillation before it decays .
the inflationary power spectrum follows from squaring this equation . in the previous subsections ,
we have been making the multi - field inflationary potential complicated .
inspired by the string landscape @xcite , it is argued that inflation may take place on a rather complicated potential @xcite .
yet there is a question : how complicated the inflationary potential could be ? to see the physics more closely , we decomposed the question into a few more concrete ones : once inflation takes place on a complicated multi - field potential , the inflationary trajectory ( white lines in the figure ) may bifurcate .
the bifurcation , if happened during observable state , and not under control , brings deserters.,scaledwidth=60.0% ] * how complicated could a potential be on one field direction ?
note that slow roll of inflationary direction already tightly constraint the potential along the inflation direction .
we postpone the discussion of possible break down of slow roll to subsection [ sec : scale - depend - feat ] . *
how many fields could there be during inflation ?
the constraint @xcite is quite weak .
nevertheless a weak constraint is stronger than no constraint @xcite .
+ note that during inflation , each massless ( actually @xmath880 should be enough ) field has a fluctuation @xmath881 per hubble volume .
thus the total energy density of @xmath335 massless fields is @xmath882 this must be much smaller than the background energy density @xmath9 .
thus we have @xmath883 this constraint can also be argued as that , near hubble crossing , the quantum perturbation of each massless field becomes classical and carry of order one bit entropy ( about its phase ) .
however , de sitter space has an entropy bound of @xmath884 .
thus the field number is bounded by .
+ note that this number is greater than @xmath21 .
thus most models are unconstrained , unless one wants the number of fields to be greater than @xmath885 or @xmath886 . *
what extras could a multi - field complicated potential bring compared with a single field one ? + as we already discussed , in multi - field inflation , much more things could happen compared to the case of single field inflation . here regarding to a complicated potential , there is at least one more : the classical trajectory of inflation may bifurcate during inflation @xcite .
inflation with such a bifurcation is called multi - stream inflation , as illustrated in fig .
[ fig : multi - stream ] .
+ in case the bifurcation of inflationary trajectory is under control ( with symmetry or fine - tuning ) , those bifurcations can be responsible to non - gaussianities @xcite or position space features such as asymmetries and/or cold spots on the cmb @xcite . or alternatively the bifurcation could happen before the observable stage of inflation @xcite . however , if the bifurcation is due to a random potential , the e - folding number difference between different trajectories are not under control . in this case
the cmb sky becomes extremely anisotropic and non - gaussian .
this kind of models should be ruled out . + it is shown that in a double - field random potential , bifurcation could happen with some not - so - fine tuning of potential parameters @xcite .
when the random potential has more dimensions , the probability for bifurcation should increase thus there may be a tighter constraint .
+ there is also a danger of rapid tunneling in a random potential , as discussed in @xcite .
we have discussed single field inflation and multi - field inflation .
definitely we can not cover even a small portion of the huge literature .
nevertheless at first look the classification of single / multiple field covers all possibilities .
however , it is actually not the case .
note that inflation is a theory with multiple energy scales built in .
the inflationary background defines an energy scale @xmath887 , whereas the perturbations defines another energy scale @xmath888 .
there is a hierarchy between those two energy scales .
the hierarchy of energy scales allows a class of inflation models , in which at the homogeneous and isotropic background level , the models look like single field inflation .
however , when we consider cosmic perturbations , which live at a higher energy scale , the physics becomes different from single field .
this class of models is called quasi - single field inflation @xcite@xcite@xcite .
for example , consider a scalar field @xmath135 with mass @xmath130 during inflation ( fig .
[ fig : qsfi - scale ] ) . at the background level
, this field stays at its minimum of the potential , because the mass is higher than the background energy scale and not slow rolling .
thus at the background level we have single field inflation .
however , at the energy scale of perturbations , @xmath130 is not too heavy .
it takes quite a few e - folds for the @xmath135 perturbation to roll back to the origin . as a result
, there is room for the @xmath135 perturbation , which appears as entropy perturbation , to convert to curvature perturbation .
energy scales during inflation .
, scaledwidth=80.0% ] note that the potential of the @xmath135 field is not restricted by slow roll condition .
thus the potential need not to be very flat . as a result , the self interactions of @xmath135 can be strong . when the entropy perturbations in the @xmath135 direction converts to curvature perturbation , the non - gaussianity of @xmath135 , from the self interactions , are also converted .
the generation and conversion of non - gaussianity is near hubble crossing .
this is different from the local non - gaussianity ( which is generated on super - hubble scales ) or equilateral non - gaussianity ( which is generated on sub - hubble scales ) . as a result , a family of new shapes of non - gaussianity arises , controlled by the mass @xmath850 of @xmath135 . as we discussed in sec .
[ sec : open - questions ] , there is an @xmath100-problem of inflation : originally massless fields ( unless protected by symmetry ) get uplifted to @xmath130 during inflation .
note that this is true not only for the inflaton field , but also for other originally massless fields .
if we have some originally massless fields in field theory . now inflationary back - reaction typically uplift them to @xmath130
. some of them may have @xmath119 by pure chance fields .
for example , assuming the mass of the field obeys gaussian distribution @xmath889 $ ] , one needs @xmath890 $ ] fields to naturally have one or more of them lying in range @xmath891 . for @xmath892 ,
16 fields are needed ; for @xmath893 , 31 fields are needed ; for @xmath894 , @xmath21 fields are needed . ] or some fine tuning .
they can behave as the inflaton , because we want inflation .
the other fields has @xmath130 .
a string theory realization of quasi - single field inflation is considered in @xcite . among those @xmath130 fields ,
some may decouple from inflaton .
the fluctuation of those fields are diluted after a few e - folds of super - hubble evolution and thus not observable .
however , others may couple to inflaton , via operators @xcite @xmath895 it is helpful to consider a simple model to see explicitly how the above argument works @xcite@xcite . to do that
, we consider a model in which the inflation trajectory is along an arc .
the angular direction is the inflaton and the radius direction is the massive field @xmath135 .
we illustrate the case in fig .
[ fig : qsfi ] . a simple example of quasi - single field inflation .
the isocurvature direction is not slow roll restricted .
thus the entropy fluctuation can be non - gaussian . during the turning of trajectory
, the non - gaussian entropy fluctuation is converted to the adiabatic perturbation , along the inflation direction.,scaledwidth=50.0% ] for simplicity , we further assume the potential for the two scalars can be written in the factorized form when written in the polar coordinates . in this case , we can write down the theory in polar coordinates as @xmath896~,\end{aligned}\ ] ] where @xmath897 is the slow roll potential in the inflation direction ; @xmath898 is the potential in the @xmath899 direction , without slow roll restriction ; and @xmath900 is the radius of the turning potential .
gravity minimally couples to this matter sector .
the background equations of motion are @xmath901 where @xmath47 is the radius of turning trajectory , which has a shift by a constant compared with @xmath902 due to the centrifugal force : @xmath903 , such that @xmath904 . as we have shown in the general discussion of adiabatic and entropy perturbations , the turning of trajectory couples the adiabatic and entropy directions .
this can be verified in the present model by expanding in perturbations .
the interaction is @xmath905 the importance of this term can be estimated as follows : near hubble - crossing , the canonically normalized field @xmath906 has fluctuation of order @xmath907 .
the @xmath825 field is massive thus has amplitude @xmath908 .
the mass dependence is because when @xmath909 , the @xmath825 fluctuation is suppressed @xcite by @xmath910 , which is the flat space estimation of propagator . to estimate how strong is the coupling @xmath911 , we can compare it with the free lagrangian @xmath912 .
we have @xmath913 because of the transfer vertex , the power spectrum gets a relative correction proportional to @xmath914 .
this situation is illustrated in fig .
[ fig : qsfi - int ] . as a result
, the power spectrum goes as @xmath915 where @xmath916 is the relative correction due to the transfer vertex .
the analytical expression of @xmath916 is given as @xmath917 in @xcite .
the @xmath918 estimate in should apply for both @xmath130 and @xmath919 .
thus in those cases , up to @xmath920 factors , @xmath921 on the other hand , when @xmath119 , there is a divergence that worth noticing .
it can be shown that in this limit , if one take the bounds in the in - in integral as @xmath922 , @xmath916 ends up with a divergence in the small @xmath850 limit : @xmath923 .
however this divergence is not real because inflation will eventually end .
if we instead take @xmath413 and cut the in - in integral as @xmath924 , we have @xmath925 .
thus in the @xmath119 limit , we have @xmath926 where we have assumed @xmath927 e - folds of observable inflation .
it turns out that when @xmath928 ( which is actually not much smaller than h ) , we will already have to consider the e - folding number limit of inflation .
the ir growth of @xmath119 case is straightforward to understand : for a massless field , when there is a fluctuation , the fluctuation will continues interacting with the adiabatic perturbation all the time until the interaction is turned off ( for example , reheating ) . in flat space qft ,
the cross sections and decay rates are defined as a finite value per unit time .
thus there is no wonder the interaction rate diverges with a diverge duration of time .
the transfer vertex in quasi - single field inflation , and the correction to the power spectrum , due to the transfer vertex.,scaledwidth=90.0% ] the amplitude of non - gaussianity can be estimated similarly to the power spectrum for quasi - single field inflation .
the bispectrum can be estimated by the coupling @xmath929 note that @xmath544 has mass dimension .
thus it is natural to have @xmath930 during inflation , similar to the @xmath130 argument . and again similarly to the @xmath100-problem , the @xmath544 term may have uv sensitivity driving @xmath931 straightly to the uv - sensitivity scale .
one can estimate @xmath932 here we shall focus on the weakly coupling case @xmath933 , leaving the more general case to sec . [ sec : general - quasi - single ] .
we should note that this interaction @xmath934 is huge , compared with the inflaton counterpart , which is offered by a slow roll potential @xmath935 . as shown in eq .
, the slow roll potential is suppressed by @xmath936 and we know this @xmath937 generates sub - dominate non - gaussianity for single field inflation with amplitude @xmath938 . thus we immediately know that the @xmath939 in the entropy perturbations should be @xmath940 , which could be huge considering @xmath941 . here
we obtain the @xmath942 factor by comparing with the single field potential .
the same estimation could either be obtained as follows : the three point function is @xmath943(coupling constant ) . on the other hand ,
the bispectrum estimator @xmath581 is multiplied with @xmath569 in its definition . to compare them we need to divide the result by @xmath944 .
now we have got the amplitude of non - gaussianity in the isocurvature direction . however
, this entropy perturbation needs to be converted to adiabatic perturbation .
we have shown the conversion is mediated by @xmath918 coupling . as a result
, the amplitude of non - gaussianity for quasi - single field inflation can be estimated as @xmath945 this amplitude is illustrated in fig .
[ fig : qsfi - int3 ] .
the interaction vertex and the three point correlation function.,scaledwidth=63.0% ] similarly , the analogue of @xmath600 and @xmath601 can also be estimated . as we have shown in sec .
[ sec : soft - limit - internal ] , @xmath601 is dominated by the diagram with an internal propagator , and @xmath600 is dominated by the other diagram .
thus we have @xmath946 when @xmath918 is small ( which is the case we can trust our calculation ) , we have @xmath947 thus the suyama - yamaguchi inequality is satisfied . the graphical understanding is given in fig .
[ fig : qsfi - sy ] .
one can also estimate the higher order correlation functions @xcite .
the @xmath601 analogue of the n - point function scales as @xmath948 where we have factorized the product into two groups , one scales with power @xmath949 and another scales with @xmath335 .
those two groups follows different interaction structures : * the power @xmath949 group shows how the non - gaussianity is generated .
this is because for three point interaction ( which generates @xmath950 , and @xmath600 belongs to yet another case ) , we need @xmath949 vertices to have @xmath335-point function . * the power @xmath335 group shows how the non - gaussianity is converted from entropy direction to curvature direction .
this is because for @xmath335-point function , we need at least @xmath335 transfer vertices to convert all external legs to adiabatic perturbation . this scaling behavior is discussed in more details in @xcite , known as the feeding mechanism .
the quasi - local analog of @xmath600 ( left panel ) and @xmath601 ( right panel).,scaledwidth=70.0% ] the suyama - yamaguchi relation .
here we have @xmath951 , as long as the transfer vertex is perturbative . in case @xmath952 , our perturbative calculation can not be trusted .
however , the suyama - yamaguchi relation should still hold .
the implication is that , resummation should suppress the large coupling.,scaledwidth=99.0% ] now comes the shape of non - gaussianity . to get some intuition ,
we may consider two limits : * when @xmath119 , quasi - single field inflation approaches multi - field inflation and the shape should approach local shape .
this is because when there is an entropy fluctuation , in the massless limit the fluctuation do not roll back . thus the interaction continues on super - hubble scales . from causality ,
the non - gaussianity should be dominated by the local shape .
* when @xmath421 , the kinetic term of @xmath135 becomes hard to excite .
thus the heavy field just stays at the effective minimal of the potential , determined by the centrifugal force @xmath953 .
to perturb this equation , additional factors of @xmath954 comes into the action of @xmath955 , which modifies the sound speed @xmath682 of @xmath955 perturbations . as a result
, the non - gaussianity should be very non - local . in general
, we expect the shape should be in between the above cases . in the squeezed limit @xmath956 , the @xmath157-dependence of the three - point function
can be calculated explicitly @xcite . here
we will not go into the details . instead ,
as observed in @xcite , the squeezed limit of quasi - single field inflation can be estimated as follows : the long wave length mode crosses the horizon at @xmath957 .
later , the short wave length mode crosses the horizon at @xmath958 . between time @xmath959 and @xmath960
, the massive field @xmath135 has a decaying solution : @xmath961 where @xmath962 . in the squeezed limit , @xmath963 . at @xmath960 , the @xmath610 mode
can still be considered as a shift of background .
but the shift has a decay factor . as a result , @xmath964
this analysis can also be made more rigorous @xcite .
also , in the special case @xmath965 , a logarithm factor appears in the squeezed limit @xcite .
the exact shape of three - point function is quite complicated and plotted in @xcite .
phenomenologically , the dimensionless shape function can be approximated by @xmath966 this shape function is simple , but unfortunately not factorized .
one can nevertheless write a factorized shape near the squeezed limit @xcite .
the shape function is plotted in fig .
[ fig : qsfi - shapes ] .
it is interesting to notice that , the shape function for @xmath967 already looks very similar to the local shape .
indeed , a fisher matrix analysis shows that the @xmath968 shape and the local shape are not distinguishable in the next 10 years experiments @xcite ( see also @xcite ) . in case the local shape
non - gaussianity fits future data , it is challenging to distinguish if the non - gaussianity comes from a massless field , or from a @xmath968 field and the corresponding quasi - local shapes . on the other hand ,
shapes from greater mass ( such as @xmath969 ) will be distinguishable from local shape . the shapes of non - gaussianity for quasi single field inflation , for @xmath970 ( i.e. @xmath965 ) , @xmath971 ( i.e. @xmath972 ) , @xmath973 ( i.e. @xmath974 ) and @xmath967 ( i.e. @xmath975 ) respectively .
the shape ansatz is used . note that the change of shapes are very sensitive to the @xmath135 mass in hubble unit.,title="fig:",scaledwidth=45.0% ] the shapes of non - gaussianity for quasi single field inflation , for @xmath970 ( i.e. @xmath965 ) , @xmath971 ( i.e. @xmath972 ) , @xmath973 ( i.e. @xmath974 ) and @xmath967 ( i.e. @xmath975 ) respectively .
the shape ansatz is used . note that the change of shapes are very sensitive to the @xmath135 mass in hubble unit.,title="fig:",scaledwidth=45.0% ] + the shapes of non - gaussianity for quasi single field inflation , for @xmath970 ( i.e. @xmath965 ) , @xmath971 ( i.e. @xmath972 ) , @xmath973 ( i.e. @xmath974 ) and @xmath967 ( i.e. @xmath975 ) respectively .
the shape ansatz is used . note that the change of shapes are very sensitive to the @xmath135 mass in hubble unit.,title="fig:",scaledwidth=45.0% ] the shapes of non - gaussianity for quasi single field inflation , for @xmath970 ( i.e. @xmath965 ) , @xmath971 ( i.e. @xmath972 ) , @xmath973 ( i.e. @xmath974 ) and @xmath967 ( i.e. @xmath975 ) respectively .
the shape ansatz is used . note that the change of shapes are very sensitive to the @xmath135 mass in hubble unit.,title="fig:",scaledwidth=45.0% ] resumming the transfer vertex .
note that the above diagrams represents a set of integral equations .
when @xmath544 and/or other interactions are turned on , the resummation becomes more complicated but yet possible.,scaledwidth=95.0% ] in the above discussion of quasi - single field inflation , we have restricted our attention to the weakly coupled case . in the strongly coupled case ,
perturbative methods of directly calculation breaks down .
however , one can still try to resum the diagrams , seeking a weakly coupled effective field theory .
for example , the transfer vertex can be in principle resummed using the schwinger - dyson equation , as illustrated in fig .
[ fig : qsfi - resum ] .
alternatively , one can generalize the weakly coupled picture without doing the resummation . especially for a conformal field theory ( cft ) the interaction structure
is highly constrained by symmetry @xcite . in cft ,
mass and interaction operators are characterized by their scaling dimensions . in this case , the squeezed limit of correlation functions are as functions of the scaling dimensions of the operators .
the scaling dimension of the mass operator is an analogue of the example of sec .
[ sec : simple - model ] .
however , a non - trivial scaling dimension of interaction operators open up new possibilities . for example , one can imagine a sector of isocurvature perturbations which is strongly coupled on sub - hubble scales , but weakly coupled on super - hubble scales .
the transition is controlled by the rg - flow and the anomalous dimension of the interaction operator . in this case
the anomalous dimension dynamically generates a scale , analogous to @xmath976 .
one can compare this scale to the hubble scale , and generate a one parameter family of non - gaussianities .
new possibilities such as the nearly orthogonal shape arise compared to the simple model that we reviewed in detail .
so far the inflation models we have reviewed are nearly scale invariant .
such models are natural because the observed power spectrum is nearly scale invariant .
however , it is still possible that there are small ripples in the inflationary power spectrum , that we have not yet observed . or the observational effects hide in higher order correlation functions or gravitational waves , which we havent yet observed . here
we review a few models of this kind .
consider single field inflation , where the potential is approximately flat but there are small features on top of that .
for example , one can model a feature as @xcite @xmath977~.\end{aligned}\ ] ] the potential is illustrated in fig .
[ fig : feature - pot ] . a feature in the potential .
the solid line denote the actual feature considered in @xcite , with @xmath978 and @xmath979 .
the feature is actually too small to be observable on the plot .
the dot line is just an illustration of the feature , choosing by hand a huge feature ( @xmath980 ) , which is clearly ruled out by experiments.,scaledwidth=50.0% ] alternatively , one can choose a feature with a non - continuous slope @xcite : @xmath981 a sharp feature can in general enhance slow roll parameters .
the enhanced slow roll parameters result in a feature in the power spectrum , as well as large non - gaussianity .
however , it is useful to estimate which slow roll parameter is enhanced most . considering the sharpness of the feature ,
higher time derivatives get enhanced most . in the third order action of standard single field inflation ,
the highest derivative is @xmath982 .
thus the interaction is dominated by @xmath983 the shape of the bispectrum can be calculated numerically .
it is intuitive to see that the non - gaussianity is now scale dependent .
this is because the far sub - hubble modes has time to relax back to the vacuum and the far super - hubble modes are conserved already .
the most - affected modes are the hubble crossing ones . in the bi - spectrum , they are the ones @xmath984 , where @xmath985 is the time when the inflaton rolls onto the feature . as a result @xcite , the shape function is modulated by an oscillation @xmath986 $ ] . here
we review a sharp turning of inflationary trajectory @xcite .
for example , there is one inflation direction , and another very massive direction .
the turning of trajectory is between those two directions , as illustrated in fig .
[ fig : sharp ] . a sharp turning of trajectory.,scaledwidth=70.0% ] during a sharp turning
, a combination of effects may happen : * _ background oscillation : _ when the trajectory turns very sharply , the kinetic energy of the inflaton gets projected on to the massive direction .
the amount of transfer energy is @xmath987 , where @xmath988 is the turning angle .
this amount of energy drives the massive direction @xmath135 to move as high as @xmath989 . in case this offset in the @xmath135 direction
is not cancelled by a linear potential , the @xmath135 field get released from this height and start to oscillate .
+ the background oscillation is not just a shift of background .
instead more dramatic things can happen .
note that the perturbations has varying physical wave length @xmath725 .
once the @xmath725 for a perturbation mode gets close to the oscillation frequency of the background , parametric resonance can happen .
this situation is similar to the case of preheating . as a result , both the linear perturbation and non - gaussianity
are enhanced .
+ it is interesting to note that , both the background oscillation itself , and the induced effects such as the resonant enhancement of perturbations , are periodic ( with a decaying envelope ) with respect to physical time . on the other hand , the comoving wave number @xmath157 is defined with respect to the conformal time .
thus in case one observes such a periodic feature in the sky , the function dependence between the physical time and conformal time is obtained .
this function dependence differs significantly between inflation and alternative to inflation models @xcite .
+ in case the turning of trajectory is not sharp enough , or a linear potential comes to cancel the centrifugal force during the turning of trajectory , the background oscillation may not be present . * _ sharp feature : _ during the turning of trajectory , the slow roll parameters got boosted thus there is an similar effect as discussed in sec .
[ sec : scale - depend - feat ] . * _ vacuum projection : _ the wave function of a quantum harmonic oscillator depends on the frequency of the oscillator .
the same is true for scalar fields .
the wave function of a massive direction is different from the wave function of a massless direction .
however , after a sharp turning of trajectory , the two wave functions gets mixed . as a result , the perturbations of the fields are no longer in the vacuum state .
+ in terms of @xmath157-modes , the perturbation modes with @xmath990 are affected .
the mixing , which can be calculated using the in - in formalism , kicks the perturbations even if there is no background oscillation . despite the wording `` slow roll ''
, the inflaton field actually runs very fast , with @xmath991 .
thus it is useful to check if the fast motion of the inflaton pollutes the inflationary environment .
for example , when other fields have coupling to the inflaton , there may be particle productions induced by the motion of the inflaton @xcite .
for example , consider an otherwise massless field @xmath117 , with the interaction lagrangian @xmath992 the motion of the inflaton provides an effective mass for @xmath117 .
the equation of motion for @xmath117 is @xmath993 when the change of background is slow , the @xmath117 field has a wkb solution . as a result @xmath117 almost stays unchanged without particle production .
however , when @xmath72 moves cross @xmath994 , the situation changes dramatically . not only that @xmath117 gets effectively massless , but also the wkb condition breaks badly : @xmath995 thus particles of @xmath117 get produced .
to see the effect more closely , one can parameterize the behavior of @xmath72 near @xmath994 as @xmath996 . solving the eom of @xmath117 ,
the occupation number can be calculated as @xmath997 after the moment @xmath998 , the produced particles quickly become massive .
the mass increases while the number density is diluted .
the back - reaction from the produced particles to the adiabatic perturbation can be calculated .
the power spectrum gets a correction . for the bispectrum , for small @xmath157
the shape of non - gaussianity is similar to local shape . for larger @xmath157
there are oscillations in the shape function @xcite .
it is also interesting to put this scenario onto a landscape type potential .
when the number of fields increase , it is exponentially likely for the inflation field to encounter particle production events ( with a small mass of @xmath117 ) @xcite .
i would like to thank my collaborators niayesh afshordi , robert brandenberger , yi - fu cai , bin chen , xingang chen , james cline , francis duplessis , xian gao , jinn - ouk gong , a.emir gumrukcuoglu , min - xin huang , bin hu , qing - guo huang , miao li , chunshan lin , alexander maloney , guy moore , shinji mukohyama , a. riotto , gary shiu , ane slosar , tower wang and wei xue for collaborations and discussions on inflation and cosmic perturbations .
i thank valentin assassi , daniel baumann , xingang chen , and david seery for discussions during writing this review .
i thank xingang chen for helpful comments on a draft of this review , and thank evan mcdonough for pointing out typos in a previous version .
this work is supported by fundings from the kavli institute for the physics and mathematics of the universe ( kavli ipmu ) , and from the japan society for the promotion of science ( jsps ) .
although the physics of inflationary perturbations is interesting , the calculations are sometimes tediously long .
fortunately , nowadays computers are in the position to assist us for this purpose .
there are quite a few choices of algebra systems in the market . among those , there are propriety software such as mathematica and maple , and free software such as sage and maxima . here
we comment the pros and cons for our particular purpose ( there are of course other choices as well ) : * _ mathematica : _ the functional programming paradigm is typically convenient for math calculations .
also there is a powerful mechanism of pattern matching , related to its functional programming nature .
there are also gr packages such as xact available on mathematica . *
_ maple : _ there is a very powerful package on general relativity : grtensor , on the maple platform . * _
sage : _ sage is a collection of free math software .
those lower level software use different languages and conventions .
and sage unifies them with an interface using the python programming language .
python is strong in object - oriented programming , in contrast with the functional programming approach taken by mathematica . *
_ maxima : _ currently sage use maxima as back - end . in the analytical aspects ,
maxima itself is feature complete , if one would not need the conveniences provided by python .
one annoying fact is that computers are usually not clever enough to combine quantities in desired forms .
especially they typically do not collect summed indices .
algebraic systems as mathematica will leave you something like @xmath999 instead of @xmath1000
. there are at least two tricks one can do for this : * _ replace the dummy indices : _ we can define a replacement rule + .... expression /.
{ d[f_,x1,x1 ] - > p2[f ] - d[f , x2,x2 ] - d[f , x3,x3 ] } .... + note the underscore here .
this underscore matches and replaces @xmath1001 acting on all variables , not only f itself .
we will also meet a lot of other combinations that we need to design rules to match and replace . *
_ keep them abstract from the beginning : _ for most of the cases , we do nt need to actually calculate the summation of spatial indices , we just need to write them in a compact form . for this purpose , we can directly write @xmath1000 in the code . it is ( typically ) elegant and not hard to do this .
we have to note : * * the default derivative operator tries to calculate @xmath1002 explicitly .
this is not the desired behavior when we decide to keep @xmath1000 as abstract . to solve this problem
, we can define derivative operators ourselves .
the definition should include linearity and the leibniz s rule for multiplication , without defining how to calculate @xmath1002 explicitly .
it is also convenient to use the `` notation '' package to overload the @xmath1003 operator . *
* dummy indices may conflict .
for example , how does mathematica write @xmath1004 as @xmath1005 instead of as @xmath1006 ?
for this purpose , the unique [ ] function is useful .
we can write + ....
( \partial _ # \partial _ # f)&[unique [ ] ] * ( \partial _ # \partial _ # f)&[unique [ ] ] .... + then mathematica automatically assign different dummies for different terms .
it is also not hard to write a function to simplify dummies . in the code calculating the action of perturbations , one frequently need to integrate by parts . the situations can be divided into 2nd order or higher order actions , and spatial or temporal directions . for 2nd order spatial direction , one can go to momentum space to get rid of integration by parts .
however , for higher order , or temporal direction , going to momentum space does not help that much .
one general way to do integration by parts is to use the mathematica function replacelist [ ] .
this function tries all possible replace rules ( which we write ourselves ) .
we just need to setup some measure to compare the result of each replacement and find a good one .
we can repeatedly call replacelist [ ] ( or multiple replacelist [ ] in a row ) until the result is simplest
. a. h. guth , `` the inflationary universe : a possible solution to the horizon and flatness problems , '' phys .
d * 23 * , 347 ( 1981 ) .
a. d. linde , `` a new inflationary universe scenario : a possible solution of the horizon , flatness , homogeneity , isotropy and primordial monopole problems , '' phys .
b * 108 * , 389 ( 1982 ) .
a. albrecht and p. j. steinhardt , `` cosmology for grand unified theories with radiatively induced symmetry breaking , '' phys .
lett . * 48 * , 1220 ( 1982 ) . v. f. mukhanov and g. v. chibisov , `` quantum fluctuation and nonsingular universe .
( in russian ) , '' jetp lett .
* 33 * , 532 ( 1981 ) [ pisma zh .
fiz . * 33 * , 549 ( 1981 ) ] .
s. w. hawking , `` the development of irregularities in a single bubble inflationary universe , '' phys .
b * 115 * , 295 ( 1982 ) .
a. a. starobinsky , `` dynamics of phase transition in the new inflationary universe scenario and generation of perturbations , '' phys .
b * 117 * , 175 ( 1982 ) .
a. h. guth and s. y. pi , `` fluctuations in the new inflationary universe , '' phys .
* 49 * , 1110 ( 1982 ) .
j. m. bardeen , p. j. steinhardt and m. s. turner , `` spontaneous creation of almost scale - free density perturbations in an inflationary universe , '' phys .
d * 28 * , 679 ( 1983 ) .
h. kodama and m. sasaki , `` cosmological perturbation theory , '' prog .
* 78 * , 1 ( 1984 ) .
v. f. mukhanov , h. a. feldman and r. h. brandenberger , `` theory of cosmological perturbations .
classical perturbations .
quantum theory of perturbations .
extensions , '' phys .
rept . * 215 * , 203 ( 1992 ) .
e. komatsu , `` the pursuit of non - gaussian fluctuations in the cosmic microwave background , '' astro - ph/0206039 .
a. riotto , `` inflation and the theory of cosmological perturbations , '' hep - ph/0210162 .
n. bartolo , e. komatsu , s. matarrese and a. riotto , `` non - gaussianity from inflation : theory and observations , '' phys .
rept . *
402 * , 103 ( 2004 ) [ astro - ph/0406398 ] .
v. mukhanov , `` physical foundations of cosmology , '' cambridge , uk : univ .
( 2005 ) 421 p d. wands , `` multiple field inflation , '' lect .
notes phys .
* 738 * , 275 ( 2008 )
[ astro - ph/0702187 [ astro - ph ] ] . s. weinberg , `` cosmology , '' oxford , uk : oxford univ . pr .
( 2008 ) 593 p k. a. malik and d. wands , `` cosmological perturbations , '' phys . rept . *
475 * , 1 ( 2009 ) [ arxiv:0809.4944 [ astro - ph ] ] .
w. h. kinney , `` tasi lectures on inflation , '' arxiv:0902.1529 [ astro-ph.co ]
. d. baumann , `` tasi lectures on inflation , '' arxiv:0907.5424 [ hep - th ] .
m. liguori , e. sefusatti , j. r. fergusson and e. p. s. shellard , `` primordial non - gaussianity and bispectrum measurements in the cosmic microwave background and large - scale structure , '' adv .
astron . *
2010 * , 980523 ( 2010 ) [ arxiv:1001.4707 [ astro-ph.co ] ] .
d. langlois , `` lectures on inflation and cosmological perturbations , '' lect .
notes phys .
* 800 * , 1 ( 2010 ) [ arxiv:1001.5259 [ astro-ph.co ] ] .
x. chen , `` primordial non - gaussianities from inflation models , '' adv .
astron . * 2010 * , 638979 ( 2010 ) [ arxiv:1002.1416 [ astro-ph.co ] ] .
j. b. hartle and s. w. hawking , `` wave function of the universe , '' phys .
d * 28 * , 2960 ( 1983 ) .
a. d. linde , `` quantum creation of an inflationary universe , '' sov .
jetp * 60 * , 211 ( 1984 ) [ zh . eksp . teor .
fiz . * 87 * , 369 ( 1984 ) ] .
a. vilenkin , `` quantum creation of universes , '' phys .
d * 30 * , 509 ( 1984 ) .
g. f. smoot , c. l. bennett , a. kogut , e. l. wright , j. aymon , n. w. boggess , e. s. cheng and g. de amici _ et al .
_ , `` structure in the cobe differential microwave radiometer first year maps , '' astrophys .
j. * 396 * , l1 ( 1992 ) . c. l. bennett , d. larson , j. l. weiland , n. jarosik , g. hinshaw , n. odegard , k. m. smith and r. s. hill _
_ , `` nine - year wilkinson microwave anisotropy probe ( wmap ) observations : final maps and results , '' arxiv:1212.5225 [ astro-ph.co ] .
g. hinshaw , d. larson , e. komatsu , d. n. spergel , c. l. bennett , j. dunkley , m. r. nolta and m. halpern _ et al .
_ , `` nine - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological parameter results , '' arxiv:1212.5226 [ astro-ph.co ] .
s. das , t. louis , m. r. nolta , g. e. addison , e. s. battistelli , j r. bond , e. calabrese and d. c. m. j. devlin _ et al .
_ , `` the atacama cosmology telescope : temperature and gravitational lensing power spectrum measurements from three seasons of data , '' arxiv:1301.1037 [ astro-ph.co ] . j. dunkley , e. calabrese , j. sievers , g. e. addison , n. battaglia , e. s. battistelli , j. r. bond and s. das _
_ , `` the atacama cosmology telescope : likelihood for small - scale cmb data , '' arxiv:1301.0776 [ astro-ph.co ] .
j. l. sievers , r. a. hlozek , m. r. nolta , v. acquaviva , g. e. addison , p. a. r. ade , p. aguirre and m. amiri _ et al .
_ , `` the atacama cosmology telescope : cosmological parameters from three seasons of data , '' arxiv:1301.0824 [ astro-ph.co ] . k. t. story , c. l. reichardt , z. hou , r. keisler , k. a. aird , b. a. benson , l. e. bleem and j. e. carlstrom _ et al .
_ , `` a measurement of the cosmic microwave background damping tail from the 2500-square - degree spt - sz survey , '' arxiv:1210.7231 [ astro-ph.co ] .
z. hou , c. l. reichardt , k. t. story , b. follin , r. keisler , k. a. aird , b. a. benson and l. e. bleem _ et al .
_ , `` constraints on cosmology from the cosmic microwave background power spectrum of the 2500-square degree spt - sz survey , '' arxiv:1212.6267 [ astro-ph.co ] .
m. li , x. -d .
li , s. wang and y. wang , `` dark energy , '' commun .
* 56 * , 525 ( 2011 ) [ arxiv:1103.5870 [ astro-ph.co ] ] .
s. r. coleman , `` the fate of the false vacuum .
semiclassical theory , '' phys .
d * 15 * , 2929 ( 1977 ) [ erratum - ibid .
d * 16 * , 1248 ( 1977 ) ] .
s. r. coleman and f. de luccia , `` gravitational effects on and of vacuum decay , '' phys .
d * 21 * , 3305 ( 1980 ) .
f. lucchin and s. matarrese , `` power law inflation , '' phys . rev .
d * 32 * , 1316 ( 1985 ) . g.
w. gibbons , s. w. hawking and j. m. stewart , `` a natural measure on the set of all universes , '' nucl .
b * 281 * , 736 ( 1987 ) .
g. w. gibbons and n. turok , `` the measure problem in cosmology , '' phys .
d * 77 * , 063516 ( 2008 ) [ hep - th/0609095 ] .
m. li and y. wang , `` the measure for the multiverse and the probability for inflation , '' jcap * 0706 * , 012 ( 2007 ) [ arxiv:0704.1026 [ hep - th ] ]
. b. a. bassett , s. tsujikawa and d. wands , `` inflation dynamics and reheating , '' rev .
phys . * 78 * , 537 ( 2006 ) [ astro - ph/0507632 ] .
r. allahverdi , r. brandenberger , f. -y .
cyr - racine and a. mazumdar , `` reheating in inflationary cosmology : theory and applications , '' ann .
nucl . part .
sci . * 60 * , 27 ( 2010 ) [ arxiv:1001.2600 [ hep - th ] ] .
l. f. abbott , e. farhi and m. b. wise , `` particle production in the new inflationary cosmology , '' phys .
b * 117 * , 29 ( 1982 ) .
a. d. dolgov and a. d. linde , `` baryon asymmetry in inflationary universe , '' phys .
b * 116 * , 329 ( 1982 ) .
a. albrecht , p. j. steinhardt , m. s. turner and f. wilczek , `` reheating an inflationary universe , '' phys .
lett . * 48 * , 1437 ( 1982 ) .
a. sen , `` tachyon matter , '' jhep * 0207 * , 065 ( 2002 ) [ hep - th/0203265 ] .
g. shiu , s. h. h. tye and i. wasserman , `` rolling tachyon in brane world cosmology from superstring field theory , '' phys .
d * 67 * , 083517 ( 2003 ) [ hep - th/0207119 ] .
j. m. cline , h. firouzjahi and p. martineau , `` reheating from tachyon condensation , '' jhep * 0211 * , 041 ( 2002 ) [ hep - th/0207156 ] .
j. h. traschen and r. h. brandenberger , `` particle production during out - of - equilibrium phase transitions , '' phys .
d * 42 * , 2491 ( 1990 ) .
a. d. dolgov and d. p. kirilova , `` on particle creation by a time dependent scalar field , '' sov .
j. nucl .
phys . * 51 * , 172 ( 1990 ) [ yad .
fiz . * 51 * , 273 ( 1990 ) ] .
y. shtanov , j. h. traschen and r. h. brandenberger , `` universe reheating after inflation , '' phys .
d * 51 * , 5438 ( 1995 ) [ hep - ph/9407247 ] .
l. kofman , a. d. linde and a. a. starobinsky , `` reheating after inflation , '' phys .
lett . * 73 * , 3195 ( 1994 ) [ hep - th/9405187 ] .
s. y. khlebnikov and i. i. tkachev , `` relic gravitational waves produced after preheating , '' phys .
d * 56 * , 653 ( 1997 ) [ hep - ph/9701423 ] .
e. j. copeland , a. r. liddle , d. h. lyth , e. d. stewart and d. wands , `` false vacuum inflation with einstein gravity , '' phys .
d * 49 * , 6410 ( 1994 ) [ astro - ph/9401011 ] .
m. kawasaki , m. yamaguchi and t. yanagida , `` natural chaotic inflation in supergravity , '' phys .
lett . * 85 * , 3572 ( 2000 ) [ hep - ph/0004243 ] .
m. kawasaki , m. yamaguchi and t. yanagida , `` natural chaotic inflation in supergravity and leptogenesis , '' phys .
d * 63 * , 103514 ( 2001 ) [ hep - ph/0011104 ] .
j. martin and r. h. brandenberger , `` the transplanckian problem of inflationary cosmology , '' phys .
d * 63 * , 123501 ( 2001 ) [ hep - th/0005209 ] .
r. h. brandenberger and j. martin , `` the robustness of inflation to changes in superplanck scale physics , '' mod .
a * 16 * , 999 ( 2001 ) [ astro - ph/0005432 ] .
r. brandenberger and j. martin , `` trans - planckian issues for inflationary cosmology , '' arxiv:1211.6753 [ astro-ph.co ] .
a. borde , a. h. guth and a. vilenkin , `` inflationary space - times are incompletein past directions , '' phys .
lett . * 90 * , 151301 ( 2003 ) [ gr - qc/0110012 ] .
p. j. steinhardt , `` natural inflation , '' in * cambridge 1982 , proceedings , the very early universe * , 251 - 266 and pennsylvania univ .
philadelphia - upr-0198 t ( 82,rec.oct . )
16 p. ( 211400 ) ( see conference index ) a. vilenkin , `` the birth of inflationary universes , '' phys .
d * 27 * , 2848 ( 1983 ) .
l. parker , `` quantized fields and particle creation in expanding universes . 1 . , ''
rev . * 183 * , 1057 ( 1969 ) .
t. s. bunch and p. c. w. davies , `` quantum field theory in de sitter space : renormalization by point splitting , '' proc .
a * 360 * , 117 ( 1978 ) .
r. l. arnowitt , s. deser and c. w. misner , `` the dynamics of general relativity , '' gr - qc/0405109 .
r. m. wald , `` general relativity , '' chicago , usa : univ .
( 1984 ) 491p j. w. york , jr .
, `` role of conformal three geometry in the dynamics of gravitation , '' phys . rev .
* 28 * , 1082 ( 1972 ) .
g. w. gibbons and s. w. hawking , `` action integrals and partition functions in quantum gravity , '' phys .
d * 15 * , 2752 ( 1977 ) .
w. unruh , `` cosmological long wavelength perturbations , '' astro - ph/9802323 . g. geshnizjani and r. brandenberger , `` back reaction and local cosmological expansion rate , '' phys .
d * 66 * , 123507 ( 2002 ) [ gr - qc/0204074 ] .
t. prokopec and j. weenink , `` uniqueness of the gauge invariant action for cosmological perturbations , '' jcap * 1212 * , 031 ( 2012 ) [ arxiv:1209.1701 [ gr - qc ] ] .
k. hinterbichler , l. hui and j. khoury , `` conformal symmetries of adiabatic modes in cosmology , '' jcap * 1208 * , 017 ( 2012 ) [ arxiv:1203.6351 [ hep - th ] ] . g. l. pimentel , l. senatore and m. zaldarriaga , `` on loops in inflation iii : time independence of zeta in single clock inflation , '' jhep * 1207 * , 166 ( 2012 ) [ arxiv:1203.6651 [ hep - th ] ] .
m. y. .khlopov , `` primordial black holes , '' res . astron
. astrophys .
* 10 * , 495 ( 2010 ) [ arxiv:0801.0116 [ astro - ph ] ] .
e. d. stewart and d. h. lyth , `` a more accurate analytic calculation of the spectrum of cosmological perturbations produced during inflation , '' phys .
b * 302 * , 171 ( 1993 ) [ gr - qc/9302019 ] .
j. e. lidsey , a. r. liddle , e. w. kolb , e. j. copeland , t. barreiro and m. abney , `` reconstructing the inflation potential : an overview , '' rev .
phys . * 69 * , 373 ( 1997 ) [ astro - ph/9508078 ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , `` planck 2013 results . xxii .
constraints on inflation , '' arxiv:1303.5082 [ astro-ph.co ] .
d. h. lyth , k. a. malik and m. sasaki , `` a general proof of the conservation of the curvature perturbation , '' jcap * 0505 * , 004 ( 2005 ) [ astro - ph/0411220 ] .
v. assassi , d. baumann and d. green , `` symmetries and loops in inflation , '' arxiv:1210.7792 [ hep - th ] .
a. a. starobinsky , `` multicomponent de sitter ( inflationary ) stages and the generation of perturbations , '' jetp lett . *
42 * , 152 ( 1985 ) [ pisma zh .
fiz . * 42 * , 124 ( 1985 ) ] .
m. sasaki and e. d. stewart , `` a general analytic formula for the spectral index of the density perturbations produced during inflation , '' prog .
phys . *
95 * , 71 ( 1996 ) [ astro - ph/9507001 ] .
d. h. lyth and y. rodriguez , `` the inflationary prediction for primordial non - gaussianity , '' phys .
* 95 * , 121302 ( 2005 ) [ astro - ph/0504045 ] .
a.a.starobinsky , in fundamental interactions , ed .
v.n.ponomarev ( mgpi press , moscow , 1984 ) p. 54 ; and in current topics in field theory , quantum gravity and strings , lecture notes in physics 246 , h.j.devega and n.sanchez , eds .
( springer , berlin , 1986 ) p. 107 .
l. susskind , l. thorlacius and j. uglum , `` the stretched horizon and black hole complementarity , '' phys .
d * 48 * , 3743 ( 1993 ) [ hep - th/9306069 ] .
n. goheer , m. kleban and l. susskind , `` the trouble with de sitter space , '' jhep * 0307 * , 056 ( 2003 ) [ hep - th/0212209 ] .
n. arkani - hamed , s. dubovsky , a. nicolis , e. trincherini and g. villadoro , `` a measure of de sitter entropy and eternal inflation , '' jhep * 0705 * , 055 ( 2007 ) [ arxiv:0704.1814 [ hep - th ] ] .
v. a. rubakov , m. v. sazhin and a. v. veryaskin , `` graviton creation in the inflationary universe and the grand unification scale , '' phys .
b * 115 * , 189 ( 1982 ) .
r. fabbri and m. d. pollock , `` the effect of primordially produced gravitons upon the anisotropy of the cosmological microwave background radiation , '' phys .
b * 125 * , 445 ( 1983 ) .
l. f. abbott and m. b. wise , `` constraints on generalized inflationary cosmologies , '' nucl .
b * 244 * , 541 ( 1984 ) .
l. a. boyle and p. j. steinhardt , `` probing the early universe with inflationary gravitational waves , '' phys .
d * 77 * , 063504 ( 2008 ) [ astro - ph/0512014 ] .
d. h. lyth , `` what would we learn by detecting a gravitational wave signal in the cosmic microwave background anisotropy ?
, '' phys .
lett . * 78 * , 1861 ( 1997 ) [ hep - ph/9606387 ] . t. fujita and s. mukohyama , `` universal upper limit on inflation energy scale from cosmic magnetic field , '' jcap * 1210 * , 034 ( 2012 ) [ arxiv:1205.5031 [ astro-ph.co ] ] .
a. d. linde and v. mukhanov , `` the curvaton web , '' jcap * 0604 * , 009 ( 2006 ) [ astro - ph/0511736 ] .
x. chen and y. wang , `` anomalous high energy dependence in inflationary density perturbations , '' arxiv:1207.0100 [ hep - th ] .
s. weinberg , `` quantum contributions to cosmological correlations , '' phys .
d * 72 * , 043514 ( 2005 ) [ hep - th/0506236 ] .
m. li and y. wang , `` multi - stream inflation , '' jcap * 0907 * , 033 ( 2009 ) [ arxiv:0903.2123 [ hep - th ] ] .
f. duplessis , y. wang and r. brandenberger , `` multi - stream inflation in a landscape , '' jcap * 1204 * , 012 ( 2012 ) [ arxiv:1201.0029 [ hep - th ] ] .
x. chen and y. wang , `` quasi - single field inflation and non - gaussianities , '' jcap * 1004 * , 027 ( 2010 ) [ arxiv:0911.3380 [ hep - th ] ] .
x. chen and y. wang , `` quasi - single field inflation with large mass , '' jcap * 1209 * , 021 ( 2012 ) [ arxiv:1205.0160 [ hep - th ] ] .
s. r. behbahani and d. green , `` collective symmetry breaking and resonant non - gaussianity , '' jcap * 1211 * , 056 ( 2012 ) [ arxiv:1207.2779 [ hep - th ] ] .
d. green , m. lewandowski , l. senatore , e. silverstein and m. zaldarriaga , `` anomalous dimensions and non - gaussianity , '' arxiv:1301.2630 [ hep - th ] .
x. chen , b. hu , m. -x .
huang , g. shiu and y. wang , `` large primordial trispectra in general single field inflation , '' jcap * 0908 * , 008 ( 2009 ) [ arxiv:0905.3494 [ astro-ph.co ] ] .
j. m. maldacena , `` non - gaussian features of primordial fluctuations in single field inflationary models , '' jhep * 0305 * , 013 ( 2003 ) [ astro - ph/0210603 ] . f. arroja and t. tanaka , `` a note on the role of the boundary terms for the non - gaussianity in general k - inflation , '' jcap * 1105 * , 005 ( 2011 ) [ arxiv:1103.1102 [ astro-ph.co ] ] .
s. renaux - petel , `` on the redundancy of operators and the bispectrum in the most general second - order scalar - tensor theory , '' jcap * 1202 * , 020 ( 2012 ) [ arxiv:1107.5020 [ astro-ph.co ] ] .
n. arkani - hamed , l. motl , a. nicolis and c. vafa , `` the string landscape , black holes and gravity as the weakest force , '' jhep * 0706 * , 060 ( 2007 ) [ hep - th/0601001 ] . q. -g .
huang , `` gravitational correction and weak gravity conjecture , '' jhep * 0703 * , 053 ( 2007 ) [ hep - th/0703039 [ hep - th ] ] .
x. chen , m. -x .
huang and g. shiu , `` the inflationary trispectrum for models with large non - gaussianities , '' phys .
d * 74 * , 121301 ( 2006 ) [ hep - th/0610235 ] .
d. k. hazra , l. sriramkumar and j. martin , `` bingo : a code for the efficient computation of the scalar bi - spectrum , '' arxiv:1201.0926 [ astro-ph.co ] .
d. k. hazra , j. martin and l. sriramkumar , `` the scalar bi - spectrum during preheating in single field inflationary models , '' phys .
d * 86 * , 063523 ( 2012 ) [ arxiv:1206.0442 [ astro-ph.co ] ] .
e. komatsu and d. n. spergel , `` acoustic signatures in the primary microwave background bispectrum , '' phys .
d * 63 * , 063002 ( 2001 ) [ astro - ph/0005036 ] .
b. chen , y. wang and w. xue , `` inflationary nongaussianity from thermal fluctuations , '' jcap * 0805 * , 014 ( 2008 ) [ arxiv:0712.2345 [ hep - th ] ] .
d. wands , `` local non - gaussianity from inflation , '' class .
* 27 * , 124002 ( 2010 ) [ arxiv:1004.0818 [ astro-ph.co ] ] .
y. -f .
cai and y. wang , `` large nonlocal non - gaussianity from a curvaton brane , '' phys .
d * 82 * , 123501 ( 2010 ) [ arxiv:1005.0127 [ hep - th ] ] .
j. -o .
gong , c. lin and y. wang , `` non - gaussianity from false vacuum inflation : old curvaton scenario , '' jcap * 1003 * , 004 ( 2010 ) [ arxiv:0912.2796 [ astro-ph.co ] ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , `` planck 2013 results . xxiv .
constraints on primordial non - gaussianity , '' arxiv:1303.5084 [ astro-ph.co ]
. t. suyama and m. yamaguchi , `` non - gaussianity in the modulated reheating scenario , '' phys .
d * 77 * , 023505 ( 2008 ) [ arxiv:0709.2545 [ astro - ph ] ] . c. lin and y. wang , `` quadra - spectrum and quint - spectrum from inflation and curvaton models , '' jcap * 1007 * , 011 ( 2010 ) [ arxiv:1004.0461 [ astro-ph.co ] ] .
g. tasinato , c. t. byrnes , s. nurmi and d. wands , `` loop corrections and a new test of inflation , '' arxiv:1207.1772 [ hep - th ] .
creminelli and m. zaldarriaga , `` single field consistency relation for the 3-point function , '' jcap * 0410 * , 006 ( 2004 ) [ astro - ph/0407059 ] .
m. li and y. wang , `` consistency relations for non - gaussianity , '' jcap * 0809 * , 018 ( 2008 ) [ arxiv:0807.3058 [ hep - th ] ] .
v. assassi , d. baumann and d. green , `` on soft limits of inflationary correlation functions , '' jcap * 1211 * , 047 ( 2012 ) [ arxiv:1204.4207 [ hep - th ] ] .
m. shiraishi , e. komatsu , m. peloso and n. barnaby , `` signatures of anisotropic sources in the squeezed - limit bispectrum of the cosmic microwave background , '' arxiv:1302.3056 [ astro-ph.co ] .
e. silverstein and a. westphal , `` monodromy in the cmb : gravity waves and string inflation , '' phys .
d * 78 * , 106003 ( 2008 ) [ arxiv:0803.3085 [ hep - th ] ] .
k. freese and d. spolyar , `` chain inflation : bubble bubble toil and trouble , '' jcap * 0507 * , 007 ( 2005 ) [ hep - ph/0412145 ] .
j. m. cline , g. d. moore and y. wang , `` chain inflation reconsidered , '' jcap * 1108 * , 032 ( 2011 ) [ arxiv:1106.2188 [ hep - th ] ] .
n. arkani - hamed , h. -c .
cheng , m. a. luty and s. mukohyama , `` ghost condensation and a consistent infrared modification of gravity , '' jhep * 0405 * , 074 ( 2004 ) [ hep - th/0312099 ] .
n. arkani - hamed , p. creminelli , s. mukohyama and m. zaldarriaga , `` ghost inflation , '' jcap * 0404 * , 001 ( 2004 ) [ hep - th/0312100 ] . c. m. bender and s. boettcher , `` real spectra in nonhermitian hamiltonians having pt symmetry , '' phys .
lett . * 80 * , 5243 ( 1998 ) [ physics/9712001 ] . c. m. bender and p. d. mannheim , `` no - ghost theorem for the fourth - order derivative pais - uhlenbeck oscillator model , '' phys .
lett . * 100 * , 110402 ( 2008 ) [ arxiv:0706.0207 [ hep - th ] ] .
j. m. cline , s. jeon and g. d. moore , `` the phantom menaced : constraints on low - energy effective ghosts , '' phys .
d * 70 * , 043543 ( 2004 ) [ hep - ph/0311312 ] .
n. barnaby and j. m. cline , `` large nongaussianity from nonlocal inflation , '' jcap * 0707 * , 017 ( 2007 ) [ arxiv:0704.3426 [ hep - th ] ] .
n. barnaby and j. m. cline , `` predictions for nongaussianity from nonlocal inflation , '' jcap * 0806 * , 030 ( 2008 ) [ arxiv:0802.3218 [ hep - th ] ] .
n. barnaby , `` nonlocal inflation , '' can . j. phys . * 87 * , 189 ( 2009 ) [ arxiv:0811.0814 [ hep - th ] ] . g. w. horndeski , `` second - order scalar - tensor field equations in a four - dimensional space , '' int .
j. theor .
* 10 * , 363 ( 1974 ) .
c. deffayet , x. gao , d. a. steer and g. zahariade , `` from k - essence to generalised galileons , '' phys .
d * 84 * , 064039 ( 2011 ) [ arxiv:1103.3260 [ hep - th ] ] .
t. kobayashi , m. yamaguchi and j. i .
yokoyama , `` generalized g - inflation : inflation with the most general second - order field equations , '' prog .
phys . * 126 * , 511 ( 2011 ) [ arxiv:1105.5723 [ hep - th ] ] .
x. gao and d. a. steer , `` inflation and primordial non - gaussianities of generalized galileons , '' jcap * 1112 * , 019 ( 2011 ) [ arxiv:1107.2642 [ astro-ph.co ] ] .
x. gao , t. kobayashi , m. shiraishi , m. yamaguchi , j. i .
yokoyama and s. yokoyama , `` full bispectra from primordial scalar and tensor perturbations in the most general single - field inflation model , '' arxiv:1207.0588 [ astro-ph.co ] .
c. armendariz - picon , t. damour and v. f. mukhanov , `` k - inflation , '' phys .
b * 458 * , 209 ( 1999 ) [ hep - th/9904075 ] .
j. garriga and v. f. mukhanov , `` perturbations in k - inflation , '' phys .
b * 458 * , 219 ( 1999 ) [ hep - th/9904176 ] .
t. kobayashi , m. yamaguchi and j. i .
yokoyama , `` g - inflation : inflation driven by the galileon field , '' phys .
lett . * 105 * , 231302 ( 2010 ) [ arxiv:1008.0603 [ hep - th ] ] .
t. kobayashi , m. yamaguchi and j. i .
yokoyama , `` primordial non - gaussianity from g - inflation , '' phys .
d * 83 * , 103524 ( 2011 ) [ arxiv:1103.1740 [ hep - th ] ] .
d. seery and j. e. lidsey , `` primordial non - gaussianities in single field inflation , '' jcap * 0506 * , 003 ( 2005 ) [ astro - ph/0503692 ] .
x. chen , m. -x .
huang , s. kachru and g. shiu , `` observational signatures and non - gaussianities of general single field inflation , '' jcap * 0701 * , 002 ( 2007 ) [ hep - th/0605045 ] .
p. creminelli , a. nicolis , l. senatore , m. tegmark and m. zaldarriaga , `` limits on non - gaussianities from wmap data , '' jcap * 0605 * , 004 ( 2006 ) [ astro - ph/0509029 ] .
f. arroja , s. mizuno , k. koyama and t. tanaka , `` on the full trispectrum in single field dbi - inflation , '' phys .
d * 80 * , 043527 ( 2009 ) [ arxiv:0905.3641 [ hep - th ] ] .
d. langlois and s. renaux - petel , `` perturbations in generalized multi - field inflation , '' jcap * 0804 * , 017 ( 2008 ) [ arxiv:0801.1085 [ hep - th ] ] .
s. renaux - petel , `` combined local and equilateral non - gaussianities from multifield dirac - born - infeld inflation , '' arxiv:1004.3693 [ astro-ph.co ] . c. cheung , p. creminelli , a. l. fitzpatrick , j. kaplan and l. senatore , `` the effective field theory of inflation , '' jhep * 0803 * , 014 ( 2008 ) [ arxiv:0709.0293 [ hep - th ] ] .
c. p. burgess , j. m. cline and r. holman , `` effective field theories and inflation , '' jcap * 0310 * , 004 ( 2003 ) [ hep - th/0306079 ] . s. weinberg , `` effective field theory for inflation , '' phys .
d * 77 * , 123541 ( 2008 ) [ arxiv:0804.4291 [ hep - th ] ] .
d. baumann and d. green , `` equilateral non - gaussianity and new physics on the horizon , '' jcap * 1109 * , 014 ( 2011 ) [ arxiv:1102.5343 [ hep - th ] ] . q. -g .
huang , `` the trispectrum in ghost inflation , '' jcap * 1007 * , 025 ( 2010 ) [ arxiv:1004.0808 [ astro-ph.co ] ] .
k. izumi and s. mukohyama , `` trispectrum from ghost inflation , '' jcap * 1006 * , 016 ( 2010 ) [ arxiv:1004.1776 [ hep - th ] ] .
a. j. tolley and m. wyman , `` the gelaton scenario : equilateral non - gaussianity from multi - field dynamics , '' phys .
d * 81 * , 043502 ( 2010 ) [ arxiv:0910.1853 [ hep - th ] ] .
s. cremonini , z. lalak and k. turzynski , `` strongly coupled perturbations in two - field inflationary models , '' jcap * 1103 * , 016 ( 2011 ) [ arxiv:1010.3021 [ hep - th ] ] . l. senatore and m. zaldarriaga , `` the effective field theory of multifield inflation , '' jhep * 1204 * , 024 ( 2012 ) [ arxiv:1009.2093 [ hep - th ] ] .
d. baumann and d. green , `` signatures of supersymmetry from the early universe , '' phys .
d * 85 * , 103520 ( 2012 ) [ arxiv:1109.0292 [ hep - th ] ] .
t. noumi , m. yamaguchi and d. yokoyama , `` effective field theory approach to quasi - single field inflation , '' arxiv:1211.1624 [ hep - th ]
. j. j. m. carrasco , m. p. hertzberg and l. senatore , `` the effective field theory of cosmological large scale structures , '' jhep * 1209 * , 082 ( 2012 ) [ arxiv:1206.2926 [ astro-ph.co ] ] .
x. chen , h. firouzjahi , m. h. namjoo and m. sasaki , `` a single field inflation model with large local non - gaussianity , '' arxiv:1301.5699 [ hep - th ] . x. chen , `` folded resonant non - gaussianity in general single field inflation , '' jcap * 1012 * , 003 ( 2010 ) [ arxiv:1008.2485 [ hep - th ] ] .
m. h. namjoo , h. firouzjahi and m. sasaki , `` violation of non - gaussianity consistency relation in a single field inflationary model , '' arxiv:1210.3692 [ astro-ph.co ] .
j. martin , h. motohashi and t. suyama , `` ultra slow - roll inflation and the non - gaussianity consistency relation , '' phys .
d * 87 * , 023514 ( 2013 ) [ arxiv:1211.0083 [ astro-ph.co ] ] .
y. -f .
cai , w. xue , r. brandenberger and x. zhang , `` non - gaussianity in a matter bounce , '' jcap * 0905 * , 011 ( 2009 ) [ arxiv:0903.0631 [ astro-ph.co ] ] .
d. wands , k. a. malik , d. h. lyth and a. r. liddle , `` a new approach to the evolution of cosmological perturbations on large scales , '' phys .
d * 62 * , 043527 ( 2000 ) [ astro - ph/0003278 ] . c. gordon , d. wands , b. a. bassett and r. maartens , `` adiabatic and entropy perturbations from inflation , '' phys .
d * 63 * , 023506 ( 2001 ) [ astro - ph/0009131 ] .
s. groot nibbelink and b. j. w. van tent , `` scalar perturbations during multiple field slow - roll inflation , '' class .
* 19 * , 613 ( 2002 ) [ hep - ph/0107272 ] .
s. groot nibbelink and b. j. w. van tent , `` density perturbations arising from multiple field slow roll inflation , '' hep - ph/0011325 . s. -h .
h. tye , j. xu and y. zhang , `` multi - field inflation with a random potential , '' jcap * 0904 * , 018 ( 2009 ) [ arxiv:0812.1944 [ hep - th ] ] .
m. s. sloth , `` suppressing super - horizon curvature perturbations ? , '' mod .
a * 21 * , 961 ( 2006 ) [ hep - ph/0507315 ] .
n. bartolo , e. w. kolb and a. riotto , `` post - inflation increase of the cosmological tensor - to - scalar perturbation ratio , '' mod .
a * 20 * , 3077 ( 2005 ) [ astro - ph/0507573 ] .
m. li , c. lin , t. wang and y. wang , `` non - gaussianity , isocurvature perturbation , gravitational waves and a no - go theorem for isocurvaton , '' phys .
d * 79 * , 063526 ( 2009 ) [ arxiv:0805.1299 [ astro - ph ] ] .
d. langlois and f. vernizzi , `` nonlinear perturbations of cosmological scalar fields , '' jcap * 0702 * , 017 ( 2007 ) [ astro - ph/0610064 ] .
t. suyama , t. takahashi , m. yamaguchi and s. yokoyama , `` implications of planck results for models with local type non - gaussianity , '' arxiv:1303.5374 [ astro-ph.co ]
. s. mollerach , `` isocurvature baryon perturbations and inflation , '' phys .
d * 42 * , 313 ( 1990 ) .
a. d. linde and v. f. mukhanov , `` nongaussian isocurvature perturbations from inflation , '' phys .
d * 56 * , 535 ( 1997 ) [ astro - ph/9610219 ] .
k. enqvist and m. s. sloth , `` adiabatic cmb perturbations in pre - big bang string cosmology , '' nucl .
b * 626 * , 395 ( 2002 ) [ hep - ph/0109214 ] .
d. h. lyth and d. wands , `` generating the curvature perturbation without an inflaton , '' phys .
b * 524 * , 5 ( 2002 ) [ hep - ph/0110002 ] .
t. moroi and t. takahashi , `` effects of cosmological moduli fields on cosmic microwave background , '' phys .
b * 522 * , 215 ( 2001 ) [ erratum - ibid .
b * 539 * , 303 ( 2002 ) ] [ hep - ph/0110096 ] .
m. sasaki , j. valiviita and d. wands , `` non - gaussianity of the primordial perturbation in the curvaton model , '' phys .
d * 74 * , 103003 ( 2006 ) [ astro - ph/0607627 ] .
k. enqvist and s. nurmi , `` non - gaussianity in curvaton models with nearly quadratic potential , '' jcap * 0510 * , 013 ( 2005 ) [ astro - ph/0508573 ] .
q. -g . huang and y. wang , ``
curvaton dynamics and the non - linearity parameters in curvaton model , '' jcap * 0809 * , 025 ( 2008 ) [ arxiv:0808.1168 [ hep - th ] ] .
m. kawasaki , t. kobayashi and f. takahashi , `` non - gaussianity from curvatons revisited , '' phys .
d * 84 * , 123506 ( 2011 ) [ arxiv:1107.6011 [ astro-ph.co ] ] .
m. sasaki , `` multi - brid inflation and non - gaussianity , '' prog .
phys . * 120 * , 159 ( 2008 ) [ arxiv:0805.0974 [ astro - ph ] ] .
q. -g .
huang , `` a geometric description of the non - gaussianity generated at the end of multi - field inflation , '' jcap * 0906 * , 035 ( 2009 ) [ arxiv:0904.2649 [ hep - th ] ] .
d. seery , d. j. mulryne , j. frazer and r. h. ribeiro , `` inflationary perturbation theory is geometrical optics in phase space , '' jcap * 1209 * , 010 ( 2012 ) [ arxiv:1203.2635 [ astro-ph.co ] ] .
g. dvali , a. gruzinov and m. zaldarriaga , `` a new mechanism for generating density perturbations from inflation , '' phys .
d * 69 * , 023505 ( 2004 ) [ astro - ph/0303591 ] .
l. kofman , `` probing string theory with modulated cosmological fluctuations , '' astro - ph/0303614 .
g. dvali , a. gruzinov and m. zaldarriaga , `` cosmological perturbations from inhomogeneous reheating , freezeout , and mass domination , '' phys .
d * 69 * , 083505 ( 2004 ) [ astro - ph/0305548 ] .
f. vernizzi , `` cosmological perturbations from varying masses and couplings , '' phys .
d * 69 * , 083526 ( 2004 ) [ astro - ph/0311167 ] .
r. bousso and j. polchinski , `` quantization of four form fluxes and dynamical neutralization of the cosmological constant , '' jhep * 0006 * , 006 ( 2000 ) [ hep - th/0004134 ] .
s. b. giddings , s. kachru and j. polchinski , `` hierarchies from fluxes in string compactifications , '' phys . rev .
d * 66 * , 106006 ( 2002 ) [ hep - th/0105097 ] .
s. kachru , r. kallosh , a. d. linde and s. p. trivedi , `` de sitter vacua in string theory , '' phys .
d * 68 * , 046005 ( 2003 ) [ hep - th/0301240 ] .
l. susskind , `` the anthropic landscape of string theory , '' in * carr , bernard ( ed . ) : universe or multiverse ?
* 247 - 266 [ hep - th/0302219 ]
. n. agarwal , r. bean , l. mcallister and g. xu , `` universality in d - brane inflation , '' jcap * 1109 * , 002 ( 2011 ) [ arxiv:1103.2775 [ astro-ph.co ] ] .
j. frazer and a. r. liddle , `` multi - field inflation with random potentials : field dimension , feature scale and non - gaussianity , '' jcap * 1202 * , 039 ( 2012 ) [ arxiv:1111.6646 [ astro-ph.co ] ] .
a. d. linde , d. a. linde and a. mezhlumian , `` from the big bang theory to the theory of a stationary universe , '' phys .
d * 49 * , 1783 ( 1994 ) [ gr - qc/9306035 ] .
q. -g .
huang , m. li and y. wang , `` eternal chaotic inflation is prohibited by weak gravity conjecture , '' jcap * 0709 * , 013 ( 2007 ) [ arxiv:0707.3471 [ hep - th ] ] .
p. adshead , r. easther and e. a. lim , `` cosmology with many light scalar fields : stochastic inflation and loop corrections , '' phys .
d * 79 * , 063504 ( 2009 ) [ arxiv:0809.4008 [ hep - th ] ] .
n. afshordi , a. slosar and y. wang , `` a theory of a spot , '' jcap * 1101 * , 019 ( 2011 ) [ arxiv:1006.5021 [ astro-ph.co ] ] .
s. li , y. liu and y. -s .
piao , `` inflation in web , '' phys .
d * 80 * , 123535 ( 2009 ) [ arxiv:0906.3608 [ hep - th ] ] .
y. wang , `` multi - stream inflation : bifurcations and recombinations in the multiverse , '' arxiv:1001.0008 [ hep - th ] .
s. -h .
henry tye , `` a new view of the cosmic landscape , '' hep - th/0611148 .
s. -h .
h. tye and d. wohns , `` resonant tunneling in scalar quantum field theory , '' arxiv:0910.1088 [ hep - th ] .
x. chen and y. wang , `` large non - gaussianities with intermediate shapes from quasi - single field inflation , '' phys .
d * 81 * , 063511 ( 2010 ) [ arxiv:0909.0496 [ astro-ph.co ] ] .
l. mcallister , s. renaux - petel and g. xu , `` a statistical approach to multifield inflation : many - field perturbations beyond slow roll , '' jcap * 1210 * , 046 ( 2012 ) [ arxiv:1207.0317 [ astro-ph.co ] ] .
s. pi and m. sasaki , `` curvature perturbation spectrum in two - field inflation with a turning trajectory , '' jcap * 1210 * , 051 ( 2012 ) [ arxiv:1205.0161 [ hep - th ] ] .
n. barnaby and s. shandera , `` feeding your inflaton : non - gaussian signatures of interaction structure , '' jcap * 1201 * , 034 ( 2012 ) [ arxiv:1109.2985 [ astro-ph.co ] ] .
e. sefusatti , j. r. fergusson , x. chen and e. p. s. shellard , `` effects and detectability of quasi - single field inflation in the large - scale structure and cosmic microwave background , '' jcap * 1208 * , 033 ( 2012 ) [ arxiv:1204.6318 [ astro-ph.co ] ] .
j. norena , l. verde , g. barenboim and c. bosch , `` prospects for constraining the shape of non - gaussianity with the scale - dependent bias , '' jcap * 1208 * , 019 ( 2012 ) [ arxiv:1204.6324 [ astro-ph.co ] ] .
x. chen , r. easther and e. a. lim , `` large non - gaussianities in single field inflation , '' jcap * 0706 * , 023 ( 2007 ) [ astro - ph/0611645 ] . j. martin and l. sriramkumar , `` the scalar bi - spectrum in the starobinsky model : the equilateral case , '' jcap * 1201 * , 008 ( 2012 ) [ arxiv:1109.5838 [ astro-ph.co ] ] .
x. chen , r. easther and e. a. lim , `` generation and characterization of large non - gaussianities in single field inflation , '' jcap * 0804 * , 010 ( 2008 ) [ arxiv:0801.3295 [ astro - ph ] ] .
a. achucarro , j. -o .
gong , s. hardeman , g. a. palma and s. p. patil , `` features of heavy physics in the cmb power spectrum , '' jcap * 1101 * , 030 ( 2011 ) [ arxiv:1010.3693 [ hep - ph ] ] .
a. achucarro , j. -o .
gong , s. hardeman , g. a. palma and s. p. patil , `` mass hierarchies and non - decoupling in multi - scalar field dynamics , '' phys .
d * 84 * , 043502 ( 2011 ) [ arxiv:1005.3848 [ hep - th ] ] .
x. chen , `` primordial features as evidence for inflation , '' jcap * 1201 * , 038 ( 2012 ) [ arxiv:1104.1323 [ hep - th ] ] .
g. shiu and j. xu , `` effective field theory and decoupling in multi - field inflation : an illustrative case study , '' phys .
d * 84 * , 103509 ( 2011 ) [ arxiv:1108.0981 [ hep - th ] ] .
s. cespedes , v. atal and g. a. palma , `` on the importance of heavy fields during inflation , '' jcap * 1205 * , 008 ( 2012 ) [ arxiv:1201.4848 [ hep - th ] ] .
a. achucarro , j. -o .
gong , s. hardeman , g. a. palma and s. p. patil , `` effective theories of single field inflation when heavy fields matter , '' jhep * 1205 * , 066 ( 2012 ) [ arxiv:1201.6342 [ hep - th ] ] .
a. avgoustidis , s. cremonini , a. -c .
davis , r. h. ribeiro , k. turzynski and s. watson , `` decoupling survives inflation : a critical look at effective field theory violations during inflation , '' jcap * 1206 * , 025 ( 2012 ) [ arxiv:1203.0016 [ hep - th ] ] .
x. gao , d. langlois and s. mizuno , `` influence of heavy modes on perturbations in multiple field inflation , '' jcap * 1210 * , 040 ( 2012 ) [ arxiv:1205.5275 [ hep - th ] ] .
a. achucarro , j. -o .
gong , g. a. palma and s. p. patil , `` correlating features in the primordial spectra , '' arxiv:1211.5619 [ astro-ph.co ] .
x. chen , `` fingerprints of primordial universe paradigms as features in density perturbations , '' phys .
b * 706 * , 111 ( 2011 ) [ arxiv:1106.1635 [ astro-ph.co ] ] .
d. j. h. chung , e. w. kolb , a. riotto and i. i. tkachev , `` probing planckian physics : resonant production of particles during inflation and features in the primordial power spectrum , '' phys .
d * 62 * , 043508 ( 2000 ) [ hep - ph/9910437 ] .
a. e. romano and m. sasaki , `` effects of particle production during inflation , '' phys .
d * 78 * , 103522 ( 2008 ) [ arxiv:0809.5142 [ gr - qc ] ] .
n. barnaby , `` on features and nongaussianity from inflationary particle production , '' phys .
d * 82 * , 106009 ( 2010 ) [ arxiv:1006.4615 [ astro-ph.co ] ] .
d. battefeld , t. battefeld , c. byrnes and d. langlois , `` beauty is distractive : particle production during multifield inflation , '' jcap * 1108 * , 025 ( 2011 ) [ arxiv:1106.1891 [ astro-ph.co ] ] . | we review the theory of inflationary perturbations .
perturbations at both linear and nonlinear orders are reviewed .
we also review a variety of inflation models , emphasizing their signatures on cosmic perturbations .
= 1 |
according to the famous dictum of gravity / gauge theory duality or maldacena s anti de sitter ( ads ) conformal field theory ( cft ) correspondence @xcite , an ads space with a black hole is dual to a field theory at a finite temperature .
ads / cft has been an excellent tool to study not only strongly coupled gauge theories at a large t hooft coupling limit but also to study hydrodynamic properties of a certain class of boundary cfts at a finite temperature .
one of the highly celebrated results over the last decade is that , for a large class of four - dimensional cfts , the ratio of the shear viscosity @xmath4 to the entropy density @xmath5 is ( in units @xmath6 ) given by @xcite @xmath7 this result has been known to hold for all gauge theories with einstein gravity dual .
this is due to the fact that in pure einstein gravity all black holes satisfy the bekenstein - hawking entropy law or the area formula of black hole entropy @xmath8 where @xmath9 is the planck constant , @xmath10 is the boltzmann constant , @xmath11 is the area of the ( black hole ) horizon corresponding to the surface at @xmath12 and @xmath13 is the newton s constant @xmath13 . in general gravity theories with higher derivative or higher - curvature corrections , however , the ratio @xmath2 can be different from @xmath3 . in the same context ,
a violation of causality might occur in the boundary cft when @xmath2 is too low ( @xmath14 ) . here
we shall argue that any such a violation of micro - causality can be related to a violation of black hole entropy bounds or a violation of certain laws of black hole thermodynamics . to show that this might be the case , and also to establish better contact with qcd via ads holography , in a holographic context ( see @xcite and references therein ) , one might like to consider the following gravitational action @xmath15,\ ] ] where @xmath16 is a dimensionless coupling and @xmath17 is a bulk cosmological term .
in fact , the regge slope or the coefficient @xmath16 appearing in the effective action ( [ main - action ] ) can be a complicated function of some microscopic parameters of the quantum gravity theory , which may even depend on the details of compactification and on dilaton couplings .
nevertheless , in the simplest scenario to be considered in this paper , one would assume that @xmath18 .
here one might also note that , especially , in an ads@xmath19 space , the coefficients of @xmath20 terms in a dual supergravity action are determined by the central charges of a cft in four dimensions . in most of our discussions we will focus to the gravity sector in a five - dimensional ads space , for which we have from the ads / cft correspondence , @xmath21 and @xmath22 , where @xmath23 is the string scale and @xmath24 the string coupling and nc is the number of color charges or rank of the gauge group . in the dual supergravity description ,
a small @xmath16 corresponds to the strong coupling limit , i.e. , @xmath25 , since @xmath26 .
the gauss - bonnet term is obtained by setting @xmath27 and @xmath28 in ( [ main - action ] ) , for which there would be no need to treat @xmath16 as small , at least , for the purpose of obtaining exact ( black hole ) solutions @xcite .
recently , kats and petrov @xcite and brigante et al .
@xcite have shown that , for a class of cfts in flat space with gauss - bonnet gravity dual , the ratio @xmath2 , which reads @xmath29 where @xmath30 , can be smaller than @xmath3 for @xmath31 and @xmath32 . based on the bulk causal structure of an ads@xmath19 black brane solution , one may actually find an even more stronger bound for @xmath33 , that is , @xmath34 which otherwise violates a microcausality in the dual cft defined on a flat space . here
we show that the critical value of @xmath33 beyond which the theory becomes inconsistent is related to the entropy bound for a large class of ads black holes .
in fact , in the holographic context , ads black hole solutions with spherical and hyperbolic event horizons allow much wider possibilities for @xmath2 @xcite .
it is quite plausible that in the presence of curvature terms like @xmath35 with @xmath36 , which generate six and higher derivatives in the metric , or the @xmath37 terms arising as @xmath38 corrections to low energy effective action of type iib string theory , the ratio @xmath2 takes a value slightly larger than @xmath3 ( see @xcite for a review ) .
however , in any consistent gravity models , interaction terms like @xmath39 , where @xmath40 , must be suppressed in a sensible derivative expansion , implying that such corrections could arise only as subleading or next - to - subleading terms . here
we show that when an underlying theory admits a large violation of the conjectured kss bound @xmath41 or the ratio @xmath2 becomes too small , then certain laws of thermodynamics can be violated in the same limit .
this effect can be seen also in terms of a violation of causality in the boundary field theory .
the bound @xmath42 may be restored only at weak couplings or in the limit of large @xmath43 , since in such cases the shear viscosity can grow faster as compared to the entropy density .
for an ads gb black hole , the entropy and hawking temperature are given by @xcite @xmath44 ( in units @xmath45 ) where @xmath46 , @xmath47 , with @xmath48 being the unit volume of the base manifold or the hypersurface @xmath49 .
the entropy of a gb black hole depends on the curvature constant @xmath50 , whose value determines the geometry of the event horizon @xmath51 and @xmath52 , respectively , for @xmath53 and @xmath54 , especially , with a constant coupling , @xmath55 , so we take @xmath32 . ] .
especially , at the @xmath54 extremal state with zero temperature , @xmath56 thus , beyond a critical coupling @xmath57 , the entropy @xmath58 becomes negative , which indicates a violation of cosmic censorship or the second law of the thermodynamics . in the ads@xmath19 case , @xmath59 .
this critical value of @xmath33 above which the theory is inconsistent nearly coincides with the bound @xmath60 required for a consistent formulation of a class of cfts in a flat space with gauss - bonnet gravity dual . to be more specific ,
let us consider small metric fluctuations @xmath61 around an ads gb black hole metric , using @xmath62 to denote this perturbation with one index raised @xmath63 and writing @xmath62 in the basis @xmath64 . ]
@xmath65 where @xmath66 and @xmath67^{1/2}$ ] and @xmath68 is the curvature of ads@xmath19 space .
the scalar metric fluctuation ( along the @xmath69 direction ) @xmath70 ( where @xmath71 ) satisfies the following ( linearized ) equation of motion @xmath72 this structure is not affected by maxwell type charges but it may well be affected by dilatonic scalar charges and rotation parameters . in the simplest scenario
one may neglect both the scalar charge and rotation parameters .
in pure gb gravity defined on an ads@xmath19 spacetime , we find @xmath73 , @xmath74 and @xmath75,\ ] ] where @xmath76 , @xmath77 , @xmath78 , @xmath79 and @xmath80 .
( [ linear - eq ] ) may be solved with the following incoming boundary conditions at the horizon @xmath81 where @xmath82 is an infinitesimal boundary source for the fluctuating field @xmath62 . to the leading order in @xmath83 , and in the limit @xmath84 ,
the solution is given by @xcite @xmath85.\ ] ] by identifying @xmath86 , where @xmath87 is a component of the energy - momentum tensor on the field theory side , we may obtain the retarded two - point green s function . indeed , gb black holes admit a stable five - dimensional anti - de sitter vacuum only if @xmath88 @xcite .
this result is independent of the types of interactions the gauss - bonnet term can have with matter fields ( including gauge and scalar fields ) . from the above result
, we can easily see that the curvature on a boundary may not affect the shear viscosity @xmath89 obtained using the kubo formula @xmath90 \equiv \lim_{w\to 0 } \frac{1}{w } { \rm i m } g^r_{12 , 12 } ( w,0).\ ] ] as is evident , this formula relates @xmath4 to zero spatial momentum ( @xmath91 ) , low frequency limit of the retarded two - point green s function @xmath92\rangle,\ ] ] which satisfies @xmath93 .
the ratio @xmath2 is now modified as @xmath94 the standard result in einstein gravity , i.e. @xmath0 , is obtained only at a fixed @xmath95 , i.e. when @xmath54 and @xmath96 .
the minimum of entropy density actually occurs at @xmath97 , implying that @xmath98 at this extremal state the shear viscosity is given by @xmath99 the positivity of extremal entropy density implies that @xmath100 and hence @xmath101 for the @xmath102 and @xmath103 for the @xmath54 solutions .
it is quite remarkable that the lower bound found above , i.e. @xmath104 , is similar to a lower value of @xmath2 found at some relativistic heavy ion collision experiments @xcite .
the bound @xmath105 was found in @xcite by considering a ricci flat horizon or assuming that the boundary theory is defined on a flat space . here
we note that , in the @xmath106 case , the lower value of @xmath2 can be slightly stronger than that in flat space .
especially , in a flat space , a non - violation of causality requires the square of local speed of graviton on a constant @xmath107-hypersurface to be less than unity , i.e. @xmath108 or @xmath109 .
this limit can be slightly altered by the maxwell term @xmath110 and also by @xmath111 type corrections to maxwell fields @xcite as a function of @xmath112 , which is the case only with the gauss - bonnet gravity , then the perturbation of the transverse gravitons can get decoupled from the fluctuations of matter fields . ] .
specifically , in the presence of a maxwell charge @xmath113 , the ratio @xmath2 is given by @xmath114 , where @xmath115 . in the standard case of an non - extremal black hole one may be required to satisfy two separate conditions @xmath116 and @xmath117 , leading to the bound @xmath118 as first reported in @xcite . for the consistency of a dual gauss - bonnet gravity action , and also from a viewpoint of classical stability of gb black holes ,
one is required to take @xmath119 respectively , in the ads@xmath120 and the ads@xmath121 cases .
next let us briefly discuss the results in the @xmath1 gravity , which is obtained by setting @xmath122 and @xmath123 in eq .
( [ main - action ] ) . in this theory
, we find @xcite should be corrected by a factor of 2 : the source of this difference is that in @xmath1 gravity the entropy function is increased by a factor of @xmath124 , in the case of a ricci - flat horizon , which was however not considered there . ]
@xmath125 this result shows that the kss bound @xmath126 can be violated for a finite @xmath127 , unless that the contribution to @xmath2 coming from next - to - leading order terms , such as @xmath128 terms , are of a comparable magnitude to that of the @xmath20 terms .
this scenario is , however , almost unlikely since the couplings associated with the cubic and higher powers in riemann tensors , such as @xmath39 with @xmath129 , must be suppressed in a sensible derivative expansion .
notice that the limit @xmath130 , as implied by eq .
( [ main - sqrm ] ) and required for positivity of @xmath2 , is the same as implied by the positivity of extremal entropy of a @xmath1-corrected ads black hole .
this result is consistent with the following observation .
the conformal anomaly of a four - dimensional cft can be identified by considering the theory in a curved spacetime and writing @xmath131 where @xmath132 and @xmath133 are , respectively , the four - dimensional euler density and the square of the weyl curvature , and @xmath134 , @xmath135 are the two central charges of a dual cft . as explained by hofman and maldacena in @xcite
( see also @xcite ) , for the consistency of a low energy action with sensible derivative expansion , one may be required to consider cfts for which @xmath136 in fact , the gravity dual of a class of cfts for which @xmath137 , must be defined in a curved background , so the choice @xmath102 may not be very physical , at least , in the presence of @xmath1 type corrections
. taking into account all three possibilities for the boundary topology that @xmath102 or @xmath138 , and demanding that @xmath139 , we find @xmath140 it is not known yet whether either of these limits applies to nuclear matters at extreme densities and temperatures , or heavy ion collision experiments , but it would be interesting to know anything specific to the universality of the result @xmath141 through numerical hydrodynamic simulations of data from high energy experiments , including the relativistic heavy ion collision ( rhic ) and large hadron collider .
recent developments in gravity / gauge theory duality , supersymmetric field theories and black hole mechanics have shown that ads black holes are excellent objects to study the properties of strongly coupled gauge theories . in this note
we have studied the limits on black hole entropy and shear viscosity for the simplest class of higher curvature - corrected black hole solutions defined on ads spaces .
we also considered the causal problem on the boundary field theory to see what kind of constraints we can get on the ratio @xmath2 ( where @xmath4 is the shear viscosity and @xmath5 is the entropy density ) so as to keep the theory phenomenologically viable and internally consistent .
it has been known for quite sometimes that the limit @xmath142 on the gauss - bonnet coupling can be viewed as a classical limit of a consistent theory of quantum gravity @xcite .
recent results coming from the studies of viscosity bound violation in higher curvature gravity @xcite have shown that the bound on the shear viscosity of any fluid in terms of its entropy density may be saturated , i.e. @xmath0 , by all gauge theories but only at large t hooft coupling , as this limit correspond to the cases where all higher - order curvature contributions are absent .
nevertheless , this bound is naturally in immediate threat of being violated in the presence of generic higher derivative and higher - order curvature corrections to the einstein - hilbert action .
it is important to note that by tuning of the gauss - bonnet coupling or the generic higher derivative couplings , the ratio @xmath2 can be adjusted to a small positive value .
causality violation might take place when @xmath2 is too low , i.e. @xmath143 .
we have shown that limits on curvature coupling can be imposed by demanding the positivity of extremal black hole entropy or by keeping boundary causality in tact , or both .
we have not found any obvious explicit bound on @xmath33 from the thermodynamics of spherically symmetric ads gauss - bonnet black holes .
this could however arise as a consequence of causality violation of a boundary cft .
the critical value of @xmath33 beyond which the theory becomes inconsistent is found to be related to the entropy bound for an ads gb black hole with a hyperbolic or euclidean anti - de sitter event horizon .
this remark applies also to the @xmath1 gravity .
some other inconsistencies of higher - derivative gravity , such as an appearance of tacyonic mode , or semi - classical instability at short distances , can also be related to classical limits on black hole entropy and viscosity bounds .
we conclude with a couple of remarks . in the presence of a maxwell type charge @xmath113 ,
the ratio @xmath2 is generally modified , which is given by @xmath144 , where @xmath145 . in the extremal limit ( @xmath146 )
, we put a restriction on the coupling @xmath33 such that @xmath147 , which guarantees that the gravitational potential of a black brane is positive and bounded .
the kss bound @xmath41 is saturated only in the extremal limit , while in general it is violated also by charged gauss - bonnet black brane solutions .
the bound @xmath148 found in @xcite with a nonzero charge is stronger than for pure einstein - gauss - bonnet gravity in flat space , namely @xmath149 .
our analysis showed that in the ads@xmath19 case , the entropy of a gb black hole could be negative for @xmath150 , leading to a possible violation of unitarity in this range .
as shown recently in @xcite , a scalar dilaton field coupled to the gauss - bonnet can have an additional non - trivial contribution to the ratio @xmath2 : a scalar charge generally lowers the kss bound . in some more recent papers @xcite , there have appeared new examples , including string theory constructions , where the standard result of einstein gravity , namely @xmath0 , is modified in a modest way .
i would like to acknowledge a fruitful collaboration and helpful discussions with naresh dadhich .
i am grateful to the organizers of the jgrg18 ( japan ) and the conference on particle physics , astrophysics and quantum field theory : 75 years since solvay ( singapore ) for their hospitality .
this work was supported by the new zealand foundation for research , science and technology grant no .
e5229 and also by elizabeth ellen dalton grant no . 5393 .
p. kovtun , d. t. son and a. o. starinets , phys .
lett . * 94 * , 111601 ( 2005 ) .
d. t. son and a. o. starinets , ann .
nucl . part .
* 57 * ( 2007 ) 95 .
y. m. cho and i. p. neupane , phys .
d * 66 * ( 2002 ) 024044 .
m. cvetic , s. nojiri and s. d. odintsov , nucl .
b * 628 * ( 2002 ) 295 ( 2002 ) .
d. l. wiltshire , phys .
b * 169 * ( 1986 ) 36 ; phys .
d * 38 * ( 1988 ) 2445 .
i. p. neupane , phys .
d * 67 * ( 2003 ) 061501 ; phys .
d * 69 * ( 2004 ) 084011 .
i. p. neupane and n. dadhich , class . quant .
* 26 * , 015013 ( 2009 ) .
a. buchel , j. t. liu and a. o. starinets , nucl .
b * 707 * , 56 ( 2005 ) .
a. buchel , r. c. myers and a. sinha , arxiv:0812.2521 [ hep - th ] .
d. molnar and p. huovinen , phys .
* 94 * ( 2005 ) 012302 ; p. romatschke and u. romatschke , phys .
* 99 * ( 2007 ) 172301 .
x. h. ge , y. matsuo , f. w. shu , s .- j .
sin and t. tsukioka , jhep * 0810 * ( 2008 ) 009 .
d. m. hofman and j. maldacena , jhep * 0805 * ( 2008 ) 012 .
[ arxiv:0803.1467 [ hep - th ] ] .
r. g. cai , z. y. nie and y. w. sun , phys . rev .
d * 78 * ( 2008 ) 126007 .
[ arxiv:0811.1665 ] .
r. g. cai , z. y. nie , n. ohta and y. w. sun , arxiv:0901.1421 [ hep - th ] .
n. banerjee and s. dutta , jhep * 0903 * , 116 ( 2009 ) .
[ arxiv:0901.3848 ] .
m. mia , k. dasgupta , c. gale and s. jeon , arxiv:0902.1540 [ hep - th ] .
y. matsuo , s. j. sin , s. takeuchi , t. tsukioka and c. m. yoo , arxiv:0901.0610 ; x. h. ge and s. j. sin , arxiv:0903.2527 . | the gravity / gauge theory duality has provided us a way of studying qcd at short distances from straightforward calculations in classical general relativity . among numerous results
obtained so far , one of the most striking is the universality of the ratio of the shear viscosity to the entropy density . for all gauge theories with einstein gravity dual ,
this ratio is @xmath0 .
however , in general higher - curvature gravity theories , including two concrete models under discussion - the gauss - bonnet gravity and the @xmath1 gravity - the ratio @xmath2 can be smaller than @xmath3 ( thus violating the conjecture bound ) , equal to @xmath3 or even larger than @xmath3 .
as we probe spacetime at shorter distances , there arises an internal inconsistency in the theory , such as a violation of microcausality , which is correlated with a classical limit on black hole entropy . |
the interplay between statistical mechanics and combinatorics has always been of great interest both for physicists and for mathematician .
the observations of razumov and stroganov in their seminal papers @xcite ( see also @xcite ) , and the body of work that followed , exemplifies an instance of such an interplay . what razumov and stroganov found was that the properly normalized components of the ground state of the dense @xmath2 model enumerate classes of so called fully packed loop ( fpl ) configurations of given topology .
the boundary conditions on the @xmath2 model are reflected on the symmetries of the fpl ( see @xcite for further explications ) .
an approach to these problems , that has revealed to be particularly powerful , was initiated by di francesco and zinn - justin @xcite .
they used the integrable structure of the dense @xmath2 model and generalized it by introducing spectral parameters , still preserving integrability .
the advantage in dealing with a more complicated problem was at the beginning a technical one : one could exploit the richer structure coming from the polynomial nature of the ground state of the generalized model to better handle it .
later it was realized that the idea of di francesco and zinn - justin opened new perspective on the problem by relating it to the study of so called qkz equations @xcite and affine hecke algebras @xcite , to algebraic geometry of certain affine varieties @xcite and allowing to push forward some further intriguing conjectures ( in part already proved @xcite ) relating the homogeneous specialization of the solution of certain qkz equation to refined enumeration of plane partitions . in the present work , we will study the qkz equations related to the so called two boundaries temperley lieb algebra @xcite . when specialized at @xmath3 the solution of such a system of equations is the ground state of the inhomogeneous @xmath2 loop model with so called open boundary conditions . for generic @xmath4 instead
, the problem is related to the study of certain laurent polynomial representations of the affine hecke algebras @xmath5 of type @xmath6 and of the doubly affine hecke algebras of type @xmath7 .
the main difficulty we encounter in dealing with two open boundaries is the absence of a completely factorized component .
the presence of such a component in the cases with other boundary conditions , previously studied , allowed to fix the unknown factor in the recursion relations which in turn allowed to derive the sum rule at the combinatorial point @xmath8 .
here we adopt the opposite strategy , we derive first the recursion relations by a method that circumvent the full knowledge of a component and then we use such recursions to fix the simplest component from which all the other can be derived . still we will show that at @xmath3 the recursion relations are not sufficient to derive the sum of the components
. therefore we must resort to a different derivation of the whole eigenstate , based on mappings to systems with a single open boundary . this way
we show that the degree is preserved and we compute the sum rule , which is given by the product of four symplectic characters , correcting some recent claims @xcite .
the paper is organized as follows . in section
[ algebra ] we introduce the two boundaries temperley lieb algebra and the representation of this algebra on extended link patterns .
the @xmath9 and @xmath10 matrices based on baxterization of the two boundaries temperley lieb algebra , solution of the yang - baxter equation and of the boundary yang - baxter equation , is presented in section [ scatt_matr ] .
then , in section [ qkz - section ] we introduce the qkz equations and explain some of the properties of their solution . in particular in section [ affine_sectio ] we explain the relation between our qkz equations and the representation theory of the ( doubly ) affine hecke algebras of type @xmath11 and in section [ app_rec_bulk ] we derive the recursion relations . in section [ sol_section ]
we concentrate on the case where the parameter @xmath12 of qkz assumes the value @xmath0 , we use in such a case the recursion relations to construct the solution of qkz . the specialization @xmath3 is studied in section [ specialization ] , where we explain how to derive the full solution of the qkz equations with open boundaries from the solution with mixed boundary conditions , by defining certain mappings of representations .
the problem we are going to consider is based on a boundary extension of the well known temperley - lieb algebra , called @xmath13 boundaries temperley lieb algebra @xcite .
the temperley lieb algebra @xmath14 can be defined as the free algebra with generators @xmath15 , for @xmath16 , and relations @xmath17 @xmath18 this algebra has an appealing graphical representation in terms of non crossing link patterns connecting @xmath19 points on the the top and @xmath19 points on the bottom of a finite strip , with the graphical rules a first extension of the tl algebra is obtained by adding to the generators of @xmath14 a so called boundary operator @xmath20 with the following commutation rules @xmath21 this algebra , which is sometimes called 1btl ( one boundary temperley - lieb algebra ) , appeared for the first time in the paper @xcite where it was called blob algebra , then it was studied in different contexts ( for example @xcite ) . here
we call it @xmath22 since it is naturally related to certain statistical mechanics loop models having closed boundary condition on one side and open boundary conditions on the other side .
graphically the generator @xmath20 can be represented as and the commutation rules correspond to the following graphical relations the algebra obtained by adding to the generators of the temperley - lieb algebra a generator @xmath23 `` to the left '' , with commutation relation @xmath24 will be called @xmath25 .
the graphical representation of the generator @xmath23 is similar to the one of @xmath20 and the commutation rules look as follows in the statistical mechanics model the boundary conditions are now exchanged with respect to the case with @xmath25 . finally we define the algebra corresponding to open boundary conditions on both sides @xmath26
if we add to @xmath14 both @xmath20 and @xmath23 and require them to commute among themselves . while the algebras @xmath14 , @xmath25 and @xmath22 are finite dimensional , this is not the case for @xmath26 .
it is quite easy to understand why by looking at the graphical representation : the commutation rules given above do not allow to erase the lines connecting the two boundaries .
this means that for example all the element in the following picture have to be considered as distinct .
however it has been proved @xcite that all finite dimensional irreducible representations come from a further the quotient of this algebra which consist in giving a weight @xmath27 to the lines going from one boundary to the other .
algebraically one has to distinguish the case @xmath19 odd and the case @xmath19 even , then introduce the following elements * * @xmath28 :* @xmath29 @xmath30 * * @xmath31 : * @xmath32 @xmath33 and take the quotients over @xmath34 in the rest of the paper we will moreover restrict to the case @xmath35 .
the reason will be explained in appendix [ recursion_bound ] .
the representation of @xmath26 we will be interested in acts on the space @xmath36 with basis labelled by _ extended link patterns_. this is the hilbert space corresponding to open - open boundary conditions .
we call `` extended link pattern '' a diagram with @xmath19 points on a line , numbered from left to right , and two more points , the first called @xmath37 is situated on the left of point @xmath38 ; the second called @xmath39 is situated on the right of point @xmath19 .
the point @xmath40 are either connected in pairs , or they are connected to the point @xmath37 or @xmath39 , by non intersecting curves .
here is an example the action of the generators @xmath26 is almost obvious from the graphical representation , for simplicity we restrict to the case @xmath41 .
in such a case each time we close a loop in the bulk we remove it and multiply the link pattern by @xmath42 . if instead we close a loop touching one of the two boundaries we simply remove it .
these rules are supplemented by the requirement that a line joining the two boundaries can be removed at the cost of multiplying the obtained link pattern a weight .
we call the weight of the removed line @xmath43 .
graphically this looks as follows in section [ projection ] we will consider also the representations of @xmath25 on the space of left extended link patterns that we call @xmath44 .
this space is the subspace of @xmath36 consisting of link patterns having no lines connected to the point @xmath39 and is the hilbert space for open - closed b.c .
considered in @xcite .
let us consider now the following @xmath45-matrix 1truecm [ rcheck ] = _ i(z , w ) = ( 0,0 ) ( 85,-5 ) where @xmath46 is the identity operator and @xmath47 ( recall that @xmath48 ) .
the matrix @xmath49 satisfies the yang - baxter equation @xmath50 in a model with boundaries , integrability is assured by the presence of boundary scattering matrix which satisfies the so called boundary yang baxter equation @xcite @xmath51 in the present case the nontrivial boundary conditions on the right are given by a right boundary scattering matrix @xmath52 of a baxterized form @xmath53 .
the nontrivial solutions of the right boundary yang baxter equation form a one parameter family , which in the case of generic @xmath54 reads [ boundary_scatt ] k_r(_r|z ) = . with @xmath55 in the case @xmath56 , which turns out to be the most interesting one for us and to which we restrict from now on , we have @xmath57 and the boundary scattering matrix reduces to k_r(_r|z ) = .
we consider left boundary scattering matrices as well but , in order to formulate the qkz equations , we require them to satisfy a modified boundary yang baxter equation in which we have a new parameter @xmath12 @xmath58 the explicit form of the left scattering matrices when @xmath59 is k_l(_l|z ) = .
given the representation of @xmath26 on the space of extended link patterns , let us consider a function @xmath60 from @xmath61 to this space .
the boundary qkz equations are the following set of equations for @xmath62 [ qkz1 ] _ i(z_i+1,z_i)(_l;z_1 , , z_i , z_i+1 , , z_n;_r ) = ( _ l;z_1 , , z_i+1 , z_i , , z_n;_r ) ; [ qkz2 ] k_r(_r|z_n)(_l;z_1 , , z_n;_r ) = ( _ l;z_1 , , 1/z_n;_r ) ; [ qkz3 ] k_l(_l|z_1)(_l ; z_1 , , z_n_r ) = ( _ l;s / z_1 , , z_n_r ) ; we expand now these equations on the natural basis of extended link patterns .
therefore we write @xmath63 where the sum runs over all the extended link patterns .
since given an extended link pattern , the notion of an arc opening or closing at a point is well defined , we will parameterize it by the location of open and closing arcs and use the following notation @xmath64 , each @xmath65 or @xmath66 standing for `` opening '' or `` closing '' .
a concrete example of our notation is 1truecm @xmath67 ( 0,0 ) ( 70,-5 ) it is useful to introduce the following operators on the space of laurent polynomials of @xmath19 variables @xmath68 @xmath69 @xmath70 these operators allows us to rewrite the qkz equations for the components of @xmath60 in the extended link patterns basis as follows . * from eq.([qkz1 ] ) it follows that if @xmath71 then we have @xmath72 otherwise , if @xmath73 then @xmath74 where the coefficients @xmath75 are defined by @xmath76 and can be either @xmath43 if in applying @xmath15 to @xmath77 we form a line joining the two boundaries or @xmath38 in case we do not form a line joining the two boundaries . * from eq.([qkz2 ] ) it follows that if @xmath78 then @xmath79 if @xmath80 there is only one preimage under @xmath20 different from @xmath81 itself , i.e. there is only one @xmath82 such that @xmath83 and the components of this preimage is given by @xmath84 the coefficient @xmath85 now is either @xmath43 if by applying @xmath20 to @xmath77 we form a line joining the two boundaries , or otherwise it is equal to @xmath38 . * finally from eq.([qkz3 ] ) it follows that if @xmath86 then @xmath87 if @xmath88 its preimage under @xmath23 , different from @xmath81 itself is unique @xmath89 the coefficient @xmath90 now is either @xmath43 if by applying @xmath23 to @xmath77 we form a line joining the two boundaries , or otherwise it is equal to @xmath38 .
it is known , from the seminal papers @xcite , that the qkz equations are related to the representation theory of affine hecke algebras .
this has been rediscovered recently in the context of the razumov stroganov conjecture @xcite .
let us explain this observation in our case ; this will lead us to consider the laurent polynomial representations of affine hecke algebras of type c introduced by noumi @xcite .
we start from eq.([qkz1 ] ) , and introduce different generators of the @xmath91 algebra @xmath92 . by recombining the terms of eq.([qkz1 ] ) we can rewrite it as @xmath93 where the operator @xmath94 acts on the polynomial part of @xmath95 .
we proceed in the same way for the other two qkz equation , by introducing two generators @xmath96 and @xmath97 , where @xmath98 .
then we can rewrite eqs.([qkz2],[qkz3 ] ) as @xmath99 the operators @xmath100 and @xmath101 , like the @xmath102 , act on laurent polynomials and are given by @xmath103 @xmath104 it is now a matter of some straightforward computations to show that the operators @xmath105 and @xmath106 satisfy separately the commutation relations of the generators of the affine hecke algebra @xmath107 of type @xmath6 @xmath108 indeed the representation of @xmath107 given by @xmath109 is well known and goes under the name of noumi representation ( actually in the noumi representation the parameters @xmath110 , @xmath111 and @xmath4 are independent ; see for example proposition 2.2 of @xcite ) . by adding also the operators @xmath112 , whose action on a polynomial is the multiplication by @xmath113 we obtain a representation of the doubly affine hecke algebra of type @xmath114 @xcite .
in the qkz equation , we are considering the action of two copies of the @xmath107 , one acting on @xmath36 , the other acting on a space of polynomial that for the moment we call @xmath115 .
then the vector @xmath95 can be interpreted as a map from @xmath36 to the dual of @xmath116 @xmath117 and the qkz equations simply state that it intertwines between the two representation .
therefore , since the representation on @xmath36 is irreducible @xcite , we conclude that @xmath118 .
this means that solving the qkz equations amounts first to find the irreducible representation of @xmath107 on laurent polynomials , dual to @xmath36 , and then to find the basis dual to the extended link pattern basis of @xmath36 . from eq.([qkz1comp1 ] ) it follows that a component in which the points @xmath119 and @xmath120 are not connected , will have the following form @xmath121 where @xmath122 is symmetric under exchange @xmath123 . in general if the consecutives points @xmath124 are not connected among themselves then @xmath125 and @xmath126 is symmetric in @xmath127 .
analogously , from eq.([qkz2comp1 ] ) , it follows that if the point @xmath19 is not connected to the right boundary @xmath39 then one has @xmath128 where @xmath129 is invariant under @xmath130 .
an analogous statement relative to the left boundary is true , namely if the point @xmath38 is not connected to the left boundary @xmath37 then the component can be written as @xmath131 where now @xmath132 is invariant under @xmath133 .
one can combine the previous remarks in order to extract more trivial factors .
for example the components having all the bulk points connected to the right boundary @xmath134 or to the left boundary @xmath135 have the following form @xmath136 @xmath137 where @xmath138 is a function symmetric under exchange @xmath139 and invariant under @xmath140 , while @xmath141 is a function symmetric under exchange @xmath139 and invariant under @xmath142 . as a consequence of the analysis in the previous section
, it follows that if we set @xmath143 , all the components of the solution of the qkz equation at length @xmath19 that do not lie in the image of @xmath15 are zero .
the subspace of @xmath26 with an arc between points @xmath119 and @xmath120 is isomorphic to the space @xmath144 .
more precisely , if we call @xmath145 the map @xmath146 which consists in adding an arc between @xmath147 and @xmath119 ( and renumbering the points ) , then @xmath145 is an isomorphism between @xmath144 and @xmath148 . since @xmath149 we can consider its preimage under @xmath145 .
we show now that such a preimage , @xmath150 , satisfies a set of modified qkz equations . in order to work this out
let us restrict for the moment to the case @xmath151 , i.e. when we add a small arc at right the end of our strip .
it is easy to show that the map @xmath152 intertwines the operators @xmath153 and @xmath154 acting on @xmath144 and the ones acting on @xmath148 ( by abuse of notation we do not adopt a different notation for operators acting on different spaces ) @xmath155 then let us consider the following matrix @xmath156 when this matrix acts on configurations in the image of @xmath157 , it can be written as @xmath158 where @xmath159 .
moreover if we notice that @xmath160 we obtain that @xmath152 intertwines between @xmath161 and @xmath162 @xmath163 therefore @xmath164 satisfies a modified set of qkz equations where eq.([qkz2 ] ) is substituted by the following equation [ qkz2modif ] c k_r(_r|z_n-2 ) p^-1_n-1_n ( , z_n-2,z_n-1 , z_n = q^2 z_n-1 , ) + = p^-1_n-1_n ( , 1/z_n-2,z_n-1 , z_n = q^2 z_n-1 , ) now let us suppose that we can find a function @xmath165 which is invariant under @xmath166 and such that @xmath167 then we have that @xmath168is a solution of the unmodified qkz equations ( [ qkz1],[qkz2],[qkz3 ] ) of length @xmath169 and .
therefore , apart from a factor independent from all the @xmath170 , it must be equal to @xmath171 @xmath172 since we are searching for laurent polynomial solutions of the qkz we require the function @xmath165 , solution of eq.([rel_rec ] ) , to be also a laurent polynomial .
this fixes the possible form of @xmath12 as function of @xmath4 .
indeed , in order to cancel the pole at @xmath173 in eq.([rel_rec ] ) , the function @xmath165 must be of the form @xmath174 the product must be such that the last term cancels the pole at @xmath175 .
this happens only if @xmath176 .
in section [ sol_section ] we will argue that the lowest value of @xmath177 for which it is possible to find a solution of the qkz equations is @xmath178 or @xmath0 .
in such a case the recursion relation takes the form @xmath179 with @xmath180 this is the recursion relation we were searching for
. it will be very important in the next section when we will construct a solution of the qkz equations .
actually we will need as well an analogous formula when we set @xmath181 for @xmath182 . of course we could repeat almost word by word the same derivation using this time a modified left boundary scattering matrix .
however , in doing so we would not be able to keep track of the normalization choice of the wave function @xmath95 , which is implicit in the function @xmath183 . we can avoid this problem by deriving the recursion relation when @xmath143 from the case @xmath151 .
the idea is simply to use eq.([qkz1 ] ) to move the spectral lines .
if we act on @xmath184 with the operator @xmath185 we obtain @xmath186 @xmath187 @xmath188 where the second equality follows from the recursion relation proved above . then we notice that when we act with @xmath189 on @xmath190 we have @xmath191 with @xmath192 this property can be restated in terms of intertwinings , just by substituting @xmath157 with @xmath152 in the l.h.s . and @xmath193 with @xmath194 in the r.h.s .
of eq.([transport1 ] )
. therefore we arrive at @xmath195 we can continue this way and move the lines @xmath196 and @xmath19 to the position @xmath119 and @xmath120 . by shuffling the indices we obtain therefore @xmath197 where the hats in the right hand side mean that the variables are removed .
we search for a solution of the qkz equations ( [ qkz1],[qkz2],[qkz3 ] ) , in the simplest situation , i.e. when we have only two internal points ( @xmath198 ) . for the moment
we leave the parameters @xmath12 and @xmath199 free and we will see whether they will be fixed by the solution .
let us start from the components @xmath200 and @xmath201 .
we have seen in section [ trivial_factors ] that they have the form reported in equation eqs.([factor - right ] , [ factor - left ] ) .
some tedious but straightforward calculations show that if we demand the laurent polynomials @xmath202 and @xmath203 to be constant in the @xmath113 , then there are no solutions of the qkz equation . if instead we allow the @xmath204s to be of degree width @xmath38 , we see that the requirement that @xmath205 determines @xmath0 and fixes completely ( except of course for an irrelevant global constant normalization ) both @xmath206 and @xmath207 @xmath208 @xmath209 we find @xmath210 by applying @xmath211 to @xmath212 @xmath213 then the qkz equations close if we have @xmath214 these conditions are satisfied only if @xmath215 .
armed with this simple solution we try to construct the solution at @xmath216 .
we notice that the representation of @xmath26 on extended link patterns at @xmath217 and @xmath218 is irreducible , therefore the discussion in section [ affine_sectio ] tells us that if we know a component of @xmath219 , then using the qkz equations as in @xcite we can reconstruct all the other components .
it turns out that making some minimal assumptions we are able to construct @xmath220 . for this
we have to come back to the bulk recursion relations of section [ app_rec_bulk ] . in order for them to be really useful
we need to determine as much as possible the unknown factor @xmath221 .
first of all using the qkz equation ( [ qkz1 ] ) we have @xmath222 moreover if we specialize at @xmath223 we have @xmath224 applying the bulk recursion relation to the left hand side and comparing it to the right hand side ( where we use eq.([factor - right ] ) ) we find a factor of @xmath225 @xmath226 repeating the same argument using @xmath227 and the recursion relation for @xmath228 we obtain a further factor @xmath229 then we assume that @xmath221 is only given by the product of the two factors above . with this expression for @xmath221 ,
the recursion relation for @xmath230 reads @xmath231 where @xmath232 the further assumption that we make is that these recursion relations completely fix @xmath230 . in section [ reconstruction ] we will show how to construct the solution of our qkz equation in the special case @xmath233 , using the known solution of the problem with mixed boundary conditions .
the solution we will found has a degree width @xmath234 in each variables @xmath113 , which means that the degree width of @xmath230 in each variables @xmath113 is @xmath235 .
if we suppose that the same remains true for generic @xmath4 at @xmath0 then the bulk recursion relations allow to fix @xmath230 completely .
indeed thanks to its symmetries , once given eq.([ref_phi ] ) , we know the values of @xmath230 for @xmath236 .
we have constructed @xmath230 , using lagrange interpolation in @xmath237 , up to @xmath238 .
it is actually quite remarkable that the interpolating formula turns out to be a laurent polynomial in the other variables even though it does nt look of this form .
this is also the case for all its expected symmetries .
once @xmath239 known , we have checked up to length @xmath238 that the qkz equations hold .
we believe that some representation theoretical argument should prove that the construction is valid for generic @xmath19 .
as mentioned in the introduction , when the parameter @xmath240 assumes the value @xmath241 , the solution of the qkz equation is also an eigenvector of the transfer matrix of the dense @xmath2 loop model with open boundary conditions .
let us recall this fact here briefly .
let us consider the lattice in the following picture the strip has horizontal length @xmath19 and vertically it extends to infinity .
each face in the bulk of the lattice is filled with the following configurations with probabilities @xmath242 and @xmath243 , while the boundary faces can be in the following configurations with probability @xmath244 , @xmath245 , @xmath246 and @xmath247 .
the end points of the lines on the bottom boundary are labelled from @xmath38 to @xmath19 .
we are interested in the probability of the connectivity patterns of such boundary points : a point @xmath119 is connected by a line either to another point @xmath248 or to the left boundary or to the right boundary .
therefore these connectivity patterns are encoded by the extended link patterns defined in the section [ extended ] . the standard way to deal with the question of finding the probability of a link pattern is trough a transfer matrix approach .
if one adds to the bottom a further double row as in the picture the probabilities of the different link patterns must be preserved .
this means that the vector @xmath249 of probabilities is an eigenvector of the double row transfer matrix defined above , with eigenvalue equal to @xmath38 .
one can show the integrability of this model by constructing a family of commuting transfer matrices @xmath250 , depending on a `` horizontal '' spectral parameter .
for this one need to introduce introduce the r and k matrices , which parametrise the probabilities by means of the spectral parameters , as in done for example in @xcite .
the existence of a family of commuting transfer matrices is a consequence of the yang - baxter and boundary yang - baxter equations .
this remains true even if we consider different bulk and boundary spectral parameters on each vertical spectral line @xcite and the double row transfer matrix depends on such parameters : @xmath251 .
when we restrict to the homogeneous problem ( all the vertical spectral parameters equal to @xmath38 ) it is not difficult to show that the following hamiltonian @xmath252 where @xmath253 and @xmath254 , commutes with the transfer matrix .
this is consequence of the fact that apart from a constant term one has @xmath255 the vector of probabilities is an eigenvector of @xmath256 with eigenvalue equal to zero , i.e. is the stationary measure of the stochastic matrix @xmath257 .
closed boundary conditions , for example to the right , are obtained by sending the boundary spectral parameter @xmath258 . if we consider systems with mixed boundary conditions ( closed - open or open closed ) or with closed bcs on both sides we can perform also only a partial homogeneous specialization .
for example if we have open bcs on the right and closed on the left we can take all the spectral parameters equal to 1 except the boundary one and the last bulk one .
the corresponding hamiltonian ( i.e. the logarithmic derivative of the transfer matrix in @xmath259 ) assumes again a simple form @xmath260 where @xmath261 .
this remark will be important in the following when we will prove certain mappings between the vectors @xmath95 with different boundary conditions and different size .
the general problem of finding the stationary measure in presence of different spectral parameter is strictly related to the qkz equations ( see @xcite for an extended discussion ) .
indeed the eigenvector of the transfer matrix @xmath262 can be normalized in such a way it is a polynomial in the spectral parameters .
moreover as a simple consequence of the ybe and of the bybe , at @xmath233 we have @xmath263 @xmath264 @xmath265 this means that ( by slightly changing the normalization , which makes it a laurent polynomial ) @xmath262 satisfies the qkz equations .
notice that at @xmath233 all the loops have weight @xmath38 . at the special point @xmath233 we can compute the sum rule for the components in the basis of extended link patterns . at this special point
each operator @xmath15 , @xmath20 and @xmath23 has an eigenvector with eigenvalue @xmath38 in the dual representation on @xmath266 , given by @xmath267 , where the sum extends to all the the link patterns .
this means that we have @xmath268 the sum of the components of @xmath95 is given by @xmath269 as a consequence of the qkz equation and of eq.([lefteigen ] ) we have @xmath270 @xmath271 and analogously @xmath272 therefore @xmath273 is a symmetric laurent polynomial invariant under @xmath274 of degree width @xmath234 in each variable , hence it determined by the values at @xmath275 distinct values of say @xmath237 .
the bulk recursion relations ( combined with the symmetry of the polynomial under exchange @xmath276 ) provide us with the value of @xmath277 at @xmath278 , which means only at @xmath279 points .
but this is not enough to construct @xmath273 by lagrange interpolation .
moreover it is also possible that at @xmath233 there could be some accidental simplifications leading to the appearance of a common factor in the solution of qkz for @xmath4 generic .
such a common factor should be canceled if one consider the lowest degree solution .
the aim of the following sections is to explain how to determine @xmath273 following a different path and will be the consequence of a stronger result . the idea will be to reconstruct the full vector @xmath219 from the knowledge of the solution of the qkz equations with a closed boundary conditions on the left and open on the right . for this
we need to construct certain intertwining maps which is done in the next subsection . from its definition
one can think for example at the algebra @xmath25 as a subalgebra of @xmath26 . however in the case
@xmath280 we can think at it as a quotient of @xmath26 or of @xmath281 .
indeed one obtains @xmath25 from @xmath26 by quotienting the relation @xmath282 , or from @xmath281 by relabelling @xmath283 and quotienting over the missing relation i.e. @xmath284 . in the present section
we want to implement this idea at the level of representations .
we construct a map : @xmath285 , which intertwines between the representation of @xmath25 and the one of @xmath26 with @xmath286 at a graphical level the map @xmath287 is obtained as follows : starting from the right remove the first two lines connected to @xmath39 and connect the unmatched points with an arc .
if the initial number of lines emanating from @xmath39 was even , repeat this procedure with the remeaning lines joining @xmath39 until the lines from @xmath39 are exhausted .
if the number of lines from @xmath39 was odd repeat the procedure until there is a single line connected to @xmath39 then remove this line and connect the unmatched bulk point with the left boundary @xmath37 .
here is an example the map @xmath287 is easier to visualize if we use a different diagrammatic representation of the components , in terms of doubly - blobbed arcs @xcite , then @xmath287 simply consists in removing the right blobs .
the other map we want to define is @xmath288 , which intertwines @xmath20 and @xmath157 @xmath289 at the diagrammatic level the map @xmath290 goes as follows : change the name of the the point the right boundary point @xmath291 .
if there are no lines connected to @xmath292 draw a line from it to @xmath37 .
if there are more than two lines connected to @xmath292 remove the second and the third one starting from the right and draw an arc connecting the unmatched bulk points . repeat this procedure until there are either one or two remaining lines . in the former case do nothing , while in the latter case remove the left most line and connect the unmatched bulk point with the left boundary @xmath37 .
here is an example the proof of eqs.([quoz1 ] , [ quoz2 ] ) is a tedious but strightforward graphical case by case analysis .
now we use the maps @xmath287 and @xmath290 to reconstruct @xmath219 from the known form of the solution of qkz with mixed boundary conditions @xcite and gain information on the sum rule .
first of all we notice that ( with a slight abuse of notation ) @xmath293 therefore we find that @xmath294 satisfies the qkz equations with mixed closed - open boundary conditions
. therefore it must be @xmath295 where @xmath296 is a symmetric polynomials in the @xmath113s invariant under @xmath297 and @xmath298 is the known solution of the qkz equations with mixed b.c .
@xcite . moreover since we have @xmath299 , where @xmath300 and the sum runs over all right extended link patterns , we obtain for the sum rule @xmath301 in @xcite it is proven that @xmath302 , this means that we have found two factors of the sum rule @xmath303 where @xmath304 is the symplectic character @xmath305 in order to find the other factors , we use the map @xmath290 . as done above with @xmath306 one can prove that @xmath307 where now @xmath308 is solution of the qkz equations with mixed boundary conditions at size @xmath292 and the parameter @xmath309 plays the role of a bulk spectral parameter .
we have also @xmath310 , and using again the results of @xcite we obtain for the sum rule @xmath311 to conclude the derivation we notice that these four factors exhaust the degree of @xmath273 ( and do not have common factors ) , therefore @xmath312 let us notice that eqs.([part_sum1],[part_sum2 ] ) provides us also with a way of finding all the components of @xmath313 without the need of solving the qkz equations for generic @xmath4 .
let us show this in a concrete example at @xmath216 .
component wise eq.([part_sum1],[part_sum2 ] ) read @xmath314 @xmath315
@xmath316 @xmath317 this system of equations is immediately solved @xmath318 @xmath319 @xmath320 giving an alternative way of computing the components of the ground state of the dense @xmath2 model with open boundary conditions .
we do not have a proof that this method allows to reconstruct the full ground state for a generic system size @xmath19 , but we have checked it up to @xmath238 , checking as well that the result coincide with the ones found by solving the qkz equation and then taking the limit @xmath321 .
in this paper we have considered the qkz equations related to the baxterization of the two boundaries temperley lieb algebra . by deriving a recursion relation satisfied by their solution we have been able to construct it explicitly . at the combinatorial point
@xmath322 the solution of the qkz equations is also the ground state of the dense @xmath2 loop model with open boundary conditions .
we have shown how to reconstruct this ground state starting from the ground state of the same model with mixed boundary conditions known from @xcite . in doing that we have also been able to prove the sum rule . in section [ affine_sectio ]
we have briefly discussed the relations between our problem and the theory of representation of ( doubly ) affine hecke algebras of type @xmath323 .
it would be interesting to develop further this observation , in particular making contact with a recent work of kasatani @xcite and hopefully with the theory of non symmetric koornwinder - macdonald polynomials
. it would be also interesting to find integral formulae as already done in the case of other boundary conditions @xcite in this paper we have only tangentially touched problems related to the combinatorics behind the homogeneous limit of the ground state of the @xmath2 model . in facts
it is not difficult to see that , making use of the mapping between model with different boundary conditions , one can easily explain certain observation reported in table 2 of @xcite .
we plan to come back to such matters and to generalization to even more exotic boundary conditions in the immediate future .
it is a pleasure for us to thank p. zinn - justin for many fruitful discussions .
we thank also the lptms ( umr 8626 du cnrs ) universit paris sud , where part of this work was done .
we acknowledge the financial support of the anr programs `` gimp '' anr-05-blan-0029 - 01 and `` sle '' anr-06-blan-0058 - 01 .
in this appendix we want to derive a boundary recursion relation similar to the bulk one , the motivation being twofold .
on one side we show that leaving the boundary weight @xmath54 free , the boundary recursion relation maps the solution of qkz at size @xmath19 to a solution of qkz at size @xmath196 and @xmath324 .
this provides a motivation to restrict to the case @xmath56 ( and similarly @xmath59 ) . on the other side
the recursion we find provides the most convenient way to write the laplace interpolation formula giving the unknown factor @xmath325 of @xmath326 . from the analysis of section [ trivial_factors ] adapted to the case @xmath327
, it follows that if @xmath328 or @xmath329 then all the components in which the point @xmath19 is not connected to the right boundary @xmath39 are zero .
the other components , namely the ones having an arc going from @xmath19 to @xmath39 are in one to one correspondence with the extended link patterns with one bulk point removed .
one simply removes the point @xmath19 and the arc from it to @xmath39 .
we call this map @xmath330 . in this section
we show that @xmath331 satisfy a modified qkz equation .
let us restrict ourselves to the case @xmath332 , the other one being completely analogous .
first of all we notice that @xmath333 satisfies the qkz eqs.([qkz1 ] ) for @xmath334 and eq.([qkz3 ] ) , these properties being a trivial consequence of the fact that @xmath335 satisfies the same equation .
then let us consider the following operator @xmath336 we see easily that @xmath337 as one expected , @xmath338 preserves the image of @xmath20 ( which are the components not identically zero when @xmath339 ) therefore if we let it act only on such a subspace we see that it assumes the following form @xmath340 where @xmath341 , @xmath342 , @xmath343 and @xmath344 is the same as @xmath345 in which we substitute @xmath54 with @xmath346 .
apart from the multiplicative factor in front of it , this has just the same form of the scattering matrix in eq.([boundary_scatt ] ) .
this is consistent with the fact that the algebra of operators @xmath23 , @xmath347 and @xmath348 is a @xmath349 with boundary loop weight @xmath350 . from now on we will restrict ourselves to the case when @xmath35 .
this implies in particular @xmath351 and we can write @xmath352 we conclude therefore that @xmath333 satisfies the following set of equations _
i < n-1(z_i+1,z_i ) p_r_n ( , z_i , z_i+1 , , z_n = q/_r ) = p_r(z_i+1 , z_i , , z_n = q/_r ) ; k_r(q^3/_r|z_n-1 ) p_r_n(
,z_n-1,z_n = q/_r;_r ) = p_r_n( ,1/z_n-1,z_n = q/_r;_r ) ; k_l(_l|z_1 ) p_r_n(_l;z_1,
,z_n = q/_r;_r ) = p_r_n(_l ,
s / z_1 ,z_n = q/_r;_r ) ; where @xmath353 . now let us suppose that we can find a function @xmath354 which is invariant under @xmath166 and such that @xmath355 then it is straightforward to see that @xmath356 is solution of the unmodified qkz equation with boundary parameter @xmath357 .
if we require that the solution @xmath165 of eq.([rel_rec1 ] ) is a laurent polynomial in @xmath358 , then we find that the parameter @xmath12 must be of the form @xmath359 and it is easy to check that the solution is @xmath360 therefore , calling @xmath361 we can rewrite eq.([rec1 ] ) as follows @xmath362 where @xmath363 is an unknown function of @xmath364 and @xmath309 , that we can fix for @xmath0 to be@xmath365 by using minimality arguments similar to the ones employed in section [ sol_section ] .
now we want to change a bit perspective on the role of the boundary parameter @xmath309 ; as a consequence of the bulk recursion relations , it follows that the function @xmath366 is a laurent polynomial in @xmath309 , invariant under @xmath367 and degree width @xmath368 .
this means that we can use the boundary recursion relation in order to construct @xmath369 by using lagrange interpolation in @xmath309 and the symmetries of @xmath369 under exchange of the @xmath113s .
forgetting a constant normalization factor that we can not fix , the result reads @xmath370 it is remarkable that this expression turn out not only to be a laurent polynomial in the spectral parameters @xmath113 , but it is also endowed with all the symmetry properties expected for @xmath371 .
we have checked for that for systems up to size @xmath238 , starting from the above expression one gets a solution of the qkz equations .
a. v. razumov , yu . g. stroganov , theor .
( 2004 ) 333 - 337 ; teor .
mat . fiz . 138
( 2004 ) 395 - 400 , arxiv : math.co/0104216 .
a. v. razumov , yu . g. stroganov , theor .
142 ( 2005 ) 237 - 243 ; teor .
fiz . 142 (
2005 ) 284 - 292 , arxiv : cond - mat/0108103 .
p. di francesco , j. stat .
( 2006 ) p09008 , arxiv : cond - mat/0607499 .
p. di francesco , j. stat .
( 2007 ) p01024 , arxiv : math - ph/0611012 .
p. di francesco and p. zinn - justin , theor . and
154 3 ( 2008 ) , 331 - 348 , arxiv : math - ph/0703015 .
p. di francesco and p. zinn - justin , j. stat .
( 2007 ) p12009 , arxiv:0709.3410 . | we consider the qkz equations based on the two boundaries temperley lieb algebra .
we construct their solution in the case @xmath0 using a recursion relation . at the combinatorial point
@xmath1 the solution reduces to the ground state of the dense @xmath2 loop model on a strip with open boundary conditions .
we present an alternative construction of such ground state based on the knowledge of the ground state of the same model with mixed boundary conditions and prove that the sum rule as of its components is given by the product of four symplectic characters .
luigi cantini |
interest in volumes of compact manifolds appearing in physics arises for several reasons .
for example , in instanton calculations it is necessary to integrate over the gauge group @xcite ; in quantum field theories over curved spaces , the volumes of the compact spaces enter very often @xcite . for entangled qubit states the geometry of the corresponding groups @xmath0 and quotient spaces @xmath1 are important ; etc .
the main purpose of this paper is to compute the volumes of several types of compact manifolds in a systematic way .
we proceed from the most symmetric cases ( spheres ) to the less symmetric ( flag manifolds ) .
essential ingredients are the scale ( `` radius '' ) of the manifold , the setting of coordinates and their ranges .
all these spaces will be riemannian , so we compute the volume with respect to that metric ; all our manifolds will also be _ homogeneous _ manifolds @xmath2 .
we define the manifold @xmath3 in terms of some parameters @xmath4 in some local charts , so as to cover the whole manifold .
the metric will look locally as @xmath5 the associated volume element is @xmath6 and the volume is ( for one chart only ) @xmath7 where @xmath8 is the range of the @xmath9 parameter ; much of our discussion will be about these ranges . in all cases
the scale of the object to be measured will be stated in a `` natural '' manner ; we shall make comparison with other choices made in the literature when convenient : it is not always easy to recognize the different conventions used .
in fact , it is only for spheres that the universal criterion is the unit radius , but we shall see that even in this case when we have a manifold described as a _ product _ of spheres , the volume will _ not _ always be the product of the natural sphere volumes ! this volume problem has been dealt with in the published literature , both physical and mathematical .
a fairly complete treatment for group manifolds is presented by marinov @xcite , corrected in @xcite ; the volume of the group is extracted from the propagator by an indirect ( and complicated ) method , in fact as a by - product of the author s study of curved path integrals @xcite . in macdonald @xcite
there is a general synthetic formula for any compact lie group , with arbitrary scale , in terms of the root lattice ; we feel that most physicists will find these formulas hard to apply . for the unitary unimodular group @xmath10
there is the very clear paper by c. bernard @xcite .
there are many other partial results in the physics literature , some of them contradictory .
it was in an effort to overcome these difficulties that we undertook this investigation .
a direct antecedent of our paper is the work of m. byrd @xcite , who studied the @xmath11 case thoroughly . in a further paper
we intend to look at the geometry of entangled states more closely @xcite ; see also the papers @xcite .
we start by collecting some formulas for _ spheres _ , as the most symmetric spaces : the isometry group ( in physics , this is called the little group ) acts transitively in the bundle of bi - planes , and therefore the _ sectional _ curvature is constant .
next is the case of _ projective _ spaces , in which the isometry group still acts transitively in the bundle of lines , and therefore the geodesics are all equal .
compact _ group manifolds _ are symmetric spaces and therefore the curvature is covariantly constant . there is a natural metric ( induced from the killing form in the lie algebra ) which generates an invariant measure ( haar , or more appropriately hurwitz ) , but also there is a structure close to the product of odd spheres : the two criteria clash .
the action of the group in the adjoint representation produces interesting orbits , the so - called _ generalized flag manifolds _ , which appear very often in geometric quantization , density matrices , entangled states , etc .
we compute also the volumes of these quotient manifolds ; there are still symmetric , or at least homogeneous manifolds , and therefore the _ scalar _ curvature is still constant @xcite .
the sphere @xmath12 is a maximally symmetric space ; in fact , the isometry group @xmath13 is transitive ( so spheres are homogeneous spaces ) and the isotropy group @xmath14 acts also _ trans _ ( from here on , we shall use _ trans _ for `` transitive '' ) in the unit tangent vectors ( hence permutes the geodesics , and all are equivalent , in particular closed with same length ) and also _ trans _ in the grassmannian of bi - planes ( hence the sectional curvature is constant , in particular the scalar curvature ) .
we have @xmath15 where @xmath16 is the space of lines and @xmath17 that of bi - planes , in the tangent space to a point .
the volume of the sphere of * unit radius * embedded in @xmath18 , @xmath19 , is calculated from the auxiliary formula @xmath20^n = \pi^{\frac{n}{2 } } = \int_{0}^{\infty } r^{n-1 } e^{-r^2 } dr \times \text{vol}(s^{n-1})\ ] ] or @xmath21 for completeness we include a proof by induction . if @xmath12 is embedded in @xmath22 as the equator , @xmath23 and @xmath24 is the latitude angle , with @xmath25 , we have for the metric @xmath26 we start by setting @xmath27 therefore as @xmath28 we find @xmath29 in particular @xmath30 explicit useful formulas for the even / odd cases are @xmath31 notice the volume is maximal for @xmath32 , and @xmath33 these `` volumes '' are numbers , they have no dimensions ( in hypersolid angles ) .
our convention for the radius @xmath34 is _ extrinsic _ , that is , depends on the embedding @xmath35 ; an equivalent _ intrinsic _ criterion is the length of the geodesic : @xmath36 applicable because all geodesics in the sphere are equivalent ( see remarks above ) .
notice also that the scalar curvature is @xmath37 = \text{tr}[\text{tr}[riemann ] ] = \frac{n(n-1)}{r^2}\ ] ] ( see @xcite ) , so @xmath38 is not a good scale as it depends on @xmath39 .
finally the volume of the unit ball is @xmath40
the projective spaces , @xmath41 , @xmath42 , @xmath43 , and @xmath44 share with the spheres the property of being two - point homogeneous spaces , or symmetric rank - one spaces @xcite : the isometry ( little ) groups act _ trans _ on directions , and hence permute the geodesics , which ( in the compact case ) would all have to be closed and of the same length .
these spaces are of constant _ covariant _ curvature , as they are symmetric , but not of constant _ sectional _ curvature , except @xmath41 of course .
one can see how the sectional curvature changes along the real bi - planes e. g. in @xmath45 @xcite .
so we have a natural characterization for the scale of projective spaces : namely , the length of a fiducial geodesic .
we shall define projective spaces as _ quotient of spheres _ , and we shall see that the natural scale for the geodesics is to have length @xmath46 , _ not _ @xmath47 as in the spheres .
@xmath41 is defined as the set of rays or lines or one - dimensional sub - spaces in @xmath48 : @xmath49 there are many equivalent characterizations : @xmath50 any line touches the unit sphere in @xmath48 in the two anti - podal points , hence the first characterizations .
the orthogonal group @xmath51 acts _ trans _ in the lines , with little group : stabilizer of a vector in @xmath52 times the two directions in @xmath53 , hence the last forms in equation .
now we _ define _ the volume of @xmath41 as a quotient : @xmath54 but this implies that the geodesic length of _ our _ @xmath41 is @xmath46 and not @xmath47 because the half - meridian from the the north pole to the south pole is already _ closed _ in @xmath41 ! notice the halving of the volume for @xmath41 is really neglecting a set of measure zero , because @xmath41 as a half - sphere , still has the anti - podal points in the equator identified ; but this is a set of measure zero .
thus we get @xmath55 , even though @xmath56 and @xmath57 : the radius has shrunk to 1/2 . the case @xmath58 is also interesting , because @xmath59 , and we shall discuss it later .
notice also @xmath60 is not orientable , hence the volume has to be properly defined with the modulus of the measure . for @xmath61 spaces ( the notation , due to besse @xcite
, means @xmath62ompact , @xmath63ank-@xmath64ne , @xmath65ymmetric , @xmath65paces , precisely the sphere and the projective spaces ) a. weinstein has established the following result @xcite , which we shall not prove but verify in many examples . if @xmath3 is a @xmath61 , dim @xmath66 , geodesic length=@xmath67 , then @xmath68 with equal normalization . for our case of real projective spaces
we indeed get @xmath69 in agreement with besse @xcite .
@xmath42 has the following definition : @xmath70 here the complex lines in @xmath71 intersect the unit sphere @xmath72 in @xmath73 along a maximal circle @xmath74 .
we again _ define _ the volume of @xmath42 as a quotient @xmath75 using equation .
again for @xmath76 we have @xmath77 , whereas @xmath78 , and @xmath79 : obviously the geodesic length of _ our _ @xmath80 is only @xmath46 , whereas the volume of the equivalent @xmath81 space is @xmath82 times as big : the quotient circle includes the anti - pode ( @xmath83 lies inside @xmath74 ) , hence the geodesics are halved .
it is remarkable that @xmath84 the weinstein integer for @xmath42 is also easy to compute : @xmath85 in full agreement with besse @xcite . for a thorough mathematical study of @xmath45 as an instanton see @xcite .
@xmath43 has the following definition : let @xmath86 be the space of quaternions and @xmath87 the set of unit quaternions and @xmath88 , the symplectic group @xmath89 .
@xmath90 with the same definition as for @xmath91 or @xmath92 , we obtain @xmath93 with _ geodesic length _
equal to @xmath46 .
again @xmath87 contains the anti - podal point .
we obtain @xmath94 , whereas @xmath95 and @xmath96 . in this case @xmath97 where @xmath98 is the modified bessel function of the first kind .
the weinstein number is @xmath99 in full agreement with besse @xcite . over the division algebra of the octonions @xmath100 ( or cayley numbers )
there are only @xmath101 , which is equal to @xmath102 and @xmath44 ( the moufang plane ) , due to the non - associativity of the octonions ; and in fact , the definition is different . for example , @xmath44 is best defined as the hermitian , idempotent , trace - one elements of the exceptional 3x3 jordan algebra over the octonions @xcite .
but the formulas work just as well and one can formally define : @xmath103 in fact , the only two cases are @xmath104 and @xmath105 @xmath106 where @xmath107 is the fourth hopf bundle @xcite , which is not principal ; for @xmath44 see @xcite .
@xmath108 is the fifty - two dimensional second exceptional lie group .
thus @xmath109 and @xmath110 we have weinstein integer @xmath111=1 obviously , and @xmath112 as it should @xcite .
we have grouped the main results from sections [ sec : two ] and [ sec : three ] in table [ tbl : a ] : @xmath34 & @xmath113 & maximally symmetric + & & geodesic length = @xmath47 & & + real projective & @xmath41 & geodesic length = @xmath46 & @xmath114 & compact , rank one , + & & & & symmetric space + & & & & [ @xmath61 ] + complex projective & @xmath42 & geodesic length = @xmath46 & @xmath116 & [ @xmath61 ] + quaternionic projective & @xmath43 & geodesic length = @xmath46 & @xmath117 & [ @xmath61 ] + moufang plane & @xmath44 & geodesic length = @xmath46 & @xmath118 & [ @xmath61 ] +
group manifolds are still symmetric spaces ( i. e. with covariant constant curvature ) of rank @xmath121 1 .
we start with the @xmath122 case which includes @xmath119 and @xmath123 which will reveal already some complications . for @xmath124 : @xmath125 which implies @xmath126 and @xmath127 ;
that is latexmath:[\[\label{fourtwo } u= \begin{pmatrix } z_{1 } & z_{2 } \\ -z_{2}^ * & z_{1}^ * \end{pmatrix } \quad \text{with } \quad is identical with @xmath87 _
( this statement becomes tautological through the equivalence @xmath129 or @xmath130 in lie algebra notation ) .
it is instructive to compute now the volume ; write @xmath131 and @xmath132 where the ranges are @xmath133 and @xmath134 , @xmath135 , and @xmath136 .
so @xmath137 to calculate the volume , write @xmath138 thus the line element is @xmath139 with @xmath140 giving of course @xmath141 for the ranges given in equations and .
we carried out this elementary calculation because the parameterization was different than the one used in equations and ( [ eq : twofive ] ) .
a compact group manifold @xmath142 has a natural measure , namely the bi - invariant haar measure ( any locally compact space has a natural left - invariant haar measure , unique up to a constant ; for compact groups , in which the volume is finite , the name hurwitz measure is more appropriate ) . the simplest way to measure _ invariantly _
the volume of @xmath119 is to start from the cartan - killing metric in the lie algebra @xmath143 of @xmath142 @xmath144 ; \quad \text{for } x , y \in l(\euscript{g})\ ] ] and to induce , by the exponential map @xmath145 , a riemannian metric on @xmath142 , and hence a finite volume element as @xmath142 is always orientable as a manifold ( in particular it is parallelizable ) and compact .
one can use any representation @xmath146 instead of the adjoint ( ad ) . let us work with the @xmath142-invariant metric for @xmath147 and see if it coincides with the embedding metric as @xmath148 .
we start now from the `` euler '' form in the _ defining representation _ @xmath149 we need to fix the ranges of the angles @xmath150 and @xmath151 ; the safest way is to convert equation to the form given in equation , namely expanding equation with @xmath152 @xmath153 and comparing with equation we obtain @xmath154 the range of @xmath155 is _ half _ of the range @xmath156 and the jacobian is @xmath157 this is seen also directly from equation ( [ eq : fourtwelve ] ) : @xmath158 we therefore can choose to write equation as @xmath159 to compute now the volume we proceed as in @xcite : 1 .
compute @xmath160 ( @xmath161 ) in terms of the @xmath162 s and @xmath163 from equation ; that is , express the holonomic vector fields @xmath164 ( @xmath161 ) in terms of the _ invariant _ anholonomic vector fields @xmath165 .
invert , to express the invariant frame in terms of the coordinate frame .
dualize , to express the invariant co - frame in terms of the one - forms @xmath166 , @xmath167 and @xmath168 .
4 . express the volume element as the determinant ( jacobian ) of the change of frames .
this is a standard procedure , it is carried out in detail e. g. in @xcite , and the net result is @xmath169 integration with respect to the ranges given in equation , yields @xmath170 notice the @xmath171 missing in equation as compared with equation ( [ eq : foureight ] ) is compensated with the halving of parameter ranges for @xmath172 .
so the _ invariant _ volume coincides with the _ embedding _ volume .
now @xmath119 has a center @xmath173 , and in fact @xmath174 also @xmath175 , as it is obvious from @xmath176 .
so let us work the @xmath120 case .
for @xmath177 we write the `` euler '' formula , namely @xmath178 where we are in the _ adjoint representation _ , e. g.@xmath179 etc .
it is well known ( e. g. @xcite p. 24 ) that the ranges are @xmath180 we compute the volume as before , starting with @xmath181 , etc .
the final result is @xmath182 ( see e. g. @xcite p. 58 ) , and the volume is @xmath183 this is the correct volume for @xmath184 with geodesic length @xmath47 , which is our case . from equation , we had obtained @xmath185 with geodesic length equal to @xmath46 .
one might wonder the factor of 2 difference in the angles from equation , namely @xmath186 and here from equation , namely @xmath187 ; it is due to the change of generators : with the @xmath162 s we have @xmath188=2 i \sigma_z$ ] , but with the @xmath189 s implicit in equation the relation is the usual one @xmath190 = i j_z$ ] .
notice also the weinstein number @xmath191 : for equal length geodesics which is our case , we have @xmath192 .
a lie group @xmath142 is a symmetric space , @xmath193 , and hence of constant scalar curvature ; for rank @xmath194 the geodesics depend on directions , in fact they can be dense , as _ it is already the case _ for the torus @xmath195 .
another normalization is necessary though . let us start with @xmath11 .
in the vector , or defining representation , @xmath11 acts in @xmath196 , the action being _ trans _ in the invariant unit sphere @xmath197 with the isotropy subgroup @xmath119 @xmath198 the latter is a principal fibre bundle , in fact , _ locally _ one certainly has @xmath199 but the _ invariant _ volume is not quite the product of the volumes of the spheres . for many purposes ,
any compact lie group can be expressed as a ( finitely twisted ) topological product of odd - dimensional spheres . for a discussion of this point see @xcite .
consider the vector * * v**=@xmath200 ; it describes the whole of @xmath201 by actions of @xmath11 ; the infinitesimal transformation is @xmath202 where @xmath203 ( @xmath204 ) are the gell - mann matrices for su(3 ) which satisfy @xmath205=2\delta_{ij}.\ ] ] on the other hand , if we describe a point in @xmath201 with locally flat infinitesimal coordinates , we would then have a jacobian between the _ invariant _ coordinates @xmath206 and the _ sphere _ coordinates @xmath207 ; in fact @xmath208 the detailed calculation is in the appendix of @xcite .
the factor @xmath209 is just the `` stretching '' ( actually , contracting ) of the @xmath210 due to : 1 . commuting with @xmath211 and 2 .
satisfying equation .
the matrix representation of @xmath210 is necessarily then @xmath212 in other words : for calculating the _ invariant volume _ it is better to think of the second sphere @xmath201 as being `` stretched '' along a _
axis by the factor @xmath209 .
therefore , _ with our trace normalization _ ( equation ) the volume of @xmath11 is @xmath213 which is in agreement with _
most _ of the physics literature , e. g. @xcite .
if we parameterize a generic element of @xmath11 as @xcite @xmath214 one set of ranges of the @xmath215 s that reproduce equation is @xcite @xmath216 @xmath11 embodies all the complications for the @xmath10 series , the calculation of @xmath217 in terms of @xmath218 by induction is now a straightforward matter . for @xmath10 write @xmath219
the `` stretching '' relative to @xmath218 occurs again _ only _ in the `` last '' @xmath203 @xmath220 where @xmath221 = 2 $ ] implies @xmath222 and therefore @xmath223 an `` invariant '' way of calculating the `` stretching '' is this : write density matrices for @xmath10 as @xmath224 where @xmath225 and the @xmath226 s satisfy equation .
we have @xmath227=1 $ ] ; for pure states @xmath228 and imposing @xmath229=1 $ ] we obtain @xmath230 so the final formula for @xmath10 turns out to be @xmath231 where we have used @xmath232 in the last step .
equation ( [ eq : fiveeleven ] ) agrees with the corrected volume in @xcite and with @xcite interpreting his scale @xmath226 ( the lebesgue measure in @xmath233 ) as our `` stretching '' factors ; it agrees also with @xcite . on the contrary , a classical textbook @xcite , and a recent paper @xcite
omit these `` stretching '' factors . for the full unitary group @xmath234
there is a _ topological _ direct product decomposition @xmath235 which can be seen , for example , by factorizing a phase in the first vector component : @xmath236 the volume of @xmath234 thus _ depends on the radius _ of the @xmath237 factor ; if it is @xmath122 , i. e. not stretched , then @xmath238 finally , for the projective unitary group @xmath239 we obtain , as @xmath240 @xmath241
for the _ orthogonal groups _ @xmath52 and @xmath242 we shall proceed in a similar manner ; the rotation group @xmath242 acting on the vector representation leaves the unit sphere , @xmath243 , invariant with the isometry group @xmath244 : @xmath245 in fact equation is the principal bundle of the tangent to the sphere @xcite .
there is no `` stretching '' factor with the lie algebra convention in the _ vector _ representation @xmath246 = 2\delta_{ij}\ ] ] because the lie algebra of @xmath242 is comprised of antisymmetric matrices , always of the type @xmath247 = 2.\ ] ] therefore the volume calculation is elementary , because the spheres act like those with radius one .
induction starts at @xmath248 .
of course @xmath249 and we have @xmath250 which , e. g. gives @xmath251 which is the same value as in equation . for n
even / odd we get @xmath252 and @xmath253 in agreement with @xcite , and also with @xcite once a trivial factor of 2 is corrected in the even case
. however we disagree with @xcite .
the orthogonal group is neither connected nor simply connected ; so we have @xmath254 where @xmath255 is the universal double covering of the rotation group ( @xmath256 ) .
so we obviously obtain the result @xmath257 notice the first equation implies a topological direct product @xmath258 .
in fact , for odd @xmath39 , this is a direct product of groups .
notice also that @xmath259 has center @xmath173 , therefore the number of central elements in @xmath260 is @xmath261 , with two classes : @xmath262 there are no irreducible _ faithful _ representations of @xmath263 , so the spin group is represented through @xmath264 , which corresponds to @xmath265 and @xmath266 , the two chiral _ irreps _ , and @xmath267 which is the vector representation . in particular , there are _ three _ subgroups in the center of @xmath263 of type @xmath173 .
this explains triality for @xmath268 , because then @xmath269 . for the case
@xmath270 we have something interesting .
we compute @xmath271 whereas @xmath272 even though @xmath273 ! the `` stretching '' factor is the culprit , of course . for the _ symplectic groups _ @xmath88
the story is pretty much the same , but now the spheres jump by four : first of all @xmath274 with volume @xmath275 ( see equation ) .
now the induction is based in the fact that @xmath88 acts in @xmath276 unitarily , and therefore @xmath277 and it is easily seen that the action on the sphere is _ trans _ with isotropy group @xmath278 . therefore @xmath279 and therefore @xmath280 it can also been seen that there is no `` stretching '' in the lie algebra matrices of @xmath88 ( @xcite , p. 188 ) ; the reason is the same as for the orthogonal group . from equations and we obtain @xmath281 in full agreement with @xcite and @xcite . in this case , the product of spheres @xmath282 is both topological and metric , with radius one spheres .
in particular we obtain @xmath283 whereas before we obtained @xmath284 but @xmath285 , corresponding to cartan s @xmath286 .
it is remarkable that the _ same _ normalization , i. e. @xmath287 = 2 $ ] produces such different volumes in similar groups .
the reason is , of course , that the normalization is performed in _
different _ representations . as for the _ exceptional groups _ we just want to add formulas for the two first cases only , namely @xmath288 and @xmath108 .
the groups in the @xmath289-series , @xmath290 , although fundamental in @xmath291-theory , are yet to be fully understood .
now @xmath288 can be _ defined _ as the automorphism group of the octonions , or cayley numbers , @xmath100 .
the reals @xmath292 are of course invariant and so is the norm : @xmath293 therefore @xmath288 leaves the set of unit , imaginary , octonions @xmath32 invariant .
one can see also that the action is _ trans _ with isotropy equal to @xmath11 ( e. g. @xcite ) : @xmath294 so we obtain @xmath295 leaving aside a scale factor @xmath296 . as for @xmath108
, it can be _ defined _ as the isometry group of the cayley - moufang plane @xmath44 ; the dimension of @xmath108 is 52 and the little group is @xmath297 @xmath298 as shown in @xcite .
so the volume is @xmath299 where we have used equations , and have left a free normalization constant @xmath296 .
we have collected some of more important volumes in the following table : equation ( [ eq : fiveeleven ] ) + @xmath234 & @xmath287=2 $ ] & equation + @xmath259 & & equation + @xmath300 & & equation + @xmath88 & & equation ( [ eq : sixfifteen ] ) +
states of quantum systems are generally elements of some homogeneous manifold , @xmath301 . for example , pure states lie in @xmath302 , the infinite projective space ; if attention is directed to a finite number of independent states , as it is the case in quantum computing , encryption , entanglement considerations , etc .
the appropriate frame is a _
finite _ dimensional hilbert space , let us say @xmath71 .
pure states here lie in @xmath42 which is equal to @xmath303 ; marginally mixed states lie within the set of hermitian , unit - trace , positive operators : @xmath304=1,\ , \text{spectrum}[\rho ] \geq 0 \},\ ] ] where @xmath305 are all the @xmath306 complex matrices . when the spectrum is @xmath307 or @xmath228 ( idempotency ) we recover the _ pure _ states .
now the spectrum properties are conserved under conjugation and therefore the _ types _ of mixed states ( including pure ones ) are related to the orbits of the set given in equation under the unitary group .
now , up to permutation , the spectral type will be indicated by the number of coincident eigenvalues , subject to the general conditions contained in equation .
permutations are carried out by the weyl group ( the group generated by reflections in hyperplanes defined by the roots , see e. g. @xcite ( p. 284 ) ) . for @xmath10
the group is just @xmath308 , the permutation of the @xmath39 eigenvalues . up to permutation , therefore , the spectral types of density matrices are in one - to - one correspondence with partitions of the number @xmath39 ; we shall explicitly show this for the case @xmath309 , the results of which are collected in table [ tbl : c ] : @xmath310 & @xmath311 & 0 & unique ( max .
entropy ) + @xmath312 $ ] & @xmath313 & @xmath314 & 8 & pure @xmath315 + & @xmath316 & same & 8 & mix @xmath315 + @xmath317 $ ] & @xmath318 & @xmath319 & 12 & mixed @xmath320 + @xmath321 $ ] & @xmath322 & @xmath323 & 14 & mixed + @xmath324 $ ] & @xmath325 & @xmath326 & 16 & mixed + @xmath327 $ ] & @xmath328 & @xmath329 & 18 & mixed + @xmath330 $ ] & & @xmath331 & 20 & flag manifold + we remind the reader that the number of partitions for large @xmath39 is only known asymptotically ; the spectral type is self - explanatory ; notice pure states lie in @xmath332 ( @xmath309 in our case ) , but _ some _ degenerate mixed states will also make a @xmath332 orbit .
the following is to be noticed in the general case @xmath333 : 1 .
all these manifolds are homogeneous manifolds , hence they have constant scalar curvature ( except @xmath334 , of course , which is just a point ) .
spaces of the type @xmath335 are called `` grassmannian . ''
explicitly , @xmath17 is the complex grassmannian of @xmath336-planes in @xmath337 , to wit @xmath338 ( other labels , such as @xmath339 are also used ) .
in particular @xmath340 we already used the grassmannian notation , @xmath341 in section [ sec : two ] .
we shall define these orbits of the unitary group in the adjoint representation as _ generalized flag manifolds _ following an extended mathematical usage .
the genuine flag manifold corresponds to the @xmath342 $ ] partition : @xmath343 they enjoy interesting mathematical properties ; for example , they are spin and khler manifolds , in particular symplectic ( see for example @xcite ) .
it is remarkable that the simplest space @xmath344 appears as a space of exceptional holonomy @xcite .
we come now to the question of the _ volumes of the generalized flag manifolds _ ; because they are always _ homogeneous _ manifolds , @xmath301 , the volume is , of course , @xmath345 but , appearances to the contrary , this does not `` cut too much ice '' because the volume of @xmath346 depends on which subgroup it is identified with !
we illustrate the case @xmath58 ( @xmath248 is trivial ) , before attacking the more general case .
we have @xmath347 now @xmath348 also , removing the same @xmath237 factor , and @xmath349 ; this @xmath237 factor is `` the long one '' in @xmath45 , if one remembers that @xmath350 , so @xmath351 and the `` last '' lambda , @xmath210 is of the form diag@xmath352 .
the @xmath237 group , in this case , has the `` stretching '' factor @xmath209 ( see section [ sec : five ] ) ; hence @xmath353 which coincides with the canonical volume for @xmath45 from equation .
for the flag manifold we have @xmath354 thus @xmath355 which coincides , as in the @xmath45 case with the `` naive '' calculation without `` stretching '' @xmath356 now the same the result holds in _ all _ generality for the generalized flag manifolds ! the `` naive '' calculation ( i. e. neglecting the `` stretching '' factors ) _ gives the correct results_. that is @xmath357 the reason is as follows : both the numerator and the denominator in equation have the same rank ! therefore , the dimension of the cartan subgroups , in fact the cartan subgroups @xmath358 , @xmath359 , up to @xmath360 are _ exactly _ the same in the numerator and the denominator , so the `` stretching '' factors themselves cancel completely , and the calculation is thus reduced to the one with the other odd spheres only ( starting from @xmath87 ) .
so there is nothing else to calculate : for the generic flag manifold the computation is @xmath361 where @xmath362 that the generalized flag manifolds have homology of the product of projective spaces times spheres can be proved easily @xcite from equation . with not too much extra effort we can extend our results to generalized _ real _
flag manifolds : there are no `` stretching '' factors at all ! we refrain to give explicit formulas , other than to remark that in @xmath363 the volumes are computed from the volumes of the corresponding orthogonal groups from section [ sec : six ] with no corrections .
we hope our normalization conventions are plausible and our volume computations useful ; as we said in the introduction , there are many different results written in the literature .
we have not tried to state the conventions ( and , in some cases , the mistakes ) of all the authors ; rather we have attempted to produce a self - consistent , and uniform way of looking at the volumes of compact groups and some quotient spaces . except for spheres
, we do nt worry much about parameterizations of manifolds ; for @xmath364 see the explicit calculations in @xcite there are a few points that have been left out , that we want to recall .
we have not attempted to calculate the canonical volume for the @xmath365 groups , as the defining realizations are rather obscure .
we share the belief @xcite that these groups have to be better understood before attempting such a calculation .
also , the exact _ algebraic _ characterization of density matrices ( e. g. equation ) is not done , except in the simplest case of @xmath248 , for then @xmath366 for @xmath256 , the generic class is the flag manifold with the matrices congruent to @xmath367 and the positivity condition is @xmath368 or , in terms of the minors , @xmath369 which give rise to complicated algebraic restrictions that we shall not try to pursue further in this paper .
finally , see the recent paper @xcite which attempts to calculate the volumes of some compact einstein manifolds which appear in @xmath291-theory .
dr . luis j. boya s work was sponsored by mcyt grant fpa2000 - 1252 and wishes to thank prof .
r. f. schwitters for his hospitality at c. p. p. ( austin ) . in @xmath45 for example
, we have a complex structure , hence real bi - planes become complex lines : there are `` holomorphic '' complex lines @xmath371 with sectional curvature @xmath372 , and orthogonal complex lines @xmath373 with @xmath374 . interpolating
complex lines have intermediate curvatures .
ljb thanks m. santander ( valladolid ) for elucidating this point .
l. c. biedenharn and j. d. louck , _ angular momentum in quantum physics : theory and application , _ in _ encyclopedia of mathematics and its applications : vol . 8 , _ ed .
by gian - carlo rota , addison - wesley , massachusetts 1981 . | we present a systematic calculation of the volumes of _ compact _ manifolds which appear in physics : spheres , projective spaces , group manifolds and generalized flag manifolds . in each case
we state what we believe is the most natural scale or normalization of the manifold , that is , the generalization of the unit radius condition for spheres . for this aim
we first describe the manifold with some parameters , set up a metric , which induces a volume element , and perform the integration for the adequate range of the parameters ; in most cases our manifolds will be either spheres or ( twisted ) products of spheres , or quotients of spheres ( homogeneous spaces ) .
our results should be useful in several physical instances , as instanton calculations , propagators in curved spaces , sigma models , geometric scattering in homogeneous manifolds , density matrices for entangled states , etc .
some flag manifolds have also appeared recently as exceptional holonomy manifolds ; the volumes of compact einstein manifolds appear in string theory .
measures on manifolds , flag manifolds 28c10 , 22f30 , 14m15 , 43a05 |
magnetic vortex , characterized by the in - plane curling magnetization and out - of - plane magnetization at the core , has attracted much attention because of its promising applications in high - density storage devices @xcite . a vortex can attach itself to a magnetic domain wall , forming a vortex domain wall ( vdw ) .
a vdw in a magnetic nanowire is a localized spin texture and can move around . thus , a vdw can be a mobile information carrier when the information is encoded in the vortex core .
although a vortex , as a topological object , can not be destroyed or created by itself in sample bulk unless another vortex with opposite topological number is involved , it could be annihilated or created at sample surfaces . on the other hand , a vortex tends to move sideways toward sample surfaces due to gyroscopic effect .
one challenge is how to keep a vortex away from sample surface when it travels in a nanostrip wire . here
, we propose a concept of magnetic vortex guide to meet this challenge .
similar to the optical waveguide that guides electromagnetic wave , the magnetic vortex guide can confine a vortex to move within a designed region and avoids possible loss of vortices and the information carried by them .
this paper is organized as follows .
the theoretical analysis of vortex dynamics and vortex velocity is presented in sec .
ii . the numerical realization and discussion of the vortex guide are illustrated in sec .
iii followed by conclusions .
a vdw in a nanowire can be driven by a static magnetic field via energy dissipation @xcite , by an electric current via spin transfer torque ( stt ) @xcite , or by a microwave via both energy exchange and angular momentum exchange @xcite . in this article
, we consider only current - driven vdw motion because of both its fundamental interest and the technological importance
. generally speaking , stt has both an adiabatic component and a non - adiabatic component although the adiabatic one is normally much bigger than the other .
the motion of a vdw in a magnetic strip wire is mainly dragged by the motion of its vortex core when an electric current is passing through the wire .
stt drives a vortex core moving in both transverse and longitudinal directions .
the transverse motion of the vortex core is detrimental for the survival of a vdw because the vortex is destroyed once its core collides with the wire edge .
the information coded in the vortex will be lost upon this annihilation .
thus , it is important to keep the vortex core away from the wire edges if vdws are used as mobile information carriers . in order to understand the working principle of the vortex guides proposed below ,
we explore analytically the vortex dynamics .
the system we consider is a very thin magnetic film , whose length , width and thickness are respectively along the @xmath0 , @xmath1 and @xmath2-axis of the cartesian coordinate system .
generally , the vortex dynamics is governed by the thiele equation @xcite , @xmath3 where @xmath4 is the external force related to magnetic field that is zero in our case , @xmath5 is gyrovector that is zero for a transverse wall and @xmath6 for a 2d vortex wall , where @xmath7 is the winding number ( + 1 for a vortex and -1 for an antivortex ) , @xmath8 is the vortex polarity ( @xmath9 for core polarization in @xmath10 direction ) , @xmath11 is the thickness of nanowire and @xmath12 is gyromagnetic ratio .
@xmath13 is dissipation dyadic , whose none zero elements for a vortex wall are @xmath14 @xcite , where @xmath15 is nanowire width and @xmath16 is the thiele s domain wall ( dw ) width @xcite .
@xmath17 is proportional to the current density @xmath18 and is along the current direction , where @xmath19 , @xmath20 and @xmath21 are the current polarization , the bohr magneton and the electron charge , respectively . for permalloy
, @xmath22 m / s corresponds to @xmath23 @xmath24 ( @xmath25 ) . in our consideration , electrons flow in the @xmath26-direction .
@xmath27 and @xmath28 are gilbert damping constant and non - adiabatic strength , respectively .
both @xmath27 and @xmath28 are dimensionless parameters @xcite .
@xmath29 is the dw velocity .
one can recast the equation in terms of @xmath29 as a function of @xmath30 @xmath31 solving these two equations simultaneously gives @xmath32 as the nanowire width is on the same order of dw width @xcite and the damping constant @xmath27 is on the order of @xmath33 , which is much smaller than 1 , the velocities can be expanded to the leading order in @xmath27 and @xmath28 as @xmath34 @xmath35 the formula ( [ vy ] ) shows that the direction of the transverse motion of a vortex core depends sensitively on @xmath36 .
the transverse velocity is always non - zero if @xmath37 @xcite . for a vortex with a given topological number @xmath38 ,
say -1 , its transverse motion is opposite for @xmath39 as shown in fig .
1a and @xmath40 shown in fig .
based on above discussion of the transverse motion of a vdw core , one natural proposal is to use a superlattice wire , as schematically illustrated in fig .
1c , to control the transverse motion of the vortex .
the superlattice is made of two basic blocks a and b with @xmath36 taking opposite signs , say @xmath41 in a and @xmath42 in b. a vdw , say with @xmath43 , shall move upward in a and downward in b. by carefully designing the lengths of blocks a and b , it is possible to compensate the transverse motion of a vdw after a complete period so that the vdw travels longitudinally on average along the superlattice wire without annihilation at the wire edges . here
it s assumed that the average velocity is the mean value of vdw velocities in blocks a and b. the average vdw velocity is approximately equal to @xmath44 where @xmath45 and @xmath46 are respectively the damping constant in blocks a and b. it is worth mentioning that such a superlattice wire is possible to realize with today s technology . a recent study @xcite demonstrated that @xmath27 of permalloy can increase by four times through a dilute impurity doping of lanthanides ( sm , dy , and ho ) .
to demonstrate the feasibility of such a magnetic vortex guide , we consider a special superlattice in which @xmath47 has almost the same magnitude but opposite sign in the two unit blocks .
the lengths of block a and block b are 1000 nm so that the vortex core moves near the medial axis of the superlattice . in our micromagnetic simulations by using mumax package @xcite , a nanowire of 520 nm wide and 8 nm thick is used
. a moving simulation window of 2000 nm long is always entered at the vortex core . to mimic permalloy wires
, we use the exchange constant @xmath48 j / m , saturation magnetization @xmath49 a / m , and zero crystalline anisotropy . the mesh size is @xmath50 @xmath51 .
the non - adiabatic stt coefficient @xmath52 is used .
the damping coefficients in blocks a and b are 0.01 and 0.0215 , respectively .
figure 1d is the time - dependence of the vortex core position ( @xmath53 ) under a current density of @xmath54 , corresponding to @xmath55 m / s .
( @xmath53 ) is the coordinate of the core spin with the minimum @xmath56component of magnetization .
the initial vortex wall has @xmath57 and @xmath58 .
after the current is switched on at @xmath59 , the vdw starts to propagate and enters the guide around @xmath60 ns , indicated by the blue line .
the black and red lines in fig .
1d are respectively @xmath61 and @xmath62 .
the vortex core moves upward by 16 nm for about 2.2 ns in block a then downward by 16 nm for about 2.2 ns in block b alternatively , as expected .
the linear behavior of @xmath62 shows that the longitudinal velocity of the vortex core is @xmath63 m / s , very close to thiele equation result .
figure 1e is the current density dependence of vdw velocity and it agrees well with the theoretical prediction ( [ barv ] ) shown by the red line . and @xmath58 in a nanowire with @xmath39 ( * a * ) and @xmath40 ( * b * ) .
the white arrows denote the magnetization directions .
the x- , y- , and the z - axis are respectively along the length direction , the width direction and the depth direction
. the color is coded by the value of @xmath64 as indicated by the right bar .
( * c * ) the superlattice structure of a magnetic vortex guide . block a : @xmath65 .
block b : @xmath66 .
the zig - zag line sketches the trajectory of a vortex core ( @xmath67 ) in the vortex guide .
( * d * ) vortex core position ( @xmath53 ) as the functions of time @xmath68 under a current density of @xmath69 @xmath24 , corresponding to @xmath55 m / s .
the black and red lines are respectively for @xmath70 and @xmath71 .
the linear behavior of @xmath70 shows a longitudinal velocity of 475 m / s .
the blue line corresponds to the time when vdw enters vortex guide .
( * e * ) current density dependence of vortex velocity .
straight line is theoretic prediction.,title="fig:",scaledwidth=45.0% ] + obviously , a perfect superlattice vdw guide requires a complete compensation of transverse motion of a vdw inside blocks a and b. this is not easy to achieve because the motion of a vdw is sensitive to many material parameters including both @xmath27 and @xmath28 .
any fluctuation or uncertainty in these parameters shall lead to an overall shift in the transverse direction , no matter how small it might be .
after a long time and/or a long travel distance , the vortex might eventually collide with the wire edges and annihilate there .
furthermore , our current understanding of current - driven dw motion does not give us the power to predict the exact compensatory length for blocks a and b for given sets of material parameters .
one can , at most , use micromagnetic simulations to provide certain guidance if one wants to build a long vdw guide in reality .
thus , it shall be ideal if one can make a self - focusing guide for vdws like an optical fiber for light .
interestingly , a bilayer wire as shown in fig .
2a can confine a vdw with @xmath72 near the interface if the top layer ( block b ) is made of material with @xmath40 and the bottom layer ( block a ) is a material with @xmath39 . according to above formula for the transverse motion of a vdw , the vdw moves downward in layer b and upward in layer a. as a result , the vortex core shall move around the interface during its lateral propagation along the interface of the layered structures .
to demonstrate such a bilayer wire is capable of confining the core of a vdw of @xmath72 near the interface of the bilayer , we consider a bilayer wire with @xmath52 in both layers , and @xmath73 in layer a and @xmath74 in layer b. both layer a and layer b have the same width of 260 nm and the same thickness of 8 nm .
figure 2b shows @xmath62 and @xmath61 of a vdw under an electric current density of @xmath69 @xmath75 , corresponding to @xmath55 m / s which is switched on at @xmath76 .
an initially created vdw enters the vortex guide around 1 ns . according to fig .
2b , @xmath77 , the vdw is confined around interface .
the longitudinal motion of the vdw is at a constant speed ( linear curve for @xmath62 ) of @xmath78 m / s .
the transverse motion of the vdw is actually oscillatory with a very small amplitude ( 0.4 nm ) .
this oscillation feature is not captured by the thiele equation .
the reason may be that a vortex is an extensive and deformable object , instead of a rigid - body object that can be approximated as a point .
thus , when its core passes through the interface , the vortex can not adjust its structure fast enough so that the force acting on the vortex does not change its sign . it shall be a challenging problem to generalize the thiele equation such that the oscillatory motion of the vortex core near the interface of the bilayer wire can be described .
figure 2c is the current density dependence of vdw velocity and it s well described by the theoretical prediction shown by the red line . .
block b : @xmath40 .
( * b * ) vortex core position ( @xmath53 ) as the functions of time @xmath68 under a current density of @xmath69 @xmath24 , corresponding to @xmath55 m / s .
the red and black lines are respectively for @xmath70 and @xmath71 .
the linear behavior of @xmath70 indicates that the longitudinal motion of the vdw is at a constant speed of 474 m / s .
( * c * ) current density dependence of vdw velocity.,title="fig:",scaledwidth=45.0% ] + the vortex guide discussed above is only for @xmath72 ( @xmath7 is the winding number and @xmath8 is the vortex polarity ) .
the vortex guide for @xmath79 can be trivially obtained by exchanging layer a with layer b. although the results presented here are for strip wires of 520 nm wide and 8 nm thick , in which a vortex wall is stable @xcite , they are also similar for wires whose widths is so small that a transverse wall is the most stable dw but a vdw is still a metastable dw . in order to generate a metastable vdw , a notch assistant process can be used .
it is known that depinning of a transverse wall at a notch starts from the transformation of a transverse wall into a vortex wall @xcite .
it is also known that for an extremely small wire , a vdw may not be able to nucleate . for permalloy
, this limit is 24 nm .
thus 24 nm is the smallest width of a vortex guide if it is made of permalloy .
in conclusion , the concept of magnetic vortex guide is raised .
two vortex guide structures are proposed and numerically studied .
one is a superlattice strip whose two unit blocks have opposite signs of @xmath36 .
such a guide requires an exact compensation of the transverse motion of a vdw in blocks a and b , and high demanding in the accurate design of two basic blocks is needed .
the second type of vortex guide is a simple bilayer strip in which @xmath36 has opposite signs in two layers .
the possible strip width of a magnetic vortex guide is determined by the minimal strip width below which a vdw can not exist .
this work was supported by china nsfc grant ( 11374249 ) and hong kong rgc grant ( 605413 ) . h.y.y .
acknowledges the support of hong kong phd fellowship .
b. van waeyenberge , a. puzic , h. stoll , k. w. chou , t. tyliszczak , r. hertel , m. fhnle , h. brckl , k. rott , g. reiss , i. neudecker , d. weiss , c. h. back , and g. schtz , nature ( london ) * 444 * , 461 ( 2006 ) . | a concept of magnetic vortex guide is proposed and numerically studied . similar to the waveguides of electromagnetic waves ,
a magnetic vortex guide allows a vortex domain wall to move along a nanostrip without annihilation at the strip edges .
it is shown by micromagnetic simulations that a magnetic nanostrip of a properly designed superlattice structure or bilayered structure can serve as vortex guides . |
it is presently believed that large scale structures ( in particular clusters of galaxies ) are formed through hierarchical clustering . from this point of view
, clusters are formed by merging of smaller clusters or by groups falling onto a larger cluster . however , the effect of merging is not the same if a cluster undergoes a major merger or accretes small groups or even single galaxies ( salvador - sol et al .
1998 ) .
notice that merging does not occur isotropically around a cluster .
groups tend to fall onto clusters along filaments , thus explaining the preferential orientations often observed ( see e.g. west et al .
1995 , durret et al .
therefore the true orientation and the true ellipticity are interesting quantities to derive .
the imprints of such merging events can be seen directly on the x - ray images , but also in substructures which can be detected at optical ( density maps , velocity structure ) , x - ray and radio wavelengths ; this was first noticed by baier ( 1977 ) from optical data .
recent studies have revealed that about 50% of clusters show multi - component structure both in their galaxy distribution ( bird 1994 , escalera et al .
1994 , west 1994 , maccagni et al . 1995 ) and in x - rays ( mohr et al .
1993 , grebenev et al .
1995 , west et al .
1995 , zabludoff & zaritsky 1995 ) .
however one may notice that , when the amount of data grows and techniques of analysis are improved , this percentage appears to grow too ( girardi et al .
this leads to question the dynamical meaning of these substructures .
the properties of relatively poor and cool clusters are somewhat more difficult to analyze because of their limited signal , but prove to be interesting as a link between rich clusters and groups .
their relaxation time scales are larger and one can wonder if they are sufficiently virialized ( e.g. mahdavi et al . 1999 and references therein ) .
the properties of the x - ray emitting gas can be different from those of rich / hot clusters ( see e.g. david et al .
in particular , various relations between optical and x - ray properties ( as the @xmath8 - t@xmath10 relation which seems linked to global properties ) , x - ray / x - ray relations ( as the l@xmath10-t@xmath10 relation ) or x - ray to dynamical properties ( as the x - ray gas fraction ) may differ from those of rich clusters ( mahdavi et al .
1997 , markevitch 1998 , arnaud & evrard 1999 ) : they are probes to test cluster formation and evolution models , particularly the fine modeling of the gas heating ( cavaliere et al .
very few low temperature clusters are included in these relations , so adding even a single cluster at the border - line is important .
we present a multi - wavelength analysis of the cluster abcg 194 based on optical data taken from the literature and on x - ray data from the rosat archive , coupled with different techniques of analysis .
abcg 194 is a linear cluster of richness 0 and bautz - morgan type iii ( struble & rood 1982 ) with central coordinates @xmath11 and @xmath12 ( chapman et al .
1988 , hereafter cgh ) .
its average redshift is 0.018 , corresponding to a heliocentric velocity of 5340 km / s , and to a spatial scaling of 1.87 mpc / degree ( h@xmath13= 50 km / s / mpc , q@xmath13= 0 ) .
the velocity interval corresponding to the cluster is 4000 - 6600 km / s according to cgh , but we will show in section 4.2 that it is in fact 4300 - 6200 km / s . we have somewhat arbitrarily divided our analysis into two scales : the global aspect at large scale , and the smaller scale features . the first description is related to the cluster as an astrophysical object by itself , while for the second description , which takes into account as much as possible the kinematics , dynamical processes are important .
[ xya194 ] our optical analysis is based on the following data : - chapman et al .
( 1988 ) have observed a list of 74 cluster member galaxies in abcg 194 ( i.e. with velocities in the 4000 - 6600 km / s range ) , located within a radius of 3@xmath14 ( 5.6 mpc ) around the cluster center . the cgh sample is 97% complete for galaxies brighter than b=16.7 in a region of radius r=30 arcmin around the cluster center , and 80% complete for galaxies brighter than b=15.5 in the entire region . out of the 74 galaxies in this sample , 48 and 62
are located within radii of 30 arcmin and 1@xmath14 of the cluster center respectively .
for all these galaxies , cgh give the following data : coordinates , radial velocity , b magnitude , morphological type , major axis position angle and ellipticity .
- we have added to this sample 22 optically fainter galaxies selected from the cf2+ssrs2 redshift survey , for which barton et al .
( 1998 ) give positions and redshifts ( hereafter the bcg sample ) ; these authors determined that the region of 1@xmath14 radius around the abcg 194 cluster is 100% complete for galaxies brighter than m@xmath15 .
for the three galaxies with numbers 15 , 30 and 54 in the cgh list , we take the radial velocity from the barton et al .
( 1998 ) paper , where the error is smaller than in cgh . -
the coordinates and redshift for an additional ( seyfert ) galaxy were taken from knezek & bregman ( 1998 ) .
the total sample therefore includes 97 galaxies located within the following range of positions relative to the cluster center : @xmath16x@xmath176480 and @xmath18y@xmath178710 arcsec .
[ opt_x ] [ opt_radio ] our x - ray study is based on a rosat pspc image of abcg 194 retrieved from the rosat data bank ; the exposure time was 24482 seconds ( p.i .
murray & stephens ) .
this image was processed with the snowden software ( snowden et al .
1994 ) . obviously , this cluster is not a strong x - ray emitter , since in spite of the relatively high exposure time the total amount of counts is 8530 , which is small .
we show in fig .
[ opt_x ] a superposition of the optical and x - ray maps , and in fig .
[ opt_radio ] that of the optical and radio maps .
notice that there is a zoom by a factor of @xmath193 between these two images . in x - rays ,
the signal to noise ratio is low : we give in fig .
[ histopix2 ] the curves which allow to obtain the number of pixels above a certain count level .
99% of the pixels contain only 1 or 2 counts ( essentially noise ) for both clusters ; however , there are much more pixels with a large number of counts ( essentially signal ) in abcg 85 than in abcg 194 .
[ histopix2 ] in the center of this cluster are located the radiosources 0123 - 016a and 0123 - 016b .
the dumbbell " source 0123 - 016a coincides with two galaxies , one of them being ngc 547 ( 3c 40 ) , while 0123 - 016b coincides with ngc 541 ( ledlow & owen 1995 , edge & rttgering 1995 ) ; ngc 547 and ngc 541 are the first are third magnitude galaxies in the cluster respectively .
the radiosource 3c 40 has a twin jet " structure ( burns et al .
the optical disk of ngc 547 is perpendicular to the radio axis ( zirbel & baum 1998 ) and its direction coincides with that joining ngc 541 and 547 ( fasano et al .
ngc 541 is a narrow - angle - tailed ( nat ) radio galaxy .
the direction of the tail coincides with that between ngc 541 and 547 ( odea & owen 1985 ) . according to brodie et al .
( 1985 ) and van breugel et al .
( 1985 ) this radio source may be interacting with minkowski s object .
lccc & disc@xmath20 & disc@xmath21 & disc@xmath22 + & & & + r@xmath23 ( arcsec ) & 370 & 1600 & 2860 + number of galaxies & 26 & 42 & 78 + @xmath24 & 0.2 & 0.8 & 0.8 + direction angle & 52 & 45 & 85 + centre ( x y ) & [ cols="^,^ " , ] [ allx ] the wavelet analysis of the x - ray image reveals no substructure at middle and large scales . at the two smallest scales ( 2 and 4 pixels , or 30 and 60 arcsec ) ,
9 components ( at least ) are detected at a 3@xmath8 level ( see fig . [
orientation]b ) .
the snowden method gives 38 sources in the same field , among which 9 in common with ours .
the difference between the numbers of detections reached by both methods raises the question of their validity .
the snowden method is based on a convolution by the psf ( which varies with radius ) , and is therefore close to the wavelet method
. however , the statistics on which the detection thresholds are determined for both methods are not the same . our wavelet software estimates the image characteristic statistics by analyzing the whole image ( see e.g. slezak et al . 1994 ) while snowden considers only circles containing between 90% and @xmath25 of the encircled energy radius of the off - axis psf .
the average background is taken into account only in the cases where there are less than 4 counts per annulus .
this may explain the difference between the numbers of sources .
note also that the purpose of both methods is not the same : snowden s software is aimed at detecting everything that is not diffuse x - ray emission , in order to eliminate these point - like sources , and therefore it tends to find a higher number of small sources ( including possible cosmic rays ) , while our purpose is to detect only x - ray sources above a certain significance level .
source a corresponds to one of the two `` dumbbell '' galaxies , which are the first and third brightest cluster galaxies ; however the x - ray pspc pixel size is too large to be able to discriminate between both galaxies .
source b can be identified with the second brightest galaxy in the cluster .
source c coincides with the seyfert galaxy reported by knezek & bregman ( 1998 ) .
the radiosources 0123 - 016ab are also strong x - ray emitters ( burns et al .
note that these sources have velocities in the cluster .
source f appears to be a star ( it is star - like in the poss and there is no quasar at this position in the vron - cetty & vron 1993 catalogue ) . the other sources do not coincide with any galaxies with available redshifts .
note that in the part of the x - ray field shown in fig .
[ orientation]b , we detect all the sources reported by lazzatti et al .
( 1998 ) except one ( their source 26 ) .
we have estimated the x - ray luminosity of the nine detected sources , proceeding in three steps : * we fit the whole data by a @xmath26-model as explained in section 3.2 ; * we subtracted to the real image the synthetic image corresponding to the values obtained by the fit ; * in the image thus obtained , we have counted the photons around the peaks corresponding to the detected sources and give the corresponding results in table 5 ; we have checked that outside the sources the background is around zero , as expected ; * the luminosities were derived from the counts assuming a temperature of 1 kev , following the method described in section [ xana ] .
the luminosities found for these sources are much lower than those given by lazzati et al .
( 1998 ) , but the reason for such a discrepancy is not clear . we have compared our source counts with the results given by the rosat sass data processing pipeline , and find that they agree within 35% except for sources b and c. note that the specific @xmath26-model used for this estimate is not very important .
we have presented a morphological and structural analysis of abcg 194 both at optical and x - ray wavelengths . from dynamical and velocity dispersion studies ,
we are left with a sample of 91 galaxies really belonging to the cluster ( 4200 - 6100 km / s ) , the mean cluster velocity being 5326 km / s with a standard deviation of 360 km / s .
this value coupled with the temperature of 2.6 kev determined for this cluster from asca data falls on the @xmath8-t@xmath10 relation ( wu et al .
1998 ) . the main underlying structure ( 53 galaxies )
has the same mean velocity but a standard deviation of 300 km / s .
the x - ray map shows an ellipticity similar to that observed in the optical .
the x - ray luminosity of abcg 194 is @xmath27erg s@xmath28 , and falls notably below the l@xmath9-t@xmath9 relation for clusters ( markevitch et al .
1998 , arnaud & evrard 1999 ) , i.e. this cluster is underluminous for its temperature . on the other hand ,
the cluster is within the dispersion of the l@xmath9-@xmath8 relation ( mahdavi et al .
the @xmath26-model fit to the x - ray gas gives @xmath29 , which appears larger than expected from the low value of the x - ray gas temperature ( arnaud & evrard 1999 ) .
a @xmath26-model fit to the galaxy density distribution also gives values of @xmath26 consistent with 1 , as in the enacs sample ( adami et al . 1998 ) .
abcg 194 appears therefore to be a very poor cluster , comparable to those recently studied by mahdavi et al . ( 1999 ) , as confirmed by the stellar , x - ray gas and dynamical masses which we have calculated . the number of counts in angular sectors from the x - ray @xmath26-model center ( fig .
[ sectorx ] ) shows priviledged directions along the major axis pa found in x - ray and optical analysis , but also along an axis perpendicular to this one . except for individual sources ,
no x - ray substructures are found with a wavelet analysis .
only 9 x - ray sources are found with our method , while 38 are detected with the snowden software .
the various analyses described above show that at large scale abcg 194 is overall a relaxed cluster , with a few superimposed groups : the velocity distribution of the `` main '' relaxed underlying cluster is not far from gaussian . however , both at optical and x - ray wavelengths , a bright linear structure is observed in the central part , and emission is much stronger south east of this line than north west .
this central structure is well defined dynamically and is also rich in radio and x - ray galaxies .
we can try to interpret these results in the framework of hierarchical structure formation . as suggested by numerical simulations such as those by tormen et al .
( 1997 ) , the time interval between two possible major mergings is large compared to the timescale of violent relaxation .
the cluster therefore has time to relax between two mergers .
this is also the interpretation proposed for detailed features observed for example in the coma cluster ( biviano et al .
we suggest that in abcg 194 a group around the bright elliptical galaxy ngc 547 may have merged into the cluster ; due to dynamical friction , this group has stopped near the cluster center and was disrupted ( gonzlez- casado et al .
the galaxy excess towards the south east could then be a remnant of this group .
abcg 194 therefore appears as one of the very few poor clusters which are well studied at various wavelengths .
such objects are intermediate cases between rich clusters and groups , and as such are interesting to compare the physical processes taking place in these objects , which may be influenced by the very different relaxation timescales of these various systems .
we are very grateful to the franco - armenian pics and jumelage for making our collaboration possible through financial support for three stays of e.n . in france .
we gratefully thank dario fadda and eric slezak for giving us their wavelet analysis programmes , and christophe adami for performing a @xmath26-model fit to our data .
we also appreciated a discussion on radio sources with jacques roland .
last but not least , we thank the referee , andrea biviano , for several interesting criticisms and suggestions . adami c. , mazure a. , katgert p. , biviano a. , 1998 , a&a 336 , 63 arnaud m. , evrard a.e .
, 1999 , mnras 305 , 631 baier f.w . , 1977 ,
298 , 151 barton e.j .
, de calvalho r.r . , geller m.j . , 1998 , aj 116 , 1573 beers t.c . , tonry j.l
, 1986 , apj 300 , 557 bird c. , 1994 , aj 107 , 1637 biviano a. , durret f. , gerbal d. , le fvre o. , lobo c. , et al . , 1996 , a&a 311 , 95 brodie j.p . , browyer s. , mccarthy p. , 1985
, a&a 259,l31 burns j.o . , rhee g. , owen f.n . , pinkney j. , 1994 , apj 423 , 94 cavaliere a. , menci n. , tozzi p. , 1998
, apj 501 , 493 chapman g.n.f .
, geller m.j . , huchra j.p . , 1988 ,
aj 95 , 999 david l.p . ,
jones c. , forman w. , 1996 , apj 473 , 692 dressler a. , 1980 , apj 236 , 351 durret f. , forman w. , gerbal d. , jones c. , vikhlinin a. , 1998 , a&a 335 , 41 edge a.c .
, rttgering h. , 1995 , mnras 277 , 1580 escalera e. , biviano a. , girardi m. et al . , 1994 , ap j 423 , 539 fadda d. , slezak e. , bijaoui a. , 1998 , a&as 127 , 335 fasano g. , falomo r. , scarpa r. , 1996 , mnras 282 , 40 fukazawa y. , makishima k. , tamura t. , et al .
1998 , pasj 50 , 187 garcia a.m. , 1993 , a&as 100 , 47 girardi m. , escalera e. , fadda d. et al . , 1997 ,
apj 482 , 41 girardi m. , giuricin g. , mardirossian f. , mezzetti m. , boschin w. , 1998 , apj 505 , 74 gonzlez - casado g. , mamon g.a .
, salvador - sol e. , 1994 , apj 433 , l61 grebenev s.a . , forman w. , jones c. , murray s. , 1995 , apj 445 , 607 den hartog r. , katgert p. , 1996
, mnras 279 , 349 kriessler j.r . , beers t.s . , 1997 , aj 113 , 80 knezek p.m. , bregman j.n .
, 1998 , aj 115 , 1737 lazzati d. , campana s. , rosati p. , chincarini g. , giacconi r. , 1998 , a&a 331 , 41 ledlow m.j . ,
owen f.n . , 1995 , aj 109 , 853 maccagni b. , garilli m. , tarenghi m. , 1995 , aj 109 , 465 mahdavi a. , bhringer h. , geller m.j . , ramella m. , 1997 , apj 483 , 68 mahdavi a. , geller m.j .
, bhringer h. , kurtz m.j . , ramella m. , 1999 , apj 518 , 69 markevitch m. , 1998 , apj 504 , 27 markevitch m. , forman w.r .
, sarazin c.l .
, vikhlinin a. , 1998 , apj 503 , 77 mohr j.j . ,
fabricant d.f .
, geller m.j . , 1993 ,
apj 413 , 492 odea c.p . ,
owen f.n . , 1985 ,
aj 90 , 927 pislar v. , durret f. , gerbal d. , lima neto g.b . , slezak e. , 1997 , a&a 322 , 53 salvador - sol e. , sanrom m. , 1989 , apj 345 , 660 salvador - sol e. , sanrom m. , gonzlez - casado g. , 1993 , apj 402 , 398 salvador - sol e. , solanes , j. m. , manrique , a. , 1998 , apj 499 , 542 scodeggio m. , solanes j.m .
, giovanelli r. , haynes m.p . , 1995 ,
apj 444 , 41 serna a. , gerbal d. , 1996 , a&a 309 , 65 slezak e. , durret f. , gerbal d. , 1994 , aj 108 , 1996 snowden s.l . , mccammon d. , burrows d.n . , mendenhall j.a . , 1994 ,
apj 424 , 714 struble m.f . ,
rood h.j . , 1982 , aj 87 , 7 tormen g. , bouchet f.r .
, white s.d.m . , 1997 ,
mnras 286 , 865 van breugel w. , filippenko a.v . , heckman t. , miley g. , 1985 , apj 293 , 83 vron - cetty m .-
, vron p. , 1993
, _ a catalogue of quasars and active nuclei _ , 6th edition , eso scientific report no .
13 west m.j . , 1994 , in `` clusters of galaxies '' , eds .
f. durret , a. mazure & tran than van , editions frontires west m.j . , jones c. , forman w. , 1995 , apj 451 , l5 wu x .-
, fang l .- z . ,
xu w. , 1998 , a&a 338 , 813 zabludoff a.j .
, zaritsky d. , 1995 , ap j 447 , l2 zirbel e.l . ,
, 1998 , apjs 114,177 . | we present a morphological and structural analysis of the richness zero cluster abcg 194 , known as a `` linear cluster '' . this study is based on a catalogue of 97 galaxies with b magnitudes and redshifts belonging to the cluster , a rosat pspc image and radio data .
we show that the overall large scale structure is rather smooth and comparable at optical and x - ray wavelengths .
the cluster is elongated along pa @xmath0 ; however it appears as `` linear '' when taking into account only galaxies in the very central part ( the axial ratio varies from 0.2 in the central region to 0.8 for a larger region ) .
we have obtained the galaxy density profile and despite the very low x - ray emission the x - ray emitting gas density profile .
we have estimated the x - ray gas and dynamical masses up to the limiting radius r@xmath1 of the pspc image ( respectively @xmath2 and 8 10@xmath3 m@xmath4 ) and the stellar mass up to r@xmath1 and to 3 mpc ( 3 @xmath5 and 7.5 10@xmath6 m@xmath4 ) . at smaller scales ,
the analysis of both positions and velocities reveals the existence of several groups but which are not strongly linked dynamically ; however a main structure with a nearly gaussian velocity distribution is exhibited .
the velocity dispersion is small ( @xmath7 = 360 ) .
a wavelet analysis of the x - ray image reveals no large scale substructures .
small scale x - ray sources are detected , mainly corresponding to individual galaxies ; we give an estimate of their luminosities .
the lack of strong substructuring suggests that abcg 194 is overall a relaxed cluster .
abcg 194 is a poor and cold cluster ; we compare how its characteristics fit into various correlations found in the literature , but generally for richer / hotter clusters , such as the @xmath8-t@xmath9 , l@xmath9-t@xmath9 , l@xmath9-@xmath8 relations or the ratios of various kinds of masses . |
type ia supernovae ( sne ia ) , the result of the thermonuclear explosion of c / o white dwarfs ( wds ) , provided the first clear evidence of the universe s accelerating expansion @xcite and have been used to measure various cosmological parameters ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the two main progenitor scenarios that likely lead to sne ia are the single - degenerate ( sd ) channel , when the wd accretes matter from a nondegenerate companion star ( e.g. , * ? ? ?
* ) , and the double - degenerate ( dd ) channel , which is the result of the merger of two wds ( e.g. , * ? ? ?
* ; * ? ? ?
while it is unclear how often either of these scenarios occur , it seems likely that both are actually present in nature . for a handful of extremely nearby sne ia
, many plausible sd scenarios have been ruled out ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the so - called super - chandrasekhar mass sne ia likely result from the dd scenario ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . on the other hand , photoionization and subsequent recombination of the circumstellar medium ( csm ) created by the progenitor system itself has been observed in a few relatively normal sne ia @xcite , and csm has likely been detected in the spectra of at least 20% of sne ia in spiral galaxies @xcite .
furthermore , extreme interaction with csm has been observed in a relatively small number of overluminous sne ia mainly via the detection of strong h@xmath0 emission .
these `` hybrid '' objects also resemble type iin sne ( sne iin ) and have been dubbed sne ia - csm @xcite .
previously , it was not completely clear whether these objects are actually sne ia or instead a new subtype of core - collapse sn ( e.g. , * ? ? ? * ; * ? ? ? * ) .
this controversy seems to have been settled by the discovery and analysis of ptf11kx @xcite
. discovered on 16 jan .
2011 ( ut dates are used throughout this paper ) by the palomar transient factory ( ptf ; * ? ? ?
* ; * ? ? ?
* ) at redshift @xmath5 @xcite and with galactic reddening @xmath6 mag @xcite , it was shown to initially resemble the somewhat overluminous type ia sn 1999aa @xcite , though with saturated h&k absorption lines and weak d lines .
this implies the object was almost certainly a sn ia that had significantly affected its immediate surroundings beginning shortly after explosion .
subsequent spectra of ptf11kx presented by @xcite showed time - variable absorption features of , , , and , which ( except for ) had not been seen in any previously observed sn ia . in addition , ptf11kx revealed a strong h@xmath0 line with a p - cygni profile ( indicative of an expanding shell of material ) whose emission component gradually increased in strength , causing the spectra of ptf11kx to eventually resemble those of other ia - csm objects @xcite .
the observations of @xcite indicate the presence of multiple csm components with slower - expanding material exterior to faster material and with velocities of @xmath3 50100 km s@xmath2 .
recent high - resolution observations of rs oph @xcite and models of circumstellar shells created in such systems @xcite seem to match many of the ptf11kx observations ; thus , @xcite suggest that ptf11kx was a _ bona fide _
sn ia with a symbiotic nova progenitor ( i.e. , a sd scenario , but see * ? ? ?
extending at least some of these findings to _ all _ of the ia - csm objects such that we can say that they are all sne ia , likely coming from a sd system , perhaps in the form of a symbiotic nova scenario , is the goal of @xcite . in [
s : spectra ] we present eight late - time spectra of ptf11kx , starting with the last spectrum shown by @xcite from 124 rest - frame days past maximum brightness and continuing through 680 d past maximum .
we measure various spectral features and analyze the spectra in [ s : analysis ] , and in [ s : comparisons ] we compare the spectra of ptf11kx with those of other sne .
we summarize our conclusions in [ s : conclusions ] .
figure 1 of @xcite shows spectra of ptf11kx from 3 d before @xmath7-band maximum brightness ( which was on 29 jan .
2011 ) to 88 d past maximum , though one extra spectrum ( observed 130 d past maximum , corresponding to 124 d in the rest frame ) is listed in their table s2 .
this observation is the first one presented in this work , where we extend the spectral coverage of ptf11kx to 680 rest - frame days past maximum brightness .
low - resolution optical spectra were obtained with the intermediate dispersion spectrograph and imaging system ( isis ) on the 4.2 m william herschel telescope ( wht ) , the low resolution imaging spectrometer ( lris ; * ? ? ?
* ) on the keck - i 10 m telescope , and the deep imaging multi - object spectrograph ( deimos ; * ? ? ?
* ) on the keck - ii 10 m telescope .
all spectra were reduced using standard techniques ( e.g. , * ? ? ? * ) .
routine ccd processing and spectrum extraction were completed with iraf , and flux calibration and removal of telluric lines were done using our own idl routines .
table [ t : ptf11kx ] lists information regarding the ptf11kx spectra analyzed here , and the data are plotted in figure [ f : ptf11kx ] . upon publication ,
all spectra presented in this paper will be available in electronic format on wiserep ( the weizmann interactive supernova data repository ; * ? ? ? * ) .
lcllrr ut date & age ( d)@xmath8 & instrument@xmath9 & range ( ) & res .
( )@xmath10 & exp . (
s ) + 2011 jun .
8 & 124 & isis & 35009498 & 3.5/7.2 & 3600 + 2011 nov . 7 & 270 & isis & 35009500 & 3.5/7.2 & 1800 + 2011 dec .
26 & 316 & lris & 57927440 & 3 & 600 + 2011 dec .
26 & 316 & lris & 340210148 & 3.9/6.1 & 900 + 2012 feb .
21 & 371 & lris & 327210250 & 3.8/6.0 & 655 + 2012 mar .
15 & 393 & lris & 32755628 , 58097427 & 3.6/3 & 600 + 2012 apr .
29 & 436 & lris & 32365630 , 57307360 & 3.7/3 & 625 + 2013 jan . 9 & 680 & deimos & 45009636 & 3 & 1800 + + + + even though more than 550 rest - frame days are spanned by the data shown in figure [ f : ptf11kx ] , there is not much spectral evolution ( until the final spectrum presented ) .
two of our highest signal - to - noise ratio ( s / n ) observations are separated by @xmath3 100 d and are nearly identical ( + 270 and + 371 d ) . most of the emission features have disappeared in the latest spectrum shown , from 680 d past maximum brightness .
all that is left is the relatively blue `` quasi - continuum , '' which is likely produced by emission from many overlapping narrow lines of iron - group elements ( iges , mostly ) excited by the csm interaction , and h@xmath0 emission , which has weakened since the previous spectrum ( see below for further details ) .
conspicuously , the near - infrared ( ir ) triplet , which is one of the strongest emission features in most of the late - time ptf11kx spectra , is almost completely gone by day + 680 .
to better characterize the spectral features seen in ptf11kx , we follow the procedure of @xcite and fit multiple gaussian components to the h@xmath0 and h@xmath4 emission lines .
the h@xmath0 profiles of our ptf11kx spectra are shown in figure [ f : hal_zoom ] .
clearly the profiles consist of a broader ( fwhm @xmath1 km s@xmath2 ) component combined with a narrow ( fwhm @xmath11 km s@xmath2 ) , unresolved component .
while p - cygni profiles were seen in high - resolution spectra of ptf11kx @xcite , the spectra presented herein have relatively low resolution , so we do not expect to observe such subtle features .
figure [ f : ew_b_hal ] shows the temporal evolution of the equivalent width ( ew ; _ top _ ) and flux ( _ bottom _ ) of the broader h@xmath0 component .
we find that the broader h@xmath0 component ( after removing the narrow component ) is mostly symmetric and slightly blueshifted , perhaps due to the csm being accelerated somewhat by the sn ejecta .
its ew increases nearly linearly with time until our latest spectrum ( from @xmath3 680 d past maximum brightness ) , when the ew drops significantly .
the h@xmath0 flux also generally increases with time until a large drop is measured in the last spectrum .
perhaps the decrease in the strength of h@xmath0 emission at + 680 d is indicative of the sn ejecta finally overtaking the majority of the csm . in their @xmath12 power - law model of sne
iin ( with @xmath13 m@xmath14 yr@xmath2 km@xmath2 s ) , the h@xmath0 flux is predicted by @xcite to decrease by a factor of @xmath3 2 from 1 to 2 yr after maximum brightness .
the factor of @xmath3 4 decrease shown by ptf11kx is larger than this prediction , but it occurs at similar epochs .
this may indicate that the density structure of the csm and/or mass - loss history of the ptf11kx system is different than what was nominally assumed by @xcite .
@xmath15{ew_b_t_ha } \\
\includegraphics[width=5in]{flux_b_t_ha } \\
\end{array}$ ] the h@xmath0 profile in the spectrum from 680 d past maximum exhibits a possible decrease in flux in the red wing as compared to the blue wing .
the dashed line on the bottom spectrum in figure [ f : hal_zoom ] is the reflection of the blue half of the h@xmath0 profile across the peak flux .
this suppression of the red wing has been seen in many sne iin and is often interpreted as new dust forming in the post - shock material ( e.g. , * ? ? ?
* ; * ? ? ?
this is also observed in all other sne ia - csm , but beginning much earlier ( @xmath3 75100 d past maximum brightness ; * ? ? ?
narrow emission lines of [ ] @xmath163727 , [ ] @xmath165007 , and h@xmath4 are apparent in many of the spectra shown in figure [ f : ptf11kx ] .
the temporal evolution of the flux of these lines , along with the flux of the narrow component of h@xmath0emission , is displayed in the top panel of figure [ f : ew_comp ] .
the flux values have been scaled ( by factors of 2 , 2.5 , and 5 for [ ] @xmath163727 , [ ] @xmath165007 , and h@xmath4 , respectively ) in order to emphasize their common evolution .
the bottom panel of figure [ f : ew_comp ] shows the ews of these narrow features , again scaled ( by factors of 2 , 4.5 , and 7.5 for [ ] @xmath163727 , [ ] @xmath165007 , and h@xmath4 , respectively ) to highlight their common changes with time .
there may be some concern that slit losses , seeing effects , or contamination from nearby regions cause the observed variation in these lines with time .
however , all observations later than 300 d past maximum brightness were taken under relatively good seeing conditions ( @xmath17 ) using a 1slit .
in addition , the observations from 316 and 371 d past maximum use position angles that differed by @xmath3 90 , and yet they yield nearly _ identical _ flux ( and ew ) measurements for all 4 narrow emission lines . on the other hand ,
the spectra from 371 and 393 d past maximum were obtained using the same position angle , but these data yield significantly _ different _ flux ( and ew ) values for each spectral feature .
the flux measurements were obtained from the spectra displayed in figure [ f : ptf11kx ] after scaling each one to nearly contemporaneous photometry of ptf11kx ( or a fit to the linear decline of the light curve ) shown in figure [ f : lc ] below .
the flux and ew of all four emission lines appear in general to decrease , remain constant for @xmath3 150 d , then perhaps increase again .
however h@xmath0 , which is a recominbation line and can have a different timescale than the forbidden lines , sometimes behaves significantly differently when compared with the other lines .
this could mean that the sn ejecta of ptf11kx are interacting with csm that is clumpy or perhaps composed of multiple thin shells , which is consistent with what was seen in the early - time spectra @xcite .
it seems unlikely to observe this behavior if the ejecta were expanding into csm with a monotonically increasing or decreasing density profile , as one might expect from a with a constant mass - loss rate .
this conclusion is supported by the extreme late - time overluminosity of ptf11kx ( see [ s : comparisons ] ) .
nevertheless , we caution that perhaps some of this variability could be caused by slit losses , seeing effects , or contamination from nearby regions .
@xmath15{flux_t_comp_scaled } \\ \includegraphics[width=5in]{ew_t_comp_scaled } \\
\end{array}$ ] when comparing both the broader and narrower components , h@xmath0 is significantly stronger than h@xmath4 at all epochs .
the h@xmath0/h@xmath4intensity ratio is @xmath18 in all of our late - time spectra of ptf11kx .
it varies throughout our observations , but it appears to generally increase with time , peaking at a value of @xmath3 15 at @xmath19 d. a large , time - variable h@xmath0/h@xmath4 intensity ratio seems to be a hallmark of the sn ia - csm class @xcite .
another likely tell - tale sign of an object being a sn ia - csm is weak emission @xcite .
ptf11kx exhibited weak @xmath165876 at early times @xcite .
figure [ f : old_spec_he ] shows the @xmath165876 region in our late - time ptf11kx spectra ; we see hints of emission from this feature in some ( but not all ) cases .
the clearest @xmath165876 emission is in the form of an unresolved narrow line in our final spectrum at 680 d past maximum brightness .
there is no evidence of emission from @xmath167065 in any of our spectra , except possibly very weak , unresolved emission again in the 680 d spectrum .
overplotted ( gray dotted line ) on the data with the `` strongest '' resolved emission in figure [ f : old_spec_he ] is the @xmath165876 profile at @xmath3 400 d after discovery of sn 2010jl , a somewhat luminous sn iin that probably came from a massive star ( thus likely _ not _ a sn ia of any flavor ; * ? ?
* ; * ? ? ?
the @xmath165876 emission is stronger and at higher s / n in sn 2010jl than any of the possible @xmath165876 detections in ptf11kx at similar epochs .
the most similar feature in the ptf11kx spectra is the possible resolved line on day + 371 ( see also fig .
[ f : ptf11kx ] ) .
in figure [ f:11kx_comp ] we plot the spectrum of ptf11kx from 316 d past maximum brightness , one of our highest s / n late - time observations .
in addition , we show late - time spectra of three other sne . at these late epochs , ptf11kx resembles the sne ia - csm more than any other sn type .
therefore , we follow the spectral feature identifications that @xcite used for the other prototypical sn ia - csm , sn 2002ic ( which are also valid for sn 2005gj ) . the spectrum of ptf11kx is dominated by relatively narrow h@xmath0emission , which is of course also seen in sne 2005gj and 2010jl ( but not in sn 1999aa ) .
the h@xmath0 emission is somewhat weaker in ptf11kx relative to that of sn 2010jl , but comparable to that of sn 2005gj .
similarly , narrow h@xmath4 emission is detected in ptf11kx , but it is substantially weaker than h@xmath0 ( as mentioned in
[ s : analysis ] ) .
the strength of the h@xmath4 feature in ptf11kx appears to be intermediate between those of sn 2010jl and sn 2005gj .
higher - order balmer emission is easily seen in sn 2010jl , but completely undetected in ptf11kx and sn 2005gj .
the broad peak in ptf11kx near 3900 and in ptf11kx and sn 2005gj near 8500 are almost certainly caused in part by h&k and the near - ir triplet , respectively , and this latter feature is one of the strongest in both spectra ( with fwhm @xmath3 10,000 km s@xmath2 ) .
note that emission from these complexes is also seen in sne iin and sne ia at late times .
in contrast to the + 316 d spectrum of ptf11kx shown in figure [ f:11kx_comp ] , there is practically no emission from the near - ir triplet in the spectrum from 680 d past maximum brightness ( see fig .
[ f : ptf11kx ] ) . a model for sne ia - csm that includes
a cool dense shell made up of many small fragments containing separate fe - poor and fe - rich zones predicts that emission should disappear once the two zones become fully mixed @xcite .
this was seen to happen in sn 1999e at between 139 and 361 d past maximum @xcite , and we find the same behavior in ptf11kx but at a later epoch ( between 371 and 680 d past maximum ) .
in all of the spectra shown in figure [ f:11kx_comp ] , the blue flux dramatically increases shortward of @xmath3 5700 .
the largest increase in flux at the blue end of the optical range appears in sn 1999aa , which comes from nebular emission lines from very iron - rich ejecta , excited by radioactivity . on the other hand ,
the smallest increase in blue flux appears in sn 2010jl , where this blue `` quasi - continuum '' probably comes from numerous blended , narrow lines of iges excited by the csm interaction .
ptf11kx ( as mentioned in
[ s : spectra ] ) and sn 2005gj have intermediate amounts of increased blue flux , and since we see evidence of strong csm interaction in these objects , this `` quasi - continuum '' is most likely a stronger version of what is seen in sn 2010jl and less like the nebular emission lines of more normal sne ia ( such as sn 1999aa , also see below ) . as most of the optical flux in ptf11kx shortward of @xmath3 5700 comes from numerous ige multiplets , it is difficult to identify individual features in that wavelength range ( aside from h&k and h@xmath4 ) , though some relatively narrow emission features can be tentatively identified . the nearly complete lack of emission at @xmath3 7700 in ptf11kx ( as well as sn 2005gj ) leads us to believe that there is little to no oxygen present in their late - time spectra .
nebular oxygen emission is a hallmark of core - collapse sne , and it appears in the spectrum of sn 2010jl presented in figure [ f:11kx_comp ] . thus , this supports the notion that ptf11kx , and all sne ia - csm in general , are in fact genuine sne ia .
furthermore , ptf11kx and sn 2005gj exhibit almost no ( see also [ s : analysis ] ) , while the spectrum of sn 2010jl shows obvious emission lines from @xmath165876 and @xmath167065 .
the almost total lack of oxygen and helium emission in the very late - time ptf11kx spectra imply that the broad , weak emission near 7300 is caused either by [ ] , as it likely is in sn 2010jl , or by [ ] and [ ] , which is what is seen in more typical sne ia at these epochs .
we favor the former option since ptf11kx shows no convincing evidence for strong , broad emission features due to [ ] , [ ] , or [ ] in any of the observations .
these forbidden emission features from iges that are typically seen in more normal sne ia at late times ( @xmath20 d ; e.g. , * ? ? ?
* ) may not be detected in ptf11kx since they have been diluted by broad - band continuum flux from the ongoing csm interaction .
@xcite found that ptf11kx is @xmath3 16 times more luminous than typical sne ia at @xmath3 100 d past maximum brightness , while more recent observations ( shown in fig . [ f : lc ] ) find that it is @xmath3 300 ( @xmath3 2600 ) times more luminous at @xmath3 460 d ( @xmath3 680 d ) past maximum . the extremely slow decline rate at late times [ @xmath3 0.2 mag/(100
d ) ] and the large late - time luminosity of ptf11kx are inconsistent with a csm model that includes a wind with a constant mass - loss rate ( i.e. , @xmath21 , * ? ? ?
this model was first proposed for sn 2002ic by @xcite , who also showed it to be inconsistent with late - time observations of that object . at these late epochs , ptf11kx is spectroscopically nearly identical to sn 2005gj ( and to the other sne ia - csm as well ; e.g. , * ? ? ?
it looks quite unlike sn 1999aa , even though it closely resembled this object spectroscopically at early times @xcite .
nebular spectra of other types of sne ia were also compared with those of ptf11kx , including the overluminous sn 1991 t @xcite and the normal sn 2011fe ( bianco et al . in prep . ) , but as with sn 1999aa , they did not provide good matches .
a variety of other sne iin were compared to ptf11kx , and in all cases they either closely resembled the spectrum of sn 2010jl presented in figure [ f:11kx_comp ] or looked completely different from both sn 2010jl _ and _ ptf11kx .
finally , we compared ptf11kx to late - time spectra of the normal type ic sn 1994i @xcite and the grb - associated type ic sn 1998bw @xcite ; we found that while both have a similar amounts of blue flux , the spectra of sne 1994i and 1998bw are dominated by [ ] @xmath166300 , [ ] @xmath22 , and ] @xmath164571 emission , which is in stark contrast to the nebular spectra of ptf11kx ( and the other sne ia - csm ; e.g. , * ? ? ? * ) .
ptf11kx was a sn ia with high levels of interaction with its csm , and it likely had a symbiotic nova progenitor @xcite . in this work
we have presented and analyzed late - time spectra of ptf11kx ranging from 124 to 680 d past maximum brightness . while the sn shows very little overall spectral evolution during these epochs ( except for the last spectrum ) , we find that the broader ( @xmath3 2000 km s@xmath2 ) h@xmath0emission appears to increase in strength with time for @xmath3 1 yr , after which time it decreases significantly , perhaps indicating that the sn ejecta have overtaken most of the csm .
there are also indications from the narrow h@xmath0 component that the ejecta of ptf11kx are interacting with csm that is clumpy or perhaps made up of multiple thin shells .
this is consistent with what was seen in the early - time spectra and the model of a symbiotic nova progenitor which had multiple eruptions prior to the sn @xcite .
we also find that ptf11kx has an unusually large h@xmath0/h@xmath4 ratio that varies with time , as well as extremely weak emission ( significantly weaker than more normal sne iin )
. both of these are hallmarks of sne ia - csm @xcite . aside from h@xmath0 , the late - time spectra of ptf11kx also show strong , broad emission features ( fwhm @xmath3 10,000 km s@xmath2 ) .
the emission disappeared nearly completely in the + 680 d spectrum , which agrees with a model of another sn ia - csm @xcite .
this spectrum also indicates the possibility of newly formed dust in the post - shock material as evidenced by a slight decrease in the red wing of h@xmath0 compared with the blue wing . at all epochs little to no emission
is detected from oxygen , , and balmer lines ( aside from h@xmath0 ) , which leads to a large observed h@xmath0/h@xmath4 intensity ratio in ptf11kx . when comparing our late - time spectra of ptf11kx with those of other sne
, we have shown that it does not resemble sne ic and only bears a passing resemblance to more typical sne ia ( ones that follow the @xcite relation ) .
ptf11kx shows a blue `` quasi - continuum '' ( due to numerous blended , relatively narrow lines of ) , but broad emission features from [ ] , [ ] , and [ ] ( which are seen in more normal sne ia at late times ) are absent .
these features are completely diluted by the broad - band continuum flux from the ongoing csm interaction .
the extremely slow decline rate at late times , @xmath3 0.2 mag/(100 d ) , and the large late - time luminosity , 49 mag brighter than a typical sn ia , are almost certainly caused by the interaction between the sn ejecta of ptf11kx and the csm .
ptf11kx was a highly special case of a rare type of sn ( @xmath3 1% of the sn ia population ) being discovered relatively early ( @xmath3 7 d after explosion ; * ? ? ? * ) .
the main observational properties of ptf11kx described above are shared by other sne ia - csm , as well as by more typical sne iin .
separating sne ia - csm from sne iin and studying the sn ia - csm class in greater detail are undertaken by @xcite . despite this work and the fortuitous discovery and extensive follow - up observations of ptf11kx , questions still remain .
models of sne iin , such as those by @xcite , seem to be somewhat applicable to sne ia - csm , and models of other sne ia - csm ( e.g. , * ? ? ?
* ; * ? ? ?
* ) appear to match the observations of ptf11kx .
these models are a great starting point for future theoretical work that we hope will utilize the observations and analysis presented herein .
we would like to thank j. s. bloom , k. clubb , a. a. miller , and a. morgan for their assistance with some of the observations , b. dilday , o. fox , and l. wang for helpful discussions , and the anonymous referee for providing comments and suggestions that improved the manuscript .
we are grateful to the staffs at the wht and the keck observatory for their support .
the wht is operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
some of the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration ( nasa ) ; the observatory was made possible by the generous financial support of the w. m. keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community ; we are most fortunate to have the opportunity to conduct observations from this mountain .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation ( nsf ) , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/ .
supernova research by a.v.f.s group at u.c .
berkeley is supported by gary and cynthia bengier , the richard and rhoda goldman fund , the christopher r. redlich fund , the tabasgo foundation , and nsf grants ast-0908886 and ast-1211916 .
work by a.g .- y . and
his group is supported by grants from the isf , bsf , gif , and minerva , an fp7/erc grant , and the helen and martin kimmel award for innovative investigation .
m.s . acknowledges support from the royal society . | ptf11kx was a type ia supernova ( sn ia ) that showed time - variable absorption features , including saturated h&k lines that weakened and eventually went into emission
. the strength of the emission component of h@xmath0 gradually increased , implying that the sn was undergoing significant interaction with its circumstellar medium ( csm ) .
these features , and many others , were blueshifted slightly and showed a p - cygni profile , likely indicating that the csm was directly related to , and probably previously ejected by , the progenitor system itself .
these and other observations led dilday et al .
( 2012 ) to conclude that ptf11kx came from a symbiotic nova progenitor like rs oph . in this work
we extend the spectral coverage of ptf11kx to 124680 rest - frame days past maximum brightness .
the late - time spectra of ptf11kx are dominated by h@xmath0 emission ( with widths of full width at half - maximum intensity @xmath1 km s@xmath2 ) , strong emission features ( @xmath3 10,000 km s@xmath2 wide ) , and a blue `` quasi - continuum '' due to many overlapping narrow lines of .
emission from oxygen , , and balmer lines higher than h@xmath0 is weak or completely absent at all epochs , leading to large observed h@xmath0/h@xmath4 intensity ratios .
the h@xmath0 emission appears to increase in strength with time for @xmath3 1 yr , but it subsequently decreases significantly along with the emission .
our latest spectrum also indicates the possibility of newly formed dust in the system as evidenced by a slight decrease in the red wing of h@xmath0 . during the same epochs , multiple narrow emission features from the csm temporally
vary in strength .
the weakening of the h@xmath0 and emission at late times is possible evidence that the sn ejecta have overtaken the majority of the csm and agrees with models of other strongly interacting sne ia . the varying narrow emission features , on the other hand , may indicate that the csm is clumpy or consists of multiple thin shells . |
the last decade has seen a renewed interest in the problem of solving an underdetermined system of equations @xmath4 , @xmath5 , where @xmath6 , by regularizing its solution to be sparse , i.e. , having very few non - zero entries .
specifically , if one aims to find @xmath7 with the least number of nonzero entries that solves the linear system , the problem is known as @xmath0-minimization:@xmath8 the problem @xmath9 is intended to seek _ entry - wise sparsity _ in @xmath7 and is known to be np - hard in general . in compressive sensing ( cs ) literature , it has been shown that the solution to often can be obtained by solving a more tractable linear program , namely , @xmath1-minimization @xcite:@xmath10 this unconventional equivalence relation between @xmath9 and @xmath11 and the more recent numerical solutions @xcite to efficiently recover high - dimensional sparse signal have been a very competitive research area in cs .
its broad applications have included sparse error correction @xcite , compressive imaging @xcite , image denoising and restoration @xcite , and face recognition @xcite , to name a few .
in addition to enforcing entry - wise sparsity in a linear system of equations , the notion of _ group sparsity _ has attracted increasing attention in recent years @xcite . in this case
, one assumes that the matrix @xmath12 has some underlying structure , and can be grouped into blocks : @xmath13 , where @xmath14 and @xmath15 .
accordingly , the vector @xmath7 is split into several blocks as @xmath16 , where @xmath17 . in this case , it is of interest to estimate @xmath7 with the least number of blocks containing non - zero entries .
the group sparsity minimization problem is posed as @xmath18 where @xmath19 is the indicator function .
since the expression @xmath20 can be written as @xmath21 , it is also denoted as @xmath22 , the @xmath23-norm of @xmath7 .
enforcing group sparsity exploits the problem s underlying structure and can improve the solution s interpretability .
for example , in a sparsity - based classification ( sbc ) framework applied to face recognition , the columns of @xmath12 are vectorized training images of human faces that can be naturally grouped into blocks corresponding to different subject classes , @xmath24 is a vectorized query image , and the entries in @xmath7 represent the coefficients of linear combination of all the training images for reconstructing @xmath24 .
group sparsity lends itself naturally to this problem since it is desirable to use images of the smallest number of subject classes to reconstruct and subsequently classify a query image .
furthermore , the problem of robust face recognition has considered an interesting modification known as the _ cross - and - bouquet _ ( cab ) model : @xmath25 , where @xmath26 represents possible sparse error corruption on the observation @xmath24 @xcite .
it can be argued that the cab model can be solved as a group sparsity problem in , where the coefficients of @xmath27 would be the @xmath28 group .
however , this problem has a trivial solution for @xmath29 and @xmath30 , which would have the smallest possible group sparsity .
hence , it is necessary to further regularize the entry - wise sparsity in @xmath27 . to this effect ,
one considers a mixture of the previous two cases , where one aims to enforce entry - wise sparsity as well as group sparsity such that @xmath7 has very few number of non - zero blocks _ and _ the reconstruction error @xmath27 is also sparse .
the _ mixed sparsity _ minimization problem can be posed as @xmath31 where @xmath32 controls the tradeoff between the entry - wise sparsity and group sparsity . due to the use of the counting norm ,
the optimization problems in and are also np - hard in general .
hence , several recent works have focused on developing tractable convex relaxations for these problems . in the case of group sparsity ,
the relaxation involves replacing the @xmath23-norm with the @xmath33-norm , where @xmath34 .
these relaxations are also used for the mixed sparsity case @xcite . in this work ,
we are interested in deriving and analyzing convex relaxations for general sparsity minimization problems . in the entry - wise case ,
the main theoretical understanding of the link between the original np - hard problem in and its convex relaxation has been given by the simple fact that the @xmath1-norm is a convex surrogate of the @xmath0-norm .
however , in the group sparsity case , a similar relaxation produces a family of convex surrogates , i.e. , @xmath35 , whose value depends on @xmath36 .
this raises the question whether there is a preferable value of @xmath36 for the relaxation of the group sparsity minimization problem ?
in fact , we consider the following more important question : _ is there a unified framework for deriving convex relaxations of general sparsity recovery problems ?
_ we present a new optimization - theoretic framework based on lagrangian duality for deriving convex relaxations of sparsity minimization problems .
specifically , we introduce a new class of equivalent optimization problems for @xmath9 , @xmath37 and @xmath38 , and derive the lagrangian duals of the original np - hard problems .
we then consider the lagrangian dual of the lagrangian dual to get a new optimization problem that we term as the _ lagrangian bidual _ of the primal problem .
we show that the lagrangian biduals are convex relaxations of the original sparsity minimization problems .
importantly , we show that the lagrangian biduals for the @xmath9 and @xmath37 problems correspond to minimizing the @xmath1-norm and the @xmath3-norm , respectively . since the lagrangian duals for @xmath9 , @xmath37 and @xmath38 are linear programs
, there is no duality gap between the lagrangian duals and the corresponding lagrangian biduals .
therefore , the bidual based convex relaxations can be interpreted as maximizing the lagrangian duals of the original sparsity minimization problems .
this provides new interpretations for the relaxations of sparsity minimization problems .
moreover , since the lagrangian dual of a minimization problem provides a lower bound for the optimal value of the primal problem , we show that the optimal objective value of the convex relaxation provides a non - trivial lower bound on the sparsity of the true solution to the primal problem .
in what follows , we will derive the lagrangian bidual for the mixed sparsity minimization problem , which generalizes the entry - wise sparsity and group sparsity cases ( also see section [ sec : results ] ) .
specifically , we will derive the lagrangian bidual for the following optimization problem : @xmath39 , ~~ { \mbox{s.t.}}~~ \begin{bmatrix } a_1 & \cdots & a_k \end{bmatrix } \begin{bmatrix } { { \boldsymbol}{x}}_1 \\ \vdots \\ { { \boldsymbol}{x}}_k \end{bmatrix } = { { \boldsymbol}{b } } , \end{split } \label{eq : primal0}\ ] ] where @xmath40 and @xmath41 .
given any unique , finite solution @xmath42 to , there exists a constant @xmath43 such that the absolute values of the entries of @xmath42 are less than @xmath44 , namely , @xmath45 .
note that if does not have a unique solution , it might not be possible to choose a finite - valued @xmath44 that upper bounds all the solutions . in this case
, a finite - valued @xmath44 may be viewed as a regularization term for the desired solution . to this effect , we consider the following modified version of where we introduce the box constraint that @xmath46 : @xmath47 , ~~ { \mbox{s.t.}}~~ a{{\boldsymbol}{x}}= { { \boldsymbol}{b}}\text { and } \|{{\boldsymbol}{x}}\|_\infty \le m , \end{split } \label{eq : primal1}\ ] ] where @xmath44 is chosen as described above to ensure that the optimal values of and are the same .
* primal problem .
* we will now frame an equivalent optimization problem for , for which we introduce some new notation .
let @xmath48 be an entry - based sparsity indicator for @xmath7 , namely , @xmath49 if @xmath50 and @xmath51 otherwise .
we also introduce a group - based sparsity indicator vector @xmath52 , whose @xmath53 entry @xmath54 denotes whether the @xmath53 block @xmath55 contains non - zero entries or not , namely , @xmath56 if @xmath57 and @xmath58 otherwise . to express this constraint ,
we introduce a matrix @xmath59 , such that @xmath60 if the @xmath61 entry of @xmath7 belongs to the @xmath62 block and @xmath63 otherwise . finally , we denote the positive component and negative component of @xmath7 as @xmath64 and @xmath65 , respectively , such that @xmath66 . given these definitions
, we see that can be reformulated as @xmath67 , ~ { \mbox{s.t.}}\text { ( a ) } { { \boldsymbol}{x}}_+ \ge 0 , \text { ( b ) } { { \boldsymbol}{x}}_- \ge 0 , \text { ( c ) } { { \boldsymbol}{g}}\in \{0,1\}^k , \!\!\\ &
\text{(d ) } { { \boldsymbol}{z}}\in \{0,1\}^n \text { ( e ) } a({{\boldsymbol}{x}}_+ - { { \boldsymbol}{x}}_- ) = { { \boldsymbol}{b } } , \text { ( f ) } \pi{{\boldsymbol}{g}}\ge \frac{1}{m}({{\boldsymbol}{x}}_+ + { { \boldsymbol}{x}}_- ) , \text { and ( g ) } { { \boldsymbol}{z}}\ge \frac{1}{m}({{\boldsymbol}{x}}_+ + { { \boldsymbol}{x}}_- ) , \end{split}\ ] ] where @xmath68 and @xmath69^\top \in { { \mathbb{r}}}^n$ ] .
constraints ( a)(d ) are used to enforce the aforementioned conditions on the values of the solution . while constraint ( e ) enforces the condition that the original system of linear equations is satisfied , the constraints ( f ) and ( g )
ensure that the group sparsity indicator @xmath70 and the entry - wise sparsity indicator @xmath71 are consistent with the entries of @xmath7 .
* lagrangial dual .
* the lagrangian function for is given as @xmath72 where @xmath73 , @xmath74 , @xmath75 , and @xmath76 . in order to obtain the lagrangian dual function ,
we need to minimize @xmath77 with respect to @xmath78 , @xmath79 , @xmath70 and @xmath71@xcite .
notice that if the coefficients of @xmath78 and @xmath79 , i.e. , @xmath80 and @xmath81 are non - zero , the minimization of @xmath77 with respect to @xmath78 and @xmath79 is unbounded below .
to this effect , the constraints that these coefficients are equal to @xmath82 form constraints on the dual variables .
next , consider the minimization of @xmath77 with respect to @xmath70 .
since each entry @xmath54 only takes values @xmath82 or @xmath83 , its optimal value @xmath84 that minimizes @xmath77 is given as @xmath85 a similar expression can be computed for the minimization with respect to @xmath71 . as a consequence , the lagrangian dual problem can be derived as @xmath86 , { \mbox{s.t.}}\\ & \text{(a ) } \forall i=1,2,4,5 : { { \boldsymbol}{\lambda}}_i \ge { { \boldsymbol}{0 } } , \text { ( b ) } \frac{1}{m}({{\boldsymbol}{\lambda}}_4+{{\boldsymbol}{\lambda}}_5)- a^\top{{\boldsymbol}{\lambda}}_3 -{{\boldsymbol}{\lambda}}_1 = { { \boldsymbol}{0}}\\ & \text{and ( c ) } \frac{1}{m}({{\boldsymbol}{\lambda}}_4+{{\boldsymbol}{\lambda}}_5)+ a^\top{{\boldsymbol}{\lambda}}_3 -{{\boldsymbol}{\lambda}}_2 = { { \boldsymbol}{0}}. \end{split}\ ] ] this can be further simplified by rewriting it as the following linear program : @xmath87 , { \mbox{s.t.}}~ \text{(a ) } { { \boldsymbol}{\lambda}}_4 \ge { { \boldsymbol}{0 } } , \text { ( b ) } { { \boldsymbol}{\lambda}}_5 \ge { { \boldsymbol}{0 } } , \text { ( c ) } { { \boldsymbol}{\lambda}}_6 \le 0 , \text { ( d ) } { { \boldsymbol}{\lambda}}_7 \le 0 , \!\!\!\!\!\!\!\!\!\!\!\!\\ & \!\!\!\!\text { ( e ) } { { \boldsymbol}{\lambda}}_6 \le { { \boldsymbol}{\alpha}}-\pi^\top{{\boldsymbol}{\lambda}}_4 , \text { ( f ) } { { \boldsymbol}{\lambda}}_7 \le { { \boldsymbol}{\beta}}-{{\boldsymbol}{\lambda}}_5 \text { and ( g ) } -\frac{1}{m}({{\boldsymbol}{\lambda}}_4+{{\boldsymbol}{\lambda}}_5 ) \le a^\top{{\boldsymbol}{\lambda}}_3 \le \frac{1}{m}({{\boldsymbol}{\lambda}}_4+{{\boldsymbol}{\lambda}}_5 ) . \end{split}\ ] ] notice that we have made two changes in going from to .
first , we have replaced constraints ( b ) and ( c ) in with the constraint ( g ) in and eliminated @xmath88 and @xmath89 from .
second , we have introduced variables @xmath90 and @xmath91 to encode the min " operator in the objective function of . * lagrangian bidual .
* we will now consider the lagrangian dual of , which will be referred to as the _ lagrangian bidual _ of .
it can be verified that the lagrangian dual of is given as @xmath92^k , \!\!\\ & \!\!\!\!\!\ !
\text{(d ) } { { \boldsymbol}{z}}\in [ 0,1]^n \text { ( e ) } a({{\boldsymbol}{x}}_+ - { { \boldsymbol}{x}}_- ) = { { \boldsymbol}{b } } , \text { ( f ) } \pi{{\boldsymbol}{g}}\ge \frac{1}{m}({{\boldsymbol}{x}}_+ + { { \boldsymbol}{x}}_- ) \text { and ( g ) } { { \boldsymbol}{z}}\ge \frac{1}{m}({{\boldsymbol}{x}}_+ + { { \boldsymbol}{x}}_- ) .
\end{split}\ ] ] notice that in going from to , the discrete valued variables @xmath71 and @xmath70 have been relaxed to take real values between @xmath82 and @xmath83 .
given that @xmath93 and noting that @xmath7 can be represented as @xmath94 , we can conclude from constraint ( g ) in that the solution @xmath42 satisfies @xmath45 .
moreover , given that @xmath70 and @xmath71 are relaxed to take real values , we see that the optimal values for @xmath95 and @xmath96 are @xmath97 and @xmath98 , respectively .
hence , we can eliminate constraints ( f ) and ( g ) by replacing @xmath71 and @xmath70 by these optimal values .
it can then be verified that solving is equivalent to solving the problem : @xmath99
\quad { \mbox{s.t.}}\text { ( a ) } a{{\boldsymbol}{x}}= { { \boldsymbol}{b}}\text { and ( b ) } \|{{\boldsymbol}{x}}\|_\infty \le m.\ ] ] this is the lagrangian bidual for .
in this section , we first describe some properties of the biduality framework in general . we will then focus on some important results for the special cases of entry - wise sparsity and group sparsity .
the optimal value of the lagrangian bidual in is a lower bound on the optimal value of the np - hard primal problem in .
[ thm : optval ] since there is no duality gap between a linear program and its lagrangian dual @xcite , the optimal values of the lagrangian dual in and the lagrangian bidual in are the same .
moreover , we know that the optimal value of a primal minimization problem is always bounded below by the optimal value of its lagrangian dual @xcite . we hence have the required result . since the original primal problem in is np - hard , we note that the duality gap between the primal and its dual in is non - zero in general .
moreover , we notice that as we increase @xmath44 ( i.e. , a more conservative estimate ) , the optimal value of the primal is unchanged , but the optimal value of the bidual in decreases .
hence , the duality gap increases as @xmath44 increases .
@xmath44 in should preferably be equal to @xmath100 , which may not be possible to estimate accurately in practice .
therefore , it is of interest to analyze the effect of taking a very conservative estimate of @xmath44 , i.e. , choosing a large value for @xmath44 . in what follows
, we show that taking a conservative estimate of @xmath44 is equivalent to dropping the box constraint in the bidual . for this purpose ,
consider the following modification of the bidual : @xmath101 \quad { \mbox{s.t.}}\quad a{{\boldsymbol}{x}}= { { \boldsymbol}{b}},\ ] ] where we have essentially dropped the box constraint ( b ) in .
it is easy to verify that @xmath102 , we have that @xmath103
. therefore , we see that taking a conservative value of @xmath44 is equivalent to solving the modified bidual in . notice that by substituting @xmath104 and @xmath105 , the optimization problem in reduces to the entry - wise sparsity minimization problem in .
hence , the lagrangian bidual to the @xmath44-regularized entry - wise sparsity problem @xmath9 is : @xmath106 more importantly , we can also conclude from that solving the lagrangian bidual to the entry - wise sparsity problem with a conservative estimate of @xmath44 is equivalent to solving the problem : @xmath107 which is precisely the well - known @xmath1-norm relaxation for @xmath9 .
our framework therefore provides a new interpretation for this relaxation : the @xmath1-norm minimization problem in is the lagrangian bidual of the @xmath0-norm minimization problem in , and solving is equivalent to maximizing the dual of .
we further note that we can now use the solution of to derive a non - trivial lower bound for the primal objective function which is precisely the sparsity of the desired solution .
more specifically , we can use theorem [ thm : optval ] to conclude the following result : [ corr : l0l1 ] let @xmath108 be the solution to .
we have that @xmath109 , the sparsity of @xmath108 , i.e. , @xmath110 is bounded below by @xmath111 . due to the non - zero duality gap in the primal entry - wise sparsity minimization problem ,
the above lower bound provided by corollary [ corr : l0l1 ] is not tight in general .
notice that by substituting @xmath112 and @xmath113 , the optimization problem in reduces to the group sparsity minimization problem in .
hence , the lagrangian bidual of the group sparsity problem is : @xmath114 as in the case of entry - wise sparsity above , solving the bidual to the group sparsity problem with a conservative estimate of @xmath44 is equivalent to solving : @xmath115 which is the convex @xmath3-norm relaxation of the @xmath23-min problem . in other words ,
the biduality framework selects the @xmath3-norm out of the entire family of @xmath33-norms as the convex surrogate of the @xmath23-norm .
finally , we use theorem [ thm : optval ] to show that the solution obtained by minimizing the @xmath3-norm provides a lower bound for the group sparsity .
[ corr : grp ] let @xmath116 be the solution to . for any @xmath117
, the group sparsity of @xmath116 , i.e. , @xmath118 , is bounded below by @xmath119 .
the @xmath3-norm seems to be an interesting choice for computing the lower bound of the group sparsity , as compared to other @xmath33-norms for finite @xmath120 . for example , consider the case when @xmath121 , where the @xmath33-norm is equivalent to the @xmath1-norm .
assume that @xmath12 consists of a single block with several columns so that the maximum number of non - zero blocks is @xmath83 .
denote the solution to the @xmath1-minimization problem as @xmath122 .
it is possible to construct examples ( also see figure [ fig : entry ] ) where @xmath123 .
hence , it is unclear in general if the solutions obtained by minimizing @xmath33-norms for finite - valued @xmath120 can help provide lower bounds for the group sparsity .
we now present experiments to evaluate the bidual framework for minimizing entry - wise sparsity and mixed sparsity .
we present experiments on synthetic data to show that our framework can be used to compute non - trivial lower bounds for the entry - wise sparsity minimization problem .
we then consider the face recognition problem where we compare the performance of the bidual - based @xmath3-norm relaxation with that of the @xmath124-norm relaxation for mixed sparsity minimization .
we use boxplots to provide a concise representation of our results statistics .
the top and bottom edge of a boxplot for a set of values indicates the maximum and minimum of the values .
the bottom and top extents of the box indicate the @xmath125 and @xmath126 percentile mark .
the red mark in the box indicates the median and the red crosses outside the boxes indicate potential outliers .
* entry - wise sparsity .
* we now explore the practical implications of corollary [ corr : l0l1 ] through synthetic experiments .
we randomly generate entries of @xmath127 and @xmath128 from a gaussian distribution with unit variance .
the sparsity of @xmath129 is varied from @xmath83 to @xmath130 in steps of @xmath131 .
we solve with @xmath132 using @xmath133 and @xmath134 , where @xmath135 .
we use corollary [ corr : l0l1 ] to compute lower bounds on the true sparsity , i.e. , @xmath136 .
we repeat this experiment @xmath137 times for each sparsity level and figure [ fig : entry ] shows the boxplots for the bounds computed from these experiments .
we first analyze the lower bounds computed when @xmath138 , in figure [ fig : m0 ] . as explained in section [ sec : entry - results ] , the bounds are not expected to be tight due to the duality gap .
notice that for extremely sparse solutions , the maximum of the computed bounds is close to the true sparsity but this diverges as the sparsity of @xmath129 reduces .
the median value of the bounds is much looser and we see that the median also diverges as the sparsity of @xmath129 reduces .
furthermore , the computed lower bounds seem to grow linearly as a function of the true sparsity .
similar trends are observed for @xmath139 and @xmath134 in figures [ fig:2m0 ] and [ fig:5m0 ] , respectively . as expected from the discussion in section [ sec : entry - results ] , the bounds become very loose as @xmath44 increases . in theory
, we would like to have _ per - instance certificates - of - optimality _ of the computed solution , where the lower bound is equal to the true sparsity @xmath136 .
nonetheless , we note that this ability to compute a per - instance non - trivial lower bound on the sparsity of the desired solution is an important step forward with respect to the previous approaches that require pre - computing optimality conditions for equivalence of solutions to the @xmath0-norm and @xmath1-norm minimization problems .
we have performed a similar experiment for the group sparsity case , and observed that the bidual framework is able to provide non - trivial lower bounds for the group sparsity also .
+ * mixed sparsity .
* we now evaluate the results of mixed sparsity minimization for the sparsity - based face recognition problem , where the columns of @xmath12 represent training images from the @xmath140 face classes : @xmath141 and @xmath142 represents a query image .
we assume that a subset of pixel values in the query image may be corrupted or disguised .
hence , the error in the image space is modeled by a sparse error term @xmath27 : @xmath143 , where @xmath144 is the uncorrupted image . a linear representation of the query image forms the following linear system of equations : @xmath145 where @xmath146 is the @xmath147 identity matrix .
the goal of sparsity - based classification ( sbc ) is to minimize the group sparsity in @xmath7 and the sparsity of @xmath27 such that the dominant non - zero coefficients in @xmath7 reveal the membership of the ground - truth observation @xmath148 @xcite . in our experiments ,
we solve for @xmath7 and @xmath27 by solving the following optimization problem : @xmath149 notice that for @xmath150 , this reduces to solving a special case of the problem in , i.e. , the bidual relaxation of the mixed sparsity problem with a conservative estimate of @xmath44 . in our experiments , we set @xmath151 and compare the solutions to obtained using @xmath152 and @xmath150 .
we evaluate the algorithms on a subset of the ar dataset @xcite which has manually aligned frontal face images of size @xmath153 for 50 male and 50 female subjects , i.e. , @xmath154 and @xmath155 .
each individual contributes 7 un - occluded training images , 7 un - occluded testing images and 12 occluded testing images .
hence , we have 700 training images and 1900 testing images . to compute the number of non - zero blocks in the coefficient @xmath7 estimated for a testing image
, we find the number of blocks whose energy @xmath156 is greater than a specified threshold .
the results of our experiments are presented in figure [ fig : mixed ] .
the solution obtained with @xmath152 gives better group sparsity of @xmath7 .
however , a sparser error @xmath27 is estimated with @xmath150 .
the number of non - zero entities in a solution to , i.e. , the number of non - zero blocks plus the number of non - zero error entries , is lower for the solution obtained using @xmath157 rather than that obtained using @xmath152 .
however , the primal mixed - sparsity objective value @xmath158 ( see ) is lower for the solution obtained using @xmath152 .
+ + we now compare the classification results obtained with the solutions @xmath7 computed in our experiments . for classification
, we consider the non - zero blocks in @xmath7 and then assign the query image to the block , i.e. , subject class , for which it gives the least @xmath159 residual @xmath160 .
the results are presented in table [ tab : mixed ] .
notice that the classification results obtained with @xmath150 ( the bidual relaxation ) are better than those obtained using @xmath152 .
since the classification of un - occluded images is already very good using @xmath152 , classification with @xmath150 gives only a minor improvement in this case .
however , a more tangible improvement is noticed in the classification of the occluded images .
therefore the classification with @xmath150 is in general better than that obtained with @xmath152 , which is considered the state - of - the - art for sparsity - based classification @xcite .
.classification results on the ar dataset using the solutions obtained by minimizing mixed sparsity .
the test set consists of 700 un - occluded images and 1200 occluded images . [ cols="^,^,^,^,^ " , ]
we have presented a novel analysis of several sparsity minimization problems which allows us to interpret several convex relaxations of the original np - hard primal problems as being equivalent to maximizing their lagrangian duals .
the pivotal point of this analysis is the formulation of mixed - integer programs which are equivalent to the original primal problems . while we have derived the biduals for only a few sparsity minimization problems , the same techniques
can also be used to easily derive convex relaxations for other sparsity minimization problems @xcite .
an interesting result of our biduality framework is the ability to compute a per - instance certificate of optimality by providing a lower bound for the primal objective function .
this is in contrast to most previous research which aims to characterize either the subset of solutions or the set of conditions for perfect sparsity recovery using the convex relaxations @xcite . in most cases ,
the conditions are either weak or hard to verify .
more importantly , these conditions needed to be pre - computed as opposed to verifying the correctness of a solution at run - time . in lieu of this , we hope that our proposed framework will prove an important step towards per - instance verification of the solutions .
specifically , it is of interest in the future to explore tighter relaxations for the verification of the solutions .
this research was supported in part by aro muri w911nf-06 - 1 - 0076 , arl mast - cta w911nf-08 - 2 - 0004 , nsf cns-0931805 , nsf cns-0941463 and nsf grant 0834470 . the views and conclusions contained in this document
are those of the authors and should not be interpreted as representing the official policies , either expressed or implied , of the army research laboratory or the u.s .
government is authorized to reproduce and distribute for government purposes notwithstanding any copyright notation herein . | recent results in compressive sensing have shown that , under certain conditions , the solution to an underdetermined system of linear equations with sparsity - based regularization can be accurately recovered by solving convex relaxations of the original problem . in this work , we present a novel primal - dual analysis on a class of sparsity minimization problems .
we show that the lagrangian bidual ( i.e. , the lagrangian dual of the lagrangian dual ) of the sparsity minimization problems can be used to derive interesting convex relaxations : the bidual of the @xmath0-minimization problem is the @xmath1-minimization problem ; and the bidual of the @xmath2-minimization problem for enforcing group sparsity on structured data is the @xmath3-minimization problem .
the analysis provides a means to compute per - instance non - trivial lower bounds on the ( group ) sparsity of the desired solutions . in a real - world application ,
the bidual relaxation improves the performance of a sparsity - based classification framework applied to robust face recognition . |
the interaction of a quantum system with an environment is generally seen as source of decoherence , leading to the loss of quantum properties and the appearance of the classical world @xcite .
in fact , due to the creation of correlation with the environment , the evolution of the quantum system is not unitary any more .
if the system is prepared in an entangled state , such entanglement is very likely to be lost @xcite .
however , environments can also create correlations among quantum systems .
for example , the interaction of a bipartite system with a common reservoir creates correlations between the parts @xcite , irrespective of the existence of a direct coupling between them @xcite . in a shared environment
highly entangled long - living states , or even sub - radiant decoherence free states can exist @xcite .
many studies have demonstrated the existence of non - zero stationary entanglement in systems of two qubits or two harmonic oscillators sharing the same reservoir @xcite .
the ability to create correlations persists if the common reservoir has non zero temperature @xcite , and is increased by memory effects in non - markovian reservoirs @xcite . in the last decade
many efforts have been done to understand the dynamics of entanglement and quantum correlations in systems which are good candidates for applications in quantum information theory and technology . however
, not all the protocols in quantum information theory rely on entanglement . on the contrary , in many cases entanglement
is simply not enough and non - local properties , expressed by the violation of a bell inequality , are required . here
i study the non - local properties , namely the violation of the chsh inequality , of a system of two qubits interacting with a common non - markovian reservoir .
i consider a factorized state of the qubits with one excitation , and study , as a function of time and of the qubit - cavity coupling , the sudden violation of chsh inequality .
the conditions for which the violation is maximum are identified . afterwards i add another dynamical ingredient and study how the dynamics of the chsh violation gets modified when the spontaneous emission of the qubits is included .
the system under investigation comprises two qubits interacting with a common zero - temperature bosonic reservoir .
the hamiltonian of the system in rotating - wave approximation , and in units of @xmath0 , is @xmath1 , @xmath2 @xmath3 where @xmath4 and @xmath5 are the pauli raising and lowering operators for qubit 1 and 2 respectively , @xmath6 is the bohr frequency of the two identical qubits , @xmath7 and @xmath8 are dimensionless environment - qubit coupling constants , @xmath9 and @xmath10 , @xmath11 and @xmath12 are the annihilation and creation operators , the frequency and the coupling constant of the field mode @xmath13 , respectively . in the following
i assume that the two qubits interact resonantly with a lossy cavity , so i choose a lorentzian spectral distribution to describe the properties of the environment , @xmath14 where @xmath15 is the width of the spectral distribution describing the cavity losses , and @xmath16 in the limit of ideal cavity ( when @xmath17 ) is proportional to the vacuum rabi frequency @xmath18 through @xmath19 with @xmath20 the collective coupling constant .
the dynamics of two qubits interacting with a common lorentzian - structured reservoir has been studied in ref .
@xcite for the case of one excitation , and in ref .
@xcite for a generic state of two identically coupled qubits .
since here i consider the dynamics of a factorized state with one excitation of the form @xmath21 i am going to use the model of ref .
there the dynamics of the qubits is expressed in the basis of super - radiant and sub - radiant states @xmath22 @xmath23 where the relative coupling strengths @xmath24 and @xmath25 have been introduced ( @xmath26 ) .
considering the dynamics in such a basis is particularly convenient since the sub - radiante state does not evolve in time .
thus the evolution of the amplitudes of the first and second qubits is just @xmath27 @xmath28 with @xmath29 and @xmath30,\ ] ] where @xmath31 . in order to evaluate the time evolution of the violation of the chsh bell inequality i use the expression of ref .
@xcite , which allows one to express the maximum of the bell function ( by an appropriate choice of angles ) as function of the two - qubit density matrix elements . in the @xmath32 basis of the qubits , the maximum of the bell function reads as follows @xmath33 with @xmath34 where @xmath35 , @xmath36 , @xmath37 and @xmath38 . for the particular initial state
considered i have @xmath39 and @xmath40 .
here i investigate the chsh inequality violation for two qubits prepared initially in the state @xmath41 i consider a cavity characterized by a high but experimentally feasible quality factor ( the width is @xmath42 , with @xmath43 the markovian decay rate of the qubits ) , and strong coupling conditions .
i set the parameter @xmath44 , indicating the strength of the coupling @xcite . in figure
[ fig : chsh ] i plot the chsh violation as a function of scaled time ( @xmath45 ) and of the relative coupling strength @xmath46 . during the evolution the system passes through highly entangled states , strongly violating the chsh inequality .
cycles of birth and death of non - locality follow one after the other , till for long times the inequality is finally satisfied .
the sudden appearance of the violation is consequence of the shared reservoir , providing an indirect coupling between the qubits .
such a reservoir - mediated interaction plays certainly a role also in the revivals of non - locality , which are also caused by the memory effects of the non - markovian reservoir .
i notice that sudden violation of the chsh inequality appears up to a certain value of the relative strength of the coupling .
in particular for the chosen coupling conditions no violation occurs when @xmath47 . for stronger coupling conditions the region of violation in the @xmath46 space of parameters widens , nevertheless it never reaches @xmath48 .
in fact for @xmath49 the qubits are symmetrically coupled to the cavity , and entanglement and non - local properties in general can not be created out of the factorized state in eq . .
( color online ) dynamics of the chsh inequality violation in a common lorentzian structured reservoir as a function of scaled time and relative coupling strength for two atoms prepared in a factorized state @xmath50 .
no spontaneous emission by the qubits is included.,width=453 ]
in this section i study how the dynamical violation of the chsh inequality is modified when including spontaneous emission for the two qubits . in ref .
@xcite i have studied the entanglement time evolution of two entangled qubits interacting with the same lorentzian structured reservoir ( leaky cavity ) and emitting independently outside the cavity due to spontaneous emission . by means of eqs .
( 1 ) and ( 2 ) of ref .
@xcite i have solved the dynamics in the case of qubits having the same transition frequency , equally and resonantly coupled with the cavity . here
i generalize that master equation to include the case of different couplings between the qubits and the cavity .
so the master equation describing the dynamics of our system ( with some renaming of parameters ) becomes : @xmath51 -\lambda(a^{\dag}a\tilde{\rho}+\tilde{\rho}a^{\dag}a-2a\tilde{\rho}a^{\dag})\nonumber\\ -\frac{{\gamma}_{1}}{2}({\sigma}_{+}^{(1)}{\sigma}_{-}^{(1)}\tilde{\rho}+\tilde{\rho}{\sigma}_{+}^{(1)}{\sigma}_{-}^{(1)}-2{\sigma}_{-}^{(1)}\tilde{\rho}{\sigma}_{+}^{(1)})\nonumber\\ -\frac{{\gamma}_{2}}{2}({\sigma}_{+}^{(2)}{\sigma}_{-}^{(2)}\tilde{\rho}+\tilde{\rho}{\sigma}_{+}^{(2)}{\sigma}_{-}^{(2)}-2{\sigma}_{-}^{(2)}\tilde{\rho}{\sigma}_{+}^{(2)}),\end{aligned}\ ] ] with @xmath52.\ ] ] where @xmath4 and @xmath5 , @xmath53 , @xmath46 and @xmath54 , @xmath15 and @xmath16 have been defined above , while @xmath55 and @xmath56 are the annihilation and creation operators for the cavity mode and @xmath57 are the spontaneous emission rates for the two qubits . for the sake of simplicity
i assume @xmath58 .
as in the previous section i consider the sudden appearance of chsh violation for two qubits prepared in the state @xmath59 .
i study the dynamics for the same strong coupling conditions @xmath44 , in particular for the same width of the spectral distribution , and vacuum rabi frequency .
the two - qubit spontaneous emission decay rates are set equal to @xmath60 , therefore they are 50 times smaller than the vacuum rabi frequency @xmath18 . in figure
[ fig : chshspont ] i show the chsh violation as a function of time and of the relative coupling parameter @xmath46 .
the effect of spontaneous emission is apparent : the long series of sharp revivals is dramatically damped out . for the chosen spontaneous emission parameter
only one small revival is present .
the amount of violation of the chsh inequality reduces with increasing decay rate , and beyond a certain threshold value the chsh inequality is always satisfied . for the coupling conditions chosen such a threshold parameter
is @xmath61 .
( color online ) dynamics of the chsh inequality violation in a common lorentzian structured reservoir as a function of scaled time and relative coupling strength for two atoms prepared in a factorized state @xmath50 . here
spontaneous emission is included and the spontaneous emission decay rate of the qubits are equal to @xmath62 , i.e. , 50 times smaller than the vacuum rabi frequency @xmath63.,width=453 ]
i have investigated the time evolution of the violation of the chsh inequality for a system of two qubits interacting with a common lorentzian structured reservoir .
the reservoir - induced correlations between the qubits drive the evolution of an initial factorized state as @xmath59 towards highly entangled states , strongly violating the chsh inequality .
after cycles of birth and death of non - locality , progressively damped out , the chsh inequality becomes always satisfied . as a second step
i consider a more realistic system in which spontaneous emission of the two qubits is taken into account .
the time evolution of the chsh violation well describes the dramatic changes that the introduction of spontaneous emission in this model brings to the dynamics of the system .
birth of non - local properties appears only if the spontaneous emission decay rate is below a certain threshold , even in that case oscillations are strongly damped and the inequality becomes permanently satisfied much earlier .
the dynamics of the chsh violation was studied for a system of two non interacting initially entangled qubits in two independent non - markovian reservoirs @xcite . there ,
death and revivals of the violation have been observed due to the memory effects of the reservoirs . in that case
however , correlations between the qubits can not be created starting from a factorized state , as a consequence an initially separable state can not evolve into an entangled one , and possibly violating the chsh inequality .
i would like to thank s. maniscalco and j. piilo for enlightening discussions , g. compagno and his group in palermo for the kind hospitality and useful suggestions .
finally i thank m. ehrnrooth foundation for financial support . | i study the dynamics of the violation of the chsh inequality for two qubits interacting with a common zero - temperature non - markovian environment .
i demonstrate sudden violation of the inequality for two qubits initially prepared in a factorized state . due to the strong coupling between the qubits and the reservoir ,
the dynamics is characterized by numerous sharp revivals .
furthermore i focus on a more realistic physical system in which the spontaneous emission for the qubits is taken into account . when including spontaneous emission even for small decay parameters , revivals in the violation are heavily damped out .
if the decay rates exceed a certain threshold , the inequality turns out to be always satisfied . |
planetary systems form from the accretion disks encircling young stars .
the composition of ice , gas and dust within the disk sets the initial chemical conditions of the planets and also regulates the physical conditions under which planets form .
the properties of the pre - planetary materials can then be compared to the composition of the present day solar system ( e.g. * ? ? ? * ) and even that of extrasolar planetary atmospheres @xcite . nonetheless , there is a missing link between these two stages , separated in time by billions of years .
it is essential to observationally capture a forming young planet in situ to put together a complete chemical ( and physical ) history of planet formation .
the upcoming capabilities of the atacama large millimeter array ( alma ) will provide extremely high sensitivity and spatially resolved observations of disks ( up to 0.007@xmath4 at 650 ghz , or 0.7 au at distances of @xmath5 pc ) . at these scales ,
detailed disk structure will be readily revealed , and observations of the local environment near forming protoplanets should be accessible .
young proto - jupiters are expected to be intrinsically hot as they accrete matter through their circumplanetary disk and liberate gravitational potential energy , thereby generating substantial accretion luminosity , @xmath6 .
theoretical models of early - stage circumplanetary disks find typical `` quiescent '' accretion levels between @xmath7 m@xmath8 year@xmath9 and @xmath10 m@xmath8 year@xmath9 @xcite .
periodically , the circumplanetary disk is theorized to undergo accretion outbursts similar to those seen in fu ori objects @xcite , where the planet s accretion rate jumps to @xmath11 m@xmath8 year@xmath9 .
accretion rates can be translated into accretion luminosities , @xmath12 @xcite , where @xmath13 and @xmath14 are the planet s mass and accretion rate , and @xmath15 is the inner radius of the circumplanetary disk , which is something like the radius of the planet . assuming a jupiter - mass planet and @xmath16 r@xmath17 , the quiescent accretion levels translate to @xmath18 l@xmath8 ( see also * ? ? ?
during an accretion burst , however , the planet can outshine even the star in bolometric luminosity , where the accretion luminosity due to the planet can be as high as @xmath19 l@xmath8 but for less than 0.1% of the planet s formation time @xcite .
@xcite reported the detection of a possible embedded protoplanet companion in lkca 15 for which they estimated a substantial accretion luminosity of @xmath6@xmath20 l@xmath8 .
planets with similar accretion luminosities will heat the nearby circumstellar dust disk and may give rise to detectable signatures with sensitive submillimeter continuum observations , though detecting the circumplanetary disk itself will be challenging @xcite .
in addition to the thermal effects on the dust , the planetary accretion heating will have a substantial impact on the chemical structure in the vicinity of the young planet .
in the present work , we explore local heating due to a single massive protoplanet ( a gas giant precursor ) on the three - dimensional chemistry of the surrounding circumstellar disk .
we do not calculate the chemistry of the hot , young circumplanetary disk , but focus instead on the larger scale effects on the cold , molecular disk within which the planet is entrenched .
we vary the planet s orbital location from the star for a fixed accretion rate of @xmath21 m@xmath8 year@xmath9 , where the physical and thermal structure are described in [ sec : phys ] . on this model , we calculate the time - dependent chemistry in 2d slices over the 3d model , focusing the calculations on the region near the planet and opposite the planet ( [ sec : chem ] ) . for species that are particularly sensitive to the presence of the planet and
have strong submillimeter transitions , we identify submillimeter emission line tracers and compute the emission based on the three - dimensional structure to determine detectability ( [ sec : emission ] ) . in
[ sec : considerations ] we relax some of the assumptions of our model and discuss potential caveats to our simple approach , and [ sec : conclusions ] summarizes our findings .
the background disk density is described by a simple power law in surface density , @xmath22 and vertically gaussian with a scale height of @xmath23 au , @xmath24 .
the gas and dust are uniformly mixed with a gas - to - dust mass ratio of 100 , and the disk gas mass contains @xmath25 m@xmath26 .
the dust is a simple mrn distribution in grain size ( @xmath27 ; * ? ? ?
* ) with minimum size of 0.005 @xmath28 m and maximum size of 1 @xmath28 m .
the dust composition is assumed to be draine and lee astrosilicates @xcite .
we fix the central star to have a mass of @xmath29 m@xmath26 , an effective temperature of @xmath30 k and @xmath31 r@xmath26 , characteristic of low mass t tauri stars .
we consider four planet locations , @xmath32 au , 10 au , 20 au , and 30 au .
for each location , we simply define a vertical gap in the disk corresponding to the hill radius of a @xmath33 m@xmath17 planet as consistent with theoretical models , which find massive planets open radial gaps comparable to the size of their hill sphere ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
the gap radii ( half the gap - width ) for the planet models at @xmath32 au , 10 au , 20 au , and 30 au are 0.4 au , 0.9 au , 1.7 au , and 2.6 au , respectively .
the dust density within the gap has a floor - value at the gap midplane of @xmath34 g @xmath35 and is optically thin to the planet s infrared radiation , which is described in section [ sec : thermal ] .
the density within the gap is a factor of six orders of magnitude lower than the disk ; however , we explore higher degrees of gap filling by dust in section [ sec : considerations ] .
we emphasize that these models are designed to understand and isolate the local effects of planetary heating , and that we do not include variations in the 3d density structure near the planet , either excess flaring near the edges of the gap ( e.g. , * ? ? ?
* ) , or accretion streams , both of which will alter the vertical and radial disk structure and should be examined in future work ( see also discussion in section [ sec : considerations ] ) .
the density structure of the bulk disk is shown in figure [ fig : fullmod]a .
disk properties away from the gap are similar for the @xmath32 au and @xmath36 au .
the central star is located at r=0 and z=0 and the plot is made for an azimuthal angle of @xmath37 ( opposite the location of planet).[fig : fullmod],scaledwidth=40.0% ] we calculate the uv and x - ray radiation field from the central star for each of the four planet locations , @xmath38 , from a single 2d slice and assume the high energy stellar radiation field is otherwise azimuthally symmetric . the procedure for calculating the 2d x - ray and fuv field are the same as described in @xcite , using the monte carlo radiative transfer code of @xcite .
the fuv radiation is used to estimate the gas temperatures ( s. bruderer , in private communication ) , which are coupled to the dust temperatures in the region where the planet s impact is the greatest , and as such , our results do not depend on the gas temperature in the upper layers where the gas and dust temperatures are decoupled .
the disk is quite opaque to the fuv stellar photons ( wavelengths between @xmath39 , including that of lyman-@xmath40 ) due to our use of an unsettled disk model .
a substantial amount of fuv continuum and line ( lyman-@xmath40 ) radiation is down - scattered into the gap , aided by the uv - exposed , optically thick outer gap edge , which tends to deflect photons both into the gap and away from the disk .
however , the scattered continuum and line photons within the gap do not penetrate far into the gap edges and do not have a substantial impact on the chemistry beyond the illuminated wall .
the integrated fuv field between @xmath41 is shown in figure [ fig : fullmod]b .
the x - ray radiation field is computed using the same transfer code as the uv with the combined gas and dust x - ray absorption cross sections of @xcite and thompson scattering .
because the x - ray photons are stopped at larger column densities much further into the disk , the presence of the gap does not have as significant of an effect as it does for the fuv ( figure [ fig : fullmod]c ) .
the gap exposes more of the disk to x - ray photons , but the overall x - ray field is much more sensitive to distribution of disk mass rather than the specific gap location ; these hill radius - sized gaps are relatively narrow compared to the path length of an x - ray photon at @xmath42 kev . for cosmic ray ionization , we use the model of @xcite for the cosmic ray ionization rate at solar maximum for the present day sun ( @xmath43 s@xmath9 per h@xmath44 ) as an upper limit to the contribution from cosmic rays , consistent with the upper limit determined from observations of the tw hya disk @xcite .
we calculate the thermal structure of the disk assuming two passive radiation sources : the self - luminous central star the protoplanet shining due to its accretion luminosity .
we assume a circumplanetary accretion rate of @xmath21 m@xmath8 year@xmath9 on an @xmath33 m@xmath17 and @xmath45 r@xmath17 planet , corresponding to an accretion luminosity of @xmath46 l@xmath8 . in this simple picture , the planet
is treated as a spherical , 1500 k blackbody that heats its self - carved gap from the inside .
the dust temperatures are calculated using the code torus @xcite assuming radiative equilibrium with the lucy method @xcite .
we note that we do not include the differential rotation of the circumstellar disk in the dust temperature calculation because radiative equilibrium at @xmath47 k is quickly attained within minutes to hours @xcite . for a planet at 10 au , this timespan is negligible compared to the @xmath48 year period of the planet s orbit , thus planetary heating of the disk is expected to remain a local phenomena and will not become highly sheared out by differential rotation .
the thermal structure for the bulk circumstellar disk ( away from the planet ) is shown in figure [ fig : fullmod]d for the planet at @xmath49 au .
the large - scale physical properties are largely similar for all four planet locations in our models ( i.e. , the presence of the gap does not significantly change the thermal / irradiation structure away from the gap ) .
all four models have an outer disk radius of 50 au and an inner disk radius of 0.2 au .
the additional heating by the planetary companion primarily results in an azimuthally extended ( @xmath50 ) but somewhat thin ( @xmath51 au ) swath of material along the gap edges , centered on the planet itself .
the temperature structures in the midplane and for a vertical slice centered on the planet s location are shown in figure [ fig : thermal08 ] .
the contours highlight the change in temperature due to the presence of the planet .
the planet increases the temperature by greater than 10% within about @xmath52 au , 1.6 au , 2.7 au , and 4 au from the edge of the gap for the 5 , 10 , 20 , and 30 au orbital radius , respectively .
the absolute change in temperature at the gap edges closest to the planet ( @xmath53 au from the wall ) due to the additional heating corresponds to an increase from 42 k to 60 k ( @xmath32 au ) , 35 k to 47 k ( @xmath36 au ) , 27 k to 34 k ( @xmath49 au ) , and 24 k to 31 k ( @xmath54 au ) for both gas and dust temperatures .
these substantial @xmath55 k changes in cold , dense molecular gas will substantially alter the local chemistry close to the planet . from the physical structure outlined above
, we can estimate molecular abundances as a function of 3d position throughout the disk .
the disk chemical code used for the calculations is presented in @xcite and further expanded in @xcite .
the chemical code itself is inherently 1 + 1d , and so to address this limitation we extract 2d azimuthal cuts from full 3d model to compute the abundances , and reconstruct the full disk profile from the individual 2d calculations .
we consider ten 2d slices at @xmath56 intervals in azimuth ranging from the planet s location to @xmath57 away from the planet , and assume that the planet is symmetric upon reflection .
for the rest of the disk beyond @xmath57 , we calculate the chemistry based upon one 2d slice on the opposite side of the disk from the planet , i.e. the `` anti - planet '' side .
differential rotation is not included in the present models but will be explored in future work ( see further discussion below ) .
the 1 + 1d @xcite code calculates the time - dependent abundances relative to hydrogen as a function of radius and height from the midplane based on a fixed grid of temperature , density , and radiation field conditions .
the chemical reaction network is based on the osu gas - phase chemical network @xcite , expanded to include grain - surface chemistry in the method of @xcite , where the grain - surface reactants `` sweep - out '' the surface at a rate related to the binding energy and mass of the reactant .
we treat the hydrogen binding energy in the method of @xcite , where we assume that the binding energy for desorption processes is the chemisorbed value such that h@xmath44 can form even at high temperatures , while the binding energy adopted for the rate of hydrogenation reactions on the surfaces of cold dust grains is assumed to be the lower , physisorbed value , or @xmath58 k , such that the h - atoms are highly mobile across the ice mantle .
the reaction network has a total of 6292 reactions , including both chemical reactions and physical processes ( i.e. , ionization , desorption , etc . ) and 697 species .
all chemical calculations are examined after 1 myr of chemical evolution .
in the following section , we present chemical abundance signatures due to the additional heating from a planetary companion as calculated over the disk .
we then simulate the emergent line emission for the fully 3d physical and chemical model .
based upon these emission models we generate alma simulations of planets embedded in disks .
the most important chemical effect from the planet is the thermal desorption of molecular species that are otherwise frozen out as ices in the midplane at the orbital distance of the planet , which we term `` primary tracers . ''
as discussed in section [ sec : thermal ] , the dust temperature rapidly equilibrates to the local irradiation conditions . to assess whether the simplification of neglecting differential rotation is important for the chemical structure
, we must compare the chemical timescales for adsorption / sublimation to the orbital time .
if the chemical timescales are longer than the orbital time , the chemical effects of the planetary heating within the gap will become sheared out over azimuth .
in contrast , short chemical timescales relative to the planet s orbital time indicate the relevant chemistry is able to adjust quickly , and thus `` follow '' the planet . in our model ,
the midplane at @xmath59 au has a density of approximately @xmath60 @xmath35 and a temperature of @xmath61 k near the planet .
the timescale for freeze - out of a molecule is related to the surface area of grains per unit volume , or @xmath62
@xmath63 @xmath35 , as well as the thermal speed of the molecule in the gas phase , @xmath64 , such that molecule @xmath65 freezes out in a characteristic time @xmath66 . at gas densities of @xmath67 @xmath35 ,
a typical grain surface area density is @xmath68 @xmath63 @xmath35 ( i.e. , 0.1 @xmath28m - sized grains at an abundance relative to h - atoms of @xmath69 ) , an h@xmath44co molecule ( for example ) with mass of 30 amu , will collide with a grain on average every @xmath70 years , which is roughly the time for circumstellar disk chemistry to reset when not directly heated by the planet .
the corresponding timescale for thermal evaporation of an h@xmath44co molecule ( assuming a desorption temperature of @xmath71 k ; * ? ? ? * ) on a 50 k dust grain based on the polyani - wigner relation ( see * ? ? ?
* ) is 0.02 years ( the timescale for molecules to evaporate when exposed to the planet s heating ) . both of these timescales are sufficiently rapid compared to the orbital time around a 1 m@xmath8 star for the planets considered here ( about 11 , 32 , and 89 years ) and therefore we expect the chemistry of the primary tracers to follow the planet and not experience strong azimuthal shear .
it is important to note that the relevant timescales will increase with height ; however , in all cases , the chemical effect of the planet is limited to a narrow vertical region close to the midplane where @xmath72 , over which the density only decreases by @xmath73
. molecules that form via gas - phase or grain - surface chemistry as a direct result of primary species desorption are `` second - order '' chemical effects
i.e. , secondary tracers .
the timescales for secondary species formed from primaries to return to the low - temperature state will depend on the particular formation pathways for the secondary species and the availability of he@xmath74 ions to break up newly formed molecules that would otherwise not be present without the planet ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this case
, the distribution of secondary species formed due to the planet are more likely to be affected by shear ( see discussion below ) . given the size of the chemical network , we took an unbiased approach to search for promising chemical tracers of planetary heating , both primary and secondary .
we calculate the vertical column density of every species in the network versus radius , and filter out species with low column density near the planet , @xmath75 @xmath76 , since these will be the most difficult to detect .
we then calculate the fractional difference between the column densities , @xmath77 , at the azimuthal location of the planet and the anti - planet side , and integrate this quantity over radius to get an estimate for the total fractional change due to the planet .
we then sort by the fractional change to identify which species are most affected by the additional heating .
the results of this `` blind search '' for gas - phase tracers are shown in figures [ fig : column_0508 ] ( @xmath32 au ) , [ fig : column_1008 ] ( @xmath36 au ) , [ fig : column_2008 ] ( @xmath49 au ) , and [ fig : column_3008 ] ( @xmath54 au ) . for certain species , the additional heating changes the vertical column density at the edges of the gap by many orders of magnitude . for most `` signpost '' molecules ,
the heating from the planet lifts the local dust temperature at the midplane from below the particular species desorption temperature to above , causing a large enhancement in the column density at the edges of the gap .
two species stand out from this trend for the planet at 5 au , where cs and ch@xmath78 instead exhibit large deficits .
the assumed binding energies of cs and ch@xmath78 are @xmath79 k and @xmath80 k , respectively . in the case of cs ,
the 5 au planet causes an initial , fast desorption of cs ( and higher column density ) , but the heating simultaneously increases the abundance of gas - phase oxygen bearing molecules . over @xmath81
year timescales , the chemical network converts the cs to ocs ice , resulting in the net chemical deficit plotted in figure [ fig : column_0508 ] .
the orbital time at 5 au is clearly much shorter than the chemical timescale for this process to occur , and so in this case , the conversion from cs to ocs ( and the corresponding cs - deficit ) is over - predicted in our simplified models . to properly model the planet
s effects on the cs chemistry in the inner disk requires models that include differential velocity shear to address these second - order time - dependent effects .
however , for planets located further out in the disk , cs shows simple net desorption behavior without further chemical processing for the more distant , 10 and 20 au planets .
the change in chemical behavior arises because the chemistry leading to ocs is not as efficient radially further out where the gas - phase oxygen is depleted . as a result
, cs becomes a `` primary '' planet tracer for the planets located further out .
thus using thermal desorption as a search tool for planets is more reliable in the more chemically inactive outer disk . for ch@xmath78
, this species low binding energy ( @xmath82 k ) allows for it to be in the gas phase regardless of the planet s location .
the deficit in ch@xmath78 is purely a second - order effect , where the ch@xmath78 at 5 au primarily forms via gas - phase channels originating from ch@xmath83 .
the presence of additional , sublimated gas - phase species that are also able to react with ch@xmath83 will form other molecules besides ch@xmath78 .
these additional pathways stymy the formation efficiency of ch@xmath78 and leads to the ch@xmath78 deficit in fig .
[ fig : column_0508 ] of about a factor of three in column density .
this process likewise occurs over much longer timescales than the planets orbital times and should be studied in further detail including disk differential rotation to fully characterize the ch@xmath78 chemistry in the presence of the planet .
there is some variation of the radial extent of the region affected by the planet as seen in the column density plots .
for example , as seen in figures [ fig : column_2008 ] and [ fig : column_3008 ] , the 20 au and 30 au planets exhibit enhanced hcn confined to a narrow region close the gap edges ( @xmath84 au ) , while ch@xmath78 is enhanced over a much wider region , @xmath85 au . the main factors that determine the chemically affected region are the gap size ( i.e. , a larger gap reduces the amount of heating at the wall s interface ; * ? ? ?
* ) and the binding energy of the species in question .
ch@xmath78 is more weakly bound to the grain surface , @xmath86 k , compared to hcn , @xmath87 k @xcite , and so the region near the planet is only sufficiently hot to evaporate hcn close to the gap walls .
in addition to hcn and ch@xmath78 , nh@xmath88 is also enhanced in the presence of the 30 au planet ( see fig . [
fig : column_3008 ] ) .
this behavior is a feature of the relatively low nh@xmath88 binding energy used in the present models , @xmath89 k @xcite , corresponding to a desorption temperature of @xmath90 k. alternatively , nh@xmath88 deposited on a h@xmath44o ice surface has a much higher binding energy , @xmath91 k @xcite , or @xmath92
k. thus if the higher desorption temperature applies , nh@xmath88 evaporation may be a more useful tracer of planets in the inner few au of a cool protoplanetary disk , or further out for warmer disks like those around herbig ae / be stars . in summary ,
the ideal chemical tracer for identifying planets is fundamentally a balance of the appropriate chemical binding energies and the disk thermal structure . from the 2d chemical model slices we reconstruct the 3d abundance profiles .
we plot the abundance structure for particularly promising species for the four planet locations in figures [ fig : tallmidplane_0508 ] , [ fig : tallmidplane_1008 ] , [ fig : tallmidplane_2008 ] , and [ fig : tallmidplane_3008 ] . from these plots , it is clear that the chemical effects extend over a larger azimuthal range than radial range for most cases .
the abundance enhancement furthermore spans the full wall height until the abundance profile merges into that set by the surface chemistry , as driven by the central star .
however , the vertical density structure is such that the densities are highest closest to the midplane and most tenuous near the surface .
thus the midplane heating by the planet can have a large effect on the total , integrated column density , similar to the direct midplane illumination of transition disks by the central star @xcite .
of all the species considered here , hcn is a particularly robust tracer with a simple midplane chemistry , such that the thermal effects from the planet are primarily its desorption .
the binding energy of hcn in our models is set to 1,760 k @xcite , i.e , a characteristic desorption temperature of @xmath93 k . for the planets beyond 10 au , hcn should otherwise be frozen - out in the midplane for the majority of the disk .
even at 5 au ( midplane temperature of @xmath94 k ) , hcn is just beginning to freeze - out in the absence of the planet .
thus hcn should be an excellent tracer of planets at orbital distance of @xmath95 au from the central star . even if the particular hcn binding energy is revised , the location hcn as a planet - tracer will scale accordingly based upon the midplane temperature profile of the disk and the molecules binding energy to the grain . furthermore ,
if the hcn binding energy varies with the properties of the substrate ( e.g. , * ? ? ?
* ) , the azimuthal region away from the planet can be used as a baseline against which the hcn enhancement can be compared independent of the specific binding energy assumed .
for all four models , the enhancement in hcn is typically an order of magnitude in column density due to the planetary heating . in the absence of the planet s effects , the baseline hcn column density on the opposite side of the disk is @xmath96 @xmath76 for the planet at 30 au and @xmath97 @xmath76 for the planet at 5 au .
the column of hcn in the absence of the planet arises almost entirely from the hcn present at the warm disk surface .
many of these `` signpost '' molecular species have strong rotational transitions that can be used to observationally identify and characterize the local physical conditions near the planet . because of its simple interpretation and large column density
, we focus on hcn emission as a tracer in the present paper , but emphasize that other , perhaps stronger tracers may exist , and the particular tracer will depend on the luminosity of the central star , which will set the thermal structure of the midplane at the disk radii probed by alma .
we use the flexible line modeling code lime @xcite to simulate the emergent line emission from the full disk , oriented face - on at a distance of @xmath5 pc .
we present hcn emission results for the planets at @xmath36 au , 20 au and 30 au as the hcn signature for the 5 au planet was not detectable or distinguishable from the background disk in any of the models considered . because lime samples the model s native grid by selecting random points and accepting / rejecting them based upon particular criteria ( in our case , normalized density and hcn abundance ) , the code can potentially not adequately sample relatively small scale local features in the disk when considering the full 3d volume of the disk . arbitrarily increasing
the point number , however , will add more points into the inner , high density regions of the disk that are already well - sampled . to address this issue , we cast two grids of points , a large grid over the full disk range ( including the planet ) , and a refined spherical grid centered on the planet that contains @xmath98 of the total number points ( @xmath99300,000 ) used to sample the model .
this insures that a sufficient number of points encompass both the disk and the region near the planet .
we furthermore tested this technique for a model without any planet ( but still with the refined local grid ) to confirm that the refined grid does not introduce emission signatures .
such sub - grids will be useful in modeling all types of substructure with lime . because hcn is fairly optically thick at the surface , we model both the hcn and h@xmath100cn isotopologue assuming an isotope ratio of @xmath101c/@xmath100c @xmath102 .
we emphasize that at the high densities where the planet is depositing its heating that selective photodissociation effects on n@xmath44 relating to hcn chemistry @xcite are not expected to play a major role .
for the emission calculations , we consider the @xmath103 and @xmath104 rotational transitions for both species , which we identify as likely strong transitions accessible with alma using preliminary radex models ( radex is a statistical equilibrium solver under the assumption of the large velocity gradient approximation ; * ? ? ?
using lime , we calculate the line optical depth for the four transitions considered . for both hcn lines ,
the disk is optically thick ( @xmath105 ) at all positions at the line center .
however , in the line wings at velocities @xmath106 km s @xmath9 , the circumstellar disk is optically thin for both transitions , while the emission near the planet is thick .
thus these lines will be useful for constraining the temperature of the emitting circumplanetary medium .
the h@xmath100cn @xmath103 line is also thick at line center across the disk ( @xmath107 ) , but becomes thin and the lines , and has an optical depth of @xmath108 near the planet .
@xmath104 is marginally optically thick at line center ( @xmath109 ) , and thin at all locations in the wings .
the lime models represent `` perfect '' images of the emission lines without any thermal noise or beam - convolution ( besides the inherent limitations of the pixel size , which is @xmath110 per pixel or 0.25 au in our models ) . to create more realistic emission models ,
we take the lime output and , using the simulation capabilities of casa we compute simulated alma observations in the same method of @xcite .
the alma.out13 antenna configuration is used for the @xmath36 au simulation of both @xmath103 lines and the alma.out10 antenna configuration was used for the @xmath111 . for the planet at @xmath49 au ( 30 au ) , alma.out16 ( alma.out14 ) was used for @xmath103 and alma.out.11 ( alma.out09 ) for @xmath111 .
configurations were chosen to provide optimal sensitivity and signal - to - noise to detect the planet .
we add realistic thermal noise with 0.25 mm precipitable water vapor , and then reconstruct the image using clean applied with natural weighting .
the synthesized beam of the simulated observations is typically between @xmath112 and @xmath113 .
the results of the simulated line observations ( with noise ) are shown in figure [ fig : em_1008 ] for the @xmath36 au planet , figure [ fig : em_2008 ] for the @xmath49 au planet , and figure [ fig : em_3008 ] for the @xmath54 au planet . in most cases ,
the emission from the planet is difficult to disentangle from the emission originating from the disk .
however , the planet s sublimation effects are confined to a @xmath114 azimuthal extent , which leaves the remaining @xmath115 as a counterpoint against which we can compare . to better isolate the signature of the planet ( without making use of a priori knowledge of the planet s location ) , we average the emission from the entire disk in annuli to estimate the radial line emission profile of the circumstellar disk ( figure [ fig : em_1008 ] , [ fig : em_2008 ] , and [ fig : em_3008 ] , middle row ) .
then , using the average disk profile , we subtract off the smooth background component from the simulated observation and examine the residual features .
this technique will only work if the asymmetric feature is small and/or weak relative to the disk s emission , otherwise a median or averaging azimuthally restricted annuli ( to mask out the feature ) should be used instead .
though this system is face - on , the technique would similarly work for systems that are not face - on if the inclination is known , i.e. , by taking a projected annular average .
the bottom row of figures [ fig : em_1008 ] , [ fig : em_2008 ] , and [ fig : em_3008 ] shows the residual emission , where the planets location is circled in all of the panels . without the additional knowledge of the planet s location ,
the signature in the @xmath103 is not strong enough to identify emission from the noise for the planet , primarily because of the high opacity of this line , which partially hides the deep emission from the midplane .
the lower opacity for @xmath104 for both isotopologues , however , makes for a substantially stronger signal from the planet , even though the noise is substantially higher at these frequencies .
the main limitation is the extremely high sensitivity required to measure the small fractional change in the emission signature from the bright circumstellar disk emission . because in our models the hcn abundance is highest along the walls of the gap near the planet , the face - on inclination provides the largest column density due to the planet .
however , if a more inclined viewing geometry may allow direct imaging of the gap wall depending on the overall scale height of the disk . we examined emission simulations of a disk oriented at @xmath116 with the planet at 90@xmath117 from the projected rotation axis ( in the gap corner ) , and were unable to recover any emission from the planet .
if the planet is on the far - edge of the disk such that the outer gap wall is more directly viewable , and the emission signature may be observable but only for a fraction of the orbit .
thus low inclination ( face - on ) disks are favorable for planet searches with chemistry .
to simulate the 3d chemical models presented here , we have made a number of simplifying assumptions that we discuss further here .
the assumed accretion rate for our models is on the high end of the range for quiescent planetary accretion as predicted by models , @xmath118 m@xmath8 year@xmath9 , and is assumed to be constant .
we found that it was difficult to detect the chemical signature of the planet for lower circumplanetary disk accretion rates , @xmath119 m@xmath8 year@xmath9 .
however , we emphasize that the corresponding accretion luminosity of the planets in our models , @xmath46 l@xmath8 is indeed similar to that observed for the tentative lkca 15 planet . furthermore ,
if young planets undergo frequent accretion outbursts ( increasing the planet s luminosity by many orders of magnitude , see * ? ? ?
* ) and if the cadence of such outbursts is competitive with the chemical timescales , the region affected by the planet may be much larger and more readily detectable .
one significant simplification in the present models is the use of a constant gas - to - dust mass ratio of 100 .
disks are observed to have a deficit of small dust grains in their upper layers , attributed to the settling of grains from the surface into the midplane @xcite .
one of the overall effects of settling is to create a warmer disk , where stellar radiation penetrates deeper .
if the disk becomes too warm , ice sublimation due to the planet may be less effective and one would have to i ) observe gas - phase tracers with relatively higher binding energies to achieve the same contrast between the planet and the background disk or ii ) look for planets at larger radii where the circumstellar disk is cooler .
the addition of dust mass ( in particular dust grain surface area ) in the midplane will also increase the rate of absorption onto grains ( i.e. , decrease the freeze - out time ) and as a consequence , this may help reduce the effects of shear by disk rotation . on the other hand ,
if small grains are quickly converted to large grains with low surface area - to - mass ratios , then the freeze - out time will increase , and there may be a lag in the removal of planet - desorbed molecules from the gas phase thus spreading out the thermal effect of the planet over a larger azimuthal area ( see section [ sec : conclusions ] ) .
we have also used a simplified density model in the present calculations .
recent work by @xcite has shown that the dust density distribution within the gap is different for small ( sub - micron ) and large ( mm ) sized grains .
furthermore , the small grains readily fill the gap while the gaps are clear of the large grain population . in that work ,
the authors find a dust density within the gap of @xmath120 g @xmath35 , which is four orders of magnitude higher than our simple , relatively empty gap model .
to explore how the addition of dust within the gap affects our results , we have calculated models for the planet at 10 au where we have increased the minimum dust density within the gap by four and five orders of magnitude to simulate filled - gaps .
the enhanced dust models have gap densities that are similar to and above that predicted in @xcite , and the dust density drop within the gap is a factor of 10 and 100 below a correspondingly gapless disk
. the results of this modeling is shown in figure [ fig : fill ] .
the most remarkable feature of these models is that the thermal structure of the protoplanetary disk near the young planet is similar regardless of the amount of dust within the gap .
moreover , the surrounding protoplanetary disk s temperature slightly increases with increasing dust in the gap .
the reason for this behavior is that some fraction of the thermal radiation emitted from the protoplanet is able to more freely escape in the vertical direction for the low density gap case and is accordingly lost to space . increasing the dust within the gap naturally traps more of the planet s heating , which is then reradiated isotropically , sending more heat into the protoplanetary disk that would otherwise be lost .
we find a small increase in the hcn column corresponding to the slight increase in protoplanetary disk heating by this effect .
the overall chemical structure is relatively unchanged .
g @xmath35 ) .
models where we have increased the dust density in the gap by four orders of magnitude ( cyan ) and five orders of magnitude ( black ) are indicated .
[ fig : fill],scaledwidth=48.5% ] another structural simplification made in this work was the assumption of axisymmetry .
massive planets will gravitationally interact with the circumstellar disk , generating density waves that can form spiral arms , changing the disk s radial and azimuthal density structure ( see the review of * ? ? ? * ) .
planets can also excite vortices , creating large azimuthal pressure traps , observationally seen as asymmetries in the dust distribution ( as seen in irs 48 ; * ? ? ?
furthermore , the presence of the gap exposes the top of the outer gap edge to direct stellar irradiation , which enhances heating near the gap at the disk surface @xcite . while the latter effect is azimuthally symmetric , the heating can be locally exacerbated by the presence of a massive planet perturbing the vertical disk structure , creating local hot / cold spots near the stellar - irradiated surface at the planet s location @xcite .
we emphasize that all of these planet induced structures / physical effects can happen in concert with planet - heating and ice - desorption in the midplane .
the presence of spiral arms can be potentially confirmed as planetary in origin if associated with a warm accreting source .
furthermore , the stellar gap - edge heating occurs primarily at the surface of the disk , away from the planet s primary effects near the midplane , such that the use of spectral lines with high critical densities may help mitigate observational confusion .
finally , the concentration of dust in pressure bumps can alter the chemical timescales such as the freeze - out time , though it should be noted that the small grains ( which dominate the freeze - out given their high surface area to mass ratio ) are still azimuthally distributed in irs 48 , providing a substrate for freeze - out .
thus ideally one would find evidence of planets through multiple signatures , both chemical , physical , and thermal ( e.g. , by detecting circumplanetary disk continuum emission ; * ? ? ?
finally , we have assumed very simple adsorption / desorption physics , with a single hcn binding energy . in reality , the hcn ice is expected to be mixed with less volatile ices , such as h@xmath44o or ch@xmath88oh . trapping can thus increase the desorption temperature to that of the strongly bound ice , @xmath121 k ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , potentially reducing the effects predicted in this work .
how trapping mediates desorption will essentially depend on how ices were initially deposited on the grain surface substrate simultaneously or sequentially , the thickness of the ice , whether the ice is amorphous or crystalline , which impacts the diffusivity of the trapped species ( e.g. , * ? ? ?
* ) , and the specific molecule in consideration , and whether it is `` co - like '' or `` h@xmath44o - like . ''
ices like hcn are expected to be intermediate species , where some fraction directly sublimates and some fraction remains trapped , while , for example , cs , is expected to desorb more readily upon heating @xcite .
while incorporating the detailed physics of the desorption process is beyond the scope of the paper , they should be taken into consideration when comparing to the inherently more complex observations .
we note that if hcn is partially trapped ( making for a weaker but existent signal ) , the technique of comparing the local ( planet ) emission signature against the disk - averaged emission will nonetheless help compensate for more complex chemical effects than considered in the present work .
we have examined the chemical structure of simple toy models of a protoplanetary disk heated by two sources , the central star and an embedded , luminous young proto - jupiter accreting at @xmath21 m@xmath8 year@xmath9 . chemical species with intermediate freeze - out temperatures , around @xmath94 k , will be particularly useful tracers of the additional planetary heating , as the midplane will be colder than 40 k outside of @xmath122 au .
a planet in a hill - radius sized gap will heat the gap edges by @xmath55 k , which will sublimate certain species and greatly enhance the column density of these species locally .
these changes in column density will translate to observable signatures in the circumstellar disk emission , which will appear as emission line asymmetries .
such asymmetries can be highlighted by subtracting off the azimuthally averaged line brightness profile of the circumstellar disk .
we found that hcn isotopologues are a particularly robust tracer of the heating effects of the planet .
additionally , multiple lines of hcn should be detectable , which will provide additional physical information ( temperature ) about the planet s local environment .
in addition to the excitation information afforded by molecular line observations , the use of gas tracers ( as opposed to continuum ) will also provide spectral information that can be used to look for velocity substructure near the planet .
this technique is most useful for planets between @xmath123 au , where the inner disk is innately too warm to provide sufficient temperature contrast , and planets in the outer disk are expected to carve larger gaps ( have larger @xmath124 ) , and as a direct consequence the planet s heating of the disk is more diluted at larger radii .
finally , the primary ( sublimated ) species can trigger additional chemical processing , creating so - called `` secondary '' planet tracers .
these secondary products may be even brighter than the primary products ; however , to fully characterize them we must take into account the differential rotation of the disk , which is beyond the scope of the present work , but will be the subject of a follow - up paper .
if the timescales for secondary product formation are a substantial fraction of the orbital time , the local effects may be sheared out over a large azimuthal range , i.e. , as an `` arc . ''
if the relevant chemical timescales are much longer than the orbital time , the emission signature of the planet will become a ring ( or double ring ) at the gap edges .
however , in the simple case of rapid sublimation / condensation , the chemical signature is predicted to `` follow '' the planet , implying that we will be able to re - observe and confirm that the emission is associated with the planet as it traverses its orbit .
furlan , e. , hartmann , l. , calvet , n. , dalessio , p. , franco - hernndez , r. , forrest , w. j. , watson , d. m. , uchida , k. i. , sargent , b. , green , j. d. , keller , l. d. , & herter , t. l. 2006 , the astrophysical journal supplement series , 165 , 568 , n. , van dishoeck , e. f. , bruderer , s. , birnstiel , t. , pinilla , p. , dullemond , c. p. , van kempen , t. a. , schmalzl , m. , brown , j. m. , herczeg , g. j. , mathews , g. s. , & geers , v. 2013 , science , 340 , 1199 | we examine changes in the molecular abundances resulting from increased heating due to a self - luminous planetary companion embedded within a narrow circumstellar disk gap . using 3d models that include stellar and planetary irradiation , we find that luminous young planets locally heat up the parent circumstellar disk by many tens of kelvin , resulting in efficient thermal desorption of molecular species that are otherwise locally frozen out .
furthermore , the heating is deposited over large regions of the disk , @xmath0 au radially and spanning @xmath1 azimuthally . from the 3d chemical models ,
we compute rotational line emission models and full alma simulations , and find that the chemical signatures of the young planet are detectable as chemical asymmetries in @xmath2 observations .
hcn and its isotopologues are particularly clear tracers of planetary heating for the models considered here , and emission from multiple transitions of the same species is detectable , which encodes temperature information in addition to possible velocity information from the spectra itself .
we find submillimeter molecular emission will be a useful tool to study gas giant planet formation in situ , especially beyond @xmath3 au . |
small amplitude vibrations of an elastic plate are governed by the three - dimensional equations of the linear theory of elasticity . for an infinite homogeneous plate ,
the frequency @xmath6 depends on the wavelength @xmath7 in the plane of the plate .
when the faces of the plate are free of traction , no energy leakage occurs , then for any real @xmath8 , the secular equations established by rayleigh @xcite yield an infinite number of real roots in @xmath9 .
the dispersion curves of these symmetric @xmath10 and anti - symmetric @xmath11 propagating modes , guided by the plate , are represented by a set of branches in the @xmath12-plane.@xcite the complete lamb mode spectrum depends on material parameters , either expressed by the longitudinal to transverse wave velocity ratio @xmath13 or by the poisson s ratio @xmath14.@xcite dispersion curves @xmath4 of high order modes start from the @xmath15 axis at a finite ordinate @xmath16 . at these cut - off frequencies @xmath17 multiple reflections of longitudinal or shear waves between the top and bottom faces of the plate , give rise to thickness - shear resonances ( modes @xmath18 or @xmath19 ) or to thickness - stretch resonances ( modes @xmath20 or @xmath21 ) at infinite wavelength .
+ in many applications , the plate undergoes a local and impulsive excitation .
the spectra of the transient waves in an elastic plate has been analyzed by weaver and pao.@xcite and santosa @xcite when surface stresses are induced by a laser source , the coupling of the thermo - elastic source with elastic waves is a complex problem.@xcite the theory and simulation of laser generated waves propagating in plates has been the object of several studies.@xcite in a recent paper , laguerre and tresseyde @xcite proposed a method to calculate the excitability of both propagating and non propagating modes . in practice , the energy deposited on the plate by the source of finite dimensions rapidly flows out of the source area except for non propagative modes .
zero group velocity ( zgv ) modes were observed experimentally with various techniques : air - coupled transducers,@xcite impact echo method,@xcite laser ultrasonics.@xcite these experiments demonstrate that the local vibration spectrum of a free plate is dominated by the resonance at the minimum frequency of the @xmath22 lamb mode .
this frequency is slightly lower than the fundamental thickness frequency @xmath23 and corresponds to the junction of @xmath22 and @xmath24 branches , where @xmath25 stands for backward wave.@xcite this is why we chose the notation @xmath26-zgv resonance.@xcite in fact , except for the first three ( @xmath27 , @xmath28 and @xmath29 ) lamb modes , all higher order modes exhibit a minimum frequency for some poisson s ratio.@xcite frequency minima always occur below the cut - off and correspond to wavelengths of the order of the plate thickness . because of there finite wavelength , zgv modes dominate the frequency spectrum of the normal surface displacement after a local impact .
however , it was observed that some thickness resonances ( infinite wavelength ) are also detected when the signal to noise ratio is sufficiently high . for example , fig . [ zgvthspectredispersion1 ] displays the resonance spectrum measured at the source point ( source spot 2.5 mm ) , on a 1 mm - thick duralumin plate .
the three observed resonances can be identified from the dispersion curves shown in fig .
[ zgvthspectredispersion2 ] .
the first and the third ones correspond to the @xmath26 and the @xmath30-zgv modes while the second one is associated to the @xmath31 thickness - shear mode .
[ zgvthspectredispersion3 ] presents the dimensionless cutoff frequencies and minimum frequencies of lamb modes versus the poisson s ratio : the horizontal lines correspond to thickness - shear modes , the dashed lines to thickness - stretch modes and the thick lines to zgv modes . for @xmath32 ( vertical line ) , only two zgv resonances exist at normalized frequencies below 3.2 .
it appears that the @xmath31 thickness - shear resonance , although 20 db below the @xmath26-zgv resonance , is clearly detected , while for example the @xmath29 or the @xmath33 thickness - shear resonances are not observed .
+ the objective of this paper is to explain these observations using theory , simulation and experiments .
it is organized as follows : in sec .
[ sec:1 ] , a simple model is proposed to estimate the temporal decay of zgv and thickness - shear resonances .
the relative amplitudes of thickness resonances , excited and detected by laser techniques , are predicted through simple approximations .
then , sec .
[ sec:2 ] presents experimental results obtained on duralumin and fused silica plates .
the temporal decay of the different resonances are observed and compared to those obtained with a finite difference code .
in order to compare the zgv and thickness resonance behavior , we propose a simple approach only valid for low attenuation materials .
the temporal decay of the @xmath26-zgv lamb mode resonance was studied in prada et al.@xcite in this paper [ eq .
( 3 ) ] , the normal surface displacement associated to a given lamb mode was expressed as @xmath34 where @xmath35 is the coefficient of thermo - elastic conversion into the normal displacement of the mode , @xmath36 is the spectral content of the laser pulse , and @xmath37 is the spatial fourier transform of the distribution of energy @xmath38 deposited on the surface . for a gaussian source beam of radius @xmath39 at @xmath40 @xmath41 where @xmath42 is the deposited energy .
the spatial 2d - fourier transform is then @xmath43 with @xmath44 and @xmath45 .
this integral can be calculated for large times @xmath46 using a taylor expansion of the dispersion relation @xmath4 in the vicinity of a resonance frequency . as the normal surface displacement vanishes for thickness - shear resonances and not for zgv resonances , the solution is derived in a different manner for each type of resonances .
the result established in @xcite is recalled . in the vicinity of a zero group velocity point
@xmath47 , the second order taylor expansion of the dispersion curve is written @xmath48 , \label{wk : fit}\ ] ] where the curvature @xmath3 at zgv point depends on the mode order and on the poisson s ratio . then the integral of eq .
can be approximated by the stationary phase method as @xmath49 it appears that the temporal decay of the resonance follows a @xmath0 law .
furthermore , the resonance amplitude is proportional to the inverse square root of the curvature of the dispersion curve @xmath3 . the coefficient @xmath3 normalized to the product of the transverse wave velocity by the plate thickness is a dimensionless coefficient @xmath50 which only depends on the poisson s ratio @xmath14 .
this parameter , numerically calculated for the first two zgv modes @xmath26 and @xmath51 , is displayed in fig .
[ curvature ] .
it is interesting to notice that for the @xmath26 mode , @xmath52 is maximum for a duralumin plate ( @xmath32 ) and approximately equal to 0.305 . and @xmath51-zgv modes as a function of the poisson s ratio .
inset , dispersion curve of @xmath26 mode in a duralumin plate .
fit ( red line ) by eq . .
] in the vicinity of a cut - off frequency @xmath53 , the slope of the dispersion curve generally vanishes . as shown by mindlin,@xcite
this is not true when there is a coincidence between a thickness - shear and a thickness - stretch resonance of the same symmetry .
the coincidence occurs for symmetrical modes @xmath20 and @xmath18 when the bulk velocity ratio @xmath13 is equal to @xmath54 , and for anti - symmetrical modes @xmath19 and @xmath21 when @xmath55 . except for these particular cases , in the vicinity of a cut - off frequency @xmath56 , the dispersion law @xmath4 can be developed to the second order as @xmath57 for a thickness - shear resonance , @xmath58 , the normal surface displacement and the conversion coefficient vanish : @xmath59 .
then , the first order taylor expansion of this coefficient can be written as @xmath60 the normal surface displacement can be approximated as @xmath61 or @xmath62 using the parameter @xmath63 , and the laplace transform @xcite equation @xmath64 with @xmath65 for @xmath66 , the displacement can be approximated by @xmath67 for times @xmath68 the displacement is @xmath69 it results that the decay of thickness mode resonance is much faster that the decay of the zgv mode resonance , this point will be discussed in the next section . in order to explain why the @xmath31 resonance is detected while , for example , the @xmath29 or the @xmath33 are not , it is necessary to compare the relative amplitude of the resonances and thus to estimate the coefficients @xmath3 and @xmath70 .
the dependence of @xmath3 and @xmath70 with respect to the poisson s ratio and the order of the resonance are now investigated . for thickness resonances ,
the curvatures of the dispersion curve were derived by mindlin and can be found in shuvalov and poncelet.@xcite the coefficient @xmath71 of a thickness - shear resonance at frequency @xmath72 is @xmath73\right ) \label{curvaturednt}\ ] ] the normalized curvature @xmath74 of the dispersion curve at the origin @xmath5 for the @xmath75 and the @xmath31 thickness - shear resonances are displayed in fig .
[ curvature_shear ] .
+ for the @xmath31 mode , the curvature diverges for @xmath76 ( @xmath77 ) when the @xmath31 shear resonance intersects the @xmath78 stretch resonance [ fig .
[ curvature_shear2 ] ] . above this point ,
the curvature is negative [ fig .
[ curvature_shear4 ] ] , which means there is a backward wave branch emanating from the thickness - shear mode and ending on the zgv mode .
then , the curvature vanishes for @xmath79 , which corresponds to the upper limit of the zgv mode . at this point , according to eq .
, the amplitude of the thickness resonance diverges .
the duralumin poisson s ratio is just above this limit and the curvature is @xmath80 which leads to @xmath81 . on the contrary , for the @xmath75 mode
the curvature is much larger [ fig .
[ curvature_shear3 ] ] , @xmath82 which explains why the @xmath75 thickness - shear resonance can not be detected . for thickness - stretch resonances of order
@xmath83 , the curvatures @xmath84 found in @xcite are equal to @xmath85\right ) \label{curvaturednl}\ ] ] the dependence of the conversion coefficient @xmath86 on the poisson s ratio is complex . here , we present a simplified approach for thickness modes .
for a metallic surface and in the thermo - elastic regime , the source mostly induces an in - plane force .
thus , one can consider that the conversion coefficient is proportional to the in - plane component @xmath87 of the mode at the surface .
as the normal displacement @xmath88 is detected at the surface , the conversion coefficient can be defined as the product of the two components normalized to the integral of the squared displacement through the plate thickness @xmath89 where @xmath90 is the half - plate thickness @xmath91 . in the vicinity of a thickness - shear mode
we have @xmath92 and @xmath93 the conversion coefficient simplifies into @xmath94 the general expressions of the in - plane @xmath95 and normal @xmath96 displacements of symmetrical @xmath97 and anti - symmetrical @xmath98 modes are given by @xcite @xmath99 where @xmath100 and @xmath101 .
the coefficients @xmath102 and @xmath103 depend on @xmath8 and satisfy the two equations in the vicinity of the thickness - shear resonance of order @xmath104 we have the following taylor expansion @xmath105 for a symmetrical ( @xmath106 ) shear resonance of order @xmath107 , we have @xmath108 and @xmath109 and eq .
gives @xmath110 = 4ikh(-1)^n a$ ] .
the expressions of the displacements at the surface become @xmath111 for an anti - symmetrical shear resonance ( @xmath112 ) of order @xmath113 , analogous equations can be written . finally , the coefficients of the leading term of @xmath86 [ eq . ]
is @xmath114 a similar derivation for the thickness - stretch modes is given in appendix [ sec : a1 ] and leads to the following coefficients @xmath115 these coefficients diverge if @xmath116 is rational which corresponds to a coincidence between thickness - shear and thickness - stretch resonances . as mentioned in sec .
[ sec:12 ] , the model is no more valid in this case . according to eq .
, the amplitude of the resonance is proportional to @xmath117 .
this parameter is plotted as a function of the poisson s ratio in fig .
[ clct1 ] for the thickness - shear modes and in fig .
[ clct2 ] for the first four thickness - stretch modes . + for sake of clarity the coefficient where plotted only for resonance frequencies below the fifth shear resonance frequency @xmath118 .
the two green dash lines indicate the poisson s ratios of fused silica and duralumin .
it appears that for fused silica the coefficient of the @xmath119 thickness - shear resonance is higher by at least one order of magnitude than for the other modes .
it means that the @xmath119 resonance will dominate the other thickness resonances .
the same observation can be done for the @xmath31 thickness - shear mode in duralumin .
this graph also indicates that the thickness - stretch resonances should not be easily detected .
is it important to underline that these curves give the relative weights of thickness resonances for a given poisson s ratio . to compare resonance amplitudes for two different materials , one should consider other parameters like optical absorption or thermal dilatation also involved in the conversion process
a heterodyne interferometer @xcite associated with a q - switched nd : yag laser at @xmath120 nm was used .
lamb waves were generated in the thermo - elastic regime by the absorption of a short pulse ( duration @xmath121 ns , energy @xmath122 mj , repetition rate of @xmath123 hz , spot diameter 2.5 mm ) . the calibration factor for mechanical displacements normal to the surface ( @xmath124 mv / nm ) is constant over the detection bandwidth ( @xmath123 khz to @xmath125 mhz ) .
measurements were made on two materials , duralumin and fused silica , having a low attenuation ( less than 1 db / m in the mhz frequency range ) . for fused silica ,
a thin aluminium layer was deposited for a sufficient optical absorption at the surface .
the spatial profile of the resonances was measured on a 1-mm thick duralumin plate .
the ratio @xmath126 is equal to 1.25 which corresponds to @xmath127 .
the normal surface displacements , detected by scanning the probe across the source along a 30-mm line , is displayed in fig .
[ dep_field_profil : a ] for the first @xmath128 .
after about @xmath129 most propagating modes have escaped the scanned area . a standing mode is clearly observed immediately after .
the temporal fourier transforms of these signals calculated on the same time - window ( @xmath130 to @xmath131 ) reveal 3 resonances at 2.923 mhz , 4.827 mhz and 9.560 mhz .
as expected , they correspond respectively to @xmath26 , @xmath31 and @xmath30 resonances .
the amplitude distributions around the resonance frequencies are displayed in fig [ dep_field_profil : b ] .
the type of resonance can be distinguished from the spatial profiles . + for zgv resonances , they follow a bessel function @xmath132 with @xmath133 mm for @xmath26 and @xmath134 mm for @xmath30 , corresponding to the interference of backward and forward waves of opposite phase velocities .
the profile of the thickness resonance is different and only corresponds to propagative forward modes .
contrary to zgv modes , the profile depends on the time window used in the fourier transform especially at short times .
the normal surface displacements were detected at the source location during @xmath135 on both duralumin and fused silica plates .
the fourier transforms of these signals are displayed in figs .
[ att_zgv_modes1 ] and [ att_zgv_modes2 ] . in both cases ,
the @xmath26 mode dominates . according to the previous analysis ,
higher order resonances for fused silica are different than for duralumin and correspond to @xmath51-zgv mode and @xmath119 thickness - shear mode .
the signals filtered ( bandwidth @xmath136 mhz ) around the resonance frequencies of the @xmath26-zgv mode and thickness - shear modes @xmath31 for duralumin and @xmath119 for fused silica are displayed in figs .
[ att_zgv_modes3 ] and [ att_zgv_modes4 ] .
+ it is clearly observed that amplitudes of thickness modes decay much faster than zgv mode amplitudes .
the temporal decays were estimated by using a hilbert transform of the filtered signals . as shown in figs .
[ att_zgv_log1 ] and [ att_zgv_log2 ] , the amplitudes decrease like @xmath0 for @xmath26-zgv modes .
similar decays were observed for @xmath30-zgv mode in duralumin and @xmath51-zgv mode in fused silica .
this result is in good agreement with the theory [ eq . ] . for thickness - shear modes ,
the power law fit provides a @xmath137 exponent for duralumin and @xmath138 for fused silica , which is slightly different than the @xmath139 predicted by the above theory [ eq . ] . using a homemade finite difference code , the local response
was also calculated and similar temporal decays ( @xmath137 ) were obtained for both resonance types as shown in figs .
[ att_zgv_log3 ] and [ att_zgv_log4 ] .
this decay is observed for time higher that @xmath140 , which is consistent with the validity condition @xmath141 , since @xmath142 is equal to @xmath143 for the @xmath31 mode in the duralumin plate . for thickness modes ,
the discrepancy between experiments and theory could be ascribed to the various approximations made in the model .
the local elastic resonances of a plate excited by a laser source was studied for low attenuation materials .
it was observed that while the dominant resonances are associated to zero group velocity lamb mode of finite wavelength , some thickness resonances are also detected .
the type of resonance can be determine from the spatial profiles . for zgv resonances
, it follows a bessel function corresponding to the interference of counter propagating modes of opposite phase velocities .
the profile of thickness resonance is very different and only correspond to interferences of forward modes . in the experiment ,
the @xmath31 thickness - shear resonance is detected in duralumin plates and the @xmath119 thickness - shear resonance is detected in fused silica , while the other thickness modes are not observed .
this is explained through a simplified model which shows that the amplitude of the resonances is proportional to @xmath2 where @xmath3 is the curvature of the dispersion law and that the temporal decay of the resonance is also different : for zgv modes it follows a power law @xmath0 , while for thickness resonances the decay is much faster .
the proposed theory predicts that the decay is as @xmath1 while experimental and numerical results reveal slightly slower decay as @xmath144 .
this discrepancy can be ascribed to the approximations made in the theory .
the development of a more complete model should be the object of further studies .
following the same approach than for thickness - shear resonances , the conversion coefficient @xmath86 for thickness - stretch resonances can be approximated as @xmath145 for a resonance of order @xmath104 : @xmath146 , and the following taylor expansions can be written @xmath147 for a symmetrical resonance of order @xmath113 and @xmath106 , we have @xmath148 and @xmath149 . using the first order development and inserting the above equation in eq . , it comes @xmath150 and the parallel and normal displacements become @xmath151 it follows that the coefficient @xmath152 is @xmath153 for an anti - symmetrical resonance of order @xmath107 and @xmath112 , we have @xmath154 and @xmath155 . using the first order development of eq .
we have @xmath156 the displacements become @xmath157 and the coefficient of the leading term of @xmath86 @xmath158 more generally , we obtain for thickness - stretch modes of order @xmath104 : @xmath159 +
this work was supported by labex wifi ( laboratory of excellence anr-10-labx-24 ) within the french program `` investments for the future '' under reference anr-10-idex-0001 - 02 psl*. r. d. mindlin , mathematical theory of vibrations of elastic plates , _
http://dx.doi.org/10.1109/freq.1956.199260[in proceedings ] of 11th annual symposium on frequency control ( u.s . army ) signal engineering laboratories , fort monmouth , nj _ , ( 1956 ) , pp .
r. d. mindlin , _
introduction to the mathematical ] theory of vibrations of elastic plates .
_ , singapore : world scientific , ed .
j. yang ( 2006 ) , pp . 2378 .
d. royer , d. clorennec , and c. prada , lamb mode spectra versus the poisson s ratio in a free isotropic elastic plate , _ http://dx.doi.org/10.1121/1.3117685[j .
soc . am . ]
_ , * 125 * ( 2009 ) , pp .
36833687 .
f. treyssde and l. laguerre , numerical and analytical calculation of modal excitability for elastic wave generation in lossy waveguides , _
_ , * 133 * ( 2013 ) , pp
. 38273837 . c. prada , o. balogun , and t. murray , laser based ultrasonic generation and detection of zero - group velocity lamb waves in thin plates , _
phys . lett . ] _ , * 87 * ( 2005 ) , p. 194109 . d. clorennec , c. prada , d. royer , and t. w. murray , laser impulse generation and interferometer detection of zero group velocity lamb mode resonance , _ http://dx.doi.org/10.1063/1.2220010[appl .
phys . lett . ]
_ , * 89 * ( 2006 ) , p. 024101 .
m. cs , d. clorennec , d. royer , and c. prada , edge resonance and zero group velocity lamb modes in a free elastic plate , _ http://dx.doi.org/10.1121/1.3607417[j .
soc . am . ]
_ , * 130 * ( 2011 ) , pp .
689694 .
a. l. shuvalov and o. poncelet , on the backward lamb waves near thickness resonances in anisotropic plates , _ http://dx.doi.org/10.1016/j.ijsolstr.2008.02.004[int . j. solids struct . ]
_ , * 45 * ( 2008 ) , pp .
34303448 .
i. s. gradshteyn and i. m. ryzhik , _
http://www.sciencedirect.com/science/book/9780122947575[table of integrals , series and products ] _ , academic press , new york , ed .
a. jeffrey and d. zwillinger ( 2000 ) , p. 1100 .
d. royer and e. dieulesaint , optical detection of sub - angstrm transient mechanical displacements , _
http://dx.doi.org/10.1109/ultsym.1986.198793[in ultrasonics symposium proceedings ] _ , ( 1986 ) , pp . 527530 . | this paper investigates the dependence on poisson s ratio of local plate resonances in low attenuating materials . in our experiments ,
these resonances are generated by a pulse laser source and detected with a heterodyne interferometer measuring surface displacement normal to the plate .
the laser impact induces a set of resonances that are dominated by zero group velocity ( zgv ) lamb modes . for some poisson s ratio ,
thickness - shear resonances are also detected .
these experiments confirm that the temporal decay of zgv modes follows a @xmath0 law and show that the temporal decay of the thickness resonances is much faster .
similar decays are obtained by numerical simulations achieved with a finite difference code .
a simple model is proposed to describe the thickness resonances .
it predicts that a thickness mode decays as @xmath1 for large times and that the resonance amplitude is proportional to @xmath2 where @xmath3 is the curvature of the dispersion curve @xmath4 at @xmath5 .
this curvature depends on the order of the mode and on the poisson s ratio , and it explains why some thickness resonances are well detected while others are not . |
macroscopic quantum tunneling is a fascinating manifestation of quantum - mechanical behavior in large - scale systems with many degrees of freedom .
it is responsible for a finite life time of a metastable state that can not decay classically by thermal activation at zero temperature . since 1980-ies , macroscopic
quantum tunneling has been studied in a number of condensed - matter systems : various types of josephson junctions , @xcite josephson junction arrays,@xcite phase - slip centers,@xcite vortices in superconductors,@xcite small ferromagnetic particles,@xcite etc .
theoretical description of macroscopic quantum phenomena is based on the concept of a collective coordinate , @xmath3 . in the simplest cases ,
it can be considered as a slow variable , which allows to integrate out the other electronic degrees of freedom and end up with an effective quantum mechanics for @xmath4 .
the resulting dynamics of the collective degree of freedom is generically non - local in time since interaction with other modes produces the retardation effect , intimately related with dissipation .
intensive studies of dissipative quantum mechanics@xcite have been triggered by the pioneering work of caldeira and leggett.@xcite among various systems , the josephson junction can be considered as a prototypical model of macroscopic quantum tunneling . here
the superconducting phase difference across the junction , @xmath3 , plays the role of a collective coordinate .
a nondissipative josephson current can run through the system , described by a certain current - phase relation @xmath5,@xcite nonsinusoidal in general ( fig . [ img : iu_chi_general]a ) , with the critical ( maximal ) current @xmath6 reached at some phase difference @xmath7 .
the current can be obtained by differentiating the free energy of the junction : @xmath8 . in the current - biased regime with the driving current @xmath9 ,
the equilibrium states ( @xmath10 ) correspond to the minima of the legendre - transformed free energy , @xmath11 , which has the standard form of a washboard potential shown in fig .
[ img : iu_chi_general]b .
the supercurrent state with @xmath12 is metastable and does decay ( due to quantum or thermal fluctuations ) into a resistive branch .
then the junction usually stays in the dissipative regime unless @xmath9 is decreased to a smaller retrapping current , resulting in a hysteretic current - voltage characteristic.@xcite early studies of the supercurrent decay in josephson junctions @xcite assumed the tunnel limit , when the superconducting terminals are coupled through an insulating layer without its own electron dynamics ( sis junctions ) . possible dissipation and charging effects
were included phenomenologically by adding an ohmic resistor and a capacitor in parallel with the junction [ resistively shunted junction ( rsj ) model].@xcite owing to advances in nanofabrication technology , current experimental interest has turned towards the study of sns junctions when two superconducting terminals are connected via a normal region @xcite ( including graphene@xcite , and the surface of a topological insulator@xcite ) . an important new physics in this case is related to the superconducting proximity effect,@xcite which renders the normal part of the junction `` partially superconducting '' .
the strength of the proximity effect is characterized by the value of the spectral minigap ( at @xmath13 ) which can be estimated as @xmath14 , where @xmath15 is the superconducting gap in the terminals and @xmath16 is the time required for an electron in the normal region to establish a contact with superconductors.@xcite for a diffusive wire of length @xmath17 with good contacts to superconductors , @xmath18 is of the order of the thouless energy , @xmath19 , where @xmath20 is the diffusion coefficient .
depending on the relation between @xmath15 and @xmath21 , one can distinguish between short ( @xmath22 , with @xmath23 ) and long ( @xmath24 , with @xmath25 ) josephson junctions .
for short junctions , @xmath5 can be expressed in terms of the transmission coefficients of the normal region,@xcite whereas long junctions require special treatment.@xcite in both cases , the critical current can be written as @xmath26 where @xmath27 is the normal - state conductance of the junction , and @xmath28 is a model - dependent factor .
equation ( [ ic ] ) is a generalization of the ambegaokar - baratoff relation to sns junctions . in this paper
we make the first step towards the theory of the supercurrent decay in josephson junctions beyond the tunnel limit and calculate the life time of the supercurrent state due to _ quantum _ fluctuations .
the normal region of the sns junction will be modeled either by a short mesoscopic conductor characterized by an arbitrary set of transmission coefficients @xmath29 or by a quasi - one - dimensional diffusive wire of an arbitrary length @xmath17 ( see fig .
[ img : sns ] ) .
the spectral gaps in superconductors are assumed to be equal .
we will work in the limit @xmath30 ( but not too close to @xmath6 so that the wkb approximation is still applicable ) . in this limit
free energy @xmath31 is flattened and @xmath3 becomes the slowest variable in the system , which makes it possible to treat electronic degrees of freedom in the adiabatic approximation ( see the justification in sec .
[ ss : applicability ] ) . at low temperatures , @xmath32 ,
thermal quasiparticles responsible for dissipation are frozen out and the junction can be described by the imaginary - time effective action with the capacitive dynamic term : @xmath33 = \int d\tau \left [ \frac{c(\chi_c)}{2 ( 2 e)^2 } \big ( \frac{\partial \chi}{\partial \tau } \big)^2 + f(\chi ) \right ] .\ ] ] such a local - in - time description of the phase dynamics is valid only in the supercurrent state with @xmath34 .
the effective capacitance , @xmath35 is a sum of the geometric , @xmath36 , and intrinsic , @xmath37 , capacitances of the junction .
the latter is determined by the response of the andreev bound states to nonstationary boundary conditions .
these states are known to be responsible for carrying the supercurrent.@xcite at lowest temperatures , their low - frequency dynamics can not be damped ( with the kernel @xmath38 in terms of matsubara frequency ) as it is not related to any dissipation , and thus it should be capacitive ( with the kernel @xmath39 ) .
thus the problem of quantum decay of a nearly critical supercurrent reduces to calculating the intrinsic capacitance @xmath40 of the junction ( evaluated at the critical phase @xmath41 ) .
the intrinsic capacitance of the tunnel ( sis ) junction was obtained in the works of larkin and ovchinnikov@xcite and ambegaokar , eckern and schn:@xcite@xmath42 ( here @xmath27 is the tunnel conductance of the barrier ) .
this result has been recently extended to the case of arbitrary short junctions by galaktionov and zaikin .
@xcite they derived the general expression for @xmath43 in terms of transmission coefficients @xmath44 [ see eq .
( [ galaktionov_zaikin_formula ] ) ] and found that , quite generally , @xmath45 .
we will generalize that result further and show that the magnitude of @xmath37 for an arbitrary sns junction is determined by the ratio of the normal - state conductance @xmath1 to the minigap @xmath2 induced in the normal part due to the proximity effect [ cf .
( [ ic ] ) for @xmath6 ] : @xmath47 the dimensionless capacitance coefficient @xmath48 is model - dependent . at the critical phase difference , @xmath49 is of the order of one , and we will evaluate it for an arbitrary short scatterer using galaktionov - zaikin theory ( see table [ tbl : from_multicharge ] ) and for a quasi - one - dimensional diffusive wire with the help of the nonlinear sigma model ( see figs .
[ img : capacity - eg ] and [ img : capacity - chi ] ) . .
various properties of some model josephson junctions ( @xmath50 marks short junctions ) , and the resulting values for the capacitance coefficient @xmath51 . [ cols="^,^,^,^,^",options="header " , ] in the case when the geometric capacitance can be neglected ( @xmath52 ) , we obtain that the instanton action , which determines the life time @xmath53 of the supercurrent state , depends only on the dimensionless conductance of the normal part , @xmath54 where @xmath55 is the conductance quantum , and the coefficient @xmath56 is model - dependent .
its values for several model josephson junctions are given by @xmath57 the paper is organized as follows .
the zero - temperature intrinsic capacitance of a short sns junction with an arbitrary normal region specified by its transmission coefficients is discussed in sec .
[ s : scattering ] on the basis of galaktionov - zaikin formula . in sec .
[ s : sigma ] we present an approach based on the nonlinear sigma model which allows us to determine the intrinsic capacitance of the josephson junction through a diffusive quasi - one - dimensional normal wire of an arbitrary length , tracing its behavior from the short to long - wire limits . the resulting expression for the life time of a slightly subcritical current due to quantum fluctuations
is presented in sec .
[ s : lifetime ] , where we also discuss the limits of applicability of our approach . in the concluding sec .
[ s : conclusion ] we summarize our results and discuss them in the context of the experimentally observed hysteresis in extended josephson junctions .
finally , numerous technical details are relegated to several appendices . in our systems of units
in this section we consider the case when the normal part can be described by the set of its transmission eigenvalues @xmath59 with the distribution function @xmath60 the scattering matrix theory proved to be a powerful and intuitively clear method for studying quantum transport in mesoscopic conductors.@xcite for normal systems , its most renowned predictions are the celebrated landauer formula for the dc conductance , @xcite @xmath61 , and the expression for the fano factor in the theory of shot noise,@xcite @xmath62 .
application of the scattering matrix approach to superconducting hybrid system is a more delicate issue.@xcite it requires the elements of the normal - state scattering matrix to be energy independent on the relevant energy scale set by the minigap @xmath2.@xcite since the scattering matrix acquires energy dependence at the scale @xmath18 , such a situation is realized for sufficiently short junctions with @xmath22 and hence @xmath23 ( a diffusive sns junction with ideal interfaces belongs to this class if @xmath63 )
. then the josephson current can be found with the help of beenakker formula:@xcite @xmath64 where @xmath65 is the temperature , and @xmath66 is the energy of the andreev bound state in the corresponding channel : @xmath67 phase dynamics of a short josephson junction has been recently studied by galaktionov and zaikin.@xcite with the help of the keldysh technique they derived an effective action for small phase fluctuations near the equilibrium phase @xmath3 .
slow phase dynamics in the zero - temperature limit is governed by the capacitance ( [ c = c+c ] ) , where the intrinsic capacitance is given by @xmath68^{-5/2 } \bigg [ 2t_\alpha \left ( t_\alpha - 2 \right ) \sin^2 \left ( \chi/2 \right ) \\ { } + 5 + t_\alpha + \frac{2 - 2 \left ( 1 + 2 t_{\alpha } \right ) \sin^2 \left ( \chi/2 \right)}{t_{\alpha } \sin^4 \left ( \chi/2\right ) } \bigg ] \bigg\}.\end{gathered}\ ] ] since the resulting expression is sufficiently involved , we find it instructive to rederive eq .
( [ galaktionov_zaikin_formula ] ) in the matsubara representation . though the general line of the derivation is very similar to that of ref .
, the absence of an additional keldysh matrix structure makes it possible to track the details of the calculation . this procedure summarized in appendix [ app : derivation_of_k_for_short_junctions ] reproduces eq .
( [ galaktionov_zaikin_formula ] ) obtained by galaktionov and zaikin . equation ( [ galaktionov_zaikin_formula ] ) can be used to evaluate numerically the intrinsic capacitance , @xmath37 , of various short josephson junctions.@xcite we calculate it at the critical phase , @xmath69 , for three types of structures , with the superconducting terminals coupled through the following links : * a tunnel barrier with all @xmath70 , * a short ( @xmath63 ) diffusive wire , with @xmath71 given by dorokhov distribution,@xcite @xmath72 * a ballistic chaotic quantum dot,@xcite with @xmath73 the results along with some characteristics of these josephson junctions are presented in table [ tbl : from_multicharge ] .
we see that the capacitance coefficient @xmath49 defined in eq .
( [ c - general ] ) is generally of the order of unity , depending on the particular distribution of transmission coefficients @xmath71 .
here we calculate the intrinsic capacitance of the sns junction made of a normal diffusive wire coupled to superconductors through highly transparent interfaces . the scattering - matrix approach used in the previous section
can not be applied to sufficiently long wires ( @xmath74 ) . in this limit ,
the energy dispersion of the scattering matrix at relevant energies @xmath75 is not negligible , and the action does not have a simple form of eq .
( [ short_junction_action]).@xcite to find @xmath37 one then has to use a more general method of the nonlinear sigma model and study spatially inhomogeneous configurations of the matrix field @xmath76 in the wire . superconducting proximity effect in a diffusive metal can be conveniently described by the replica sigma model in the imaginary time @xcite .
its action for the quasi - one - dimensional metallic wire of fig .
[ img : sns ] can be written as @xmath77 = \frac{\gn}{16 g_q } \int_{-1/2}^{1/2 } dr { \mathop{\rm tr}}\left [ \left ( \nabla q \right)^2 - 4 \varepsilon \tau_3 q \right],\end{gathered}\ ] ] where the spacial coordinate @xmath78 is measured in the units of the wire length @xmath17 , summation goes over matsubara energies @xmath79 , and @xmath80 with the thouless energy @xmath81 .
in general , the field @xmath76 subject to the constraint @xmath82 acts as a matrix in the replica space , but since the phase difference @xmath3 carries no replica index , the structure of the instanton describing quantum tunneling is trivial in the replica space and we can omit the replica index everywhere . then @xmath83 becomes a matrix in the nambu - gorkov space ( pauli matrices @xmath84 ) , as in appendix [ app : derivation_of_k_for_short_junctions ] .
the stationary saddle point @xmath85 for the sigma - model action ( [ sigma_model_action ] ) satisfies the usadel equation@xcite @xmath86=0,\end{gathered}\ ] ] where prime stands for the derivative with respect to @xmath78 . assuming perfect ns interfaces , we write the boundary conditions with the antisymmetric choice of superconducting phases as @xmath87 , where @xmath88 is defined in eq .
( [ q_s_general ] ) . in the standard parametrization in terms of the spectral angles
@xmath89 and @xmath90 , @xmath91 the usadel equation reduces to two coupled equations : [ usadel_equations ] @xmath92 with the boundary conditions at @xmath93 : @xmath94 the stationary supercurrent is given by @xmath95 [ its conservation is guaranteed by eq .
( [ usadel_equation_1 ] ) ] . temperature and length dependence of the critical current of diffusive sns junctions was studied in ref . .
our further analysis will be limited to the @xmath96 case . in the presence of a time - dependent phase difference across the junction , @xmath97 ,
the @xmath76 matrices in the leads become functions of two time arguments [ cf .
( [ qlr ] ) ] : @xmath98 in order to get the action as a functional of @xmath97 one has to integrate out @xmath99 in the wire with the boundary conditions ( [ q - chi - bc ] ) . near the criticality , @xmath34 ,
variations of @xmath100 on the instanton trajectory are small and hence induced ( non - diagonal in energy ) deviations of @xmath76 from the saddle - point @xmath101 can be treated perturbatively .
perturbative expansion of the diffusive sigma - model near the supercurrent state in josephson junctions has been developed in ref . , and here we generalized it to the case of nonstationary boundary conditions ( an alternative approach would be to work on the level of the dynamic usadel equations as it was done in ref . ) . in that paper
the authors studied the real part of admittance for the case of the long diffusive sns junction , while the intrinsic capacitance can be extracted from the low - frequency expansion of its imaginary part .
first we present the stationary saddle point @xmath102 in the form @xmath103 with the diagonal - in - energy position - dependent matrix @xmath104}e^{-i\tau_{3}\varphi_\epsilon(r)/2}$ ] .
then we parametrize small fluctuations near @xmath101 as @xmath105 in terms of the field @xmath106 satisfying @xmath107 and @xmath108 .
the former constraint implies @xmath109 , and we combine @xmath110 and @xmath111 into a vector object @xmath112 where the unitary matrix @xmath113 is introduced to simplify the quadratic action [ eq .
( [ expression_for_s_ww_bulk ] ) below].@xcite its explicit form can be found in appendix [ app : s_w_and_s_ww ] .
the pauli matrices acting in the @xmath114 space will be referred to as @xmath115 .
@xcite physically , the field @xmath116 describes soft diffusive modes on top of a superconducting state ( in the presence of a supercurrent they can no longer be classified as difusons and cooperons ) .
expansion of the action in the powers of @xmath116 reads : @xmath117 = s[q_0 ] + s^{(w)}[w ] + s^{(w^2)}[w ] + \dots , \end{gathered}\ ] ] where each term is a combination of the bulk and boundary contributions : @xmath118 = s_\text{bulk}^{(w^n)}[w ] + s_\text{bound}^{(w^n)}[w ] .\end{gathered}\ ] ] since @xmath101 is the saddle point , the linear term @xmath119 is totally due to the boundary , but it will contribute to the quadratic - in-@xmath120 action [ see eq .
( [ quadratic_in_x_part_of_action ] ) below ] .
the boundary contributions are given by @xmath121 where @xmath122 , and the operators @xmath123 and @xmath124 read @xmath125 the bulk contribution is given by @xmath126 where explicit expression for the effective potential @xmath127 can be found in appendix [ app : s_w_and_s_ww ] .
the effective action for the phase variable @xmath128 is determined by the saddle - point trajectory of the action ( [ expansion_of_action_in_w ] ) with the boundary conditions ( [ q - chi - bc ] ) .
it can be written as a series in @xmath120 : @xmath129 with @xmath130 .
as we are interested in quadratic - in-@xmath120 terms , it suffices to follow only @xmath131 and @xmath132 . due to nonlinearity of the theory
, the linear terms @xmath131 may influence the equation of motion for the quadratic term @xmath132 .
however , as shown in appendix [ app : higher_orders ] , this mechanism gives no contribution to the effective action .
therefore we may consider linear equations of motion obtained by varying @xmath133 : @xmath134 \bb_{\epsilon_1,\epsilon_2}(r ) = 0 , \end{gathered}\ ] ] and process each order in @xmath120 independently . the boundary conditions for eq .
( [ equations_of_motion_for_b ] ) can be obtained by comparing eq .
( [ q - chi - bc ] ) with the parametrization ( [ w_parametrization ] ) at @xmath93 .
the first two terms are given by [ boundary_cond_for_w_in_1_and_2_orders_in_x ] @xmath135 where [ bc - w ] @xmath136 and @xmath137/2 $ ] . in what follows we will need only diagonal elements of @xmath132 , so we introduce the notation @xmath138 .
with the boundary conditions ( [ bc - w ] ) , the solution of eq .
( [ equations_of_motion_for_b ] ) can be written in the form : [ w_through_xi_r_dep ] @xmath139 where the functions @xmath140 and @xmath141 obey the same eq .
( [ equations_of_motion_for_b ] ) as @xmath142 and satisfy the boundary conditions : [ bc_xi ] @xmath143 to get the quadratic - in-@xmath120 part of the action , we substitute eqs .
( [ w_through_xi_r_dep ] ) to eq .
( [ expansion_of_action_in_w ] ) : @xmath144 + s^{(w ) } [ w^{(2 ) } ] , \end{gathered}\ ] ] which takes the form of eq .
( [ quadratic_terms_in_the_expansion_of_multicharge ] ) with the kernel @xmath145 after the substitution @xmath146 and @xmath147 .
then the intrinsic capacitance should be determined from eq .
( [ c_through_k ] ) . to find it numerically , we expand the matrix @xmath148 and the solutions @xmath149 in powers of @xmath150 as @xmath151 and write down the system of linear equations for the first three coefficients : @xmath152 which should be supplemented by the boundary conditions obtained from the expansion of @xmath153 in @xmath150 . extracting @xmath154 from eq .
( [ quadratic_in_x_part_of_action ] ) , we obtain @xmath155 [ the term originating from the @xmath156 part of eq .
( [ b_definition ] ) equals zero ] . according to eq .
( [ c_through_k ] ) , the intrinsic capacity of the junction is determined by the integral of eq .
( [ komegaomega_through_xi ] ) over @xmath157 .
the first step of numerical simulation is to find the solutions for the usadel equations ( [ usadel_equations ] ) .
one way is to use the explicit expression in terms of elliptical functions , @xcite that reduces to finding two constants from a system of algebraic equations numerically .
however in order to obtain the functions @xmath89 and @xmath90 numerically we find it more convenient to follow the other way and to simulate the original differential eqs .
( [ usadel_equations ] ) directly .
obtained spectral angles are used to determine the matrices @xmath158 , @xmath159 , and @xmath160 in eq .
( [ omega - in - omega ] ) . then the system ( [ three_eqs_on_b ] ) is solved numerically , the solutions are substituted to eq .
( [ komegaomega_through_xi ] ) and the integral over @xmath157 in eq .
( [ c_through_k ] ) is calculated with a proper energy grid . as a check of the numerical method , it is instructive to calculate @xmath161 .
it is easy to see that it should be proportional to the derivative of the current - phase relation @xmath162 . by checking that this quantity indeed crosses zero at the critical phase @xmath7 we can provide an independent test for our numerical calculations . the resulting dependence of the capacitance coefficient @xmath163 evaluated at the critical phase difference , @xmath7 , on the ratio @xmath164 is presented in fig .
[ img : capacity - eg ] ( for the dependence of @xmath2 on @xmath165 see , e. g. , fig . 5 of ref .
) . in general
, @xmath49 is of the order of unity , varying from @xmath166 for short wires to @xmath167 for long wires [ in that limit , @xmath168 ( ref . ) and @xmath169 ( ref . ) ] .
the maximal value of @xmath170 is achieved at @xmath171 , corresponding to the wire length @xmath172 . for completeness , in fig .
[ img : capacity - chi ] we present the phase dependence of the intrinsic capacitance coefficient @xmath173 in the long - wire limit . in the vicinity of @xmath174 , where the gap closes , the intrinsic capacitance diverges as @xmath175 a similar behavior obtained for short diffusive junctions in ref .
is attributed to the presence of almost open channels ( @xmath176 ) in dorokhov distribution ( [ dorokhov ] ) .
decay of the dissipationless supercurrent state governed by the action ( [ our_effective_action ] ) is equivalent to a quantum - mechanical tunneling of a massive particle under a barrier . near the criticality , at @xmath34 , the potential barrier @xmath177 can be approximated by a cubic parabola : @xmath178 , \ ] ] where @xmath179 is defined in eq .
( [ ic ] ) , and @xmath180 is the dimensionless curvature of the current - phase characteristic at @xmath7 . parameters @xmath179 and @xmath181 for some model josephson junctions are listed in table [ tbl : from_multicharge ] .
the instanton trajectory for the potential ( [ cubic ] ) takes the form @xmath182 where @xmath183 is the plasma frequency describing small phase oscillations near the minimum of @xmath184 . in the short - wire limit
, @xmath183 was calculated in ref . , and the general expression is given by @xmath185 the life time of the subcritical current state in the wkb approximation can be estimated as : @xmath186 where @xmath187 is the instanton action , @xmath188 this expression generalizes the result for the tunnel limit@xcite to the case of an arbitrary josephson junction .
equations ( [ attempt - freq])([instanton_action ] ) supplemented by the expressions ( [ c = c+c ] ) and ( [ c - general ] ) provide full description of the supercurrent decay due to quantum fluctuations .
the numerical coefficients @xmath179 , @xmath181 and @xmath49 for a number of model josephson junctions are summarized in table [ tbl : from_multicharge ] .
if the geometric capacitance is sufficiently small , @xmath189 , the instanton action ( [ instanton_action ] ) can be simplified .
substituting @xmath69 from eq .
( [ c - general ] ) , we arrive at eq .
( [ s - gamma ] ) , with the model - dependent parameter @xmath190 for some model cases the coefficient @xmath191 is given by eq .
( [ gamma - res ] ) .
in our analysis we rely on the adiabatic approximation for @xmath97 justified at @xmath30 . in this limit ,
variation of the phase is weak and small , that allows to expand the action . on the other hand , in the very vicinity of @xmath6 , the instanton action @xmath192 and the wkb method fails .
therefore the present theory is applicable as long as @xmath193^{2/5 } \ll 1 - i / i_c \ll 1 $ ] .
if the geometric capacitance is negligible that reduces to @xmath194 finally , we discuss the condition on the temperature range when the @xmath96 description of the phase tunneling is applicable .
the principal limitation is related to the crossover to thermal decay with the activation exponent @xmath195 , where @xmath196 is the height of the free energy barrier .
quantum description is applicable as long as @xmath197 , which translates into the constraint @xmath198 ^ 2 \ll 1 - i / i_c $ ] . in the important limit of a small geometric capacitance
, this condition reduces to a simple inequality @xmath199 the condition ( [ constraint_for_t ] ) also guarantees that the phase action can still be written in a capacitive form ( [ our_effective_action ] ) neglecting quasiparticle damping effects . indeed ,
dissipative terms in the action can be roughly described by a large shunting resistance @xmath200 , which has no effect on phase tunneling since the corresponding @xmath201 time is much larger than the period of plasma oscillations . to estimate the predictions of our theory
consider a 1@xmath202m - long sns junction similar to that fabricated in refs . . with the parameters @xmath203 , @xmath204mk , and @xmath205 ,
the intrinsic capacitance is expected to be of the order of @xmath206pf .
the condition ( [ constraint_for_i ] ) determines a wide range of allowed bias currents : @xmath207 , in which the predicted life time ( when the intrinsic capacitance dominates ) varies from nanoseconds to practically unlimited values .
for example , the decay time @xmath208s is achieved at @xmath209 . according to the constraint ( [ constraint_for_t ] ) , the quantum tunneling regime is realized in the sub-100mk temperature range .
our calculation was performed in the assumption of ideally transparent sn interfaces , whereas experimentally studied junctions may have a finite conductance @xmath210 of interfaces .
our theory is valid as long as @xmath211 , but it can be straightforwardly extended beyond this limit by adding the corresponding boundary terms@xcite to the sigma - model action ( [ sigma_model_action ] ) . in the opposite limit , @xmath212 , the normal region can be treated as a chaotic quantum dot as it was done in ref . .
in the present paper we have found the life time of a slightly subcritical dissipationless supercurrent state in an sns josephson junction due to quantum fluctuations . at low temperatures ,
the superconducting phase difference across the junction behaves as a massive quantum - mechanical variable , with the mass determined by the sum of the geometric ( @xmath36 ) and intrinsic ( @xmath40 ) capacitances of the junction .
while the former describes charging effects , the latter is associated with the dynamics of the andreev bound states .
we obtain that , generically , the intrinsic capacitance is expressed through the normal - state conductance @xmath1 and the spectral minigap @xmath2 induced in the normal part of the junction by the relation @xmath213 , where @xmath214 is a model - dependent coefficient .
an analogous expression for @xmath40 in the long - wire limit was conjectured in ref . , based on the requirement that the @xmath215 time of the junction is governed by @xmath216 .
we determine the model - dependent numerical factor @xmath49 for two classes of josephson junctions : an arbitrary short scatterer described by the set of its transmission eigenvalues ( see table [ tbl : from_multicharge ] ) and a diffusive metallic wire of arbitrary length ( see fig .
[ img : capacity - eg ] ) . in the former case
we rely on galaktionov - zaikin expression ( [ galaktionov_zaikin_formula ] ) , rederived in appendix [ app : derivation_of_k_for_short_junctions ] . in the latter case
we employ the nonlinear sigma model formalism . depending on a particular junction
, the capacitance coefficient @xmath49 can vary by a factor of three that may have a pronounced effect on the quantum tunneling rate due to its exponential sensitivity to the junction parameters . for long junctions with negligible geometric capacitance
, the decay rate is determined only by @xmath1 and proximity to criticality , as given by eq .
( [ s - gamma ] )
. measuring the coefficient @xmath191 in that relation and comparing it with the results ( [ gamma - res ] ) may provide an independent tool to determine the type of the josephson junction .
our analysis of the quantum decay of the supercurrent is limited to low temperatures , @xmath32 . for higher temperatures , finite population of quasiparticle states
leads to dissipative phase dynamics , and the problem should be treated in the spirit of ref .
, with an additional complications due to inapplicability of the simple rsj model , as well as possible issues on inelastic relaxation and thermalization in the normal part of the junction.@xcite capacitive phase dynamics is realized only in the nondissipative supercurrent state near the criticality .
once it is destroyed by fluctuations , the junction switches to the resistive branch and the phase dynamics becomes dissipative ( like for high temperatures ) .
though we can not access this regime , our results can be used for a qualitative description of the hysteretic behavior observed experimentally in lateral junctions .
@xcite assuming that an oversimpified rsj model can qualitatively describe the sns junction with the choice of the resistance @xmath217 and capacitance @xmath218 , one finds that the mccumber parameter @xmath219 ( which is the square of the quality factor @xmath220 ) is generally of the order of one [ formal substitution @xmath221 and @xmath222 yields @xmath223 , which is larger than 6 for long wires ] . in this model , large mccumber parameter , @xmath224 ,
is required for a hysteretic behavior .
though for long junctions the geometric capacitance is too small to explain the hysteresis , an account of the intrinsic capacitance provides sufficiently large @xmath225 necessary for observing a hysteretic behavior.@xcite recently it was demonstrated that low retrapping current is a consequence of electron overheating in the normal region , when the electron temperature can be several times larger than the bath temperature.@xcite the retrapping current is then identified with the critical current at the elevated electron temperature , explaining large hysteresis . by contrast , we would like to emphasize that the intrinsic junction capacitance due to dynamics of the andreev bound states may itself lead to a hysteretic behavior , even for a perfect thermal contact with the environment .
we are grateful to j. p. pekola , v. v. ryazanov and k. s. tikhonov for stimulating discussions .
this work was partially supported by rfbr grant no .
13 - 02 - 01389 .
in this appendix we rederive galaktionov - zaikin formula ( [ galaktionov_zaikin_formula ] ) in the matsubara formalism .
an arbitrary scatterer with an energy - independent scattering matrix ( short - wire limit ) sandwiched between two terminals can be described by the action@xcite @xmath226 \right\rangle,\end{gathered}\ ] ] where @xmath227 and @xmath228 are the quasiclassical green functions in the leads .
@xmath229 is a function of two energy arguments and acts as a matrix in nambu - gorkov space ( n ) . when multiplying @xmath76-matrices and taking trace , integration over @xmath230 and @xmath231 is assumed to be done .
the time representation is defined in the conventional way as @xmath232 in a stationary uniform superconductor , @xmath233 depends only on the time difference , @xmath234 , and hence is diagonal in the energy representation , @xmath235 , where @xmath236 and @xmath237 is superconducting phase .
now we proceed to the derivation of the capacitive term in the action ( [ our_effective_action ] ) . we assume zero superconducting phase on the left lead and a time - dependent phase @xmath97 on the right lead ( see fig .
[ img : sns ] ) .
we assume that @xmath238 , where @xmath128 is a small and slow function of matsubara time @xmath239 : [ qlr ] @xmath240 where @xmath241 is the pauli matrix in the n space . substituting these expressions into eq .
( [ short_junction_action ] ) we obtain the action for the phase difference @xmath97 . in the limit @xmath34 , the free energy barrier protecting the supercurrent state @xmath12 is small , and one can expand the action in powers of @xmath242 . the quadratic term can be written in the form @xmath243 with the kernel @xmath244 calculaed below .
also , the same limit guarantees that phase dynamics is slow and thus can be described ( at @xmath96 ) by the term @xmath245 in eq .
( [ our_effective_action ] ) , with the intrinsic capacitance of the junction given by @xmath246 here we calculate the kernel @xmath248 starting with the action ( [ short_junction_action ] ) for an arbitrary distribution of @xmath59 ( for brevity we omit brackets that denote averaging over @xmath59 ) .
we write @xmath249 as @xmath250 where @xmath251 corresponds to the stationary phase @xmath252
. then we insert expressions ( [ qlr ] ) into eq .
( [ short_junction_action ] ) and expand the action in powers of @xmath253 .
the first variation of the action with respect to @xmath120 reads : @xmath254 = - \frac{i}{2 } \sum_{\alpha } \sum_{k=0}^{\infty } \left(\frac{t_{\alpha}}{4}\right)^{k+1 } { \mathop{\rm tr}}\delta\tilde{q}^{2k+1 } [ \tau_{z}\hat{x},\tilde{q}_{r}],\ ] ] where we denote @xmath255 and @xmath256 . in the following expression for the second variation
we will omit for simplicity terms that contain two @xmath257 operators without any @xmath76-matrix in between , since such terms will give no contribution to the intrinsic capacitance . also after taking the variation , we put @xmath258 as we are interested in the second variation near the stationary solution .
the result can be written as @xmath259 \delta q^{2k - l}\tau_{z}\hat{x}q_{r}+\delta q^{2k}\tau_{z}\hat{x}q_{r}\tau_{z}\hat{x}q_{r}+\delta q^{2k+1}\tau_{z}\hat{x}q_{r}\tau_{z}\hat{x } \\
-\sum_{l=1}^{2k}\delta q^{l } [ \tau_{z}\hat{x},q_{r } ] \delta q^{2k - l}q_{r}\tau_{z}\hat{x}+\delta q^{2k+1}\tau_{z}\hat{x}q_{r}\tau_{z}\hat{x}+\delta q^{2k}q_{r}\tau_{z}\hat{x}q_{r}\tau_{z}\hat{x } \bigg\ } , \end{gathered}\ ] ] where @xmath260 is diagonal in the energy representation : @xmath261 switching to the energy representation , we arrive at the action ( [ quadratic_terms_in_the_expansion_of_multicharge ] ) with the kernel @xmath262 \\
+ \sum_{l=1,3,\dots}^{2k-1 } x_{2k , l } \left [ 2 { \mathop{\rm tr}}\tau_{z}q_{r}(\epsilon)q\tau_{z}q_{r}(\epsilon')q - { \mathop{\rm tr}}\tau_{z}q_{r}(\epsilon)q q_{r}(\epsilon)\tau_{z}q - { \mathop{\rm tr}}\tau_{z}q_{r}(\epsilon')q q_{r}(\epsilon')\tau_{z}q\right ] \\
+ x_{2k,0 } + x_{2k,2k } + x_{2k+1,0 } { \mathop{\rm tr}}q\tau_{z}q_{r}(\epsilon)\tau_{z } + x_{2k+1,2k+1 } { \mathop{\rm tr}}q\tau_{z}q_{r}(\epsilon')\tau_{z } \bigg\},\end{gathered}\ ] ] where @xmath263 , and @xmath264^{n}}{\left(\epsilon^{2}+\delta^{2}\right)^{l/2}\left(\epsilon'^{2}+\delta^{2}\right)^{(n - l)/2 } } .\ ] ] evaluating traces one gets : @xmath265 performing summation over @xmath266 and @xmath267 one obtains for the kernel @xmath268 : @xmath269 + \left\ { \epsilon \leftrightarrow \epsilon ' \right\}.\end{gathered}\ ] ] finally , expanding this expression to the second order in @xmath150 and integrating over @xmath157 we arrive at eq .
( [ galaktionov_zaikin_formula ] ) obtained by galaktionov and zaikin .
the rotation by the matrix @xmath113 in eq .
( [ w - cd ] ) is introduced in order to get rid of the first derivative in the bulk action @xmath270 .
its matrix elements are given by @xcite @xmath271/2 } , \ ] ] where @xmath115 are pauli matrices in the @xmath272 space , and @xmath273 after such a rotation , @xmath270 acquires the form ( [ expression_for_s_ww_bulk ] ) , with the matrix potential @xmath274 given by ( here and below we omit the spacial coordinate @xmath78 for brevity)@xcite @xmath275 where @xmath276/4 \\ { } + \varepsilon_2 \cos \theta_{\epsilon_2 } - \left [ \theta_{\epsilon_2}'^2 + ( \sin \theta_{\epsilon_2 } \varphi_{\epsilon_2 } ' ) ^2 \right]/4,\end{gathered}\ ] ] and @xmath277 and the function @xmath278 is given by ( as usual , @xmath80 ) @xmath279 ^ 2 } .\ ] ]
in this appendix we show that for finding the capacitive effective action ( [ our_effective_action ] ) it is sufficient to use the linear equation ( [ equations_of_motion_for_b ] ) [ resulting from varying only @xmath133 ] and neglect higher order terms in eq .
( [ expansion_of_action_in_w ] ) .
the only possible extra contribution to the quadratic - in-@xmath120 part of the action can originate from the influence of @xmath131 on the equation of motion for @xmath132 arising from cubic nonlinearity in the action ( [ expansion_of_action_in_w ] ) . that would induce a correction @xmath281 to @xmath282 which satisfies eq .
( [ equations_of_motion_for_b ] ) with a nonzero right - hand side : @xmath283 where the indices @xmath284 , @xmath285 , and @xmath267 refer to the @xmath272 space , and the operator @xmath286 may contain derivatives with respect to @xmath78 .
the crucial point is that the boundary conditions for @xmath281 are trivial : @xmath287 since they are already satisfied by @xmath282 [ see eq .
( [ bc_xi_x2 ] ) ] .
when substituted into @xmath119 , @xmath281 could have produced a correction to @xmath288 .
but since @xmath119 [ eq . ( [ expression_for_s_w ] ) ] is proportional to @xmath116 at the boundary , condition ( [ boundary_condition_for_correction_to_b_from_higher_orders ] ) guarantees the absence of corrections to @xmath288 . | a nondissipative supercurrent state of a josephson junction is metastable with respect to the formation of a finite - resistance state .
this transition is driven by fluctuations , thermal at high temperatures and quantum at low temperatures .
we evaluate the life time of such a state due to quantum fluctuations in the limit when the supercurrent is approaching the critical current .
the decay probability is determined by the instanton action for the superconducting phase difference across the junction . at low temperatures , dynamics of the phase
is massive and is determined by the effective capacitance , which is a sum of the geometric and intrinsic capacitance of the junction .
we model the central part of the josephson junction either by an arbitrary short mesoscopic conductor described by the set of its transmission coefficients , or by a diffusive wire of an arbitrary length .
the intrinsic capacitance can generally be estimated as @xmath0 , where @xmath1 is the normal - state conductance of the junction and @xmath2 is the proximity minigap in its normal part .
the obtained capacitance is sufficiently large to qualitatively explain hysteretic behavior of the current - voltage characteristic even in the absence of overheating . |
since the discovery of graphene @xcite a great progress has been made in production of high quality samples with nearly ballistic behavior of dirac electrons @xcite . even though the experimental results agree well with the ballistic theory of transport @xcite , the role of metallic contacts in electronic properties of sub - micron graphene samples
is debated @xcite . in this paper
we introduce a notion of metallic proximity effect in graphene and show that this effect can lead to a non - uniform gating of small graphene flakes . despite the intense graphene studies , the scattering mechanisms , which limit electron mobility in the best available graphene flakes , are not clearly identified @xcite .
still , the experimental evidence that strong impurities , such as adatoms and vacancies , dominate the scattering at low temperatures is currently building up @xcite .
we demonstrate below that even the point - like impurities may introduce large spacial variations of electronic density at distances of the order of the system size from the impurity provided the quasiparticle energies are close to the dirac degeneracy point .
an impurity can be characterized by at least two length scales : its physical size @xmath0 and the characteristic scattering length @xmath1 .
the latter is determined by the scattering cross - section and , therefore , depends on the energy of dirac quasiparticles that scatter on the impurity .
a single weak impurity , i.e. the impurity with @xmath2 , has practically no effect on transport and electronic density of states in ballistic sample .
the effect becomes essential only for a substantial concentration of weak impurities @xcite .
in contrast , a single strong impurity , i.e. the one with @xmath3 , can have a noticeable effect on transport @xcite .
the strong impurities are of a particular importance at zero doping , i.e. in the situation when the chemical potential in graphene is tuned precisely to the dirac point .
it is theoretically predicted that each strong impurity can form a resonant `` mid - gap '' state @xcite , which contributes a universal value to the conductance and shot noise @xcite provided the wave - length of the dirac quasiparticle , @xmath4 , exceeds the distance between the metallic leads , @xmath5 .
the resonant conditions are equivalent to the limit @xmath6 .
evidently , in this regime , the properties of graphene flakes are strongly influenced by both the impurities and the sample boundaries .
the values of conductance and shot noise observed in experiments with sub - micron graphene flakes @xcite agree with the predictions of ballistic theory @xcite for @xmath7 , which makes us to believe that this limit is experimentally achievable .
both the ballistic theory of ref .
@xcite and experiment @xcite deal with rectangular graphene samples with large aspect ratio , @xmath8 , where @xmath9 is the sample width and @xmath5 is the distance between the metallic leads .
for @xmath7 the proximity to metal gives rise to the so - called pseudo - diffusive charge transport @xcite in the model with abrupt boundary conditions at graphene - metal boundaries
. we shall demonstrate below that in this model the local density of states ( ldos ) takes on the most simple form , @xmath10^{-1}$ ] , which corresponds to the lowest curve in figure [ fig : clean ] .
the enhancement of the density near the graphene - metal interfaces illustrates the metallic proximity effect .
the ldos in the presence of a single impurity is shown in figure [ fig : impurity ] with a contour plot for resonance and off - resonance conditions .
such a strong spacial dependence of the ldos suggests that , in the presence of electron - electron interactions , the sample is not uniformly charged by the gate electrode .
the ldos is defined by the general formula , @xmath11 , where the retarded green s function yields the matrix equation @xmath12 with @xmath13 . in this paper the dirac hamiltonian of ideal graphene , @xmath14 ,
is taken in the valley symmetric representation , where @xmath15 m/s is an effective `` speed of light '' .
the impurity potential , @xmath16 , is considered to be a scalar in the dirac equation , which assumes that the corresponding physical potential is smooth on the scale of the graphene lattice .
the vector @xmath17 is a two - component vector of pauli matrices and the energy , @xmath18 , determines the wave - length of free dirac quasiparticle as @xmath19 . following earlier works we describe the metal - graphene interfaces at @xmath20 with a step function model @xmath21 where @xmath22 is regarded as the fermi wave length in the metallic lead .
we naturally assume that @xmath23 . in order to avoid a detailed discussion of the sample edges we impose periodic boundary conditions in @xmath24 direction .
we start with a solution of equation ( [ green ] ) in the absence of impurity potential .
this solution is most easily obtained in the channel representation , which is defined by the fourier transform , @xmath25 where the summation runs over the discrete values of transversal momentum , @xmath26 . selecting the solutions which decay in the leads we obtain the effective boundary conditions on graphene - metal interfaces in the limit @xmath23 @xcite , @xmath27
using these boundary conditions the green s function inside the clean sample is readily found as @xmath28 where @xmath29 , @xmath30 and @xmath31 in equation ( [ gr0 ] ) the following definitions were used , @xmath32 where @xmath33 . the quantity , @xmath34 , introduced in equation ( [ defs ] ) has a meaning of @xmath35 component of the quasiparticle momentum , which can take on complex values . thus , the ldos of a clean system is given by the expression @xmath36 where the factor @xmath37 takes into account the spin and valley degeneracies .
the ldos of the clean setup has no dependence on @xmath24 due to the translational symmetry of the system in this direction .
the @xmath35-dependence of the ldos is plotted in figure [ fig : clean ] for @xmath38 and several values of energy .
the expression ( [ dens_clean ] ) can be approximated in the doped graphene , @xmath39 , by averaging over the friedel oscillations on the scale @xmath4 .
away from the boundaries , @xmath40 , the evanescent modes , which are characterized by complex momenta @xmath34 , can be disregarded and the averaging over the friedel oscillations can be replaced by the averaging over the phase @xmath41 . for @xmath42 ,
the averaged ldos value is given by @xmath43 , which is shown in figure [ fig : clean ] with dotted lines .
near the sample edge , @xmath44 , the density is dominated by the metallic proximity effect , so that it diverges at the metal - graphene interfaces in the limit @xmath45 . for @xmath46 and @xmath42 ,
the ldos does not depend on energy and can be approximated as @xmath47^{-1}$ ] .
in contrast , for @xmath7 , the ldos is fully determined by the evanescent modes .
in particular , for @xmath48 ( @xmath49 ) , the result can be obtained by taking @xmath50 , @xmath51 , and @xmath52 in equation ( [ dens_clean ] ) , which leads to the expression @xmath53 in the limit @xmath42 , the sum over the transversal momenta , @xmath54 , can be replaced by the integration , with the result @xmath55 where the small term @xmath56 in the denominator is effective only at the edges , @xmath57 and @xmath58 , in which case the sum ( [ rho0general ] ) equals the number of channels in the lead .
the result ( [ refldos ] ) shows that the ldos at @xmath48 is dominated , in the limit @xmath42 , by the states originating in the leads .
these boundary states , which are also referred to as the evanescent modes , are responsible for the so - called pseudo - diffusive transport regime in this setup @xcite , which is characterized by universal full counting statistics @xcite .
we see , therefore , that the strong metallic proximity effect is the natural property of the abrupt interface model ( [ leads ] ) , which is compatible with the transport measurements of ref .
in this section we consider the effect of a scalar point - like impurity on the ldos at @xmath48 for a rectangular graphene sample with @xmath42 .
the calculation of the green s function is analogous to the one presented in ref .
@xcite for the charge transport and can be carried out straightforwardly .
we shall start with the real space representation of the free green s function ( [ gr0 ] ) at @xmath48 , that reduces , in the limit @xmath42 , to @xcite @xmath59 where @xmath60^{-1}.\ ] ] one can see from the equations ( [ real],[c ] ) that the free green s function obeys the chiral symmetry @xmath61{^\dagger}=-\sigma_z
g^r_0({\boldsymbol}{r},{\boldsymbol}{r}')\sigma_z = g^{a}_{0}({\boldsymbol}{r},{\boldsymbol}{r}'),\ ] ] which is , therefore , preserved in the abrupt interface model described by equation ( [ leads ] ) .
one can also see from equation ( [ real ] ) that the green s function at the coinciding arguments can be decomposed into the sum of the free green s function in an infinite system , @xmath62 , and the proximity - induced part , @xmath63 , such that @xmath64 where @xmath62 is off - diagonal and singular , @xmath65 while @xmath66 is proportional to the product of the ldos ( [ refldos ] ) and the unit matrix , @xmath67 in order to simplify the calculation we choose the impurity potential in the form of a disk with sharp boundaries , @xmath68 , where the theta function , @xmath69 , equals @xmath70 for @xmath71 and zero otherwise .
we also take the limit of point - like impurity , @xmath72 . in this limit
the full t - matrix of the impurity can be written , in the symbolic notations , as @xmath73 where we used the decomposition ( [ decompose ] ) and introduced the reduced t - matrix , @xmath74^{-1}$ ] , which is represented , in the detailed notations , by the infinite sum @xmath75 the reduced t - matrix given by equation ( [ tinf ] ) does not contain any information on the metallic leads and sample boundaries .
therefore , it can be obtained from the consideration of scattering of free dirac quasiparticles on a single disk impurity .
this problem has been addressed by several authors @xcite .
the result in the s - wave approximation and in the limit @xmath76 takes the form @xmath77 where @xmath78 is the bessel function and the t - matrix , @xmath79 , is proportional to the unit matrix in spin , valley , and sublattice space . the equation ( [ tresult ] )
completes the calculation of the green s function in the presence of a single impurity , @xmath80 where @xmath81 specifies the impurity location .
we stress that the characteristic impurity size , @xmath1 , determined by the reduced t - matrix , strongly depends on the impurity potential , @xmath82 , and diverges at the resonance . from equation ( [ gfinal ] )
we find the ldos at the point @xmath83 , which can be written as @xmath84 ^ 2}.\ ] ] this result is plotted in the figure [ fig : impurity ] for an impurity placed in the middle of the sample , @xmath85 . for the resonant conditions , @xmath86 , ( i.e. for @xmath87 ) the disk impurity supports a quasi - bound state at @xmath48 @xcite with only one pseudo - spin component being localized .
in this case , the ldos is strongly enhanced at distances of the order of @xmath5 from the impurity . away from the resonant condition , the effect of a single impurity on the ldos is predictably small .
it was argued on the basis of the dft analysis @xcite that some single - atom impurities such as hydrogen naturally create the mid - gap quasi - bound states .
the resonance condition for such scatterers are always fulfilled near the dirac point .
therefore , even few impurities in the sample will have a great impact on the ldos .
we expect that this effect can be observed with the stm measurements .
finally we note that the denominator in the second term of equation ( [ final ] ) coincides with the denominator in the impurity contribution to the conductance in the same setup @xcite . near the resonance , the second term in equation ( [ final ] )
has a lorentzian dependence on the position of the impurity level with respect to the dirac point .
this characteristic dependence reflects the resonant nature of charge transport that originates in the impurity assisted tunneling in undoped graphene @xcite .
we found from equation ( [ dens_clean ] ) that the ldos in ballistic rectangular sample with @xmath42 can be approximated as @xmath88 where @xmath89 is the distance to the closest graphene - metal interface .
this result shows that the metal proximity effect is fully dominating the ldos for energies close to the dirac point , @xmath90 .
it also demonstrates that the ldos is never vanishing in graphene with metal boundaries . from equation ( [ rough ] )
we readily estimate the integrated density of states as @xmath91 where @xmath92 .
thus , for @xmath90 , the integrated density increases logarithmically with the system size , @xmath5 , due to the electronic states at the metal - graphene interfaces .
peculiar transport properties of ballistic graphene @xcite are originating in the metallic proximity effect .
similarly , the presence of the resonant impurity has a great effect on the ldos near the dirac point since the corresponding electronic states localize not only on the metal interfaces but also on the impurity .
this effect leads to a highly inhomogeneous electronic density profile for @xmath90 , which is illustrated in figure [ fig : impurity ] .
this result implies that electron - electron interactions may prevent the uniform gating of the sample in the vicinity of the dirac point . assuming that the interaction potential is given by @xmath93 we can estimate , up to a constant value , an additional potential profile in the single - electron problem ( [ green ] ) as @xmath94 where @xmath95 is the fermi distribution function . since only the spatially varying part of the potential @xmath96 is relevant
, we can let @xmath97 .
finally the ldos and the potential profile , @xmath96 , have to be found self - consistently from equation ( [ green ] ) with @xmath98 and equation ( [ int ] ) .
the detailed analysis of the non - uniform gating is , however , beyond the scope of the present paper .
the work was supported by the dfg
center for functional nanostructures and by eurohorcs / esf euryi award ( i.v.g . ) . the discussions with alexander mirlin , pablo san - jose , henning schomerus , and alexander schuessler are gratefully acknowledged .
) in ballistic graphene sample with metal boundaries and @xmath38 .
different curves correspond to different energy values @xmath99 . in the middle of the sample ( i.e. for @xmath100 , where @xmath101 ) the ldos
is approximately given by @xmath102 ( dotted lines ) .
near the metal boundaries ( with @xmath103 ) the ldos behaves as @xmath104^{-1}$ ] , where @xmath89 . ] calculated from the equation ( [ final ] ) or out - of - resonance ( left ) and resonance ( right ) conditions .
the physical size of impurity is chosen as @xmath105 , while the effect on the ldos is determined by the effective length @xmath1 . at resonance conditions , @xmath6 , the ldos
is strongly enhanced in the large area , which is of the order of the system size @xmath5 . ] | we study the effect of resonant scatterers on the local density of states in a rectangular graphene setup with metallic leads .
we find that the density of states in a vicinity of the dirac point acquires a strong position dependence due to both metallic proximity effect and impurity scattering .
this effect may prevent uniform gating of weakly - doped samples .
we also demonstrate that even a single - atom impurity may essentially alter electronic states at low - doping on distances of the order of the sample size from the impurity . |
spectral lines from he - like ions are prominent throughout the electromagnetic spectrum .
he i is responsible for many lines in the optical and infrared spectra and they are commonly observed in nebular plasmas .
thus , the low temperature recombination spectrum ( without dielectronic recombination ) of he i has been extensively studied for over four decades ( mathis 1957 ; burgess & seaton 1960a , b ; pottasch 1961 ; robbins 1968 , 1970 ; robbins & robinson 1971 ; brocklehurst 1972 ; bhatia & underhill 1987 ; almog & netzer 1989 ; smits 1991 , 1996 ; hummer & storey 1998 ; benjamin , skillman & smits 1999 ) .
the tabulated results from these calculations have then been extensively used in the analysis of observed spectra .
alternatively , a few models for the hei system have been constructed , based on the treatment by cota ( 1987 ) , for implementation into plasma modeling codes ( e.g. cloudy by ferland 1986 ; ion by netzer 1987 ; xstar v.1 by kallman & krolik 1995 ) .
the emission from he - like ions other than he i is important in the x - ray spectra of astronomical and laboratory plasmas .
although the spectral lines from these ions that result from the n=2 levels have received much attention the treatment of recombination excitation has alway been approximated , while no detailed calculation of these ions recombination spectra has been reported .
the theory for computing the line intensities for transitions from the n=2 levels in coronal plasmas was developed by gabriel & jordan ( 1969 , 1970 , 1972 , 1973 ) .
this theory includes a treatment of dielectronic recombination contributions in addition to collisional excitation .
the next major advance in the subject was made by mewe & schrijver ( 1978a , b , c ) who introduced approximated treatments for dielectronic and radiative recombination , collisional excitation , and inner - shell ionization .
all of this was based on poorly known atomic data which they fit according to ionic nuclear charge along the whole isoelectronic series .
pradhan ( 1985 ) studied again the emission from n=2 states paying special attention to recombination contributions to the emission in non - coronal plasmas .
thus , most currently available spectra modeling codes simply extrapolate effective recombination rates for any he - like ion from the he i rates using hydrogenic scaling rules .
there are at least two other difficulties with available he - like emission models and calculations .
one is that all of them are restricted to densities less than @xmath6 3 . yet
, various kinds of astronomical x - ray sources of interest could be much denser , for example low - mass x - ray binaries ( bautista 1998 ) and accretion disks near extragalactic black holes ( nayakshin , kazanas , and kallman 2000 ) .
the other difficulty is that none of the calculations and models correctly approach lte under proper conditions .
this invalidates those results in the presence of strong radiation fields and in nearly optically thick plasmas . in the present paper
we start by presenting the atomic data used for the calculation .
then we described the calculations of level population and recombination spectra . in section 4
, we explain how the recombination - collisional - radiative models were constructed . in section 5
we compare the present models and their results with previous work .
finally , our conclusions are presented in section 6 .
in this section we describe the atomic data used in the present work .
it is important to note that all of the present calculations are carried out in ls coupling since fine - structure recombination rate coefficients are unavailable .
the ls approximation is expected to be accurate as long as @xmath7=0 radiative transitions are small and relativistic effects in the photoionization cross sections and radiative transition rates remain small . although the present ls calculations are expected to be accurate additional work is currently in progress to compute the atomic data necessary for fine structure recombination calculations .
term energies and line wavelengths are taken from the compilation of the national institute of standards and technology ( martin , sugar , and musgrove 1999 ) when available .
energies for other @xmath8 terms are from the topbase .
energies for @xmath9 terms not known experimentally are assumed as hydrogenic .
we use the photoionization cross sections for n@xmath10 and @xmath11 of fernley , taylor , & seaton ( 1987 ) calculated as part of the opacity project ( op hereafter ; seaton 1987 ) and which are available from topbase ( cunto 1993 ) .
in principle one could obtain total , radiative and dielectronic , recombination rate coefficients by means of the milne relation using cross sections that resolve all autoionizing resonances .
there are , however , some limitations in the op cross sections : ( 1 ) the cross sections are tabulated with an energy mesh too coarse to properly resolve the resonances .
in fact , the number of points in the tabulated cross sections is so limited that even the radiative recombination rates can not be computed with accuracy better than @xmath125% ; ( 2 ) the he i cross sections do not contain all contributing series of resonances , and the cross sections for other ions do not include any resonances at all ; ( 3 ) the op cross sections for he i are total cross sections , including photoionization into the n=2 states of the hydrogen target ion . however , under most conditions the bulk of the recombining ions are in the ground state , so the calculation of recombination rates should only use state specific photoionization cross sections to the ground state of the target ion . to solve these problems with the op he i cross sections we remove the resonances by interpolation between points on opposite sides of the resonances . later
, we truncate the cross sections at the energy points right before the first n=2 target threshold and extrapolate the cross sections to higher energies using the quantum defect method of peach ( 1967 ) . in doing so
, these quantum defect high energy cross sections are scaled to match the op cross sections at the crossing point . for levels with n@xmath13 and @xmath8 we use cross sections from the quantum defect method . for all levels with @xmath14 hydrogenic cross section from the method of burgess and seaton ( 1960a ) were used .
the state specific dielectronic recombination ( dr ) coefficients for n@xmath10 and @xmath15 were taken from adas ( badnell 1995 ) .
these adas rate coefficients are given in tabulated form at temperatures between @xmath16 and @xmath17k which we fit with an expression of the form @xmath18},\ ] ] for @xmath19 .
here , @xmath20 is the ionization energy of the ion and @xmath21 represents the energy of the autoionizing resonance @xmath22 .
the choice of such an expression and the values of @xmath23 and @xmath21 to fit the dr coefficients are explained is section 4.1 .
the resulting fits to the adas data are accurate to better than 1% in all cases .
the dr coefficients for @xmath24 were estimated by n@xmath25 extrapolation of the coefficients for lower n. dr coefficients for states with angular momentum , @xmath26 , greater than 4 are expected to be very small and are neglected altogether .
the topbase data provides oscillator strengths , @xmath27 , and transitions probabilities , @xmath28 , for @xmath29 and @xmath8 for he - like ions .
these transition probabilities are generally accurate to within a few per cent .
for he i more accurate data was calculated by kono and hattori ( 1984 ) .
recently , nahar and pradhan ( 1999 ) computed fine structure @xmath27-values for fe xxv .
their results when summed over fine structure agree well with the topbase data .
for @xmath24 the a - values were computed using the coulomb approximation methods described in smits ( 1991 ) .
transition probabilities for the forbidden , @xmath30 , and intercombination , @xmath31 , transitions were taken from mewe and schrijver ( 1978a ) , which are based on computations by drake ( 1969 , 1971a , 1971b ) , johnson and lin ( 1974 ) , and drake and dalgarno ( 1969 ) .
we adopt collisional data recommended by dubau ( 1994 ) and kato and nakazaki ( 1989 ) .
these reviews consider data for levels up to @xmath32 at most .
collisional rates for other @xmath33 changing transitions were calculated using the impact parameter method ( seaton 1962 ) .
rates for angular momentum changing transitions by collisions with electrons , protons , and alpha particles were computed using the impact parameter method but with the modifications suggested by hummer and storey ( 1987 ) .
collisional ionization rates for excited levels are computed by using the semi - empirical formula by sampson and zhang ( 1988 ) .
this formula is based on extensive distorted wave calculations of collisional excitation and is expected to be accurate for the he - like ions .
however , the formula neglects the contributions of collisional excitation - autoionization which , in the case of ionization from the ground state , sometimes dominate the ionization rate .
unfortunately , there is no data for excitation - ionization from excited levels .
thus we estimate that collisional excitation rates from excited states may have uncertainties of around 50% .
three body recombination rates are obtained from the collisional ionization rates by means of detailed balance .
@xmath34 were @xmath35 is the collisional ionization rate coefficient , @xmath36 and @xmath37 are the statistical weights of the recombined and recombining ions , @xmath38 is the ionization energy of the @xmath22 state .
under ionization - excitation equilibrium conditions the population of any level @xmath39 is given by @xmath40 & = \sum_{k < i } n_k a_{ki } + \sum_{l < i } n_l ( n_e c_{li}^e + n_pc_{li}^p)\\ & n^+n_e \alpha_i + n^+ n_e^2 c_{\infty i}\end{aligned}\ ] ] where @xmath41 is the radiative transition rate from level @xmath22 to level @xmath42 . @xmath43 and @xmath44 are the density per unit volume of electrons / protons and the transition rates by collisions with electrons / protons respectively
. @xmath45 is the density of the next ionization stage of the element , i.e. the h - like ionic state in this case .
@xmath46 is the rate coefficient for recombination ( radiative plus dielectronic ) to level @xmath22 . @xmath35 and @xmath47 are the collisional ionization and 3-body recombination rate coefficients respectively . in principle , the sums in this equation includes energy levels 1 through infinity .
however , in the present calculation it was found sufficient to explicitply include levels up to @xmath48 only , while cascades from higher levels were accounted for by extrapolation .
this recombination - collision - cascade problem was solved with the matrix condensation technique ( burgess and summers 1976 ) using the computer code developed by smits ( 1991 ) .
the population of all levels up to n=50 of every ion were calculated for several temperatures covering the entire range between @xmath49 and @xmath1k and nine electron densities from 10@xmath50 to @xmath2 3 .
the high density calculations required special attention to account for the effects of lowering of the continuum and suppression of dielectronic recombination .
the treatment of these high density effects is described next .
the lowering of the continuum effect is very important under the high plasma densities considered here .
this effect comes about from cutting the high-@xmath33 orbitals of ions due to particle packing , debye shielding , stark broadening , and collisional broadening . under the conditions of interest here particle packing is the most important of these mechanisms .
particle packing occurs when the mean inter - nuclear separation in the plasma is smaller than the distance from the nucleus to the high-@xmath33 ionic orbitals . by comparing the mean inter - nuclear separation in the plasma with the mean size of ionic orbitals of principal quantum number @xmath33
, one can define a continuum level as ( hahn 1997 ) : @xmath51 where @xmath52 is the nuclear charge of the ion considered and @xmath53 is the density of nuclei in the plasma in 3 .
debye shielding is the mechanism by which high-@xmath33 orbitals are shielded from the electrostatic attraction to the nucleus by a high density of free electrons . according to this mechanism the continuum level is given by ( hahn 1997 ) : @xmath54 where @xmath55 is in kelvin and @xmath52 effective charge if the ion ( i.e. @xmath56 for he - like ions ) . under a high concentration of singly charged ions in a plasma a micro - electric - field is formed which will lead to stark broadening of the atomic levels .
then , for sufficiently high @xmath33-numbers the atomic levels will merge with each other lowering the continuum . in this case
the continuum level is given by ( inglis & teller 1939 ) : @xmath57 ^{2/15}.\ ] ] for temperatures lower than @xmath58 k@xmath59 free electrons contribute to the broadening through the static stark effect .
so the density @xmath53 in the last equation should include both positive and negative charges . at higher temperatures
the electrons contribute to the broadening by means of collisions , but this is smaller than the stark effect of the same electrons at lower temperatures .
thus , only positive charges are considered in this case .
furthermore , one may define @xmath60 as the minimum of @xmath61 and @xmath62 .
then , only orbitals with @xmath63 may be treated as bound , while orbitals with @xmath64 are mixed with the continuum and must be excluded from the calculation of the recombination process .
one can see that in the case of hydrogen like oxygen ( z=8 ) , for example , under conditions of t=@xmath58 k and @xmath53= @xmath653 only levels with @xmath66 are bound , at @xmath67 3 only levels with @xmath68 are bound , and at @xmath69 3 only levels with @xmath70 are bound .
the effects of high electron densities and radiation fields on dr were studied by burgess and summers ( 1969 ) .
dr occurs by means of three basic processes : \(1 ) resonance capture , autoionization : @xmath71 $ ] ; \(2 ) stabilization : @xmath72\to x^{+(z-1)}(i , nl)+h\nu$ ] ; \(3 ) cascade : @xmath73 ; where the brackets indicate an autoionizing state . under low electron density conditions the cascade process proceeds until the @xmath74 ion reaches its ground state , so that the recombination rate is just the total rate of all stabilizations .
high electron densities may affect the reactions ( 1 ) , ( 2 ) , or ( 3 ) as follows : \(a ) the autoionizing states of the recombined ion , @xmath72 $ ] in the right hand side of ( 1 ) , could be affected by collisions with electrons . this would cause a redistribution among the different @xmath26-states , leading to an increase in the recombination rate coefficient .
\(b ) collisional deexcitation could assist the stabilization ( process ( 2 ) ) which would enhance the dr rate .
\(c ) collisional ionization could reduce the number of bound states involved in stabilization process ( @xmath75 in ( 2 ) ) . in other words , the lowering of the continuum effect , discussed in the previous section , will suppress dr in similar fashion as it reduces the total radiative recombination rate . for the range of densities considered here , @xmath76 3 , the effects ( a ) and ( b ) can be neglected .
this is because collisional transition rates , normally of the order of @xmath77 , have to compete with the much greater autoionization rates , of order @xmath78 1 .
furthermore , the net effect of high electron densities considered here is to suppress dr .
this has been calculated for some ions by various authors ( e.g. burgess and summers ( 1969 ) and summers 1972 ) .
on the other hand , jordan ( 1969 ) computed an empirical correction factors to dr as function of the maximum bound orbital of the ion before the continuum .
this approach allowed jordan to calculate ionization balance in dense plasmas in good agreement , @xmath79 , with the more elaborate computations of summers ( 1972 ) .
we adopt jordan s method to compute dr suppression factors .
the computed level populations of the he - like ions that were obtained for density vs. temperature grids allow us to construct ionization - excitation - spectra ( ies ) models for each ion .
these models are capable of simultaneously calculating the h / he - like ionization balance and the line emissivities under a variety of conditions .
our models are similar to those described for the he i case by cota ( 1987 ) , almog and netzer ( 1989 ) and benjamin , skillman , and smits ( 1999 ) .
however , several improvements have been made with respect to those models to allow for accurate treatment of high densities , optical depth effects , and convergence towards lte .
cota ( 1987 ) and almog and netzer ( 1989 ) explicitly consider a finite number of levels up to @xmath80 and adopt four fictitious levels that account for all more excited levels .
the recombination cascades from the fictitious levels to the explicit levels are accounted for by using averaged radiative transition probabilities weighted by the level population , which are assumed to be in lte .
this assumption introduces errors in the calculated spectrum , particularly at low temperatures . in the model of benjamin ( 1999 )
cascades from levels with @xmath81 to the lower are summed over to define `` indirect '' recombination rates which are fit over a temperature range .
this approach avoids the use of fictitious levels and yields higher accuracy under some conditions .
however , the method of benjamin introduces new problems when considering optical depth effects .
let us consider for example a model with indirect recombination rates computed under optically thin conditions ( case a ) . when using this model to solve the case b problem one would shut down all transitions to n=1 .
then , the total recombination obtained this way would be the total case b recombination minus the fraction of recombination into @xmath81 which cascades down to n=1 . for even greater optical depths which affect transition to n=2,3 , etc .
the error in the total recombination rate obtained by the benjamin approach will become worse .
our ionization - excitation model for he - like ions consists of all spectroscopic ls terms with n@xmath825 ( i.e. 29 levels ) , 2 so called superlevels " , a level that represents the ground state of the h - like ion , and 6 doubly excited levels ( @xmath83 and @xmath84 ) .
the superlevels are meant one for the singlets and one for the triplet states and are built to account for all @xmath85 levels . to do that we define : - recombination rates ( rr+dr and three body recombination ) , @xmath86 : @xmath87 - radiative transition probabilities , @xmath88 : @xmath89 - collisional deexcitation rates , @xmath90 : @xmath91 here
, the sums include all levels with @xmath85 , @xmath92 represents the population of the level @xmath93 which is known from the recombination calculation described in section 3 .
all of these rates for transitions involving the superlevels were computed for every pair of points in the temperature vs. density grid . from these values
the transition rates at any temperature and density of interest can be obtained with accuracy of approximately 2% by linear interpolation .
in addition , all the inverse processes to these rates are calculated by means of detailed balance .
it is important to notice that by separating the collisional and radiative cascades from the superlevels one can correctly treat the problem of high density and high optical depth where radiative decays from excited levels may be suppressed by self - absorption but cascades still occur via collisional deexcitation . the doubly excited levels in the ies models were included to account for the most important satellite lines to h - like ly@xmath0 emission .
excitation energies , line wavelengths , radiative transition probabilities and autoionizing rates were taken from vainshtein and safronova ( 1978 ) .
we also included electron impact excitation rates from bound to doubly excited levels and between doubly excited levels from the calculations of sampson , goett , and clark ( 1983 ) and goett , sampson , and clark ( 1983 ) .
explicit treatment of doubly excited levels introduces the additional practical complication that these levels will contribute to the dr to the bound levels .
thus , these dr contributions had to be subtracted from the total dr to each bound level as taken from the adas database ( badnell 1985 ) . on the other hand
, the detailed treatment of doubly excited levels tends to improve the behavior of the models at high densities .
this is because by considering collisional transitions involving these levels one accounts for the effects of redistribution of @xmath26 states and collisional assisted stabilization in the dr process ( see section 3.2 ) .
in addition , collisional excitation from bound to doubly excited levels results in excitation - autoionization contributions to the collisional ionization rates which are missing in the rates of sampson and zhang ( 1988 ) .
it is important , however , to be cautious about the predictions of satellite lines due to the large uncertainties in some of the current atomic data , particularly of the collisional rates for doubly excited levels which are based on coulomb - born - exchange collision strengths that neglect resonances .
these resonances can dominate the collisional rates for doubly excited levels .
for example , bautista ( 2000 ) recently showed that in the case of doubly excited levels of fe xvi the resonances in the collision strengths would enhance the excitation rates by up to three orders of magnitude .
furthermore , improved calculations of collision strengths and radiative and autoionization rates for he - like ions are currently underway
. one important feature of the present ies models is their ability to converge to lte under appropriate condition .
this characteristic is essential if they are to be applicable to conditions of dense plasmas and/or high radiation fields . for an ies model to properly converge to lte every transition rate must be balanced by its detailed balance inverse . in the present models the transitions rates of interest and their inverse are : rcl @xmath94&- & @xmath95@xmath96&-&@xmath97@xmath98&-&@xmath99@xmath100&-&@xmath101 & & @xmath102 the collisional and radiative transition rates
are readily balanced from the effective collision strengths , @xmath103 , and a - values .
three - body recombination and collisional ionization balance each other through eqn.(2 ) .
the balance between recombination ( rr+dr ) and photoionization is somewhat more complicated to achieve due to the fact we have only background photoionization cross sections ( i.e. without autoionizing resonances ) .
these cross sections balance only the rr component of the total recombination . to balance
dr one needs to work backwards from the dr rates to find at least the dominant autoionizing resonances .
let us consider a photoionization cross section with zero background and narrow autoionizing resonances which can be approximated by delta functions at photon energies @xmath21 , i.e. @xmath104 from this cross section and using the milne relation the recombination rate is @xmath105},\ ] ] where @xmath106 and @xmath107 are the statistical weights of the recombined and recombining levels respectively and @xmath20 is the ionization threshold energy .
this equation is equivalent to equation ( 1 ) , used to fit the adas dr rates , if one defines @xmath108 for @xmath21 in ryd and @xmath109 in mb .
the choice of the location of the autoionizing resonances comes from the realization that the most prominent resonances are those that belong to the rydberg series converging to the n=2 states of the h - like recombining ions .
these thresholds are located at energies of @xmath110 ryd above @xmath20 .
thus , the resonances are approximately @xmath111 ryd below the threshold , with @xmath112=2 , 3 , 4 , 5 , ... thus , @xmath113 in practice , we found that at most four resonances are needed to fit the adas dr rates to an accuracy better than 1% .
then , these resonances were added to the background photoionizations cross section in order to guarantee that the models converge to lte under proper conditions .
the present section reports the results of the present ies models .
we start by comparing our effective recombination coefficients with other calculations
. then we focus on the case of coronal ionization equilibrium and discuss various line ratio diagnostics for optically thin plasmas .
finally , the spectra from photoionized plasmas is discussed . figs .
1(a)-(b ) show the effective recombination rate coefficients ( direct recombination and cascades ) vs. temperature for the n=2 terms @xmath114 and @xmath115 respectively .
the rates are shown for various he - like ions ( o vii , s xv , and fe xxv ) . also in the same figures we show the rate coefficients from mewe and schrijver ( 1978a ) .
one can see from these figures that the present recombination rates differ significantly from those of mewe and schrijver .
particularly problematic is the high temperature where recombination is dominated by dr . here
the discrepancies between mewe and schrijver and the present calculation can exceed a factor of four . also problematic
are the recombination coefficients of mewe and schrijver for the @xmath116 state which are systematically overestimated by a factor of @xmath117 .
this is shown in fig .
2 ( upper - left panel ) . confirmation about the mewe and schrijver coefficients being overestimated can be obtained from the he i case . at t=@xmath118k the direct recombination rate coefficient to @xmath116
is @xmath119 @xmath4s@xmath5 ( see osterbrock 1989 and benjamin 1999 ) and our present total rate coefficient , including cascades , to this state is @xmath120 @xmath4s@xmath5 , which is 16% lower than sum of direct " and indirect " recombination rates of benjamin . on the other hand ,
the mewe and schrijver is @xmath121 @xmath4s@xmath5 which is clearly too high by a factor of 3 .
2 presents the effective recombination coefficients for n=2 terms against ionic nuclear charge ( @xmath122 ) for temperatures of @xmath123 k and low electron density .
also in this figure , are the results of mewe and schrijver ( solid curves ) and the hydrogenic extrapolation with @xmath124 from neutral helium ( dashed curves ) , i.e. @xmath125 there is good agreement between the present results and those of mewe and schrijver for @xmath126 and @xmath127 , but some differences are present in for the @xmath128 term as well as the systematic factor of 3 discrepancy for the @xmath116 term . on the other hand ,
the hydrogenic extrapolation of the rate coefficients reproduces very well the recombination rates for @xmath116 and @xmath127 but it systematically overestimates the rate coefficients for the @xmath126 and @xmath128 terms . for the latter term ,
the hydrogenic extrapolation and the results of mewe and schrijver are almost indistinguishable .
it is important to note that the @xmath127 level is responsible for the well known forbidden ( @xmath27 ) line of he - like ions which is used in density and temperature line ratio diagnostics ( see section 5.2 ) .
furthermore , the overestimation of the effective recombination rate to this level by ms yields erroneous diagnostics .
the emission lines that originate from transitions from n=2 levels to the ground state are the most intense in the spectra of he - like ions . these are three resolvable features in medium resolution spectra , i.e. the forbidden ( @xmath27 ) @xmath30 , intercombination ( @xmath22 ) @xmath31 , and resonant ( @xmath129 ) @xmath130 transitions . strictly speaking ,
the @xmath22 feature is composed of two fine structure lines , i.e. @xmath131 and @xmath132 .
historically , these lines are also referred to as @xmath133 and @xmath134 lines , and the @xmath27 and @xmath129 lines are also called @xmath52 and @xmath135
. under collisional ionization conditions line ratios among the @xmath136 , and @xmath129 features are useful density and temperature diagnostics . in photoionized plasmas , however , these diagnostics are misleading ( section 5.3 ) .
the intensity ratio @xmath137 is sensitive to electron density above a critical value .
this is because while the @xmath127 and @xmath138 states are populated at nearly constant proportions for a wide rage of conditions the population of @xmath127 can be excited to the @xmath128 state for densities such @xmath139 .
the ratio @xmath140 is particularly sensitive to temperature .
this is because at high temperatures all the lines are mostly collisionally excited , but at lower temperatures recombination dominates the population of the triplet states .
blends of satellite lines with the @xmath27 and @xmath22 lines complicate the picture as they too have a temperature dependence .
the satellite lines come from transitions of the type @xmath141 .
dielectronic recombination is the most important excitation mechanism for the majority of these lines , but for some satellites collisional inner - shell excitation is important .
the intensity ratio of dielectronic satellites to the the @xmath129 line is roughly proportional to @xmath142 .
3 shows the @xmath143 ratios vs. for several ions .
the density sensitivity of the @xmath143 ratios starts at around @xmath144 3 for o vii and around @xmath65 3 for fe xxv .
the systematic behavior of the @xmath143 and @xmath145 ratios along the helium isoelectronic sequence were explored by pradhan ( 1982 ) . under low density conditions the @xmath143 ratios exhibit a relatively weak temperature dependence .
@xmath145 ratios vs. temperature plots are shown in fig .
4 ( solid curves ) . these ratios present strong temperature dependence which makes them good temperature diagnostics . also in this plots we show the ratios when the satellite contributions are included ( dashed curves ) .
these satellite lines lie to the blue side but very close to the @xmath27 and @xmath22 lines .
thus satellites will enhance the apparent g ratio as measured in low to medium resolution spectra . as a consequence ,
neglecting the contributions of unresolved satellites in the spectra may result in underestimation of the temperature by up to one order of magnitude .
the traditional line ratio diagnostics discussed above are only applicable to coronal plasmas . on the other hand
, photoionization may be the dominant ionization mechanism in a variety of astronomical plasmas .
the nature of the ionization mechanism ( photoionization or collisional ionization ) of the observed plasma must be established before diagnostics can be performed .
4 shows the g ratio vs. temperature under photoionization conditions ( dotted curves ) .
it can be seen that observed ratio from photoionized plasmas departs considerably from that in the coronal case indicated by the solid curve .
thus , failure to recognize photoionization contributions to the emitted spectra will result in misleading diagnostics
. direct evidence of photoionized plasmas can be found in the strength of lines from the @xmath146 levels .
the strongest of these lines are : @xmath147 and @xmath148 .
table 1 compares the relative strengths of these lines to the @xmath129 line in a coronal and photoionized plasmas .
the results are given for a @xmath149k coronal plasma , near the temperature of maximum o vii abundance , and at a temperature of @xmath58k , near the thermal equilibrium temperature , for a photoionized plasma .
in general , lines from @xmath146 levels are very weak or undetectable in coronal plasmas , while photoionized plasmas do produce these lines with strengths of up to @xmath150 of the @xmath129 line .
cll @xmath130 & 1.0 & 1.0 @xmath30 & 0.57 & 3.47 @xmath31 & 0.16 & 0.98 @xmath151 & 0.002 & 0.10 @xmath148 & 0.001 & 0.27 @xmath147 & 0.014 & 0.20 under photoionization conditions a good temperature diagnostic can be obtained from the ratio @xmath152 ( p ratio hereafter ) .
this ratio is plotted against temperature in fig .
5 . this ratio conforms a reliable diagnostic without blending with satellite lines and with very high critical densities for collisional deexcitation .
we have carried out detailed calculation of recombination spectra of the he - like ions he i , c v , n vi , o vii , ne ix , mg xi , si xiii , s xv , ar xvii , ca xix , and fe xxv .
the calculations include radiative and dielectronic recombination , collisional ionization and three - body recombination , and the effects of high density lowering of the continuum .
these calculations allow us to build ionization - excitation - spectral models for fast computation of the spectra for modeling and/or spectral diagnostics .
this models include all levels up to @xmath32 and the most important satellite lines to the h - like spectra .
the ies models are applicable to densities of up to @xmath2 3 and converge to lte under proper conditions .
the results of the ies models are compared with calculations by other authors .
it is found the present effective recombination rates differ significantly from earlier estimates by mewe and schrijver ( 1978a ) .
for example , we find that the mewe and schrijver recombination rates for the @xmath116 state are overestimated by a factor of three .
the @xmath116 level decays to the ground state either two - photon continuum or collisional transitions .
thus , the discrepancy found in the recombination rates for this level will be important in trying to simultaneously model line and continuum emission .
the difference in recombination to the @xmath116 level may also affect the line emission spectrum under high density or high optical depth conditions where the @xmath116 level contributes to the population of other levels .
further , we present several line ratio diagnostics under coronal and photoionization conditions . it is shown that blends of satellite lines with the traditionally studied forbidden and intercombination lines can considerably enhance the apparent @xmath145 ratio in coronal plasmas . also , we find that the traditional @xmath143 and @xmath145 line ratio diagnostics used for coronal plasmas are misleading when the plasma is photoionized .
such photoionized plasmas can be recognized by the presence of lines arising from @xmath1533 levels in the spectra . moreover , these lines can used for reliable temperature diagnostics .
the present ies models , data files and a computer code , is available upon request to the authors .
these models can be implemented in spectra modeling codes or used as a stand alone tool for spectral diagnostics . in this sense ,
the present ies models for he - like ions and similar ones for h - like ( bautista 1998 ) are already implemented in the photoionization modeling code xstar v.2 ( kallman and bautista 2000 ) . for isoelectronic sequences other than h and he - like
xstar v.2 accounts only for direct recombination to every level , but calculations of more detailed recombination models like those described in this paper are underway .
this work has been supported in part by a grant from the smithsonian observatory for the emission line project . | we calculate the recombination spectra of the he - like ions he i , c v , n vi , o vii , ne ix , mg xi , si xiii , s xv , ar xvii , ca xix , and fe xxv .
we include the following physical processes : radiative recombination , dielectronic recombination , three - body recombination , electron impact ionization , and collisional excitation by electrons , protons , and @xmath0-particles
. the calculations also account for the effects of lowering of the continuum at high densities and high density corrections to dielectronic recombination . from the populations of all levels in the recombined ions we construct models for he - like ions for fast computation of their spectra .
every model includes 29 bound levels up to n=5 , a pair of `` superlevels '' which accounts for radiative and collisional cascades from highly excited levels , 6 doubly excited levels that account for the most important satellite lines , and a level that represents the hydrogenic recombining ion .
the models are constructed in a way that allows for proper approach to lte under appropriate conditions .
these models can simultaneously solve for the h / he - like ionization balance in photoionized or collisionally ionized plasmas and compute emission spectra including the combined effects of radiative and dielectronic recombination , collisional excitation , photoionization from excited levels , fluorescence , and line trapping .
the models can be used for any temperature between 100 and @xmath1k and electron densities of up to @xmath2 3 .
the models can be easily used within spectral modeling codes or as stand - alone tools for spectral analysis .
we present comparisons between the results of the present models and previous work .
significant differences are found between the present effective recombination rate coefficients to the @xmath3 and those of previous estimates .
later , we study various emission line ratio diagnostics under collisional ionization and photoionized conditions . 3@xmath4 1s@xmath5 |
negative energy sturmian functions ( s - f ) , introduced in the 60ties by m. rotenberg @xcite , found many useful applications , such as in the calculation of electron induced ionization collisions @xcite , @xcite , in the identification of resonances in nucleon - nucleus scattering @xcite , in the solution of a schrdinger equation with non - local potentials @xcite , for a separable representation of scattering @xmath5matrices @xcite and in the solution of three - body faddeev equations @xcite .
for the applications involving long range coulomb forces , the analytical expressions for the coulomb green s functions , initially developed by l. hosteler and r. pratt @xcite have found many applications @xcite .
sturmian functions form a complete , discrete set of eigensolutions of a sturm - liouville differential ( or integral ) equation , and form a complete set of basis functions . however
the expansion of a wave function into a set of sturmian functions in many cases does not converge well @xcite , and methods to improve the convergence , such as pad approximations , have frequently been utilized @xcite , @xcite . the atom - atom potentials that occur in atomic physics calculation often have strong repulsive cores , in addition to a long range attractive part , and for such potentials it has been difficult to obtain a reliable set of sturmian functions .
a method based on spectral expansions in terms of chebyshev functions has been recently developed @xcite that overcomes this difficulty , and hence the use of sturmian functions is again of interest .
nevertheless , for numerical calculations the expansions have to be truncated , and because of the slow convergence of sturmian expansions , an iterative method to correct for the truncation error becomes desirable .
it is the purpose of this study to introduce such a method , based on the original quasi - particle method of weinberg @xcite , and , for the purpose of testing it , apply it to the scattering of a particle from a potential as described by the lippmann - schwinger ( l - s ) integral equation . since
, the method described here should be applicable to more general kernels of an integral equation , the present study is done in anticipation of solving the more complicated two - dimensional integral equations that occur in the solution of three - body equations in configuration space @xcite , @xcite .
the expected accuracy is to be better than 6 to 7 significant figures , desirable for doing atomic physics calculations . by comparison ,
the solution of three - body equations for nuclear physics applications , done commonly in momentum space @xcite , achieve an accuracy not better than four significant figures @xcite .
the present paper examines the convergence of several iteration methods for the solution of the l - s equation describing the scattering of a particle with positive energy @xmath6 by a local potential @xmath7 in one dimension .
the insights gained from this example is to guide one for solving a more general integral equation .
the method initially suggested by weinberg @xcite , also called the `` quasi - particle '' method q - p , requires sturmian functions that are the eigenstates for the exact integral operator .
the present method , denoted as method @xmath8 differs from the q - p method in that it uses auxiliary sturmian functions that are based on an auxiliary potential @xmath9 that is not the same as the potential @xmath10 since it may be computationally easier to obtain the auxiliary sturmians rather than the sturmians for the original integral operator method @xmath11 may prove advantageous for the case of very complicated integral kernels . for the @xmath12 example , the auxiliary sturmians are used to approximate the scattering potential @xmath13 by a non - local separable representation @xmath14 in terms of @xmath15 auxiliary sturmian functions and subsequently correcting for the truncation error @xmath16 by means of iterations .
we find that the rate of convergence of the iterations does not depend significantly on the choice of @xmath17 , provided that the range of @xmath17 is sufficiently larger than the range of @xmath7 .
the sturmian functions in all these cases are calculated in configuration space for a positive energy @xmath6 , so that the asymptotic form of the approximated wave function is the same as the exact wave function . when solving for a bound - state wave function , it is customary the represent the potential by negative energy sturmians that are real , but for scattering cases , positive energy sturmians should be preferable .
an advantage of the sturmian expansion method over the fourier - grid method @xcite for positive energies is that the sturmian method emphasizes only the spatial region where the potential is non - negligible , the asymptotic part of the wave function being already incorporated into the sturmian basis , while in a fourier - grid method , the asymptotic part has to be obtained explicitly . since an accuracy of the scattering solution to @xmath18 is envisaged , the sturmian functions are also required to have the same , or better , accuracy .
our approach for solving the sturm - liouville eigenvalue equation @xcite utilizes a `` spectral '' expansion method into chebyshev polynomials @xcite , @xcite , that permits one to prescribe a desired accuracy . with this method
one can obtain sturmian eigenvalues and eigenfunctions with an accuracy of , for example , @xmath19 and because of its stability , one can incorporate the effect of long - range tails of potentials , as was demonstrated in the calculation of the bound state eigenvalue of a helium - helium dimer @xcite .
the present calculations are performed in configuration space , and can be generalized to include asymptotic coulomb functions .
contrary to what is often done at negative energies , our approach does not expand the green s functions into a separable set of sturmians , but rather expands the potential @xmath13 into such a set .
the present method to calculate sturmian functions is similar to a method introduced previously @xcite , in that it does not use a square well potential sturmian basis set in terms of which the desired sturmians are expanded , and , being based on a spectral method , is considerably more precise .
this additional feature permits a more accurate study of the iteration convergence properties . in section
ii we present the formalism that defines the sturmian functions ; section iii describes the approximate q - p methods @xmath20 and @xmath11 that extend weinberg s q - p original iterative method so that the use of sturmian functions for the integral kernel are avoided , and auxiliary sturmians are used instead ; in section iv numerical examples are presented for the case that the integral kernel is of the lippmann - schwinger type , and section v contains the summary and conclusions .
appendices a , b , c and d contain further details .
the one - dimensional schrdinger equation for the @xmath21 radial wave function @xmath22 is@xmath23 where @xmath24 is the differential operator , @xmath6 is the energy assumed positive here , @xmath7 is the scattering potential .
these quantities are in units of inverse length squared , and were transformed from their energy units by multiplication with the well known factor @xmath25 the wave number @xmath26 is related to the energy by @xmath27 the corresponding lippmann - schwinger integral equation for @xmath28 is@xmath29 the undistorted green s function is given by @xmath30 where @xmath31 if @xmath32 and @xmath33 if @xmath34 . in the present zero angular momentum
case@xmath35 according to eqs .
( [ 3 ] ) and ( [ 4 ] ) the asymptotic form of @xmath28 is@xmath36 with @xmath37 near the origin @xmath38 because both @xmath39 as well as the integral term in eq .
( [ 3 ] ) go to zero .
the sturmian functions @xmath40 , @xmath41 obey eq .
( [ 1 ] ) with @xmath7 replaced by @xmath42@xmath43 where @xmath44 is the sturmian potential chosen conveniently , that need not be equal to the scattering potential @xmath13 , and the @xmath45 s are the eigenvalues to be determined . the sturmian functions @xmath46 obey the boundary conditions@xmath47 where the constant @xmath48 is determined by the normalization of the sturmian function .
the functions @xmath46 obey the integral equation version of eq .
( [ 8 ] ) @xmath49 with the green s function given by eq .
( [ 4 ] ) for positive energies .
the @xmath50 s are eigenvalues of the operator @xmath51 , they form an infinite set with point of convergence at @xmath52 , and are related to the @xmath45 s according to @xmath53 with the choice ( [ 4 ] ) and ( [ 5 ] ) of the green s function the sturmians @xmath46 obey the boundary conditions ( [ 9 ] ) , and @xmath54 the @xmath55 s are not square integrable for positive energies or orthogonal to each other . however , they are orthogonal if one includes the potential @xmath17 as a weight function@xmath56 in the bra - ket notation above , @xmath57 is _ not _ the complex conjugate of @xmath46 , as is usually implied . the validity of ( [ 11 ] ) can be shown by replacing @xmath58 by @xmath59 in the integral ( [ 11 ] ) , and subsequently replacing @xmath60 by @xmath61 .
the result is @xmath62 , and if @xmath63 , the above identity becomes absurd unless @xmath64 because of the completeness of the sturmian functions , one has the identity@xmath65 which can also be written as @xmath66 depending on whether the delta function applies to the left or to the right of the integrand in an integral .
one can understand intuitively the properties of the @xmath55 s as follows @xcite .
as the index @xmath67 increases , the corresponding values of @xmath68 increase , and hence the potential @xmath69 increases in magnitude .
if @xmath17 is real and attractive and the real part of @xmath68 is positive , then the real part of @xmath69 becomes more attractive , and the corresponding eigenfunction @xmath46 becomes more oscillatory inside of the attractive region of the well .
so , from one @xmath67 to the subsequent one @xmath70 the eigenfunction acquires one more node inside of the well . according to flux considerations the imaginary part of @xmath71 has to be positive , i.e. , the well has to be emissive @xcite .
near the origin the flux is @xmath72 since @xmath73 however asymptotically the outgoing form of the wave function produces a positive outgoing flux .
this outgoing flux is generated by an emissive imaginary potential , exactly the opposite of the case of an optical potential , that absorbs flux .
these properties will be verified in the numerical section below . in the examples given below
, the scattering potential @xmath7 is of the morse type with a repulsive core near the origin ( @xmath74 ) , and the sturmian potential @xmath17 is either also of the morse type with the repulsive core suppressed ( @xmath75 , but of the same range as the scattering potential @xmath74 , or it is a potential of the woods - saxon type ( @xmath76 ) with a range larger than that of either @xmath74 or @xmath77 the morse potentials are formed by the combination of two exponentials@xmath78 , \label{21}\ ] ] with parameters given in table [ table1 ] [ c]|l||l|l|l| & @xmath79 & @xmath80 & @xmath81 + @xmath74 & @xmath82 & @xmath83 & @xmath84 + @xmath85 & @xmath82 & @xmath83 & @xmath52 + , and the woods - saxon potential is given by@xmath86\ } \label{21ws}\ ] ] with @xmath87 and @xmath88the resulting dependence of these potentials on @xmath89 is illustrated in fig .
[ fig1 ] . [ ptb ]
potentials.eps the spectrum of the eigenvalues @xmath68 for the two morse - type potentials , with @xmath90 is illustrated in figs .
[ fig2 ] and [ fig3 ] .
[ ptb ] lambdas.eps [ ptb ] lambdap_05.eps since the potential @xmath85 is entirely attractive , the real parts of @xmath68 are positive , while the imaginary parts are negative , in accordance with the argument given above .
these sturmian eigenfunctions are obtained by an adaptation of hartree s iterative method designed to obtain energy eigenvalues to the schrdinger equation , and described in appendix c. a list of these eigenvalues precise to @xmath91 significant figures is given in the appendix d. some of the eigenfunctions are illustrated in figs .
[ fig4 ] and [ fig5 ] , that show that at large distances where the potential becomes small , these functions become linearly dependent .
the @xmath46 used in the present examples are normalized such that asymptotically they equal the function @xmath92 , with unit coefficient .
[ ptb ] re_phi_s_1_4.eps [ ptb ] imag_phi_s_1_4.eps because potential @xmath74 has both a repulsive and an attractive part , the eigenvalues fall into two categories . in category
i the eigenvalues @xmath45 have a positive real part and a negative imaginary part , and the corresponding eigenfunctions are large mainly in the attractive regions of the potential well .
examples are given in figs .
[ fig6 ] and [ fig7 ] .
[ ptb ] re_phi_p_05_1_5.eps [ ptb ] imag_phi_p_1_5.eps in category ii the real parts of @xmath45 are negative so as to turn the repulsive piece of the potential near the origin into an attractive well , and the formerly attractive valley into a repulsive barrier .
examples of the corresponding sturmian for indices @xmath93 and two of these functions are shown in figs .
[ fig8 ] and [ fig9 ] .
[ ptb ] phi_p_4_05.eps [ ptb ] phi_p_7.eps for @xmath94 the function @xmath95 has a magnitude less than @xmath96 the sturmian for @xmath97 is similar to that for @xmath98 in that it is also large near the origin ( with an amplitude of @xmath99 and has a node near @xmath100 the functions for @xmath98 and @xmath101 are `` resonant '' in the radial region @xmath102 $ ] , while the one for @xmath103 is non - resonant . for a larger energy the effect of the barrier for the functions of class ii decreases , and the absolute value of the eigenvalues @xmath104 decreases for @xmath105 as is illustrated if fig .
[ fig11 ] .
[ ptb ] abs_eta_p_05_15.eps as will be shown in section iii the functions for which @xmath106 play an important role in the q - p method for the iterative correction of the truncation errors .
further , in the expansion of a wave function in terms of sturmians , such as described in section iv - a below , a dominator @xmath107 is likely to appear . for the values of @xmath67 for which @xmath108 , and @xmath109
, the corresponding sturmians make a resonant contribution to the expansion . in fig .
[ fig14 ] some of the @xmath110 are illustrated in the form of an argand diagram for three values of @xmath111 which shows that for @xmath103 the values of @xmath112 for all three values of @xmath26 satisfy this resonance criterion by lying close to unity .
[ ptb ] eta_p_05_15.eps
the original q - p iteration procedure was developed for the solution of the @xmath12 eq .
( [ 3 ] ) , but is couched here in terms of a more general integral kernel @xmath113 for a one dimensional integral equation @xmath114 this equation is to be solved for @xmath28 , and @xmath39 is a known `` driving '' term . the shorthand form of eq . ( [ 22 ] ) is@xmath115 one procedure , convenient if @xmath113 has a semi - separable nature , as is the case for green s functions in configuration space , is to expand the unknown function into a set of chebyshev polynomials , as is done in the iem method @xcite , and obtain an answer to high accuracy without iterations .
if however the operator is very complex and of large norm , so that a born series would not converge , then an alternative method based on expansions into sturmian functions , described by weinberg @xcite as the `` quasi - particle '' method , is as follows .
one approximates the operator @xmath113 by a separable part of rank @xmath15 , @xmath116and the remainder @xmath117 is defined as@xmath118 and iterates on the remainder . if the norm of @xmath117 is less than unity , the iterations should converge .
since the numerical complexity of performing iterations is less than solving a linear equation with a matrix of large dimension , weinberg s method can be computationally advantageous .
weinberg s method to obtain @xmath119 is to calculate the sturmians @xmath120 associated with the operator @xmath113 which satisfy the same boundary conditions as @xmath28 @xmath121 and approximate @xmath113 by a separable expansion into those sturmians @xmath120 whose eigenvalues @xmath122 , @xmath123 have a magnitude larger than unity
. iterations on the remainder will then converge , since they involve powers of @xmath122 which for @xmath124 are less than unity .
a numerical example ( example @xmath125 ) will be given for the case that @xmath126 where the potential @xmath127 of eq .
( [ 21 ] ) has a repulsive core and an attractive well . since the computational effort to obtain the eigenvectors and eigenvalues of a general operator
@xmath113 can be larger than the effort to solve eq .
( [ 22 ] ) directly , a simplified method ( method @xmath20 @xmath128 will be described below , that uses auxiliary sturmian functions .
these methods avoid the rigorous calculation of the @xmath120 by expanding the operator @xmath113 into a finite set of auxiliary sturmian functions @xmath46 , @xmath129 which satisfy eq .
( [ 10 ] ) where @xmath17 is an auxiliary potential , and which obey the same boundary conditions as the function @xmath130@xmath131 these sturmian functions can be obtained reliably and with high accuracy , using a spectral expansion method @xcite , @xcite.however , the auxiliary sturmians are not used to obtain the eigenfunctions for the operator @xmath132 as is done in other studies @xcite where the emphasis was in obtaining the scattering resonances , but rather to set up a well converging iteration scheme . in method
@xmath20 , designed to solve eq . ( [ 22 ] ) iteratively one first defines a function @xmath133 which is the solution of @xmath134 since @xmath135is given by the separable representation ( [ 25 ] ) , a solution of eq .
( [ 30 ] ) is of the form @xmath136 with the coefficients @xmath137 defined as @xmath138 these coefficients are obtained algebraically by solving the matrix equation ( [ 44 ] ) for the @xmath137@xmath139 with @xmath140 this equation is obtained @xcite by multiplying eq .
( [ 42 ] ) on the left by @xmath141 and integrating over @xmath89 from @xmath52 to @xmath142 the iterative solution of eq .
( [ 22 ] ) @xmath143 is obtained by solving @xmath144 iteratively with @xmath145 this series will converge if the norm of @xmath117 is less than unity .
the advantage of this procedure is that the functions @xmath146 can also be obtained algebraically , using the same matrix as required for the calculation of the coefficients @xmath137 , and the evaluation of @xmath147 only involves integrals over known functions , rather than the solution of linear equations , which would be more computer intensive .
the matrices are of dimension @xmath148 , where @xmath15 is of order @xmath149 or less .
numerical examples are given below .
in the approximate q - p method @xmath150 one solves the once iterated form of eq .
( [ 22a ] ) @xmath151@xmath152 the iterative treatment is similar to that of method @xmath20 , with the driving function @xmath39 replaced by @xmath153 , and the operator @xmath113 in eq .
( [ 22a ] ) replaced by @xmath154 this method lends itself to a decomposition of the matrices of @xmath155 in terms of the singular value decomposition method ( svd ) @xcite , @xcite , that takes the place of finding the eigenvalues and eigenvectors @xmath122 and @xmath120 of the operator @xmath113 , and provides a measure of the norm of @xmath156 this method is described in ref .
@xcite , but is not used for the numerical calculations described in the text .
the operator @xmath157 for all of the three numerical examples described below is @xmath158 where @xmath159 is the green s function defined in eqs .
( [ 4 ] ) and ( [ 5 ] ) , @xmath7 is the scattering potential given by @xmath74 , defined in eq .
( [ 21 ] ) and table [ table1 ] .
the integral equation being solved is the conventional lippmann - schwinger ( @xmath12 ) scattering equation , eq .
( [ 3 ] ) . for most cases
the wave number @xmath26 that defines the energy @xmath160 in eq .
( [ 2 ] ) has the value @xmath161 as illustrated in fig .
[ fig1 ] , @xmath74 has a repulsive core near the origin followed by an attractive well that decreases exponentially with distance .
the original @xmath162 method is illustrated by example @xmath163 and the approximate q - p method @xmath11 is illustrated for two sturmian potentials @xmath85 and @xmath164 the size @xmath15 of the sturmian basis is @xmath165 , @xmath149 , and @xmath166 for potentials @xmath85 , @xmath167 , and @xmath168 respectively in this case the sturmian potential is equal to the scattering potential , @xmath169 and hence the sturmian functions @xmath46 , defined by eq .
( [ 10 ] ) are the same , to within a normalization constant , as the eigenfunctions @xmath170 of the operator @xmath171 , and the eigenvalues @xmath110 are also the same as the eigenvalues @xmath172 the @xmath46 used in the present examples are normalized such that asymptotically they equal the function @xmath92 , with unit coefficient . if an eigenfunction @xmath173 is normalized differently , then it can be renormalized according to@xmath174 where @xmath175 the value of @xmath26 is @xmath176 , and @xmath177 .
some of these sturmians are illustrated in figs .
[ fig6 ] to [ fig9 ] and the absolute values of @xmath110 are displayed in fig .
[ fig11]@xmath178 the operator @xmath119 becomes @xmath179 in the present example @xmath180 , and @xmath181 .
the matrix elements in eq .
( [ 44 ] ) become diagonal , the coefficients @xmath137 , defined in eq .
( [ 44 ] ) , are given by @xmath182,$ ] and the function @xmath133 , solution of eq .
( [ 30 ] ) , is given by@xmath183 the quantity @xmath184 is defined by the asymptotic limit of @xmath185 @xmath186 and is given by @xmath187 , where @xmath188 and the iterative corrections to @xmath28 are @xmath189 as shown in the appendix a. the values of @xmath190 are shown in fig .
[ fig12 ] .
they begin to decrease nearly monotonically for @xmath191 but the rate of decrease is slow .
for example @xmath192 and @xmath193 this shows that without the additional iterations described next , the accuracy of @xmath194 given by @xmath184 alone@xmath195 would be good to only one or two significant figures .
for example , the value of @xmath184 obtained from @xmath133 is @xmath196 while the `` exact '' value of @xmath194 , obtained via the spectral integral equation method ( iem ) @xcite is @xmath197 in fig .
[ fig13 ] the improvement due to the successive weinberg iterations are displayed .
[ ptb ] ci_p_05.eps [ ptb ] iter_05.eps for both methods @xmath20 and @xmath11 the operator @xmath113 is again @xmath199 with @xmath127 , and the sturmian potential @xmath17 can be either @xmath200 eqs . (
[ 21 ] ) , or @xmath201 the latter defined in appendix @xmath202 in this section the normalization of the sturmian functions is such that@xmath203 for method @xmath20 the elements of the matrix @xmath204 , eq.@xmath205 become @xmath206 , and all other equations described in section iii apply . the separable approximation to the operator @xmath132 eq .
( [ 25 ] ) , is given by @xmath207 and as a result of the normalization ( [ 55 ] ) the @xmath208th power of @xmath135is@xmath209 where the matrix @xmath210 has the matrix elements@xmath211 equation ( [ 57 ] ) shows that if the norm of the matrix @xmath210 is larger than unity , iterations performed with @xmath212will not converge .
for this reason separating from @xmath210 a part that has a norm less than unity , and performing iterations for @xmath213 on that part , will converge .
that is the reason for solving the iterated equation ( [ 48 ] ) rather than the original eq .
( [ 3 ] ) , as will be explained further below . for more complicated integral kernels
a similar separation is also feasible with the singular value decomposition method ( svd ) @xcite but is not needed for the examples given here . in method @xmath11
one decomposes @xmath214 the square of the full operator , into two parts @xmath215 according to eq .
( [ 57 ] ) ( @xmath216 is given by @xmath217 and one @xmath218 is defined as @xmath219 in order to solve the once iterated @xmath12 equation ( [ 48 ] ) , one defines the function @xmath220 as the solution of @xmath221 given in the present case by@xmath222 where @xmath223 @xmath224is the solution of the linear equation@xmath225 the terms @xmath226 required for the subsequent iterations,@xmath227 are obtained by solving @xmath228 with @xmath229 a justification for the advantage of method @xmath11 over @xmath20 is as follows .
formally , eq . ( [ 66 ] ) can be written as @xmath230 ^{n}\mathcal{f}_{2}^{(n)}$ ] , and hence the rate of convergence of the iterations depends on the norm of the operator @xmath231 since the norm of @xmath232 is expected to be smaller than unity , the rate of convergence should be faster than the powers of the norm of @xmath233 in view of the completeness of the sturmian functions , @xmath218 is also given by@xmath234 by defining @xmath235@xmath236 and keeping in mind that @xmath237 one can show that the powers of @xmath218 are given by@xmath238 and hence the norm of the iterations should decrease with the order of the iteration @xmath208 as@xmath239 where @xmath240 is some constant .
by contrast , for method @xmath20 @xmath241 and by a reasoning similar to the one above , one expects@xmath242 these convergence estimates are supported by figs .
[ 21 ] and [ 24 ] .
[ ptb ] iter_s2_ws2_p.eps although one would expect from eq .
( [ 70 ] ) that the magnitude of the iteration error decreases as the square of the iteration order @xmath208 , ( i.e. a power @xmath243 ) the fit in fig .
[ 21 ] decreases by a larger power , @xmath244 .
nevertheless , the convergence of method @xmath20 is considerably slower than for @xmath245 as illustrated in fig . [ 24 ] , and is in accordance with eq .
( [ 72 ] ) .
[ ptb ] iter_b_b2_s2g_p_.eps the results for method @xmath11 , displayed in fig .
( [ fig21 ] ) , also show that the iterations using sturmians that are based on a potential with a longer range , @xmath168 than the potential @xmath85 converge faster .
this feature can be understood from the fact that at large distances the sturmians become linearly dependent , since asymptotically they become proportional to each other , as can be seen from figs .
[ fig4 ] to [ fig7 ] , and hence at large distances they do not form a good basis set . indeed , at short distances the convergence to the scattering wave function is much faster than at large distances , as can be seen from fig .
[ fig22 ] , obtained with sturmians for @xmath77 the results labeled @xmath246 and @xmath247 in fig .
[ fig24 ] are obtained with methods @xmath194 and @xmath248 respectively , for sturmians based on potential @xmath249 this potential , defined in eq .
( [ 21ws ] ) and further illustrated in the appendix , is identical to potential @xmath74 at large distances , but has its repulsive core decreased by a factor that decreases to zero near the origin . both potentials @xmath167 and @xmath85 decay in a similar fashion at large distances while @xmath167 has a barrier near the origin , and @xmath85 does not . the result that both sets of sturmians give nearly indistinguishable results for the iteration , as shown in fig .
[ fig24 ] , shows that the behavior of the sturmians near the origin does not significantly affect the results .
[ ptb ] error_psi_1_14_s_05.eps however , using the sturmians defined with the longer range potential @xmath76 the convergence is significantly better at large distances , as is shown in fig .
[ fig23 ] .
[ ptb ] error_psi_1_14_ws_05.eps the open circles in fig . [ 24 ] , labeled as @xmath250 were obtained with a green s function iteration method , described in appendix @xmath246 , in which potential @xmath251 is divided into @xmath252 the @xmath12 equation with potential @xmath167 is solved to produce the function @xmath185 and the corrections due to @xmath253 are obtained iteratively in a born - series manner as approximation to the exact function @xmath28 .
the asymptotic value of @xmath185 is much closer to that of @xmath28 than for method @xmath11 , but the rate of convergence of the green s function iterations is not as fast as that of method @xmath198
the main result of this paper is to present a viable iteration procedure to correct for the truncation errors of a sturmian expansion used for the solution of an integral equation .
the basic idea is to construct a separable ( and truncated ) approximation to the integral kernel of the integral equation , that leads to an algebraic solution for the approximated integral equation , denoted as @xmath254 the corrections to @xmath185 are performed iteratively , and the rate of convergence of the iteration is investigated .
the advantage of doing iterations rather than solving the integral equation numerically directly , is that the numerical complexity for obtaining the algebraic solution for @xmath185 and the subsequent iterations can be substantially less than for the direct numerical solution , especially if the kernel of the integral equation is very complicated .
an example would be the solution of a schrdinger equation with nonlocal potentials , recast into a lippmann - schwinger ( @xmath255 integral form .
the numerical examples are done in configuration space for a one - dimensional ( @xmath255 equation for which the integral kernel is of the form @xmath199 where @xmath159 is the undistorted green s function and @xmath7 the scattering potential , and the scattering energy is positive .
however the formalism can be generalized to more general integral kernels .
several methods are compared .
one is the `` classical '' quasi - particle @xcite method , that requires a basis of sturmians that are eigenfunctions of the kernel of the integral equation . for the application of this method to the scattering of a particle from a potential with a repulsive core , @xmath256
the iterations converge very fast ( denoted as method @xmath257 and to high accuracy , as illustrated in figs .
[ fig13 ] , [ 21 ] , and [ 24]@xmath178 if the sturmian eigenstates of the above mentioned integral kernel are not available , as is assumed here , then a set of auxiliary sturmian functions are introduced , and the convergence of the iterations is examined .
two methods are presented for this case . in method
@xmath20 the iterative solution of the original integral equation is examined , while in method @xmath11 the solution of a _ once iterated _ integral equation is considered .
it is predicted and also found that the iterations of method @xmath11 converge faster than for method @xmath20 , and both methods converge faster than the iterations performed with a green s function born approximation procedure , but more slowly than the original q - p method .
a key condition for obtaining an accurate approximation to the wave function at large distances is that the auxiliary sturmian functions be based on an auxiliary potential whose range is sufficiently larger than that of the scattering potential .
an alternate method for solving the once iterated ( but complicated ) integral equation , based on the singular value decomposition of a matrix , is described in appendix @xmath258 but not used in the numerical examples .
an accuracy of at least eight significant figures is obtained for all the numerical calculations performed .
this accuracy is achieved by using the spectral method for solving integral equations @xcite , both for the scattering wave function and for the sturmian functions . as a result
, previous investigations with positive energy sturmians @xcite , hampered by lack of this type of accuracy , were improved upon and expanded in the present study . a proof of eq .
( [ 39 ] ) is as follows : the operator @xmath119 , given by eq .
( [ 34 ] ) , projects any function into the space spanned by the sturmians @xmath46 , @xmath259 , while the operator @xmath117 projects any function into the sturmian space for @xmath260 .
the latter property follows from the orthogonality given by eq .
( [ 11 ] ) , and the fact that the sturmians form a complete set . hence @xmath261 eq.([27 ] ) , can also be written as @xmath262\langle \phi_{s}v\label{a15}\ ] ] this expression for @xmath117 also shows that the norm of @xmath117 is less than unity if @xmath263@xmath264 for @xmath265 in view of eqs .
( [ 30 ] ) , and ( [ a15 ] ) one finds that @xmath266 by writing@xmath267 and by by making use of eqs .
( [ 47 ] ) for @xmath268 , @xmath269 @xmath270 hence @xmath271 lies entirely in the sturmian space with @xmath272 and @xmath273 in view of eq .
( [ 47 ] ) the same argument can be made for @xmath274 @xmath275 and eq .
( [ 39 ] ) is proven for the case that @xmath276 and @xmath277
. at large distances potential @xmath167 is identical to @xmath74 , but at short distances the repulsive core is reduced by the factor @xmath278 $ ] , as given by eq .
( [ 21ws])@xmath279 this potential is illustrated by the open circles in fig .
[ fig25 ] this potential was defined for the purpose of performing a green s function iteration , as follows .
the potential @xmath74 is divided into two parts @xmath167 and @xmath280 .
the first part has a suppressed repulsive core , and the second part , @xmath281 adds the repulsive core again .
the iterations are based on green s functions that explicitly include the distortion due to @xmath167 , and iterate over the effect of @xmath281 in a fashion similar to the born approximation .
the green s function method has many applications , but in view of the factor @xmath282 in the greens functions @xmath39 and @xmath283 eq .
( [ 4 ] ) , this method does not converge well @xcite for low values of the wave number @xmath284 in what follows the eigenvalue subscripts @xmath67 will be dropped , and the iterations which lead to the solution for a particular fixed value of @xmath67 will be described . in the differential equation ( [ 8 ] )
the exact values of @xmath45 and @xmath55 are initially unknown , but the equation can still be solved for a guessed value @xmath285 .
one solutions , denoted as @xmath286 satisfies the correct boundary conditions at the origin , and is integrated from inside outward.@xmath287 the other solution , denoted as @xmath288 , satisfies the correct boundary conditions asymptotically , @xmath289 and is integrated inward towards the intermediate point @xmath290 chosen to lie within the region where @xmath17 is non negligible .
the three solutions at the point @xmath290 are denoted as @xmath291,and @xmath292 one can renormalize @xmath293 by a factor @xmath294 @xmath295 so that @xmath296 if one multiplies eq .
( [ 8 ] ) on both sides by @xmath297 and eq .
( [ 12 ] ) by @xmath55 , integrates over @xmath89 from @xmath52 to @xmath290 , and subtracts the results one from the other , after an integration by parts one obtains@xmath298 where the prime denotes a derivative with respect to @xmath299 by a similar procedure one obtains@xmath300 by adding eqs .
( [ 15 ] ) and ( [ 16 ] ) , by noting that @xmath301 because @xmath302 and @xmath55 vanish at the origin , and that @xmath303 because @xmath304 and @xmath55 obey the same boundary condition asymptotically ( to within a normalization constant ) , one obtains@xmath305
. \label{17}\ ] ] the above equation is still rigorously valid .
the approximation consists in replacing @xmath306 in the first integral by @xmath307 and by @xmath288 in the second integral , and further , by replacing @xmath308 in the left hand side by either @xmath309 or @xmath292 after making these approximations , and by dividing both sides of eq .
( [ 17 ] ) by @xmath310 one obtains the final result@xmath311 in the above , @xmath312 replaces @xmath45 as the iteratively corrected value for @xmath285 ; also the factor @xmath240 disappeared because it canceled in the numerator and denominator of eq .
( [ 18 ] ) . equation ( [ 18 ] ) is the generalization to sturmian eigenvalues of the iterative calculation of energy eigenvalues , eq .
( 7 ) of ref .
it can also be applied to negative energies , and can be generalized to the cases of coupling potentials or of non - local potentials , which however goes beyond the scope of the present study .
the functions @xmath293 and @xmath304 can be calculated as the solutions of the differential eqs .
( [ 12 ] ) and ( [ 13 ] ) by any convenient finite difference method , or they can be obtained as the solutions of the integral equations@xmath313 and@xmath314 or else the whole iteration procedure can be bypassed by obtaining the eigenvalues @xmath110 and eigenfunctions @xmath46 of the integral operator @xmath315 contained in eq .
( [ 10 ] ) . in the numerical calculations described below
, this latter procedure is carried out ( with questionable accuracy according to ref @xcite ) so as to provide the initial values @xmath316 for the iteration , eq .
( [ 18 ] ) .
the present implementation obtains the solution of eqs .
( [ 19 ] ) and ( [ 20 ] ) by using the spectral chebyshev expansion method , @xcite , @xcite because its high accuracy can be predetermined by the specification of an accuracy parameter @xmath317 , and the method is by now well tested .
[ c]|l||l|l| & real part of @xmath68 & imag . part of @xmath68 + @xmath319 & @xmath320 & @xmath321 + @xmath243 & @xmath322 & @xmath323 + @xmath324 & @xmath325 & @xmath326 + @xmath84 & @xmath327 & @xmath328 + @xmath329 & @xmath330 & @xmath331 + @xmath82 & @xmath332 & @xmath333 + @xmath334 & @xmath335 & @xmath336 + @xmath337 & @xmath338 & @xmath339 + @xmath340 & @xmath341 & @xmath342 + @xmath101 & @xmath343 & @xmath344 + @xmath91 & @xmath345 & @xmath346 + @xmath347 & @xmath348 & @xmath349 + @xmath350 & @xmath351 & @xmath352 + @xmath1 & @xmath353 & @xmath354 + @xmath355 & @xmath356 & @xmath357 + @xmath358 & @xmath359 & @xmath360 + @xmath361 & @xmath362 & @xmath363 + @xmath364 & @xmath365 & @xmath366 + @xmath367 & @xmath368 & @xmath369 + @xmath3 & @xmath370 & @xmath371 + @xmath372 & @xmath373 & @xmath374 + @xmath375 & @xmath376 & @xmath377 + m. rotenberg , ann . of phys .
( n.y . ) * 19 * , 262 ( 1962 )
; m. rotenberg , `` theory and applications of sturmian functions '' in adv . in atomic and molec .
phys . , ed by d.r .
bates and i. esterman , ( academic press , n.y .
( 1970 ) ) , vol * 6 , * p233 - 268 ; j. s. ball and d. wong , phys . rev .
* 169 * , 1362 ( 1968 ) ; g. h. rawitscher and l. canton , phys . rev .
* c 44 * , 60 ( 1991 ) ; l. canton and g. h. rawitscher , j. of phys .
g : nucl . and
* 17 * , 429 ( 1991 ) ; d. eyre , h. g. miller , phys .
b * 153 b * , 5 , ( 1985 ) ; z. c. kuruoglu and f. s. levin , annals of phys . * 163*,120 ( 1985 ) ; a. c. fonseca and m. t. pena , phys . rev . * a 38 * , 4967 ( 1998 ) ; s. yu and j. h. macek , phys . rev .
* a 55 * , 3605 ( 1997 ) ; j. m. randazzo , l. u. ancarani , g. gazaneo , a. l. frapiccini and f. d. colavecchia , phys . rev . * a 81 * , 042520 ( 2010 ) ; a. maquet , v. veniard , t. a. marian , j. phys * b 31 * , 3743 ( 1998 ) ; c. brezinski , _
pade -type approximation and general orthogonal polynomials , _ ( birkhauser , basel , 1980 ) ; c. brezinski and j. van iseghem , _ pade approximation _ , in handbook of numerical analysis ( p. g. ciarlet and j. l. lions , eds . ) , vol .
47 222 , north - holland , amsterdam , 1994 ; r. shakeshaft and x. tang , phys
. rev . * a 35 * , 3945 ( 1987 ) , r. shakeshaft , _ ibid . _ * a 37 * , 4488 ( 1988 ) ; s. weinberg , phys . rev . * 131 * , 440 ( 1963 ) ; _ ibid .
_ * 133 b * 232 ( 1964 ) ; s. weinberg and m. scaldron , _ ibid .
_ * 133 b * 1589 ( 1964 ) ; s. weinberg in _ lectures on particles and field theory , brandeis summer institute in theoretical physics _
( prentice - hall , englewood cliffs , 1964 ) vol .
2 , chaps . 4 and 5 ; w. glckle and g. rawitscher , nucl . phys . * a 790 * , 282c ( 2007 ) ; w. glckle and g. rawitscher , `` three - atom scattering via the faddeev scheme in configuration space '' , physics/0512010 at arxiv.com ; g. rawitscher and w. glckle , epj web of conferences * 3 * , 05012 ( 2010 ) ; e. o. alt , p. grassberger and w. sandhas , phys . rev . *
d 1 * , 2581 ( 1970 ) ; c. y. chen and k.t .
chung , phys .
rev * a 2 , * 1449 ( 1970 ) * ; * p. j. kramer and j. c. y. chen , phys .
rev * a 3 * , 568 ( 1971 ) ; w. glckle and r. offerman , phys . rev . * c 16 * , 2039 ( 1977 ) ; r. kosloff in _ dynamics of molecules and chemical reactions , _ edited by r.e .
wyatt and j. z. h. zhang ( marcel dekker , new york , 1966 ) ; v. kokoouline , o. dulieu , r. kosloff and f. masnou - seeuws , j. of chem .
phys . * 110 * , 9865 ( 1999 ) ; r. a. gonzales , j. eisert , i koltracht , m. neumann and g. rawitscher , j. of comput . phys . * 134 * , 134 - 149 ( 1997 ) ; r. a. gonzales , s .- y .
kang , i. koltracht and g. rawitscher , j. of comput . phys . * 153 * , 160 - 202 ( 1999 ) ; a. deloff , ann . phys .
( ny ) 322 , 13731419 ( 2007 ) ; l. n. trefethen , _ spectral methods in matlab _ , ( siam , philadelphia , pa , 2000 ) ; john p. boyd , _ chebyshev and fourier spectral methods , _
2nd revised ed .
( dover publications , mineola , ny , 2001 ) ; g. rawitscher and i. koltracht , computing sci .
* 7 * , 58 ( 2005 ) ; g. rawitscher , _ applications of a numerical spectral expansion method to problems in physics : a retrospective , _ in operator theory , advances and applications , vol .
203 , edited by thomas hempfling ( birkuser verlag , basel , 2009 ) , pp . 409426 . | years ago s. weinberg suggested the `` quasi - particle '' method ( q - p ) for iteratively solving an integral equation , based on an expansion in terms of sturmian functions that are eigenfunctions of the integral kernel .
an improvement of this method is presented that does not require knowledge of such sturmian functions , but uses simpler auxiliary sturmian functions instead .
this improved q - p method solves the integral equation iterated to second order so as to accelerate the convergence of the iterations .
numerical examples are given for the solution of the lippmann - schwinger integral equation for the scattering of a particle from a potential with a repulsive core .
an accuracy of @xmath0 is achieved after @xmath1 iterations , and @xmath2 after @xmath3 iterations .
the calculations are carried out in configuration space for positive energies with an accuracy of @xmath4 by using a spectral expansion method in terms of chebyshev polynomials .
the method can be extended to general integral kernels , and also to solving a schrdinger equation with coulomb or non - local potentials . |
since the fundamental review by paul hodge in 1971 the known dwarf population of the local group has more than doubled .
the current census of dwarf galaxies , faint objects with luminosities @xmath5 , is around 40 , including all objects that lie within or very close to the zero - velocity surface of the local group ( mateo 1998 ; van den bergh 1999 ) .
these tiny galaxies are either dwarf irregulars ( dirrs ) , dwarf ellipticals ( des ) or dwarf spheroidals ( dsphs ) ( grebel 1998 ; van den bergh 1999 ) .
dirrs ( @xmath6 ) are irregularly shaped galaxies with recent or ongoing star formation and generally quite low surface brightness ( @xmath7 mag arcsec@xmath8 ) .
the des have ellipsoidal shapes , are fairly luminous ( @xmath9 ) and typically have a central surface brightness ( @xmath10 mag arcsec@xmath8 ) higher than that of bright dirrs like the lmc or ic10 ( mateo 1998 ) .
similarly to their counterparts found in galaxy clusters ( ferguson & binggeli 1994 ; james 1991 ) some of them have blue nuclei ( bica et al .
1990 ; jones et al . 1990 ) .
they contain mainly old ( @xmath11 gyr ) and intermediate - age ( 1 - 10 gyr ) stellar populations and show in part recent star formation . the dsphs number some of the least luminous ( @xmath12 ) and lowest surface brightness galaxies known to date ( @xmath13 mag arcsec@xmath8 ) .
they do not have nuclei nor a significant central concentration and are dominated by old or intermediate - age stellar populations .
star formation in some of them has ceased a few gyr ago , while in others it has been more extended in time and episodic ( hurley - keller et al .
1998;grebel 1999 ) .
a few galaxies have properties intermediate between dirrs and dsphs and may be evolving from dirrs to dsphs ( grebel 1999 ) .
a clear progression from dirrs to dsphs is evident in their hi - to - total baryonic mass ( stars + gas ) , ranging between 7% and 50 % for dirrs , between 1% and 10% in transition systems , while being lower than 0.1 % in dsphs and des .
the local group dwarf galaxies shows a striking morphology - density relation ( grebel 1998 ) , which seems to be featured even in other nearby loose groups ( karachentsev et al .
2000 ) : dirrs are found far away from either the milky way or m31 , while the early - type dwarfs , dsphs and des , lie within 300 - 400 kpc from the center of the primary galaxies .
the distances of dsphs and des from the center of the giant spirals are sufficiently small to postulate that they are satellites of the dominant spirals , moving on bound orbits in their large dark matter halos ( peebles et al .
1989 ) .
kinematical analysis of dwarf galaxies in the local group highlights an important distinction between early - type dwarfs and dirrs , namely _ their different angular momentum content_. the stars in dsphs and des are supported by velocity dispersion ( typically @xmath14 , see mateo 1998 and bender et al .
1992 ) , while the neutral hydrogen is known to rotate in nearly all dirrs ( @xmath15 ) .
the kinematics of the stellar component is still not available for dirrs but equilibrium arguments suggest that the stars should be arranged in a rotating disk component like the gas .
only the faintest lg dirrs , like sagdig and gr8 ( @xmath16 ) , are dominated by rotation inside their half - right radius but are pressure supported at large radii ( carignan et al .
1990 ; hoffman et al .
1996 ; lo et al .
1993 ) . despite these differences ,
the surface brightness profiles of dirrs and early - type dwarfs are quite similar .
traditionally , fits with exponential profiles are preferred for dirrs while king profiles are used for dsphs and des ( ferguson & binggeli 1994 ) .
however , in most cases an exponential law also reproduces the surface brightness profiles of early - type dwarfs well ( mateo 1998 ; ferguson & binggeli 1994 ; faber & lin 1983 ; hodge et al .
1991a , b ; irwin & hatzidimitriou 1995 ) .
in addition , the luminosity grows with increasing surface brightness in dsphs and des , similarly to dirrs and spirals ( ferguson & binggeli 1994 ; binggeli & cameron 1991 ; bender et al .
. moreover , both dsphs and dirrs are dark matter dominated and show an anti - correlation between their mass - to - light ratio and their luminosity ( mateo 1998 ) : draco and ursa minor , the faintest dsphs ( @xmath17 ) , have @xmath18 ( lake 1990 ) ) and the dimmest dirrs , gr8 and sagdig ( @xmath16 ) have @xmath19 . in summary , the similarities existing between dsphs and dirrs are enough to postulate an evolutionary link between them .
for this to happen one needs a mechanism which not only removes the cold gas from dirrs but also drains internal angular momentum and heats the stellar disks .
supernovae feedback is by far the most widely quoted mechanism to expel gas from dwarf galaxies . in the model by dekel & silk ( 1986 ) supernovae can induce substantial gas removal in galaxies with a low circular velocity , @xmath20 100 km / s . for a dwarf dominated by its dark matter halo
, any gas loss would have a marginal dynamical effect on the stellar system that is left behind ( navarro et al .
1996 ; burkert & silk 1997 ; gelato & sommer - larsen 1999 ) : as a consequence , after star formation has stopped and the galaxy has faded , the `` newly born '' dsph will maintain the profile and the correlation between luminosity and surface brightness present in its former dirr state .
however , mac low and ferrara ( 1999 ) , who perform a detailed calculation of the energy balance , have shown that complete blow out of gas occurs only in very small halos , with @xmath21 km / s , while , for example , all bright des likely reside in substantially more massive halos ( mateo 1998 ) .
an alternative scenario has been proposed in which the gas is kept hot and star formation is inhibited at high - redshift by the photoionizing cosmic uv background in dwarfs with halos corresponding to @xmath22 km / s ( babul & rees 1992 ; efstathiou 1992 ; quinn et al .
1996 ; bullock et al .
2000 ) : the stellar component observed in dsphs would have formed prior to the reionization epoch , while dirrs would have appeared at low redshift , when the intensity of the background has sufficiently dropped .
however , neither of these mechanisms acts preferentially close to the primary galaxies and thus no explanation is provided for the morphology - density relation .
moreover , they offer no way to explain the different angular momentum content of the two types of dwarfs . among mechanisms related to the local environment
, the ram pressure stripping scenario has been invoked frequently for the dwarfs in the local group ( einasto et al .
1974 , faber & lin 1983 , van den bergh 1999 , blitz & robishaw 2000 ) : a hot gaseous halo surrounding the disk of the milky way or m31 would strip the gas from the shallow potential wells of dwarfs moving through ( gunn & gott 1972 ) .
sophisticated numerical simulations have shown the effectiveness of this mechanism in the inner regions of galaxy clusters like coma ( abadi et al . 1998 ; quilis et al .
however , observations indicate that the hot gaseous corona of the milky way is three order of magnitude less dense than the core of rich clusters .
a recent work by murali ( 2000 ) has shown that a coronal gas density @xmath23 @xmath24 , at @xmath25 kpc , is necessary for the magellanic stream not to be dissolved by ram - pressure driven evaporation in less than 500 myr .
halo models derived from rosat x - ray observations also yield comparable upper limits ( kalberla & kerp 1998 ) , while models in which lg dwarfs lose most of their gas due to ram pressure ( gallart et al .
2000 ; einasto et al .
1974 ) require gas densities more than one order of magnitude higher .
in addition , even if the coronal gas density were substantially larger than currently estimated ( e.g. blitz & robishaw 2000 ) , ram pressure stripping would be unable to affect the structure and kinematics of the pre - existing stellar component .
thus , although we can not exclude that ram pressure is at play , it hardly seems to be the key evolutionary driver for lg dwarfs . in galaxy clusters , fast fly - by collisions of galaxies with the largest cluster members plus a secondary contribution of the tides raised by the global potential of the cluster , are capable of transforming disk galaxies in objects resembling s0 or spheroidal galaxies ( moore et al .
1996 , 1998 , 1999 ) by removing both mass and angular momentum .
this mechanism , known as _ galaxy harassment _
affects all galaxy components and is purely gravitational . for harassment to be effective , 3 - 4 collisions per galaxy with a galaxy as massive as the milky way , namely having a circular velocity @xmath26 @xmath27 km / s , corresponding to roughly @xmath28 ( where @xmath29 is the circular velocity of a coma - sized cluster , @xmath30 km / s ) , are required .
this occurs in clusters , where the collision rate @xmath31 is around 1/gyr , as follows from @xmath32 , where n is the number density of milky way - sized galaxies ( @xmath33 mpc@xmath34 ) , in a rich coma - like cluster ( of radius @xmath35 mpc ) , @xmath36 is the velocity dispersion of the cluster ( @xmath37 km / s ) and @xmath38 is the cross section of the target galaxy ( here we are considering the extent of a typical @xmath39 galaxy out to the virial radius of the dark matter halo , @xmath40 kpc , as it will take a few gyr for a galaxy to approach the cluster core and be significantly tidally truncated ) . rescaling the calculation to the case of an lg dwarf satellite of either the milky way or m31 colliding against a fairly large satellite with @xmath41 km / s ( namely comparable to the small magellanic cloud and having @xmath42 , where @xmath43 is the circular velocity of the milky way ) one obtains only @xmath44 0.1/gyr : this is because the number of observed satellites in the halo of the milky way or m31 is much lower than the number of mw - like galaxies in clusters ( moore et al .
1999 ; klypin et al .
1999 ) . thus
, unless we are missing a substantial population of satellites , either because they are completely dark or because their baryons have not yet been turned into stars due to some cooling - halting mechanism ( blitz et al . 1999 ) , harassment is negligible in the local group . if close - by encounters between dwarfs are rare in the local group , only the tidal field of the primary halos can substantially reshape these small galaxies .
gnedin et al .
( 1999 ) found , using a generalized semi - analytical formulation based on the impulsive approximation and including adiabatic corrections , that the heating of a satellite by the tides of the primaries can be very strong when the orbits are eccentric or have pericenters close to the center of the primary .
these results are also supported by the more sophisticated time - dependent perturbation theory of gravitational shocks developed by weinberg ( 1994a , b , c ) .
typical orbits of subhalos in structure formation models are fairly eccentric , having an average apocenter to pericenter ratio @xmath45 ( ghigna et al .
1999 ) . the orbital time in the milky way
halo is roughly the same as in a coma - like cluster , as @xmath46 and @xmath47 for virialized structures ( white & frenk 1991 ) , and is @xmath48 gyr for an object on a circular orbit halfway from the center of the primary halo .
however , the milky way satellites should have performed a higher number of orbits than galaxies in virialized clusters because the latter typically appeared 5 gyr ago ( rosati et al .
2000 ) , while the galaxy was already in place at least 10 gyr ago ( van den bergh 1996 ) .
semi - analytical approaches limit the analysis to the amount of energy that can be injected by tides into a satellite , being unable to handle with the nonlinear mechanisms possibly excited by such energy input , such as gravitational instabilities in the stellar component , that can lead to substantial morphological evolution . in mayer
et al . ( 2001 ) we investigated for the first time the fate of `` tidally stirred '' dwarf irregulars orbiting the halo of the milky way using high - resolution n - body / sph simulations .
we found that the general properties of their remnants match well those of the dsphs and des in the local group , additionally providing a natural explanation for their observed morphological segregation . in this paper
we carefully describe the techniques used to construct realistic replicas of dwarf irregulars , and we study in depth their dynamical evolution and the internal structure of their remnants . in addition , by carrying out an unprecedented number of high - resolution n - body simulations ( more than 50 ) , we explore a much wider parameter space in terms of both the orbital parameters and the internal structure of the satellites , obtaining a statistical sample for a detailed comparison with observations .
finally , with our present study we aim to identify the basic conditions for tidal stirring to be effective as well as at testing its predictions within the framework of hierarchical structure formation against observations .
the simulations were performed with pkdgrav ( stadel & quinn , in preparation ) , a fast , parallel binary treecode , or , when a gas component is also included , with a newer version of the same code , gasoline , which implements an sph algorithm for solving the hydrodynamical equations and radiative cooling ( wadsley , stadel & quinn , in preparation ) .
the outline of the paper is as follows . in section 2 and 3
we illustrate , respectively , the initial dwarf galaxy models and orbital configurations . in section 4 we follow the dynamical evolution of the satellites , while section 5 is devoted to the structural properties of the remnants .
in section 6 we test the tidal stirring scenario , varying the internal structure of the satellites , the orbits and the structure of the primary halo . in section 7
we derive the star formation history of the satellites , and section 8 concerns the gross observational properties of the remnants .
finally , in section 9 we discuss the main results and draw our conclusions .
the models of dwarf irregular galaxies are constructed using the technique of hernquist ( 1993 ) , which has been extensively shown to produce stable configurations for systems with more than one component . in the present work
, galaxies normally comprise an exponential stellar disk embedded in an extended truncated isothermal dark matter halo . in some cases
a cold gas component is also present in the disk .
the profile of the dark matter halo is given by : @xmath49 where @xmath50 is the halo mass , @xmath51 serves as a cut off radius , @xmath52 is the `` core '' radius , and @xmath53 is a normalization constant defined by @xmath54 ^{-1 } \ $ ] , where @xmath55 .
the stellar disks initially follow the density profile ( in cylindrical coordinates ) : @xmath56 where @xmath57 is the disk mass , @xmath58 is the radial scale - length , and @xmath59 is the vertical scale - thickness .
the particles used to represent the gaseous disk follow roughly the same density profile of the disk stars but their vertical scale height depends on the temperature of the gas ; as an example , for a gas temperature @xmath60 k the gas layer is thinner than the stars for vertical stellar dispersions similar to those in the milky way disk .
our procedure for `` building '' dwarf galaxies involves the joint use of observations and theoretical models of structure formation .
we build two basic galaxy models , a high surface brightness ( hsb ) dwarf and a low surface brightness ( lsb ) dwarf ( hm1 and lm1 , respectively , described in section 2.1 ) .
we then construct other models by keeping fixed the mass of the halo and disk while varying the other structural parameters ( section 2.1 ) .
we also construct models with different total masses ( section 2.2 and 2.3 ) using the scaling relations for objects forming in cold dark matter cosmogonies ( e.g. mo , mao & white 1998 ) . in a scenario in which galaxies form from the cooling of gas inside dark matter halos ( white & rees 1978 , white & frenk 1991 ) ,
the baryons settle into a disk rotating in centrifugal equilibrium within the halo , and thus the rotational velocity equals the circular velocity of the halo .
therefore , once we fix the circular velocity @xmath26 of the halo , we determine the luminosity of the embedded disk according to the tully - fisher relation . for our basic models , lm1 and hm1
, we choose a halo circular velocity @xmath61 km / s .
zwaan et al .
( 1995 ) have shown that hsb and lsb disk galaxies obey the same tully - fisher ( t - f ) relation in the b - band ( although lsbs show more scatter ) , @xmath62 . a rotational velocity of @xmath63 km / s implies @xmath64 according to the t - f relation ; these parameters correspond to the brightest dwarf irregular galaxy in the lg , the lmc ( mateo 1998 ; sofue 1999 ) .
= 9truecm in order to satisfy the same t - f relation lsbs must have larger disk scale - lengths compared to equally luminous hsbs , because @xmath65 ( where @xmath66 is the surface luminosity ) .
for the disk scale length @xmath58 of the hsb we chose a value of @xmath67 kpc , while the lsb model has a disk scale length of @xmath68 kpc .
both values are in agreement with the bimodal @xmath69 relation in zwaan et al .
( 1995 ) . to fix the disk mass of both hm1 and
lm1 we assume @xmath70 , following bottema ( 1997 ) , which yields a mass of @xmath71 .
once the circular velocity is known , the virial mass @xmath72 and virial radius @xmath73 of the halo ( which we always equate to the tidal radius of the halo , @xmath51 , that appears in eq.(1 ) ) follow immediately from @xmath74 where , for the basic models , we assume @xmath75 and @xmath76 km s@xmath77 mpc@xmath77 the normalization in the above equations follows from assuming that the critical overdensity for the collapse of objects is @xmath78 , where @xmath79 is the critical density of the universe , as appropriate for a scdm cosmology . in other more currently favored models , like in lcdm ,
the value of the critical overdensity is marginally different , and thus the final sizes and masses of halos are only slightly affected ( eke et al .
according to ( 1 ) we assign to our basic dwarf galaxy models a ( halo ) virial mass @xmath80 and a virial radius @xmath81 kpc by changing the _ concentration _ @xmath82 at fixed @xmath51 and halo mass we reproduce the rotation curves of both lsb and hsb dwarfs , as done by de blok & mcgaugh ( 1997 ) for a large sample of galaxies of both types ( see also persic & salucci 1997 ) .
following again the same authors , we assume that the halo and disk scale lengths are related and set @xmath83 , thus making hsbs halos more than twice as concentrated as lsb halos given the difference in their disk scale lengths .
we also build two models in which @xmath58 and @xmath52 are not equal to further explore the parameter space .
the rotation curves of the basic hsb and lsb models ( lm1 and hm1 ) and of the latter two models ( lm1rc03 and hm1rc03 ) are shown in figure 1 .
= 8truecm = 8truecm in the models described so far the scale height of the disk is fixed at @xmath84 , as suggested by observations of late - type galaxies ( van der kruit &
de grijs 1999 , de grijs et al .
1997 ) , and we set @xmath85 at @xmath86 ; detailed numerical studies have shown that a toomre parameter @xmath87 is a necessary condition for avoiding the growth of spontaneous bars ( athanassoula & sellwood 1986 ; friedli 1999 ) .
different choices of @xmath88 and @xmath89 can determine , respectively , a different susceptibility to the growth of bars and to bending instabilities , the latter being vertical oscillations responsible , for instance , for the buckling of bars ( merritt & sellwood 1994 ; raha et al .
given the chosen parameters , the basic models have structurally normal disks . to further explore the issue of instabilities we construct a small set of models with different values of @xmath88 and @xmath89
the rotation curves are not affected by the latter variation of parameters , and they look the same as those of models lm1 and hm1 shown in figure 1 .
= 8truecm figure 2 shows the stability properties of the various models employed ( including the model of gr8 described in subsection 2.3 ) in terms of @xmath88 and @xmath90 ( @xmath91 is the epicyclic frequency , @xmath92 is the disk radius and @xmath66 is the disk surface density ) , the latter being related to the swing amplifier mechanism that can determine a rapid growth of the @xmath93 modes responsible for the bar instability ( binney & tremaine 1987 ) . in table 1
the structural parameters of the models are listed ; note that by assuming a smaller halo core radius @xmath52 , ( smaller than the disk scale length @xmath58 ) the @xmath94 parameter increases .
n - body simulations by mihos et al .
( 1997 ) have shown that in encounters of equal - mass galaxies , @xmath95 in the inner disk is enough to stabilize systems with slowly rising rotation curves like those of lsbs , to be stable against bar formation , while @xmath96 is required for disks with steep rotation curves like those of hsbs .
as in our simulations the perturber s mass ( the milky way ) is at least 30 times larger than the mass of the target dwarf galaxy , we will be able to test the validity of this criterion in a radically different situation .
finally , we will test the susceptibility to vertical bending instabilities with model lmh ( table 1 ) , identical to model lm1 except for a larger scale - height ( @xmath97 ) ; this should suffer little vertical heating ( see raha et al .
1991 ) . to fix the parameters of model galaxies with different ( smaller ) masses
we choose a value for the new mass and then we make use of equations ( 3 ) to determine the new circular velocity @xmath98 and virial radius @xmath73 of the dark matter halo _ assuming @xmath75 and thus @xmath99_. the disk masses and scale lengths are assumed to simply follow the halo scaling ( mo et al .
the resulting models are hm2 and lm2 ( see table 1 ) .
as shown by eqs.(3 ) , masses and radii of objects formed in hierarchical cosmogonies depend on redshift _ at fixed circular velocity_. the scaling is contained in the hubble constant : @xmath100}^{1/2}\ ] ] and implies that radii and masses decrease at the same rate , i.e. @xmath101 . as a consequence the density of objects grows as @xmath102 .
some of the dwarf spheroidal galaxies inhabiting the local group must have formed at @xmath103 or higher given the very old age of the bulk of their stellar populations .
therefore , we also build models with masses and radii rescaled as they were formed at @xmath103 ( models lz and hz , see table 1 ) to investigate how the scaling with redshift implied by the hierarchical scenario can affect their evolution .
the intrinsic stability properties of all these rescaled models are unchanged with respect to the basic models because equations 3 and 4 define an invariant transformation with respect to both @xmath88 and @xmath94 .
moreover , their rotation curves look exactly the same as those of the basic models shown in figure 1 , albeit on a different scale .
so far we have used the tully - fisher relation to link circular velocities of halos to luminosity .
hoffman et al . (
1996 ) have shown that the t - f relation exhibits a large scatter at @xmath104 .
the cosmological scaling offers an attractive alternative to set the physical parameters for models of the faintest dwarf irregular galaxies .
examples of very faint dirrs are found in the lg ; both gr8 and sagdig have @xmath16 .
it turns out that galaxies that faint should correspond to halos with @xmath105 , and that such halos form typically at @xmath106 in cold dark matter cosmogonies ( lacey & cole ( 1993 , 1994 ) ) .
we build a model resembling the gr8 dwarf using the kinematics of the gas at large radii to construct its global potential . at 2 - 3 disk scale lengths the neutral hydrogen in gr8
is pressure supported , with a peak velocity dispersion @xmath107 km / s ( carignan et al . 1990 ) . assuming that the gas is in virial equilibrium with a surrounding isothermal dark matter halo
, the latter should have a virial circular velocity @xmath108 km / s : this fixes the virial mass and virial radius of the halo through equations ( 3 ) and equation ( 4 ) , which are evaluated for @xmath109 .
the disk scale length ( @xmath110 pc ) and stellar disk mass ( @xmath111 ) are the observed values ( carignan et al . 1990 ) .
the chosen @xmath58 and @xmath73 , assuming @xmath112 , imply a very concentrated halo and a very high halo / disk mass ratio ( see table 1 ) .
the resulting central dark matter density is very high , @xmath113 .
this value is slightly higher than inferred for gr8 ( @xmath114 ) but similar to that inferred for draco and ursa minor ( lake 1990 ; mateo 1998 ) .
the mass - to - light ratio at the optical radius ( @xmath115 ) is @xmath116 32 , like that of gr8 not counting the contribution of the gaseous disk .
the rotation curve of the `` gr8 '' model is shown in figure 3 .
we also construct an lsb and an hsb model with a cold gaseous disk which should represent the neutral hydrogen present in observed dwarf irregular galaxies .
dwarf irregular galaxies are generally gas - rich , with an average total hi - to - stellar mass ratio larger than one ( hoffman et al . 1996 ) .
however , the gaseous disks are typically two times more extended than the stellar disks ( cote et al . 1997 ; de blok & mcgaugh , 1997 ; young & lo 1997 ) , whereas within the optical radius ( @xmath116 3 disk scale lengths ) , the neutral hydrogen fraction often drops to less than @xmath117 of the stellar mass ( jobin & carignan 1990 ; cote et al .
1990 ) .
= 9truecm as we will show , material lying outside @xmath118 is almost entirely stripped when the satellites orbit in the milky way halo and thus would not affect the bound components .
therefore , the gaseous disk included in our model galaxies extends only out to the radius of the stellar disk and its mass is 30% of the total baryonic mass . the total disk mass ( stars + gas ) as well as all the other structural parameters are shown in table 1 ( models lmg2 and hmg2 ) .
the density drops to zero at a radius @xmath119 to mimic the holes in neutral gas distribution existing in the center of dwarf irregular galaxies ( cote et al .
the temperature of the gas is fixed at 5000 k after stability tests in the adiabatic regime .
the rotation curves of both models are shown in figure 4 .
n - body simulations are subject to two - body relaxation effects
. these can be particularly severe when systems of different masses or a system of several components are present .
massive halo particles may heat up the disk even in absence of any external perturbation by colliding with lighter stellar or gas particles : a similar numerical effect is known to be responsible for the premature evaporation and disruption of small subhalos in cosmological simulations with insufficient resolution ( moore , katz , & lake 1996 ) . to minimize numerical disk heating we use lighter halo particles in the central part of the halo ( precisely within 10 disk scale lengths ) where orbital times are shorter and therefore collisions with disk particles are more frequent .
we perform several tests in isolation in order to find the optimal balance between reasonable stability and a moderate total number of particles ( see figure 5 ) .
the final models have a disk of @xmath120 particles while the halo is sampled by 208350 in the lsb model and 180860 in the hsb model ( the hi - res halo component in the lsb extends out to a larger fraction of the virial radius because the stellar disk does as well , resulting in a higher total number of particles ) .
for the gr8 model we had to use more than @xmath121 particles in the halo due to the exceptionally high halo / disk ratio : this is the _ highest resolution model ever made for a single galaxy _ with a particle mass of only @xmath122 . in the simulations we use a spline kernel for the gravitational softening as well as for the smoothing length in the sph calculations ( hernquist & katz 1989 ) .
the softening length is a fixed fraction of the disk scale length @xmath58 , thus ensuring the same _ intrinsic _ spatial resolution for hsbs and lsbs and for all their rescaled versions . in code units ( for which @xmath123 ) the softenings are @xmath124 for the stars , @xmath125 for the dark matter particles in the hi - res component of the halo and @xmath126 for those in the low - res component .
we adopt an opening angle @xmath127 for the force calculation and expand multipoles to hexadecapole order .
the timestep @xmath128 is assigned using a multistepping criterion based on the local acceleration of particles , namely @xmath129 , where @xmath130 is chosen to be 0.3 .
= 5truecm in the collisionless runs the timestep varies in the range @xmath131 , where @xmath132 is the crossing time of the disks at @xmath133 . in the gasdynamical
runs we use a courant parameter = 0.4 and the timesteps can be as small as @xmath134 .
radiative cooling is implemented for a primordial mixture of hydrogen and helium .
we use 20.000 particles to sample the gas component and 40.000 particles for the stellar disk , minimizing any eventual collisional heating of the gaseous disk by the star particles since the ratio of the total masses of the two components is @xmath135 ( otherwise two - body heating by stellar or dark particles may suppress radiative cooling of the gas , see steinmetz & white 1997 ) .
halos , in this case , have 250.000 particles in total .
the satellites are placed on bound orbits in the potential of a milky - way sized primary halo .
the latter is modeled as the fixed potential of an isothermal sphere with a total mass @xmath136 inside a virial radius @xmath137 kpc , consistent with both recent measures based on radial velocities of distant satellites ( wilkinson & evans 1999 ) and with generic models of structure formation ( peebles et al .
the core radius is @xmath138 kpc and the resulting circular velocity at the solar radius is @xmath139 km / s .
dynamical friction is neglected because orbital decay times are expected to be longer than the hubble time given the small mass of the satellites ( they are at least @xmath33 time less massive than the primary ) and the delay introduced by tidal stripping ( colpi , mayer , & governato , 1999 ) . in the majority of the runs , which we will call the `` standard runs '' , the orbits have an apocenter close to the virial radius of the milky way halo , in agreement with the results of n - body simulations of structure formation ( ghigna et al .
the same simulations show also that the orbits of satellites have an average apo / peri @xmath45 , while sporadic measurements of the orbits of lg dsphs yield , although with large errors , nearly circular orbits , with average apo / peri=2 ( schweitzer et al . 1995 ; ibata & lewis 1998 ) .
we explore mainly orbits with apo / peri between 4 and 10 but we also consider orbits with apo / peri as low as 1.5 - 2 .
the orbital energy @xmath140 is usually such that the corresponding circular orbit would have a radius @xmath141 and a typical period of @xmath48 gyr .
we will refer to these orbits as to the _
standard orbits_. as the orbits considered have typically pericenters larger than 40 kpc , the contribution of the milky way disk to the primary potential should be negligible ; we will test the validity of this hypothesis by adding an axisymmetric component to the external potential in a few runs .
= 9truecm satellites that entered the milky way halo at high redshift ( @xmath142 ) became bound on orbits with a significantly smaller apocenter compared to satellites infalling later because the primary halo itself was typically 2 - 3 times smaller ( see equations ( 3 ) and ( 4 ) ) .
therefore , we also consider the regime of _ tightly bound orbits _ , with apocenters ranging from @xmath143 to @xmath144 of the present - day virial radius of the milky way halo , thus having a orbital energy @xmath140 lower than that of the standard orbits .
the inclination of the disk spin with respect to the orbital angular momentum is also varied , going from direct to retrograde spin - orbit orientations .
the combination of different orbital parameters and different models of satellites has led us to perform more than 50 high - resolution runs . the orbital parameters and satellite models employed in the various runs
are indicated in table 2 and table 3 .
the dwarf galaxy models are evolved for several gyrs in the external potential of the milky way halo .
the strong tidal disturbance exerted by the primary halo produces a dramatic morphological evolution of the small satellites after 2 - 3 pericenter passages , corresponding to 7 gyr on the standard orbits .
the satellites are first tidally truncated down to the radius imposed by their pericenter distance and then continue to lose mass as each subsequent tidal shock pumps energy into them and decreases their potential wells ( gnedin & ostriker 1999 ) . in this section we will focus on the evolution of the models with the `` standard '' initial stability properties ( i.e. @xmath85 , @xmath145 and @xmath146 ) , while we will discuss the effects of changing such properties in section 6.1 . in general ,
lsbs lose more than @xmath147 of their halo mass and @xmath148 of their stellar mass , while hsbs , although still losing most of their halo mass , retain , on average , at least @xmath149 of their initial disk mass . a high - density dwarf model like gr8 is barely perturbed on a standard orbit ( run gr81 ) , while it loses @xmath150 of its disk mass but most of its outer halo mass if it falls on a short - period , tightly bound orbit ( @xmath151 gyr ) as expected at @xmath152 ( run gr82 ) , performing five pericenter passages in less then 7 gyr .
= 9truecm the evolution of the mass distribution for the various models is reflected in the variation of the circular velocity profiles ( figures 6,7,8 ) .
after 7 gyr lsbs have stellar masses @xmath153 while hsbs have stellar masses closer to @xmath154 .
lsbs are more fragile than hsbs both because they have a low - concentration halo and because they have large disk scale lengths . for the same disk mass ,
the difference in scale length is a factor 2.4 .
correspondingly , if @xmath155 is the crossing time at the disk half - mass radius and @xmath156 is the collision time ( i.e. the time spent close to pericenter @xmath157 while moving at the velocity @xmath158 ) , for @xmath159 kpc as in most of the cases ( see table 2 and table 3 ) we have @xmath160 for the lsbs , while @xmath161 for the hsbs , and thus the latter will respond more adiabatically to the perturbation ( gnedin et al .
orbits with progressively larger pericenter obviously produce a lower damage even in lsbs .
finally , limited stripping occurs in hz and lz models that have a disk scale length 2.83 times smaller than their z=0 counterparts ( figures 6,7 ) .
close to the first pericentric passage both lsbs and hsbs develop a central bar , its scale length being always @xmath162 .
the higher intrinsic stability of lsbs to bar formation ( suggested by the @xmath94 parameter in figure 2 ) is thus not indicative when the perturber , namely the milky way , is several times more massive than the target satellite , as was also argued by miwa & noguchi ( 1998 ) .
the gr8-like model develops very extended spiral arms after the first tidal shock .
the second shock triggers the formation of a central bar which is more compact than usual , being @xmath163 in length .
= 8truecm the appearance of the bar structure corresponds to a drastic change of the orbits of stars inside the bar radius : nearly circular orbits turn turn into nearly radial box or tube orbits supporting the elongation of the bar .
this change is illustrated in figure 9 , where the specific angular momentum of stars inside @xmath58 is plotted against time .
the bar leads to an increase in the central concentration of the stellar density profiles , as explained in the next section .
however , the degree of central concentration may vary depending on the strength of the bar . in the case of lsbs
the bar is weakened by tidal stripping as stars are removed even inside the bar radius when the pericenter distance is below 50 kpc ( e.g. run l01 ) . instead ,
for larger pericenters or for satellite models with smaller initial disk scale length and/or higher central densities ( all hsb models and also lz models ) , the bar is quite stable to stripping .
the rotating bar soon slows down shedding angular momentum outwards to the halo and outer stellar material as a result of dynamical friction ( hernquist & weinberg 1989 : fux et al .
1995 ) . in the meantime
, stripping removes the high - angular momentum stellar material outside the bar radius .
the bar starts to buckle soon after the first orbit , when the radial anisotropy has increased in such a way that @xmath164 , where @xmath165 and @xmath166 are , respectively , the stellar velocity dispersion normal and parallel to the plane defined by the two longest principal axes of the remnant ( see also raha et al .
1991 and merritt & sellwood 1994 ) .
the buckling gradually erases the bar symmetry , ultimately leading to a more isotropic , spheroidal configuration : the transition is highlighted by the evolution of both @xmath165 and the minor / major axis ratio of the stellar mass distribution , @xmath167 ( figure 9 and figure 10 ) .
once excited , the tidally induced bar and bending modes grow on the dynamical time scale and thus the morphological transition is faster in high - redshift satellites ( hz and lz ) , which reach a spheroidal configuration @xmath168 gyr after the bar appears while , other satellites remain in a transitory , bar - like configuration for a few gyr .
= 8truecm = 8truecm after @xmath169 gyr of evolution , nearly all the disks have been transformed into spheroidals ( figure 11 ) .
heating induced by bar buckling is more important than direct heating by the tidal field : indeed satellites inclined by 90 degrees with respect to their orbital plane ( which suffer the strongest direct tidal heating in the direction normal to the disk plane ) undergo less vertical heating compared to those on a coplanar orbit ( where a stronger bar is induced ) , as shown in figure 9 ( see caption ) .
it is thus understood why even a very stiff object like the gr8 model , that is barely stripped , is heated into a spheroidal configuration .
the joint action of loss of angular momentum and induced vertical heating determine a marked decrease of the @xmath1 of the bound stellar component ( figure 10 ) after the morphological transformation the satellites remain in a nearly stable dynamical state as was verified by pushing a few simulations to over 10 gyr .
= 8truecm = 8truecm when the satellites are placed on retrograde orbits the initial disk structure is partially preserved . except for the case of lsbs on orbits with the smallest pericenters , tidal stripping
is remarkably reduced and weaker tidal tails are observed due to an anti - resonant coupling between orbital and internal motions ( toomre & toomre 1972 ; barnes 1988 ; springel & white 1998 ; mihos et al .
the bar forms when the satellite performs the second pericenter passage , after @xmath116 5 gyr .
vertical heating is not as strong as usual and the satellite remains quite flattened .
the final @xmath1 within @xmath170 is still close to 1 at the end .
we perform two dissipational runs placing models lmg2 and hmg2 on orbits with apo / peri=9 ( runs l26 and h14 in table 2 and table 3 , respectively ) .
figure 12 shows two important results : the gas is either stripped or funneled to the center of the galaxy .
the gas loses angular momentum after the first pericentric passage , due to the torquing by the stellar bar , which is misaligned with respect to the gas ( heller & shlosman 1994 ; mihos et al .
1997 ) . in the hsb only @xmath171 of the initial gas mass
is tidally stripped , while all the rest is driven to the center soon after the first pericentric passage . as a result
, a central spike appears in the gas density profile ( the maximum density increases by more than two orders of magnitude , as seen in figure 13 ) and then the profile evolves little for the following gyr .
instead , in the lsb more than half of the gas is stripped , while the remaining part is still funneled towards the center but at a slower pace , the density growing by only one order of magnitude ( figure 13 ) .
= 5truecm stripping dominates in the lsb because the gaseous disk is more loosely bound ( initial gas profiles follow the stellar density profile in our models ) and the central inflow is moderate as a result of both the weak bar and the long dynamical time . in the lsb , as the gas slowly drifts inwards , it has time to feel the effect of tidal compression at the second pericenter passage , which leads to the largest increase in the central gas density ( figure 13 ) .
tidal compression occurs at the turning points of the orbit because the direction of the tidal force relative to orbital motion is reversed : the leading edge of the galaxy is thus pushed towards the trailing edge .
the compression will be larger at pericenter because there the magnitude of the tidal force is greater .
= 8truecm = 8truecm
the final stellar surface brightness profiles do not deviate substantially from the initial exponential form , although they tend to become steeper ( figures 14 and 15 ) . in general , when the galaxy is structurally stiff or the pericenter distance is fairly large ( @xmath172 kpc ) , the profile is bi - modal , having a steeper slope inside the half mass radius @xmath170 of the galaxy ( see figures 14 and 15 ) : this is because the bar instability leads to a more concentrated stellar component , which persists unless the subsequent tidal shocks are strong enough to strip the system down to the central part . as a result ,
both single exponential and bi - modal profiles are present among lsb satellites , while the stiffer hsbs or the high - redshift dwarf models , in particular gr8 , are always fairly stable to tides within the bar radius and always exhibit bi - modal profiles .
= 8truecm = 8truecm the typical sizes of the remnants can be measured by means of their final half - mass radius @xmath170 ( figure 16 ) .
the more pronounced compactness of the remnants of hsbs versus those of lsbs resembles an analogous difference seen in des versus dsphs ( ferguson & binggeli 1994 , kent 1991 , carter & sadler 1990 ) . at the end hsbs
have a central surface brightness only slightly smaller than the initial one ( even including fading according to the model described in section 9 ) , while their stellar mass is decreased by a factor of 3 or more : therefore , they turn out to have a surface brightness higher than any dirr with similar luminosity , just as it is typically observed for des in the lg and in nearby groups and clusters ( mateo 1998 , james 1991 ) .
= 9truecm when fitting the profile , for some of the lsbs a single exponential law works pretty well , while king profiles with a concentration @xmath173 reproduce better the bi - modal profiles of the remnants of hsbs and of gr8 ( see figure 17 ) .
because of the greater compactness of hsb remnants , their core radii resulting from the fits with king profiles are typically as small ( @xmath174 pc ) as those of lsb remnants ten times less massive .
this matches the observations well : bright des like ngc147 and ngc185 have core radii as small as those of dsphs like carina and leo i ( mateo 1998 ) , the latter being fainter by @xmath175 magnitudes .
= 9truecm in the dissipative runs some dynamical feedback of the gas on the stellar component occurs .
the latter is different in hsbs vs. lsbs owing to the different dynamical evolution of their gas component .
the strong central gas concentration that appears in the hsbs soon after the first pericentric passage immediately produces a response of the stellar potential and the final surface brightness profile is remarkably steeper in the center compared to the collisionless case , as shown in figure 18 . on the other end , in the lsb
the stellar density profiles evolve nearly in the same way with and without gas because the weaker bar does not create a dense central gas knot and a substantial fraction of the gas is stripped away ; the gas - to - stellar mass ratio inside @xmath176 is roughly the same as at the beginning .
the kinematics of the satellites changes drastically by the end of the simulations .
the final @xmath1 measured inside @xmath170 is typically lower than @xmath177 , this being a distinctive feature of dsphs and des not only in the local group but even in galaxy clusters ( mateo 1998 ; bender & nieto 1990 , bender et al .
1992 ; ferguson & binggeli 1994 ) , a slightly larger ( by @xmath178 ) residual rotation around the minor axis is present at larger radii , due to high angular momentum , loosely bound material which is about to be tidally ejected .
the transition in the dwarfs kinematics is the major success of our model as it provides an explanation for the _ most important dynamical difference _ between dirrs and dsphs : _ their angular momentum content_. the bulk of the result is independent from the viewing projection , as shown in figure 19 , which also shows that all the remnants fall below the theoretical prediction for rotationally flattened spheroidals , @xmath179 , where @xmath180 is the ellipticity projected onto the plane of the sky ( binney & tremaine 1987 ) , which means that any flattening is due to anisotropy in their velocity dispersion , like in giant ellipticals , as suggested by observations ( ferguson & binggeli 1994 ) .
= 8truecm the final average velocity dispersions vary from 10 to 20 km / s among the remnants of lsbs , while they are around 30 - 35 km / s for those of hsbs ( figure 20 ) .
the initial models of lsbs and hsbs span exactly the same range of masses and both have initial velocity dispersions of a few km / s ( only @xmath181 larger in the hsbs ) , being rotationally supported .
it is the larger mass retained by the hsb satellites together with their typically more concentrated final stellar profile that determines such a kinematical distinction by the end of the simulations .
the velocity dispersions of the remnants of lsbs are in good agreement with those inferred for dsphs , while those of the remnants of hsb match those of des ( mateo 1998 ) .
the measured velocities are very close to the expectation from virial equilibrium when the residual rotation is negligible . in other words
, we do not find that tidal forces inflate the velocity dispersions in the bound remnants , in agreement with piatek & pryor ( 1995 ) and pryor ( 1994 ) , who did a similar analysis on spherical satellites .
= 8truecm the velocity dispersion profiles of the remnants are typically slowly decreasing ( figure 21 ) , as found in some of the dsphs ( hargreaves et al .
1994a , b , 1996 ) and des ( ngc205 , held et al . 1990 , 1992 ; carter & sadler 1990 ) .
the final radial velocity anisotropy , measured by the parameter @xmath182 , is higher in the more flattened lsbs ( figure 22 ) .
the average value of @xmath183 is @xmath184 for hsbs and @xmath185 for lsbs : thus remnants are on average only mildly anisotropic . to measure the shape of the remnants we use the triaxiality parameter @xmath186 ( franx , illingworth & de zeeuw 1991 ) , where @xmath187 , @xmath188 and @xmath189 are , respectively , the major , intermediate and minor axis of the remnants .
@xmath190 represents a purely prolate halo , while we have @xmath191 for a purely oblate halo .
objects can be defined nearly prolate if @xmath192 , nearly oblate if @xmath193 and triaxial if @xmath194 . in measuring the shape of the remnants within @xmath170
we find that the majority of the remnants are nearly prolate spheroidals and a fair fraction is triaxial , while oblate shapes are very rare .
the tendency towards prolateness is due to the residual radial anisotropy originated during the bar stage of the evolution and can be enhanced by tidal distortion , especially for the loosely bound lsbs .
= 8truecm = 8truecm the parameter @xmath167 provides a measure of the maximum observable flattening of the remnants : the higher the flattening the lower will be @xmath167 . figure 22 clearly indicates that the remnants of lsbs are more flattened than those of hsbs , with average @xmath167 of @xmath185 and @xmath177 , respectively .
this is expected because vertical heating occurs faster in hsbs due to their smaller dynamical times .
in addition , the more flattened remnants of both hsbs and lsbs correspond to cases of nearly retrograde encounters that better preserve the original disk structure .
an anti - correlation between final surface brightness and final flattening is present in the remnants .
interestingly , there are indications that bright des , both in the lg and in galaxy clusters , are rounder than low surface brightness dsphs ( van den bergh 1986 , ferguson & sandage 1991 )
. however , the comparison with observations is not trivial because we have to take into account projection effects . for example , if viewed along the major axis all the remnants would appear rounder than when viewed along either of the two other axes , as seen in figures 22 .
the mean ratio between two of the principal axes , obtained averaging over the three projections , is thus higher than that associated with the measure of @xmath167 , being around @xmath177 for the remnants of lsbs and around @xmath195 for the remnants of hsbs : we can consider the latter values as yielding the _ mean observable flattening_. = 8truecm ellipticities of lg dsphs range from the extreme value of @xmath196 in sagittarius and @xmath197 for ursa minor to @xmath198 for sextans , with the average being around @xmath199 ( mateo 1998 ) . we can safely conclude that the ellipticities of the remnants fall in the observed range .
when the gas is included in the model satellites we observe an increase in the final rotation of the stars inside @xmath170 , by @xmath200 in the lsb and by about @xmath201 in the hsb .
the reason is that the gas spins up the stellar bar while losing angular momentum ( friedli 1999 ; friedli & benz 1993 ) .
this mechanism is suggested by figure 23 , where we compare the evolution of the specific angular momentum of the bound stellar component with and without gas . because rotation is higher , the remnant of the lsb is also flatter ( @xmath202 instead of @xmath203 ) than in the collisionless case . in the case of the hsb ,
the dominant effect is instead the stronger central concentration of both the gas and the stellar components , and thus the remnant is rounder , notwithstanding the higher rotation ( @xmath204 instead of @xmath205 ) .
overall , the final @xmath1 increases slightly for the lsb while it is almost unchanged for the hsb because a higher velocity dispersion is achieved due to the more concentrated potential .
a rounder shape of gaseous remnants was also found by moore et al .
( 1998 ) for harassment in galaxy clusters as in that case the model galaxies were structurally more similar to our hsb models .
the circular velocity profiles ( figures 6 , 7 and 8) show the relative contribution of dark matter and stars after 7 gyr of evolution inside the milky way potential .
remnants of lsbs , being severely stripped down to their inner parts , have a `` central '' @xmath206 ( measured inside @xmath170 ) around 1 - 2 , while their high - redshift counterparts ( lz models ) can have ratios as high as 4 - 5 .
hsbs , although quite resistant to tides , have in general @xmath207 as they start off with a ratio @xmath206 ( within the optical radius ) a factor of 2 lower than that in lsb models .
contrary to common belief , not all dsphs have extremely high dark matter content ; most of them have inferred central mass - to - light ratios ( measured at the core radius ) between 5 and 20 ( mateo 1998 ) , which , for the stellar mass - to - light ratios expected in objects dominated by old stars ( ( @xmath208 ) , implies dark matter halo masses within the range of our results .
mass - to - light ratios in des turn out to be moderate , @xmath209 ( carter & sadler 1990 ) , in agreement with those of the remnants of hsbs .
our evolutionary model is able to explain also the extreme dark matter content of the faintest dsphs , draco and ursa minor ( lake 1990 ; hargreaves et al .
1994a , b ; mateo 1998 ) .
the gr8-like model almost preserves the initial @xmath206 ; with @xmath210 within @xmath170 , the mass - to - light ratios would be @xmath211 for @xmath212 .
= 8truecm a dwarf as dense as gr8 needs to enter early into the milky way halo to be transformed into a dsph , requiring a high number of tidal shocks and thus a small orbital time .
the hierarchical structure formation scenario naturally meets these conditions because the densest objects form , on average , earlier and are accreted by the main halos on orbits with smaller radii ( see section 3 ) .
we have shown that the transformation occurs on a orbit with apocenter @xmath213 kpc , in agreement with the present distances of draco and ursa minor ( mateo 1998 ) .
as time goes on , lower - density dwarfs will form and enter into an enlarged milky way halo : these will be on wider orbits and will be more akin to the standard lsb models , thus ending up in dsphs with lower mass - to - light ratios . an anti - correlation between distance from the primary and mass - to - light ratio
should thus arise and is indeed suggested by observations ( mateo 1998 ; irwin & hatzidimitriou 1995 ) .
early - type dwarf galaxies in the local group show a certain spread in their total central densities ( including baryons and dark matter , see mateo 1998 ) . comparing these data with the densities of our remnants
shows a remarkably good agreement .
the final central densities of the remnants of lbs are as low as those of most dsphs ( e.g. sextans , sagittarius and fornax ) , namely @xmath214 ( mateo 1998 ) the remnant of gr8 is comparable to the densest dsphs , draco , sculptor and ursa minor , having @xmath215 ( lake 1990 ) .
even higher densities are reached in the inner part of the remnants of hsbs thanks to the highly concentrated stellar potential , as found in des .
however , central densities as high as those of ngc205 ( @xmath216 , see mateo ( 1998 ) ) appear only in the gas - dynamical runs ( model hm2 g ) , where the profile of the stellar component steepens remarkably .
= 8truecm = 8truecm it is important to note that the spread of the central densities of the remnants is only slightly larger than that of the initial models .
therefore , our ability to reproduce the vast range of properties of early - type dwarfs results from the choice of initial models spanning a wide range of central densities .
the spread in the concentration @xmath189 in the initial models is a factor of 4 , which corresponds to a factor of almost 20 in the initial central densities ( @xmath217 for truncated isothermal spheres ) . in cold dark matter models
the typical concentration of halos on a scale @xmath218 is as high as that of our gr8 model .
cosmological n - body simulations show that the scatter of the concentration at a given mass scale is large ( bullock et al .
2000 ) but no as much as to encompass the extremely low values chosen for our lsb models .
interestingly , the latter values seem to be very common for small halos formed in wark dark matter ( wdm ) cosmogonies ( bode , ostriker , & turok 2000 ; eke , navarro , & steinmetz 2000 ) .
= 4truecm = 4truecm = 4truecm = 4truecm the issue of the high mass to light ratios found in some of the dsphs is critical for understanding in general the nature of dark matter ( lake 1990 ) .
many authors have pointed out that some steps of the core fitting procedure used to derive @xmath219 for dsphs and/or tidal effects could lead to overestimate them ( piatek & pryor 1995 ; irwin & hatzidimitriou 1995 ) .
it is usually assumed that mass follows light in dsphs . in figure 24
we see the comparison between final total mass and stellar mass profiles for three `` prototype '' models . for hsb and lsb
remnants one would actually slightly underestimate @xmath219 ( by 10 - 30 % ) by assuming that mass follows light .
in contrast , the difference is remarkable for models as stiff as gr8 , where we would underestimate the dark matter mass by a factor @xmath220 .
we can also ask whether the assumptions that ( 1 ) the measured velocity dispersion can be interpreted as resulting from virial equilibrium and that ( 2 ) such equilibrium is isotropic hold for our remnants .
nearly all of our remnants have reached a nearly - stable equilibrium by the end of the simulations . however , in some of them a fair degree of rotation is present in addition to the velocity dispersion and this should be taken into account when estimating the virial mass of the object .
in a few cases , in which @xmath221 , one would underestimate the true mass - to - light ratio by nearly a factor of 2 considering only the contribution of the velocity dispersion to the kinetic energy ( because m @xmath222 ) .
therefore , if , for example , the rotation apparently detected in ursa minor by hargreaves et al .
( 1994a ) is intrinsic , the true @xmath219 could be even higher than that usually assumed .
= 8truecm = 8truecm according to figure 22 the maximum value of the anisotropy parameter @xmath183 is @xmath223 , which means that velocity dispersion in the plane of the galaxy can be a factor @xmath220 higher than that normal to the plane ( here we are assuming that the velocity dispersions along the two principal axes lying _ within _ the plane are comparable , as found in most of the remnants ) ; at most one would overestimate the one - dimensional velocity dispersion by only @xmath224 ( from @xmath225 , which leads to an overestimate of a factor @xmath226 for @xmath219 ( from @xmath227 ) as an upper limit .
however , only @xmath150 of the remnants exhibit such high final radial anisotropies .
a more interesting issue is provided by projection effects : the velocity dispersion measured along the line of sight ( and thus the @xmath219 ) could be overestimated if the unbound , fast moving stars that make up the tidal tails are included by chance in spectroscopic samples .
these `` extra tidal '' stars produce a radial positive gradient in the velocity dispersion profile ( figure 25 ) .
the larger variations are seen in the lsbs due to its more pronounced mass loss .
= 8truecm as we can measure only projected distances , unbound stars , although lying far away from the center , could appear to be located near the core of the dwarf .
from our simulations we can get an idea of how these projection effect can alter the observed kinematics .
we derived an upper limit for the velocity dispersion overestimate for both lsbs and hsbs , by selecting the worst case possible " , namely two runs with direct orbits ( l01 and h01 , see table 2 and table 3 ) , in which the largest tails are produced .
we viewed the remnant along the orbital plane close to apocenter , where the stream density is at one of its density maxima and the galaxy spends most of the time ( at the turning points of the orbits the stream density is higher because of conservation of phase space density , e.g. helmi & white 1999 ) .
we measure the average velocity dispersion of the bound and unbound stars projected within a distance equivalent to @xmath170 and find the average of the dispersions of the two individual samples by weighting according to the number of stars in each of them .
we find that in this extreme case the velocity dispersion can be overestimated by @xmath228 for the lsbs , while for the hsbs the difference is of a few percents . as a consequence
, the mass - to - light ratio can be overestimated by a factor @xmath4 for the lsbs , while essentially no overestimate occurs for the hsbs ( the latter statement would be even more true for dense dwarfs like the gr8 model ) .
the remarkable difference in the apparent dispersions of lsbs vs. hsbs simply arises because in the former case a much larger fraction of the stars is located in the tidal streams that follow the original orbit .
= 8truecm = 8truecm although not negligible , such tidal effects can not rule out the need of high dark matter contents in some of the dsphs : objects like draco and ursa minor ( @xmath18 ) would still have @xmath229 even with the maximum velocity dispersion enhancement just described . on the other end
, the velocity gradient shown in figure 25 suggests that the rotation detected in ursa minor ( hargreaves et al .
1994a ) might be only apparent .
so far we have based our description of the numerical results on the set of satellite models that we consider more akin to observed dirrs and we have placed them on the typical orbits found in n - body simulations of structure formation .
furthermore , we have considered only one form of the primary potential , whose global parameters were fixed again based on structure formation models . in order to figure out how general the `` tidal stirring '' mechanism explored in this paper is , in this section we will relax each of these hypotheses , one at a time , investigating a much wider parameter space . as already mentioned in section ( 2.1 ) ,
we have built a special set of models which should be the least prone to bar and/or bending instabilities by varying @xmath88 , the initial disk thickness or the core radius of the halo ( see table 1 for the models parameters and figure 2 for the intrinsic stability properties ) .
we mostly placed such models on a direct orbit with moderate eccentricity ( apo / peri=4 , @xmath230 kpc ) : this configuration , in which the excited bar is barely damaged by tidal stripping even in the case of the fragile lsbs , is ideal to study the vertical heating caused by bar - buckling .
the results are illustrated in figure 26 .
an lsb model with a small core radius ( run l21 ) undergoes a smaller increase of @xmath165 ( by about @xmath231 ) compared to the standard reference case ( run l02 ) .
a weaker bar develops because the higher halo central density damps the growth of @xmath93 modes and the buckling is thus reduced due to the lower radial anisotropy .
an analogous result is obtained with an hsb disk embedded in a more compact halo ( model hm1rc03 ) , which has an initial rotation curve very similar to that recently derived by sofue ( 1999 ) for the lmc .
if placed on a nearly circular orbit ( run h11 ) as implied by recent astrometric measurements ( kroupa & bastian 1997 ) , the kinematics remain disk - like at @xmath232 ( stripping is very modest ) , but a thick spheroidal component develops after 5 gyr .
these results qualitatively agree with those obtained by weinberg ( 2000 ) in a detailed study of the response of the lmc to the tidal field of the milky way : the thick spheroid could be responsible for some of the microlensing events seen towards the lmc ( evans & kerins 2000 ) . in the models with @xmath233 ( run l23 ) ,
namely considerably `` hotter '' than the standard models , the bar is also weak and @xmath165 increases by less than a factor of 2 .
= 8truecm we then explore models with a larger scale height ( @xmath234 ) and @xmath85 ( model lmh in run l22 ) : the vertical dispersion increases by only @xmath235 , a factor @xmath48 less than in run l02 .
this is not surprising because in this model @xmath165 is initially 2 times larger than usual , placing the dwarf well above the critical radial anisotropy level necessary for the onset of bending ( see raha et al .
1991 ) . despite the different degree of evolution reported in these `` peculiar models ''
the final @xmath1 inside the half - mass radius is always less than 1 , with the exception of run l22 .
however the latter case is rather unrealistic as late - type spiral and dwarf irregular galaxies show a tendency towards very thin disks ( see van der kruit & de grijs 1999 ) .
in general , our evolutionary model seems quite robust : tidal shocks are too strong for the initial stability properties of the models to change substantially the final outcome of tidal stirring . in this section
we will investigate to what extent the effectiveness of tidal stirring depends on the orbital parameters and , in particular , on the eccentricity of the orbits .
although eccentric orbits are expected for satellites in the framework of hierarchical structure formation ( ghigna et al .
1998 ) , so far observational data on dsphs have been in favour of nearly circular orbits ( e.g. schweitzer et al .
we thus perform a number of runs in which satellite models are placed on low - eccentricity orbits .
lsb satellites on orbits with apo / peri @xmath35 maintain @xmath15 and a disk - like appearance even after 7 - 8 gyr if their pericenters are larger than 150 kpc ( run l08 and l17 ) , irrespective of the disk / orbit orientation . on the other end , if the orbits have equally low eccentricity but considerably smaller pericenter ( run l16 and h09 ) , both lsbs and hsbs do transmute into a spheroidal galaxy after a few orbits . in the latter case the apocenter ( @xmath236 kpc )
is comparable to the present distance of the farthest dsph , leo i , and more and stronger tidal shocks are suffered by the satellite because the orbital time is shorter than in runs l08 and l17 ( @xmath48 gyr instead of @xmath237 gyr ) and the pericenter distance are nearly a factor of 2 smaller .
however , compared to the standard runs employing high - eccentricity orbits ( e.g. run l1-l15 and
h01-h08 ) the velocity at pericenter is sensibly lower ( @xmath238 km / s instead of @xmath239 km / s ) and therefore the encounter takes place in a `` less impulsive '' regime .
therefore , the satellite should respond more adiabatically to the perturbation .
evidently , the lower velocity is not able to counterbalance the effect of the large perturbing mass of the milky way halo when the pericenter distance is sufficiently small . therefore , orbital energy ( or pericenter distance ) is at least as important as orbital eccentricity in setting the effect of tides , as found also by gnedin et al.(1999 ) using a semi - analytical approach .
= 8truecm from our wide exploration of parameter space we can conclude that the orbital time @xmath240 is the most significant parameter and that tidal stirring has been effective within the past 10 gyr only for satellites having @xmath241 4 gyr , which corresponds to a circular orbit with a radius @xmath242 ( @xmath243 being the virial radius of the milky way ) .
very dense satellites require smaller orbital times , allowing for them _ more and stronger shocks _ ( a smaller orbital time also implies a smaller @xmath244 , see section 4.1 ) , as shown by the simulations of gr8 ; in the framework of hierarchical structure formation this comes out automatically because , on average , they formed earlier and thus fell into the milky way halo when it had a smaller virial radius . thanks to the rescaling properties of halos ( see section 2 ) , the condition on the orbital time is automatically generalized for every redshift @xmath245 if we assume that @xmath243 is the virial radius of the main halo _ at the redshift of infall_. the present distances of most of the dsphs in the milky way or m31 systems are within @xmath246 ( assuming @xmath247 400 kpc ) and should thus satisfy the constraint on the orbital time .
however , the farthest dsph , leo i , having a distance of about 270 kpc , probably lies at the boundary of the `` allowed '' region .
thus , if leo i was tidally stirred by the milky way , it must be on a fairly eccentric orbit in order to have a sufficiently small orbital time despite its large apocenter distance . at more than 800 kpc from the milky way the tucana dwarf ( gallart et al . 1999 ) could be the greatest challenge for our model if future measurements of its internal kinematics will confirm that it is a dsph in all respects . a few satellites on nearly radial orbits with apocenters just inside the turnaround radius , namely far exceeding the virial radius of the primary halo , are indeed found in n - body simulations ( ghigna et al .
1998 ) .
= 8.3truecm = 8truecm we also performed a few runs to explore the regime of _ very tightly bound orbits _ , namely those with a corresponding circular radius @xmath248 .
in particular , we consider the case of nearly circular and circular orbits with apocenters @xmath249 kpc ( runs l19 , l20 , lz7 , h10 ) .
orbits of this type would be consistent with the the present distances of some of the milky way satellites , the lmc , sculptor , ursa minor , draco and sagittarius what would be the fate of a fragile lsb satellite on such orbits ?
our simulations indicate that lsbs are completely disrupted on nearly circular or even exactly circular orbits at a distance of @xmath250 kpc ( run l19 and l20 ) .
the mechanism is simply one of continuous mass loss ( figure 27 ) : the disks of lsbs are so large that the tidal radius falls within them at these short orbital distances and thus they are dramatically destabilized .
the system responds to strong mass loss by expanding and thus reducing even further its binding energy ( binney & tremaine 1987 ; gnedin & ostriker 1999 ) .
the destruction is faster on a circular orbit ( figure 27 ) because the orbit - averaged tidal force is stronger .
this is in agreement with n - body results on the survival of cosmological halos on tightly bound orbits ( kravtsov & klypin 1999 ) .
instead , hsbs or lz models ( run h10 and lz7 ) have disks lying inside their tidal radius and thus lose mass only after tidal heating has sufficiently loosen their potentials : they lose @xmath251 of their stellar mass and transform into a spheroidal but survive for several gyr .
we conclude that low surface brightness satellites can not survive too close to the milky way or m31 .
sagittarius might be the nearly - disrupted remnant of such a satellite : what remains of the central part of the lsb satellite , in run l19 and l20 , is recognizable as a density enhancement along the streams for a few gyr and has a maximum elongation @xmath252 , similar to that inferred for sagittarius ( mateo 1998 ) .
stable satellites at short distances from the milky way must have high central dark matter densities , as indeed seems to be the case for the nearby draco and ursa minor dsphs ( mateo 1998 ; van den bergh 1999 ) .
a primary halo as massive and extended as adopted so far is expected in structure formation models and is consistent with the observed radial velocities of satellites once even the farthest of the group , leo i and leo ii , are taken into account ( zaritsky et al .
1993 ; wilkinson & evans 2000 ) .
a more conservative approach is based on the use of the magellanic stream as halo mass tracer : the stream is likely bound to the milky way and its origin is explained as the result of gas being tidally stripped from the lmc ( putman et al .
1998 ) . under the latter hypothesis
the milky way halo could extend out to only 50 kpc and could have a total mass of only @xmath253 solar masses ( little & tremaine 1983 ) .
assuming then an isothermal profile with a core radius of 4 kpc still yields a circular velocity of 220 km / s at the solar radius , as observed .
we placed an lsb model on a bound eccentric orbit with apocenter of 220 kpc and pericenter of 25 kpc ( run l24 , see table 2 ) in this `` minimal halo '' .
the apocenter of this orbit is still sufficiently large to be consistent with the distance of the second most distant dsph satellite , leo ii ( @xmath254 kpc ) . due to the truncation of the potential
the orbital time is larger than for orbits with similar parameters in the standard potential ( @xmath45 gyr instead of @xmath48 gyr ) and thus the satellite performs only one pericenter passage in 7 gyr .
the lower number of shocks as well as the longer time spent by the dwarf far from any significant influence of the tidal field produce a remnant with a still recognizable disk structure and @xmath255 , larger than that of any dsph . in figure 28
we see the evolution of @xmath1 for this run ( l24 ) and also a comparison with run l17 ( see table 2 ) , the latter corresponding to an orbit with similar apocenter and lower eccentricity placed in the standard halo model . in figure 28
we can also see that the circular velocity profile evolves much less than in the case of lsb satellites in the standard primary halo ( figure 6 ) , indicating that stripping is comparatively modest .
these results strengthen the point made in the last section , namely that the orbital time is the key parameter determining the transformation .
moreover , they highlight that primary halos as massive and extended as those formed in cdm cosmogonies are required to explain the origin of _ all _ lg dsphs , out to the distances of leo i and leo ii , with tidal stirring .
= 8truecm the plunging orbits chosen in many of our simulations penetrate quite close to the center of the primary halo , at 30 - 40 kpc .
the milky way and m31 have disks that extend at least out to 15 - 20 kpc .
however , their mass is expected to be small compared to that of their halos , even if they do not have @xmath206 ratios as high as those of dwarf galaxies ( persic & salucci 1997 ) . in order to test the validity of our results when a central disk component is added to the primaty
, we placed an lsb satellite ( the most sensible to variations in the tidal forces ) on a orbit with pericenter of 40 kpc in the usual isothermal halo and we added to the latter a disk with a structure similar to that of the milky way ( i.e. with a mass @xmath256 and scale length of 3 kpc ) .
the disk was modeled using a miyamoto - nagai potential @xmath257 where @xmath258 is the disk radius in cylindrical coordinates , @xmath187 is the scale length ( @xmath259 kpc ) and @xmath188 is the scale height .
we set @xmath260 ( binney & tremaine 1987 ) . comparing the results obtained in this run ( l25 ) with an analogous run in the standard primary potential ( l13 ) we see only marginal differences : as illustrated in figure 29 , the mass of the remnant after 6 gyr is very similar , although the addition of the a central disk component to the primary halo slightly enhances stripping .
we thus conclude that the tidal shock is always dominated by the halo , even for quite small pericenter distances .
from the evolution of the gas surface density observed in the gasdynamical runs ( section 4.2 ) we can calculate the evolution of the star formation rate ( sfr ) of the dwarfs using the kennicutt law ( kennicutt 1998 ) .
this is a phenomenological law , calibrated using large samples of disk galaxies , from dwarf irregulars to large spiral galaxies , and relates the _ specific _ star formation rate @xmath261 ( measured in @xmath262 yr@xmath77 kpc@xmath8 ) to the gas surface density @xmath263 . at a given location in a galaxy
we have : @xmath264 where @xmath265 is the dynamical time of the gas .
the efficiency @xmath53 is @xmath266 for the best fit indicated by kennicutt ( 1998 ) using a large dataset , although such data show a large scatter , especially for faint galaxies : for non - starburst dwarf galaxies , with sfrs @xmath267 yr@xmath77 , @xmath53 ranges between 0.01 and 1 .
we calibrate @xmath53 for our model galaxies by requiring that the satellites , from their formation until the time they enter into the milky way halo , have a constant star formation rate high enough to produce the stellar mass of our initial models ( models lmg2 and hmg2 , see table 1 ) .
therefore @xmath53 may vary depending on their formation epoch .
we consider two formation epochs for the dwarfs , either 12 or 15 gyr ago , as the age of the oldest stars in dsphs falls within that range ( grebel 1999 , hernandz et al .
2000 ) , while we always assume that they entered into the milky way halo 10 gyr ago .
the sfrs and related parameters are reported in the caption of figure 30 . with this procedure ,
the specific sfrs for lsb satellites turn out to be of the same order of those recently derived for local dirrs like sextans a and gr8 using hst photometry ( dohm - palmer et al .
1997 , 1998 ) and are pretty close to those derived for isolated lsb galaxies using n - body / sph simulations ( gerritsen & de blok 1999 ) .
lsbs require high values of @xmath53 because of the low surface density of their disks : instead , for hsbs the efficiencies are closer to the `` best fit '' given by kennicutt ( 1998 ) , which is not surprising since the galaxies in that sample are mostly normal , high - surface brightness disks .
= 8truecm = 8truecm from the time at which we start the simulations up to the final time , we compute , for each output , the specific sfr as a function of radius calculating the values of @xmath263 and @xmath265 in equally spaced radial bins ( in calculating @xmath268 we always take into account the gas mass that has already being consumed at the previous output in a given bin ) .
we then calculate the average sfr in the dwarf integrating over its area ( only out to @xmath269 because the gas is rapidly stripped at larger radii ) and weighting by the gas mass in each bin .
the resulting star formation histories are shown in figure 30 . in the case of the lsb the star formation rate has a periodic behavior , with peaks of activity quite well correlated to each pericenter passage .
the exact number and height of the peaks depends on the efficiency parameter @xmath53 : when this is high , the gas is mainly consumed in the first two bursts and there is only a small amount left when the satellite approaches the primary for the third time .
the small starbursts are separated by a timescale of @xmath270 gyr , roughly the orbital time of the dwarf galaxy .
qualitatively , these star formation histories bear a striking resemblance to those recently derived for leo i and carina by hernandez et al .
( 2000 ) using hst color - magnitude diagrams .
do they also correspond quantitatively ? if we integrate the specific sfrs at the peaks obtained for the lsb over the areas of carina and leo i ( using directly the size of the lsb remnant would be misleading as this corresponds more to an extended dsph like sextans ) , for a tidal radius of @xmath116 750 pc ( see mateo 1998 ) , we obtain , for @xmath271 , a sfr between 150 and 350 @xmath262/myr , pretty close to the values of 150 - 200 @xmath262/myr estimated by hernandz et al .
( 2000 ) . instead , for a higher star formation efficiency the maximum sfrs are sensibly higher than the observational estimate .
in the case of the hsb satellite , only one , very stronger burst occurs after first pericentric passage , depleting all the gas in less than @xmath220 gyr , before the satellite completes the second orbit . because gas consumption is always extremely fast compared to the orbital time
, the results do not depend sensitively on the efficiency parameter @xmath53 .
the maximum star formation rate is almost a factor of 10 higher than in the case of the lsb .
this second type of star formation history is present also among observed early type dwarfs : draco and ursa minor are well known examples and even m31 s des appear to have had one major episode of star formation ( grebel 1998 ; mateo 1998 ) .
supernovae feedback , which is not included in the derived star formation histories , could have a significant effect , especially in the case of hsbs , given their strong burst .
their star formation rate during the burst is of the order of that measured in the dwarf galaxies analyzed by martin ( 1999 ) , which all show strong outflows of gas ( even an lg dirr , sextans a , is included in such a sample sample ) .
mass loss rates of about @xmath272/yr are thought to be associated with the prominent h@xmath53 shells observed .
the velocity of the inflow of gas associated with the bar is of the order of @xmath273 km / s , while the expanding shell produced by supernovae explosions ( which would occur almost instantly compared to the dynamical time of the galaxy ) would be moving at up to 100 km / s ( meurer et al .
1992 ) and could in principle reverse the motion of the gas .
however , maclaw & ferrara ( 1999 ) , using a high - resolution eulerian code to model the ism in a star - bursting dwarf within a massive dark matter halo , have shown that , unless the dwarf has a total mass as low as @xmath274 , the gas mass lost during these outflows is only a few percent of the total mass even for starbursts as intense as @xmath275 erg / s ( comparable to the intensity calculated for the hsb satellite ) .
the holes produced in the gaseous disk recollapse under the influence of the gravity of the galaxy and will settle down again @xmath276 yr after the explosion , switching on again star formation .
we can imagine these process repeating several times as new supernovae explosions occur . as a result
, the same gas mass will be turned into stars on a longer time scale than in our model : we should still see a first peak of activity , but this will be lower , being soon depressed by the first consistent supernovae explosion , and will be then followed by a less intense , extended activity regulated more by feedback than by the bar inflow .
this might explain why in the des , after a first burst , some star formation took place even recently ( grebel 1999 ) .
= 8truecm we can ask how the sfr history of the gr8-like dwarf would like if we added gas .
the disk stellar surface density of gr8 is pretty close to that of the hsbs and this would be likely true even for the gaseous disk . therefore , the specific sfr would be necessarily close to that of hsbs .
a clear bar pattern appears after the second pericenter passage and would likely produce a strong burst .
the star formation in draco and ursa minor seems to have lasted @xmath48 gyr according to hernandz et al .
( 2000 ) and this constraint would be approximately satisfied if we adopt for gr8 a sf history similar to the hsb .
hernandz et al .
( 2000 ) invoke feedback to explain the abrupt truncation of the star formation for ursa minor : in our model this would result automatically from gas consumption . however , the gr8 model has a mass so low to fall in the regime of substantial blow - away of the gas by supernovae according to mac law & ferrara ( 1999 ) . adding the effect of supernovae
would actually match the observations better , lowering the intensity of the burst ( those derived by hernandz et al .
( 2000 ) for draco and ursa minor are substantially weaker ) while still drastically reducing the star formation immediately after it .
in section 4 we have seen that , while remnants of hsbs have little scatter in their structural properties , remnants of lsbs have a large scatter , being very sensible to the orbital configuration .
on orbit with large pericenters , for example , the bar instability is stronger because stripping is less efficient : in this case a star formation history closer to that of hsbs may result .
therefore , a strong prediction of our model is that dsphs must have a large variety of star formation histories , while des should have very similar histories .
the data seems to support this conclusion ( grebel 1999 ; mateo 1998 ) .
= 5truecm from the final density profiles of the gas ( figure 13 ) we can argue that a compact stellar nucleus would form in the remnants of hsbs . a crude conversion of the gas surface density directly into stellar surface density implies that the central surface brightness should be boosted by @xmath220 magnitudes inside a scale of 500 pc assuming the same @xmath277 of the pre - existing stellar component ( this is a conservative choice shortly after the burst has started because the blue luminosity would be higher at that time ) .
such a central luminosity spike would probably resemble that associated to the blue nucleus in ngc205 ( ferguson & binggeli 1994 ) .
the fraction of the newly formed stellar population with respect to the total stellar mass would be @xmath278 in both the lsbs and hsbs , similar to the initial gas fraction .
this is not surprising because gas and pre - existing stars are stripped to the same extent and most of the gas that remains bound is transformed into stars .
however , this means also that if the initial gas / stars ratio in the models were @xmath279 , as in some dirrs ( e.g. sagdig in the lg ) , the satellite could in principle form most of its stars during the interaction . due to the bar driven inflow , young stars in the remnant would be principally located in its central part .
this is what is observed in many lg early type dwarfs , with fornax , carina , sextans and ngc147 being well known examples ( grebel 1999 ) .
the much stronger bursts in hsb satellites should give rise to a higher metal enrichment compared to lsb satellites because the supernovae rate should be higher .
as tidal stirring necessarily implies that the remnants of the robust hsbs are more luminous than the remnants of the fragile lsbs , one would expect a correlation between metallicity and luminosity , similar to that observed in the lg , where bright des are more metal - rich than dsphs ( mateo 1998 ) . on the other end
, a higher metallicity spread should be present in dsphs as a consequence of their star formation being more extended in time .
radial metallicity gradients , with the more metal rich population sitting in the center of the galaxy , as observed in fornax ( grebel 1999 ) , are also a natural consequence of the present model .
figure 31 shows the final cumulative profile of the cold gas component in the tidally stirred dwarfs once consumption by star formation is taken into account .
only the results for the lsb are shown as all the gas is consumed in the hsb , notwithstanding the chosen efficiency .
the final @xmath280 in the lsb remnant is @xmath281 for @xmath271 and @xmath282 for @xmath283 .
the limits derived for the lsb remnant for the lower value of @xmath53 , which yields star formation rates closer to those observed for dsphs , are slightly higher than the limits inferred for @xmath284 in dsphs ( mateo 1998 ) .
however , our simulations lack heating sources , like supernovae , that could have ionized or even blown out some of the remaining gas .
= 8truecm the final distribution of the unbound gas yields some interesting predictions . leading and trailing streams of gas form that parallel the stellar streams .
these keep following the satellite s trajectory in the primary potential for all the duration of the simulations ( more than 7 gyr ) . in the case of the lsb satellite
the gaseous streams comprise more than 60 % of the total initial gas mass .
such extra - tidal gas is never reaccreted by the dwarf galaxy , contrary to recent claims ( lin & murray 2000 ) : particles that end up in the remnant have always been inside the final tidal radius , as was verified by tracing back into time the position of the particles . however , the streams tend to clump at apocenter , where the orbital velocity is low , and chance alignments along the line of sight are possible , like in the case of the stellar streams ( section 5.4 ) . when viewed from within the orbital plane , the streams _ projected onto the plane of the sky
_ could look like two gas clouds symmetrically distributed around the bound stellar component ( figure 32 ) .
the resulting configuration is very reminiscent of that of the outer hi distribution in sculptor ( carignan et al .
in addition , similar sistemic velocities would be measured for the galaxy and the clouds , as these are moving on the same orbit : these results could explain the origin of the outer gas component recently discovered around several lg dsphs ( blitz & robishaw 2000 ) .
using stellar population synthesis models ( bruzual & charlot 1993 ) we calculated the evolution of the luminosity and of the stellar mass - to - light ratio of the remnants according to the derived star formation histories .
we adopt synthesis models for stellar populations with a metallicity of @xmath285 the solar value .
gerritsen & de blok ( 1999 ) find such low values for the metallicity of lsb galaxies and dsphs in the local group have even lower metallicities in their older stellar populations ( mateo 1998 ) .
given a star formation rate @xmath286 , the evolution of the total luminosity @xmath287 of the galaxy in the b band is given by : @xmath288 where @xmath289 is the luminosity contributed by stars of age @xmath290 .
we simply assume @xmath291 , where @xmath292 is the constant sfr prior to the first pericenter passage , equal to @xmath272 yr@xmath77 , and @xmath293 a exp@xmath294 represent the bursts ( @xmath295=0 for the hsb ) .
the time and normalization constants of the bursts are derived by naively fitting the star formation histories obtained for the highest value of @xmath53 ( see previous section ) in order to evaluate an upper limit for the final luminosity . in figure 33
we can see the resulting evolution of the b band stellar mass - to - light ratio , @xmath277 , for both the lsb and the hsb .
the final @xmath277 are @xmath45 in both cases after 7 - 8 gyr ( the earlier truncation of the star formation in the hsb is compensated by the much stronger burst ) decreasing by more than a factor of 2 with respect to their initial value
. we can now use such information to compute the luminosities of the remnants from @xmath277 and their final stellar masses .
being conservative , we consider the value of @xmath277 reached after 7 gyr , which is equal to 5 , except in the case of gr8 , where we consider the value reached after 10 gyr , @xmath296 , since in our scenario we assumed a much earlier infall time for the latter , corresponding to @xmath106 .
the surface brightness can be obtained as well : the resulting profiles were actually shown in section 5 . in section 5.3
we gave lower limits on the final value of @xmath219 .
we can now compute more precisely the final @xmath219 by multiplying the obtained @xmath277 times the ratio @xmath206 found in the remnants .
the resulting @xmath219 are in the range 6 - 30 , except for the remnant of gr8 , that would have @xmath297 ( @xmath206 is @xmath298 ) , matching well the values inferred for draco and ursa minor .
= 8truecm as shown in mayer et al .
( 2001 ) , it is possible to construct the fundamental plane for the remnants coupling the information directly provided by the simulations with the model for the photometric evolution described in this section .
such fundamental plane reproduces well that of observed early - type dwarfs ( ferguson & binggeli 1994 ) .
here we show the correlation between luminosity and surface brightness , the @xmath299 relation ( figure 34 ) , which also matches well the observations ( ferguson & binggeli 1994 ) .
the range of surface brightness and luminosity spanned by the remnants of lsbs and hsbs corresponds to that spanned by , respectively , dsphs and des ( see mateo 1998 ) .
we have shown that disk - like dwarf galaxies resembling nearby dirrs can be transformed into dsphs and des as a result of 2 - 3 strong tidal shocks at the pericenters of their orbits within the primary halo .
our model naturally leads to the morphology - density relation observed in the local group ; present - day lg dirrs , being close to the boundary of the lg zero velocity surface ( van den bergh 1999 ) , have just decoupled from the hubble flow and have not yet suffered any tidal forcing by the primary galaxies , while dsphs are the descendants of dwarf disk galaxies that crossed the boundary of the primary halos several gyr ago .
the transformation results from a combination of mass loss by tidal stripping plus removal of angular momentum and vertical heating due tidally induced dynamical instabilities .
the different impact of mass stripping and of triggered instabilities in hsb dirrs versus lsb dirrs naturally explains the different structural properties of des versus dsphs .
the fact that , at least in the local group , the number of des is lower compared to that of dsphs ( only 3 around m31 against nearly 20 dsphs in the whole lg ) can be understood in terms of the skewness of the number density of disk galaxies towards lower surface brightness , especially at low luminosities ( sprayberry et al .
1997 ; bothun 1997 ; oneil & bothun 2000 ) .
photometric surveys of other nearby groups will be useful to ascertain if the relative proportion of dsphs versus des is a general trend among faint dwarf satellites . according to our model
this can be explained with a prevalence of objects with shallow halo profiles ( namely lsb dirrs ) among the progenitors of early - type dwarfs .
this might have an impact on our understanding of structure formation .
in fact , in cdm cosmogonies the average central density of halos on the mass scale of dwarfs is higher than that assumed in lsb models ( the gr8 model being a notable exception ) , the latter being closer to the typical central density of small halos in wdm models ( bode et al .
2000 ; eke et al .
2000 ) . on the other end , the origin of the extremely dark matter dominated dsphs , draco and ursa minor , is well explained assuming a cdm scenario .
in fact , in the latter model one expects that the dwarf satellites forming at high redshift have a typical central dark matter density as high as that of the faint dirr gr8 .
these small galaxies are very robust to tides , but if they fall at @xmath300 into the primary halo their orbital times are so small that they can double the number of pericenter passages relative to the satellites infalling at @xmath301 .
by @xmath75 they transmute into tiny spheroidal galaxies but maintain very high dark matter contents , just like draco and ursa minor ; one can actually use such high density dwarfs to place a lower limit on the mass of the dark matter particle in wdm models ( dalcanton & hogan 2000 ) .
in this paper we have shown that _
tidal stirring _ is effective if satellites have sufficiently low orbital times , notwithstanding the moderate eccentricities of their orbits .
lsb satellites on circular orbits very close to the milky way can even disrupt due to their large , low density disks and will produce long - lasting density enhancements that resemble the dwarf spheroidal sagittarius .
however , the farthest dsphs , leo i and leo ii , must be moving on fairly eccentric orbits ( apo / peri @xmath302 ) in order to be transformed .
these distant dwarfs place also a constraint on the structure of the primary halo , which must be as large and massive as expected in hierarchical models for them to have sufficiently small orbital times .
our evolutionary mechanism is able to explain the difference in gas content between dwarf spheroidals and dwarf irregulars as well as the recent observations of extra - tidal gas around some lg dsphs .
in addition , we obtained both periodic star formation histories and single - burst star formation histories for , respectively , lsb and hsb satellites , reproducing the variety observed among lg early - type dwarfs . with tidal stirring operating in a hierarchical structure formation scenario
, we expect dsphs closer to the primary galaxies to have stars typically older then those in more distant ones because they should have , on average , an earlier infall time and thus an earlier bursting phase : van den bergh ( 1994 ) has pointed out that there are indications of such a correlation among lg dsphs .
van den bergh ( 2000 ) and grebel ( 1999 ) have recently noted that some peripheral globular clusters in lg dsphs ( e.g. fornax ) are younger ( less than 10 gyr ) than globular clusters in the outer halo of the galaxy . in our scenario
they could have formed in the strongest phase of the tidal shock , immediately following pericenter passage , but in this case we would expect to find them quite close to the center : subsequently , dynamical friction can only take them even closer to the center , apparently leading to a conflict with observations .
however , oh , lin & richer ( 2000 ) have found that tidal mass loss provoked by the primary galaxy would cause a diffusion of the clusters orbits towards the outer regions as they preserve their radial and azimuthal actions during the gradual decline of the potential of the dwarf .
therefore , we can argue that tidal stirring might provide a mechanism that explains , at the same time , the formation of globular clusters in dsphs and their present spatial distribution .
the stars stripped in both the hsb and lsb should be prevalently old ( with an age @xmath303 gyr ) , as we know that the new star formation takes place in the inner , bound regions of the satellite . as a result
, we expect the outer stellar halo of the milky way to be abundant in old , low metallicity stars .
if , as suggested by the results of the lsb runs , we assume that carina , fornax , leo i , leo ii , sculptor , sextans and sagittarius have typically lost @xmath304 of their original stellar mass , we find out that dwarf spheroidals should have contributed at least @xmath305 solar masses to the old stellar halo
. these should be still recognizable as part of distinct streams tracing the orbits of the satellites , with a maximum surface brightness ( at the turning points of the orbits ) of just 30 mag arcsec@xmath8 ( in the b band ) .
spectroscopic evidence for stellar streams from the dsph carina has been claimed ( majewski et al .
2000 ) and the northern stream of sagittarius or a new stream associated to another disrupted nearby dwarf has recently been observed ( martinz - delgado et al .
future astrometric missions , like sim and gaia ( gilmore et al .
1998 ; helmi et al .
2000 ) should reveal such faint features and will also carry out high - quality measurements of proper motions for many satellites of the milky way , thus providing a test for the orbital configurations used in this model .
stellar streams projected on the plane of the sky close to the line of sight can produce subtle observational effects in the case of fragile lsb satellites , thereby inflating the measured velocity dispersions .
their signature should be an apparent rotation of the dwarf and could have been observed in ursa minor ( hargreaves et al .
1994a ) . instead
, velocity dispersion anisotropy can not inflate significantly the measures of the mass - to - light ratios . from the evolution of the stellar mass - to - light ratio
, it turns out that hsb satellites have @xmath277 as low as @xmath199 during the bursting phase : the high blue luminosity would originate mainly from the central part of the galaxies , which could thus resemble the blue compact dwarfs(bcds ) observed by guzmn et al .
( 1997 ) at intermediate redshift ( @xmath306 ) .
the velocity widths of these galaxies correspond to those of the largest among our hsb satellites ( between 50 and 75 km / s ) and the total blue luminosities would also be comparable ( model hm1 reaches a luminosity of about @xmath307 in this phase ) .
interestingly , guzmn et al . (
1997 ) conclude that fading would transform bcds into objects similar to ngc205 and the other des after several gyr and that none of them could fade so much as to resemble dsphs .
the latter conclusion is very reminiscent of our claim that hsb dirrs are the progenitors of des and not of the faint dsphs .
future investigations aimed at searching for massive galaxies in the vicinity of blue compact dwarfs may shed light on the connection with tidally stirred dwarfs .
the authors thank g.gavazzi , c.gallart , m.vietri , r.bower , a.helmi and s.ghigna for stimulating discussions .
simulations were carried out at the cineca ( bologna ) and arsc ( fairbanks ) supercomputing centers .
rosati , p. , borgani , s. , della ceca , r. , stanford , a. , eisenhardt , p. , lidman , c. , large scale structure in the x - ray universe , proceedings of the 20 - 22 september 1999 workshop , santorini , greece , eds .
plionis , m. & georgantopoulos , i. , atlantisciences , paris , france , p.13 c|c|c|c|c|c|c|c|c|c|c model & @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath51 & @xmath58 & @xmath26 & @xmath312 & @xmath313 & @xmath94 + & & & @xmath314 & & ( kpc ) & ( kpc ) & ( km s@xmath315 & & & + hm1 ( hsb ) & 0.032 & 0.016 & 6 & 0.02 & 140 & 2 & 75 & 22 & 2 & 2 hm2 ( hsb ) & 0.01 & 0.016 & 6 & 0.02 & 95 & 1.35 & 50 & 22.5 & 2 & 2 hmg2 ( hsb ) & 0.01 & 0.016 & 6 & 0.02 & 95 & 1.35 & 50 & 22.5 & 2 & 2 hz ( hsb ) & 0.013 & 0.016 & 6 & 0.02 & 49.5 & 0.7 & 75 & 21.5 & 2 & 2 hm1rc03 ( hsb ) & 0.032 & 0.016 & 10 & 0.006 & 140 & 2 & 75 & 22 & 2 & 2.7 lm1 ( lsb ) & 0.032 & 0.016 & 12 & 0.036 & 140 & 4.8 & 75 & 24 & 2 & 3 lmh ( lsb,@xmath316 ) & 0.032 & 0.016 & 12 & 0.036 & 140 & 4.8 & 75 & 24 & 2 & 3 lm2 ( lsb ) & 0.01 & 0.016 & 12 & 0.036 & 95 & 3.2 & 50 & 24.5 & 2 & 3 lmg2 ( lsb ) & 0.01 & 0.016 & 12 & 0.036 & 95 & 3.2 & 50 & 24.5 & 2 & 3 lz ( lsb ) & 0.013 & 0.016 & 12 & 0.036 & 140 & 1.69 & 75 & 23.5 & 2 & 3 lm1rc03 ( lsb ) & 0.032 & 0.016 & 15 & 0.0108 & 140 & 4.8 & 75 & 24 & 2 & 5 lm1q4 ( lsb ) & 0.032 & 0.016 & 12 & 0.036 & 140 & 4.8 & 75 & 24 & 4 & 3 gr8 & @xmath317 & @xmath318 & 32 & 0.014 & 7.8 & 0.076 & 17 & 22.5 & 2 & 6 the structural parameters of the dwarf galaxy models are shown .
@xmath319 is the mass of the milky way halo , @xmath57 is the mass of the satellite s disk , while @xmath72 is the virial mass of the halo of the satellite ; @xmath51 is the truncation radius of its halo (= @xmath73 ) , @xmath58 is its disk scale length and @xmath52 is its halo core radius .
the circular velocity @xmath26 is measured at the virial radius of the satellite ( i.e. @xmath320 ) .
the mass - to - light ratio at the optical radius (= @xmath321 ) is @xmath322 and @xmath312 is the b band central surface brightness in mag arcsec@xmath8 .
model lmg2 and hmg2 have a gaseous component in the disk , with mass @xmath323 .
the stability parameter @xmath94 is estimated at @xmath58 , while @xmath88 is estimated at the optical radius .
c|c|c|c|c run & model & @xmath324 & @xmath325 & @xmath326 + & & ( kpc ) & ( kpc ) & degrees + l01 & lm1 & 40 & 360 & 0 l02 & lm1 & 80 & 360 & 0 l03 & lm1 & 40 & 360 & 180 l04 & lm1 & 40 & 360 & 40 l05 & lm1 & 40 & 360 & 63 l06 & lm1 & 80 & 360 & 63 l07 & lm1 & 40 & 360 & 125 l08 & lm1 & 40 & 360 & 150 l09 & lm1 & 240 & 360 & 63 l10 & lm2 & 40 & 360 & 0 l11 & lm2 & 80 & 360 & 0 l12 & lm2 & 80 & 360 & 180 l13 & lm2 & 40 & 360 & 40 l14 & lm2 & 40 & 360 & 63 l15 & lm2 & 80 & 360 & 90 l16 & lm2 & 40 & 360 & 130 l17 & lm2 & 125 & 250 & 63 l18 & lm2 & 200 & 360 & 90 l19 & lm1 & 75 & 120 & 63 l20 & lm1 & 75 & 75 & 63 l21 & lm1rc03 & 80 & 360 & 0 l22 & lmh & 80 & 360 & 0 l23 & lm1q4 & 80 & 360 & 0 l24 & lm1 ( minimal milky way potential ) & 25 & 220 & 40 l25 & lm1 ( miyamoto disk included ) & 40 & 360 & 40 l26 & lmg2 & 40 & 360 & 40 initial conditions of the runs .
we indicate the galaxy model used , the pericenter distance @xmath324 , the apocenter distance @xmath325 , and the angle between the spin of the disk and the orbital angular momentum ( @xmath326 ) .
c|c|c|c|c run & model & @xmath324 & @xmath325 & @xmath326 + & & ( kpc ) & ( kpc ) & degrees + lz1 & lz & 40 & 360 & 0 lz2 & lz & 80 & 360 & 0 lz3 & lz & 40 & 360 & 180 lz4 & lz & 40 & 360 & 40 lz5 & lz & 40 & 360 & 63 lz6 & lz & 40 & 360 & 125 lz7 & lz & 75 & 120 & 63 h01 & hm1 & 40 & 360 & 0 h02 & hm1 & 80 & 360 & 0 h03 & hm1 & 40 & 360 & 180 h04 & hm1 & 40 & 360 & 40 h05 & hm1 & 40 & 360 & 63 h06 & hm1 & 40 & 360 & 125 h07 & hm1 & 40 & 360 & 150 h08 & hm2 & 40 & 360 & 40 h09 & hm2 & 125 & 250 & 40 h10 & hm2 & 75 & 120 & 63 h11 & hm1rc03 & 50 & 120 & 0 h12 & hmg2 & 40 & 360 & 40 hz1 & hmz & 40 & 360 & 0 hz2 & hmz & 40 & 360 & 180 hz3 & hmz & 40 & 360 & 40 hz4 & hmz & 40 & 360 & 63 hz5 & hmz & 40 & 360 & 125 gr81 & gr8 & 40 & 360 & 63 gr82 & gr8 & 12 & 110 & 63 initial conditions of the runs .
we indicate the galaxy model used , the pericenter distance @xmath324 , the apocenter distance @xmath325 , and the angle between the spin of the disk and the orbital angular momentum ( @xmath326 ) . | we present results from high - resolution n - body / sph simulations of rotationally supported dwarf irregular galaxies moving on bound orbits in the massive dark matter halo of the milky way .
the dwarf models span a range in disk surface density and the masses and sizes of their dark halos are consistent with the predictions of cold dark matter cosmogonies .
we show that the strong tidal field of the milky way determines severe mass loss in their halos and disks and induces bar and bending instabilities that transform low surface brightness dwarfs ( lsbs ) into dwarf spheroidals ( dsphs ) and high surface brightness dwarfs ( hsbs ) into dwarf ellipticals ( des ) in less than 10 gyr .
the final central velocity dispersions of the remnants are in the range @xmath0 km / s and their final @xmath1 falls to values @xmath2 , matching well the kinematics of early - type dwarfs .
the transformation requires the orbital time of the dwarf to be @xmath3 gyr , which implies a halo as massive and extended as predicted by hierarchical models of galaxy formation to explain the origin of even the farthest dsph satellites of the milky way , leo i and leo ii .
we show that only dwarfs with central dark matter densities as high as those of draco and ursa minor can survive for 10 gyr in the proximity of the milky way : this is naturally achieved within hierarchical models , where the densest objects should have small orbital times due to their early formation epochs .
part of the gas is stripped and part is funneled to the center due to the bar , generating one strong burst of star formation in hsbs and smaller , multiple bursts in lsbs .
therefore , the large variety of star formation histories observed in lg dsphs naturally arises because different types of dirr progenitors respond differently to the external perturbation of the milky way .
our evolutionary model automatically explains the morphology - density relation observed in the lg and in other nearby loose groups .
extended low - surface brightness stellar and gaseous streams originate from lsbs and follow the orbit of the dwarfs for several gyr . due to their high velocities , unbound stars projected along the line of sight can lead to overestimate the mass - to - light ratio of the bound remnant by a factor @xmath4 , but this does not eliminate the need of extremely high dark matter contents in some of the dsphs . |
consider a @xmath24 immersion @xmath25 , where @xmath26 is an open set . here
@xmath27 is a hypersurface element ( or a position vector ) , and @xmath28 .
the first fundamental form is a symmetric , positive definite metric tensor of @xmath29 , given by @xmath32 .
its matrix elements can also be expressed as @xmath33 , where @xmath34 is the euclidean inner product in @xmath35 , @xmath36 .
let @xmath37 be the unit normal vector given by the gauss map @xmath38 , @xmath39 where the cross product in @xmath35 is a generalization of that in @xmath40 . here
@xmath41 is the normal space of @xmath29 at point @xmath42 .
the vector @xmath43 is perpendicular to the tangent hyperplane @xmath44 at @xmath45 .
note that @xmath46 , the tangent space at @xmath45 . by means of the normal vector @xmath43 and tangent vector @xmath47 ,
the second fundamental form is given by @xmath48 the mean curvature can be calculated from @xmath49 where we use the einstein summation convention , and @xmath50 .
let @xmath51 be an open set and suppose @xmath52 is compact with boundary @xmath53 .
let @xmath54 be a family of hypersurfaces indexed by @xmath4 , obtained by deforming @xmath29 in the normal direction according to the mean curvature .
explicitly , we set @xmath55 we wish to iterate this leading to a minimal hypersurface , that is @xmath56 in all of @xmath57 , except possibly where barriers ( atomic constraints ) are encountered . for our purpose ,
let us choose @xmath58 , where @xmath1 is a function of interest .
we have the first fundamental form : @xmath59 the inverse matrix of @xmath60 is given by @xmath61 where @xmath62 is the gram determinant .
the normal vector can be computed from eq .
( [ normal ] ) @xmath63 the second fundamental form is given by @xmath64 i.e. , the hessian matrix of @xmath65 . | we introduce a novel concept , the minimal molecular surface ( mms ) , as a new paradigm for the theoretical modeling of biomolecule - solvent interfaces .
when a less polar macromolecule is immersed in a polar environment , the surface free energy minimization occurs naturally to stabilizes the system , and leads to an mms separating the macromolecule from the solvent . for a given set of atomic constraints ( as obstacles ) ,
the mms is defined as one whose mean curvature vanishes away from the obstacles .
an iterative procedure is proposed to compute the mms .
extensive examples are given to validate the proposed algorithm and illustrate the new concept .
we show that the mms provides an indication to dna - binding specificity .
the proposed algorithm represents a major step forward in minimal surface generation .
the stability and solubility of macromolecules , such as proteins , dnas and rnas , are determined by how their surfaces interact with solvent and/or other surrounding molecules .
therefore , the structure and function of macromolecules depend on the features of their molecule - solvent interfaces @xcite . molecular surface was proposed @xcite to describe the interfaces and has been applied to protein folding @xcite , protein - protein interfaces @xcite , protein surface topography @xcite , oral drug absorption classification @xcite , dna binding and bending @xcite , macromolecular docking @xcite , enzyme catalysis @xcite , calculation of solvation energies @xcite , and molecular dynamics @xcite .
it is of paramount importance to the implicit solvent models @xcite .
however , the molecular surface model suffers from it being probe dependent , non - differentiable , and being inconsistent with free energy minimization .
minimal surfaces are omnipresent in nature .
their study has been a fascinating topic for centuries @xcite .
french geometer , meusnier , constructed the first non - trivial example , the catenoid , a minimal surface that connects two parallel circles , in the 18th century . in 1760 ,
lagrange discoved the relation between minimal surfaces and a variational principle , which is still a cornerstone of modern mechanics .
plateau studied minimal surfaces in soap films in the mid - nineteenth century .
in liquid phase , materials of largely different polarizabilities , such as water and oil , do not mix , and the material in smaller quantity forms ellipsoidal drops , whose surfaces are minimal subject to the gravitational constraint .
the self - assembly of minimal cell membrane surfaces in water has been discussed @xcite .
the schwarz p minimal surface is known to play a role in periodic crystal structures @xcite .
the formation of @xmath0-sheet structures in proteins is regarded as the result of surface minimization on a catenoid @xcite .
a minimal surface metric has been proposed for the structural comparison of proteins @xcite . however , to the best of our knowledge , a natural minimal surface that separates a less polar macromolecule from its polar environment such as the water solvent has not been considered yet . the objective of this report is to introduce the theory of and algorithm to generate minimal molecular surfaces ( mmss ) .
since the surface free energy is proportional to the surface area , a mms contributes to the molecular stability in solvent .
therefore , there must be a mms associated with each stable macromolecule in its polar environment .
although minimal surfaces are often generated by evolving surfaces with predetermined curve boundaries @xcite , there is no algorithm available that generates minimal surfaces with respect to obstacles , such as atoms . here
, we develop such an algorithm based on the theory of differential geometry @xcite . for a given initial function @xmath1 that characterizes domain encompassing the biomolecule of interest
, we consider an evolution driving by the mean curvature @xmath2 @xmath3 where @xmath4 is a small parameter , @xmath5 is the gram determinant , and @xmath6 .
our procedure involves iterating eq .
( [ evolut ] ) until @xmath7 everywhere except for certain protected boundary points where the mean curvatures take constant values .
physically , the vanishing of the mean curvature is a natural consequence of surface free energy minimization . consider the surface free energy of a molecule as @xmath8 , where @xmath9 is boundary of the molecule , @xmath10 the energy density and @xmath11 .
the energy minimization via the first variation leads to the euler lagrange equation , @xmath12 where @xmath13 . for a homogeneous surface , @xmath14 , a constant , eq .
( [ el ] ) leads to the vanishing of the mean curvature @xmath15 . for a given set of atomic coordinates , we prescribe a step function initial value for @xmath1 , i.e. , a non - zero constant @xmath16 inside a sphere of radius @xmath17 about each atom and zero elsewhere .
alternatively , a gaussian initial value can be placed around each atomic center .
the value of @xmath1 is updated in the iteration except for at obstacles , i.e. , a set of boundary points given by the collection of all of the van der waals sphere surfaces or any other desired atomic sphere surfaces . here
@xmath18 and @xmath19 can be approximated by any standard numerical methods . for simplicity , we use the standard second order central finite difference . due to the stability concern , we choose @xmath20 , where @xmath21 is the smallest grid spacing .
the mms is differentiable , probe independent , and consistent with the surface free energy minimization .
[ cols="^,^,^ " , ] finally , we employ our mms to study the mechanism of molecular recognition in protein - dna interactions .
nmr and molecular dynamics studies suggest that antennapedia achieves specificity through an ensemble of rapidly fluctuating dna contacts @xcite . while x - ray structure indicates a well - defined set of contacts due to side chains constraints @xcite . in the present work ,
we reveal flat contacting interfaces which stabilizing the protein - dna complex . figs . [ fig.9ant2](a ) and [ fig.9ant2](c ) depict the mmss of antennapedia and dna ( pdb i d : 9ant ) , generated by using @xmath22 and @xmath23 .
clearly , the binding site of the dna ( middle groove ) has a large facet , which is absent from the top and bottom grooves of the dna .
interestingly , the mms of the protein exhibits a complimentary facet . for a comparison ,
the molecular surfaces ( mss ) generated by using the program msms @xcite with the same set of van der waals radii and a probe radius of 1.5 are depicted in figs .
[ fig.9ant2 ] ( b ) and [ fig.9ant2 ] ( d ) .
apparently , it is very difficult to recognize the complementary binding interfaces from mss .
it is interesting to note that the mms also better reveals the skeleton of the dna s double helix structure . to quantitate the affinity at the contacting site , we compute the mean distance between the mmss of the protein and dna by using about 7200 surface vertices over the binding domain . a small mean distance of 0.4054 unveils a close contact between two facets .
relatively small standard deviation of 0.3401 indicates the smoothness of the contacting facets .
in contrast , inconclusive mean ( 0.8697 ) and standard deviation ( 0.5818 ) were found from the corresponding mss .
this study indicates the great potential of the proposed mms for biomolecular binding sites prediction and recognition .
we have introduced a novel concept , the minimal molecular surface ( mms ) , for the modeling of biomolecules , based on the speculation of free energy minimization for stabilizing a less polar molecule in a polar solvent .
the mms is probe independent , differentiable , and consistent with surface free energy minimization .
a novel hypersurface approach based on the theory of differential geometry is developed to generate the mmss of arbitrarily complex molecules .
numerical experiments are carried out on few - atom and many - atom systems to demonstrate the proposed method .
it is believed that the proposed mms provides a new paradigm for the studies of surface biology , chemistry and physics , in particular , for the analysis of stability , solubility , solvation energy , and interaction of macromolecules , such as proteins , membranes , dnas and rnas .
it has potential applications not only in science , but also in technology , such as vehicle design and packaging problems . 99 l.a .
kuhn , m. a. siani , m. e. pique , c. l. fisher , e. d. getzoff and j. a. tainer , the interdependence of protein surface topography and bound water molecules revealed by surface accessibility and fractal density measures , _
j. mol .
biol .
_ , * 228 * , 13 - 22 ( 1992 ) .
f.m .
richards , areas , volumes , packing and protein structure , _ annu
. rev .
biophys .
bioeng .
_ , * 6 * , 151 - 176 ( 1977 ) .
m.l .
connolly , analytical molecular surface calculation . ,
_ j. appl .
crystallogr .
_ , * 16 * , 548 - 558 ( 1983 ) .
r.s .
spolar and m.t .
jr .
record , coupling of local folding to site - specific binding of proteins to dna , _ science _ , * 263 * , 777 - 184 ( 1994 ) .
p.b .
crowley and a. golovin , cation - pi interactions in protein - protein interfaces , _ proteins - struct .
func .
bioinf .
_ , * 59 * , 231 - 239 ( 2005 ) . c.a.s .
bergstrom , m. strafford , l. lazorova , a. avdeef , k. luthman and p. artursson , absorption classification of oral drugs based on molecular surface properties , _ j. medicinal chem .
_ , * 46 * , 558 - 570 ( 2003 ) . a.i .
dragan , c.m .
read , e.n .
makeyeva , e.i .
milgotina , m.e.a .
churchill , c. crane - robinson and p.l .
privalov , dna binding and bending by hmg boxes : energetic determinants of specificity , _
j. mol .
biol .
_ , * 343 * , 371 - 393 ( 2004 ) .
r.m .
jackson and m.j .
sternberg , a continuum model for protein - protein interactions : application to the docking problem , _ j. mol .
biol .
_ , * 250 * , 258 - 275 ( 1995 ) .
v.j .
licata and n.m .
allewell , functionally linked hydration changes in escherichia coli aspartate transcarbamylase and its catalytic subunit , _ biochemistry _ , * 36 * , 10161 - 10167 ( 1997 ) . t.m .
raschke , j. tsai and m. levitt , quantification of the hydrophobic interaction by simulations of the aggregation of small hydrophobic solutes in water , _ proc .
natl .
acad .
sci .
usa _ , * 98 * , 5965 - 5969 ( 2001 ) .
b. das and h. meirovitch , optimization of solvation models for predicting the structure of surface loops in proteins , _ proteins _ ,
* 43 * , 303 - 314 ( 2001 ) .
j. warwicker and h.c .
watson , calculation of the electric - potential in the active - site cleft due to alpha - helix dipoles , _ j. mol .
biol .
_ , * 154 * , 671 - 679 ( 1982 ) .
b. honig and a. nicholls , classical electrostatics in biology and chemistry , _ science _ , * 268 * , 1144 - 1149 ( 1995 ) .
s. andersson , s.t .
hyde , k. larsson and s. lind , minimal surfaces and structures from inorganic and metal crystals to cell membranes and bio polymers , _ chem .
rev . _ , * 88 * , 221 - 242 ( 1998 ) .
m.w .
anderson , c.c .
egger , g.j.t .
tiddy , j.l .
casci and k.a .
brakke , a new minimal surface and the structure of mesoporous silicas , _ angew .
chem .
int .
ed .
_ , * 44 * , 3243 - 3248 ( 2005 ) .
d. pociecha , e. gorecka , n. vaupotic , m. cepic and j. mieczkowski , spontaneous breaking of minimal surface condition : labyrinths in free standing smectic films , _ phys .
rev .
lett .
_ , * 95 * , no . 207801 ( 2005 ) .
j.m .
seddon and r.h .
templer , cubic phases of self - assembled amphiphilic aggregates , _ philos .
t. royal soc .
london ser .
a - math .
phys .
engng .
sci .
_ , * 244 * , 377 - 401 ( 1993 ) .
chen bl , eddaoudi m , hyde st , okeeffe m , yaghi om , interwoven metal - organic framework on a periodic minimal surface with extra - large pores , _ science _ , * 291 * , 1021 - 1023 ( 2001 ) .
e. koh and t. kim , minimal surface as a model of beta - sheets , _ prot .
struct .
func .
bioinf .
_ , * 61 * , 559 - 569 ( 2005 )
. a. falicov and f.e .
cohen , a surface of minimum area metric for the structural comparison of proteins , _
j. mole .
biol .
_ , * 258 * , 871 - 892 ( 1996 ) .
d.l .
chopp , computing minimal - sufaces via level set curvature flow , _ j. comput
. phys .
_ , * 106 * , 77 - 91 ( 1993 ) .
t. cecil , a numerical method for computing minimal surfaces in arbitrary dimension , _
j. comput .
phys .
_ , * 206 * , 650 - 660 ( 2005 )
. a. gray , modern differential geometry of curves and surfaces with mathematica , _ second edition _ , ( crc press , boca raton , 1998 ) .
m. billeter , homeodomain - type dna recognition , _ progr .
biophys .
mol .
biol .
_ , * 66 * , 211 - 225 ( 1996 ) .
e. fraenkel and c.o .
pabo , comparison of x - ray and nmr structures for the antennapedia homeodomain / dna complex , _ nature struc .
mol .
biol .
_ , * 5 * , 692 - 697 ( 1998 ) .
m.f .
sanner , a.j . olson and j.c .
spehner , reduced surface : an efficient way to compute molecular surfaces , _ biopolymers _ , * 38 * , 305 - 320 ( 1996 ) . |
molecular magnet materials have attracted a great deal of interest in recent years , exhibiting fascinating properties such as cooperative phenomena , magnetic memory , quantum tunneling , and unusual relaxation behavior that are most commonly associated with mesoscopic solids .
one prototype single molecule magnet is [ mn@xmath0o@xmath0(ch@xmath2coo)@xmath3(h@xmath4o)@xmath5@xmath62ch@xmath2cooh@xmath6 4h@xmath4o , denoted mn@xmath0 .
it consists of eight mn@xmath7 ( @xmath8 ) and four mn@xmath9 ( @xmath10 ) ions , held together by oxygen atoms , acetate ligands , and waters of crystallization ; the ferrimagnetic spin arrangement yields @xmath11=10 @xcite .
mn@xmath0 crystallizes in a tetragonal lattice , with weak exchange coupling and no long range magnetic ordering @xcite .
though a number of efforts have been made to understand the energetics of the mn@xmath0 system , a quantitative theory is still lacking .
recent prospective hamiltonians include anisotropy , zeeman splitting , spin - phonon interaction , and transverse terms , as well as spin operators up to fourth order @xcite .
mn@xmath0 initially attracted attention due to the striking steps and hysteresis loop in the magnetization , indicative of quantum tunneling . at present , these steps are thought to result from the double well potential that separates spin states ; magnetization reorientation transitions have optimal probability when the `` spin up '' and `` spin down '' levels of the different magnetic quantum numbers align with applied magnetic field @xcite .
high field epr@xcite , neutron scattering@xcite , and sub - mm @xcite techniques have been used to measure the excitation energies between levels in these magnetic clusters .
the highest energy excitation ( m@xmath12=10 @xmath13 m@xmath12=9 ) occurs in the very far infrared , near 10 @xmath1 ( 300 ghz ) . that the magnetic dipole transition energies in the mn@xmath0 system are irregular ( especially near the top of the anisotropy barrier )
clearly shows the presence of higher than second order terms in the spin hamiltonian@xcite .
early heat capacity measurements revealed the irreversible / reversible effects between the two wells below / above the blocking temperature@xcite .
the exact value of the blocking temperature , @xmath14 ( @xmath15 3 k ) , depends on the probe , due to the dynamic nature of the blocking process . above 3 k ,
the magnetization relaxation is exponential in time and reversible ; this is the thermally activated regime .
notable deviations from exponential relaxation are observed below 2 k @xcite . despite an explosion of interest in the low - energy quantum behavior ,
little is known about the vibrational characteristics of mn@xmath0 and many other prototypical molecular magnet materials .
spin - phonon coupled modes have been investigated in other transition metal oxides such as euo , lamno@xmath2 , @xmath16@xmath17-nav@xmath4o@xmath18 , and cuo ; the frequencies at which they appear correspond to vibrational modulation of the superexchange integral in the materials @xcite , rather than the aforementioned low - energy , phonon - assisted relaxation processes . in order to provide further information on the electrodynamic response of single molecule molecular magnets
, we have measured the far - infrared spectra of mn@xmath0 as a function of both temperature and applied magnetic field in the thermally activated regime .
we use this data to assess spin - phonon coupling in this material .
high quality single crystals of mn@xmath0 were synthesized following the original procedure described by lis@xcite .
the single crystals of mn@xmath0 were ground with paraffin at 77 k to prepare pellets of various concentrations suitable for transmittance measurements in the far infrared @xcite . a sample with @xmath153.5% of mn@xmath0 by mass proved to be optimal for most of the frequency range under investigation . a more concentrated pellet ( @xmath1585% )
was used for measurements between 30 and 110 @xmath1 .
infrared transmission measurements were performed in our laboratory and at the national high magnetic field laboratory ( nhmfl ) in tallahassee , florida , using a bruker 113v fourier transform infrared spectrometer .
spectra were taken using the 3.5 , 12 , 23 , and 50 @xmath19 m mylar beam splitters , covering a frequency range of 25650 @xmath1 .
both absolute and relative transmittance spectra were measured as a function of temperature using a bolometer detector and a continuous flow helium cryostat .
small differences in the transmittance spectra were assessed at low temperature using absorption differences , calculated from the relative transmittance as @xmath20 $ ] .
the magnetic field dependence of the transmittance at 5 k was measured at the nhmfl using a 20 t superconducting magnet and a transmission probe equipped with a bolometer detector@xcite .
again , transmittance ratio spectra were used to investigate small deviations from unity due to the applied field . by taking the natural log of the transmittance ratio @xmath21)$ ] ,
we obtain the absorption difference at each field : @xmath22 . upon examination of the absorption difference curves over the entire aforementioned frequency range
, we identified three features that change with applied magnetic field . in order to distinguish signal from noise in a quantitative way , we calculated the standard deviation from the mean for each feature .
in addition , the same analysis was performed on representative nearby frequency ranges containing no field - dependent features .
we use these standard deviations from the mean to quantify the effects of the field and to characterize the intrinsic noise level in regimes away from the magneto - optic signatures @xcite .
field sweeps on an empty hole were also carried out for reference ; as expected , no field dependence was observed .
figure 1 displays the far infrared transmittance of mn@xmath0 as a function of temperature . as expected for a molecular solid , intramolecular vibrational modes of the mn@xmath0 molecule
appear above 150 @xmath1 @xcite . in agreement with previous authors
@xcite , we also observe a weak low - energy structure centered near 35 @xmath1 ( fig .
1 , panel a ) .
this magnetic - dipole allowed excitation has been cited as evidence for other excited state multiplets with @xmath11@xmath2310 existing in mn@xmath0 @xcite . within our sensitivity
, we have found that the 35 @xmath1 feature displays limited temperature dependence and no magnetic field dependence .
the far infrared spectra of mn@xmath0 display a large number of intramolecular modes due to the relatively low symmetry of the molecule .
they are grouped together in clusters and superimposed upon one another .
these structures sharpen and harden with decreasing temperature ( fig .
recent 20 k neutron diffraction studies by langan @xmath24 @xcite assessed the low - temperature molecular distortion ; compared to the 300 k structure , major characteristics include a gradual displacement of mn(3 ) and carboxylate groups , and more extensive solvent interactions including a 4-center hydrogen bond .
( note that mn(3 ) is part of the crown , according to the numbering scheme of ref .
@xcite , with connections / interactions to the bridges , ligands , and solvent . )
these results motivated us to more carefully investigate the far infrared response in the low - temperature regime .
figure [ temp]a displays the absorption difference spectra of mn@xmath0 at low temperature @xcite . the gradual distortion of the cluster framework was previously assessed by neutron diffraction studies @xcite . from a magnetic properties point of view , the behavior of the mn ions is most important ; experiment indicates that mn(3 ) on the crown displays the most significant low - temperature displacement @xcite .
a number of vibrational features are modified in this temperature range and therefore likely contain a substantial mn(3 ) , carboxylate , and acetate ligand contribution . as will be discussed in the next section ,
three modes ( 284 , 306 , and 409 @xmath1 ) are of particular interest because of their dependence on magnetic field .
the 284 @xmath1 mode shows an intensity variation at low temperature and a small frequency shift at higher temperatures ( fig .
[ temp]b ) . the standard deviation from the mean in the vicinity of the 284 @xmath1 feature ( fig .
[ temp]c ) quantifies these changes , with an inflection point around 20 k. the maximum in the derivative of the fourth - order polynomial fit to this data better illustrates the position of this inflection point .
the 306 @xmath1 structure ( not shown ) is sensitive to this temperature regime as well .
although the feature centered at 409 @xmath1 also exhibits changes in this temperature range , no peculiarities were observed around 20 k. preliminary molecular dynamics simulations @xcite suggest that a majority of atoms in the mn@xmath0 molecule are , in some way , involved in each vibration in this frequency range .
the modes are many , both because of the low molecular symmetry and the large number of atoms . using mode visualization and the spectra of several model compounds as a guide ,
particularly that of mn(ii)-ac@xmath4 ( fig .
[ model ] ) , we propose the following general assignments .
the low - energy motions below 300 @xmath1 include a great deal of acetate ( ligand ) motion .
complex low energy motions of the core and crown begin around 180 @xmath1 .
these motions include bending , rocking , shearing , twisting , and wagging .
asymmetric and symmetric stretching of the core and crown seem to begin slightly below 500 @xmath1 , thus providing a likely assignment for the complicated vibrational cluster observed in the spectrum centered near 540 @xmath1 .
the simulations show that modes in this energy range contain sizable ( core and crown ) oxygen contributions .
comparison of the spectral data from the model compound mn(ii)-ac@xmath4 and mn@xmath0 confirm that the mode clusters centered near 360 , 400 , and 540 @xmath1 in mn@xmath0 are mainly motions of the magnetic center rather than the ligands .
related model compounds such as mno , mno@xmath4 , and mn@xmath4o@xmath2 also have characteristic vibrational features in the far infrared ( below 600 @xmath1 ) . although the structures are different and mn valences unmixed compared to the title compound , the reference spectra of these model mn - based solids @xcite indicate that the far infrared response is rich and highly relevant to the mn o motion .
measurements on the mn(ii)-ac@xmath4 model compound ( fig .
3 ) suggest that the structures centered at 590 and 635 @xmath1 in the spectrum of mn@xmath0 are acetate - related .
the similar temperature dependence of these modes in the two samples supports this assignment . in order to understand and model physical properties that contain important phonon contributions ( for instance @xmath25@xmath26 or heat capacity ) in the thermally activated regime , it is helpful to know realistic values of the most important low - energy infrared active phonons of mn@xmath0 .
our data suggests that use of a relatively small series of characteristic frequencies may be adequate .
for instance , the three most intense structures are centered at 375 , 410 , and 540 @xmath1 .
more complete models might take additional modes , the detailed mode clustering , and relative intensities into account .
figure 4 displays the magnetic field dependence of three features in the far infrared spectrum of mn@xmath0 .
the center positions of these structures ( @xmath15284 , 306 , and 409 @xmath1 ) are indicated with arrows on the absolute transmittance spectra in fig .
it is interesting that the 284 and 306 @xmath1 features do not correspond to major modes in the absolute transmittance spectrum . as shown in the left - hand panels ( figs .
4a , 4c , and 4e ) , the field dependence as measured by the absorption difference spectra is well - defined for the mode near 284 @xmath1 , whereas it is more modest for the structures centered at 306 and 409 @xmath1 .
the filled symbols and solid lines in the right - hand panels ( figs . 4b , 4d , and 4f )
provide a quantitative view of the field dependent trends .
the standard deviation from the mean for the three modes of interest shows a clear upward trend , in contrast to that of nearby spectral regions which shows no field dependence and provides an estimate of the intrinsic noise level @xcite .
the feature at 284 @xmath1 is the most prominent of the three field dependent structures
. the absorption decreases with applied field and shows a tendency toward saturation at 17 t , as indicated in fig .
[ ratio]b .
the standard deviation from the mean in nearby spectral ranges ( for instance , similarly sized regimes centered at @xmath15 277 and 299 @xmath1 ) shows no field dependence .
the response of the 284 @xmath1 feature is therefore well above the noise and characteristic of mn@xmath0 .
similar trends and a tendency towards saturation of the standard deviation from the mean are observed for the more complicated absorption difference structures centered near 306 and 409 @xmath1 , although the overall coupling of these modes to the applied field is weaker and the line shapes are different from that of the 284 @xmath1 feature .
the changes observed near 306 @xmath1 are mainly caused by a 1 @xmath1 softening , whereas the 409 @xmath1 structure is due to both softening and broadening .
the field dependence of these three features in the absorption difference spectra is confirmed by repeated measurements using different experimental parameters @xcite .
further , as seen in fig . 4 , the trend with applied field in both the difference spectra and the standard deviation from the mean near 306 and 409 @xmath1 can be easily distinguished from that of the non - field - dependent intervals .
it is of interest to understand the microscopic character of the 284 , 306 , and 409 @xmath1 modes and why they are affected by the magnetic field .
although there are a number of mechanisms that might explain this behavior , such as the effect of magnetic field on hybridization of ligand - metal bonds @xcite , we believe the most promising is that of spin - phonon coupling . here
, variations in the applied field modulate the magnetic system , thereby affecting the phonons coupled to it .
such an interaction has been thought to be important for describing the slow magnetization relaxation behavior of mn@xmath0 , making it an important term in the total hamiltonian @xcite . that only three far - infrared features are observed to change with applied field up to 17 t suggests that these modes are related to motions that maximally change the low temperature spin interactions .
one possible candidate for such motion might contain bending between the core and crown structures affecting the mn - o - mn angle , therefore modulating the superexchange between mn ions .
an alternate explanation involves the especially curious sensitivity of the 284 @xmath1 mode to temperature variation ( in the range 5 - 35 k ) as well as to the applied magnetic field .
this coincidence suggests that the magneto - elastic response at 284 @xmath1 may be intimately related to the mn(3 ) motion on the crown .
interestingly , the 306 @xmath1 structure also displays some sensitivity to the displacement of the molecular framework .
further work is clearly needed to untangle these complex interactions .
the observed magnetic field dependencies at 284 , 306 , and 409 @xmath1 suggest that @xmath27 can be augmented by including contributions from these three vibrational modes .
spin - lattice interactions in the far - infrared energy range have been observed in a number of other transition metal oxides in the past @xcite .
however , the energy scale of the 284 , 306 , and 409 @xmath1 features is larger than might be expected for relevant spin - phonon processes in mn@xmath0 , even in the thermally activated regime .
indeed , previous studies on the title compound have focused on acoustic phonons as the major contributors to @xmath27 @xcite .
thus , if these far - infrared phonons are connected with the mn@xmath0 relaxation processes in any way , virtual states would likely be involved .
the energy scale of the 284 , 306 , and 409 @xmath1 phonons may be more relevant to high field transport or specific heat measurements .
we have measured the far - infrared response of mn@xmath0 as a function of temperature and magnetic field . the features in this region involve vibrations from all of the groups of atoms ( core , crown , and ligands ) in the molecule .
three structures related to intramolecular vibrations near 284 , 306 , and 409 @xmath1 are observed to change in applied magnetic field , suggesting that they are coupled to the spin system .
of these three , the feature centered at 284 @xmath1 displays the strongest coupling .
based on the similarity between the temperature and field dependent absorption difference data , we speculate that this mode involves mn(3 ) motion on the crown .
the data reported here may be helpful in understanding the role of vibrations in the theoretical models of magnetization relaxation in mn@xmath0 and related systems .
@xmath28 current address : department of chemistry , university of tennessee , knoxville , tn 37996 .
funding from the national science foundation ( dmr-9623221 ) to support work at suny - binghamton is gratefully acknowledged .
the work at florida state university was also partially funded by the nsf .
a portion of the measurements were performed at the nhmfl , which is supported by nsf cooperative agreement no .
dmr-9527035 and by the state of florida .
we thank z.t .
zhu for technical assistance .
b. barbara , w. wernsdorfer , l.c .
sampaio , j.g . park , c. paulsen , m.a .
novak , r. ferr , d. mailly , r. sessoli , a. caneschi , k. hasselbach , a. benoit , and l. thomas , _ j. magn
mater . _ * 140 - 144 * , 1825 ( 1994 ) .
standard deviation from the mean was used to quantify the features in the absorbance difference spectra rather than integrated intensity due to the complicated line shapes of some of the features .
this method was also used to estimate the intrinsic noise level by analysis of nearby spectral regions .
thus , for consistency , standard deviation from the mean was used for all treatments of the difference spectra . | we report the far - infrared spectra of the molecular nanomagnet mn@xmath0-acetate ( mn@xmath0 ) as a function of temperature ( 5300 k ) and magnetic field ( 017 t ) .
the large number of observed vibrational modes is related to the low symmetry of the molecule , and they are grouped together in clusters .
analysis of the mode character based on molecular dynamics simulations and model compound studies shows that all vibrations are complex ; motion from a majority of atoms in the molecule contribute to most modes .
three features involving intramolecular vibrations of the mn@xmath0 molecule centered at 284 , 306 and 409 @xmath1 show changes with applied magnetic field .
the structure near 284 @xmath1 displays the largest deviation with field and is mainly intensity related .
a comparison between the temperature dependent absorption difference spectra , the gradual low - temperature cluster framework distortion as assessed by neutron diffraction data , and field dependent absorption difference spectra suggests that this mode may involve mn motion in the crown . |
there is a continuous and rapid global growth in data storage needs .
archival and backup storage form a specific niche of importance to both businesses and individuals .
a recent market analysis from idc stated that the global revenue of the data archival business is expected to reach $ 6.5 billion in 2015 .
the necessity to cost - effectively scale - up data backup systems to meet this ever growing storage demand poses a challenge to storage systems designers .
when a large volume of data is involved , deploying a networked distributed storage system becomes essential , since a single storage node can not scale .
furthermore , distribution provides opportunities for fault tolerance and parallelized i / o . examples of such distributed storage systems are readily found in datacenter environments , including distributed file systems such as gfs @xcite or hdfs @xcite , distributed key - value stores like dynamo @xcite or cassandra @xcite , for storing huge volume of scientific or multimedia content @xcite as well as in ad - hoc end user resource based peer - to - peer ( p2p ) settings such as oceanstore @xcite and friend - to - friend ( f2f ) storage systems @xcite , a special kind of peer - to - peer systems often considered particularly suitable for personal data backup .
an important design aspect in distributed storage systems is redundancy management .
data replication provides a simple way to achieve high fault - tolerance , while erasure codes such as reed - solomon codes @xcite are more sophisticated alternatives , capable of significantly reducing the data storage footprint for different levels of fault - tolerance @xcite .
various trade - offs in adopting erasure codes in storage systems , such as storage overhead & fault - tolerance , access frequency & decoding overheads , but also due to the need of replenishment of lost redundancy , repair bandwidth & repair time after failures , have been studied in the literature , revealing in particular that erasure codes are particularly suited for backup and archival storage , where data access is infrequent , and hence the effects of decoding are marginal .
consequently , many modern data center oriented storage and file systems such as microsoft s azure @xcite , hdfs @xcite and the next generation of google s file system ( colossus ) have incorporated erasure codes to enhance the systems storage efficiency . however , a relatively unexplored aspect of the usage of erasure codes in storage systems both in data center as well as peer - to - peer environments is that of the time required for inserting the data ( along with the necessary redundancy for fault - tolerance and durability ) . when using replication , a source node aiming to store new data can upload one replica of this data to the first storage node , which can concurrently forward the same data to a second storage node , and so on . by applying such a pipelining mechanism ,
the load for redundancy insertion can be shared , and the source node does not need to upload any redundant information itself .
replication thus naturally supports `` in - network '' generation of redundancy , that is , generation of new redundancy within the network , through data exchange among storage nodes , which in turn leads to fast insertion of data .
in contrast , in erasure encoded systems , the source node is the only one responsible for computing and uploading all the encoded redundant data to the corresponding storage nodes .
this has traditionally been the case , because typically the coding process is centralized in nature , and hence all encoded data fragments are generated at a single node , which then also bears the load of inserting the encoded fragments at other storage nodes .
the amount of data the source node uploads is then considerably larger , including the data object and its corresponding redundancy , resulting in lower data insertion throughput .
insertion throughput may further be exacerbated when the source node and the set of storage nodes have additional ( mismatched ) temporal constraints on resources availability , in which case in - network redundancy generation can provide partial mitigation .
we elaborate the effect of temporal resource ( un)availability issues with two distinct example scenarios , which we also use later in our experiments to determine how ( much ) in - network redundancy generation may improve the data insertion throughput : a. in datacenters , storage nodes might be used for computation processes which require efficient access to local disks .
since backup processes consume large amounts of local disk i / o , system administrators might want to avoid backup transfers while nodes are executing i / o intensive tasks e.g. , mapreduce tasks .
b. in peer - to - peer settings , users exchange some of their spare disk resources in order to realize a collaborative data backup service .
such resource sharing may furthermore be driven by other constraints such as trust or friendship .
however , desirable users may be online at different times of the day , complicating the data insertion process . in both cases
the insertion of new redundancy by the source node is restricted to the periods when the availability windows of the source overlap that of the storage nodes . unlike replication , where in - network redundancy generation is achieved trivially ,
traditional erasure codes are not easily amenable .
our solution is based on new classes of erasure codes designed for better repair efficiency , obtained through local repairability , that is the property of being able to repair a failure by contacting only a small number of live nodes ( e.g. @xcite ) . in particular , to provide a concrete instance that will be used later for experimentation , we will focus on a novel family of erasure codes called self - repairing codes ( src ) @xcite , whose salient property is that the encoded data stored at each node can be easily regenerated by using information from typically only two live storage nodes .
as we will show , local repairability is the key property to achieve in - network redundancy generation .
however , src have strict constraints on how storage nodes can combine their data to generate content for other nodes , which , along with the temporal availability constraints of nodes , complicate the design of efficient in - network redundancy generation .
in this paper we provide an analytical framework to define valid transfer schedules for the in - network redundancy generation using src .
we then identify the main requirements that valid scheduling algorithms must satisfy , and show that determining optimal schedules is computationally intractable .
accordingly , we explore several heuristic algorithm implementations which aim at maximizing the utilization of the spare resources of the storage nodes , thus improving the backup throughput .
we determine the efficacy of our approach in different environments by using google data center availability and workload traces , and availability traces from real friend - to - friend ( f2f ) @xcite and peer - to - peer ( p2p ) applications @xcite .
our results show that for data center traces the algorithms proposed for the in - network redundancy generation can increase the throughput of the storage process up to 90% as compared to classical naive storage approaches , while only requiring 50% more of the network resources , but using it only when it is otherwise unused . for
p2p / f2f traces the throughout can increase up to 60% while requiring 38% additional ( but again , spare ) network resources .
the main contributions of this paper can be summarized as follows : a. we introduce the concept of _ in - network redundancy generation _ for reducing data insertion latency in erasure code based storage systems , and demonstrate its feasibility using one specific instance of locally repairable code , namely self - repairing codes .
b. we define an _ analytical framework _ to explore valid data transfer schedules where the in - network redundancy generation process maximizes the use of the available network resources .
c. we show that besides requiring a node availability prediction , determining the optimal data insertion schedule is computationally intractable .
d. we propose a set of _ heuristics _ for efficient in - network redundancy generation .
e. we determine the efficacy of in - network redundancy generation in diverse distributed storage environments using real workload and availability trace driven simulations .
the rest of the paper is organized as follows . in section [
s : background ] we provide the background on erasure codes and self - repairing codes , a particular instance of locally repairable codes . in section [ s : redgen ]
we define our in - network redundancy generation process and in section [ s : scheduling ] we show how to schedule the redundancy generation to speedup the insertion of new data .
due to the complexity of determining optimal schedules , we propose in section [ s : algorithm ] several heuristic scheduling algorithms , which are evaluated in section [ s : eval ] using real availability traces . finally in section [ s : conclusions ] we state our conclusions and further research directions .
the p2p research community has long studied the applicability of erasure codes in low availability environments with limited storage capacity @xcite .
the growing interest in applying erasure codes in data centers is more recent , and aims at reducing the storage costs @xcite .
it further triggered a line of research around repairability of storage systems , i.e. , how lost redundancy can be replenished , which includes the application of network coding @xcite to carry out the repair process in a decentralized manner , designing novel codes with inherent local repairability @xcite , as well as engineering solutions such as applying multiple levels of encoding on the same @xcite or even across different data objects @xcite . while fault - tolerance , storage overhead , repairability , but also i / o and bandwidth are well recognized critical bottlenecks for the storage of huge amounts of data , existing literature does not explore yet how data insertion can be optimized in the context of erasure codes based storage .
this work leverages on one of these recent approaches on distributed storage systems repairability , namely , novel codes with local repairability , in order to improve the very process of data insertion . in the following ,
we provide some background on erasure codes as classically used for distributed storage , as well as on one specific instance of locally repairable codes , self - repairing codes ( src ) , which we use in the rest of the paper to demonstrate the feasibility and quantify the benefits of in - network redundancy generation in erasure code based distributed storage systems .
a classical @xmath0 erasure code allows to redundantly encode an object of size @xmath1 into @xmath2 redundant fragments of size @xmath3 , each to be stored in a different storage node . the data storage overhead ( or redundancy factor )
is then given by @xmath4 , and the stored object can be reconstructed by downloading an amount of data equal to @xmath1 , from @xmath5 or more different nodes out of @xmath2 .
one of the main drawbacks of using classical erasure codes for storage is that redundant fragments can only be generated by applying coding operations on the original data .
the generation of new redundancy is then restricted to nodes that possess the original object ( or a copy ) , namely : the source node , storage nodes that previously reconstructed the original object , or possibly were storing a copy ( as is the case in a hybrid model where a full copy of the object is kept , together with encoded fragments ) . when the original raw object is not available , repairing a single node failure consequently entails downloading an amount of information equivalent to the size of the original object , causing a significant communication overhead . in order to mitigate this communication overhead ,
a new family of erasure codes called regenerating codes ( see e.g. @xcite and reference therein ) was recently designed by adopting ideas from network coding @xcite , a popular mechanism deployed to improve the throughput utilization of a given network topology .
the main advantage of regenerating codes is that new redundant fragments can be generated by downloading an amount of data @xmath6 from @xmath7 other redundant fragments , where @xmath8 , and @xmath9 . unlike in classical erasure codes
, regenerating codes can thus repair missing fragments by downloading only an amount of data equal to @xmath10 , where usually @xmath11 .
however , the maximum communication savings occur for large values of @xmath7 , in which cases however , the chance to find @xmath7 available nodes might be very low , limiting the practicality of such codes . self - repairing codes ( src ) @xcite are a new family of erasure codes designed to minimize the maintenance overhead by reducing the number of nodes @xmath7 required to be contacted to recreate lost fragments . a specific family of src , named homomorphic self - repairing codes ( hsrc ) , has the property that two encoded fragments can be xored for such a regeneration , i.e. @xmath12 , as long as not more than half of the nodes have failed .
this locally repairable property makes hsrc suitable for the in - network redundancy generation since partial redundant data stored in two different nodes can be used to generate data for a third node , without requiring the intervention of the source node .
however , as we will show below , the pairs of nodes used for that purpose can not be arbitrary chosen .
let us recall briefly the construction of hsrc .
we denote finite fields by @xmath13 .
the cardinality of @xmath13 is given by its index , that is , @xmath14 is the binary field with two elements ( the two bits 0 and 1 ) , and @xmath15 is the finite field with @xmath16 elements . if @xmath17 , for some positive integer @xmath18 , we can fix a @xmath19 of @xmath15 and represent an element @xmath20 using an @xmath18-dimensional vector @xmath21 where @xmath22 , @xmath23 .
let @xmath24 be the object to be stored over a set of @xmath2 nodes , which is represented as a data vector of size @xmath25 bits , @xmath26 , i.e. : @xmath27 given these @xmath5 original elements , the @xmath2 redundant fragments are obtained by evaluating the polynomial @xmath28 \label{e : poly}\ ] ] in @xmath2 non - zero values @xmath29 of @xmath30 , yielding the redundant vector @xmath31 of size @xmath32 bits , i.e. : @xmath33 in particular we need the code parameters @xmath0 to satisfy @xmath34
the main important property of hsrc is its homomorphic property . from @xcite
we have that : let @xmath35 and let @xmath36 be the polynomial defined in ( [ e : poly ] ) , then @xmath37 .
[ l : homo ] this implies that we can generate a redundant element @xmath38 from @xmath39 and @xmath40 if and only if @xmath41 .
this homomorphic property is one way of obtaining local repair by contacting only @xmath12 nodes .
[ ex : k3n7_1 ] consider a @xmath42 hsrc and an object @xmath43 of size @xmath44 bits , where @xmath45 , @xmath46 .
we write @xmath47 , @xmath48 , @xmath49 , from which we compute @xmath50 .
we evaluate @xmath36 in @xmath51 values of @xmath52 , represented in vector form as @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 , yielding : @xmath60 with corresponding redundant vector @xmath61 we can check that @xmath62 which illustrates the local repairability of the code . note that we have not used the vector @xmath63 here , which would have resulted in a longer code @xmath64 .
we now discuss how hsrc operate in two different scenarios : ( i ) when the source introduces data in the system , and ( ii ) during the in - network redundancy generation .
the homomorphic property described in lemma [ l : homo ] has been introduced to repair node failures , though it can similarly serve to generate redundancy from the source .
recall from lemma [ l : homo ] that @xmath65 , where both @xmath66 can be seen as @xmath18-dimensional binary vectors , by fixing a @xmath67-basis of @xmath68 .
let us denote this basis by @xmath69 .
thus @xmath70 can be written as @xmath71 , @xmath72 , and by virtue of the homomorphic property , we get that @xmath73 this means that the source only needs to compute @xmath74 for a given basis @xmath69 , after which all the other encoded fragments are obtained by xoring pairs of elements in @xmath75 .
thus , when using an @xmath76 hsrc , the source computes @xmath5 ( @xmath26 ) encoded fragments @xmath77 , where @xmath78 are linearly independent , for example , @xmath79 , and then performs the corresponding xoring .
the source then injects the @xmath2 encoded fragments in the network .
[ ex : basis ] in example [ ex : k3n7_1 ] , we have @xmath80 , and a natural @xmath67-basis for @xmath52 is @xmath81 , @xmath82 , @xmath83 , @xmath84 .
the source can generate redundancy by first computing @xmath85 , @xmath86 , @xmath87 , then @xmath88 , @xmath89 , @xmath90 and @xmath91 .
the @xmath51 encoded fragments are then ready to be sent over the network .
further notice that the set @xmath92 can be seen as a basis for the set of redundant fragments , since they are linearly independent , and can be combined to generate every redundant fragment .
let us now consider the case where the source might not inject the whole set of @xmath2 encoded fragments , but only a subset @xmath93 of the encoded fragments .
we use the triplet notation @xmath94 to represent the possibility to generate the element @xmath95 by xoring @xmath96 and @xmath97 , @xmath98 .
note that due to the commutativity property of the additive operator , triplets @xmath94 and @xmath99 can be indistinguishably used to denote the same redundancy generation process .
we denote by @xmath100 the set with all the feasible repair triplets from a set of @xmath2 redundant elements .
finally , let us define the following two sets : let @xmath101 be the set of all the possible @xmath94 triplets where fragment @xmath96 is used to generate some other fragment : @xmath102 [ d : outset ] let @xmath103 be the set of all the possible @xmath94 triplets that can be used to create @xmath95 : @xmath104 [ d : inset ] finally , given a number of redundant elements @xmath105 , for any positive integer @xmath106 , we have from @xcite that : @xmath107 , we have that @xmath108 previously we detailed how to encode a data vector @xmath24 of size @xmath25 bits into a redundant vector @xmath31 of size @xmath32 bits .
we showed that hsrc allow this encoding by using data from the source node as well as by using data from other storage nodes . in this subsection
we describe one method to practically implement hsrc to encode larger data objects of size @xmath1 , where @xmath109 . the first step to encode an object of size
@xmath1 is to split it into @xmath110 vectors is multiple of @xmath25 , otherwise the object can be encoded by parts and/or zero - padded to meet this requirement . ] of size @xmath25 bits .
let us represent the object to be encoded as @xmath111 .
after the splitting process , @xmath112 , where @xmath113 , @xmath114 , @xmath115 .
each of these vectors @xmath116 is individually encoded using the polynomial ( [ e : poly ] ) to obtain an encoded vector @xmath117 , with @xmath118 , @xmath119 .
finally , the vector @xmath120 with the @xmath2 fragments to be stored in the system is obtained by concatenating individual elements of @xmath121 so that @xmath122 , namely @xmath123 contains those coefficients @xmath124 of @xmath125 that have @xmath126 as second index .
[ x : hsrc ] consider the @xmath42 hsrc and the object @xmath127 , where @xmath128 .
we split the object into @xmath129 vectors @xmath130 , where @xmath131 , @xmath132 and @xmath133 . after encoding each of the individual vectors we obtain the set of redundant vectors @xmath134 ( @xmath116 is encoded to obtain @xmath135 ) , where @xmath136 .
finally , we can obtain @xmath137 , and similarly for all the fragments .
note that this encoding technique allows stream encoding .
as soon as the source node receives the first @xmath25 bits to store , it can generate the vector @xmath138 , encode it to @xmath139 , and distribute @xmath140 to the @xmath2 storage nodes .
similarly , when a storage node receives @xmath141 it can forward it to other nodes for in - network redundancy generation , for instance , when the source does not have adequate bandwidth to upload all the @xmath2 redundant fragments . to implement computationally efficient codes one can set @xmath142 , or @xmath143 , for
which addition can simply be done by xoring system words , and for which efficient arithmetic libraries are available @xcite . in the rest of this paper
we will assume that hsrc are implemented using the method described here .
we will use the term * _ redundant fragment _ * to refer to each of the redundant elements @xmath144 , i.e. , each node stores one redundant fragment . and
similarly we will use the term * _ redundant chunk _ * to refer to each of the sub - elements @xmath145 stored in each node @xmath126 , i.e. , each node @xmath126 can store up to @xmath146 redundant chunks .
in - network redundancy generation has the potential to speedup the insertion of new data in distributed storage systems .
however , the magnitude of actual benefit depends on two factors : ( i ) the availability pattern of the source and storage nodes , which determines the achievable throughput , and ( ii ) the specific schedule of data transfer among nodes subject to the constraints of resource availability , which determines the actual achieved throughput for data backup . in this section
, we explore the scheduling problem , demonstrating that finding an optimal schedule is computationally very expensive even with a few simplifying assumptions , and accordingly motivate some heuristics instead .
let @xmath147 be a source node aiming to store a new data object to @xmath2 different storage nodes , and let @xmath126 , @xmath148 , represent each of these @xmath2 storage nodes .
we model our system using discrete time steps of duration @xmath149 , where at each time step nodes can be available or unavailable to send / receive redundant data .
the binary variable @xmath150 denotes this availability for each node @xmath126 for the corresponding time step @xmath151 . using this binary variable
we can define the maximum amount of data that node @xmath126 can upload during time step @xmath151 _ upload capacity _ ) by @xmath152 where @xmath153 is the upload bandwidth of node @xmath126 during time step @xmath151 .
similarly , the amount of data each node can download during time step @xmath151 ( _ download capacity _ ) is given by @xmath154 where @xmath155 represents the download bandwidth of node @xmath126 during time step @xmath151 .
then , we define the _ in - network redundancy generation network _ as a weighted temporal directed graph @xmath156 , @xmath157 , with the set of nodes @xmath158 , and the set of edges @xmath159 . the amount of data that nodes might send among themselves is a mapping @xmath160 , denoted by @xmath161 , @xmath162 , @xmath163 .
hsrc characteristics constrain the mapping @xmath164 since nodes can only send or receive data trough valid redundancy creation triplets : @xmath165 furthermore , we assume ( for algorithmic simplicity ) that nodes send data through each of the redundancy generation triplets symmetrically : @xmath166 for ease of notation we will refer to the data sent through each of the redundancy generation triplets simply by @xmath167 . similarly , because of the upload / download bandwidth constraints , the mapping @xmath164 must also satisfy the following constraints : * the amount of data the source uploads is constrained by its upload capacity : @xmath168 * the amount of data storage nodes upload is also constrained by their upload capacity : @xmath169 * the amount of data storage nodes download is restricted by their download capacity : @xmath170 a _ bandwidth - valid in - network redundancy generation scheduling _ is any mapping @xmath164 on @xmath171 that satisfies the constraints defined in equations ( [ e : c : triplets ] ) , ( [ e : c : symmetry ] ) , ( [ e : c : uploadsrc ] ) , ( [ e : c : uploadbw ] ) and ( [ e : c : downloadbw ] ) .
let @xmath172 be the amount of data that node @xmath126 had received at the end of time step @xmath151 . for sufficiently large enough files and small values of @xmath18 ( e.g. @xmath142 ) , we can assume without loss of generality that @xmath173 corresponds to the index of the last _ redundant chunk _ received by node @xmath126 .
let @xmath174 denote the size of the largest possible file that a schedule @xmath164 can store in @xmath175 time steps , which represents a data insertion throughput of @xmath176 . then , by definition of erasure codes , to consider that a file of size @xmath174 has been successfully stored after @xmath175 time steps , each node must receive an amount of data equal to @xmath177 .
using this fact we can define @xmath174 as : @xmath178 for a given network @xmath171 and a duration @xmath175 , an in - network redundancy generation scheduling @xmath164 is then _ optimal _ if it maximizes @xmath174 .
note that to maximize @xmath174 an optimal schedule will tend to even out the amount of data @xmath179 sent to each node , obtaining a maximum insertion throughput and a minimum consumption of network resources when @xmath180 . we can measure the overall network traffic required by any schedule @xmath164 , after @xmath175 time steps , namely @xmath181 , by : @xmath182 accordingly , we define an in - network redundancy generation schedule @xmath164 to be an _ optimal minimum - traffic _ schedule if besides maximizing @xmath174 , it also minimizes @xmath181 . from ( [ e : innettraf ] ) we also want to note that in the total traffic required by the schedule @xmath164 , @xmath181 , the amount of redundant data each node @xmath126 receives comes from two distinct components ( the two summands within parentheses ) : @xmath183 and @xmath167 , where @xmath184 .
the first component represents the redundant data inserted by the source while the second component represents the redundant data created through the in - network redundancy generation process . since in the last case
there are two nodes ( i.e. , @xmath185 and @xmath186 ) uploading data to node @xmath126 the value has two be multiplied by two , leading to the following remark : the new redundant data created through the in - network redundancy generation process requires twice the traffic required by the source redundancy generation .
[ r : sourcetraffic ] however , although the use of the in - network redundancy generation increases the required network traffic , it has the potential to increase the opportunities of redundancy generation , and hence , increase the data insertion throughput . in section [ s :
eval ] we will show that the relative increase of insertion throughput surpass the relative increase of network traffic , demonstrating the practicality of the in - network redundancy generation process and its scalability . in this section we elaborate that while being a bandwidth - valid schedule is a necessary condition , it is not a sufficient condition for the schedule to be actually valid .
for that , we will use example [ x : hsrc ] , an in - network redundancy generation network using hsrc with parameters @xmath42 , where the redundant fragments @xmath187 have to be stored in nodes @xmath188 respectively .
recall also that each redundant fragment @xmath189 is composed of 3 redundant chunks , hence @xmath190 .
for ease of notation we will assume that each redundancy generation triplet @xmath191 , @xmath192 , satisfies the property @xmath193 where @xmath194 denotes the _ bitwise xor operation_. based on example [ x : hsrc ] , we consider three different scheduling policies , all depicted in figure [ f : onions ] .
we assume that due to the limited upload capacity of the source node it can only upload three redundant fragments simultaneously . in the first scenario , at time @xmath195 the source node sends to nodes 1 , 2 and 3 their first redundant chunk ; at time time
@xmath196 it does the same for nodes 5 , 6 and 7 .
note that if at time step @xmath196 the mapping @xmath164 tries to make use of the in - network redundancy generation triplets @xmath197 , @xmath198 and @xmath199 ; nodes 5 , 6 and 7 end up receiving the same redundant fragment twice . in this case
the in - network redundancy traffic does not contribute in speeding up the backup process and only consumes communication resources . although avoiding this problem is implicit in the definition of a _ minimum - traffic _ scheduling , it needs to be explicitly considered during the scheduling .
consider a second scheduling policy trying to solve the previous problem by sending to nodes 5 , 6 and 7 the second chunk instead of the first .
it allows these nodes to receive two different fragments by time @xmath196 .
however , it appears a circular dependency problem with triplets @xmath197 , @xmath198 and @xmath199 . to show this dependency ,
imagine that we want to generate fragment @xmath200 using non - source data .
note that @xmath200 requires of @xmath201 , and @xmath201 requires of @xmath202 , which at the same time requires of the fragment we aim to generate , @xmath200 .
although it is a _ bandwidth - valid _ schedule , the circular dependency problem makes it an unfeasible schedule .
finally , in the third case we see how the circular dependency problems can be avoided if the source sends uncorrelated fragments at each time step .
it is easy to see from this example that a valid schedule needs to be not only bandwidth - valid , but also ensure that : ( i ) nodes do not receive duplicated data , and ( ii ) circular triplet dependencies are prevented . we show that finding an optimal schedule satisfying all the previous requirements is computationally very expensive , even under further simplifying assumptions : the amount of data that the source node @xmath147 sends during each time step @xmath151 to any storage node @xmath126 , @xmath183 , is a constant value and is not part of the optimization problem . [ a : simpleprob1 ] storage nodes can only receive redundant chunks sequentially .
it means that node @xmath126 will never receive chunk @xmath203 before previously receiving chunk @xmath204 .
[ a : simpleprob2 ] it is easy to see that the simplified problem subject to these two assumptions corresponds to a specific instance of the generic case described above .
the interesting property about this simplified version of the problem is that we can reduce the decision of choosing the optimal schedule @xmath164 to an algorithm `` sortedvector '' which sorts @xmath100 , as it is shown in algorithm [ a : algo ] .
it is also easy to see , how due to the iterative use of redundancy generation triplets , algorithm [ a : algo ] avoids both the `` duplicate data '' and the `` circular dependencies '' problems .
however , since @xmath205 , it means that there are @xmath206 possible ways of sorting @xmath100 , and thus , @xmath207 different scheduling possibilities .
thus , a brute force algorithm to determine the best schedule would have a @xmath208 cost .
@xmath209 @xmath210 @xmath211 @xmath212 $ ] / * @xmath191 * / availdatabybw @xmath213 min(@xmath214 ) availdatabyindex @xmath213 min(@xmath215 availdatabyindex @xmath213 max(availdatabyindex , 0 ) availdata = min(availdatabybw , availdatabyindex ) @xmath216 availdata @xmath217 availdata .
we assume in this example a hypothetical system where @xmath218 . ]
if we focus on a single time step @xmath151 , then the scheduling problem can be restated as how to choose the best permutation of @xmath100 .
we can represent this decision problem using a permutation tree as is depicted in figure [ f : permtree ] .
the weight of the edges in this permutation tree correspond to the negative amount that choosing each edge contributes to @xmath219 .
choosing the best scheduling algorithm tis the same than finding the shortest path between vertices @xmath220 and @xmath7 in the permutation tree .
the bellman - ford algorithm can find the shortest past with cost @xmath221 where @xmath222 and @xmath223 respectively represent the number of edges and vertices in the permutation tree .
however , in our permutation tree the number of edges and vertices are both @xmath206 , which makes finding the optimal schedule for even the simplified problem computationally exorbitantly expensive , even for small number of nodes @xmath2 .
hence , we consider the general problem described in [ s : optimal ] to be also intractable .
in this section we investigate several heuristics for scheduling the in - network redundancy generation .
we split the scheduling problem into two parts , following the strategy presented in algorithm [ a : algo ] .
the heuristics do not require assumption [ a : simpleprob1 ] , thus allowing the source node to send different amounts of data to each storage node .
we however still rely on assumption [ a : simpleprob2 ] , which allows us to model the decision problem with a sorting algorithm , as previously outlined in algorithm [ a : algo ] .
thus , the overall scheduling problem is decomposed into the following two decisions : ( i ) how does the source node schedule its uploads ? ( ii )
how are redundancy generation triplets sorted ? recall that generating redundancy directly from the source node involves less bandwidth than doing it with in - network techniques ( remark [ r : sourcetraffic ] ) .
thus , a good source traffic scheduling should aim at maximizing the source s upload capacity utilization .
furthermore , the schedule must also try to ensure that the source injected data can be further used for the in - network redundancy generation . given a @xmath76 hsrc , where @xmath224 , any subset of @xmath5 linearly independent encoded fragments
forms a basis , denoted by @xmath225 ( see example [ ex : basis ] for an illustration ) .
let @xmath226 be the set of all the possible bases @xmath225 .
since each storage node stores one redundant fragment , we use @xmath227 to represent all the basis of @xmath226 whose corresponding storage nodes are available at a time step @xmath151 ( and likewise , refer to each combination of such nodes as an _ available basis _ ) : @xmath228 from the set of available basis , @xmath227 , the source node selects one basis @xmath225 and uploads some data to each node @xmath229 .
the amount of data the source uploads to each node @xmath229 is set to guarantee that at the end of time step @xmath151 , all these nodes have received the same amount of data , @xmath230 . from equations ( [ e :
throughput ] ) and ( [ e : innettraf ] ) we know that evening out the data all nodes receives allows to minimize network traffic and maximize insertion throughput .
to even out the amount of data each node in basis receives and maximizing the utilization of the upload capacity of the source , the source needs to send to each node @xmath229 an amount of data equal to @xmath231 besides determining how to distribute the upload capacity of the source between nodes of a basis @xmath225 , the source node also needs to select the basis @xmath225 from all the available ones @xmath227 ( if more than one basis is available ) .
we consider the following heuristic policies for the source node to select a specific basis @xmath225 : * * random : * @xmath225 is randomly selected from @xmath232 . repeating this procedure for several time steps is expected to ensure that all nodes receive approximately the same amount of data from the source . *
* minimum data : * the source selects the basis @xmath225 that on an average has received less redundant data .
it means that @xmath225 is the basis that minimizes @xmath233 .
this policy tries to homogenize the amount of data all nodes receive . *
* maximum data : * the source selects the basis @xmath225 that on an average has received more redundant data .
it means that @xmath225 is the basis that maximizes @xmath233 .
this policy tries to have a basis of nodes with enough data to allow the in - network redundancy generation for the entire data object even when the source may not be available . * * no basis : * the source does not considers any basis and instead uploads data to all the online nodes .
the upload bandwidth of the source is also distributed to guarantee that , after time step @xmath151 , all online nodes have received the same amount of data . at each time
step @xmath151 , once the source allocates its upload capacity to nodes from a specific available basis @xmath225 , the remaining upload / download capacity of the available nodes is used for in - network redundancy generation . for that purpose , the list of _ available triplets _ , @xmath234 ,
is determined as follows : @xmath235 then , the set of available triplets @xmath234 is sorted , and the available upload / download capacity of storage nodes allocated according to this priority ( i.e. , the first available triplets have more preference ) .
we consider the following sorting heuristics : * * random : * repair triplets are randomly sorted .
this policy tries to uniformly distribute the utilization of network resources to maximize the amount of in - network generated data . *
* minimum data : * the list of available triplets are sorted in ascending order according to the amount of data @xmath236 the destination is the destination of a triplet @xmath237 , @xmath191 . ]
node @xmath5 has received .
this policy tries to prioritize the redundancy generation in those nodes that have received less redundant data . *
* maximum data : * similarly to the _ minimum data _ policy , however , triplets are sorted in descending order .
this policy tries to maximize the amount of data some specific subset of nodes receive , to allow them to sustain the redundancy generation process even when the source is not available . * * maximum flow : * the triplets are sorted in descending order according to the amount of redundant data these nodes can help generate .
note that the amount of data a triplet @xmath237 can generate at each time step @xmath151 , where @xmath191 , is given by : @xmath238 this policy tries to maximize the amount of new redundancy generated per time step .
.different policy combinations . [ cols="^,^,^",options="header " , ] he have proposed four different policies for the source traffic scheduling problem and four policies for the triplets sorting problem . however , after an extensive experimental evaluation of all polices we will only report for each case the two best policies ( in terms of achieved throughput ) . at the source ,
the _ random _ and _ minimum data _ policies consistently outperform the others , and at the storage nodes , the _ maximum flow _ and _ minimum data _ sorting policies for the triplets likewise outperform the others .
we will refer to each of the combinations as denoted in table [ t : policies ] .
first , the interpretation of the good performance of the random policy in the source node is that the use of random bases favors the diversity of information among the nodes , which in turn enables more redundancy generation triplets .
second , it is interesting to note that the _ minimum data _ policy obtains good storage throughput in both cases , which leads us to infer that _ in general , prioritizing redundancy generation in those nodes that have received less data _ is a good strategy to maximize the throughput of the backup process .
we considered a @xmath42-hsrc , which is a code that can achieve a static data resiliency similar to a 3-way replication , but requiring only a redundancy factor of @xmath239.@xcite using this erasure code we simulated various backup processes with different node ( un)availability patterns for a fixed number of time steps @xmath175 .
in all the simulated cases we consider three different metrics : a. the maximum amount of data that can be stored in @xmath175 time steps , @xmath240 . b. the amount of data the source node uploads per unit of useful data backed up , @xmath241 c. the total traffic generated per unit of useful data stored , @xmath242 .
we evaluate the three metrics for a system using an in - network redundancy generation algorithm and we compare our results with a system using the naive erasure coding backup process , where the source uploads all the data directly to each storage node .
our results depict the savings and gains , in percentage , of using an in - network redundancy algorithm with respect to the naive approach . regarding the ( un)availability patterns of nodes and their bandwidth constraints we consider two different distributed storage cases : a. a p2p - like environment where we assume , to simplify simulations , that nodes have an upload bandwidth uniformly distributed between 20kbps and 200kbps , and an asymmetric download bandwidth equal to four times their upload bandwidth . nodes in this category follow two different availability traces from real decentralized application : ( i ) traces from users of an instant messaging ( i m ) service @xcite and traces from p2p nodes in the amule kad dht overlay @xcite . in both cases
we filter the nodes that on average stay online more than 4 , 6 and 12 daily hours , obtaining different mean availability scenarios .
b. real availability traces collected from a google datacenter .
the traces contain the normalized i / o load of more than 12,000 servers monitored for a period of one month .
we consider that a server is available to upload / download data when its i / o load is under the @xmath243-percentile load .
we consider three different percentiles , @xmath244 , giving us three different node availability constraints . finally the time step duration is set to @xmath245hour and we obtain the results by averaging the results of 500 backup processes of @xmath246 time steps each ( 5 days ) . before discussing the results
, we will like to note that the experiments make a few simplifying assumptions .
furthermore , real deployments have somewhat different workload characteristics than what have been considered above .
hence , the quantitative results we report are only indicative ( and many more settings could possibly be experimented ) , and instead the specific choices help us showcase the potential benefits of our approach in only a qualitative manner . [ [ storage - throughput . ] ] storage throughput .
+ + + + + + + + + + + + + + + + + + + in figure [ f : stored ] we show the increment of the data insertion throughput achieved by the in - network redundancy generation process .
we can see how the gain is higher when nodes are more available for redundancy generation .
this fact is a consequence of the constraint in eq .
( [ e : c : symmetry ] ) requiring redundancy generation triplets to be symmetric , which requires the three involved nodes in each triplet to be available simultaneously .
the higher the online availability , the higher the chances to find online three nodes from a triplet .
further , we observe that the _ rndflw _ policy achieves significantly better results in comparison to other policies ; the second best policy is _
minflw_. it is easy then to see that the _ maximum flow
_ heuristic plays an important role on the overall redundancy generation throughput , which tries to maximize the use of those nodes that can potentially generate more redundancy .
additionally , a _
random _ source selection policy provides more benefits than the _ minimum data _ policy .
[ [ network - traffic . ] ] network traffic .
+ + + + + + + + + + + + + + + + in figure [ f : traffic ] we show the increment on the required network traffic of the in - network redundancy generation strategy as compared to the traditional redundancy generation .
as noted previously ( in remark [ r : sourcetraffic ] ) , the total traffic required for in - network redundancy generation can be up to twice the needed by the traditional process ( i.e. 100% traffic increment ) .
however , since the in - network redundancy generation can not always take place due to the node availability constraints , the traffic increment is always below 100% .
as it is expected then , the traffic increment is minimized when nodes are less available , in which case the source has to generate and introduce larger amounts of redundancy ( i.e. , less reduction in the data uploaded by source , as shown in figure [ f : source ] ) .
it is also important to note that the increase in traffic is approximately the same or even less than the increase in storage throughput even for low availability scenarios .
thus the in - network redundancy generation scales well by achieving a better utilization of the available network resources than the classical storage process .
[ [ data - uploaded - by - the - source . ] ] data uploaded by the source .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + in figure [ f : source ] we show the reduction of data uploaded by the source . in the traditional approach , the source needs to upload @xmath239 times the size of the actual data to be stored ; @xmath247 of this data is redundant , however the in - network redundancy generation process allows to reduce the amount of data uploaded by the source . in this figure
we can see how in the best case ( _ rndflw _ policy ) our approach reduces the source s load by 40% ( out of a possible 57% ) , yielding 40 - 60% increase in storage throughput
. finally , we want to note that the in - network redundancy performance requires finding three available nodes simultaneously , which becomes difficult on environments with fewer backup opportunities . to solve this problem , we would need to look at more sophisticated in - network redundancy generation strategies not subjected to the symmetric constraint ( defined in eq .
( [ e : c : symmetry ] ) ) , so that nodes can forward and store partially - generated data .
however , the scheduling problem will be much more complicated , and is beyond the reach of this first work .
furthermore , in real traces , nodes will have correlation ( e.g. , based on batch jobs ) , which are missing in the synthetic traces , and such correlations can be leveraged in practice . exploring both
these aspects will be part of our future work .
in this work we propose and explore how storage nodes can collaborate among themselves to generate erasure encoded redundancy by leveraging novel erasure codes local - repairability property . doing so not only reduces a source node s load to insert erasure encoded data , but also significantly improves the overall throughput of the data insertion process .
we demonstrate the idea using self - repairing codes .
we show that determining an optimal schedule among nodes to carry out in - network redundancy generation subject to resource constraints of the system ( nodes and network ) is computationally prohibitive even under simplifying assumptions .
however , experiments supported by real availability traces from a google data center , and p2p / f2f applications show that some heuristics we propose yield significant gain in storage throughput under these diverse settings , proving the practicality of not only the idea in general , but also that of the specific proposed heuristics .
calder , b. , wang , j. , ogus , a. , nilakantan , n. , skjolsvold , a. , mckelvie , s. , xu , y. , srivastav , s. , wu , j. , simitci , h. , haridas , j. , uddaraju , c. , khatri , h. , edwards , a. , bedekar , v. , mainali , s. , abbasi , r. , agarwal , a. , haq , m.f.u . ,
haq , m.i.u .
, bhardwaj , d. , dayanand , s. , adusumilli , a. , mcnett , m. , sankaran , s. , manivannan , k. , and rigas , l. `` windows azure storage : a highly available cloud storage service with strong consistency . '' in _ acm symposium on operating systems principles ( sosp)_. 2011 .
ford , d. , labelle , f. , popovici , f.i .
, stokely , m. , truong , v.a . , barroso , l. , grimes , c. , and quinlan , s. `` availability in globally distributed storage systems . '' in _ usenix conference on operating systems design and implementation ( osdi)_. 2010 .
hastorun , d. , jampani , m. , kakulapati , g. , pilchin , a. , sivasubramanian , s. , vosshall , p. , and vogels , w. `` dynamo : amazon s highly available key - value store . '' in _
symposium on operating systems principles ( sosp)_. 2007 .
kubiatowicz , j. , bindel , d. , chen , y. , czerwinski , s. , eaton , p. , geels , d. , gummadi , r. , rhea , s. , weatherspoon , h. , weimer , w. , wells , c. , and zhao , b. `` oceanstore : an architecture for global - scale persistent storage . '' in _ intl .
conference on architectural support for programming languages and operating systems ( asplos)_. 2000 .
liu , s. , schulze , j.p . ,
herr , l. , weekley , j.d .
, zhu , b. , van osdol , n. , plepys , d. , and wan , m. `` cinegrid exchange : a workflow - based peta - scale distributed storage platform on a high - speed network . ''
_ future generation comp .
_ , 27(7):966976 , 2011 . | erasure coding is a storage - efficient alternative to replication for achieving reliable data backup in distributed storage systems . during the storage process , traditional erasure codes require a unique source node to create and upload all the redundant data to the different storage nodes .
however , such a source node may have limited communication and computation capabilities , which constrain the storage process throughput
. moreover , the source node and the different storage nodes might not be able to send and receive data simultaneously
e.g. , nodes might be busy in a datacenter setting , or simply be offline in a peer - to - peer setting which can further threaten the efficacy of the overall storage process .
in this paper we propose an `` in - network '' redundancy generation process which distributes the data insertion load among the source and storage nodes by allowing the storage nodes to generate new redundant data by exchanging partial information among themselves , improving the throughput of the storage process .
the process is carried out asynchronously , utilizing spare bandwidth and computing resources from the storage nodes .
the proposed approach leverages on the local repairability property of newly proposed erasure codes tailor made for the needs of distributed storage systems .
we analytically show that the performance of this technique relies on an efficient usage of the spare node resources , and we derive a set of scheduling algorithms to maximize the same .
we experimentally show , using availability traces from real peer - to - peer applications as well as google data center availability and workload traces , that our algorithms can , depending on the environment characteristics , increase the throughput of the storage process significantly ( up to 90% in data centers , and 60% in peer - to - peer settings ) with respect to the classical naive data insertion approach .
* keywords : distributed storage systems , data insertion , locally repairable codes , self - repairing codes * |
[ lissajous definition ] a lissajous knot @xmath2 in @xmath0 is one that can be represented parametrically by @xmath3 for integer frequencies @xmath4 , and @xmath5 , real phase shifts @xmath6 , and @xmath7 , and @xmath8 . in order for the parameterized curve to actually be a knot ,
that is , to not have any self - intersections , it follows easily that the three frequencies , @xmath4 and @xmath5 must be pair - wise relatively prime .
furthermore , there are restrictions on the phase shifts . if we assume that @xmath9 , then the @xmath10 and @xmath11 phase shifts must _ not _ be of the form @xmath12 ( see @xcite or @xcite . ) note that by replacing @xmath13 with @xmath14 we may change the phase shifts , making any one of them zero for example .
lissajous knots were first studied in @xcite where a few examples were given and a few properties of lissajous knots were derived . in particular it was shown that a lissajous knot with all odd frequencies is strongly plus amphicheiral and one with a single even frequency is @xmath15-periodic .
these symmetries are easy to see . in the first case , consider the orientation reversing homeomorphism @xmath16 given by negating each coordinate , or in other words , reflecting through the origin . if @xmath17 is a lissajous knot with all odd frequencies then @xmath18 furthermore , it is not difficult to check that the orientation of @xmath2 is preserved .
thus @xmath2 is strongly plus amphicheiral .
if one of the frequencies is even , say @xmath19 , then a similar argument will show that a 180 degree rotation of @xmath0 around the @xmath10-axis will take @xmath2 to itself .
thus @xmath2 is 2-periodic .
being strongly plus amphicheiral is a fairly restrictive property and among all prime knots to 12 crossings only 3 have this quality @xcite .
these are 10a103 ( @xmath20 , 10a121 ( @xmath21 ) , and 12a427 .
knot names are given in both the dowker - thistlethwaite ordering of the hoste - thistlethwaite - weeks table @xcite and , in parenthesis , the rolfsen @xcite ordering ( for knots with 10 or less crossings ) .
symmetries of the knots in the htw table were computed using _
snappea _ as described in @xcite .
the fact that 10a103 is lissajous is given in @xcite and to date , it is the only one of these knots that has been shown to be lissajous .
if @xmath2 is any knot and @xmath22 is its mirror image , then @xmath23 is strongly plus amphicheiral .
currently , only three lissajous knots of this form have been found , namely when @xmath2 is 3a1 ( @xmath24 ) , 5a1 ( @xmath25 ) , and 6a3 ( @xmath26 ) .
a list of lissajous knots with small crossing number which includes these examples is given in @xcite .
in the case where one of the frequencies is even , and the knot is @xmath15-periodic , actually more is known . in this case , it is shown in both @xcite and @xcite that the knot must have linking number @xmath27 with the rotational axis of symmetry .
both the trefoil knot and the figure eight knot for example are @xmath15-periodic . but
neither has such a symmetry where their linking number with the axis of rotation is @xmath27 .
this is because , as we will see shortly , neither can be a lissajous knot !
the symmetry of a lissajous knot has a strong effect on its alexander polynomial .
hartley and kawauchi @xcite show that a strongly plus amphicheiral knot must have an alexander polynomial which is a perfect square . for knots that are @xmath15-periodic and
link their axis of rotation @xmath27 times , murasugi @xcite has shown that the alexander polynomial must be a square modulo 2 . from these facts
, lamm shows that many large classes of knots are not lissajous .
( see also @xcite . ) for example , if @xmath28 and @xmath29 are relatively prime integers larger than 1 , then the @xmath30-cable of any knot is not lissajous . in particular , torus knots ,
iterated torus knots and non - trivial algebraic knots can not be lissajous .
( thus the trefoil is not lissajous . )
lamm also points out that these conditions on the alexander polynomial also imply that fibered knots with odd genus , and fibered 2-bridge knots are not lissajous .
( thus the figure eight knot is not lissajous . )
it is well known that the arf invariant of a knot may be derived from its alexander polynomial ( see the proof of lemma [ alexpolyformula ] ) , and in @xcite it is shown that if the alexander polynomial is a square modulo 2 then the arf invariant is zero .
thus the trefoil and figure eight knots can not be lissajous as each has an arf invariant of 1 .
the fact that the arf invariant of a lissajous knot is zero was also proven in @xcite by a different method .
all of the necessary conditions mentioned so far for a knot to be lissajous are derived from the symmetry of the knot : either strongly plus amphicheiral , or 2-periodic with linking number @xmath27 with the axis of rotation . in general , the implied statements about the alexander polynomial are strictly weaker and the fact that the arf invariant must be zero is weaker still . in rolfsen ) has arf invariant zero but its alexander polynomial is not a square modulo 2 .
the knot 8n1 ( @xmath31 in rolfsen ) has alexander polynomial which is a square modulo 2 but it is not @xmath28-periodic for any @xmath28 .
( see table i in @xcite ) .
finally , the knot 8n1 ( @xmath31 in rolfsen ) has alexander polynomial a perfect square , but it is not strongly plus amphicheiral . ] to date these are the only known necessary conditions for a knot to be lissajous and it seems unlikely that these necessary conditions are sufficient .
however , in this paper we show that these conditions are sufficient for the class of twist knots .
in fact , a twist knot is lissajous if and only if its arf invariant is zero . for those twist knots that are not lissajous we know that infinitely many are _ second order _ lissajous .
a second order lissajous knot is defined as follows .
a second order lissajous knot @xmath2 is one that can be represented parametrically by @xmath32 for integer frequencies @xmath33 , and @xmath34 , real phase shifts @xmath35 , and @xmath36 , and @xmath8 .
we believe the following conjecture to be true , and hope to include this result in a forthcoming paper .
every twist knot with arf invariant 1 is second order lissajous .
[ arf1conjecture ] note that a second order lissajous knot has a _ lissajous projection _ ( in the @xmath10-@xmath11 plane ) . by employing large enough frequencies
it seems reasonable to make the following conjecture .
every knot has a lissajous projection .
[ lpconjecture ] while not explicitly stated as a conjecture in @xcite , this belief is evident in their paper and we echo their sentiment here .
if conjecture [ lpconjecture ] is true , then the height function @xmath37 can be estimated by a finite sum of cosine functions .
whether just two ( or one ) cosine functions would then suffice in all cases , as we believe it does for all twist knots , is an interesting question . in section [ twist knots ]
we show that a twist knot is lissajous if and only if its arf invariant is zero .
next , in section [ lp ] we give a general result that can be used to show that a knot has a lissajous projection .
we use this to show that all 2-bridge knots have lissajous projections .
finally , in the last section we continue with the techniques in section [ lp ] to show that all @xmath1-torus knots have lissajous projections . * acknowledgement * this research was carried out during the summer of 2005 as part of a _ research experiences for undergraduates _ program administered by the claremont colleges with funding from the national science foundation .
the authors wish to thank the claremont colleges for their support and hospitality as well as both the claremont colleges and the national science foundation for funding the reu site .
* postscript * after completing this paper we learned that lamm has proven conjecture [ lpconjecture ] .
the twist knot @xmath38 is defined as in figure [ twistknot ] .
there are @xmath39 crossings in the `` twisty '' part and then two more crossings in the left - handed `` clasp . ''
if @xmath40 then the given diagram is alternating ( and reduced ) and thus depicts a knot with crossing number equal to @xmath41 .
if @xmath42 , then the crossings in the `` twisty '' part have the opposite handedness . for example , @xmath43 is the figure eight knot , @xmath44 is the left - handed trefoil , and @xmath45 is the unknot .
it is not difficult to show that @xmath38 and @xmath46 are mirror images of each other .
we will consider two knots the same if there is a homeomorphism of @xmath0 to itself taking one knot to the other and therefore will not distinguish between a knot and its mirror image .
thus , it suffices to consider only the case where @xmath47 is even .
we begin with the following lemma .
the alexander polynomial and the arf invariant of the twist knot @xmath48 are [ alexpolyformula ] @xmath49 * proof : * the first part is well known .
see @xcite for example . for any knot @xmath50
, the arf invariant may be derived from the alexander polynomial as follows .
@xmath51 in our case , @xmath52 which is 1 mod 8 if @xmath53 is even and 5 mod 8 if @xmath53 is odd
. @xmath54 from lemma [ alexpolyformula ]
we see that the alexander polynomial of a twist knot is never a perfect square and hence twist knots are never strongly plus amphicheiral .
thus none can be lissajous knots with all odd frequencies .
it remains to consider the possibility that one frequency is even .
the following lemma narrows the scope of our investigation . for twist knots
@xmath38 , the following are equivalent : [ equivalent conditions for twist knots ] 1 .
@xmath38 is 2-periodic and has linking number @xmath55 with its rotational axis of symmetry .
the alexander polynomial @xmath56 is a square modulo 2 .
the arf invariant of @xmath38 is zero .
* proof : * for any knot , @xmath57 .
thus it remains to prove @xmath58 . as mentioned already
, we may assume that @xmath47 is an even integer .
suppose the arf invariant of @xmath48 is zero .
thus lemma [ alexpolyformula ] implies that @xmath53 must be even .
figure [ symmetry ] now shows that @xmath38 is 2-periodic . in the figure ,
the box labeled @xmath53 indicates a total of @xmath53 crossings .
the axis of rotation is perpendicular to the plane of the diagram and appears as a heavy dot in the center of the figure .
the reader should check that the linking number with the axis is @xmath27 if @xmath53 is even and @xmath59 if @xmath53 is odd .
@xmath54 because of lemma [ equivalent conditions for twist knots ] it is quite natural to ask whether or not twist knots with arf invariant equal to zero are lissajous . by exploring lissajous knots with small frequencies
one quickly finds that if @xmath60 and @xmath61 is odd , then the lissajous knot has a projection which _ could _ be the projection of a twist knot if all of the crossings are arranged in the appropriate way .
for example , consider the two lissajous knots shown in figure [ m=2 or 3 ] where @xmath62 or 7 and @xmath63 . arbitrarily choosing @xmath63 ,
we are led to the following question : if @xmath64 and @xmath63 , can we choose @xmath5 and @xmath7 so as to produce a twist knot ? in order to study this question we begin by investigating the double points of the @xmath65-projection of a general lissajous knot .
let @xmath17 be the lissajous knot parameterized as in definition [ lissajous definition ] .
suppose @xmath66 and @xmath67 give a double point in the projection to the @xmath65-plane .
then the set of such pairs @xmath68 fall into two types .
[ cols= " < , > , < " , ] moreover , there are @xmath69 doubles points of type i , and @xmath70 double points of type ii .
* proof : * we must find all pairs of times @xmath68 such that @xmath71 and @xmath72 . using the fact that @xmath73 if and only if @xmath74 is an integer multiple of @xmath75 , one is led to a number of cases that must be analyzed .
see @xcite for further details .
let @xmath76 and @xmath77 denote the crossings that occur at the double points of type i and ii respectively .
by @xmath78 or @xmath79 we will mean the @xmath10 and @xmath11 values at that crossing .
suppose now that @xmath64 and @xmath63 .
then there are @xmath80 double points of type i. for each of these , @xmath81 and @xmath82 .
the time pairs are @xmath83 it is easy to show that @xmath84 for all @xmath85 and hence each type i crossing lies on the @xmath10-axis .
furthermore , it is not difficult to show that no type i crossing can occur at the origin and the number of type i double points that fall on the positive @xmath10-axis is @xmath86 .
if we now look at the type ii double points we see that there are @xmath87 of these given by the time pairs @xmath88 where @xmath89 if @xmath90 or @xmath91 and @xmath92 if @xmath93 .
if we compute the value of @xmath11 at these crossings we obtain @xmath94 by computing the @xmath10 value of the type ii crossings we obtain @xmath95 for all @xmath96 .
furthermore , the number of type ii crossings @xmath77 with @xmath97 is @xmath98 . of these
, @xmath99 have @xmath93 while @xmath86 have @xmath90 or @xmath91 .
finally , since @xmath60 , the knot is rotationally symmetric around the @xmath10-axis .
thus the signs of @xmath100 and @xmath101 are the same as are the signs of @xmath102 and @xmath103 .
we now want to understand how the choice of @xmath5 and @xmath7 determines the crossing at each of the double points in the @xmath65-projection .
note first that increasing the phase shift by @xmath104 will reflect the knot though the @xmath65-plane .
thus we may assume that @xmath105 .
suppose that at a double point of either type , @xmath106 .
this occurs if and only if the sum or difference of @xmath107 and @xmath108 is an integer multiple of @xmath75 .
if we examine the type ii double points first , we find that all of the crossings @xmath102 become singular precisely when @xmath9 or @xmath104 .
all the other type ii crossings become singular precisely when @xmath109 . furthermore , both of these statements are independent of @xmath5 . in particular , once @xmath5 is chosen , we are free to vary @xmath7 between @xmath110 and @xmath111 without changing any of the type ii crossings !
we now turn our attention to the type i crossings .
since @xmath5 is odd , it is not hard to show that the difference of @xmath107 and @xmath108 can not be an even multiple of @xmath104 .
however , their sum is an even multiple of @xmath104 if and only if @xmath112 for some integer @xmath113 . therefore , there are @xmath114 values of @xmath7 between @xmath110 and @xmath104 where all the type i crossings become singular . these values , together with @xmath111 , give @xmath115 phases that must be avoided .
the locations of these singular phases depend on @xmath5 but the number of them depend only on @xmath39 .
thus , for a given choice of @xmath39 there are @xmath116 intervals where we may safely choose @xmath7 and varying @xmath7 within each interval will not cause any of the crossings to become singular .
hence there are at most @xmath116 different lissajous knot types with @xmath117 and @xmath118 . at this point
it is a simple matter to program a computer to search through a large number of pairwise relatively prime frequencies @xmath119 and for each of these construct the @xmath120 possible lissajous knots .
looking for twist knots in this way led to the following theorem .
[ arf0ifflissajous ] a twist knot is lissajous if and only if its arf invariant is zero . in particular , the lissajous knot @xmath121 for @xmath122 is equivalent to @xmath123 if @xmath39 is even and @xmath124 if @xmath39 is odd .
* proof : * if @xmath125 or @xmath126 it is easy to verify the result .
thus , in everything that follows we assume that @xmath127 .
diagrams for @xmath128 and 3 are shown in figure [ m=2 or 3 ] . a nice pattern exists that we will show persists for all @xmath39 .
consider the checkerboard coloring of the diagram and assume the unbounded region has been colored white .
there are exactly @xmath39 white regions each having 4 edges and 4 vertices .
we call these _ diamonds_. each diamond lies on the @xmath10-axis and contains two type i crossings and two type ii crossings .
there are a total of @xmath39 diamonds and when @xmath39 is even , half lie on each side of the @xmath11-axis . when @xmath39 is odd , @xmath129 lie in the right half - plane and @xmath130 lie in the left half - plane .
we claim that each diamond must appear as one of the four possibilities shown in figure [ band crossing ] , which we call _ left band crossings _ , _ clasps _ , or _
right band crossings_. in figure [ m=2 or 3 ] for example , notice that as we move from left to right the diamonds form a ( possibly empty ) string of left band crossings , then a clasp , and then a string of right band crossings .
thus all of the right band crossings `` unravel '' and the clasp of the twist knot is formed at the diamond that is of type b ) or c ) in figure [ band crossing ] . from this
it clearly follows that the knot is indeed a twist knot .
it suffices to prove the following four claims : 1 .
if @xmath39 is even , then all type i crossings in the right half - plane are left - handed and all type i crossings in the left half - plane , _ except _ the one closest to the @xmath11-axis , are right - handed .
if @xmath39 is odd , then all type i crossings in the left half - plane are right - handed and all type i crossings in the right half - plane , _ except _ the one closest to the @xmath11-axis , are left - handed .
type ii crossings with @xmath131 are left - handed if and only if they lie in the right half - plane .
type ii crossings with @xmath81 or @xmath91 are right - handed if and only if they lie in the right half - plane . from these claims
it is easy to establish 1 .
if @xmath39 is even , the first diamond to the left of the @xmath11-axis is a clasp as shown in figure [ band crossing]b .
if @xmath39 is odd , the first diamond to the right of the @xmath11-axis is a clasp as shown in figure [ band crossing]c . in either case ,
all diamonds to the right of this clasp are right band crossings and all diamonds to its left are left band crossings .
once statements 15 are proven we obtain the desired result as follows .
each of the left band crossings contributes one full left - handed twist to the _ band _ of the twist knot and each of the type ii crossings with @xmath131 in the left half - plane contributes half a left - handed twist .
notice that since the edges of the band are oriented oppositely , these are actually right - handed crossings . finally , the left - most type i crossing is right - handed and contributes half a left - handed twist to the band .
thus if @xmath39 is odd for example , all of these crossings add up to @xmath132 right half - twists in the band .
the clasp of the twist knot is right - handed .
thus this knot is the mirror image of @xmath133 which is equivalent to @xmath124 . in order to establish statements
14 we study the signs of the crossings .
suppose @xmath68 is a pair of times that gives a double point , and let @xmath134 be the tangent vector to the @xmath65-projection at @xmath135 .
the cross product @xmath136 is now a vector perpendicular to the @xmath65-plane .
the following lemma is proven in @xcite .
[ sign of crossing]the sign of the crossing corresponding to the times @xmath137 is @xmath138.\ ] ] using lemma [ sign of crossing ] , somewhat lengthy , but nevertheless straightforward , calculations show that statements 14 are correct .
we illustrate the proof of statement 3 and leave the others to the reader . considering a type ii crossing with @xmath93 , we obtain @xmath139 @xmath140 hence the sign of @xmath141 is the same as the sign of @xmath142 next
, we consider @xmath143 .
as mentioned earlier , varying @xmath7 between @xmath110 and @xmath144 will not change the type ii crossings with @xmath93 .
thus we may replace @xmath37 with @xmath145 .
we obtain , @xmath146 with the sign of @xmath147 equal to @xmath148 $ ] .
finally , using lemma [ sign of crossing ] , the sign of @xmath102 is @xmath149&=&\mbox{sign}\left[(-1)^k\sin\frac{2 k \pi}{2 m+1 } ( -1)^{k+1}\sin\frac{4 k \pi}{2 m+1}\right]\\ & = & -\mbox{sign}\left[\sin\frac{2 k \pi}{2 m+1 } \sin\frac{4 k \pi}{2 m+1}\right]\\ & = & -\mbox{sign}\left[\cos\frac{2 k \pi}{2 m+1}\right]\\ & = & -\mbox{sign}\left [ x(c^{ii}_{k,2})\right].\end{aligned}\ ] ] @xmath54
theorem [ arf0ifflissajous ] shows that every twist knot with arf invariant equal to zero has a lissajous projection with @xmath60 and @xmath61 some odd integer .
our first goal in this section is to show that all 2-bridge knots have lissajous projections . to do this we first introduce the concepts of a _ lissajous arc _ and a _ lissajous braid_. figure [ lissajous projection ]
illustrates the lissajous projection with frequencies @xmath150 and @xmath151 and phase shifts @xmath152 and @xmath153 . in figure
[ lissajous arc ] we have set both phase shifts to zero .
notice that the first figure appears like the projection of a long thin twisted disk or band whose `` core '' is the arc of the second figure . as the phase shift @xmath154
is increased from zero , the arc in the second figure expands to the band of the first figure . a double point in figure
[ lissajous arc ] becomes four crossings in figure [ lissajous projection ] . but
additional crossings are also introduced because the band does not lay flat in the plane of the projection . instead
, think of the band as twisted , with these additional crossings located at the extreme points of the arc in both the @xmath10 and @xmath11 directions .
these observations motivate the following two definitions .
a lissajous arc with frequencies @xmath19 and @xmath61 is the plane figure parameterized by @xmath155 for @xmath156 , where @xmath19 and @xmath61 are relatively prime positive integers .
notice that if small neighborhoods of all the maxima and minima in the @xmath11 direction are removed from a lissajous arc , we are left with the projection of a braid on @xmath61 strings .
suppose @xmath157 is an @xmath53-string braid expressed as a product of the standard braid generators @xmath158 and their inverses .
here @xmath159 denotes the braid where string @xmath160 passes over string @xmath113 as one travels down the braid .
let @xmath161 denote the obvious projection of the braid obtained by replacing @xmath162 with a double point which we will denote as @xmath163 .
a lissajous braid with frequencies @xmath19 and @xmath61 is any braid @xmath157 on @xmath61 strings whose projection @xmath161 is obtained from a lissajous arc of the same frequencies by removing small neighborhoods of all the minima and maxima in the @xmath11 direction . the following characterization of lissajous braids is straightforward and is left to the reader .
let @xmath157 be a lissajous braid with frequencies @xmath19 and @xmath61 .
define @xmath164 and @xmath165 as @xmath166 where @xmath167 and @xmath168 are the largest even and odd integers less than @xmath61 . then @xmath169 the number of crossings in the braid ( or arc of the same frequencies )
is @xmath170 .
the number of crossings in a lissajous projection with these frequencies is @xmath171 .
thus a lissajous braid has a projection which is an alternating product of @xmath164 and @xmath165 .
however not all such products are possible .
for example , the projection can never begin and end with @xmath165 and the fact that @xmath19 and @xmath61 are relatively prime impose further restrictions as well .
the following two lemmas will play a key part in showing that every 2-bridge knot has a lissajous projection .
suppose @xmath157 is any @xmath53-string braid with @xmath172 of the form @xmath173 where @xmath174 .
thus @xmath175 .
then there exist equivalent braids @xmath176 and @xmath177 such that @xmath178 [ first replacement lemma ] * proof : * by using two type i and one type iii reidemeister moves we may change @xmath157 to @xmath179 we now have @xmath180 .
a similar argument gives @xmath177 .
@xmath54 any 3-string braid @xmath157 is equivalent to a braid whose projection is an alternating product of @xmath181 and @xmath182 .
[ 3 string braids are nearly lissajous ] * proof : * we proceed by induction on the length ( that is , the number of crossings ) of @xmath157 .
if @xmath157 has length one , then @xmath161 equals either @xmath181 or @xmath182 and we are done .
there are now two cases to consider : either @xmath183 or @xmath184 , where @xmath185 is a shorter braid .
suppose @xmath184 . by induction
, we may assume that @xmath185 is equivalent to a braid , which we continue to call @xmath185 , whose projection alternates between @xmath181 and @xmath182 .
if @xmath186 begins with @xmath181 , then we are done , because @xmath161 now alternates between @xmath182 and @xmath181 .
if @xmath186 begins with @xmath182 , then @xmath161 begins with @xmath187 , which we can replace with @xmath188 by lemma [ first replacement lemma ] .
a similar argument works if @xmath183 .
@xmath54 the closure of a lissajous arc is the knot projection obtained by connecting the ends of a lissajous arc with an arc whose interior is disjoint from the lissajous arc and which introduces no additional crossings .
the significance of this definition will be borne out in the following theorem .
[ closure of la implies lp ] suppose the knot @xmath2 has a projection which is the closure of a lissajous arc with frequencies @xmath19 and @xmath61 .
then @xmath2 has a lissajous projection with the same frequencies .
* proof : * we demonstrate the idea of the proof with the following example . consider the trefoil knot as shown on the left in figure [ trefoil2lissajous ] .
this projection is the closure of a lissajous arc .
the next step is illustrated in the middle and right - hand diagrams in figure [ trefoil2lissajous ] .
imagine picking up the right - hand strand of the diagram ( this is the arc that was used to close the lissajous arc ) and then laying it back down atop the figure , retracing the lissajous arc .
as the lissajous arc is retraced we also introduce crossings as needed to achieve a lissajous projection .
the additional crossings are located at each of the extreme points in both the @xmath10 and @xmath11 directions .
note that the final lissajous projection is equivalent to the one shown in figure [ lissajous projection ] .
@xmath54 we now have everything in place to prove the following theorem .
[ 2 bridge implies lp ] every 2-bridge knot has a lissajous projection with @xmath189 .
* proof : * every 2-bridge knot has a diagram which is the plat closure of a 4-string braid that uses only @xmath190 and their inverses .
we may think of this braid as a 3-string braid @xmath157 that has had a trivial fourth string added along the right side . by lemma [ 3 string braids are nearly lissajous ] we may alter @xmath157 so that its projection alternates between @xmath181 and @xmath182 . since we are adding a fourth string to @xmath157 and then taking its plat closure ,
we may remove all beginning and trailing appearances of @xmath191 or its inverse from @xmath157 .
hence we may assume that the projection of @xmath157 is of the form @xmath192 for some @xmath96 .
if @xmath193 mod 3 , then it is not hard to check that the plat closure will produce a link rather than a knot .
thus we must have that @xmath96 is either zero or one mod 3 . in either case
@xmath157 is now a lissajous braid with frequencies @xmath194 and @xmath151 .
finally , we now have that the diagram is the closure of a lissajous arc and theorem [ closure of la implies lp ] applies .
the simplest torus knots are the @xmath195-torus knots .
since these are 2-bridge knots , theorem [ 2 bridge implies lp ] implies the following result .
we may think of the @xmath1-torus knot as the plat closure of the 6-string braid @xmath196 . our strategy is to focus on the five left - most strings , changing this braid into a lissajous braid .
the torus knot will then have a projection which is the closure of a lissajous arc and theorem [ closure of la implies lp ] will apply .
let @xmath185 be a 6-string braid with projection @xmath206 .
then there is a 6-string braid @xmath204 with projection @xmath207 such that the plat closure of @xmath185 is equivalent to the knot obtained from @xmath204 by using ordinary plat closure at the top of @xmath204 and the modified plat closure shown in figure [ fourth 6-string lemma figure ] at the bottom of @xmath204 .
[ fourth 6-string lemma ] * proof : * consider the right - most string in the left projection of figure [ fourth 6-string lemma figure ] .
let @xmath208 be the crossing involving this string and having the projection @xmath209 .
if @xmath210 , then we may slide the right - most string over all the other strings as shown in the figure . if @xmath211 , then we may slide the string under all the other strings .
we may now use lemmas [ first 6-string lemma ] to [ fourth 6-string lemma ] to prove that all @xmath1-torus knots have lissajous projections .
as we have already mentioned , no torus knot is lissajous .
thus the height function corresponding to these projections can not be given by a single cosine function .
by lemma [ first 6-string lemma ] , we may replace @xmath185 with a braid whose projection is now @xmath219 if we now apply the first part of lemma [ second 6-string lemma ] we may further alter the braid to one whose projection is @xmath220 finally , using lemma [ fourth 6-string lemma ] we may change the braid so that its projection is @xmath221 this last change requires that we change the way the braid is closed at the bottom as described in the lemma .
we have now reached a projection which is the closure of a lissajous arc with frequencies @xmath212 and @xmath222 .
hence , by theorem [ closure of la implies lp ] , the torus knot has a lissajous projection with these same frequencies .
we begin by rewriting the projection of @xmath185 as @xmath224 applying the second part of lemma [ second 6-string lemma ] we may change the braid so that its projection is now @xmath225 finally , using lemma [ third 6-string lemma ] , we may change to a projection of @xmath226 as in case i , we have now reached a projection which is the closure of a lissajous arc with frequencies @xmath212 and @xmath227 . | knots in @xmath0 which may be parameterized by a single cosine function in each coordinate are called _ lissajous _ knots .
we show that twist knots are lissajous knots if and only if their arf invariants are zero .
we further prove that all 2-bridge knots and all @xmath1-torus knots have lissajous projections . |
we may identify the zero - point energy of a system of quantum fields with the vacuum expectation value of the field energy , @xmath0 in the vacuum both sides of this equality are divergent and meaningless .
what is observable is the _ change _ in the zero - point energy when matter is introduced . in this way we can calculate the casimir forces . for a massless scalar field ,
the canonical energy - momentum tensor is @xmath1 the vacuum expectation value may be obtained by taking derivatives of the casual green s function : @xmath2 alternatively , we can calculate the stress on the material bodies .
consider the original geometry considered by casimir , where he calculated the quantum fluctuation force between parallel , perfectly conducting plates separated by a distance @xmath3 @xcite .
the force per unit area @xmath4 on one of the plates is given in terms of the normal - normal component of the stress tensor , @xmath5 for electromagnetic fields , the relevant stress tensor component is @xmath6 we impose classical boundary conditions on the surfaces , @xmath7 and the calculation of the vacuum expectation value of the field components reduces to finding the classical te and tm green s functions . in general ,
one further has to subtract off the stress that the formalism would give if the plates were not present , the so - called volume stress , and then the result of a simple calculation , which is sketched below , is @xmath8=-{\pi^2\over 240 a^4 } \label{casimirsforce } \hbar c,\ ] ] an attractive force .
the dependence on the plate separation @xmath3 is , of course , completely determined by dimensional considerations .
numerically , the result is quite small , @xmath9 and will be overwhelmed by electrostatic repulsion between the plates if each plate has an excess electron surface density @xmath10 greater than @xmath11 , from which it is clear that the experiment must be performed at the @xmath12 m level .
nevertheless , over twenty years , many attempts to measure this force directly were made @xcite .
( the cited measurements include insulators as well as conducting surfaces . ) until recently , the most convincing experimental evidence came from the study of thin helium films @xcite ; there the corresponding lifshitz theory @xcite has been confirmed over nearly 5 orders of magnitude in the van der waals potential ( nearly two orders of magnitude in distance ) .
quite recently , the casimir effect between conductors has been confirmed at the 5% level by lamoreaux @xcite , and to perhaps 1% by mohideen and roy @xcite .
( in order to achieve the stated accuracy , corrections for finite conductivity , surface distortions , and perhaps temperature must be included .
for a brief review see ref .
here we wish to concentrate on dimensional dependence . for simplicity
we consider a massless scalar field @xmath13 confined between two parallel plates in @xmath14 spatial dimensions separated by a distance @xmath3 .
assume the field satisfies dirichlet boundary conditions on the plates , that is @xmath15 the casimir force between the plates results from the zero - point energy per unit ( @xmath16-dimensional ) transverse area @xmath17 where we have set @xmath18 , and introduced normal modes labeled by the positive integer @xmath10 and the transverse momentum @xmath19 .
this may be easily evaluated by introducing a proper - time representation for the square root , and by analytically continuing from negative @xmath16 we obtain @xmath20 which reduces to the familiar casimir result at @xmath21 : @xmath22 this is , as expected , 1/2 of the electromagnetic result ( [ casimirsforce ] ) .
this less - than - rigorous calculation can be put on a firm footing by a green s function technique .
define a reduced green s function by @xmath23 the ( interior ) solution of which vanishing at @xmath24 , @xmath3 , being ( @xmath25 ) @xmath26 where @xmath27 ( @xmath28 ) is the greater ( lesser ) of @xmath29 and @xmath30 .
the force per unit area on the surface @xmath31 is obtained by taking the discontinuity of the normal - normal component of the stress tensor : @xmath32 this is easily evaluated by doing a complex rotation in frequency : @xmath33 : @xmath34 evidently , eq .
( [ scalarforce ] ) is the negative derivative of the casimir energy ( [ scalarenergy ] ) with respect to the separation between the plates : @xmath35 this result has now been obtained by a completely well - defined approach , so arguments about the conceptual validity of the casimir effect are seen to be without merit . the force per unit area , eq . ( [ scalarforce ] ) , is plotted in fig .
[ fig2 ] , where @xmath36 and @xmath37 .
this general result was first derived by ambjrn and wolfram @xcite .
the effect of massless fermionic fluctuations between parallel plates embedded in three dimensional space , subject to `` bag model '' boundary conditions , @xmath38 where @xmath39 is the normal to the surface , was first calculated by ken johnson @xcite . in place of eq .
( [ zeropt2 ] ) , the fermionic casimir energy for @xmath21 is formally @xmath40 so once the @xmath19 integral is performed the energy is proportional to @xmath41 thus , the result is 7/4 times the scalar casimir energy , @xmath42 the casimir effect so implemented breaks supersymmetry .
however , in a susy theory , if supersymmetric boundary conditions are imposed ( the unconstrained components of the fermion and scalar fields are all periodic , for example ) , the fermionic casimir energy will just cancel that due to the bosons .
( for a simple example of how this works , see ref .
@xcite . ) because of the rather mysterious dependence of the sign and magnitude of the casimir stress on the topology and dimensionality of the bounding geometry , we have carried out a calculation of te and tm modes bounded by a spherical shell in @xmath43 spatial dimensions .
we first consider massless scalar modes satisfying dirichlet boundary conditions on the surface , which are equivalent to electromagnetic te modes .
again we calculate the vacuum expectation value of the stress on the surface from the green s function . the green s function @xmath44 , eq .
( [ vevgreen ] ) , satisfies the inhomogeneous klein - gordon equation , or @xmath45 where @xmath46 is the laplacian in @xmath43 dimensions .
we solve the above green s function equation by dividing space into two regions , the interior of a sphere of radius @xmath3 and the exterior of the sphere . on the sphere we impose dirichlet boundary conditions
@xmath47 in addition , in the interior we require that @xmath48 be finite at the origin @xmath49 and in the exterior we require that @xmath48 satisfy outgoing - wave boundary conditions at @xmath50 , that is , for a given frequency , @xmath51
. the radial casimir force per unit area @xmath4 on the sphere is obtained from the radial - radial component of the vacuum expectation value of the stress - energy tensor : @xmath52 to calculate @xmath4 we exploit the connection between the vacuum expectation value of the stress - energy tensor @xmath53 and the green s function at equal times @xmath54 : @xmath55_{{\bf x}={\bf x}',~|{\bf x}|=a}. \label{8}\end{aligned}\ ] ] adding the interior and the exterior contributions , and performing the usual imaginary frequency rotation , we obtain the expression for the stress @xcite : @xmath56 it is easy to check that this reduces to the known case at @xmath57 , for there the series truncates only @xmath58 and 1 contribute , and we easily find @xmath59 which agrees with eq .
( [ scalarforce ] ) for @xmath60 and @xmath36 .
in general , we proceed as follows : * analytically continue to @xmath61 , where the sum ( [ e4 ] ) converges , although the integrals become complex . *
add and subtract the leading asymptotic behavior of the integrals .
* continue back to @xmath62 , where everything is now finite .
the results of numerical evaluations for the total stress @xmath63 on the sphere are as shown in fig .
[ figvdc ] .
note the following features for the scalar modes : * poles occur at @xmath64 , @xmath65 .
* branch points occur at @xmath66 , @xmath67 , and the stress is complex for @xmath61 . *
the stress vanishes at negative even integers , @xmath68 , @xmath65 , but is nonzero at @xmath69 : @xmath70 . *
the case of greatest physical interest , @xmath71 , has a finite stress , but one which is much smaller than the corresponding electrodynamic one : @xmath72 .
( this result was confirmed in ref .
@xcite . )
the tm modes are modes which satisfy mixed boundary conditions on the surface , @xmath73 the results are qualitatively similar , and are also shown in fig .
[ figvdc ] . in particular
, removing the @xmath58 contribution from the sum of the te and tm contributions , we recover the repulsive boyer result in three dimensions @xcite , @xmath74 for the 3-sphere , the fermionic casimir energy subject to the boundary condition ( [ bmbc ] ) is a factor of two smaller @xcite @xmath75 a similar calculation of the electromagnetic casimir effect of a perfectly conducting infinite right circular cylinder in three dimensions was performed by deraad and me twenty years ago .
the calculation is rather more involved , and the regularization of the divergences more subtle .
the result for the casimir energy per unit length , or the force per unit area , is @xcite @xmath76 unlike the three - dimensional sphere , the cylinder experiences an attractive casimir stress .
two recent calculations have confirmed this result using zeta - function techniques @xcite .
over 40 years ago , lifshitz and collaborators @xcite worked out the corresponding forces between dielectric slabs .
imagine we have a permittivity which depends on @xmath29 as follows : @xmath77 then the lifshitz force between the bodies at zero temperature is given by ( @xmath78 , @xmath79 is the imaginary frequency ) @xmath80 here the denominators are given by , for the electric ( tm ) green s function , @xmath81 the magnetic ( te ) green s function has the same form as the electric one but with the replacement @xmath82 except in the exponentials ; the corresponding denominator is denoted by @xmath83 . from this
, we can obtain the finite temperature expression immediately by the substitution @xmath84 @xmath85 the prime being a reminder to count the @xmath58 term with half weight .
( for a fuller discussion of temperature dependence , see ref .
@xcite . )
if the central medium is tenuous , @xmath86 , and is surrounded by vacuum , for large distances @xmath87 , where @xmath88 is a characteristic wavelength of the medium , we can expand the above general formula and obtain a dispersion - free result : @xmath89 for this regime , this should be derivable from the sum of van der waals forces , obtained from an intermolecular potential of the form @xmath90 we do this by computing the energy ( @xmath91 density of molecules ) @xmath92^{\gamma/2}}.\end{aligned}\ ] ] if we disregard the infinite self - interaction terms ( analogous to dropping the volume energy terms in the casimir calculation ) , we get @xmath93 so then , upon comparison with ( [ longdist ] ) , we set @xmath94 and in terms of the polarizability , @xmath95 we find @xmath96 or , equivalently , we recover the casimir - polder retarded dispersion potential @xcite , @xmath97 whereas for short distances ( @xmath98 ) we recover the london potential @xcite , @xmath99 given the divergences of the above calculation , and the essentially one - dimensional restriction , it is of interest to consider a tenuous dielectric sphere .
the theory of the casimir energy for a dielectric ball was first worked out by me 20 years ago @xcite .
the general expression is of course quite complicated @xmath100 where @xmath101 ^ 2-\xi^2[s_l(x')e'_l(x)+s_l'(x')e_l(x)]^2,\end{aligned}\ ] ] where the @xmath102 , @xmath103 are spherical bessel functions of imaginary argument , the quantity @xmath104 is @xmath105 where @xmath106 , @xmath107 represent the permittivity and permeability in the interior , the corresponding unprimed quantities refer to the exterior , the time - splitting regularization parameter is denoted by @xmath108 , and @xmath109 it is easy to check that this result reduces to that for a perfectly conducting spherical shell @xcite if we set the speed of light inside and out the same , @xmath110 , as well as set @xmath111 . however , if the speed of light is different in the two regions , the result is no longer finite , but quartically divergent , and indeed the schwinger result @xcite follows for that leading divergent term .
although , in general , this expression is not finite , there are several methods of isolating the divergences , at least if the ball is tenuous , ( @xmath86 ) , and the finite repulsive observable casimir energy is @xcite @xmath112 what is most remarkable about this result is that it coincides with the van der waals energy calculated two years earlier for this nontrivial geometry .
that is , starting from the casimir - polder potential ( [ caspol ] ) we summed the pairwise potentials between molecules making up the media .
a sensible way to regulate this calculation is dimensional continuation , similar to that described above .
that is , we evaluate the integral @xmath113 where @xmath114 is the angle between @xmath115 and @xmath116 , by first regarding @xmath117 so the integral exists .
the integral may be done exactly in terms of gamma functions , which when continued to @xmath71 , @xmath94 yields eq .
( [ nobulk ] ) @xcite
. thus there can hardly be any doubt that the casimir effect , in the tenuous limit , coincides with the van der waals attraction between molecules .
this seems to go some way toward providing understanding of this zero - point fluctuation phenomenon .
but the subject is not closed .
two years ago romeo and i demonstrated that for a dilute cylinder the van der waals energy is _ zero _ @xcite .
presumably the same holds for the casimir energy , in order @xmath118 , but a demonstration of that is not yet at hand .
( it turns out to be quite difficult to compute the casimir effect for a dielectric cylinder . )
remarkably , for a dilute cylinder with constant speed of light inside and out , @xmath119 , it has been demonstrated that the casimir energy vanishes , that is , the energy is of order @xmath120 @xcite , but that would seem to be a completely different case .
_ sonoluminescence _ refers to that remarkable phenomenon in which a small bubble of air injected into a container of water and suspended in a node of a strong acoustic standing wave emits light .
more precisely , if it is driven with a standing wave of about 20,000 hz at an overpressure of about 1 atm , the bubble expands and contracts in concert with the wave , from a maximum radius @xmath121 cm to a minimum radius of @xmath122 cm .
exactly at minimum radius roughly 1 million optical photons are emitted , for a total energy liberated of 10 mev . for a review of the experimental situation as of a few years ago ,
see ref .
@xcite .
julian schwinger , informed of these experiments by putterman , immediately assumed the effect derived from a dynamical version of the casimir effect , and published in the last few years of his life a series of papers in the pnas - usa attempting to account for the observations in this way @xcite .
subsequently a number of other have jumped on this bandwagon @xcite .
the problem is that the `` dynamical casimir effect '' remains largely unknown .
schwinger and his followers had to rely on the known results for the casimir effect with static boundary conditions .
two possible avenues then appeared : * one could employ the adiabatic approximation , which does not seem unreasonable , since the time scale for emission of the photons , @xmath123 s , is far longer than the characteristic optical time scale , @xmath124 s. schwinger @xcite , and carlson et al .
@xcite , then used the divergent bulk energy term @xmath125 if a reasonable cutoff is inserted , the required 10 mev of energy is indeed present .
my objection to this is that it now seems unequivocal that the casimir energy for a ( dilute ) dielectric ball is given by the finite expression ( [ nobulk ] )
. the divergent expression ( [ bulk ] ) is properly absorbed in a renormalization of the material properties .
( [ nobulk ] ) is used , the casimir energy is _ ten orders of magnitude too small to be relevant .
_ * the second avenue , followed by schwinger @xcite and liberati et al .
@xcite , is to use the instantaneous or sudden approximation , in which the static bubble simply disappears .
then the photon production rate can be calculated from the overlap of the two static configurations , through the bogoliubov coefficients .
again , reasonable agreement with observations is reported .
my objection here is that impossibly short time scales are required . for the sudden approximation to make sense , the time scale @xmath126 of collapse must be small compared to @xmath127 s. but in fact , the overall collapse time scale is @xmath128 s , and the emission time scale is probably @xmath129 s. an estimate of what one expects with reasonable numbers may be obtained from the unruh temperature @xcite @xmath130 where @xmath131 is the acceleration of the surface .
estimating @xmath131 by @xmath132 , @xmath3 being some relevant bubble radius , and recognizing that the observations are consistent with a photon temperature of order 20,000 k , we estimate the required time scale by @xmath133 which once again seems physically unrealizable . , the leading approximation in ref .
@xcite , is used .
see also ref .
a great many isolated facts about the casimir effect are now known : * the dimensional dependence for planes and hyperspheres is now known . * the equivalence of the van der waals force and the casimir force for dilute media is definitively established .
* moreover , important applications of the casimir effect to many fields , such as cosmology , have been made .
* we can not predict , _ a priori _ , the sign of the effect .
* there are divergences occurring , for example with the dielectric ball , the nature of which is not well understood . * physically , why does the casimir energy for a sphere in even dimensions diverge ?
* most importantly , can one make progress in understanding the dynamical casimir effect ?
* acknowledgements .
* this work was supported in part by the us department of energy .
i am grateful to the organizers of the fradkin memorial conference for allowing me to make this presentation in honor of the many great works of efim fradkin .
i thank peter van nieuwenhuizen and gerhard soff for helpful discussions at the conference .
b. v. deriagin and i. i. abrikosova , _ zh . eksp .
fiz . _ * 30 * ( 1956 ) 993 ; * 31 * ( 1956 ) 3 [ english transl . : _
soviet phys .
jetp _ * 3 * ( 1957 ) 819 ; * 4 * ( 1957 ) 2 ] ; b. v. deriagin , i. abrikosova , and e. m. lifshitz , _ quart . rev . _
* 10 * ( 1968 ) 295 ; a. kitchener and a. p. prosser , _ proc .
( london ) a _ * 242 * ( 1957 ) 403 ; m. y. sparnaay , _ physica _ * 24 * ( 1958 ) 751 ; w. black , j. g. v. de jongh , j. t. g. overbeck , and m. j. sparnaay , _ trans .
faraday soc . _
* 56 * ( 1960 ) 1597 ; a. van silfhout , _ proc .
wetensch .
b _ * 69 * ( 1966 ) 501 ; d. tabor and r. h. s. winterton , _ nature _ * 219 * ( 1968 ) 1120 ; d. tabor and r. h. s. winterton , _ proc .
( london ) a _ * 312 * ( 1969 ) 435 ; r. h. s. winterton , _ contemp .
* 11 * ( 1970 ) 559 ; j. n. israelachivili and d. tabor , _ proc .
( london ) a _ * 331 * ( 1972 ) 19 .
e. m. lifshitz , _ zh . eksp .
* 29 * ( 1956 ) 94 [ english transl . : _
soviet phys .
jetp _ * 2 * ( 1956 ) 73 ] ; i. d. dzyaloshinskii , e. m. lifshitz , and l. p. pitaevskii , _ usp .
nauk _ * 73 * ( 1961 ) 381 [ english transl . : _
soviet phys .
* 4 * ( 1961 ) 153 ] .
t. h. boyer , _ phys .
_ * 174 * ( 1968 ) 1764 .
boyer s result has been confirmed by many authors , including b. davies , _ j. math .
* 13 * ( 1972 ) 1324 ; r. balian and b. duplantier , _ ann .
_ * 112 * ( 1978 ) 165 ; k. a. milton , l. deraad , jr . , and j. schwinger , _ ann . phys .
_ * 115 * ( 1978 ) 388 .
j. schwinger , _ lett .
* 24 * ( 1992 ) 59 , 227 ; _ proc .
usa _ * 89 * ( 1992 ) 4091 , 11118 .
for a history of schwinger s developments , see k. a. milton in _ the casimir effect 50 years later _ , ed .
m. bordag ( world scientific , singapore , 1999 ) , p. 20 .
i. brevik , v. n. marachevsky , and k. a. milton , _ phys .
* 82 * ( 1999 ) 3948 ; g. barton , _ j. phys . a _ * 32 * ( 1999 ) 525 ; m. bordag , k. kirsten , and d. vassilevich , _ phys
* 59 * ( 1999 ) 085011 ; j. s. hye and i. brevik , _ j. stat
_ * 100 * ( 2000 ) 223 . c. eberlein , _ phys .
* 76 * ( 1996 ) 3842 . c. e. carlson , c. molina - pars , j. perez - mercader , and m. visser , _ phys .
_ * b395 * ( 1997 ) 76 ; _ phys .
d _ * 56 * ( 1997 ) 1262 ; c. molina - pars and m. visser , _ phys .
d _ * 56 * ( 1997 ) 6629 .
s. liberati , m. visser , f. belgiorno , and d. w. sciama , _ phys .
* 83 * ( 1999 ) 678 ; _ phys . rev .
d _ * 61 * ( 2000 ) 085023 ; * 61 * ( 2000 ) 085024 ; s. liberati , f. belgiorno , m. visser , and d. w. sciama , _ j. phys . a _ * 33 * ( 2000 ) 2251 ; f. belgiorno , s. liberati , m. visser , and d. w. sciama , phys
a * 271 * ( 2000 ) 308 .
g. t. moore , _
j. math .
* 11 * ( 1970 ) 2679 ; e. sassaroli , y. n. srivastava , and a. widom , _ phys .
a _ * 50 * ( 1994 ) 1027 ; s. a. fulling and p. c. w. davies , _ proc .
london _ , ser .
a * 348 * ( 1976 ) 393 ; l. h. ford and a. vilenkin , _ phys .
rev.a _ * 25 * ( 1982 ) 2569 ; for example .
p. c. w. davies , _
j. phys . a _ * 8 * ( 1975 ) 365 ; w. g. unruh , _ phys . rev .
d _ * 14 * ( 1976 ) 870 ; n. d. birrell and p. c. w. davies , _ quantum field theory in curved space _ ( cambridge university press , cambridge , 1982 ) ; s. a. fulling , _ aspects of quantum field theory in curved space - time _ ( cambridge university press , cambridge , 1989 ) . | zero - point fluctuations in quantum fields give rise to observable forces between material bodies , the so - called casimir forces . in this lecture
i present some results of the theory of the casimir effect , primarily formulated in terms of green s functions .
there is an intimate relation between the casimir effect and van der waals forces .
applications to conductors and dielectric bodies of various shapes will be given for the cases of scalar , electromagnetic , and fermionic fields .
the dimensional dependence of the effect will be described .
finally , we ask the question : is there a connection between the casimir effect and the phenomenon of sonoluminescence ? |
efficient computing of x - ray ( and neutron ) scattering from crystals has been the subject of intense work since computers became available . except in the case of small structures ( @xmath21000 atoms ) or small number of reflections ( @xmath21000 ) , the method of choice has long been to use the fast - fourier transform @xcite of the crystal s scattering density . by computing this density inside the crystal s unit cell over a suitable grid @xcite , it is possible to compute structure factors at nodes of the reciprocal lattice . in the case of strained @xcite or disordered @xcite crystals , the scattering must take into account a large part of the crystal ( or possibly the entire crystal ) instead of a single unit cell , in order to describe the departure from an infinite , triperiodic structure .
this requires either using approximations , or a large computing power .
moreover , both strain and disorder lead to non - discrete scattering , so that the scattered amplitude must be evaluated on a fine grid around or between bragg diffraction peaks .
this type of computations can greatly benefit from fast calculations , which we will present in this paper .
this article is organized as follows : in section [ secscattering ] we describe the formulas used for computing the scattering from an atomistic model , how it can be efficiently computed using a graphical processing unit ( gpu ) , and what performance can be achieved . in section [ secpynx ] we present the open - source package pynx which can be used to easily compute scattering with little programming knowledge . in section [ secapplication ]
a few examples are given .
x - ray and neutron scattering can generally be calculated , in the kinematic approximation , as : @xmath3\ ] ] where @xmath4 is the scattered amplitude , @xmath5 is the scattering vector , @xmath6 represents the scattering density ( electronic or nuclear ) at position @xmath7 inside the crystal , and @xmath8 denotes the fourier transform .
this equation can be used to determine the scattering from a crystal as long as @xmath6 is described on a grid fine enough to resolve the atomic positions , which is easy if the crystal can be described from a single unit cell . in the case of an aperiodic object ( crystal with an inhomogeneous strain or disordered ) , it is simpler to compute the scattering from an atomistic model , which can be obtained using reverse monte - carlo @xcite , atomic potentials combined with molecular dynamics or a direct minimization of the crystal energy @xcite .
the scattered amplitude is then derived from the atomic positions : @xmath9 where @xmath10 is the scattering length ( either the thomson scattering factor for x - rays or the nuclear scattering length for neutrons ) of atom @xmath11 and @xmath12 its position .
the number of floating - point operations ( @xmath13 ) required to evaluate equation ( [ eqn : eqnscattatoms ] ) is approximately equal ( see section [ secimplemgpu ] ) to : @xmath14 for a structure with @xmath15 ( e.g. a cube of silicon of @xmath16 ) and @xmath17 points in reciprocal space , this corresponds to @xmath18 , which can be compared to the current computing power of today s consumer micro - processors of @xmath19 per core . in the case of large nano ( and micro)-structures , for which the description of the atomic structure is not possible in practice , a model based on continuum elasticity can be used , either with an analytical or numerical approach ( see and references within ) .
the most popular method in the case of epitaxial nanostructures currently is the finite element method- see for a recent discussion .
this model can then be used to calculate the scattering using groups of atoms : @xmath20 where @xmath21 is the structure factor for the @xmath22 block of atoms ( generally a group of unit cells ) , @xmath23 its original position , and @xmath24 its displacement from the ideal structure .
assuming that all blocks of atoms are identical and on a triperiodic grid , it is possible to rewrite eqn .
[ eqn : eqnscattblock ] as : @xmath25\ ] ] where @xmath26 denotes the bragg peak position around which the calculation is made , @xmath27 is the structure factor calculated for a block of atoms , and @xmath28 the displacement field inside the crystal . if the composition of the blocks of atoms vary ( e.g. due to interdiffusion ) , it is also possible to include a variation of the average scattering density in the @xmath8 @xcite : @xmath29\ ] ] where @xmath30 is the relative scattering density in the crystal .
both equation ( [ eqn : eqnscattblockft ] ) and ( [ eqn : eqnrhoftdispl ] ) allow the use of a _ fast _ fourier transform , but are only valid as long as : @xmath31 moreover , use of equations ( [ eqn : eqnscattblockft ] ) and ( [ eqn : eqnrhoftdispl ] ) with a _
fast _ fourier transform restricts the computation of scattering on a triperiodic grid in reciprocal space - this is a limitation since modern data collection often use 2d detectors , and the measured points in reciprocal space are located on a _ curved _ surface ( the projection of the detector on ewald s sphere ) .
furthermore , as the resolution in reciprocal space is inversely proportional to the size in real space , analysis of high - resolution data using a fft calculation demands a large model - even if the extent in reciprocal space is very limited .
therefore , even if the speed of the fft is optimal for large crystalline structures - for @xmath32 points in real space , @xmath32 points in reciprocal space are calculated with a cost proportional to @xmath33 instead of @xmath34 - it is still interesting to consider a _
direct _ computation using equation ( [ eqn : eqnscattatoms ] ) or ( [ eqn : eqnscattblock ] ) because it allows computation for : * _ any _ assembly of points in reciprocal space * from _ any _ structural model ( no matter how severely distorted or disordered ) in order to achieve the calculations in a reasonable time , it is useful to consider current graphics cards as general - purpose gpu . this has already been reported in the scope of crystallography , for computing scattering maps from disordered crystals @xcite , powder pattern computing using the debye equation @xcite , and for single - particle electron microscopy @xcite . to summarize basic principles behind gpu computing ,
it is possible to accelerate any calculation provided that : 1 .
it is * highly parallel * , _ i.e. _ the same formula must be applied on large amounts of data , independently 2 .
the number of * memory * transfers required is much smaller than the number of mathematical operations 3 .
the calculation pathway is determined in advance ( at compilation time ) , which excludes any _ if ... then ... else _ operation in the inner computation loop moreover , many classical functions ( e.g. : @xmath35 , @xmath36 , @xmath37 , fused @xmath38 evaluation , ... ) are highly optimized on gpus - an algorithm requiring many such operations will be greatly accelerated .
equation ( [ eqn : eqnscattatoms ] ) fulfills all requirements , assuming that both the number of atoms and the number of points in reciprocal space are large ( @xmath391000 ) .
[ figspeed ] indicates the number of reflections for the gpu calculations ( black lines ) .
the cpu ( central processing unit ) curves ( red lines ) correspond to a computing using a vectorized ( sse - optimized ) c++ code running on a _ single _ core of an intel core2 quad q9550 running at 2.83 ghz , for @xmath40 ( the curves for @xmath41 and @xmath42 are almost identical).,title="fig:",scaledwidth=95.0% ] the implementation presented in this article uses the cuda @xcite toolkit .
it is beyond the scope of this article to detail the exact algorithm used for computation , as the implementation is freely available as an open - source project ( see section [ secpynx ] ) .
however , it should be noted that the calculations are made in parallel for all reflections , with the atomic coordinates shared between parallel threads ( to minimize memory transfers ) - this method is optimal for large number of atoms . for some configurations ( large number of reflections and small number of atoms ) , it may be more optimal to parallelize on the atoms and share the reflection coordinates between parallel processes .
the achieved speed is shown in fig .
[ figspeed ] , for a calculation of scattering for a random list of points in reciprocal space , and random coordinates for the atoms - the occupancy of all atoms is assumed in this test to be equal to 1 , and the atomic scattering factor is not evaluated- in practice the atomic scattering factors can be factorized and represent a negligible amount of computing ( see section [ sec : secexampleinas ] ) - the same is true for debye - waller factors . as can be seen in fig .
[ figspeed ] , there is a strong dependence of the speed with the number of reflections and the number of atoms per second - the maximum speed ( @xmath43 ) is only reached if both numbers are larger than @xmath42 .
each couple ( reflection , atom ) corresponds to 8 floating - point operations ( 3 multiplications , 4 additions , one @xmath44 evaluation ) operation is hardware - accelerated , it is 4 to 8 times slower than a simple addition . if the @xmath44 evaluation is counted as 4 @xmath13 , the achieved speed is @xmath45500 @xmath46 . ] , so that the overall speed is equal to @xmath45367 @xmath46 .
this can be compared to the peak theoretical speed of 1.7 @xmath47 for this graphics card , which is only achieved when using fused add - multiply operations , without any bottleneck due to memory transfers . by comparison , when computing on the cpu ( see fig .
[ figspeed ] ) , the maximum speed is reached sooner : for 100 atoms and @xmath481000 reflections , or for @xmath481000 atoms when using @xmath48100 reflections .
the top speed for a _ single _ core ( intel core2 quad q9550 running at 2.83 ghz ) , using sse - vectorized sine and cosine functions @xcite , is @xmath49 - @xmath45380 times slower than the gpu version .
un_-optimized ( without using sse code ) c++ code runs about 3 times slower , or @xmath451000 times slower than the gpu version .
using multiple cores , the speed increases linearly ( except for small number of atoms ) with the number of cores .
as was already pointed out by , accuracy is an important issue since gpus are most efficient when using single precision floating - point operations .
moreover , the accuracy of operations can be slightly relaxed @xcite compared to ieee standards @xcite . for example , since single - precision floating - point use 24 bits mantissa ( _ i.e. _ a relative accuracy of @xmath50 ) , precision may be expected to become problematic when the total scattered amplitude varies on more than 7 orders of magnitude . a simple test can be made : computing the scattering for a linear , perfectly periodic chain of identical atoms , and comparing it to the analytical formula : @xmath51 ( where @xmath52 is the reciprocal lattice unit and @xmath32 the number of atoms in the chain ) .
this is shown in fig.[figaccuracy ] , for chains of atoms of different lengths .
the discrepancy between the analytical calculation and the single - precision gpu calculation is clearly visible in the regions where the intensity is minimal - however in practice , the dynamic range where the calculation is reliable is always larger than @xmath53 , and most of the time ( @xmath54 of the points ) around @xmath55 - these numbers refer to the intensity ( squared modulus of the scattered amplitude ) .
[ figaccuracy ] ( b ) @xmath42 and ( c ) @xmath56 atoms , using a perfectly periodic chain calculated using the gpu ( black line ) , the analytical model ( red line ) , and for a chain where the atoms are randomly displaced with a gaussian distribution with a standard deviation of @xmath57 ( gray line ) .
the atoms where located at @xmath58 , and the h coordinates are located every 0.001 in reciprocal lattice units ( r.l.u.).,scaledwidth=95.0% ] such a dynamic range should be sufficient for most applications , as the practical range for experimentally measured intensities is usually lower , except in the case of perfect crystals . in fig.[figaccuracy ] a gray curve is superposed to the simulations , and corresponds to the gpu calculation for a chain of atoms with random displacements with a gaussian standard deviation of @xmath57 of their fractional coordinate . the error due to
the single - precision computing is generally lower than the noise level represented by the gray curve .
we have found that errors due to single precision floating point calculations were not significant _ in practice _ : indeed , most of the time structural models for which this type of computation is used are not ideal ( see examples of simulated calculations using our code in and ) and therefore do not present a very large dynamic range ( larger than 8 orders of magnitude ) .
it should however be noted that gpus can also perform calculations using double precision floating point , but with a lower performance , as the number of available processing units are generally smaller ( 8 times in the case of cuda graphics cards with capability less than 2.0 @xcite ) than for single precision calculations .
more recent graphics cards ( available since mid-2010 ) , using the fermi architecture ( ` http://www.nvidia.com/object/fermi_architecture.html ` ) provide a higher computing power dedicated to double - precision computing ( about half the speed of single - precision ) .
writing programs using gpu computing is a relatively complex process , as it is necessary to fine - tune the algorithm , notably in order to optimize memory transfers - which can make a very significant difference in terms of performance .
for example an early version of the presented algorithm did not perform in a synchronized way between parallel computing threads , and its performance was @xmath59 slower than the final algorithm used .
moreover , all data has to be allocated both in the computer s main memory as well as on the graphics card , which can be tedious to write . for this reason
, we have written an open - source library , pynx `` python tools for nanostructure xtallography '' , using the python language .
the main features of this software package are the following : * computing of scattering for a given list of atomic positions and points in reciprocal space does not require _ any _ gpu - computing knowledge * it is possible to input either a list of @xmath60 coordinates , or also include their occupancy @xmath61 * the shape and order of the @xmath60 and @xmath62 coordinates ( i.e. 1d , 2d or 3d , sorted or not ) is irrelevant - all calculations are made _ in fine _ on 1d vectors * the computation can be distributed on several gpus - e.g. such cards as nvidia s gtx 295 are seen as two independent gpu units - the calculation is distributed transparently over the two gpu * a pure sse - optimized cpu computation is also available when no gpu is available , and can take advantage of all the computing cores available .
three modules are available : * ` pynx.gpu ` , which is the main module allowing fast , parallel computation of @xmath63 , either using a gpu or the cpu * ` pynx.fthomson ` , which gives access to the analytic approximation for the x - ray atomic scattering factors @xcite extracted from the cctbx library @xcite * ` pynx.gid ` , which provides transmission and reflection coefficients at an interface @xcite , which is required for grazing incidence diffraction analysis @xcite using the distorted wave born approximation ( dwba ) , as is demonstrated in section [ sec : dwba ] .
[ figstrainedcube ] .
intensities are color - coded on a logarithmic scale.,scaledwidth=95.0% ] to compute the scattering around the @xmath65 reflection of a simple cubic structure with a lateral size of 100 unit cells , the following code is used : .... # import libraries from numpy import arange , float32,newaxis , log10,abs from pynx import gpu # create array of 3d coordinates , 100x100x100 cells x = arange(-50,50,dtype = float32 ) y = arange(-50,50,dtype = float32)[:,newaxis ] z = arange(0,100,dtype = float32)[:,newaxis , newaxis ] # hkl coordinates as a 2d array h = arange(-.1,.1,0.001 ) k=0 l = arange(3.9,4.1,0.001)[:,newaxis ] # the actual computation fhkl , dt = gpu.fhkl_thread(h , k , l , x , y , z , gpu_name="295 " ) # display using matplotlib from pylab import imshow imshow(log10(abs(fhkl)**2),vmin=0 , extent=(-.1,.1,3.9,4.1 ) ) .... in this example , the calculation takes 0.93s on a gtx 295 graphics card .
the library used for graphics display is matplotlib ( http://matplotlib.sourceforge.net/ ) . of course scattering from this cube
could be calculated analytically - if we introduce a simple displacement field in the z - direction : @xmath66 , the following line can be inserted after the `` ` z = arange ... ` '' instruction : .... z = z+1e-6*z*(x**2+y**2 ) .... the computed diffraction map is shown in fig.[figstrainedcube ] . in the previous example , the atomic scattering factor is not taken into account - since this factor is the same for all atoms of the same type , it is easy to group all atoms of the same type together and calculate first @xmath63 , and then multiply it by the value of the atomic scattering factor dependent on the position in reciprocal space ( and the energy if anomalous scattering terms are to be taken into account ) , as well as the debye - waller factor .
these atomic scattering factors can be extracted from the ` pynx.fthomson ` module .
let us consider an inas nano - structure , for which we have atomic coordinates in separate files ` in.dat ` and `
as.dat ` , each file having 3 columns corresponding to the x , y , z orthonormal coordinates ( in nanometers ) .
the scattering around reflection @xmath65 for this data can be calculated in the following way ( the f and f " resonant terms were taken manually from the cctbx library @xcite ) : .... # import libraries from numpy import arange , newaxis , sqrt , abs , loadtxt from pynx import gpu , fthomson # hkl coordinates as a 2d array h = arange(-.1,.1,0.001 ) k=0 l = arange(3.9,4.1,0.001)[:,newaxis ] # load orthonormal coordinates xas , yas , zas = loadtxt("as.dat",unpack = true ) xin , yin , zin = loadtxt("in.dat",unpack = true ) # convert to fractional coordinates xas/=.6036 yas/=.6036 zas/=.6036 xin/=.6036 yin/=.6036 zin/=.6036 # compute scattering fhklin , dt = gpu.fhkl_thread(h , k , l , xin , yin , zin , gpu_name="295 " ) fhklas , dt = gpu.fhkl_thread(h , k , l , xas , yas , zas , gpu_name="295 " ) # apply scattering factors at 10kev s=6.036/sqrt(h**2+k**2+l**2 ) fas = fthomson.fthomson(s,"as")-1.64 + 0.70j fin = fthomson.fthomson(s,"in")+0.09 + 3.47j # full structure factor fhkl = fhklas*fas + fhklin*fin .... a specific module ( ` pynx.gid ` ) is available for grazing incidence diffraction - this module allows to compute the complex refraction index of a crystalline material ( the substrate ) and determine the reflection and transmission coefficients at the interface @xcite .
it is therefore possible to simulate grazing incidence x - ray scattering using the dwba approximation , by taking into account the reflections before and after diffraction by the object at the surface , which influence the shape of the scattering in reciprocal space @xcite . in fig.[figdwba ]
is shown the simulated scattering for a germanium quantum dot on a silicon substrate .
for the sake of demonstrating the use of pynx , a simple analytical model is used : * the quantum dot shape corresponds to a portion of a sphere , with a radius equal to 50 unit cells and a height of 20 unit cells . *
the germanium content varies linearly from @xmath67 ( bottom ) to @xmath68 ( top ) * the relaxation ( x , y , z being the fractional coordinates relative to the silicon substrate lattice ) are : + @xmath69 moreover , in this example only the scattering from the quantum dot is calculated , ignoring any contribution from the substrate .
the corresponding script is provided as a supplementary file . a more complete description of the scattering , taking into account scattering from the ( strained ) substrate @xcite ,
could also be added in the future .
[ figdwba ] reflection , and is plotted against @xmath52 ( reciprocal lattice unit ) and the outgoing angle @xmath70 .
the location ( outgoing angle ) of the intensity maximum as a function of @xmath52 varies as the in - plane strain changes with the height of the corresponding layer in the dot , due to interferences between the four scattered beams @xcite .
see text for details.,scaledwidth=95.0% ]
the pynx library is freely available from the project website at : http://pynx.sourceforge.net .
it is open - source software , distributed under the cecill - b license ( http://www.cecill.info/index.en.html ) , a permissive license similar to the freebsd license . although this library has been developed and tested only under linux , it should work on any operating system ( including macos x and windows ) supported by the pycuda @xcite library .
it has been tested on a variety of graphics card ( 9600 gt , gtx 295 , 8800 gtx ) .
although it is recommended to use a dedicated graphics card ( not used for display ) for gpu computing , it is not a requirement - the library automatically cuts the number of atoms in order to decompose the calculation in batches which last less than 5s ( a limit imposed by the cuda library for graphics cards attached to a display ) . and
it is also possible to use the cpu for calculations .
this library uses the scipy ( http://www.scipy.org ) and pycuda @xcite libraries , and optionally the cctbx @xcite to determine the refraction index for the computing of transmission and reflection coefficients for grazing incidence x - ray scattering .
the main interest from this computing project is the ability to compute scattering for non - ideal structures without any approximation .
this is particularly important for strained nano - structures where the calculation often uses a fast fourier transform approximation , even though the displacements from the ideal structure are large .
this could also be useful for coherent diffraction imaging in bragg condition for strained nano - structures @xcite , especially in order to extend this method to severely distorted lattices ( e.g. near an epitaxial interface with dislocations , a grain boundary , ... ) .
a current limitation of this project is related to the toolkit used - the cuda development package is the most popular gpu computing tool available at the moment , but it depends on a single manufacturer , and remains proprietary . an important development in that regard is the creation of the opencl language ( http://www.khronos.org/opencl/ ) , which is intended to allow gpu - computing _ independently of the graphics card_. a future implementation of the proposed algorithm could use opencl and ensure its usability on a larger range of computing equipment . | scattering maps from strained or disordered nano - structures around a bragg reflection can either be computed quickly using approximations and a ( fast ) fourier transform , or using individual atomic positions . in this article
we show that it is possible to compute up to @xmath0 @xmath1 using a single graphics card , and we evaluate how this speed depends on number of atoms and points in reciprocal space .
an open - source software library ( pynx ) allowing easy scattering computations ( including grazing incidence conditions ) in the python language is described , with examples of scattering from non - ideal nanostructures .
coraux richard renevier scattering calculations from atomistic models using graphical processing units are presented , and compared to the speed achieved using normal cpu calculations . an open - source software toolbox ( pynx ) is presented , with a few examples showing the fast calculation of scattering maps from strained nanostructures , including grazing - incidence conditions . |
most elliptical galaxies are poor with respect to interstellar gas . also , elliptical galaxies invariably contain central massive black holes ( bhs ) , and there exists a tight relationship between the characteristic stellar velocity dispersion @xmath1 and the bh mass @xmath0 @xcite , and between @xmath0 and the host spheroid mass in stars , @xmath3 @xcite .
are these two facts related ? here we focus on a scenario in which the mass of the central bh grows within gas rich elliptical progenitors until star formation has reduced the gas fraction in the central regions to of order 1% of the stellar mass . then radiative feedback during episodes when the bh luminosity approaches its eddington limit drives much of the gas out of the galaxy , limiting both future growth of the bh and future star formation to low levels .
many works already recognized the importance of feedback as a key ingredient of the mutual bh and galaxy evolution @xcite .
what is new about this work is the stress on one component of the problem that has had relatively little attention : the radiative output of the central bh is not conjectural
it must have occured and the high energy component of that radiative output will have a dramatic and calculable effect in heating the gas in ellipticals . using the average quasar spectral output derived in @xcite , we show below and in more detail in @xcite that the limit on the central bh produced by the above argument coincides accurately with the observed @xmath0@xmath1 relation . not only the slope , but also the small normalization factor is nicely reproduced .
the present work is complementary to @xcite in that , while it does not attempt to model the complex flaring behavior of an accreting bh with an efficient hydrodynamical code , it does do a far more precise job of specifying the input spectrum and the detailed atomic physics required to understand the interaction between that spectrum and the ambient interstellar gas in elliptical galaxies .
below we assess the conditions required for the central bh radiation to significantly heat the ism over a substantial volume of the galaxy . in this section
we shall _ assume _ that the central bh has a mass as given by the observed @xmath0@xmath1 relation for local ellipticals and bulges @xcite : = 1.510 ^ 8()^4 .
[ sigma_mbh ] this assumption will be dropped in [ origin ] , where we _ predict _ the @xmath0@xmath1 relation . in @xcite
we computed for the average quasar spectrum the equilibrium temperature @xmath4 ( at which heating due to compton scattering and photoionization balances cooling due to line and continuum emission ) of gas of cosmic chemical composition as a function of the ionization parameter @xmath5 , where @xmath6 is the bh bolometric luminosity . in the range @xmath7@xmath8k , ( )
210 ^ 2k , [ tstat_eq ] while at @xmath9 and @xmath10 , @xmath11k and @xmath12k , respectively . on the other hand ,
the galactic virial temperature is given by 3.010 ^ 6k ( ) ^2 .
[ tvir ] we can then find the critical density @xmath13 defined by ( l / r^2)=[ncrit ] as a function of distance @xmath14 from the bh .
gas with @xmath15 will be heated above @xmath16 and expelled from the galaxy .
we show in fig .
[ nr_tvir ] the resulting @xmath17 diagrams for a small and large bh / galaxy .
plane for a galaxy with @xmath18kms@xmath19 ( @xmath20k , @xmath21 , upper panel ) , and with @xmath22kms@xmath19 ( @xmath23k , @xmath24 , lower panel ) . in the dashed area , gas can be heated above @xmath16 by radiation from the central bh emitting at the eddington luminosity .
the upper boundary of this area scales linearly with luminosity .
vertical boundaries are @xmath25 , @xmath26 , @xmath27 and @xmath28 . ] in reality , provided that @xmath29 , significant heating will take place only out to a certain distance that depends on the luminosity and duration of the quasar outburst .
since the bh releases via accretion a finite total amount of energy , @xmath30 , there is a characteristic limiting distance : = ( ) ^1/2 = 400pc()^1/2 ( ) ^1/2 .
[ rc ] inside this radius , a low density , fully photoionized gas will be heated to the compton temperature @xmath31kev characteristic of the quasar spectral output .
more relevant for the problem at hand is the distance out to which low density gas will be compton heated to @xmath32 : r_1=()^1/2= 1,300pc()^1/2 .
[ r1 ] yet another characteristic radius is out to which gas of critical density @xmath13 will be heated to @xmath32 via photoinization and compton scattering : r_2=r_1 [ ( n_crit)/_c]^1/2 , [ r2 ] where @xmath33 and @xmath34 are the compton and total heating rates , respectively . depending on gas density ( @xmath35 ) , the outer boundary of the `` blowout region '' will be located somewhere between @xmath26 and @xmath27 .
the size of the heating zone can be compared with the galaxy effective radius ~4,000pc ( ) ^2 .
[ reff ] the different characteristic distances defined above are shown as a function of @xmath0 in fig .
[ mbh_radii ] .
one can see that a bh of mass @xmath36 should be able to unbind the ism out to several @xmath28 . in the case of more massive bhs / galaxies with @xmath37@xmath38
, the heating will be localized to innermost @xmath20.30.5@xmath28 .
( dotted line ) , @xmath26 ( short - dashed line ) , and @xmath27 ( long - dashed line ) , and the galactic effective radius ( solid line ) , as a function of @xmath0 . ]
we now consider the following general idea .
before the bh grows to a certain critical mass , @xmath39 , its radiation will be unable to efficiently heat the ambient gas , and the bh will accrete gas efficiently .
once the bh has grown to @xmath39 , its radiation will heat and expell a substantial amount of gas from the central regions .
feeding of the bh will then become self - regulated on a time scale of order the cooling time of the low density gas .
subsequent central activity will be characterized by a very small duty cycle ( @xmath20.001 ) , as predicted by hydrodynamical simulations @xcite and suggested by observations @xcite .
bh growth will be essentially terminated .
suppose that the galaxy density distribution is that of a singular isothermal sphere , with the gas density following the total density : _
gas(r)=. [ rho_gas ] here @xmath40 and @xmath41 are the gas mass and total mass within the region affected by radiative heating .
the size of the latter is uncertain but is less than a few kpc ( see [ heatfar ] ) , so that @xmath41 is dominated by stars rather than by dark matter .
radiation from the central bh can heat the ambient gas up to 6.510 ^ 3k ( ) ^-1 ( ) ^2 , [ tstat ] this approximate relation being valid in the range @xmath42@xmath8k .
remarkably , @xmath4 does not depend on distance for the adopted @xmath43 density distribution .
we then associate the transition from rapid bh growth to slow , feedback - limited bh growth with the critical condition = , [ tstat_tvir ] where @xmath44 and @xmath16 is given by ( [ tvir ] ) .
once heated to @xmath45 , the gas will stop feeding the bh .
the condition ( [ tstat_tvir ] ) will be met for = 4.610 ^ 10()^4 .
[ mcrit ] therefore , for fixed values of @xmath46 , @xmath47 and @xmath48 we expect @xmath49 , similarly to the observed @xmath0@xmath1 relation .
equally important is the normalization of the @xmath0@xmath1 relation . by comparing ( [ mcrit ] ) with ( [ sigma_mbh ] )
we find that the observed correlation will be established if /m=310 ^ -3 ^ -1l/. [ mgas_crit ] the gas - to - stars ratio is thus required to be low and approximately constant for spheroids of different mass at a certain stage of their evolution . as for the eddington ratio , it is reasonable to expect @xmath500.11 during quasar outbursts .
@xmath1 correlation resulting from heating of the ism by the radiation from the central bh assuming @xmath51 and @xmath52 .
thin solid line corresponds to @xmath53 and @xmath54 .
dashed line is the observed @xmath55 correlation in the range @xmath56a few @xmath38 , extrapolated to lower and higher bh masses .
dotted lines are @xmath57 and @xmath58 laws . ] the approximately linear @xmath59 dependence [ see ( [ tstat_eq ] ) ] was crucial to the above argument leading to the @xmath49 result . however , the @xmath59 function becomes nonlinear outside the range @xmath60 @xcite . in fig .
[ s_mbh ] we show the predicted correlation between @xmath39 and @xmath1 for @xmath61 and @xmath62 . it can be seen that the @xmath55 behavior is expected to break down for @xmath63 and for @xmath64 a few @xmath38 .
it is perhaps interesting that the range of masses shown in fig .
[ s_mbh ] for which @xmath55 is obtained from considerations of atomic phyisics ( and the observed agn spectra ) corresponds closely with the range of masses for which this power law provides a good fit to the observations . exploring the @xmath0@xmath1 relation observationally near @xmath38 would be a sensitive test of the importance of radiative feedback .
in @xcite we addressed in a more quantitative way the bh growth in the context of the parent galaxy evolution .
we adopted a physically - motivated one - zone model , taking into account the mass and energy return from the evolving stellar population .
this model predicts that after an initial `` cold '' phase dominated by gas infall , once the gas density becomes sufficiently low the gas heating dominates and the galaxy switches to a `` hot '' solution .
the gas mass / stellar mass ratio at that epoch ( @xmath20.003 ) is remarkably close to the value inferred above from the argument leading to the right @xmath0@xmath1 relation .
other predictions of the toy model are also in satisfactory agreement with observations .
the `` cold '' phase would probably be identified observationally with the lyman break and scuba galaxies , while the `` hot '' phase with normal , local ellipticals . | we discuss the role of feedback via photoionization and compton heating in the co - evolution of massive black holes at the center of spheroidal galaxies and their stellar and gaseous components .
we first assess the energetics of the radiative feedback from a typical quasar on the ambient interstellar gas .
we then demonstrate that the observed @xmath0@xmath1 relation could be established at a relatively early epoch in galactic evolution when the formation of the stellar bulge was almost completed and the gas - to - stars mass ratio was reduced to a low level @xmath20.01 such that cooling could not keep up with radiative heating .
a considerable amount of gas was expelled at that time and black hole accretion proceeded at a much lower rate thereafter . |
( 350,130)(0,10 ) ( 330,135)kek - preprint-94 - 162 ( 330,120)nwu - hep 94 - 07 ( 330,105)dpnu-94 - 59 ( 330,90)tit - hpe-94 - 013 ( 330,75)tuat - hep 94 - 07 ( 330,60)ocu - hep 94 - 07 ( 330,45)pu-94 - 692 ( 330,30)ins - rep 1077 ( 330,15)kobe - hep 94 - 06 ( -10,30)=3.5 cm r.itoh@xmath6 , m.yamauchi@xmath6 , a.yamaguchi@xmath7 k.abe@xmath8 , t.abe@xmath8 , i.adachi@xmath6 , k.adachi@xmath7 , m.aoki@xmath8 , m.aoki@xmath9 , s.awa@xmath7 , k.emi@xmath10 , r.enomoto@xmath6 , h.fujii@xmath6 , k.fujii@xmath6,t.fujii@xmath11 , j.fujimoto@xmath6 , k.fujita@xmath12 , n.fujiwara@xmath7 , h.hayashii@xmath7 , b.howell@xmath13 , n.iida@xmath7 , y.inoue@xmath12 , h.iwasaki@xmath6 , m.iwasaki@xmath7 , k.kaneyuki@xmath9 , r.kajikawa@xmath8 , s.kato@xmath14 , s.kawabata@xmath6 , h.kichimi@xmath6 , m.kobayashi@xmath6 , d.koltick@xmath13 , i.levine@xmath13 , s.minami@xmath9 , k.miyabayashi@xmath8 , a.miyamoto@xmath6 , k.muramatsu@xmath7 , k.nagai@xmath15 , k.nakabayashi@xmath8 , e.nakano@xmath8 , o.nitoh@xmath10 , s.noguchi@xmath7 , a.ochi@xmath9 , f.ochiai@xmath16 , n.ohishi@xmath8 , y.ohnishi@xmath8 , y.ohshima@xmath9 , h.okuno@xmath14 , t.okusawa@xmath12,t.shinohara@xmath10 , a.sugiyama@xmath8 , s.suzuki@xmath8 , s.suzuki@xmath9 , k.takahashi@xmath10 , t.takahashi@xmath12 , t.tanimori@xmath9 , t.tauchi@xmath6 , y.teramoto@xmath12 , n.toomi@xmath7 , t.tsukamoto@xmath6 , o.tsumura@xmath10 , s.uno@xmath6 , t.watanabe@xmath9 , y.watanabe@xmath9 and a.yamamoto@xmath6 + the lla ( leading log approximation ) parton shower model well reproduces the various distributions of observables for hadronic final states of @xmath4 collisions , when the `` coherence '' effect of soft gluons is taken into account .
several monte carlo programs were written based on this scheme ( jetset63@xcite , for example ) and were used in various experiments to study qcd . however , since the coherence effect is the consequence of higher order corrections , the effect could only be inserted by hand in the lla scheme in these monte carlo programs . on the other hand , even in low @xmath17 region , the momentum spectrum of gluons in the parton shower process
can be analytically calculated using the modified leading log approximation(mlla)@xcite .
the coherence effect is taken into account in mlla by consistently importing a part of next - to - leading order corrections .
the distribution of the particle spectra is expressed as a function of two parameters , @xmath18=log(@xmath19 ) and @xmath20=log(@xmath21 ) : here @xmath23 is the momentum of a particle normalized by the beam energy @xmath24 , @xmath25 is the opening angle of the jet cone , and @xmath26=@xmath27 , @xmath28 , @xmath29=@xmath30 and @xmath31 for five quark flavors , respectively .
the two functions , @xmath32 and @xmath33 , are two solutions of the confluent hyper - geometric equation .
the quantities @xmath34 and @xmath35 are the qcd scale parameter and the energy cut - off of the parton evolution , respectively .
this calculation predicts depletion of soft partons as a consequence of the destructive interference of soft gluons .
this depletion shows up clearly in eq .
[ eq : mlla ] when written as a function of @xmath36 .
this function has a maximum at a certain @xmath37 value and decreases in the larger @xmath37 region .
the coherence effect reduces the available phase space in this region and therefore the effect can be studied by comparing the measured inclusive cross section with this calculation .
however , since this expression is calculated for partons , it can not be compared directly with the measurement unless the distribution at the level of final state hadrons is ensured to be similar to that of partons .
the concept of local parton hadron duality ( lphd ) dictates that the distribution of the final state hadrons is closely related to that of partons@xcite .
the conversion of partons into hadrons , which occurs at low virtuality scale , includes only small momentum transfer , and hence leaves the distribution essentially unchanged . in this article
we compare the calculated parton spectrum ( eq .
[ eq : mlla ] ) directly with the measured hadron spectrum assuming lphd .
the @xmath35 , which was primarily introduced to regularize collinear singularity , gives a cut - off on parton energies .
therefore , the value of @xmath35 should be close to the mass of the particle being considered@xcite , while @xmath34 should be common to all the particle species .
these dependences can be experimentally tested by measuring the values of @xmath35 and @xmath34 for various particle species . the momentum spectrum might still be distorted by fragmentations and decays in spite of lphd .
this complication can be avoided partially when the energy evolution of the distribution is considered .
the peak position of the distribution ( eq . [ eq : mlla ] ) is given using the limiting spectra calculation in which @xmath34 is assumed to be equal to @xmath35@xcite : @xmath38 where @xmath39 is the number of colors ( @xmath40 ) .
this should be compared with the variation as @xmath41 expected from phase space consideration alone .
the data used in this analysis are accumulated by the topaz detector at the tristan @xmath4 collider at center - of - mass energies between 52.0 and 61.4 gev .
the average energy is 58.0 gev , and the total integrated luminosity is 113.7/pb .
details of the topaz detector is described elsewhere@xcite . in this analysis ,
the data from a time projection chamber ( tpc ) is mainly used for tracking and de / dx measurement for charged particles@xcite . the trigger condition relevant to
this analysis is the track trigger which requires two or more charged tracks in the fiducial volume of the tpc , and the energy trigger which requires 4 gev or more energy deposit in the barrel lead glass calorimeter .
the event trigger is generated by a logical or of those two , and the trigger efficiency for multihadronic events is practically 100% . out of the triggered events ,
multihadronic events are selected by the following conditions ; ( a ) five or more charged tracks having transverse momenta with respect to the beam axis larger than 0.15 gev/@xmath42 are originating from the interaction point with the angle larger than 37@xmath43 with respect to the beam axis , ( b ) total visible energy is larger than 1/2 of the center - of - mass energy , and ( c ) momentum imbalance along the beam direction is smaller than 0.4 . by these conditions , the event selection efficiency is estimated to be 67.1% with a background contamination of less than 2.0% .
two - photon processes and @xmath44 productions are the main sources of the background .
in addition to these cuts , the jet axis is required to have an angle larger than 40@xmath43 with respect to the beam axis to ensure that the event is well contained in the detector acceptance .
applying these criteria , 11247 events remain to be used in the analysis .
the particle species @xmath0 , @xmath1 and @xmath3 are identified by measuring the de / dx of tracks detected in tpc .
the detail of the particle identification technique is described in ref.@xcite .
the typical resolution of the de / dx measurement for minimum ionizing pions is 4.6% . fig .
[ fig : dedx ] shows de / dx distribution as a function of the track momentum for a part of the event sample .
the cross section is calculated from the number of tracks in each momentum slice . in the low momentum region ,
the number of tracks for each particle species are directly obtained by counting the number of tracks in each de / dx band . the de
/ dx bands are , however , not well separated in the higher momentum region . to extract the number of each particle species in this region ,
the de / dx distribution in a momentum slice is fitted with a superposition of four gaussians corresponding to @xmath45 , @xmath46 s , @xmath47 s and @xmath3 in each momentum slice .
the widths and the centers of the gaussians are determined from the measurement for bhabha events and cosmic ray @xmath48 s . only the normalizations of the gaussians are the free parameters of the fit .
a typical fit to the de / dx distribution is shown in fig .
[ fig : dedxfit ] .
because of a hardware calibration problem of the tpc , the de / dx resolution is time dependent .
therefore the data sample is divided into two groups and the procedure to count the number of tracks of each particle species is done separately for these two groups .
@xmath49 s are identified by searching for their daughter charged pion pairs in tpc .
a pair of tracks must satisfy following conditions to be identified as a @xmath49 : a ) the distance of the closest approach of the two tracks is less than 0.8 cm ; b ) the distance from the interaction point to the decay vertex is longer than 2.0 cm where the decay vertex is defined as the center of the closest approach of two tracks ; c ) the angle formed by the vector from the interaction point to the decay vertex and the momentum sum of the two tracks at the decay vertex is less than 8 degrees if the position of decay vertex is longer than 6.0 cm ; or c ) the closest distance from the vector sum to the interaction point is less than 0.6 cm if the decay length is shorter than 6.0 cm ; d ) the closest approach of one of the tracks to the interaction point is more than 0.5 cm if the momentum sum of the tracks is less than 2gev / c ; e ) tracks are identified as @xmath46 by the de / dx measurement in the tpc ; and f ) the tracks are not from the gamma conversion . for
each of track pairs remained after these selections , the invariant mass is calculated assuming the tracks are charged pions .
[ fig : ks - mass ] shows the mass distribution for the reconstructed @xmath49 s .
the distribution is fitted by a sum of a gaussian and a background function ( exponential + polynomial ) and the number of reconstructed @xmath49 s is obtained to be 771 @xmath50 61 . in the same way , the numbers of these @xmath49 s are counted for each momentum slice . the inclusive cross section , @xmath51 is then calculated from the number of particles observed in each momentum slice using following formula : @xmath52 where @xmath53 is the ith slice of the momentum fraction , @xmath54 is the total hadronic cross section , @xmath55 is the width of the slice , @xmath56 is the number of hadronic events in the event sample , and @xmath57 is the number of particles counted in the slice , respectively .
@xmath58 is the factor to correct the counted number for the effects of detector acceptance and initial state radiation .
@xmath58 is calculated for each momentum slice separately with the monte carlo simulation as @xmath59 where @xmath60 is the generated number of particles in the slice @xmath53 without including the effects of the detector acceptance and the initial state radiation .
@xmath61 is the corresponding total number of generated events .
@xmath62 is the number of detected particles in the slice @xmath53 generated with the effects of detector acceptance and initial state radiation , and @xmath63 is the corresponding total number of generated events .
the jetset6.3@xcite and 7.3@xcite monte carlo programs are used to obtain these numbers combined with the topaz detector simulation program . as for @xmath49 , to convert the counted number to the cross section of @xmath2 , @xmath58 is multiplied by 2 .
the cross sections are measured as a function of @xmath37 .
the measured cross section for all charged particles is shown in table [ table : xsec0 ] .
table [ table : xsec1 ] shows the cross sections measured for each of @xmath0 , @xmath1 , @xmath3 , while that for @xmath2 is shown in table [ table : xsec2 ] . from eq .
[ eq : peak ] , the mlla prediction of peak position can be approximated in the linear form @xmath64 where @xmath65 to be 0.5 .
if the gluon coherence effect is not taken into account ( phase space only ) , @xmath65 should become 1.0 . to obtain the value of @xmath65 from the measured peak positions , a linear fit is performed to the peak positions as a function of @xmath5 for all charged particles measured by tasso@xcite , topaz , and aleph as shown in fig .
[ energy_evolution ] . from the fit
, the value of @xmath65 is obtained to be @xmath66 .
a similar fit is also done to the peak positions for charged pions ( tasso , pep4/tpc , topaz and aleph ) and @xmath65 is measured to be @xmath67 .
these values are consistent with the mlla predictions .
the inclusive cross sections are measured as a function of @xmath68 for all charged particles and for each of @xmath0 , @xmath1 , @xmath3 and @xmath2 in the hadronic events taken at @xmath5=58gev .
the cross sections were compared with the mlla calculation assuming lphd .
the mlla formula describes the observed distributions very well for all particle species over the wide momentum range . by this comparison ,
@xmath35 and @xmath34 in the mlla expression are determined for each particle species .
the determined @xmath35 for each of the particle species is close to the mass of the particle , while the @xmath34 is almost constant for these particles .
the @xmath35 values are also determined with @xmath34 fixed at 200mev .
the obtained @xmath35 also coincides with the mass of each particle . the same analysis is carried out for the measurements at @xmath69 by pep4/tpc and similar tendency is observed .
these results show that the momentum distribution of particles is identical to that of partons at the end - point of the parton shower evolution independently of the energy scale .
this supports the mlla + lphd conjecture .
the energy evolution of the peak position in @xmath70 is studied by comparing our data with the measurements by pep4/tpc and aleph .
the measured peak positions of light particles are well reproduced by the limiting spectra calculation .
the measured peak positions are fitted to a linear function of log of the beam energy with the measurements at other energies .
the obtained slope of the linear function is consistent with the mlla prediction , while it excludes models without the gluon coherence effect .
a. basseto et al . , phys .
* 100 * ( 1983 ) 201 ; yu . l. dokshitzer et al . , rev . of mod .
, * 60 * ( 1988 ) 373 ; yu . l. dokshitzer et al .
, in perturbative qcd , ed
. a. h. mueller ( world scientific , singapore , 1989 ) 241 ; v. a. khoze , nato asi : z0 physics 1990 ( 1990 ) 419 ; yu . l. dokshitzer , v. a. khoze and s. i. troyan , lu tp 91 - 12 ( 1991 ) .
r. enomoto et al . , nucl .
methods * a269 * ( 1988 ) 507 ; t.tsukamoto , m.yamauchi and r.enomoto , nucl . instrum .
methods * a297 * ( 1990 ) 148 ; a. imanishi et al . , nucl .
methods * a269 * ( 1988 ) 513 ; t. kamae et al .
methods * a252 * ( 1986 ) 423 ; a. yamamoto et al .
* 25 * ( 1986 ) l440 ; s. kawabata et al .
methods * a270 * ( 1988 ) 11 ; j. fujimoto et al .
methods * a256 * ( 1987 ) 449 ; s. noguchi et al .
methods * a271 * ( 1988 ) 404 ; k. fujii et al .
methods * a264 * ( 1988 ) 297 .
hemingway , `` spectra of identified particles '' , presented at xxvii international conference on high energy physics , glasgow ( 1994 ) .
pta collaboration was formed to study the energy dependence of qcd .
this is the joint effort of pep4/tpc , topaz and aleph collaborations . | inclusive momentum spectra are measured for all charged particles and for each of @xmath0 , @xmath1 , @xmath2 , and @xmath3 in hadronic events produced via @xmath4 annihilation at @xmath5=58gev .
the measured spectra are compared with qcd predictions based on the modified leading log approximation(mlla ) .
the mlla model reproduces the measured spectra well .
the energy dependence of the peak positions of the spectra is studied by comparing the measurements with those at other energies .
the energy dependence is also well described by the mlla model .
-0.2 in 8.5 in 0.0 in 5.75 in .3 in |
the infrared dynamics of qcd can be viewed only dimly via presently available tools . because of that , there may be surprises , as it has already happened with our understanding of nonperturbative dynamics in @xmath1=1 supersymmetric qcd @xcite . in particular
, those studies showed that the conventional wisdom which has accepted that the asymptotic freedom implies confinement is not always true .
recently , there has been considerable interest in the existence of a nontrivial conformal dynamics in 3 + 1 dimensional non - supersymmetric vector like gauge theories , with a relatively large number of fermion flavors @xmath0 @xcite .
the roots of this problem go back to a work of banks and zaks @xcite who were first to discuss the consequences of the existence of an infrared - stable fixed point @xmath2 for @xmath3 in vector - like gauge theories .
the value @xmath4 depends on the gauge group : in the case of su(3 ) gauge group , @xmath5 in the two - loop approximation . a new insight in this problem @xcite has been , on the one hand , connected with using the results of the analysis of the schwinger - dyson ( sd ) equations describing chiral symmetry
breaking in qcd ( for a review , see refs.@xcite ) and , on the other hand , with the discovery of the conformal window in @xmath6 supersymmetric qcd @xcite .
in particular , appelquist , terning , and wijewardhana @xcite suggested that , in the case of the gauge group su(@xmath7 ) , the critical value @xmath8 separates a phase with no confinement and chiral symmetry breaking ( @xmath9 ) and a phase with confinement and with chiral symmetry breaking ( @xmath10 ) .
the basic point for this suggestion was the observation that at @xmath9 the value of the infrared fixed point @xmath11 is smaller than a critical value @xmath12 , presumably needed to generate the chiral condensate @xcite .
the authors of ref.@xcite considered only the case when the running coupling constant @xmath13 is less than the fixed point @xmath11 . in this case
the dynamics is asymptotically free ( at short distances ) both at @xmath10 and @xmath14 .
yamawaki and the author @xcite analyzed the dynamics in the whole ( @xmath15 ) plane and suggested the ( @xmath15)-phase diagram of the su(@xmath7 ) theory ( see fig . 1 below ) .. in particular , it was pointed out that one can get an interesting non - asymptotically free dynamics when the bare coupling constant @xmath16 is _ larger _ than @xmath11 , though not very large .
the dynamics with @xmath17 admits a continuum limit and is interesting in itself .
also , its better understanding can be important for establishing the conformal window in lattice computer simulations of the su(@xmath7 ) theory with such large values of @xmath0 . in order to illustrate this ,
let us consider the following example .
for @xmath18 and @xmath19 , the value of the infrared fixed point @xmath20 is small : @xmath210.04 ( see below ) . to reach the asymptotically free phase
, one needs to take the bare coupling @xmath16 less than this value of @xmath20 .
however , because of large finite size effects , the lattice computer simulations of the su(3 ) theory with such a small @xmath16 would be unreliable .
therefore , in this case , it is necessary to consider the dynamics with @xmath22 .
the existence of the phase transition with respect to @xmath0 in qcd raises the following question : what are the infrared properties of the qcd @xmath23 function for different @xmath0 and how the @xmath23 function structure reflects the existence of this phase transition ?
i will discuss those issues at the end of my talk but first i will discuss the dynamics in the conformal window of qcd in detail . in particular , i will consider the spectrum of low energy excitations in that dynamics in the presence of a bare fermion mass @xcite .
we will see that in this case , unlike the familiar qcd with a small @xmath0 ( @xmath0=2 or 3 ) , glueballs are much lighter than bound states composed of fermions , if the value of the infrared fixed point is not too large .
another characteristic point is a strong ( and simple ) dependence of the masses of all the colorless bound states on the bare fermion mass , even if the latter is tiny .
this talk is based on papers @xcite .
i begin by recalling the basic facts concerning the two - loop @xmath23 function in an su(@xmath7 ) theory .
the @xmath23 function is @xmath24 with @xcite @xmath25 while these two coefficients are invariant under change of a renormalization scheme , the higher - order coefficients are scheme dependent . actually , there is a renormalization scheme in which the two - loop @xmath23 function is ( perturbatively ) exact @xcite .
we will use such a renormalization scheme .
if @xmath26 ( @xmath27 ) and @xmath28 , the @xmath23 function has a zero , corresponding to a infrared - stable fixed point , at @xmath29 when @xmath0 is close to @xmath30 , the value of @xmath20 is small .
for example , from eqs.([b ] ) , ( [ c ] ) , and ( [ alpha^ * ] ) , one gets @xmath31 0.04 , 0.14 , 0.28 , and 0.47 for @xmath7=3 and @xmath0=16 , 15 , 14 , and 13 , respectively .
the value of @xmath20 becomes equal to @xmath32 at @xmath0 close to @xmath33 .
and the fixed point disappears at the value @xmath34 , when the coefficient @xmath35 becomes positive ( @xmath36 is @xmath37 for @xmath7=3 ) .
the @xmath23 function ( [ beta ] ) leads to the following solution for the running coupling : @xmath38 we emphasize that this solution is valid both for @xmath39 and @xmath22 .
let us first consider the case with @xmath39 .
it is convenient to introduce the parameter [ 1 ] @xmath40 .
\label{lambda}\ ] ] then , eqs .
( [ solution ] ) and ( [ lambda ] ) imply that @xmath41 taking @xmath42 , we find that @xmath43 one may think that @xmath44 plays here the same role as @xmath45 in the confinement phase .
however , as we will see , its physical meaning is somewhat different .
( [ solution1 ] ) implies that @xmath46 for @xmath47 ( the usual behavior in asymptotically free theories ) , and @xmath48 for @xmath49 , governed by the infrared fixed point @xmath20 .
let us turn to a less familiar case with @xmath22 .
one still can use eq.([solution ] ) .
introduce now the parameter @xmath50 as @xmath51 \label{tilde}\ ] ] ( compare with eq.([lambda ] ) ) .
then , eqs.([solution ] ) and ( [ tilde ] ) imply @xmath52 what is the meaning of @xmath50 ? it is a landau pole at which @xmath53 . indeed , taking @xmath54 in eq.([solution2 ] ) , one gets @xmath55 the only solution of this equation is @xmath56 .
the presence of the landau pole implies that the dynamics is not asymptotically free . to get a more insight in this dynamics ,
let us introduce an ultraviolet cutoff @xmath57 with the bare coupling constant @xmath58 .
now all momenta @xmath59 satisfy @xmath60 .
eq.([solution2 ] ) implies that at finite @xmath61 , the cutoff @xmath57 is less than @xmath50 , with @xmath56 .
therefore the landau pole is unreachable in the theory with cut off @xmath57 and with @xmath62 .
still one can of course use @xmath50 ( [ tilde ] ) for a convenient parametrization of the running coupling @xmath63 ( see eq.([solution2 ] ) ) .
however , one should remember that momenta @xmath59 satisfy @xmath64 .
eq.([solution2 ] ) implies that @xmath65 for @xmath66 , and @xmath67 when @xmath63 is close to @xmath20 , i.e. when @xmath68 .
thus , now @xmath63 approaches the fixed point @xmath20 from above ( compare with eq.([ir ] ) ) . and
, in general , eq.(11 ) implies that @xmath63 monotonically decrease with @xmath59 , from @xmath69 at @xmath70 to @xmath71 at @xmath72 .
does a meaningful continuum limit exist in this case ?
the answer is of course `` yes '' . as it follows from eq.([solution2 ] ) , when @xmath57 ( and therefore @xmath50 ) goes to infinity , and the bare coupling @xmath17 is arbitrary but fixed , @xmath63 is equal to the fixed value , @xmath71 , for all @xmath73 .
therefore it is a non - trivial conformal field theory .
so far we considered the solution for @xmath63 connected with the perturbative ( and perturbatively exact in the t hooft renormalization scheme @xcite ) @xmath23 function ( [ beta ] ) .
however , unlike ultraviolet stable fixed points , defining dynamics at high momenta , infrared - stable fixed points ( defining dynamics at low momenta ) are very sensitive to nonperturbative dynamics leading to the generation of particle masses .
for example , if fermions acquire a dynamical mass , they decouple from the infrared dynamics , and therefore the perturbative infrared fixed point ( [ alpha^ * ] ) will disappear . the phase diagram in the ( @xmath74)-plane in this theory was suggested in ref.@xcite .
it is shown in fig .
the left - hand portion of the curve in this figure coincides with the line of the infrared - stable fixed points @xmath75 in eq.([alpha^ * ] ) .
it separates two symmetric phases , @xmath76 and @xmath77 , with @xmath78 and @xmath79 , respectively .
its lower end is @xmath80 ( with @xmath81 if @xmath12 ) : at @xmath82 the infrared fixed point is washed out by generating a dynamical fermion mass .
the horizontal , @xmath80 , line describes a phase transition between the symmetric phase @xmath76 and the phase with confinement and chiral symmetry breaking .
as it was suggested in refs.@xcite , based on a similarity of this phase transition with that in quenched @xmath83 @xcite and in @xmath84 @xcite , there is the following scaling law for @xmath85 : @xmath86 where the constant @xmath87 is of order one and @xmath88 is a scale at which the running coupling is of order @xmath89 .
it is a continuous phase transition with an essential singularity at @xmath80 .
the characteristic point of this phase transition is that the critical line @xmath80 separates phases with essentially different spectra of low energy excitations @xcite and the different structure of the equation for the divergence of the dilatation current ( i.e. with essentially different realizations of the conformal symmetry ) @xcite .
it was called the conformal phase transition in ref.@xcite . at present
it is still unclear whether the phase transition on the line @xmath80 is indeed a continuous phase transition with an essential singularity or it is a first order phase transition @xcite .
however , anyway , the two properties ( the abrupt change of the spectrum of excitations and the different structure of the equation for the divergence of the dilatation current in those two phases ) have to take place . at last ,
the right - hand portion of the curve on the diagram occurs because at large enough values of the bare coupling , spontaneous chiral symmetry breaking takes place for any number @xmath0 of fermion flavors .
this portion describes a phase transition called a bulk phase transition in the literature , and it is presumably a first order phase transition .
is large enough , is valid at least for lattice theories with kogut - susskind fermions .
notice however that since the bulk phase transition is a lattice artifact , the form of this portion of the curve can depend on the type of fermions used in simulations ( for details , see ref.@xcite ) . ]
the vertical line ends above @xmath0=0 since in pure gluodynamics there is apparently no phase transition between weak - coupling and strong - coupling phases . up to now
we have considered the case of a chiral invariant action . but how will the dynamics change if a bare fermion mass term is added in the action ?
this question is in particular relevant for lattice computer simulations : for studying a chiral phase transition on a finite lattice , it is necessary to introduce a bare fermion mass .
we will show that adding even an arbitrary small bare fermion mass results in a dramatic changing the dynamics both in the @xmath76 and @xmath77 phases .
recall that in the case of confinement su(@xmath7 ) theories , with a small , @xmath10 , number of fermion flavors , the role of a bare fermion mass @xmath90 is minor if @xmath91 ( where @xmath45 is a confinement scale ) .
the only relevant consequence is that massless nambu - goldstone pseudoscalars get a small mass ( the pcac dynamics ) .
the reason for that is the fact that the scale @xmath45 , connected with a scale anomaly , describes the breakdown of the conformal symmetry connected _ both _ with perturbative and nonperturbative dynamics : the running coupling and the formation of bound state .
certainly , a small bare mass @xmath91 is irrelevant for the dynamics of those bound states .
now let us turn to the phase @xmath76 and @xmath77 , with @xmath9 . at finite @xmath44 in @xmath76 and @xmath50 in @xmath77 ,
there is still conformal anomaly : because of the running of the effective coupling constant , the conformal symmetry is broken .
it is restored only if @xmath92 in @xmath76 and @xmath93 in @xmath77 . however , the essential difference with respect to confinement theories is that both @xmath44 and @xmath50 have nothing with the dynamics forming bound states : since at @xmath9 the effective coupling is relatively weak , it is impossible to form bound states from @xmath94 fermions and gluons ( recall that the @xmath76 and @xmath77 phases are chiral invariant ) . therefore the absence of a mass for fermions and gluons is a key point for _ not _ creating bound states in those phases .
the situation changes dramatically if a bare fermion mass is introduced : indeed , even weak gauge , coulomb - like , interactions can easily produce bound states composed of massive constituents , as it happens , for example , in qed , where electron - positron ( positronium ) bound states are present . to be concrete ,
let us first consider the case when all fermions have the same bare mass @xmath90 .
it leads to a mass function @xmath95 in the fermion propagator @xmath96 .
the current fermion mass @xmath97 is given by the relation @xmath98 for the clearest exposition , let us consider a particular theory with a finite cutoff @xmath57 and the bare coupling constant @xmath99 being not far away from the fixed point @xmath20 .
then , the mass function is changing in the `` walking '' regime @xcite with @xmath100 .
it is @xmath101 where the anomalous dimension @xmath102 @xcite .
eqs.([m ] ) and ( [ m(q ) ] ) imply that @xmath103 there are two main consequences of the presence of the bare mass : \(a ) bound states , composed of fermions , occur in the spectrum of the theory .
the mass of a n - body bound state is @xmath104 ; \(b ) at momenta @xmath105 , fermions and their bound states decouple .
there is a pure su(@xmath7 ) yang - mills theory with confinement .
its spectrum contains glueballs . to estimate glueball masses ,
notice that at momenta @xmath105 , the running of the coupling is defined by the parameter @xmath106 of the yang - mills theory , @xmath107 therefore the glueball masses @xmath108 are of order @xmath109 for @xmath18 , we find from eqs.([b ] ) , ( [ c ] ) , and ( [ barb ] ) that @xmath110 is @xmath111 , @xmath112 , @xmath113 , and @xmath114 for @xmath0=16 , 15 , 14 , and 13 , respectively
. therefore at @xmath0=16 , 15 and 14 , the glueball masses are essentially lighter than the masses of the bound states composed of fermions .
the situation is similar to that in confinement qcd with heavy quarks , @xmath115 .
however , there is now a new important point : in the conformal window , _ any _ value of @xmath90 ( and therefore @xmath97 ) is `` heavy '' : the fermion mass @xmath97 sets a new scale in the theory , and the confinement scale @xmath116 ( [ ym ] ) is less , and rather often much less , than this scale @xmath97 .
this leads to a spectacular `` experimental '' signature of the conformal window in lattice computer simulations : glueball masses rapidly , as @xmath117 , decrease with the bare fermion mass @xmath90 for _ all _ values of @xmath90 less than cutoff @xmath57 .
few comments are in order : \(1 ) the phases @xmath76 and @xmath77 have essentially the same long distance dynamics .
they are distinguished only by their dynamics at short distances : while the dynamics of the phase @xmath76 is asymptotically free , that of the phase @xmath77 is not .
in particular , when all fermions are massive ( with the current mass @xmath97 ) , the continuum limit @xmath118 of the @xmath77-theory is a non - asymptotically free confinement theory .
its spectrum includes colorless bound states composed of fermions and gluons .
for @xmath119 the running coupling @xmath63 is the same as in pure su(@xmath7 ) yang - mills theory , and for all @xmath120 @xmath63 is very close to @xmath20 ( `` walking '' , actually , `` standing '' dynamics ) . for those values @xmath0 for which @xmath20 is small ( as @xmath0=16 , 15 and 14 at @xmath7=3 ) , glueballs are much lighter than the bound states composed of fermions .
notice that , unlike the case with @xmath121 , there exists an s - matrix in this theory .
\(2 ) in order to get the clearest exposition , we assumed such estimates as @xmath81 for @xmath122 and @xmath123 for the anomalous dimension @xmath124 . while the latter should be reasonable for @xmath125 ( and especially for @xmath126 ) @xcite , the former is based on the assumption that @xmath127 which , though seems reasonable , might be crude for some values of @xmath7 .
it is clear however that the dynamical picture presented in this paper is essentially independent of those assumptions .
\(3 ) so far we have considered the case when all fermions have the same bare mass @xmath90 . the generalization to the case
when different fermions may have different bare masses is evident .
\(4 ) lattice computer simulations of the su(3 ) theory with a relatively large number of @xmath0 @xcite indeed indicate on the existence of a symmetric phase .
however , the value of the critical number @xmath122 is different in different simulations : it varies from @xmath128 @xcite through @xmath129 @xcite .
i hope that the signature of the conformal window considered in this talk can be useful to settle this important issue .
how does the structure of the qcd @xmath23 function reflect the existence of the phase transition with respect to @xmath0 ?
this problem has been addressed in the work @xcite .
the impetus for that work was the structure of the @xmath23 function of the @xmath130 susy gluodynamics which is known exactly if no matter fields are present @xcite : @xmath131 ; \;\;x\equiv\frac{\alpha}{\pi}. \label{nsvz}\ ] ] this structure shows the following two noticeable features : \a ) eq.([nsvz ] ) implies that , besides the conventional asymptotically free phase with @xmath132 , there exists a strong ultraviolet phase in which the coupling @xmath133 is greater than the @xmath23-function pole at @xmath134 @xcite . in that phase ,
the value @xmath135 is an ultraviolet fixed point . as in the case of the infrared fixed point considered in the previous section ,
these two phases share common infrared properties . because of that
, the @xmath23-function pole is called an infrared attractor . unlike theories with an infrared fixed point , theories with an infrared attractor correspond to the confinement phase . in ref .
@xcite , we addressed whether pad - summations of the @xmath136 qcd @xmath23 function for a given number of flavors exhibit an infrared fixed point , or alternatively , an infrared attractor .
the main results are the following .
below an approximant - dependent flavor threshold @xmath137 , the pad - summation @xmath23 functions incorporating @xmath138 , [ 1|2 ] , [ 2|2 ] , [ 1|3]$ ] , and @xmath139 $ ] approximants whose maclaurin expansions match known higher - than - one - loop contributions to the @xmath23-function series always exhibit a positive pole prior to the occurrence of their first positive zero , precluding any identification of this first positive zero as an infrared fixed point .
this result is shown to be true regardless of the magnitude of the presently - unknown five - loop @xmath23-function contribution explicitly appearing within pad - summation @xmath23 functions incorporating @xmath140 , [ 1|3]$ ] , and @xmath139 $ ] approximants . like in the case of supersymmetric gluodynamics
, the pole in question suggests the occurrence of dynamics in which both a strong and an asymptotically free phase share a common infrared attractor . as @xmath0 increases above an approximant - dependent flavor threshold , pad - summation @xmath23 functions exhibit dynamics controlled by an infrared fixed point .
this fixed point decreases in magnitude with increasing flavor number . thus utilizing pad - summation qcd @xmath23 functions
, we obtain a good degree of agreement with infrared properties predicted @xcite via the t hooft renormalization scheme @xcite in which the @xmath23 function is truncated subsequent to two - loop order .
it is noticeable that the infrared structure of the @xmath23 function we obtained in qcd with a small number of flavors is similar to that of the @xmath23 function in susy gluodynamics : there are strong arguments in the literature in the support of essentially the same mechanism of confinement in those theories .
i am grateful to the organizers of the tmu - yale symposium , in particular , hisakazu minakata and noriaki kitazawa , for their warm hospitality .
my special thanks to koichi yamawaki for his hospitality during my stay at nagoya university .
this work was supported by the grant - in - aid of japan society for the promotion of science no .
11695030 .
b. holdom , phys . rev . *
d24 * , 1441 ( 1981 ) ; phys . lett . *
150b * , 301 ( 1985 ) ; k. yamawaki , m. bando , and k. matumoto , phys . rev . lett . * 56 * , 1335 ( 1986 ) ; t. akiba and t. yanagida , phys . lett . *
169b * , 308 ( 1986 ) ; t. appelquist , d. karabali , and l. c. r. wijewardhana , phys .
lett . * 57 * , 957 ( 1986 ) . | the dynamics in qcd with different number of fermion flavors @xmath0 is discussed .
the emphasis is on the description of the conformal phase transition with respect to @xmath0 separating a phase with no confinement and chiral symmetry breaking and a phase with confinement and with chiral symmetry breaking . |
neutrinoless double-@xmath0 decay ( @xmath27 ) is one of the most promising probes for physics beyond the standard model .
the process is sensitive to lepton number violation , the nature of the neutrino ( majorana or dirac particle ) , the absolute neutrino mass and the neutrino mass hierarchy ( see @xcite and references therein ) . in particular , the investigation of neutrinoless double - electron capture ( @xmath28 ) and of electron capture with positron emission ( @xmath29 ) could clarify the possible contribution of the right - handed currents to the @xmath30 decay rate @xcite .
the osmium contains two potentially double-@xmath0 active isotopes : @xmath2os ( energy of decay @xmath31 kev @xcite ; isotopic abundance @xmath32% @xcite ; allowed decay channels : @xmath33 and @xmath34 ) and @xmath6os ( @xmath35 kev @xcite ; @xmath36% @xcite ; @xmath37 ) .
the decay schemes of @xmath2os and @xmath6os are presented in figs . 1 and 2 , respectively .
there is a possibility of a resonant enhancement of the @xmath38 double - electron capture in @xmath2os to a few excited levels of @xmath2w .
the most promising of them is the level @xmath39 1322.2 kev @xcite .
we have investigated the radiopurity of the osmium sample and realized the first search for double-@xmath0 processes in @xmath2os and @xmath6os with the help of an ultra - low background hpge @xmath1 spectrometer .
preliminary results of the experiment were presented in @xcite .
an ultra - pure osmium sample ( more than 99.999% purity grade @xcite ) with a mass of 173 g was used in the experiment .
the material was obtained using electron - beam melting of osmium powder with further purification by electron - beam zone refining .
the estimated density of the metal is 23 g/@xmath40 ( the tabulated value is 22.57 g/@xmath40 @xcite ) .
the sample contains @xmath41 nuclei of @xmath2os and @xmath42 nuclei of @xmath6os assume the natural isotopic composition @xcite .
the experiment was carried out at the gran sasso national laboratories of the infn ( italy ) with the ultra - low background hpge detector gecris with a volume of 465 @xmath40 .
the detector is shielded by lead ( @xmath43 cm ) and copper ( @xmath44 cm ) .
the fwhm energy resolution of the spectrometer is 2.0 kev for the 1333 kev @xmath1 quanta of @xmath11co .
the data with the sample were accumulated over 2741 h , while the background spectrum was taken over 1046 h. the spectra are presented in fig .
3 . the peaks in the energy distributions can be ascribed to @xmath1 quanta of u / th daughters , @xmath10k , @xmath11co , @xmath7cs , @xmath8os and @xmath9bi .
the response functions of the detector to decays of these nuclides as well as to the double-@xmath0 processes in the osmium isotopes were simulated by egsnrc @xcite and geant4 @xcite packages with initial kinematics given by the decay0 event generator @xcite .
both simulations gave consistent results quanta with the energy 418.8 kev , expected to be emitted in the resonant @xmath28 process in @xmath2os , is 3.14% ( 3.12% ) with egsnrc ( geant4 ) . ] .
some excess of events in the 662 kev peak of @xmath7cs was observed with an activity ( @xmath45 ) mbq / kg .
the radioactive @xmath8os ( electron capture decay , @xmath46 d @xcite ) was also detected in the sample with an activity ( @xmath47 ) mbq / kg .
we assume that this radionuclide was generated before the installation of the sample into the set - up by capture of neutrons by @xmath2os and by spallation processes induced by cosmic rays on heavier osmium isotopes .
the presence of @xmath9bi at the level of ( @xmath48 ) mbq / kg can be explained by contamination of the sample .
the activities of @xmath7cs , @xmath8os and @xmath9bi , as well as upper limits of @xmath10k , @xmath11co and u / th daughters are presented in table 1 .
.radioactive impurities in the osmium sample .
the upper limits are presented at 90% cl , the uncertainties are given with 68% cl .
the reference date for activities is october 2011 . [
cols="<,<,<",options="header " , ] we do not observe any peaks in the energy distribution accumulated with the osmium sample which could indicate double-@xmath0 activity of @xmath2os or @xmath6os
. therefore only lower half - life limits ( @xmath49 ) can be set according to the formula : @xmath50 where @xmath51 is the number of potentially @xmath52 unstable nuclei , @xmath53 is the detection efficiency ( in which the yields of the specific @xmath1 quanta in accordance with the scheme of the decay , fig . 1 , are included ) , @xmath54 is the measuring time , and @xmath55 is the upper limit on the number of events of the effect searched for which can be excluded at a given confidence level ( cl ) .
all the half - life limits and the values of @xmath55 are presented in this paper at 90% cl .
one positron can be emitted in the @xmath34 decay of @xmath2os with an energy up to ( @xmath56 ) kev .
the annihilation of the positron should give two 511 kev @xmath1 s leading to an extra rate in the annihilation peak . to estimate @xmath55 for the decay ,
the energy spectra were fitted in the energy interval ( @xmath57 ) kev ( see fig . 4 )
. there are ( @xmath58 ) events in the 511 kev peak in the data accumulated with the osmium sample .
the area of the annihilation peak in the background spectrum is ( @xmath59 ) counts , corresponding to ( @xmath60 ) counts after normalising to the exposure of the osmium sample .
the difference in the areas of the annihilation peak with and without sample [ ( @xmath61 ) counts ] gives no indication of the effect . in accordance with the feldman - cousins procedure
@xcite we should take @xmath62 counts which can be excluded at 90% cl .
taking into account almost the same detection efficiency for 2@xmath63 ( @xmath53=9.0% ) and 0@xmath63 ( 8.9% ) modes , we have obtained the following limit on the half - life of @xmath2os for the @xmath64 and @xmath38 modes of @xmath34 decay ( see also table 2 where the energies of @xmath1 quanta used in the analysis , the detection efficiencies and the values of @xmath55 are given ) : @xmath65os , g.s .
@xmath66 g.s.)@xmath67 yr . the @xmath68 decay of @xmath2os is also allowed to the first excited level of @xmath2w with an energy of 111.2 kev .
the strongest restrictions on the decay were obtained by analysis of the annihilation peak .
taking into account slightly different detection efficiencies for @xmath64 ( 9.0% ) and @xmath38 ( 8.8% ) modes , we have obtained the following limits : @xmath69os , [email protected] kev@xmath71 yr , @xmath72os , [email protected] kev@xmath73 yr . in the case of the @xmath74 capture in @xmath2os
, a cascade of x rays and auger electrons with individual energies up to 69.5 kev is expected .
the most intensive x ray lines of tungsten are @xcite : 58.0 kev ( the yield of the x ray quanta is 27.4% ) , 59.3 kev ( 47.0% ) , 67.0 kev ( 5.4% ) , 67.2 kev ( 10.3% ) and 69.1 kev ( 3.6% ) . to derive a limit on the decay ,
the energy spectrum was fitted in the energy interval ( @xmath75 ) kev by the model consisting of five gaussians ( to describe the expected x ray peaks ) plus a polynomial function of the first degree ( background ) .
the fit provides the total area of the @xmath74 effect ( @xmath76 ) counts , which corresponds to @xmath77 counts ( see fig .
the detection efficiency of the whole effect in this case is calculated as : @xmath78 , where @xmath79 are the efficiencies for the specific x lines with energies from 58.0 kev to 69.1 kev .
taking into account @xmath80 , one can calculate the following half - life limit : @xmath81os , [email protected].@xmath82 yr . the sensitivity of the experiment to the @xmath83 decay of @xmath2os to the 111.2 kev excited level of @xmath2w is estimated from the fit of the energy spectrum accumulated with the osmium sample in the vicinity of the energy 111.2 kev .
the fit gives an area for the @xmath1 peak of ( @xmath84 ) counts , that leads to @xmath85 counts ( see fig . 5 ) .
the detection efficiency for the 111.2 kev @xmath1 quanta , taking into account the fact that the transition from the 111.2 kev excited level to the ground state of @xmath2w is efficiently converted to electrons ( the experimental coefficient of conversion is @xmath86 @xcite ) , was calculated to be 0.076% .
generally speaking , due to the large coefficient of conversion to electrons , the sensitivity of the experiment should be different for @xmath87 , @xmath88 and @xmath89 captures .
for example , since both @xmath90 electrons are already captured in the @xmath87 process , the conversion from the @xmath90 shell is prohibited in further 111.2 kev deexcitation process .
the coefficients of conversion from different atomic shells can be calculated with the bricc program @xcite .
however , the excited 111.2 kev level of @xmath2w has quite a long half - life of 1.251 ns @xcite compared to the characteristic atomic relaxation time of @xmath91 s ( see e.g. @xcite ) .
thus , we consider that the relaxation processes in the w atomic shell after the initial @xmath83 capture are finished before the 111.2 kev deexcitation and we can apply the value @xmath92 for all @xmath83 captures .
this gives : @xmath93os , [email protected] kev@xmath94 yr . to estimate limits on the @xmath83 decay of @xmath2os to the @xmath95 903.3 kev
, @xmath96 1002.5 kev and @xmath95 1121.4 kev excited levels of @xmath2w , the energy spectrum accumulated with the osmium sample was fitted in the energy intervals where intense @xmath1 peaks from the de - excitation process are expected .
for instance , we have obtained the following limit on the @xmath83 decay of @xmath2os to the excited 903.3 kev level of @xmath2w by the fit of the data in the energy interval ( @xmath97 ) kev where the 903.3 kev @xmath1 peak is expected : @xmath93os , g.s.@xmath98kev@xmath99 yr . the limits obtained for the @xmath83 decay of @xmath2os to the excited levels of @xmath2w are presented in table 2 . in the neutrinoless double - electron capture in addition to x rays
we suppose that only one bremsstrahlung @xmath1 quantum is emitted @xcite to take away the rest of the energy , which in the @xmath64 process is taken by the neutrinos . in the case of the @xmath38
double- electron capture from @xmath90 and @xmath100 shells , the energy of the @xmath1 quanta is expected to be equal to @xmath101 , where @xmath102 are the binding energies of the captured electrons on the @xmath90 and @xmath100 atomic shells of tungsten @xcite : @xmath103 kev , @xmath104 kev , @xmath105 kev , @xmath106 kev .
therefore , the expected energies of the quanta for the @xmath28 capture in @xmath2os to the ground state of @xmath2w are : @xmath107 kev for @xmath87 , @xmath108 kev for @xmath88 and @xmath109 ) kev for @xmath89
. there are no clear peaks in the data with these energies ( see fig .
we have estimated the values of @xmath55 by fitting the data in the energy intervals with the simple model : a gaussian function ( to describe the peaks searched for ) plus a polynomial function ( continuous background ) . taking into account the calculated efficiencies to detect @xmath1 quanta with energies of @xmath110 mev ( @xmath111 ) ,
we set the following limits on the processes searched for : @xmath112os , [email protected].@xmath114 yr , @xmath115os , [email protected].@xmath116 yr , @xmath117os , [email protected].@xmath118 yr . the @xmath119 energy of @xmath2os allows also the population of several excited levels of @xmath2w .
the limits obtained for the processes of @xmath28 decay to the ground and to the excited levels of the daughter nuclei are presented in table 2 .
the neutrinoless double - electron capture of @xmath2os to the excited levels of @xmath2w with energies of 1322.2 kev , 1386.3 kev and 1431.0 kev may occur with higher probability due to a possible resonant enhancement .
the double capture on the @xmath39 1322.2 kev level was considered as the most promising with a half - life in the range ( @xmath120 ) yr for the majorana neutrino mass of 1 ev @xcite value for the double- @xmath0 decay of @xmath2os @xcite , the authors have concluded that the half - life of the transition exceeds @xmath121 yr . ] .
limits on resonant double - electron capture in @xmath2os from @xmath90 and @xmath100 shells were obtained by analyzing the experimental spectrum in the energy intervals where the most strongest @xmath1 lines from the de - excitation of these levels are expected .
for example , the greatest sensitivity to the resonant @xmath87 capture in @xmath2os to the excited @xmath122 level of @xmath2w with an energy of 1322.2 kev is obtained by searching for the de - excitation @xmath1-ray with an energy 903.3 kev , yielding the limit : @xmath123os , g.s.@xmath124 kev@xmath125 yr .
limits on the resonant double - electron captures to the excited levels @xmath95 1386.3 kev and @xmath95 1431.0 kev were obtained in a similar way .
the half - life limits on the resonant @xmath33 processes in @xmath2os are presented in table 2 . to set a limit on the @xmath37 transition of @xmath6os to the @xmath95 316.5 kev excited level of @xmath6pt ,
the energy spectrum was fitted by a straight line ( which represents the background model ) and a gaussian at 316.5 kev .
the energy spectrum in the vicinity of the peak is presented in fig .
7 . taking into account the detection efficiency ( @xmath127 both for the @xmath64 and @xmath38 modes of the decay )
, we have obtained the following limit : @xmath128os , g.s.@xmath129 kev@xmath130 yr .
the radiopurity of an osmium sample has been investigated and the first experiment to search for @xmath52 processes in @xmath2os and @xmath6os was carried out by using ultra - low background hpge @xmath1 spectrometry .
after 2741 h of data taking with a 173 g ultra - pure osmium sample limits on double - beta processes in @xmath2os have been established at the level of @xmath131 ) yr . a possible resonant neutrinoless double - electron capture in @xmath2os to the excited 1322.2 kev , 1386.3 kev and
1431.0 kev states of @xmath2w are bound at the level of @xmath132 yr .
the @xmath37 decay of @xmath6os to the first excited level of @xmath6pt is restricted to @xmath133 yr at 90% cl .
the osmium sample has subsequently been placed in a well - type ultra low background hpge detector especially designed for low energy @xmath1 rays spectrometry decay of @xmath2os and @xmath134os to the excited levels of their daughter nuclei ( see @xcite where an indication for @xmath135 decay of @xmath2os to the first excited level ( 103.5 kev ) of @xmath136w with the half - life @xmath137 yr was obtained ) . ] .
in such a way we expect to improve the sensitivity of the experiment at least to the @xmath74 decay , which is the most probable double - beta process in @xmath2os .
further progress can be achieved by using osmium enriched in @xmath2os , although this will require new enrichment techniques to be developed .
the authors from the institute for nuclear research ( kyiv , ukraine ) were supported in part by the space research program of the national academy of sciences of ukraine .
99 yu.g .
zdesenko , rev .
74 ( 2002 ) 663 .
elliot , p. vogel , ann .
nucl . part .
52 ( 2002 ) 115 .
vergados , phys . rep . 361
( 2002 ) 1 .
elliot , j. engel , j. phys .
g 30 ( 2004 ) r183 .
avignone iii , g.s .
king , yu.g .
zdesenko , new j. phys . 7
( 2005 ) 6 .
h. ejiri , j. phys .
japan 74 ( 2005 ) 2101 .
avignone iii , s.r .
elliott , j. engel , rev .
mod . phys . 80 ( 2008 )
klapdor - kleingrothaus , int .
e 17 ( 2008 ) 505 .
w. rodejohann , int . j. mod
e 20 ( 2011 ) 1833 .
elliott , mod .
lett . a 27 ( 2012 ) 1230009 .
vergados , h. ejiri , f. simkovic , rep .
phys . 75 ( 2012 ) 106301 .
m. hirsch et al . , z. phys . a 151 ( 1994 ) 347
c. smorra et al . , phys .
c 86 ( 2012 ) 044604 .
m. berglund , m.e .
wieser , pure appl . chem . 83
( 2011 ) 397 .
g. audi , a.h .
wapstra , c. thibault , nucl .
a 729 ( 2003 ) 337 .
krivoruchenko et al . , nucl .
phys . a 859 ( 2011 ) 140 .
baglin , nucl .
data sheets 111 ( 2010 ) 275 .
baglin , nucl .
data sheets 113 ( 2012 ) 1871 .
p. belli et al .
, first search for double beta decay of osmium by low background hpge detector , to be published in proc . of 4@xmath138 int .
npae - kyiv2012 , 37 september 2012 , kyiv , ukraine .
azhazha , g.p .
kovtun , g.f .
tihinsky , the obtaining and metallophysics of high - pure metals , metallofizika i noveishie tekhnologii 22 ( 2000 ) 21 ( in russian ) .
lide ( ed ) , crc handbook of chemistry and physics , 84th edition .
crc press .
boca raton , florida , 2003 .
i. kawrakow and d.w.o .
rogers , nrcc report pirs-701 ( 2003 ) s. agostinelli et al .
instrum . meth . a 506 ( 2003 ) 250
; j. allison et al . , ieee trans .
53 ( 2006 ) 270 .
ponkratenko et al .
, phys . at .
( 2000 ) 1282 ; + v.i .
tretyak , to be published .
firestone et al . , _ table of isotopes _ , 8-th ed . , john wiley , new york , 1996 and cd update , 1998 .
feldman , r.d .
cousins , phys .
d 57 ( 1998 ) 3873 . t. kib@xmath139di et al . , nucl . instrum . meth . a 589 ( 2008 ) 202
; + http://www.nndc.bnl.gov / nndcscr / ensdf@xmath140pgm / analysis / bricc/. m. drescher et al . ,
nature 419 ( 2002 ) 803 .
winter , phys .
rev . 100 ( 1955 ) 142 . | a search for double-@xmath0 decay of osmium has been realized for the first time with the help of an ultra - low background hpge @xmath1 detector at the underground gran sasso national laboratories of the infn ( italy ) .
after 2741 h of data taking with a 173 g ultra - pure osmium sample limits on double-@xmath0 processes in @xmath2os have been established at the level of @xmath3 yr .
possible resonant double - electron captures in @xmath2os were searched for with a sensitivity @xmath4 yr . a half - life limit @xmath5 yr was set for the double-@xmath0 decay of @xmath6os to the first excited level of @xmath6pt .
the radiopurity of the osmium sample has been investigated and radionuclides @xmath7cs , @xmath8os and @xmath9bi were detected in the sample , while activities of @xmath10k , @xmath11co , @xmath12ra and @xmath13th were limited at the @xmath14 mbq / kg level . * first search for double-@xmath0 decay of @xmath2os and @xmath6os * 1.0 cm * p. belli@xmath15 , r. bernabei@xmath16 , f. cappella@xmath17 , r. cerulli@xmath18 , f.a .
danevich@xmath19 , s. dangelo@xmath20 , a. di marco@xmath20 , a. incicchitti@xmath17 , g.p .
kovtun@xmath21 , n.g .
kovtun@xmath21 , m. laubenstein@xmath18 , d.v .
poda@xmath19 , o.g .
polischuk@xmath22 , a.p .
shcherban@xmath21 , v.i .
tretyak@xmath19 * 0.3 cm @xmath15_infn , sezione roma `` tor vergata '' , i-00133 rome , italy
_ @xmath23_dipartimento di fisica , universit@xmath24 di roma `` tor vergata '' , i-00133 rome , italy _ @xmath25_infn , sezione roma `` la sapienza '' , i-00185 rome , italy _ @xmath26_dipartimento di fisica , universit@xmath24 di roma `` la sapienza '' , i-00185 rome , italy _ @xmath18_infn , laboratori nazionali del gran sasso , 67100 assergi ( aq ) , italy _
@xmath19_institute for nuclear research , msp 03680 kyiv , ukraine _ @xmath21_national science center `` kharkiv institute of physics and technology '' , 61108 kharkiv , ukraine _ 0.5 cm 0.4 cm _ pacs _ : 23.40.-s ; 23.60.+e 0.4 cm _ keywords _ : double - beta decay ; @xmath2os ; @xmath6os ; ultra - low background hpge spectrometry |
we gratefully acknowledge financial support from the national research foundation , the csir nlc , and the nccr must research instrument of the swiss national science foundation .
xun gu , lin xu , mark kimmel , erik zeek , patrick oshea , aparna p. shreenath , rick trebino , and robert s. windeler .
frequency - resolved optical gating and single - shot spectral measurements reveal fine structure in microstructure - fiber continuum .
, 27(13):11741176 , 2002 .
john dudley , xun gu , lin xu , mark kimmel , erik zeek , patrick oshea , rick trebino , stphane coen , and robert windeler .
cross - correlation frequency resolved optical gating analysis of broadband continuum generation in photonic crystal fiber : simulations and experiments .
, 10(21):12151221 , 2002 .
jia liu , yahui feng , hao li , peifen lu , haifeng pan , jian wu , and heping zeng .
supercontinuum pulse measurement by molecular alignment based cross - correlation frequency resolved optical gating .
, 19(1):4046 , 2011 . | we demonstrate that time - domain ptychography , a recently introduced ultrafast pulse reconstruction modality , has properties ideally suited for the temporal characterization of complex light pulses with large time - bandwidth products as it achieves temporal resolution on the scale of a single optical cycle using long probe pulses , low sampling rates , and an extremely fast and robust algorithm . in comparison to existing techniques , ptychography minimizes the data to be recorded and processed , and drastically reduces the computational time of the reconstruction .
experimentally we measure the temporal waveform of an octave - spanning , 3.5 ps long supercontinuum pulse generated in photonic crystal fiber , resolving features as short as 5.7 fs with sub - fs resolution and 30 db dynamic range using 100 fs probe pulses and similarly large delay steps .
supercontinuum ( sc ) generation has become a scientific and commercial success story , in particular driven by photonic crystal fiber ( pcf ) technology . tailored fiber - based sc sources delivering spectral bandwidths of one octave and more
are now routinely applied in bio - photonic imaging , microscopy , frequency metrology , and nonlinear pulse compression , to name only a few @xcite . despite such wide - spread use , the full characterization of such sc pulses remains extremely challenging as their temporal waveforms are amongst the most complex to be found in ultrafast optics .
their envelope is typically several picoseconds long , but contains fine structure on the scale of just a few femtoseconds .
combined with their ultrabroad and equally complex sprectrum they easily reach time - bandwidth products ( tbp ) in the order of @xmath0 @xcite . since no known detector is fast enough to measure such pulses directly ,
several schemes for the indirect reconstruction of temporal intensity and phase have been devised , of which cross - correlation frequency resolved optical gating ( xfrog ) in various implementations has been most successful in adequately measuring complex sc pulses @xcite .
it is based on recording the spectrogram of the unknown object pulse cross - correlated via a nonlinear interaction with a series of known time - delayed gate pulses and iteratively reconstructing the electric field using a derivative of the generalized projections algorithm ( gpa ) @xcite . for the accurate reconstruction of pulses with large tbp high sensitivity , large ranges , and fine resolution in both temporal and spectral domains
have to be achieved simultaneously , which in the case of xfrog entails the recording of a large number of time - gated spectra using short gate pulses . for such complex pulses
the gpa demands the interpolation of the measured spectrogram on a time - frequency grid with a number of points in the order of @xmath1 or more @xcite , all of which makes the experimental measurement and the reconstruction of the pulse a rather cumbersome and computationally intensive task .
recently time - domain ptychography was introduced as a new modality for ultrafast pulse characterization @xcite , with an application range even extending down to the attosecond regime @xcite .
originating from lensless spatial imaging , ptychography is based on a powerful phase retrieval algorithm , the ptychographic iterative engine ( pie ) @xcite .
while the experimental realization is similar to xfrog in the sense that the cross - correlation spectrogram is recorded , the data processing and the reconstructing algorithm are conceptually very different and offer several key advantages , which become especially relevant in the context of measuring ultra - complex sc pulses : 1 ) the temporal resolution is independent of both the duration of the gate pulse and the time delay increment .
in fact , the gate pulse should be much longer than the object pulse ( or longer than the relevant temporal features contained therein ) .
2 ) typically only a few spectra have to be recorded since long gate pulses and large delay steps can be used .
3 ) the computational grid size required by pie is orders of magnitudes smaller than the one used in the gpa without compromising temporal resolution , resulting in fast processing and convergence . here
we demonstrate through simulations and experiments that these properties make time - domain ptychography an excellent choice for the efficient and accurate measurement of complex pulses with large tbp .
fig .
[ fig : setup ] shows a schematic setup for the implementation of time - domain ptychography of sc pulses .
a modelocked ti : sapphire oscillator generates fourier - limited pulses with 100 fs full width at half maximum ( fwhm ) duration centered at 800 nm with 1 w average power at 76 mhz repetition rate .
the pulses are split into two copies with variable power ratio using a polarizing beam splitter ( pbs ) and a half - wave plate ( hwp ) and serve both as the pump for the sc generation as well as the gate for the recording of the spectrogram .
the pump pulse is coupled into the fast axis of a 22.5 cm long polarization maintaining pcf ( nkt nl-750 pm ) with a zero - dispersion wavelength ( zdw ) of 760 nm .
the generated sc is collimated and passes through a set of achromatic hwp and pbs that allows to adjust the polarization direction and discard any sc components that might have become depolarized in the pcf .
the cross - correlated spectrogram @xmath2 is then generated by non - collinearly overlapping the sc field @xmath3 with the time - delayed gate @xmath4 in a @xmath5 m thick nonlinear bbo crystal and recording the spectrum of the sum - frequency generation ( sfg ) signal for different time delays @xmath6 .
note that it is also possible to use other forms of nonlinear interaction with different gate functions , such as phase - only gates @xcite .
we first perform numerical simulations closely modelled after the experimental setup in order to verify the performance of pie for pulses with large tbp .
we model the sc generation process by numerically solving the generalized nonlinear schrdinger equation ( gnlse ) using the dispersive and nonlinear properties provided by the pcf manufacturer @xcite .
the model has a temporal and angular spectral resolution of @xmath7 fs and @xmath8 thz , we assume a pump pulse peak power of 2.5 kw , and also take the collimating , focusing , and polarizing optics after the pcf into account .
since it is well known that the generated sc might suffer from substantial pulse - to - pulse fluctuations if the nonlinear broadening mechanisms are sensitive to shot noise @xcite , we also included quantum noise terms into the gnlse and calculated the spectrally averaged first order degree of coherence @xmath9 from a set of multiple simulations with random noise seeds @xcite . for our conditions @xmath10 , indicating that the spectrogram averaged over a large number of shots does not differ significantly from a single shot measurement .
hence we implement our experiment in a simple multi - shot setup assuming an identical pulse is available at every shot .
note , however , that a single - shot scheme for measuring complex spectrograms has recently been devised for a variant of xfrog @xcite , which could easily be adapted for ptychography should this become necessary for measuring pulses with larger fluctuations .
we use the result of the gnlse simulation to calculate the spectrogram from eq .
( [ eq : spectrogram ] ) and pass it to the pie reconstruction algorithm , which works on two essentially independent grids .
the object and gate pulse are sampled on a temporal grid with @xmath11 samples equally spaced by @xmath12 , which is determined by the resolution and the range of the spectrometer used such that @xmath13 . if the spectrogram contains the full spectral range of the pulse , zero padding can be used to increase @xmath11 and consequently adjust the desired temporal resolution of the reconstruction .
we choose @xmath14 such that @xmath7 fs , both for simulations and experiments .
the second grid is that of time delays and consists of @xmath15 samples equally spaced by @xmath16 .
ptychographic reconstruction works well as long as @xmath16 is smaller than the fwhm width @xmath17 of the gate pulse , i.e. for a sampling rate @xmath18 .
the resulting redundancy is the basis for the accuracy and robustness of the pie algorithm . thus @xmath16 can be chosen freely within the constraint @xmath19 .
@xmath15 is typically orders of magnitude smaller than @xmath11 , for example in this paper @xmath15 ranges between 3 and 103 , depending on @xmath20 and the desired delay range to be reconstructed .
hence pie requires @xmath15 cross - correlated spectra recorded at different time delays @xmath21 between the object and gate pulse , which are combined in the spectrogram @xmath22 sampled on a @xmath23 grid .
no further interpolation is required .
in contrast , gpa based schemes work with a spectrogram that has to fulfill the discrete fourier transform relations and hence needs to be interpolated on the delay axis to a @xmath24 grid @xcite . for achieving the same temporal resolution
as we do with pie , an enormous grid size of @xmath25 would have to be processed , independent of the actual delay range of interest .
pie therefore operates with 100 - 3000 times smaller grid size than the gpa . with typically hundreds of iterations needed before convergence
is reached , pie saves computational time in the order of hours compared to gpa on a standard modern pc .
this impressively demonstrates the power of ptychographic reconstruction , which becomes especially apparent when large temporal and spectral ranges have to be combined with fine resolution .
the simulated and reconstructed spectrograms ( @xmath26 arrays ) with @xmath27 fs and @xmath28 fs ( i.e. @xmath29 ) are displayed in figs .
[ fig : simulation](a ) and [ fig : simulation](b ) on a logarithmic intensity scale with 30 db dynamic range .
the spectrum spans the range 570 - 1050 nm , the duration of the pulse is about 3.5 ps , and the tbp is > 500 .
the agreement between simulation and reconstruction is excellent , both are visually indistinguishable .
the spectrogram is dominated by the group index profile of the pcf modified by a tilt caused by the dispersive optics in the setup ( dashed line in fig .
[ fig : simulation](a ) ) , and consists of a normal ( top ) and anomalous ( bottom ) dispersion arm . the compact objects in the anomalous dispersion arm
can be identified as solitons , while the low - level radiation in the top arm are the corresponding dispersive waves .
the temporal profile of the pulse is dictated by the overlap of these two dispersive arms and the resulting interference structures , which are particularly sensitive to the relative phase of the overlapping components .
two especially challenging structures for the reconstruction are highlighted in fig .
[ fig : simulation](c ) : ( i ) the densly packed solitons around @xmath30 , where small relative phase errors will result in large intensity errors ; and ( ii ) the overlap of the strongest soliton with a dispersive wave at @xmath31 fs , which causes an extremely fast temporal beating with 6.5 fs period , equal to the frequency of their spectral separation of about 150 thz .
both structures are perfectly resolved in the reconstruction . finally , the comparison of the spectra in fig .
[ fig : simulation](d ) shows that the ptychographic reconstruction is accurate even down to the -30 db level and hence achieves precise phase retrieval even in sections of the pulse where the spectral intensity is low .
this is a particularly important feature for sc pulse characterization , because the temporal profile critically depends on a correct relative phase across the whole bandwidth .
the conclusions of the visual comparison are confirmed by the root mean square error ( rms ) between original and reconstructed spectrogram , @xmath32 ^ 2}}{n m } } , \label{eq : rms}\ ] ] which we found to be limited only by machine precision and the computational time required to achieve it .
here we aborted the pie algorithm when @xmath33 was reached after about 300 iterations .
next we investigate the behaviour of pie for lower sampling rates by keeping @xmath27 fs fixed and increasing @xmath16 .
fig .
[ fig : delay ] shows the results of the temporal intensity reconstruction for @xmath16 equal to 70 fs , 90 fs , and 120 fs , focusing on the soliton group around zero delay for better visibility .
for @xmath34 fs ( @xmath35 ) , the reconstruction is of equal quality ( @xmath36 ) as the one presented in fig .
[ fig : simulation ] , but requires only half the number of spectra and half the grid size ( @xmath37 ) . for @xmath38 fs ,
@xmath20 approaches one , further decreasing the required number of spectra , but also the redundancy in the measurement and consequently the rebustness of the reconstruction . nevertheless , the main features are well recovered , with only small relative phase errors leading to slight discrepancies in the intensity of interference features ( @xmath39 ) . for @xmath40 fs , the measurement is undersampled due to @xmath41 , leading to larger errors and we find that @xmath42 does not fall below @xmath43 .
this set of simulations therefore confirms that pie is able to accurately reconstruct the full complexity of the pulse even for large @xmath16 as long as @xmath19 . note that this condition implies that the number of spectra can be further reduced by using longer gate pulses .
experimentally a limit to this rule might be posed by the required peak power in the nonlinear process that creates the cross - correlated signal .
fs showing spectrogram , temporal intensity and spectral phase for ( a ) @xmath44 and ( b ) @xmath45 spectra . ]
lastly , we highlight the capability of pie to do partial reconstructions , a feature that is particularly useful if only short sections of an sc pulse are of interest . as an example , we focus on the fast temporal beating on the envelope of the soliton at @xmath46 fs .
we again consider @xmath47 fs and @xmath28 fs as in fig .
[ fig : simulation ] , but reduce the size of the sampled delay window to the region of interest , thus simulating a partial measurement of the pulse . fig .
[ fig : slices ] shows the resulting spectrograms and reconstructions of temporal intensity and spectral phase for delay windows of 490 fs and 70 fs , corresponding to @xmath44 and @xmath45 spectra , respectively . in both cases , the temporal waveform is surprisingly well reconstructed .
the retrieved absolute spectral phase does not match the simulation , because there is no available information about the intermediary phase between the two isolated spectral components .
intererestingly , however , the relative phase modulo @xmath48 between the two spectral components is determined accurately , apparent from the correct reconstruction of the phase of the interference pattern in the temporal waveform .
note that in this example pie correctly resolves features on the time scale of two optical cycles using 100 fs probe pulses and requiring only three spectral measurements .
we now proceed to the experimental realization using the setup already described above ( fig .
[ fig : setup ] ) .
a practical challenge of this implementation is that the sfg phase - matching bandwidth of the bbo crystal is smaller than the maximum spectral bandwidth of the sc .
fast angle - dithering of the crystal during the measurement has been shown to solve this problem @xcite , but since pie only requires a small number of spectra we decided to simply repeat the measurement for different crystal angles and obtain the full spectrogram from a superposition after correcting for the phase - matching efficiency versus wavelength .
a maximum of three measurements with suitably chosen crystal angles were sufficient to cover the complete sc bandwidth .
the comparison of the spectral marginal , i.e. the integral of the measured spectrogram with respect to delay , to an independently measured spectrum ensured the accuracy of the post - processing @xcite .
standard background correction and fourier noise filtering were also applied .
fig .
[ fig : exp ] shows measured and reconstructed spectrograms with @xmath47 fs after conversion to fundamental wavelengths for pump peak powers between 200 w and 2.5 kw coupled into the pcf . following the results of the simulations ,
@xmath34 fs was chosen .
the sampling grid size is @xmath49 , where @xmath50 $ ] for fig .
[ fig : exp](a)-(c ) , respectively , resulting in a temporal resolution equal to that used previously in the simulations , i.e. @xmath7 fs .
we achieved excellent sensitivity allowing us to record all spectrograms with signal - to - noise ratios ( snr ) exceeding 30 db and enabling us to follow the development of the sc from low to high pump powers . in fig .
[ fig : exp](a)-(c ) we observe self - phase modulation , soliton fission , and fully developed sc spectrum with a bandwidth of about one octave , respectively . in all cases , the reconstruction is of excellent quality for an experimentally recorded data set , with @xmath51 for ( a ) , ( b ) and @xmath52 for the broadest spectrum ( c ) , limited mainly by the snr .
the reconstruction required a computational time of less than 1 minute on a standard laptop computer .
it is important to confirm that both the measured spectrogram and the reconstruction are correct . in fig .
[ fig : exp ] we therefore also show the measured spectral marginal and the retrieved spectrum , and compare both to an independently measured osa spectrum . in all cases
the agreement is very good , confirming the accuracy of both the measurement procedure and the reconstruction . since there is also a good qualitative agreement between simulated and measured sc spectrogram ( compare figs .
[ fig : simulation ] ( b ) and [ fig : exp ] ( c ) ) , we also know that the shortest temporal feature of the fully developed sc pulse should be the beating caused by the overlap of soliton and dispersive wave at @xmath53 fs with a beat frequency equal to their spectral separation . from the measured spectrogram
we determine that separation to be @xmath54 thz , resulting in an expected oscillation period of about 5.7 fs .
fig .
[ fig : pulse ] shows the reconstructed temporal waveform with a full range of 3.5 ps , where this expected period is exactly recovered and fully resolved .
interestingly , we also resolve the much weaker temporal beating with a period of 8.2 fs at the edge of the soliton at @xmath55 fs , caused by the overlap with a low - level dispersive wave separated by about 120 thz .
the amplitude of this oscillation is as low as @xmath56 on the normalized scale .
hence we achieve a dynamic range close to 30 db in the temporal domain . in conclusion
we have demonstrated that time - domain ptychography is a powerful technique for the temporal characterization of complex ultrashort pulses with large tbp .
it is able to accurately resolve features on the scale of a single optical cycle using 100 fs ( or longer ) gate pulses and equally large delay steps , minimizes the number of time - gated spectra to be recorded and processed , and is superior to comparable gpa based techniques in terms of memory requirements and computational speed . |
local starburst galaxies are excellent local laboratories to study the physics of galaxy building ( cf .
heckman 1998 ) .
it is now clear that the deposition of mechanical and thermal energy by multiple supernovae in starbursts leads to a global outflow of metal - enriched gas .
these outflows are called `` superbubbles '' during their early dynamical evolution , and `` superwinds '' after they blow out of the galaxy s ism ( e.g. , heckman 2000 and references therein ) .
such flows are expected to play an especially important role in the evolution of dwarf galaxies , whose relatively shallow potential wells make them susceptible to wind - driven loss of gas and newly - created metals ( e.g. , dekel & silk 1986 ; martin 1999 ) . since similar outflows appear to be common in high - redshift galaxies as well ( pettini et al .
1998 , 2000 ; tenorio - tagle et al . 1999 ) they are perhaps the most plausible mechanism by which the mass - metallicity relation in galactic spheroids was established ( e.g. , lynden - bell 1992 ) and the intergalactic medium was heated and chemically - enriched ( e.g. , ponman , cannon , & navarro 1999 ; gibson , loewenstein , & mushotzky 1997 ) .
one of the major uncertainties concerning starburst - driven outflows is the importance of radiative cooling : what fraction of the kinetic energy supplied by supernovae is carried out in the flow rather than being radiated away ? while the available x - ray data and models establish that radiative losses from hot gas ( @xmath18k ) are not severe ( see strickland & stevens 2000 ) , up until now there has been no direct observational probe of the coronal - phase gas ( @xmath19 to @xmath20 k ) that could dominate the radiative cooling . the recent launch of the @xmath21 ( @xmath0 - moos et al .
2000 ) provides access to the best probe of rapidly - cooling coronal gas in starburst outflows : the ovi@xmath22@xmath221032,1038 doublet .
accordingly , we have obtained @xmath0 spectra of a small sample of the nearest and brightest starbursts . in this paper , we present the first @xmath0 detection of @xmath11 in the galaxy ngc 1705 .
ngc 1705 is an ideal test - case for the study of coronal - phase gas .
this nearby ( d = 6.2 mpc ) dwarf starburst galaxy has the second highest vacuum - ultraviolet flux of any starburst galaxy in the extensive compilation of iue spectra in kinney et al .
the detailed investigation by meurer et al .
( 1992 ) established ngc 1705 as a prototypical example of a dwarf starburst undergoing mass - loss .
they were able to delineate a kpc - scale fragmented ellipsoidal shell of emission - line gas that was expanding at roughly 50 km s@xmath5 along our line - of - sight .
they also showed that the population of supernovae in the young super star cluster ( ngc 1705 - 1 ) was energetically - sufficient to drive this flow .
the expulsive nature of the flow was later confirmed by @xmath23 observations that showed that the ultraviolet interstellar absorption - lines towards ngc 1705 - 1 were blueshifted by 70 to 80 km s@xmath5 relative to the galaxy systemic velocity ( heckman & leitherer 1997 ; sahu & blades 1997 ; sahu 1998 ) .
hensler et al . (
1998 ) reported the detection of soft x - ray emission , presumably from hot gas inside the expanding emission - line nebula .
two @xmath0 observations of ngc1705 ( @xmath24 ; @xmath25 ) were obtained on 4 - 5 february 2000 .
the central super star cluster ngc 1705 - 1 was centered in the large ( lwrs , @xmath26 ) aperture of the lif1 ( guiding ) channel for each observation by the standard guide - star acquisition procedure .
the two observations resulted in 13 exposures totaling 21.3 @xmath27 of on - target exposure time .
approximately 95% of the observing time occurred during orbital night , which greatly reduced the amount of terrestrial and airglow entering the fuse apertures .
flux was recorded through the lwrs apertures in both long wavelength ( lif , @xmath28 ) channels and both short wavelength ( sic , @xmath29 ) channels .
the astigmatic heights of the lif1 spectra near 1030 were roughly 1/3 of the aperture width , consistent with the compact photometric structure of ngc 1705 .
images at 2200 show that roughly 40% of the light that would be admitted into the @xmath0 lwrs aperture would come from the single brightest star cluster ngc 1705 - 1 , and 70% of the light would come from the region within a radius of 2.8 arcsec of ngc 1705 - 1 ( meurer et al 1995 ) .
a comparison of the hst @xmath30 ( 1.7 arcsec aperture ) and @xmath31 ( 10 by 20 arcsec aperture ) spectra implies that these fractions should not be strongly wavelength dependent .
thus , while the @xmath0 data sample the entire inner region of ngc 1705 ( projected size of roughly 900 by 900 pc ) , the sightlines within about 80 pc of ngc 1705 - 1 are very strongly weighted . ]
the data are preserved in the @xmath0 archive with observation identifications a0460102 and a0460103 .
the raw time - tagged photon event lists for each exposure were processed with the standard @xmath0 calibration software ( @xmath32 ) available at the johns hopkins university as of august 2000 .
the lists were screened for valid data , and corrections for geometric distortions , spectral motions , and doppler shifts were applied ( see sahnow et al .
the data were not affected by the detector event bursts that plagued many of the early fuse spectra .
the 13 individual calibrated extracted spectra for each channel were cross - correlated , shifted to remove residual velocity offsets due to image motion in the apertures , and combined to a produce composite spectrum for each channel .
these composite spectra in the 10001070 region were then compared , and any remaining velocity offsets were removed by referencing to the lif1 channel data .
the wavelengths were put into the local standard of rest ( lsr ) reference frame by requiring that the galactic ism lines fall at @xmath33 km s@xmath5 , the approximate velocity of the milky way lines as observed at longer wavelengths with the space telescope imaging spectrograph ( stis ) on the @xmath34 .
these new stis data will be presented in future papers ( sembach et al .
2001 ; heckman et al 2001 ) .
the @xmath0 data have a wavelength uncertainty of @xmath35 km s@xmath5 ( @xmath36 ) .
the spectrum of ngc 1705 is shown in figure 1 . in this article
, we do not combine the data from all four channels because the instrumental resolution and sensitivity changes as a function of wavelength .
therefore , when measuring the strengths of absorption features , we compare the individual measurements ( w@xmath37 ) for the two channels having the highest sensitivity at the wavelengths of the absorption features of interest ( usually lif1 and lif2 for @xmath38 , or sic1 and sic2 for @xmath39 ) .
the integrated equivalent widths derived from separate channels generally agree very well .
table 1 contains these measurements for selected milky way and ngc1705 absorption features observed along the sight line .
the data have a velocity resolution of @xmath40 km s@xmath5 and s / n ratios of 16 , 13 , 10 , and 9 per resolution element at 1032 in the lif1 , lif2 , sic1 , and sic2 channels , respectively .
-5 cm -35 mm
we confirm the principal conclusion of sahu & blades ( 1997 ) and sahu ( 1998 ) : there are three separate systems of interstellar absorption - lines along the sight - line to ngc 1705 ( table 1 ; figure 2 ) .
these arise in the milky way , the periphery of the high velocity cloud hvc 487 , and the interstellar medium of ngc 1705 . in the present paper
, we will focus our analysis on ngc 1705 , but briefly summarize our results on hvc 487 . as pointed out by sahu & blades ( 1997 ) , hvc 487
is located 2@xmath41 away from the ngc 1705 sight - line .
sahu ( 1998 ) proposes that hvc 487 is associated with the magellanic stream , so the implied impact parameter would be roughly 2 kpc .
the properties of the absorption associated with hvc 487 will be discussed in detail in a future paper which will combine the @xmath0 data with echelle spectra taken with stis on @xmath23 ( sembach et al .
2001 ) . here
, we note only that hvc 487 is strongly detected in absorption in the @xmath81032 , @xmath42977 , @xmath431036 , and lyman - series lines . assuming that the @xmath81032 line is optically - thin , the implied @xmath11 column is [email protected] @xmath2 10@xmath45 @xmath46 .
this is typical of other hvc sight - lines studied with @xmath0 ( sembach et al .
the lyman series , ciii@xmath22977 , and cii@xmath221036 absorption - lines from hvc 487 are at @xmath47 = 270@xmath4415 km s@xmath5 .
these are close to the @xmath4821 cm velocity of hvc 487 of 232@xmath4414 km s@xmath5 ( sahu 1998 ) .
in contrast , the @xmath81032 line is centered at @xmath47 = 326@xmath4410 km s@xmath5 , an offset of 94@xmath4417 km s@xmath5 from the @xmath4821 cm velocity .
the @xmath11 line is very broad , with an observed fwhm = 100@xmath4415 km s@xmath5 .
this breadth is comparable to the sound speed in coronal gas where @xmath11 would be abundant , and is similar to those of other @xmath11 hvcs observed with @xmath0 ( sembach et al .
2000 ) .
we expect our @xmath0 data to probe four phases of interstellar gas in ngc 1705 : molecular gas , neutral atomic gas , warm gas photoionized by hot stars , and collisionally - heated coronal gas .
the neutral atomic gas is traced by species with ionization potentials of creation @xmath49 1 ryd , the warm gas by ions with @xmath50 = 1 to 4 ryd ( up to the heii edge ) , and the coronal gas by ions with @xmath514 ryd .
no molecular hydrogen was detected in any of the first three @xmath52 levels ( 0 - 2 ) , and the limit on the total @xmath53 column density is @xmath54 14.6 ( corresponding to a molecular hydrogen fraction of @xmath55 - based on the observed hi column given below ) .
similar results have been reported for the metal - poor dwarf starburst izw18 ( vidal - madjar et al .
the intrinsic reddening in ngc 1705 is very small : heckman et al .
( 1998 ) give @xmath56 0.1 . in the particular case of ngc 1705 ,
the low values for @xmath57 and @xmath58 are quite consistent with galactic sightlines with similarly - low @xmath59 and @xmath60 ( savage et al .
1977 ) .
lines tracing the other three phases are well - detected ( table 1 ; figures 1 and 3 ) .
in particular , the @xmath81032 line is independently detected at the 10@xmath61 level in both the lif1 and lif2 detectors ( figure 2 ) .
the weaker member of the doublet ( @xmath81038 ) is blended with the much stronger galactic @xmath621039 line and is not convincingly detected .
the profiles shown are from the lif1 channel , except for ciii @xmath63 , oi@xmath64 , and nii@xmath65 , which are from the sic2 channel
. data of roughly comparable quality exists for the lif2 and sic1 channels ( see table 1 ) .
the identifications of the lines are listed underneath each spectrum .
additional lines arising in the ism or within ngc 1705 within the velocity range shown are also indicated ( without wavelengths ) . in some cases , lines from various species blend and can not be resolved .
the vertical dashed lines indicate the lsr velocities of the ism ( 20 km s@xmath5 ) , hvc 487 ( @xmath66 km s@xmath5 compared to the @xmath4821 cm emission velocity of 232 km s@xmath5 ) , and the systemic velocity of ngc 1705 ( 622 km s@xmath5 ) .
note that the ngc 1705 absorption lines are substantially blue - shifted relative to the systemic velocity of the galaxy.,height=680 ] we begin by considering the kinematic properties of the gas . in what follows
, we will adopt an lsr systemic velocity for ngc 1705 of @xmath67 = 622 km s@xmath5 , which is derived from the hi@xmath2221 cm rotation curve ( meurer et al .
1998 ) .
the mean radial velocity of the neutral gas lines listed in table 1 is 590@xmath4411 km s@xmath5 as determined by fitting gaussian components to the observed absorption features .
this implies a blue - shift of 32 km s@xmath5 relative to @xmath67 , which is significantly larger than the uncertainty 11 km s@xmath5 derived from the standard deviation of the line centroids . in contrast , the coronal gas traced by @xmath11 is significantly more blueshifted : @xmath68 = 545@xmath4410 km s@xmath5 , or
@xmath69 = -77 km s@xmath5 . the warm photoionized gas appears to have intermediate velocities with the mean line centroid at 569@xmath4410 km s@xmath5 ( @xmath69 = -53 km s@xmath5)977 line , which is discussed separately ] . further evidence for differing dynamics between the neutral and coronal gas comes from the line widths ( see figure 3 ) .
the weak ( unsaturated ) lines from the neutral phase have observed lines widths ( fwhm ) of roughly 70 km s@xmath5 , while the @xmath81032 line has a broader structure ( fwhm @xmath4 100@xmath4410 km s@xmath5 ) .
the lines from the photoionized gas are roughly the same width as the @xmath11 line .
the strongly saturated @xmath42977 and @xmath431036 lines allow us to probe the low - column density gas at extreme radial velocities .
relative to @xmath67 , the centroid of the @xmath70 line is blueshifted by 95@xmath4420 km s@xmath5 . the blueward edge of the line is blended with the hvc feature , but absorption extends to at least -260 km
s@xmath5 with respect to @xmath67 . relative to @xmath67 , the @xmath431036 line centroid is blueshifted by 40@xmath4420 km s@xmath5 km s@xmath5 and absorption is present from at least -180 km s@xmath5 ( where the line is blended with the hvc feature ) to roughly + 70 km s@xmath5 .
the hi@xmath2221 cm maps of ngc 1705 published by meurer et al ( 1998 ) show emission from -94 to + 86 km s@xmath5 relative to @xmath67 .
the velocity range of the _ redshifted _ gas seen in @xmath431036 is thus consistent with the kinematics of the normal neutral ( turbulent ? ) ism , but an additional component of highly blueshifted ( outflowing ) neutral gas is also present .
we have determined column densities by converting the observed absorption profiles into optical depth profiles and integrating over velocity ( see savage & sembach 1991 ) .
we measured only the weaker ( and thus , more - optically - thin ) metal lines with equivalent widths @xmath71 250 m . the results are given in table 2 .
for the lines in the neutral phase , we estimate the following column densities ( log@xmath72 ) : @xmath73 ( 14.0 ) , @xmath74 ( 15.6 ) , @xmath75 ( 14.6 ) , @xmath76 ( 13.5 ) , and @xmath77 ( 14.5 ) .
heckman & leitherer ( 1997 ) estimated that @xmath78 = 1.5@xmath210@xmath79 @xmath46 column towards ngc 1705 - 1 based on fitting the red side of the damped @xmath80 profile in the hst @xmath30 data .
we have fit the ly@xmath81 and higher - order lyman series lines in our @xmath0 spectrum , and find @xmath78 = 2@xmath441 @xmath210@xmath79 @xmath46 ( @xmath825 km s@xmath5 ) .
we adopt the mean of these two estimates : log@xmath78 = [email protected] .
since these species above should be the dominant state of their respective element in the neutral gas , we can estimate the following abundances ( anders & grevesse 1989 ) : [ n / h]=-2.2 , [ o / h]=-1.5 , [ si / h]=-1.1 , [ ar / h]=-1.3 , and [ fe / h]=-1.2 .
the absolute abundances are uncertain by 0.3 dex . with the exception of nitrogen , the abundances are marginally consistent with the metallicity of [ o / h]=-0.9 derived for the nebular emission - line gas ( heckman et al .
this is in agreement with expectations that we see ambient interstellar gas , rather than the much hotter gas that is polluted by recent supernova ejecta .
we confirm sahus ( 1998 ) finding of a low relative @xmath72 abundance , based on the @xmath73 column density .
as sahu points out , the @xmath83 ion can also make a significant contribution to the total @xmath72 column in the @xmath84 phase ( see sofia & jenkins 1998 ) .
we estimate log@xmath85 = 14.7 .
unless most of this is associated with the @xmath84 phase , @xmath72 remains selectively underabundant .
in fact , most of the observed @xmath83 must be associated with the warm photoionized gas that dominates the total gas column along our line - of - sight ( see below ) .
this is implied by the kinematics of the @xmath861084 line , which has a significantly larger blueshift and fwhm than the lines arising in the neutral phase , but agrees with the other lines arisiing in the warm photoionized gas ( table 1 ; and see above ) .
a low nitrogen abundance is typical of low - metallicity gas in dwarf galaxies , and is a consequence of a partly secondary nucleosynthetic origin for @xmath72 ( e.g. van zee , haynes , & salzer 1997 ) .
we do not confirm the low relative @xmath87 abundance noted by sahu & blades ( 1997 ) .
finally , we see no evidence for the systematic gas - phase depletion of refractory elements ( e.g. compare @xmath87 to @xmath88 ) .
the neutral gas in the outflow thus appears largely dust - free ( e.g. , savage & sembach 1996 ) , consistent with the lack of detectable reddening in the uv spectrum of ngc 1705 ( e.g. , meurer , heckman , & calzetti 1999 ) .
the total column density in the warm photoionized gas is more uncertain .
the weak @xmath891012 and @xmath901063 lines imply column densities log@xmath91 = 15.0 and log@xmath92 = 14.4
. the @xmath93 ion can also be adundant in photoionized gas . while the hst @xmath30 spectrum covers a somewhat different sight - line ( the [email protected] arcsec large science aperture
was centered on ngc 1705 - 1 ) , the @xmath93 column density log@xmath94 = 15.0 from sahu & blades ( 1997 ) is a useful additional constraint .
the total implied column of ionized @xmath95 is log@xmath96 = 15.3 . for an assumed value [ s / h]=-0.9 ( based on nebular emission - line metallicity ) ,
the total @xmath97 column in the photoionized gas is log@xmath98 = 20.9 , or almost an order - of - magnitude greater than the @xmath84 column .
the observed @xmath81032 line yields an @xmath11 column density log@xmath99 = 14.3 . deriving the total column density of the coronal phase gas
requires uncertain assumptions . for gas in collisional ionization equilibrium
, @xmath11 reaches its peak relative abundance ( @xmath420% of the total @xmath100 abundance ) at @xmath1013@xmath210@xmath3 k ( sutherland & dopita 1993 ) . if we assume that [ o / h]=-0.9 in the coronal phase gas ( similar to the value in the optical emission - line gas ) , the implied minimum total @xmath97 column in the coronal gas is then log@xmath102 = 19.0 .
while uncertain , we note that this is a substantially smaller column than seen in the cooler gas .
in addition to the blueshifted @xmath11 absorption - line produced by coronal gas on the frontside of the outflow , redshifted @xmath11 _ emission _ from the backside of the outflow must also be present .
we do not detect such emission , but can set an interesting upper limit to its intensity . combining our lif1 and lif2 spectra
, we obtain a 3@xmath61 upper limit on the flux of the redshifted @xmath81032 emission - line of @xmath10310@xmath104 erg @xmath46 s@xmath5 .
this limit assumes that the line would have a breadth similar to the blueshifted absorption - line ( @xmath4 100 km s@xmath5 ) .
we first correct this flux for forground dust extinction .
the galactic @xmath84 column towards ngc 1705 is 1.3@xmath210@xmath79 @xmath46 , based on both the damped @xmath80 profile ( sahu 1998 ) and radio @xmath2221 cm observations ( reported in hensler et al .
1998 ) . adopting the standard extinction curve of mathis ( 1990 ) ,
the implied extinction at 1032 is 0.3 magnitudes .
the data discussed in section 3.2.2 imply that there is a negligible amount of dust extinction intrinsic to ngc 1705 . in the discussion below
, we will be interested in the total @xmath11 luminosity of ngc 1705 . to derive a global upper limit
, we make a simple aperture correction to our observed limit .
the portion of the backside of the ngc 1705 outflow included in our @xmath0 aperture represents about 15% of the total surface area of the outflow ( front and back sides ) , as mapped by meurer et al .
thus , the global extinction - corrected @xmath81032 flux from ngc 1705 is @xmath17 6.5@xmath210@xmath105 erg @xmath46 s@xmath5 , and the corresponding luminosity is @xmath17 3@xmath210@xmath106 erg s@xmath5 .
the simplest physical model that can be compared to our data is an adiabatic expanding superbubble whose expansion is driven by the energy supplied by multiple supernovae ( weaver et al .
1977 ; koo & mckee 1992 ) .
this is a natural model to use , since it appears to be a good first - order description of the properties of the @xmath107 emission line nebula in ngc 1705 ( meurer et al .
1992 ; marlowe et al .
1995 ) . in this simple model , there are six concentric zones .
from inside - out these are : @xmath108 an innermost region inside which energy is injected by supernovae ( the starburst ) .
@xmath109 a region of supersonic flow fed by the hot gas created in zone 1 .
@xmath110 a region of hot gas ( material from zone @xmath111 that has passed through an internal shock ) .
@xmath112 a conductive interface of intermediate - temperature gas created as the hot gas in zone @xmath113 heats the relatively cool , dense gas in zone @xmath114 . @xmath115 a thin dense shell of ambient gas that has been swept - up ( shocked and then radiatively cooled ) as the `` piston '' of hot gas in zone @xmath113 expands into the ism . depending on the available flux of ionizing radiation from the starburst ,
this shell may be partially or fully photoionized . @xmath116 the undisturbed ambient ism .
the dynamical evolution of a superbubble in the type of plane - parallel ism appropriate to a starburst has been extensively discussed ( see maclow & ferrara 1999 and references therein ) .
once the radius of the superbubble is several times the vertical scale - height of the ism , its expansion will accelerate and rayleigh - taylor instabilities will cause the outer shell ( zone @xmath114 ) to fragment
. this will allow the hot gas from the interior to escape from the ruptured superbubble and flow out into the galactic halo .
this `` blow - out '' or `` break - out '' stage marks the transition from superbubble to superwind .
before considering the superbubble model in more detail , it is worthwhile to make some rough estimates of the mass and kinetic energy in the gas and compare these to expectations . to convert the observed column densities and outflow velocities into masses and kinetic energy , we will assume the following idealized model for the ngc 1705 outflow .
meurer et al .
( 1992 ) show that the morphology and kinematics of the emission - line nebula can be described as the expansion of a hollow prolate ellipsoid with a semi - major axis of 1.5 kpc and semi - minor axes of 0.5 kpc .
multiplying the surface area of this ellipsoid by our estimated column densities ( see above ) implies total gas masses of 5@xmath210@xmath117 , 1@xmath210@xmath117 , and 6@xmath210@xmath3 @xmath14 respectively in the warm photoionized gas , the neutral gas , and the coronal gas .
the outflow speeds then yield corresponding kinetic energies of 1.6@xmath210@xmath118 , 1@xmath210@xmath119 , and 3@xmath210@xmath120 ergs .
the supernovae in the ngc 1705 starburst will supply kinetic energy at a mean rate of roughly 2@xmath210@xmath121 erg s@xmath5 ( marlowe , meurer , & heckman 1999 ) .
the dynamical age of the expanding emission - line nebula is 10 to 15 myr ( marlowe et al .
1995 ; meurer et al .
1992 ) , so the total amount of kinetic energy supplied by the starburst during this time is @xmath12210@xmath123 ergs . in the standard adiabatic superbubble model
, the kinetic energy of the swept - up shell ( zone @xmath114 ) will be about 19% of the total injected energy ( e.g. maclow & mccray 1988 ) .
given the nature of our estimates , the rough agreement between the observed and available / predicted kinetic energies is gratifying .
the amount of mass returned directly by supernovae and stellar winds during the last 10 - 15 myr will only be @xmath410@xmath124 ( leitherer & heckman 1995 ) .
this is tiny compared to shell s mass ( see above ) , consistent with the basic superbubble model in which the shell is swept - up ambient ism . in this case , the total gas column density seen by @xmath0 ( 10@xmath125 @xmath46 ) can be no larger than the total column density of the ism prior to the expansion of the superbubble .
the hi@xmath2221 cm maps in meurer et al .
( 1998 ) show that the mean column density in the inner part of ngc 1705 is about 2@xmath210@xmath125 @xmath46 , consistent with this requirement .
the similarity between the total gas column of outflowing gas measured by @xmath0 and the typical hi column in ngc 1705 implies that the dimensions of the superbubble must be at least comparable to the characteristic thickness of the ism .
this suggests that the superbubble is early in the blow - out stage of it dynamical evolution .
such an inference is also consistent with the patchy , filamentary morphology of the superbubble s h@xmath126 emission ( meurer et al .
1992 ; marlowe et al . 1995 ) , which suggests that the superbubble shell has begun to accelerate and fragment ( see above ) . the absorbing @xmath84 column in ngc 1705 ( log@xmath78 = [email protected] ) is roughly an order - of - magnitude smaller than the column observed in emission at @xmath2221 cm in the same region ( with the @xmath430 arcsec @xmath127 beam - meurer et al .
this implies that roughly 90% of the @xmath84 seen in the radio map lies _ behind _ the dominant source of far - uv light ( the region around the super star cluster ngc 1705 - 1 ) .
that is , we infer that the superbubble is not symmetrically located within the hi disk of ngc 1705 , but must be on the near side . thus , the blow - out is apparently one - sided at present ( cf .
maclow , mccray , & norman 1989 ) .
the column density of gas along our line - of - sight is dominated by the warm ionized gas .
this implies that the shell of swept - up material ( zone @xmath114 in the superbubble ) has been mostly photoionized by lyman continuum radiation from the starburst .
it is natural to identify the gas seen in absorption in the @xmath0 data with the emission - line nebula studied by marlowe et al .
( 1995 ) and meurer et al .
( 1992 ) .
the kinematics are consistent with this identification .
recall that the @xmath0 sight - line is heavily weighted towards material within about 3 arcsec of the bright super star cluster ngc 1705 - 1 .
the echelle spectra of this region analyzed by marlowe et al .
( 1995 ) and meurer et al .
( 1992 ) show double - peaked h@xmath126 emission - line profiles , with the blueshifted ( redshifted ) peak corresponding to emission from the front ( back ) side of the expanding superbubble .
the measured lsr radial velocity of the blueshifted component is 560@xmath4420 km s@xmath5 along the different position angles covered by their combined data sets . within the uncertainties , this agrees with the mean velocity for the corresponding @xmath0 absorption - lines ( 569@xmath4410 km s@xmath5 ) .
since we now have a mass for the warm photoionized gas ( 4.2 ) , its h@xmath126 luminosity ( marlowe et al .
1995 ) can be used to _
directly _ determine that the mean electron density in this material is @xmath128 1 @xmath129 .
we emphasize that this makes no assumptions about the volume filling - factor of the gas .
taking @xmath10110@xmath10 k ( typical of photoionized gas ) , the corresponding thermal pressure is p / k @xmath4 2@xmath210@xmath10 k @xmath129 .
can the starburst in ngc 1705 keep this gas photoionized ? taking @xmath128 1 @xmath129 , the measured radius of the superbubble ( @xmath4 500 pc ) , and the lyman continuum luminosity of ngc 1705 ( @xmath130 = 10@xmath120 s@xmath5 ) , the stromgren thickness of the photoionized supershell will be @xmath410@xmath125 cm and the column density of the ionized layer is then @xmath410@xmath125 @xmath46 .
this is in good agreement with the column density of photoionized gas we infer from the @xmath0 data .
we can compare the basic dynamical properties of the warm ionized gas in ngc 1705 to the predictions of the superbubble model .
in convenient units , the radius and expansion speed of the superbubble are given by : @xmath131 @xmath132 for an adiabatic superbubble inflated by a kinetic energy injection rate @xmath133 ( in units of 10@xmath121 erg s@xmath5 ) into a uniform medium with nucleon density @xmath134 for a time @xmath135 ( in units of 10@xmath117 years ) .
the age of the dominant super star cluster ( ngc 1705 - 1 ) is well - constrained to be about 10@xmath117 years ( de mello , leitherer , & heckman 2000 ) .
this agrees with the dynamical age ( @xmath136 ) of the superbubble ( marlowe et al .
1995 ; meurer et al . 1992 ) . the estimated kinetic energy injection rate due to supernovae is @xmath133 = 2 ( marlowe , meurer , & heckman 1999 ) . finally , the hi column in the center of ngc 1705 implies a mean nucleon density of @xmath1371 @xmath129 for an ism thickness of 1 kpc .
our estimate of the swept - up mass of gas in the supershell ( 5@xmath210@xmath117 @xmath14 ) divided by volume enclosed by the supershell ( @xmath138 @xmath139 ) likewise implies @xmath1371 @xmath129 .
equations 1 ) and 2 ) above then predict a radius of 0.8 kpc and an expansion velocity of 50 km s@xmath5 for the supershell .
given the overly - simplistic model ( spherical symmetry , constant density ) , we regard the agreement with the data as satisfactory .
the neutral gas probed by @xmath0 is significantly less blueshifted than the ionized gas , and must therefore have a different origin . for this gas to remain neutral
, it must be optically - thick to the lyman continuum radiation from the starburst .
this condition implies that a slab of gas with a column density log@xmath72 = 20.2 located inside the superbubble ( and thus at a distance @xmath140 500 pc from an ionizing source with a lyman continuum luminosity of @xmath141 s@xmath5 ) must have a density @xmath142 8 @xmath129 .
recall that the mean ism density in ngc 1705 is roughly 1 @xmath129 .
moreover , log@xmath72 = 20.2 and @xmath142 8 @xmath129 implies that the neutral absorbers must be very small ( smaller than a few pc ) .
thus , only relatively small dense clouds would be neutral .
on - average , the clouds are outflowing , so most can not be in any undisturbed ambient medium in front of the superbubble s outer shell .
we therefore identify the outflowing material with clouds in the superbubble s interior .
these were presumably clouds in the ism of ngc 1705 that were overtaken and engulfed as the superbubble propagated through the more tenuous inter - cloud medium .
this would be a natural consequence of a multi - phase ism ( e.g. , white & long 1991 ) .
the dynamical model here is one of dense clouds exposed to the outflowing hot gas in zones @xmath111 and @xmath113 of the superbubble .
we would expect to see absorbing material that is injected from quiescent material at or near @xmath67 , and which is then accelerated up to some terminal velocity as it is transported outward by the hot outflow .
following heckman et al .
( 2000 ) , an interstellar cloud with column density @xmath72 , originally located a distance @xmath143 from a starburst , will be accelerated by a spherically - symmetric outflow that carries an outward momentum flux @xmath144 up to a terminal velocity given by : @xmath145 we have chosen values for these parameters that are appropriate to the neutral gas in ngc 1705 .
the predicted velocity range of absorption relative to @xmath67 would be @xmath4 0 to -100 km s@xmath5 .
this agrees tolerably well with the velocities of the blueshifted neutral gas observed in our @xmath0 data .
the detection of the interstellar @xmath11 absorption - line ( the first in a starburst galaxy ) is the most important result in this paper .
we will organize our discussion of the coronal gas in ngc 1705 as follows .
we begin by making some simple inferences about the basic physical properties of the gas that are independent of any specific model for the origin of the gas ( 4.5.1 ) . in particular , in
4.5.2 we point out that - independent of the detailed thermal / dynamical history of the gas - the observed @xmath11 column density and line width imply that we are observing gas that has been heated to @xmath146 and which then cools radiatively .
these model - independent results can then be used to constrain the radiative cooling rate from the coronal gas and show that it is small compared to the supernova heating rate .
we then turn our attention to specific models for the @xmath11
. we will not exhaustively consider all the possible alternative models ( for example , conduction fronts associated with either the supershell fragments or the hi clouds discussed above ) .
instead we will describe the shortcomings of the standard superbubble model ( 4.5.3 ) , and then consider what we regard as an especially plausible alternative model in which the @xmath11 arises in a hydrodynamical interaction between hot outrushing gas and the dense , cooler fragments of the superbubble shell during the `` blow - out '' phase .
this idea is considered analytically in 4.5.4 in the context of models of turbulent mixing layers , and then using numerical hydrodynamical models in 4.5.5 .
the combination of a detection of blueshifted @xmath81032 in absorption and an upper limit to corresponding redshifted emission allows us to place an upper bound on the density and pressure in the coronal gas ( since we know the column density , and have an upper limit to the emission - measure ) .
this calculation only assumes reasonable symmetry between the front and back sides of the outflow averaged over the kpc - scale projected @xmath0 aperture . the @xmath11 ion is abundant only over a rather narrow temperature range ( e.g. sutherland & dopita 1993 ) , so we will assume @xmath9 3@xmath210@xmath3 k. the upper limit to the extinction - corrected @xmath81032 flux in the @xmath0 aperture corresponds to an upper limit to the @xmath11 intensity of @xmath14710@xmath148 photons @xmath46 s@xmath5 ster@xmath5 .
following shull & slavin ( 1994 ) , our observed @xmath11 column density then leads to an upper limit to the electron density in the coronal gas of @xmath149 0.1 @xmath129 .
the corresponding upper limit to the pressure is @xmath15010@xmath10 k @xmath129 .
we have estimated a pressure in the warm photoionized gas of @xmath151 2@xmath210@xmath10 k @xmath129 .
the pressure in the coronal - phase gas probed by the @xmath11 ion should be the same , since zones @xmath113 through @xmath114 in the adiabatic superbubble model are isobaric ( maclow & mccray 1988 ) .
we note that the isobaric condition could be invalidated by the presence of a dynamically - significant magnetic field ( in which case the thermal pressure in the coronal gas should be higher than in the denser photoionized material ) .
however , our model - independent upper bound of @xmath15010@xmath10 k @xmath129 in the coronal gas is only a factor of three higher than the pressure implied by an assumption of isobaric conditions . for @xmath9 3@xmath210@xmath3
k , our estimated pressure then implies that the characteristic density in the coronal gas is @xmath1523@xmath210@xmath153 @xmath129 . for a metal abundance of 1/8 solar in ngc 1705 ( heckman et al .
1998 ) the implied radiative cooling time from @xmath9 3@xmath210@xmath3 k is @xmath154 = 1@xmath210@xmath155 years ( interpolating between the collisional - ionization - equilibrium models for different metallicities in sutherland & dopita 1993 ) .
the assumption of collisional - ionization - equilibrium is probably reasonable for @xmath11 , since the relevant recombination times at our assumed density and temperature ( nahar 1999 ) are roughly an order - of - magnitude less than the radiative cooling time .
the estimated column density in the coronal gas in ngc 1705 is @xmath410@xmath156 @xmath46 , so the above density implies a characteristic thickness for the absorbing material of @xmath4 100 pc . if this path length is contributed by @xmath72 total clouds , the sound crossing time of a cloud is @xmath157 years .
thus , the ratio of sound - crossing and radiative cooling times in a cloud is @xmath158 and the assumption that the coronal gas is in rough pressure - balance with its surroundings is therefore plausible .
we have estimated that the total mass of coronal - phase gas is @xmath159 .
the above cooling time then implies a cooling rate of @xmath160 0.6 @xmath14 per year .
this can be compared to the average rate at which mass has been swept up by the superbubble over its lifetime : @xmath161 years @xmath4 5 @xmath14 per year .
the implied cooling luminosity is @xmath162 erg s@xmath5 .
this is about 5% of the estimated rate at which the starburst supplies mechanical energy . thus , radiative cooling associated with the coronal gas should not dominate the dynamical evolution of the outflow .
we note that radiative cooling from the hotter gas detected in soft x - rays ( @xmath163 erg s@xmath5 - hensler et al .
1998 ) is significantly smaller than our estimate of @xmath164 , and is therefore negligible .
edgar & chevalier ( 1986 ) have computed the expected column densities of various ions for the generic situation in which gas is heated to a temperature @xmath165k and then cools radiatively .
the total column density of cooling gas is just given by @xmath166 , where @xmath154 is the radiative cooling time and @xmath167 is the rate of cooling per unit area .
for a flow speed @xmath7 , mass conservation in the cooling flow implies @xmath168 .
thus , the cooling column can also be written at @xmath169 .
it is important to note that @xmath170 is independent of density , since @xmath171 .
moreover , at coronal temperatures the cooling is dominated by metals , so @xmath172 ( where @xmath173 is the metallicity ) . since @xmath174 , the characteristic @xmath11 column density in radiatively cooling coronal gas is essentially independent of density and metallicity and depends only on the value for @xmath168 .
edgar & chevalier ( 1986 ) calculate that @xmath175 @xmath46 ( @xmath7/100 km / s ) .
calculations of @xmath11 production in high - speed radiative shocks give similar values ( dopita & sutherland 1996 ) .
this fiducial column density agrees well with our measured values for @xmath99 and @xmath7 in ngc 1705 .
this good agreement has two immediate implications . first , independent of the detailed dynamical and thermodynamical history of the @xmath11 , this gas it is almost certainly the result of the radiative cooling of initially hotter gas .
second , we can now estimate the implied cooling rate . using the edgar & chevalier ( 1986 ) models ,
our observed value @xmath176 = 14.3 implies that @xmath177 = 5 @xmath2 10@xmath155 cm s@xmath5 . in their models
the gas cools from @xmath178 10@xmath155 k ( and we note that significantly higher values for @xmath179 in ngc 1705 are excluded by its low x - ray luminosity ) .
since we estimate @xmath180 = 2@xmath210@xmath10 k @xmath129 , it follows that @xmath134 = 0.01 @xmath129 at @xmath181k and thus @xmath167 = 5 @xmath2 10@xmath10 @xmath46 s@xmath5 . to calculate the implied rate at which gas is cooling ( @xmath182 ) , we take a surface area @xmath183 @xmath184 for the ngc 1705 superbubble ( see above ) , and find that @xmath185 per year . as the gas cools from 10@xmath155 k
its cooling luminosity is @xmath162 erg s@xmath5 .
this is about 5% of the estimated rate at which the starburst supplies mechanical energy , which agrees well with the more naive estimate in 4.5.1 above .
the above estimates of the cooling rate depend upon our assumed pressure .
however , the upper limit to the luminosity of @xmath81032 emission - line ( 3.2.3 ) yields a pressure - independent upper limit to the cooling rates of @xmath186 0.3 @xmath14 per year and @xmath187 4@xmath210@xmath15 erg s@xmath5 ( edgar & chevalier 1986 ) .
these limits are consistent with the above estimates . in the simple superbubble model
there are two plausible origins for the coronal gas .
first , if the speed of outer shock driven into the ambient ism is high enough , @xmath11 ions will be abundant behind this shock ( in zone @xmath114 ) .
this does not appear to be feasible in ngc 1705 , since the minimum required shock speed ( post - shock temperature ) is @xmath4 150 km s@xmath5 ( 3@xmath188 k ) - see shull & mckee ( 1979 ) and dopita & sutherland ( 1996 ) .
this is significantly higher than the expansion speed measured in for the shell in ngc 1705 : @xmath4 53 km s@xmath5 , corresponding to a post - shock temperature of only 40,000 k. second , thermal conduction will transfer heat from zone @xmath113 to zone @xmath114 and create coronal - phase gas at the interface ( zone @xmath189 ) .
weaver et al .
( 1977 ) predict an @xmath11 column density in this material of @xmath190 where @xmath191 is the oxygen abundance relative to solar . for @xmath191 = 1/8
, the predicted value for @xmath99 is a factor of @xmath415 smaller than we measure . in this case
, @xmath99 is smaller than the value for radiatively cooled gas ( 4.5.2 ) and has a direct dependence on the metallicity .
this is because cooling in the conductively heated zone is dominated by adiabatic expansion losses ( weaver et al .
1977 ) .
the relative velocity of @xmath11 absorption in ngc 1705 is also inconsistent with the simple superbubble model , which predicts that the outflow velocity in this material ( zone @xmath189 ) will be substantially less than the outflow speed of the superbubble shell ( zone @xmath114 ) .
our data instead show that the @xmath11 outflow speed is probably even larger than that of the shell ( 77@xmath4410 vs. 53@xmath4410 km s@xmath5 - see above ) .
we conclude that the simple superbubble model does not account for the observed properties of the @xmath11 absorption - line in ngc 1705 .
we have argued in
4.5.2 that the @xmath11 arises in a flow of gas that has been heated to an initial temperature of @xmath192 and which is then cooling radiatively .
as discussed in 4.5.3 , this heating / cooling process does not correspond to the outer shock in the superbubble ( zone 5 ) because the observed expansion velocity of the superbubble is too slow to produce @xmath11 behind this shock . in this section ,
we therefore consider a plausible alternative model .
we have argued above that that the superbubble in ngc 1705 is in the process of breaking out of the ism of ngc 1705 . during this phase ,
the expansion speed of the superbubble shell accelerates and rayleigh - taylor instabilities will cause the shell ( zone @xmath114 ) to fragment .
this allows the hot x - ray emitting gas in zone @xmath189 to push its way out through the fragmented shell , and kelvin - helmholtz instabilities are believed to lead to the turbulent mixing of this hot outrushing gas with the cooler shell fragments ( e.g. maclow , mccray , & norman 1989 ) .
slavin , shull , & begelman ( 1993 hereafter ssb ) have investigated the emission and absorption - lines produced by the intermediate - temperature ( coronal phase ) gas in these `` turbulent mixing layers '' ( tmls ) .
we emphasize that the kinematics of the @xmath11 absorbers in ngc 1705 are consistent with tmls .
the blueshift of the @xmath11 relative to the shell material is expected as the hot gas rushes out through `` cracks '' in the shell .
the magnitude of this blueshift follows from ssb : the velocity of the intermediate temperature gas in the tml relative to the cool gas from which it is created is given by @xmath193 , where @xmath194 is the relative flow speed between the gas at @xmath195 and the outrushing hot gas at @xmath196 . for the photoionized shell fragments that dominate the mass in ngc 1705 @xmath197 k. based on x - ray spectroscopy of dwarf starburst galaxies ( della ceca et al .
1996,1997 ; hensler et al . 1998 ) , @xmath198 few@xmath210@xmath155 to 10@xmath117 k. as this hot gas flows past the shell fragments , its maximum relative velocity will be roughly its sound speed ( i.e. @xmath199 500 km s@xmath5 ) .
thus , for @xmath200 10@xmath201 , @xmath20216 km s@xmath5 for @xmath194 = 500 km s@xmath5
. this can be compared with ngc 1705 where the @xmath11 absorption line is blueshifted by 24@xmath4410 km s@xmath5 relative to the photoionized shell material . as pointed out by ssb ,
the ionic column density per tml is independent of the pressure in the gas , but is directly proportional to the relative velocity of the cool and hot gas ( @xmath194 ) .
also the @xmath11 column will be appreciable only if the maximum temperature of the tml exceeds @xmath203 k. their models only extend up to @xmath194 = 100 km s@xmath5 and @xmath204 = 5.5 , and these predict @xmath11 columns that are of - order 10@xmath205 @xmath46 per tml .
these values could be adjusted upward in tmls with larger values for @xmath194 and @xmath204 . in terms of the arguments in 4.5.2 , the low @xmath11 columns in the tml models are partly attributable to the small relevant flow velocity in the tml : @xmath206 10 km s@xmath5 .
it appears that of - order 10@xmath207 tmls per line - of - sight are needed in ngc 1705 ( we see a ` sea' of shell fragments , each with its own tml ) .
proportionately fewer tmls would be needed for larger flow velocities . as an initial attempt to investigate qualitatively the dynamics of gas in the complex transitional stage between superbubble and superwind ,
we have performed several 2-dimensional hydrodynamical simulations of a superbubble blowing out of a dwarf galaxy .
we calculate the column densities and velocity structure of simulated lines of sight through these models , in particular the relative velocities of the warm and coronal phases that the @xmath0 observations of ngc 1705 show can not be explained by the classic superbubble model .
the simulations were performed using the hydrodynamical code described in strickland & stevens ( 2000 ) , with the initial conditions altered to roughly correspond to ngc 1705 .
these simulations were performed in cylindrical coordinates , assuming rotational symmetry around the @xmath208-axis ( the minor axis of the galaxy ) .
the hydrodynamical grid covers a region 2.5 kpc in radius by 2.5 kpc high along the @xmath208-axis with 500 by 500 equally sized cells .
the initial ism was set up in rotating hydrostatic equilibrium .
we use a king model for the gravitational potential , producing a rotational velocity at @xmath209 kpc of @xmath210 km s@xmath5 .
the ambient gas was given a temperature of @xmath211 k , to approximate turbulent pressure support of the ism , based on the @xmath212 km s@xmath5 velocity dispersion of the hi gas .
the resulting initial ism density distribution has a peak number density ( in the nucleus ) of @xmath213 @xmath129 , a total column density through the galaxy along the minor axis of @xmath214 @xmath46 , a vertical scale height of 0.2 kpc , and a total mass of @xmath215 ( within @xmath216 kpc ) .
these properties are all crudely similar to those observed in ngc 1705 ( meurer et al .
1998 ) .
we modeled the return of mass and mechanical energy into the ism from starburst event as a single instantaneous burst of star - formation , using the low metallicity models ( @xmath217 ) of leitherer & heckman ( 1995 ) .
appropriate amounts of mass and thermal energy were added at each computational time step to those cells within our assumed starburst region , a cylindrical region @xmath218 pc in radius and @xmath219 pc thick vertically .
our fiducial model has a time - average mechanical luminosity of @xmath220 erg s@xmath5 , somewhat weaker than the observed starburst ngc 1705 , but chosen to give blow - out at @xmath221 myr ( approximately the observed dynamical age of the superbubble in ngc 1705 ) . at blow - out
the radius of superbubble is @xmath222 pc in the plane of the galaxy , and about twice as large along the minor axis .
we assume an ism metal abundance of 1/10 solar for the purposes of calculating radiative cooling and column densities .
we also performed other simulations to crudely explore parameter space : first , a simulation with a more powerful starburst ( @xmath223 erg s@xmath5 ) , which blows out earlier at @xmath224 myr but otherwise is very similar to the fiducial model in other properties .
second , a model using solar abundance to investigate the scaling of column density with metal abundance .
third , a higher resolution model with twice the resolution in every dimension , to investigate the dependence of calculated properties on numerical resolution .
an example of the model is shown in figure 4 , which plots the density and temperature of the gas in our fiducial low - metallicity model at t = 12.5 myr .
the @xmath11 absorption seem in these simulations arises primarily in material at the interfaces between cool dense gas , e.g. in the shell of the superbubble or shell fragments after blow - out , and the hot ( @xmath18 k ) gas filling the bubble volume ( see figure 4 ) . in almost all cases
these interfaces are numerically unresolved .
the three panels show ( a ) log number density ( @xmath129 ) , ( b ) log gas temperature ( k ) and ( c ) the location of the @xmath11-absorbing material ( gas with @xmath225 log t ( k ) @xmath226).,height=264 ] column density , covering all lines of sight and all epochs in the three low metallicity numerical simulations ( solid line ) and the simulation using solar metallicity ( dashed line ) .
( b ) corresponding histograms of the mean @xmath11 absorption line velocity ( weighted by the @xmath11 column density).,height=264 ] we calculate the column density as a function of line - of - sight velocity profiles in each model at three different epochs ( pre- , during , and post- blow - out , typically at @xmath227 and @xmath228 myr respectively ) , for four different lines of sight from the nuclear starburst ( assuming galaxy inclination angles of @xmath229 & @xmath230 ) , and for two different ionic species that probe different observed phases in superwinds ( warm gas : @xmath83 ; coronal gas : @xmath11 ) .
fractional ion abundances as a function of temperature were taken from sutherland & dopita ( 1993 ) , which assume collisional ionization equilibrium .
this a reasonable assumption in the case of @xmath11 ( see above ) .
it is also a reasonable mock - up of @xmath83 ( in reality , photoionization by stars in ngc 1705 keeps the gas containing @xmath83 near @xmath9 10@xmath10 , close to the temperature floor of 1.6 @xmath210@xmath10 k in the simulations ) .
much of the uv light in ngc 1705 comes from the compact central star cluster , but a significant fraction comes from a more diffusely spread component ( meurer et al 1995 ) .
the width of @xmath231 fuse aperture used corresponds to a projected width of @xmath232 pc at the distance of ngc 1705 , so the observed spectrum is a sum over many slightly different lines of sight . to investigate the effect of this on the simulated absorption line profiles we explored two cases , which should bracket the real situation . in the first case we assumed
all the light comes from a single point - like source at the center of the galaxy , and in the second case that the light comes from uniform surface brightness disk of radius @xmath233 pc .
we found very little difference in the simulated absorption line profiles between the two methods , and so quote only results using the point - like light source .
for any particular model , the column densities and kinematics of the gas vary strongly between different sight lines , due to the structural complexity that arises in 2-d simulations ( see suchkov et al .
1994 , strickland & stevens 2000 ) .
the total column densities ( in either @xmath11 or @xmath83 ) do not change systematically with time in these simulations .
we find that the mean and variance in total column density is the same before , during and after blow - out , considering all 12 simulated lines of sight at each specific epoch .
as shown in figure 5 , we find @xmath11 column densities in the range @xmath234 in the low metallicity models ( @xmath235 ) , and @xmath236 in the solar metallicity model . only about 10% of the lines - of - sight in the low metallicity models have @xmath11 columns comparable to the observed value in ngc 1705 .
one reason for the failure of these simulations to reproduce the observed ovi column densities may be that there is insufficient numerical resolution to resolve the cooling zones between hot and cool gas , where the ovi - absorbing gas exists . evidence to support
this idea comes from comparing the predicted column densities in the fiducial model with those in the high resolution model .
although the predicted columns for low ionisation species always agree to within a factor 2 , the predicted ovi column densities agree to within a factor 2 in only 50% of the sight lines , with the remaining sight lines differing by a factor of up to 30
. the lack of thermal conduction in these models may also be partially responsible for the low ovi columns , although the arguments presented in
4.5.3 suggest thermal conduction alone can not produce the observed ovi columns .
our simulations do show that the column density of lower ionization material is considerably higher than that of coronal material , and that this phase contains most of the mass in the outflow ( consistent with our observations ) .
the column - weighted mean velocities for the coronal gas along the different sight - lines generally lie in the range @xmath237 to @xmath238 km s@xmath5 , although the full range covers @xmath239 to @xmath240 km s@xmath5 ( figure 5 ) .
this gas is typically more blue shifted , and has a wider range of outflow velocities , than the cooler gas in our simulations .
for example , the typical velocity range for @xmath83 is @xmath241 to @xmath238 km s@xmath5 and the full range seen only covers @xmath242 to @xmath243 km s@xmath5 .
the velocity range covered appears very similar in both low and solar metallicity models , so it appears that the kinematics of the gas in these simulations is independent of the metal abundance . for @xmath11 , the mean outflow velocity ( and range of velocities ) ,
does appear to increase with time , moving from pre - blow - out through to post - blow - out .
this is not the case for lower ionization species , e.g. the distribution of mean velocity for @xmath83 is very similar at all three epochs . in summary ,
while the numerical models underpredict the @xmath11 columns , they _ can _ explain one aspect of the observed data on ngc 1705 .
the mean outflow velocity of high - ionization gas probed by @xmath11 is as high as , or higher than , that of low - ionization gas .
this is in contrast with the standard superbubble model in which the outflow speed in the coronal gas ( zone @xmath189 ) is less than that of the superbubble shell .
this result supports the idea that blow - out is responsible for the kinematics of the coronal phase in ngc 1705 .
as summarized in 1 , probably the greatest uncertainty in assessing the impact of starburst - driven outflows on the evolution of dwarf galaxies is the importance of radiative cooling ( which will directly determine the outflow s dynamical evolution ) . in the absence of severe radiative cooling ,
both analytic arguments and hydrodynamical simulations imply that outflows like the one in ngc 1705 will easily blow - out of their galaxy s ism ( e.g. , maclow & mccray 1988 ; marlowe et al .
1995 ; martin 1998,1999 ; maclow & ferrara 1999 ; strickland & stevens 2000 ) .
the hot gas that blows out of the superbubble will then be able to heat and chemically enrich the inter - galactic medium ( e.g. , ponman et al .
1999 ; tozzi , scharf , & norman 2000 ) and/or an extended gaseous halo surrounding the dwarf galaxy ( silich & tenorio - tagle 1998 ) .
we will not repeat these arguments here , but simply note that within the context of the standard mckee & ostriker ( 1977 ) formalism , the lack of radiative cooling would be a result of the exceedingly high supernova rate per unit volume , which leads to a porosity near unity in the ism ( i.e. , the ism within which the supernovae detonate is filled almost entirely with hot , diffuse gas ) .
this is especially pertinent to ngc 1705 , in which nearly half of the supernovae will detonate inside the pc - scale super - star - cluster ngc 1705 - 1 .
existing x - ray data establish that radiative cooling from hot ( @xmath244k ) gas is energetically insignificant in ngc 1705 ( hensler et al .
1998 ) and other starburst galaxies ( cf .
heckman 2000 and references therein ) .
the new data from @xmath0 are very instructive , since they directly probe for the coronal - phase ( @xmath245 k ) gas near the peak of the cooling curve where hitherto - unobserved radiative cooling might be severe . as we showed in 4.5.2 above
, the observed @xmath11 column density ( @xmath246 @xmath46 ) and flow speed ( @xmath13 km s@xmath5 ) implies that we are seeing gas in the outflow that has been heated to @xmath247 k and then radiatively cooled .
we have estimated that the implied rate of radiative cooling is only @xmath46% of the rate of supernova heating ( 4.5.2 ) .
thus , there is no place left to hide the radiation necessary to quench the supernova - driven outflow in ngc 1705 .
what then is the fate of the outflowing gas ? for an isothermal gravitational potential that extends to a maximum radius @xmath248 , and has a virial velocity @xmath249 , the escape velocity at a radius @xmath250 is given by : @xmath251^{1/2}\ ] ] the rotation curve for ngc 1705 implies @xmath249 = 62 km s@xmath5 ( meurer et al . 1998 ) , and the rotation curve remains relatively flat out to the last measured point at @xmath250 = 7 kpc .
thus , the minimum escape velocity at the present location of the superbubble wall ( @xmath252 0.5 kpc ) will be roughly 170 km s@xmath5 .
this is significantly higher than the velocities of most of the outflowing gas that we observe in - absorption in ngc 1705 .
thus , there is no direct evidence that most of this material will escape ngc 1705 .
maps of the hi in ngc 1705 ( meurer et al .
1998 ) show that the large - scale ism in ngc 1705 will not be strongly affected by the outflow ( which will only puncture a few - kpc - scale hole in the central region of the hi disk ) .
on the other hand , the bulk of kinetic energy and most of the newly - created metals will reside in the hotter gas that is driving the superbubble s expansion .
the temperature of this gas will be @xmath253 k ( della ceca et al .
1996,1997 ; hensler et al .
we have argued that we are witnessing the blow - out phase in ngc 1705 in which this gas is vented through the fragmented superbubble wall .
following blow - out , in the absence of significant radiative cooling this hot gas will feed a galactic wind with a terminal velocity of at least 300 km s@xmath5 ( chevalier & clegg 1985 ) .
this is comfortably above @xmath254 for any plausible value for @xmath248 .
once the hot gas vents out of the ruptured superbubble , there is no known external medium to impede the resulting high - speed wind .
we conclude that ngc 1705 can eject most of the newly - created metals and kinetic energy returned by the central starburst over the past @xmath4 10@xmath117 years . if typical of dwarf starbursts , this process could help create the low metal abundances in dwarf galaxies ( e.g. , dekel & silk 1986 ; lyden - bell 1992 ) and contribute to the heating and chemical enrichment of the intergalactic medium ( ponman et al .
1999 ; gibson et al . 1997
; ellison et al .
we have presented new @xmath0 far - uv spectroscopy of the prototypical dwarf starburst galaxy ngc 1705 .
previous optical and uv spectroscopy has established that the starburst in this galaxy is driving a large - scale outflow ( meurer et al .
1992 ; marlowe et al .
the @xmath0 data are especially important because they probe the coronal - phase ( @xmath9 10@xmath3 to 10@xmath155 k ) gas that may dominate the radiative cooling of the outflow , and thereby largely determine its dynamical evolution . the high quality of the data ( spectral resolution of @xmath4 30 km s@xmath5 and a @xmath255 16 ) gives us important new insight into all the phases in the ism of this galaxy .
firstly , we do not detect any @xmath53 , with an upper limit of @xmath256 14.6 .
this is similar to the results obtained by vidal - madjar et al .
( 2000 ) for the dwarf starburst izw18 . in the specific case of ngc 1705
the low @xmath53 abundance is consistent with galactic sight - lines with similarly low @xmath59 ( savage et al . 1977 ) .
we do detect absorption from three other phases of the ism : neutral gas , warm photoionized gas , and coronal - phase gas ( probed with the @xmath81032 line ) .
the total column densities ( @xmath46 ) in the three phases are @xmath257 = 20.2 , 20.9 , and @xmath25819.0 respectively .
the first is directly measured from lyman series , while the latter two are estimated from the measured ionic columns assuming a metal abundance of 12% solar ( as measured in the optical emission - line gas ) .
all the interstellar absorption - lines associated with ngc 1705 are broad ( @xmath4 10@xmath207 km s@xmath5 ) and blueshifted .
thus , gas is flowing out of ngc 1705 with @xmath259 = -32 , -53 , and -77 km s@xmath5 for the neutral , photoionized , and coronal gas respectively . the mass and kinetic energy in the outflowing gas probed by @xmath0 is dominated by the warm photoionized gas , which is also seen via its optical line - emission .
the size , morphology , outflow speed , mass , kinetic energy , and dynamical age of this structure are all consistent with a simple model of an adiabatic superbubble whose expansion is driven by a piston of hot gas created by the cumulative effect of the supernovae in the starburst ( e.g. , weaver et al . 1977 ) . in contrast , we have shown that the neutral gas must reside in relatively small dense clumps in order to be optically - thick in the lyman continuum .
we attribute the outflowing material to clouds in the ism that have been overtaken by the superbubble and accelerated by the outflowing hot gas .
the properties of the coronal gas probed with @xmath81032 are inconsistent with the predictions of the simple superbubble model . the observed expansion speed of the superbubble ( 53@xmath4410 km s@xmath5 ) is too small to produce @xmath11 behind its shock - front .
the blueshift of the @xmath11 line is larger than that of superbubble shell ( while the model predict that it should be substantially smaller ) . finally , the observed @xmath11 column ( @xmath176 = 14.3 ) is a factor of @xmath415 too large compared to the superbubble model ( in which such gas arises in a conductively - heated and adiabatically - cooled interface between hot gas and the cool outer superbubble shell ) .
we therefore proposed the following origin for the @xmath11 absorber , and have investigated its plausibility via numerical hydrodynamical models .
the morphology and relative size - scale of the superbubble implies that it has begun to `` break out '' or `` blow out '' of the ism in the disk of this galaxy . during the blow - out phase ,
the superbubble shell will accelerate and then fragment ( e.g. , maclow & mccray 1988 ) .
as it fragments , the piston of hot gas inside the superbubble will flow out through the `` cracks '' , and the resulting hydrodynamical interaction between this outrushing gas and the shell fragments will create intermediate - temperature coronal gas that can produce the observed @xmath11 absorption ( e.g. , slavin et al . 1993 ) .
this process accounts nicely for the observed kinematics of the @xmath11 absorption .
we also emphasized that for the observed flow speed of @xmath13 km s@xmath5 , the observed @xmath11 column density is just what is expected for gas that has been collisionally - ionized and then radiatively - cooled , independent of its density or metallicity and independent of any specific hydrodynamical model ( edgar & chevalier 1986 ) .
we have argued that the coronal gas should be in rough pressure - balance with the warm photoionized gas ( whose pressure we determined to be @xmath260 k @xmath129 ) .
this allowed us to estimate cooling rates of @xmath40.07 @xmath14 per year and @xmath410@xmath15 erg s@xmath5 in the coronal gas , based on the @xmath11 absorption - line .
independent of the gas pressure , the lack of redshifted @xmath11 _ emission _ from the backside of the outflow implies upper limits to the cooling rates of @xmath2610.3 @xmath14 per year and @xmath262 erg s@xmath5 .
the rate of radiative cooling due to hotter ( @xmath244 k ) gas is negligible in comparison ( hensler et al .
1998 ) . since the rate of supernova - heating in ngc 1705 is @xmath263 erg s@xmath5 ( marlowe et al . 1999 ) , we have argued that radiative cooling is not dynamically important in the ngc 1705 outflow . in the absence of significant radiative cooling ,
the superbubble in ngc 1705 should be able to blow out of the ism .
based on the relatively low outflow velocities , it is unlikely that the absorption - line gas we observe with @xmath0 will escape from the galaxy .
likewise , the bulk of the global ism ( @xmath84 ) in ngc 1705 will be retained ( cf .
de young & heckman 1994 ; maclow & ferrara 1999 ) .
however , we have argued that the hotter gas that drives the outflow ( and which is now being vented out of the ruptured superbubble ) can escape and thereby carry away most of the kinetic energy and newly - created metals supplied by the starburst .
this process has potentially important implications for the evolution of both dwarf galaxies and the igm .
we thank the members of the @xmath0 team for providing this superb facility to the astronomical community .
we thank colin norman and robin shelton for enlightening discussions , and an anonymous referee for a detailed and thoughtful report that improved the paper .
this work was supported in part by nasa grants nag5 - 6400 and nag5 - 9012 .
dks is supported by nasa through _ chandra _ postdoctoral fellowship award number pf0 - 10012 , issued by the _
chandra _ x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 39073 .
anders , e. , & grevesse , n. 1989 , geochim .
acta . , 53 , 197 chevalier , r. , & clegg , a. , 1986 , nature , 317 , 44 dekel , a. , & silk , j. 1986 , apj , 303 , 39 della ceca , r. , griffiths , r. , & heckman , t. 1997 , apj , 485 , 581 della ceca , r. , griffiths , r. , heckman , t. , & mackenty , j. 1996 , apj , 469 , 662 de mello , d. , leitherer , c. , & heckman , t. 2000 , apj , 530 , 251 de young , d. , & heckman , t. 1994 , apj , 431 , 598 dopita , m. , sutherland , r. 1996 , apjs , 102 , 161 edgar , r. , & chevalier , r. 1986 , apjl , 310 , l27 ellison , s. , songaila , a. , schaye , j. , & pettini , m. 2000 , aj , 120 , 1175 gibson , b. , loewenstein , m. , & mushotzky , r. 1997 , mnras , 290 , 623 heckman , t. 1998 , in origins , asp conf .
series , 148 , 127 heckman , t. 2000 , astro - ph/0009075 heckman , t. , & leitherer , c. 1997 , aj , 114 , 69 heckman , t. , robert , c. , leitherer , c. , garnett , d. , & van der rydt , f. 1998 , apj , 503 , 646 heckman , t. , lehnert , m. , strickland , d. , & armus , l. 2000 , apjs , 129 , 493 heckman , t. , sembach , k. , leitherer , c. , martin , c.l .
, meurer , g. , de mello , d. 2001 , in preparation hensler , g. , dickow , r. , junkes , n. , & gallagher , j. 1998 , apjl , 502 , l17 kinney , a. , bohlin , r. , calzetti , d. , panagia , n. , & wyse , r. 1993 , apjs , 86 , 5 koo , b. , mckee , c. 1992 , apj , 388 , 93 leitherer , c. , & heckman , t. 1995 , apjs , 96 , 9 lynden - bell , d. 1992 , in elements and the cosmos , ed .
m. edmunds & r. terlevich ( cambridge : cambridge univ . press ) , 270 maclow , m. , & mccray , r. 1988 , apj , 324 , 776 maclow , m. , & ferrara , a. 1999 , apj , 513 , 142 maclow , m. , mccray , r. , & norman , m. 1989 , apj , 347 , 141 marlowe , a. , heckman , t. , wyse , r. , & schommer , r. 1995 , apj , 438 , 563 marlowe , a. , meurer , g. , & heckman , t. 1999 , apj , 522 , 183 martin , c. 1998 , apj , 506 , 222 martin , c. 1999 , apj , 513 , 156 mathis , j. 1990 , ara&a , 28 , 37 mckee , c. , & ostriker , j. 1977 , apj , 218 , 148 meurer , g. , freeman , k. , dopita , m. , & cacciari , c. 1992 , aj , 103 , 60 meurer , g. , heckman , t. , leitherer , c. , kinney , a. , robert , c. , & garnett , d. 1995 , aj , 110 , 2665 meurer , g. , stavely - smith , l. , & killeen , n. 1998 , mnras , 200 , 705 meurer , g. , heckman , t. , & calzetti , d. 1999 , 521 , 64 moos , h.w .
2000 , apj , 538 , l1 morton , d. 1991 , apjs , 77 , 119 nahar , s. 1999 , apjs , 120 , 131 pettini , m. , kellogg , m. , steidel , c. , dickinson , m. , adelberger , k. , & giavalisco , m. 1998 , 508 , 539 pettini , m. , steidel , c. , adelberger , k. , dickinson , m. , & giavalisco , m. 2000 , apj , 528 , 96 ponman , t. , cannon , d. , & navarro , j. 1999 , nature , 397 , 135 sahnow , d.s . ,
2000 , apj , 538 , l7 sahu , m. 1998 , aj , 116 , 1205 sahu , m. , & blades , j.c .
1997 , apjl , 484 , l12 savage , b. , & sembach , k. 1991 , apj , 379 , 245 savage , b. , & sembach , k. 1996 , ara&a , 34 , 279 savage , b. , bohlin , r. , drake , j. , & budich , w. 1977 , apj , 216 , 291sembach , k. , & savage , b. 1992 , apjs , 83 , 147 sembach , k. et al . 2000 , apjl , 538 , l31 sembach , k. , heckman , t. , leitherer , c. 2001 , meurer , g. , martin , c.l .
, & de mello , d. 2001 , in preparation shull , j.m . , & mckee , c. 1979 , apj , 227 , 131 shull , j.m . , & slavin , j.d .
1994 , apj , 427 , 784 slavin , j. , shull , j.m .
, & begelman , m. 1993 , apj , 407 , 83 steidel , c. , giavalisco , m. , pettini , m. , dickinson , m. , & adelberger , k. 1996 , apjl , 462 , l17 silich , s. , & tenorio - tagle , g. 1998 , mnras , 299 , 249 sofia , u. , & jenkins , e. 1998 , apj , 499 , 951 strickland , d. , & stevens , i. 2000 , mnras , 314 , 511 suchkov , a. , balsara , d. , heckman , t. , & leitherer , c. 1994 , apj , 430 , 511 sutherland , r. , & dopita , m. 1993 , apjs , 88 , 253 tenorio - tagle , g. , silich , s. , kunth , d. , terlevich , e. , & terlevich , r. 1999 , mnras , 309 , 332 tozzi , p. , scharf , c. , & norman , c. 2000 , apj , 542 , 106 van zee , l. , haynes , m. , & salzer , j. 1997 , aj , 114 , 2497 vidal - madjar , a. et al .
2000 , apjl , 538 , l77 weaver , r. , mccray , r. , castor , j. , shapiro , p. , & moore , r. 1977 , apj , 218 , 377 white , r. , & long , k. 1991 , apj , 373 , 543 lrrrrr h2 q(1 ) & 1009.770 & @xmath1635(3@xmath61 ) & @xmath1635(3@xmath61 ) & ... & ... + c ii & 1036.337 & 510@xmath4430 ( l1 ) & @xmath264680 ( l1)c & + 18 & + 582 + & & 492@xmath4422 ( l2 ) & @xmath264670 ( l2)c & & + c iii & 977.020 & 560@xmath4430 ( s2 ) & @xmath264800 ( s2)b & + 28 & + 555 + & & 520@xmath4435 ( s1 ) & @xmath264700 ( s1)b & & + n i & 1134.165 & 121@xmath4410 ( l1 ) & 23@xmath4410 ( l1 ) & + 28 & + 582 + & & 134@xmath4410 ( l2 ) & 24@xmath4410 ( l2 ) & & + n i & 1134.415 & 155@xmath4410 ( l1 ) & 32@xmath4410 ( l1 ) & + 24 & + 590 + & & 156@xmath4410 ( l2 ) & 23@xmath4410 ( l2 ) & & + n i & 1134.980 & 171@xmath4411 ( l1 ) & 35@xmath4411 ( l1 ) & + 25 & + 602 + & & 161@xmath4410 ( l2 ) & 35@xmath4410 ( l2 ) & & + n ii & 1083.990 & 356@xmath4425 ( s2 ) & 338@xmath4435 ( s2 ) & + 7 & + 573 + o i & 1039.230 & 260@xmath4420 ( l1)a & 243@xmath4415 ( l1 ) & + 14 & + 592 + & & 252@xmath4420 ( l2)a & 237@xmath4415 ( l2 ) & & + o i & 988.773 & 302@xmath4425 ( s2)a & 480@xmath4430 ( s2 ) & + 10 & + 595 + & & 283@xmath4430 ( s1 ) & 540@xmath4445 ( s1 ) & & + o vi & 1031.926 & 252@xmath4422 ( l1 ) & 171@xmath4418 ( l1 ) & + 25 & + 545 + & & 220@xmath4422 ( l2 ) & 191@xmath4419 ( l2 ) & & +
si ii & 1020.699 & 129@xmath4416 ( l1 ) & 94@xmath4424 ( l1 ) & + 16 & + 589 + & & 104@xmath4412 ( l2 ) & 65@xmath4420 ( l2 ) & & + s iii & 1012.502 & 143@xmath4420 ( l1)b & 250@xmath4415 ( l1 ) & + 8 & + 578 + & & 128@xmath4420 ( l2)b & 215@xmath4420 ( l2 ) & & + s iv & 1062.662 & @xmath1636 ( 3@xmath61 ) & 85@xmath4415 ( l1 ) & ... & + 555 + & & @xmath1636 ( 3@xmath61 ) & 82@xmath4415 ( l2 ) & & + ar i & 1048.220 & 76@xmath4410 ( l1 ) & 60@xmath4410 ( l1 ) & + 20 & + 566 + & & 83@xmath4410 ( l2 ) & 80@xmath4420 ( l2 ) & & + fe ii & 1063.176 & 125@xmath4411 ( l1 ) & 162@xmath4416 ( l1 ) & + 11 & + 597 + & & 140@xmath4413 ( l2 ) & 177@xmath4421 ( l2 ) & & + fe ii & 1144.938 & 192@xmath4412 ( l1 ) & 262@xmath4417 ( l1 ) & + 20 & + 601 + & & 206@xmath4411 ( l2 ) & 241@xmath4417 ( l2 ) & & + h2 ( j=0 ) & & & @xmath1614.2 & @xmath1614.2 & @xmath1614.2 + h2 ( j=1 ) & & & @xmath1614.2 & @xmath1614.2 & @xmath1614.2 + h2 ( j=2 ) & & & @xmath1614.2 & @xmath1614.2 & @xmath1614.2 + h i & 1025.7 & 1.909 & 20.2 & 20.2 & [email protected] +
n i & 1134.1 & 1.182 & @xmath1614.17 & @xmath1614.30 & @xmath1614.30 + & 1134.4 & 1.483 & @xmath1614.08 & @xmath1614.07 & @xmath1614.07 + & 1135.0 & 1.660 & 13.97 & 13.98 & [email protected] + n ii & 1084.0 & 2.048 & & 14.70 & [email protected] + o i & 1039.2 & 0.980 & 15.64 & 15.63 & [email protected] + o vi & 1031.9 & 2.137 & 14.26 & 14.26 & [email protected] + si ii & 1020.7 & 1.460 & 14.68 & 14.51 & [email protected] + s iii & 1012.4 & 1.556 & 15.02 & 15.00 & [email protected] + s iv & 1062.7 & 1.628 & 14.45 & 14.43 & [email protected] + ar i & 1048.2 & 2.408 & 13.44 & 13.49 & [email protected] + fe ii & 1144.9 & 2.084 & 14.53 & 14.54 & [email protected] + & 1063.2 & 1.805 & 14.54 & 14.55 & [email protected] + | we report @xmath0 far - uv spectroscopy of the prototypical dwarf starburst galaxy ngc 1705 .
these data allow us for the first time to directly probe the coronal - phase ( @xmath1 = few @xmath210@xmath3 k ) gas that may dominate the radiative cooling of the supernova - heated ism and thereby determine the dynamical evolution of the starburst - driven outflows in dwarf galaxies .
we detect a broad ( @xmath4100 km s@xmath5 fwhm ) and blueshifted ( @xmath6@xmath7 = 77 km s@xmath5 ) @xmath81032 absorption - line arising in the previously - known galactic outflow . the mass and kinetic energy in the outflow we detect
is dominated by the warm ( @xmath9 10@xmath10 k ) photoionized gas which is also seen through its optical line - emission .
the kinematics of this warm gas are compatible with a simple model of the adiabatic expansion of a superbubble driven by the collective effect of the kinetic energy supplied by supernovae in the starburst . however , the observed properties of the @xmath11 absorption in ngc 1705 are _ not _ consistent with the simple superbubble model , in which the @xmath11 would arise in a conductive interface inside the superbubble s outer shell .
the relative outflow speed of the @xmath11 is too high and the observed column density ( @xmath12 14.3 ) is much too large .
we argue that the superbubble has begun to blow out of the ism of ngc 1705 . during this blow - out phase the superbubble shell accelerates and fragments .
the resulting hydrodynamical interaction as hot outrushing gas flows between the cool shell fragments will create intermediate - temperature coronal gas that can produce the observed @xmath11 absorption . for the observed flow speed of @xmath13 km s@xmath5 ,
the observed @xmath11 column density is just what is expected for gas that has been heated and which then cools radiatively .
assuming that the coronal - phase gas is in rough pressure balance with the warm photoionized gas , we estimate a cooling rate of - order @xmath40.1 @xmath14 per year and @xmath410@xmath15 erg s@xmath5 in the coronal gas .
the latter represents @xmath1610% of the supernova heating rate .
independent of the assumed pressure , the lack of observed redshifted @xmath11 emission from the backside of the outflow leads to upper limits on the cooling rate of @xmath17 20% of the supernova heating rate . since the x - ray luminosity of ngc 1705 is negligible , we conclude that radiative - losses are insignificant in the outflow
. the outflow should therefore be able to fully blow - out of the ism of ngc 1705 and vent its metals and kinetic energy .
this process has potentially important implications for the evolution of dwarf galaxies and the igm . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.