article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
although it is well understood that two - dimensional axisymmetric models are important for studying the hadley circulation ( schneider , 1977 ; held and hou , 1980 ) , this kind of model has not been thought useful for large - scale motions in the extratropics , since non - axisymmetric baroclinic waves are essential in the real atmosphere . until the beginning of this century
, however , meridional cellar structure in an axisymmetric situation had been a major problem of the general circulation of the atmosphere ( e.g. , thomson , 1892 ; lorenz , 1967 ) : not only the hadley circulation but also the cellar structure in the extratropics .
views of the general circulation has changed from that given by hadley(1735 ) in which only a direct cell exists in each hemisphere ( upward motion in the equatorial side and downward motion in the polar side ) to that given by ferrel(1856 ) and thomson(1892 ) in which a direct cell co - exists with an indirect cell ( downward motion in the equatorial side and upward motion in the polar side ) .
although various cellar structures had been presented , the problem of the cell shapes was almost settled by the requirement of the angular momentum balance ; the direct cell must override the indirect cell in the axisymmetric circulations , and the cell boundary is inclined to the pole as in the upper layers .
this cellar structure is presented by bjerknes(1937 ) and eliassen(1951 ) .
the numerical experiments by schneider(1977 ) and held and hou(1980 ) have qualitatively reproduced this structure .
satoh(1994a ) has calculated the axisymmetric circulations of a moist atmosphere by using a two - dimensional model in the global domain ( hereafter , referred to as s94 ) .
the surface temperature is externally fixed and the free atmosphere is cooled by the gray radiation model .
large - scale convective motions are naturally resolved in the model and the stratification is stabilized , even if no cumulus parameterization except for the large - scale condensation is introduced .
the results show that , in the low - latitudes , there exists a large - scale direct cell which corresponds to the hadley cell .
the cell widths are broader as in the upper layers , so that the cell shape is very similar to that presented by bjerknes(1937 ) and eliassen(1951 ) . in the extratropics , however , unsteady multiple cells exist instead of an indirect cell and they propagate toward the equator ( s94 , fig .
3 ; reproduced as fig .
2(c ) of the present paper ) .
hereafter , the multiple cells calculated in s94 in the mid- and high - latitudes are referred as the symmetric cells .
the circulations in the mid - latitudes of s94 are different from those obtained by schneider(1977 ) and held and hou(1980 ) .
the difference may be ascribed to the maintenance mechanism of the stratification of the models . in s94
, the stratification tends to be unstable because of the radiative cooling in the troposphere and the heating from the surface .
convective motions must be generated in order to suppress the unstable stratification .
in contrast to this , schneider(1977 ) and held and hou(1980 ) used a stable temperature profile as a reference state of the newtonian cooling , so that no local convection exist in their models . in the extratropics of the real atmosphere ,
non - axisymmetric motions such as baroclinic waves are dominant , and the symmetric cells calculated in s94 is hardly observed . as will be shown in sections 2 and 3 , however
, the symmetric cells generally exist in all experiments of the two - dimensional model , and play an important role in the transport of the angular momentum .
it is required that dynamical mechanisms of the symmetric cells should be clarified , in particular , for their propagation and horizontal scales , to gain a complete view of the axisymmetric systems .
the symmetric cells are also valuable , since they will provide a further insight into the hadley circulation .
it will be shown that the hadley cell can be regarded as one of the symmetric cells . in this paper ,
the circulations in the mid - latitudes of the axisymmetric atmospheres are focused . a two - dimensional model in horizontal - vertical section
is utilized to investigate the dynamics of the symmetric cells .
two kinds of coordinates are used ; the model in the spherical coordinate is referred as the global model , and the model in the cartesian coordinate is referred as the @xmath3plane model .
formulation of the model is described in the appendix . in section 2 ,
the angular momentum budget of the standard experiment of s94 is analyzed , and additional results for the global model are shown . in sections 3 and 4 ,
the symmetric cells are discussed with the results for the @xmath3plane model .
section 5 summarizes the relationship between the symmetric cells and the hadley circulations and discuss the generalization of these findings .
the angular momentum budget plays a key role in meridional circulations of the axisymmetric atmosphere .
when steady direct and indirect cells co - exist as depicted by bjerknes(1937 ) and eliassen(1951 ) , the cell shapes are determined by the requirements of the angular momentum balance in the following manner .
the direct cell transports the angular momentum upward , and the indirect cell transports it downward , so that there must be an exchange of the angular momentum between both cells . in the axisymmetric system , the viscous stress is the only process of the momentum exchange between steady cells .
if coefficients of the viscosity are isotropic , horizontal transport caused by the viscous stress is much smaller than vertical transport by the viscous stress , since the aspect ratio of the atmosphere is very small .
thus , the angular momentum is vertically transported by the viscous stress which is associated with the vertical shear of the zonal winds . in order for the zonal shear to be maintained , the direct cell must override the indirect cell , and the cell boundary
is inclined to the pole as in the upper layers . in the beginning , we examine the angular momentum budget of the results for the global model of s94 , and compare it with the above consideration . in the standard experiment of s94 , the surface temperature is given by @xmath4 where @xmath5 is the latitude , @xmath6 k , and @xmath7 k. the diffusion coefficient of momentum flux is 5 m@xmath8/s , and that of heat and moisture fluxes is 1 m@xmath8/s . in the standard experiment , the hadley cell exists in the low - latitudes , and a multiple cellar structure which propagates equatorward exists in the mid- and high - latitudes . in order to isolate transports by the symmetric cells in the mid- and high - latitudes
, we divide convergence of the angular momentum into steady and unsteady parts : @xmath9 @xmath10 is the absolute angular momentum , @xmath11 is the meridional velocity , and @xmath12 is the divergence in the meridional plane .
@xmath13 denotes time average and @xmath14 the deviation from the average .
1 shows the respective terms in eq .
( [ eq : lcnv ] ) . the total component ( a ) shows a large divergence ( gain of the angular momentum in the atmosphere ) in the lower boundary layers of the hadley cell . at the polar boundary of the hadley cell ( the latitudes 10@xmath1520@xmath15 ) , divergent and convergent regions are vertically adjacent to each other .
both regions are inclined to higher latitudes in the upper layers .
the steady component ( b ) is uniformly divergent in the lower layers in every latitude , while it is convergent in the upper layers in the extratropics .
conversely , the unsteady component ( c ) is divergent in the upper layers , while it is convergent in the lower layers .
1 indicates that it is the symmetric cells that transport the angular momentum downward in the mid - latitudes .
the angular momentum is supplied to the atmosphere in the surface easterly regions in the low - latitudes , and is transported upward and poleward by the hadley cell .
it is transferred downward by the viscous stress at the polar boundary of the hadley cell , where a large vertical shear of the westerlies exists . in the mid- and high - latitudes , westerly shear
is maintained by the poleward transport of the angular momentum by a weak steady direct cell , which has a hemispheric scale as shown by fig .
on the other hand , the unsteady symmetric cells transport the angular momentum downward in these regions .
although both easterlies and westerlies exist at the surface boundary of the symmetric cells , the westerly component is a little stronger , hence the angular momentum is transferred from the atmosphere to the ground due to the net effect of the surface friction .
the role of the hadley cell in the angular momentum budget is the same as that depicted by bjerknes(1937 ) and eliassen(1951 ) . in the extratropics , however , the symmetric cells play a role of the downward transport of the angular momentum instead of a steady indirect cell .
additional experiments of the global model are performed , so as to investigate the dependencies of the symmetric cells on the surface temperature differences @xmath16 and the viscosity coefficients @xmath17 . fig .
2 shows the time sequences of the latitudinal distribution of the precipitation for @xmath18 0k , 20k , 40k ( the standard experiment ) , and 80k .
the temperature at the equator is @xmath19 300k for every case . in the standard experiment
, the symmetric cells are propagating equatorward at a speed of approximately 1 m / s , and have a period of about 20 days .
the speed of the propagation becomes faster as @xmath0 becomes larger . in the case of @xmath1 ,
the precipitation patterns are periodic and they do not change their positions except for the equatorial latitudes .
these results imply that the gradient of the surface temperature and the coriolis parameter affect the propagation of the symmetric cells .
it should be noted that , in the case of @xmath20 0 ( fig .
2(a ) ) , precipitation is not concentrated at the equator but randomly distributed in the low - latitudes .
that is , the hadley cell no longer exists in this model , when the surface temperature gradients are sufficiently small near the equator .
3 displays the dependencies of the precipitation on the viscosity coefficients of the free atmosphere ( @xmath17 ) and of the surface layer ( @xmath21 ): @xmath22 1 and 25 m@xmath8/s . in the case of @xmath22 25 m@xmath8/s , the symmetric cells around the latitudes 30@xmath15
do not propagate .
this result indicates the propagation of the symmetric cells is related with the angular momentum transport , since the viscous diffusion controls the vertical momentum transport .
in this section , an `` @xmath3-plane model '' is used to investigate the symmetric cells in the mid - latitudes ; the model is two - dimensional ( horizontal - vertical ) with a uniform rotation rate in the cartesian coordinate . as summarized in the appendix ,
the formulation of the model , i.e. the basic equations , the discretization , and the physical processes , is almost the same as that of the global model . only the metrics and the domain are different .
the model is based on the primitive equations with a @xmath23 coordinate ( @xmath24 ; @xmath25 the pressure and @xmath26 the surface pressure ) .
the lateral widths of the domain is @xmath27 m. here , the lateral coordinate is denoted by @xmath28 , which corresponds to the latitude of the global model ; larger @xmath28 indicates the northward direction .
the lateral boundaries are assumed to be walls with slip and insulating conditions .
the number of grid points is 100 in the horizontal direction and 50 in the vertical direction .
the grid intervals are @xmath29 0.02 and @xmath30 0.02 , respectively .
the coriolis parameter is prescribed as @xmath31 rad / s , which is the value at the latitude 30@xmath15 ( @xmath32 is the rotation rate of the earth ) . as for the radiation process , a uniform cooling rate 2k / day is given below the level @xmath33 to the ground just for simplicity .
the level @xmath33 can be regarded as the tropopause .
the surface temperature is externally fixed , and the ground surface is assumed to be wet everywhere .
moisture is supplied from the surface to the atmosphere by the bulk formula . in the same way as section 3a of s94 , only the `` large - scale condensation '' is employed as the moist process in the atmosphere , and any other cumulus parameterization is used ; phase change of moisture is explicitly calculated at each of the grid point , and the liquid part of the air is removed as precipitation when it is saturated .
diffusion coefficients for the momentum , heat , and moisture are set to be constant as in s94 .
the surface temperature is linearly prescribed as @xmath34 the temperature difference in the domain is denoted by @xmath16 .
4 shows time variations ( 0 100 days ) of precipitation for ( a ) @xmath35 300 k , ( b ) @xmath19 305 k , @xmath36 295 k , ( c ) @xmath19 310 k , @xmath36 290 k , and ( d ) @xmath19 320 k , @xmath36 280 k. in every case , there exists a systematic cellar structure similar to that in the mid - latitudes of the global model .
the precipitation occurs in the upward motion regions of the symmetric cells .
when there is no temperature difference ( @xmath20 0 ) , the precipitation patterns have a periodicity of about 20 days .
when there is a temperature difference ( @xmath37 0 ) , the precipitation patterns propagate to the warmer regions ( smaller @xmath28 ) . as
the temperature difference becomes larger , the velocity of the propagation becomes larger . although the lateral intervals between the precipitating regions are not sensitive to @xmath0 , they are larger in the warmer regions .
the periodicity or the propagation of the precipitation is associated with the circulations of the symmetric cells .
structure and mechanism of the symmetric cells will be examined in the following subsections .
the symmetric cells in the case @xmath20 0 ( fig .
4(a ) ) are examined here .
5 shows the time variation of the surface zonal winds for the days 0 100 .
westerlies exist in the southern side of the precipitating regions , whereas easterlies exist in the northern side .
their phases correspond to those of the precipitation .
6(a ) and ( b ) show the zonal winds in the meridional plane ( @xmath28-@xmath23 ) for the averages of the days 4655 and the days 5665 , respectively .
7(a ) and ( b ) are the temperature perturbation from the @xmath28-average for the same periods .
the phase of the precipitation changes at about the day 55 .
for instance , precipitation occurs at @xmath38 0.52 during the days 4655 , and it jumps to @xmath38 0.41 and 0.62 at the day 55 . at the precipitating regions , the circulations of the lower layers are cyclonic with convergence of lateral winds , while those of the upper layers are anticyclonic with divergence of lateral winds .
the temperatures in the upward motion regions are warmer than those in the downward motion regions irrespective of height .
the temperatures are almost in thermal wind balance with the zonal winds .
the periodicity of the symmetric cells can be explained as follows .
associated with the convective circulation , cyclonic winds develop in the lower layers , and anticyclonic winds develop in the upper layers . in the northern part of a precipitating region ( upward motion region ) ,
easterlies develop in the lower layers while westerlies develop in the upper layers . as the convective motion continues ,
the westerlies in the upper layers gradually come down . by the time they reach the ground , the lateral gradient of pressure changes its sign due to the geostrophic adjustment to the westerlies .
the inverted pressure gradient force causes lateral winds of the opposite direction near the ground .
finally , the circulations are reversed ; the downward motions are generated in the previously upward motion region , and upward motions are generated in the previously downward motion region .
the period is given by the overturning time @xmath2 , which is defined by @xmath39 where @xmath40 is the height of the troposphere and @xmath41 is the speed of the downward motion .
( this is also estimated by eq .
( 29 ) of s94 . )
the lateral scale of the symmetric cells @xmath42 can be estimated in the following manner .
note that there are two cells between adjacent precipitating regions , so that the lateral interval of the precipitating regions is expressed by @xmath43 . if one denotes the lateral temperature difference in the middle of the troposphere by @xmath44 , the thermal wind balance is expressed by @xmath45 where @xmath46 is
the maximum value of the zonal wind at the tropopause @xmath40 , and @xmath47 is the mean temperature . by estimating @xmath48 from the angular momentum conservation at the height @xmath40 , one obtains @xmath49 where @xmath50 is the acceleration of gravity .
the value of @xmath44 should be known in order to obtain @xmath42 .
as shown below , only an upper bound of @xmath44 can be given from the consideration of the stratification .
however , it will be shown that @xmath44 is actually close to the upper bound .
if one supposes a thermal wind balance of a large @xmath42 , @xmath44 also becomes larger as @xmath42 is increased . in this case , the temperature profile in the upward motion regions does not change , since it is prescribed by a moist adiabat .
thus , the temperature in the downward motion regions must be cooler in this balance .
since the surface temperature is uniform , the stratification of the downward motion regions becomes unstable .
this implies that another convective motion will occur from these regions .
although the downward motion regions are , in general , conditionally unstable , they would be absolutely unstable in the balance of a sufficiently large @xmath42 . therefore , the upper bound of @xmath44 is given by that in the case when temperature lapse rate in the downward motion regions is equal to the dry adiabat .
8 displays a time variation of lapse rate ( @xmath51 ) at @xmath52 which is just above the mixed layers .
the values above 8 k / km are shown by hatching .
comparison with fig .
4(a ) shows that the lapse rate is about 9k / km at the downward motion regions when the circulation changes its direction .
this lapse rate is close to the dry adiabat @xmath53 9.8k / km ( @xmath54 is specific heat for constant pressure ) . from this
, one may conclude that the temperature difference @xmath44 is given by this upper bound .
the temperature difference is given by @xmath55 where @xmath56 is the lapse rate of the moist adiabat .
one obtains , therefore , by using eq .
( [ eq : l_f ] ) and ( [ eq : deltat_f ] ) , @xmath57^{1/2 } = \frac{nh}{f } \equiv r_{n } , \label{eq : l_f2}\ ] ] where @xmath58 is the brunt visl frequency for the moist adiabatic profile . in conclusion ,
the lateral scale is given by the rossby deformation radius @xmath59 .
it is estimated as @xmath60 for the values in the case of fig .
4(a ) : @xmath61 [ sec@xmath62 , @xmath63 [ m ] and @xmath64 [ sec@xmath65 . fig .
4(a ) shows that the lateral interval of the precipitating regions is approximately 0.18 @xmath66 ; that is , @xmath67 , which is very close to the above estimation .
one might suspect that the result that the scale of the convective cell is given by the rossby deformation radius is only a trace of linear theories ( e.g. , gill , 1982 ; emanuel , 1983 ; bretherton , 1987 ) .
however , the present case is considered in the nonlinear statistically equilibrium states , to which the linear theories are not directly applied . the nonlinear advection of the angular momentum and the asymmetry between the lower and the upper boundary conditions for the momentum ( the angular momentum is exchangeable at the lower surface ,
but it is not at the upper boundary ) are key points of the above derivation .
if , instead , one applies the slip conditions in which momentum exchange is forbidden both at the lower and the upper boundaries , the angular momentum will be horizontally uniform in the whole layer around the cumulus region in the nonlinear equilibrium states . in this case ,
the vertical component of the absolute vorticity becomes zero , so that the circulation field associated with cumulus heating will not feel the basic rotation @xmath3 and the scale of the convective cell will be larger than the rossby deformation radius . in the present study , however , the circulation field can feel the basic rotation through the angular momentum exchange at the lower surface , and , as a result , the rossby deformation radius becomes a natural scale for the convective cell . as an example for @xmath37 0 , the wind fields in the case of @xmath19 320 k and @xmath36
300 k are examined .
9 shows a time variation ( the days 0 100 ) of the precipitation .
the wind fields are averaged for 50 days along the propagating precipitating region a @xmath68 b of fig . 9 ( a : @xmath38 0.68 at the day 51 , b : @xmath38 0.45 at the day 100 ) .
10 ( a ) , ( b ) show the zonal winds and the stream functions , respectively .
the line a @xmath68 b corresponds to @xmath38 0.5 of fig .
10 . in the lower layers ,
the northern side of the precipitating region ( @xmath38 0.5 ) is westerly , while the southern side is easterly . in the upper layers ,
contrary to this , the northern side of @xmath38 0.5 is easterly , while the southern side is westerly .
the region of the upper westerlies is connected with that of the surface westerlies ( @xmath38 0.6 0.66 ) , which are associated with the inflow to the next precipitating region ( @xmath38 0.66 ) . as shown by the stream functions , each precipitating region has both northern and southern cells . the northern cell is broader than the southern cell especially in the upper layers , and the northern cell spreads over the next southern cell which is associated with the next precipitating region ( @xmath38 0.66 ) .
that is , the direct cell ( positive sign of the stream functions ) is in contact with the indirect cell ( negative sign ) by an inclined boundary ( the contour of zero ) .
the angular momentum is transported upward by the direct cell , and is transported downward by the indirect cell , hence the angular momentum should be transferred from the direct cell to the indirect cell .
the inclination of the cell boundary makes possible the vertical transfer between the cells by the viscosity .
this cellar structure is similar to the relation between the hadley cell and the ferrel cell depicted by eliassen(1951 ) , and is essential for the angular momentums transport between the two cells .
11 shows the temperature and the normalized angular momentum @xmath69 at the day 100 .
the upper angular momentum and temperature is uniform in the regions @xmath38 0.22 0.4 and @xmath38 0.41 0.64 , for example .
these uniform regions are associated with the precipitation at @xmath38 0.24 and 0.46 , respectively ( fig .
the lateral gradient of the temperature is larger as in the lower layer , and the temperature is in the thermal wind balance . in this respect ,
the structure of the symmetric cells is very similar to that of the hadley cell ; the angular momentum is conserving in the upper layers and the temperature is balanced with the vertical shear ( compare with figs . 2 and 4 of s94 ) .
as for the relationship with the hadley cell , the symmetric cells in the @xmath3-plane model should be investigated in terms of the argument by plumb and hou(1992 ) , in which the symmetric circulations are classified into two regimes : a non - meridional circulation regime in thermal wind balance with the prescribed heating ( a thermal equilibrium ( te ) solution ) and a meridional circulation regime with an angular momentum conserving flow in the upper layer ( an amc solution ) .
they argued that the response takes the form of the te solution if the thermal forcing is weak or broad or it is far from the equator ( eq .
( 8) and appendix of plumb and hou ) .
although the newtonian cooling is used as the thermal forcing in plumb and hou , the similar argument is partly applied to the present study if their reference temperature @xmath70 is replaced by the surface temperature @xmath71 . in the @xmath3-plane model ,
( 8) of plumb and hou corresponds to @xmath72{t_{s}}{y } < f. \label{eq : te}\ ] ] if this condition is satisfied , the angular momentum balanced with the surface temperature does not have an extremum in the interior of the atmosphere and the te solution is regular provided . plumb and hou further argued that the amc solution can not exist when the te solution is regular provided .
if the surface temperature is linearly distributed ( eq .
( [ eq : ts_f ] ) ) , the left hand side of ( [ eq : te ] ) vanishes and the inequality holds .
however , it should be pointed out that the proof in their appendix is insufficient for the @xmath3-plane model , since they implicitly assumed that two hadley cells coexist : the summer cell and the winter cell .
their relation ( a13 ) , which is derived from this assumption , does not hold in the present situation .
thus , the possibility of the amc solution is not precluded even if ( [ eq : te ] ) is satisfied .
in contrast to plumb and hou , s94 derived the condition for the amc solution and the latitudinal scale of the hadley cell from the relation between the surface temperature @xmath71 and the temperature in the middle layer @xmath73 which is in the thermal wind balance with the angular momentum conserving flow ( figs .
11 and 13 of s94 ) . from this
, the amc solution can exist when @xmath73 is higher than @xmath71 around the upward motion region .
this condition is rewritten in terms of the gradients : @xmath74 the temperature deviation in the middle layer from that at the upward motion region of the amc solution is given by @xmath44 in eq .
( [ eq : l_f ] ) . replacing @xmath42 by the distance from the upward motion region @xmath75 and @xmath44 by @xmath76 , one obtains @xmath77 from this ,
the right hand side of ( [ eq : amc ] ) becomes zero as @xmath78 0 , so that the amc solution can be possible if there is a non - zero surface temperature gradient .
if one denotes the deviation of @xmath71 from that at the upward motion region by @xmath79 , the relation between @xmath80 and @xmath81 gives the scale of the cell @xmath42 as shown by fig .
12(b ) . by equating the change of @xmath81 in @xmath42 with that of @xmath80
: @xmath82 one obtains with eq .
( [ eq : t_amc ] ) , @xmath83 this indicates that @xmath42 increases with @xmath0 .
the scale of the cell @xmath42 corresponds to the cell width of the northern side of the precipitating region ; the cell width is defined by the upper spread of the stream functions of the cell .
( remind that @xmath42 is not the interval of the precipitating regions . in the case of @xmath20 0 , the cell width is @xmath42 whereas the interval of the precipitating regions is @xmath43 . ) for @xmath84 k and @xmath85 270k , eq .
( [ eq : l_delt_f ] ) estimates @xmath86 , which is much smaller than the cell width shown by fig .
furthermore , eq . ( [ eq : l_delt_f ] ) approaches zero as @xmath87 and contradicts eq .
( [ eq : l_f2 ] ) .
one of the reasons for this underestimation is that the effect of static stability , which is important in the case of @xmath20 0 , is not taken into account in the derivation of eq .
( [ eq : l_delt_f ] ) . as shown by fig .
11 , the temperatures in the middle layers are maximum around @xmath88 where the precipitation occurs , and the temperatures are minimum near @xmath38 0.40 and 0.64 .
the northern and the southern cells exist in the regions between these minimum and maximum temperatures ; the northern cell spreads in the region @xmath38 0.46 0.64 , and the southern cell spreads in the region @xmath38 0.40 0.46 . as in the case of @xmath20 0 , the static stability is smaller around the minimum temperatures .
one can take the effect of the static stability on the cell scale into account by simply adding eq .
( [ eq : deltat_f ] ) to eq .
( [ eq : deltat_f2 ] ) ( fig .
12(c ) ) : @xmath89 which yields with eq . ( [ eq : t_amc ] ) , @xmath90 from this , one obtains @xmath91 where @xmath92 is the width of the northern cell and @xmath93 is the width of the southern cell .
if one notes @xmath94 then the interval of the precipitating regions is given by @xmath95 , \label{eq : l3_approx } \end{aligned}\ ] ] which is not sensitive to @xmath0 .
this result is consistent with fig .
4 , in which the intervals of the precipitating regions are almost the same regardless of @xmath0 : 0.18 @xmath66 in the case of @xmath20 0 and 0.2 @xmath66 in the case of @xmath20 40k . the reason for the propagation of the symmetric cells is suggested as follows . in the case of @xmath20 0 ,
as shown in the previous subsection , the symmetric cells are periodic rather than propagating .
it has been argued that , if the upper angular momentum descends in the northern side , zonal winds become westerlies , and that these westerlies lead to the change of direction of the circulation and hence the periodicity . in order for the symmetric cells to preserve their circulation structure , the surface winds must be easterly in the northern side of the precipitating region .
the zonal winds can be easterly only if the symmetric cells propagate southward during the overturning period , since the upper angular momentum descends to a more southern place than the place where it is supplied ( this is schematically shown by fig . 13 ) .
the above consideration suggests that , as a convenient estimation , the upper bound of the speed of the propagation of the symmetric cells is given by @xmath96 , where @xmath2 is the overturning time and @xmath42 is the cell length .
this corresponds to the ultimate case in which the angular momentum is conserved ( fig .
the value of the angular momentum is reduced by the viscous transfer with the adjacent indirect cell in the northern side , and it is also preferable for the easterly .
the viscosity also affects the speed as shown by fig .
3 , though this effect is not easy to be taken into account in the simple scaling argument .
it should be noted , however , that the above consideration does not tell which is the preferable direction for the propagation .
the experiments show that the symmetric cells always propagate southward ( toward the warmer regions ) .
the direction of the propagation seems to be controled by the requirement of the surface easterlies .
one can suppose a basic state that has a westerly shear in thermal wind balance with the prescribed surface temperature gradient .
the upper westerly does not change its direction if it descends at the same @xmath28 . only if it descends southward
, it becomes easterly .
there must be both easterlies and westerlies in the surface zonal winds on the constraint of the angular momentum conservation .
the symmetric cells should move southward as a whole in order to cancel the basic westerly component and to produce surface easterlies by the advection .
if they move northward , westerlies would be stronger at the surface .
in this section , the symmetric cells in the case of @xmath37 0 are examined in terms of the symmetric instability ( si ) ( e.g. , stone , 1966 ; emanuel , 1979 , 1982 ) .
it should be noted that it is impossible in the strict sense to say which instability is the origin of the symmetric cells , since the symmetric cells of the present study are disturbances in the statistically equilibrium states of a long time integration .
stability of a prescribed initial field and time evolution of the field are mainly concerned in the instability problems .
when one relates disturbances in the equilibrium states to some kind of instability , one should appropriately choose a reference state that is thought to be an initial state for the instability .
a reference state is sometimes defined by an idealistic balanced state of forcings prescribed to the systems under consideration .
an alternative definition of a reference state is given by a simple average ( in time or in space ) of the equilibrium states . in the following
, relation between the symmetric cells and instabilities is considered according to the two kind of the reference states .
forcings imposed on the present system are the radiative cooling below the tropopause ( @xmath97 0.1 ) and the sensible and the latent heats from the surface .
if any motion is suppressed , a balanced state of the forcings is in principle statically unstable .
convective motions will be produced to release the unstable stratification . in this respect , it is natural to consider the symmetric cells to be the circulations derived from the `` convective instability''(ci ) rather than si .
as for the reference state as an average of the equilibrium states , a time - mean or space - mean state has , in general , no relation with the initial state for the instability ; the initial state will be lost when instability occurs , and may be no longer similar to the state of the full development of the instability .
it is well known , however , that the equilibrium state is sometimes very close to the critical state for the instability , or becomes a slightly unstable state , as represented by `` convective adjustment '' and `` baroclinic adjustment '' .
if this idea is also true for the present study , one may infer the origin of the symmetric cells by examining a reference state as a time - average of the equilibrium states .
14 shows the time - mean fields over the days 50 100 in the case of fig .
9 ( @xmath19 320 k , @xmath36 300 k ) : relationships between the normalized angular momentum @xmath98 and ( a ) potential temperature @xmath99 , ( b ) equivalent potential temperature @xmath100 and ( c ) saturation equivalent potential temperature @xmath101 . since the contours of @xmath10
are inclined more vertically than those of @xmath99 ( i.e. , the richardson number is much greater than one ) , it is stable for the `` dry '' si ( stone , 1966 ) . as for the `` moist '' si , bennetts and hoskins ( 1979 ) discussed that moist conditional si
is preferable if the contours of @xmath100 are more inclined than those of @xmath10 .
14(b ) shows that it is stable in the upper layers . in the lower layers ,
however , it is conditionally `` convectively '' unstable rather than conditionally symmetrically unstable . in this respect , therefore , the symmetric cells seem to be ascribable to ci . as seen above , the reference states of the present study is preferable for ci rather than si
, then it seems difficult to relate the properties of the symmetric cells to si . in the moist situation , however , a view of the moist si may still be applied to the symmetric cells , if one isolates circulations associated with cumulus heating from the effect of moist process on stratification .
emanuel ( 1982 ) considered moist si as cisk in a conditionally statically unstable state with a baroclinic flow .
he pointed out that a symmetric mode which propagates toward warmer regions exists .
this result is consistent with the present study , in which the symmetric cells are always propagating toward the warmer regions .
it should be reminded that , in the present case , it is not clear whether moist circulations can be separately considered from its effect on the stratification as in emanuel , since the stratification of the system is maintained by the symmetric cells themselves . as for the scale of the symmetric cells , it does not seem to be relevant to the scale predicted by the theories of si .
according to the dry si theory , there are two horizontal scales : the horizontal wavelength and the horizontal projection of the isentropes based on the layer depth ( emanuel , 1979 ; thorpe and rotunno , 1989 ) .
the wavelength of the dry si is controlled by frictional process particularly in the inviscid limit , but the viscosity is not important for the cell scale of the present study , although it may control the scale of the upward branch .
the second scale , the horizontal projection of the isentropes , is given by @xmath102 where @xmath40 is the layer depth .
this scale seems to correspond to the wavelength of the moist si ( emanuel , 1982 ; bennetts and hoskins , 1979 ) .
emanuel(1982 ) normalized horizontal length by @xmath103 , and showed the wavelength of the most growing mode is proportional to @xmath103 ( eq .
( 8) of emanuel(1982 ) ) . in the numerical experiments by bennetts and hoskins(1979 )
, the cell scale is also approximately @xmath103 , although they did not clearly argue about the scale .
however , the scale of the present symmetric cells is irrelevant to @xmath103 ; the scale is rather related to the slope of the angular momentum : @xmath104 where the effect of stability is neglected as in eq .
( [ eq : l_delt_f ] ) .
14(a ) shows that this scale is much smaller than @xmath103 .
the symmetric cells which exist in the two - dimensional model are examined through the dependency on the surface temperature gradient . in the case of no surface temperature gradient ( @xmath20 0 ) , the circulations of the symmetric cells oscillate .
the oscillation period is given by the convective time @xmath2 , and the cell scale by the rossby deformation radius .
advection of the angular momentum by the convective motion of the cells is essential for the oscillation . in the case of a surface temperature gradient
( @xmath37 0 ) , the symmetric cells propagate toward warmer regions .
each symmetric cell consists of a cell in the warmer side of the precipitating region and a cell in the colder side .
the cell shapes are different between the two types of the cells ; the width of the cells in the colder side is broader as in the upper layers , whereas the width in the warmer side is smaller as in the upper layers .
successive cells are sloping and vertically in contact with each other , so that the angular momentum is vertically transferred by diffusive process from the upper cell ( in the warmer region ) to the lower cell ( in the colder region ) .
the hadley cell in the axisymmetric model can be viewed as an extreme case of the symmetric cells ; it is the only symmetric cell that does not move .
it is because a pair of the hadley cells of both hemisphere is symmetric about the equator and the upward motion of the hadley cells is located at the warmest region ( the equator ) .
the structure of the symmetric cells is very similar to that of the symmetric hadley cell ; the cell width of colder regions ( higher latitudes ) is broader as in the upper layers .
the scale of the symmetric cells is also determined by a similar mechanism to that of the hadley cell .
it is determined by the relation between the surface temperature and the vertical mean temperature which is in the thermal wind balance with the angular momentum flow in the upper layers . in the hadley case ,
the latitudinal dependency of the coriolis parameter ( the equatorial @xmath105 effect ) must be considered in the estimation of the cell width ( s94 , fig .
13 ) . to show how the hadley cell and the symmetric cells are related with each other
, we calculate the motion of the position of the symmetric cells in the global condition . if the surface temperature is given by eq .
( [ eq : ts ] ) , the substitution of a local coriolis parameter and a surface temperature gradient into eq .
( [ eq : l_delt_f ] ) yields @xmath106 where @xmath107 is the temperature difference from the pole to the equator , @xmath108 is the radius of the earth , and @xmath109 . if the latitude of a precipitating region is denoted by @xmath110 , the propagation speed is estimated by @xmath111 integration for time under the initial condition @xmath112 yields @xmath113 fig .
15 shows the calculated paths of the cells , where @xmath114 20 days and @xmath115 0.068 are used . in this figure ,
the paths in the higher latitudes than 60@xmath15 are omitted .
15 shows that eq .
( [ eq : cell_pos ] ) reproduces a similar pattern to that of fig .
according to eq .
( 46 ) of s94 , the hadley scale is given by @xmath116 .
it is comparable to the scale of the symmetric cells at the moment they reach the equator . as for the results for @xmath20 0 as shown by fig .
4 , the concentration of the precipitation to the equator becomes less clear as the surface temperature gradient becomes smaller , while a cellar structure still exists in the mid - latitudes .
this difference may be attributed to the effect of rotation ; moist convective motions are organized into a cellar structure if there is sufficient rotation .
the discussion by schneider(1977 ) should be reconsidered in terms of these results .
schneider estimated the hadley width on the assumption that a reference temperature is horizontally uniform .
this assumptions corresponds to the case @xmath20 0 of the present study .
however , the hadley cell no longer exists in this case and schneider s estimation is not applicable .
of course , it is too early to say that the hadley cell in a three - dimensional model will show the same behavior as in the present two - dimensional models in the case of @xmath20 0 .
it has been reported that the concentration of the precipitation to the equator is sensitive to cumulus parameterization , in particular , when the surface temperature gradient is small near the equator ( hess et al . , 1993 ; numaguti , 1993 ) .
it is not clear how the symmetric cells are relevant to the phenomena in the observed atmospheres .
the latitudinal motion of the cumulus bands such as itcz may have a common mechanism of the movement of the symmetric cells .
goswami and shukla(1984 ) simulated a northward propagation of the indian monsoon by using an axisymmetric gcm . in their calculation , there exists a disturbance propagating toward the warmer region with period around 15 days , and this disturbance has many similarities to the observed fluctuation of the indian monsoon .
the structure of their disturbance is confined to the lower atmosphere and they pointed out the interaction with the land surface through the evaporation plays a key role . these properties are not seen in the present study , so it is not clear their disturbance and the symmetric cells of the present study have the same origin .
although the symmetric cells are convective modes , they should be re - examined in terms of the symmetric instability ( e.g. , stone , 1966 ; emanuel , 1979 , 1982 ) . in particular , emanuel(1982 ) pointed out that a symmetric mode which propagates toward warmer regions exists in a shear flow of moist atmosphere .
his results are consistent with the present study .
it will be also interesting to study the relationship with the motion of the spiral bands of typhoon or with the phenomena in the planetary atmospheres such as the striped pattern of jupiter .
the symmetric cells are also expected to exist in the dry atmosphere or in the boussinesq fluid , although all the experiments in the present study are for the moist atmosphere .
they will oscillate under the uniform surface temperature condition , and will propagate toward warmer regions if there is a surface temperature gradient , since these are the consequences of the conservation of the angular momentum
. the difference will be in their horizontal scale ; horizontal viscosity will play a major role in the scale of the cells in the dry atmosphere or in the boussinesq fluid , as in the case of the bnard convection .
the symmetric cells should be discussed in a more general context , in which they are viewed as a convective motion on a rotating plane with a rigid boundary below and a slip boundary above .
it should be also examined how the symmetric cells are modified in a large - scale three - dimensional field , particularly in the situation where both vertical convection and baroclinic waves co - exist in a vertically unstable basic stratification .
the comments by two referees are very helpful for the improvement of the manuscript .
this work was supported by the center for the climate system research , university of tokyo .
numerical calculations were performed with hitac s-3800 and m-880 at the computer center of the university of tokyo , and nec sx-3 at the national institute for environmental studies .
figures were produced by gfd - dennou libraries and gtool3 by dr .
numaguti .
the model is two - dimensional with a lateral direction and a vertical direction .
the primitive equations with a @xmath23-coordinate are used . the global model ,
that is , a two - dimensional axisymmetric model with a spherical coordinate is described in sec 2 of s94 .
formulation of the @xmath3-plane model ( with the cartesian coordinate ) is almost similar to that of the global model .
the symbols are summarized in table .
let the direction perpendicular to the two - dimensional plane be denoted by @xmath117 ( longitude , in the spherical case ) , the lateral direction by @xmath28 , the vertical direction by @xmath118 , and the corresponding metrics by @xmath119 , @xmath120 , and @xmath121 , respectively .
we also define @xmath122 .
the lateral coordinate is bounded by @xmath123 , where @xmath124 and @xmath125 if @xmath28 is non - dimensionalized by the length of the lateral domain @xmath66 in the cartesian coordinate system ( only in this appendix , @xmath28 is non - dimensional ) . in the spherical coordinate system , @xmath126 ( @xmath5 is the latitude ) is used , and hence @xmath127 and @xmath125 .
the metrics and the angular momentum @xmath10 are expressed , in the cartesian coordinate , by @xmath128 whereas , in the spherical coordinate , by @xmath129 where @xmath130 is the coriolis parameter , @xmath131 is the rotation velocity , and @xmath108 is the radius of the earth .
the surface pressure @xmath26 , the angular momentum @xmath10 , the lateral velocity @xmath132 , the moist enthalpy @xmath133 , and the specific humidity @xmath134 are predicted variables in this system . the choice of @xmath10 and @xmath133 as the predicted variables automatically guarantees the conservations of the angular momentum and the moist enthalpy .
the basic equations are expressed as follows : @xmath135 \nonumber \\ & & + p_{s } f_{c } + p_{s } f_{y } , \label{eq : model_v}\\ 0 & = & - p_{s } - \rho \partial_{\sigma } \phi , \label{eq : model_seiriki}\\ \partial_{t } ( p_{s } h ) & = & - \frac{1}{\sqrt{g } } \partial_{y } ( h_{1 } p_{s } v h ) - \partial_{\sigma } ( p_{s } \dot{\sigma } h ) \nonumber \\ & & + \frac{p_{s}}{\rho } \dd{}{t } p + \frac{p_{s}}{\rho } q^{dif } + \frac{p_{s}}{\rho } q^{rad } + \frac{p_{s}}{\rho } q^{fric } , \label{eq : model_h } \\
\partial_{t } ( p_{s } q ) & = & - \frac{1}{\sqrt{g } } \partial_{y } ( h_{1 } p_{s } v q ) - \partial_{\sigma } ( p_{s } \dot{\sigma } q ) + \frac{p_{s}}{\rho } s^{dif } - \frac{p_{s}}{\rho } s^{prc } , \label{eq : model_q}\end{aligned}\ ] ] where @xmath136 is the gravitational potential and @xmath137 is the centrifugal force , which is given , in the spherical coordinate , by @xmath138 where @xmath139 is the centrifugal potential . in the cartesian coordinate
, it corresponds to the coriolis force @xmath140 .
the vertical @xmath23-velocity @xmath141 is given by the integration from @xmath23 to @xmath97 1 of eq .
( [ eq : model_ps ] ) : @xmath142 .
\label{eq : model_sigdot}\end{aligned}\ ] ] the diffusion terms @xmath143 , @xmath144 , @xmath145 , @xmath146 , and @xmath147 in the basic equations are given by the same form as in section 2 of s94 .
@xmath148 of the humidity equation is a term representing the precipitation , which is calculated as a removal of a liquid part of the humidity after each of time integration ( this process is referred as the `` large - scale condensation '' ) .
the radiative cooling @xmath149 is specified as a constant value in the @xmath3-plane model .
boundary conditions are @xmath150 at @xmath97 0 , 1 , and @xmath151 at @xmath152 , @xmath153 . in the spherical coordinate , although @xmath154 at the poles ( @xmath155 ) , the angular velocity @xmath156 does not generally vanish .
the fluxes at the boundaries at @xmath155 , and @xmath157 are zero , while those at @xmath158 are given by bulk formulae .
the grid points are defined as follows . as shown by fig .
16 , the lateral boundary @xmath152 is denoted by @xmath159 1/2 , and @xmath160 by @xmath161 , while the vertical boundary @xmath158 is denoted by @xmath162 and @xmath157 by @xmath163 .
the grid points are defined at both half - integer @xmath164 ( @xmath165 ) and integer @xmath166 ( @xmath167 ) .
the half - integer point which is placed between @xmath166 and @xmath168 ( @xmath167 and @xmath169 ) is denoted by @xmath170 ( @xmath171 ) .
the number of the half - integer points is @xmath172 in the lateral direction , while it is @xmath173 in the vertical direction .
@xmath174 is even in the case of the spherical coordinate system .
the grid intervals of the lateral direction are equal both for integer and half - integer points .
as for the vertical direction , the half - integer points are equally placed , while the integer points are based on arakawa and suarez(1983 ) : @xmath175 where @xmath176 .
the variables are defined at the following points : @xmath177 vertical discretization by arakawa and lamb(1977 ) and arakawa and suarez(1983 ) is used so as to conserve total kinetic energy . the angular momentum and the equations of motion
are discretized so as to conserve total kinetic energy(satoh , 1994b ) .
emanuel , k. a. , 1983 : elementary aspects of the interaction between cumulus convection and the large - scale environment . in _
mesoscale meteorology theories , observations and models _ ( eds .
lilly , d.k . and gal - chen , t. ) , reidel , 551 - 575 .
hess , p. g. , d. s. battisti and p. j. rasch , 1993 : maintenance of the intertropical convergence zones and the large - scale tropical circulation on a water - covered earth , _
_ , * 50 * , 691 - 713 . fig
: : convergence of the angular momentum . the standard experiment of the global model .
( a ) total convergence @xmath178 , ( b ) steady component @xmath179 , and ( c ) unsteady component @xmath180 .
the contour intervals are ( a ) 100 m@xmath8/s@xmath8 , and ( b ) , ( c ) 60 m@xmath8/s@xmath8 .
2 . : : time variation of the meridional distribution of the precipitation .
dependencies on the pole - to - equator temperature difference : ( a ) @xmath1 k , ( b ) @xmath84 k , ( c ) @xmath181 k ( the standard experiment ) , and ( d ) @xmath182 k. the contour intervals are 5 @xmath183 10@xmath184 kg / m@xmath8 s. fig .
3 . : : time variation of the meridional distribution of the precipitation .
dependencies on the coefficient of the viscosity : ( a ) @xmath185 m@xmath8/s , ( b ) @xmath186 m@xmath8/s , where @xmath17 is the viscosity in the free atmosphere and @xmath21 is the viscosity at the surface .
the contour intervals are 5 @xmath183 10@xmath184 kg / m@xmath8 s. fig .
4 . : : time variation of the lateral distribution of the precipitation in the @xmath3-plane model .
the surface temperature is linearly distributed from @xmath187 at @xmath188 to @xmath189 at @xmath190 : ( a ) @xmath191 k , ( b ) @xmath192 k , @xmath193 k , ( c ) @xmath194 k , @xmath195 k , and ( d ) @xmath196 k , @xmath197 k. the contour intervals are 2 @xmath183 10@xmath198 kg / m@xmath8 s. fig . 5 . : : time variation of the zonal winds ( winds perpendicular to the model domain ) at the surface .
surface temperature is uniform : @xmath191 k. the contour intervals are 2 m / s .
: : distribution of the zonal winds ( winds perpendicular to the figures ) .
surface temperature is uniform : @xmath191 k. ( a ) 10 days average for the days 46 55 , and ( b ) 10 days average for the days 56 65 .
the contour intervals are 5 m / s .
: : as in fig . 6 but for the temperature difference from the horizontal average .
the contour intervals are 1 k. fig .
: : time variation of the temperature lapse rate ( @xmath199 ) at @xmath200 .
surface temperature is uniform : @xmath191 k. the contour intervals are 2 k / km .
the hatched area indicates the values larger than 8 k / km .
: : time variation of the lateral distribution of the precipitation in the @xmath3-plane model for @xmath19 320 k and @xmath36 300 k. the contour intervals are 2 @xmath183 10@xmath198 kg / m@xmath8 s. a @xmath68 b denotes a reference line for the composite maps . fig . 10 .
: : composite maps averaged with the propagation of the symmetric cells from @xmath201 at the day 51 to @xmath202 at the day 100 ( shown by a @xmath68 b in fig .
the center of the symmetric cell is placed at @xmath203 .
( a ) zonal winds(winds perpendicular to the figure ) with the contour interval of 5 m / s and ( b ) stream functions with the contour interval of 1,000 kg / m s. ( stream functions are those per unit length perpendicular to the figure . ) fig . 11 .
: : temperature ( solid lines ) and normalized angular momentum @xmath204 ( broken lines ) at the day 100 in the case of @xmath19 320 k and @xmath36 300 k. the contour intervals are 5k for the temperature and 0.02 for the normalized angular momentum . fig
: : the scales of the symmetric cells .
( a ) scale in the case of @xmath20 0 : @xmath205 , ( b ) scale in the case of @xmath37 0 when the effect of static stability is not included : @xmath206 , and ( c ) scale in the case of @xmath37 0 when the effect of static stability is included : the warmer side @xmath207 and the colder side @xmath92 .
thick allows indicate the precipitating region .
the curves are the temperature difference from the precipitating region in the middle of the troposphere @xmath81 , and the sloping lines are that of the surface temperature @xmath80 .
the three vertical segments of lines shown by @xmath208 in ( a ) and ( c ) are the same temperature difference @xmath209 . in
( b ) , @xmath210 , and , in ( c ) , @xmath211 . fig . 13 .
: : relationship between surface zonal winds and motion of a parcel which starts from the origin @xmath212 0 shown by a thick arrow .
@xmath212 0 is the position of precipitation at the time when the parcel starts .
if the angular momentum is conserved along the motion of the parcel , zonal wind becomes easterly if the parcel comes down to the negative @xmath28 , whereas it becomes westerly if the parcel comes down to the positive @xmath28 .
zonal wind is zero only if the parcel returns to the starting position @xmath212 0 . in order for the surface wind to be easterly
, the precipitation must move toward the negative direction during the cycle of the parcel .
: : comparison of time averaged distributions of ( a ) potential temperature @xmath99 , ( b ) equivalent potential temperature @xmath100 and ( c ) saturation equivalent potential temperature @xmath101 with distribution of normalized angular momentum @xmath204 .
the normalized angular momentum is the same in ( a)-(c ) , and is shown by broken lines with the contour interval of 0.02 .
the potential temperatures , @xmath99 , @xmath100 and @xmath101 , are shown by solid lines with the contour interval of 5k .
these are 50 days average for the days 51 - 100 in the case of @xmath19 320 k and @xmath36 300 k. fig .
: : calculated patterns of the symmetric cells propagation by eq .
( [ eq : cell_pos ] ) .
: : arrangement of the grid points . | as a tool for understanding the meridional circulation of the atmosphere , a two - dimensional ( latitude height ) numerical model is used to clarify the relationship between the hadley circulation and large - scale motions associated with moist convection . the model is based on the primitive equations including the moist process , and two kinds of coordinates are used : the spherical coordinate and the cartesian coordinate with a uniform rotation .
the surface temperature is externally fixed and the troposphere is cooled by the radiation ; unstable stratification generates large - scale convective motions .
dependencies on the surface temperature difference from north to south @xmath0 are investigated .
the numerical results show that a systematic multi - cell structure exists in every experiment .
if the surface temperature is constant ( @xmath1 ) , convective motions are organized in the scale of the rossby deformation radius and their precipitation patterns have a periodicity of the advective time @xmath2 . as @xmath0 becomes larger , the organized convective system tends to propagate toward warmer regions .
the convective cells calculated in the cartesian coordinate model is very similar to those of the mid - latitudes in the spherical coordinate model . in particular
, the hadley cell can be regarded as the limit of the convective cells in the equatorial latitudes .
-.4 in 6.4 in |
there has been high interest in recent years in the development and analysis of adaptive finite element methods ( afem ) for approximating eigenvalues and eigenfunctions of elliptic operators .
we consider the following model eigenvalue problem : find @xmath0 such that @xmath1 and @xmath2 here @xmath3 , @xmath4 , is a polyhedral domain , @xmath5 and @xmath6 .
there is then a sequence of eigenvalues @xmath7 and corresponding @xmath8-orthonormal eigenfunctions @xmath9 satisfying .
given a nested sequence of adaptively generated simplicial meshes @xmath10 with associated finite element spaces @xmath11 ( @xmath12 ) , the corresponding discrete eigenvalue problem is : find @xmath13 such that @xmath14 and @xmath15 we seek to approximate an eigenvalue cluster @xmath16 and associated invariant subspace @xmath17 .
our index set @xmath18 is given by @xmath19 for some @xmath20 , @xmath21 .
the corresponding discrete sets are @xmath22 and @xmath23 .
afem for eigenvalues are typically based on the standard loop @xmath24 to the finite element error , afem employs local error indicators @xmath25 , where @xmath26 is a standard residual error indicator for the residual @xmath27 .
let @xmath28 be a given parameter .
@xmath29 is used in to select a smallest set @xmath30 satisfying the drfler ( bulk ) @xcite criterion @xmath31 proofs of convergence and optimality of afem for approximating have been given in several papers .
the first proof of optimality of afem for controlling simple eigenvalues and eigenfunctions was given in @xcite .
other papers concerning convergence of afem for simple eigenvalues include @xcite .
the paper @xcite contains a proof of optimal convergence of standard afem for an eigenvalue with multiplicity greater than one , while @xcite proves a similar result for clustered eigenvalues .
these papers mirror afem convergence theory for source problems ( cf .
@xcite ) in that they first prove that the afem contracts at each step .
an optimal convergence rate dependent on membership of the eigenfunctions in suitable approximation classes is then obtained by standard methods .
all require that the maximum mesh diameter in the initial mesh be sufficiently small to suitably resolve the nonlinearity of the problem .
the behavior of afem for eigenvalues in the pre - asymptotic regime was studied in @xcite , where the authors proved plain convergence results ( with no rates ) starting from any initial mesh .
these results guarantee convergence of afem for general elliptic eigenproblems to some eigenpair , but not generically to the correct pair . the works of dai et al . @xcite and gallistl @xcite are most relevant to ours .
in @xcite the authors establish convergence of an afem for a multiple eigenvalue of a symmetric second - order linear elliptic operator for arbitrary - degree finite element spaces .
approximation of eigenclusters of the laplacian using piecewise linear elements is considered in @xcite . .
this leaves open the question of convergence results for afem for eigenvalue clusters using lagrange spaces of arbitrary polynomial degree .
we fill this gap by showing that standard afem for eigenvalue clusters using polynomials of arbitrary degree also converge optimally . while we consider only conforming simplicial meshes , we also provide a key step in extending such analysis to quadrilateral elements of any degree , meshes with hanging nodes , and discontinuous galerkin methods ; cf .
@xcite for analysis of the source problem .
the analysis of @xcite additionally does not immediately apply in the case of non - constant diffusion coefficients .
in contrast , our results extend as in @xcite to general symmetric second - order linear elliptic operators ( see remark [ rem : genops ] ) .
we briefly explain the difficulty which we resolve . a key step in standard afem convergence proofs establishing a certain continuity between error indicators on adjacent mesh levels . in the case of multiple or clustered eigenvalues ,
the ordering and alignment of the discrete eigenfunctions may change between mesh levels even on fine meshes .
thus @xmath32 and @xmath33 may not approximate the same eigenpair , making the comparison between @xmath34 and @xmath35 used in standard afem convergence proofs irrelevant .
a critical contribution was made in @xcite , where this problem was circumvented by first analyzing a theoretical ( non - computable ) afem based on error indicators @xmath36 aligned to the _ fixed _ continuous cluster .
@xmath37 may be viewed as an indicator for a `` pseudo - residual '' @xmath38 , where @xmath39 and @xmath40 are projections onto @xmath41 defined later .
given @xmath42 $ ] , the marking strategy consists of selecting the smallest set @xmath43 of @xmath44 with @xmath45 optimality is guaranteed ( see @xcite ) provided the initial mesh is fine enough and @xmath46 for some @xmath47 only depending on @xmath44 and the initial subdivision @xmath48 . to deduce optimal properties of the practical algorithm based on @xmath49 instead of @xmath50 , an elementwise equivalence of @xmath51 and @xmath37
is used to show that the corresponding afem are essentially equivalent .
we provide ( lemma [ lem : main ] ) a more direct proof of the equivalence between @xmath37 and @xmath52 which is valid for finite element spaces of arbitrary degree .
extension of the afem convergence results for eigenvalue clusters of @xcite to higher - degree finite element spaces is then immediate ( see theorem [ t : opt ] and corollary [ c : opt ] ) .
equivalence constants between @xmath37 and @xmath52 which are independent of essential quantities on sufficiently refined meshes , leading to a final result which is more robust with respect to the properties of the target cluster . in view of and , optimal afem require @xmath53 , where @xmath54 is independent of @xmath55 , @xmath56 , and @xmath57 ( see remark [ r : theta ] ) .
in contrast , the analysis of @xcite implies that a stricter ( smaller ) choice of @xmath58 may be necessary as @xmath55 and the ratio @xmath59 increase .
as numerical experiments confirm ( [ sec : numerics ] ) , this improvement is important from a practical standpoint as it establishes that no knowledge of cluster properties is needed in order to choose @xmath58 correctly
. it also potentially increases computational efficiency by confirming that @xmath58 may be chosen reasonably large even when computing large clusters . on the other hand
, @xcite implies that @xmath58 should be taken to be small as @xmath55 increases , thus yielding a large number of adaptive loops and decreasing computational efficiency .
the remainder of the note is structured as follows . in [ sec : prelims ] we give further assumptions and preliminaries . in [ sec : main_result ] we prove our main result and briefly sketch its application in the proof of afem convergence for eigenvalue clusters .
a brief set of numerical experiments in [ sec : numerics ] illustrates our theoretical results .
finally note that we do not give a full proof of eigenvalue convergence , as most of the analysis of @xcite holds verbatim for higher - order elements given the equivalence of the theoretical and practical error estimators .
we also largely employ the notation of that work .
familiarity with @xcite is thus essential to the reader .
let @xmath44 be a polyhedral domain and denote by @xmath60 the set of all conforming refinements of @xmath61 obtainable by the newest - vertex bisection algorithm typically used in afem or its generalization to higher dimensions ; cf .
@xcite . by construction
all @xmath62 are uniformly shape regular .
given @xmath63 , we denote by @xmath64 the space of continuous piecewise polynomials of arbitrary but fixed degree @xmath65 subordinate to @xmath66 .
for each @xmath67 , we abbreviate @xmath68 .
the afem produce sequences of subdivisions @xmath69 such that @xmath70 is a refinement of @xmath71 . thus @xmath72 .
we also define several projection operators .
let @xmath73 be the @xmath8 projection onto @xmath41 , and let @xmath74 be the ritz projection .
that is , @xmath75 in addition , we define @xmath76 . finally , @xmath77 is the @xmath8 projection onto the continuous invariant subspace @xmath78 corresponding to the target eigenvalue cluster .
given a discrete eigenpair @xmath79 with @xmath80 and @xmath81 , the computable local error indicator @xmath49 is given by @xmath82 here @xmath83 is the jump across the element boundary , @xmath84 is the @xmath8 norm over @xmath85 and @xmath86 .
let @xmath87 be the energy norm .
the computable error estimator @xmath88 reliably controls up to higher - order terms which can be proved to be negligible under the assumption that @xmath89 is sufficiently small .
the theoretical local indicator @xmath50 is given by @xmath90 a slight modification of proposition 4.1 of @xcite shows that @xmath91 also bounds @xmath92 up to higher - order terms which are negligible if @xmath93 is sufficiently small . in @xcite the volume term is simplified to @xmath94 due to the use of piecewise linear finite element spaces , and the definition in the higher - order case is left open with our choice given as one possibility ( cf .
remark 9 ( f ) of @xcite ) .
our definition of @xmath37 also differs from that given in @xcite .
proofs of optimal convergence rates for adaptive fem require assumptions on the eigenspaces and their resolution by the initial mesh .
first , the continuous eigencluster must be separated from the remainder of the spectrum , i.e. , @xmath95 ( here we take @xmath96 ) .
in addition , the discrete cluster respects this separation in that @xmath97 and @xmath98 .
let then @xmath99 .
the implied requirement on the initial mesh may be more or less strict depending on the separation of the cluster from the remainder of the spectrum .
reliability of our a posteriori estimators is only guaranteed for sufficiently fine initial mesh @xmath61 .
we have for example from proposition 4.1 of @xcite that latexmath:[\ ] ] we now define the vectors @xmath108 $ ] , @xmath109 $ ] , @xmath110 $ ] , and . here
these quantities vary with @xmath112 ( for @xmath113 and @xmath114 ) or @xmath115 lying on the mesh skeleton ( for @xmath116 and @xmath117 ) , but this fact has little significance for the time being .
we also define the matrix @xmath118 $ ] , with @xmath119 the row and @xmath120 the column index .
the relationships may now be written @xmath121 let @xmath122 denote the operator @xmath123-norm and @xmath124 the euclidean length .
the proof of lemma [ lem : main ] is thus reduced to showing that @xmath125 under the stated hypotheses .
we now employ several basic facts from linear algebra . letting @xmath126
, we have that @xmath127 and thus @xmath128 . because @xmath129 is symmetric and positive
, @xmath130 is equal to the maximum eigenvalue of @xmath129 .
a short computation shows that @xmath131 $ ] .
a standard result from linear algebra is that @xmath132 and @xmath133 are isospectral for square @xmath134 , so @xmath135 is isospectral with @xmath129
. therefore we have that @xmath136 , and both quantities are equal to the maximum eigenvalue of @xmath135 .
the matrix @xmath135 was analyzed extensively in the proof of lemma 5.1 of @xcite .
in particular , @xmath135 is nonsingular under the condition , and by ( 5.2 ) and following of @xcite , we have @xmath137 gershgorin s theorem thus gives that the eigenvalues @xmath138 of @xmath135 satisfy @xmath139 thus @xmath140 , which is the first inequality in .
the invertibility of @xmath135 guarantees the invertibility of @xmath141 .
computing as above yields @xmath142 .
because @xmath135 is positive and diagonalizable , we have from that @xmath143 , thus completing the proof of the second inequality in . in @xcite
it is shown for piecewise linear elements that under the assumption , @xmath144 the first bound in holds with no restriction on @xmath93 .
our proof thus yields an improved cluster - independent bound under the fineness assumption .
also gives an improved bound for @xmath52 in terms of @xmath37 .
significantly loosening the restriction appears to be substantially more difficult .
@xmath141 is invertible for @xmath145 , but the obtained upper bound for @xmath146 degenerates as @xmath147 .
the condition @xmath148 is not substantially weaker than , and control over the equivalency constants is lost as @xmath147 , so we retain from @xcite . in lemma 3.3 of @xcite it is shown that if @xmath149 and @xmath150 is of arbitrary polynomial degree , @xmath151 where @xmath152 as @xmath153 and @xmath154 is similar to @xmath37 . in our analysis
the obtained bounds are independent of @xmath55 and the double sum term on the right is completely absent .
the absence of the double sum term on the right is important , as ; cf .
remark 9 ( f ) of @xcite .
[ rem : genops ] in @xcite afem optimality is analyzed for eigenvalues of arbitrary second - order symmetric operators @xmath155 . letting @xmath156
, our equivalence analysis extends immediately to the general case .
we consider only @xmath157 , @xmath158 because analyzing optimality in the general case involves consideration of data oscillation , which would make our presentation much more involved ; cf @xcite .
let given lemma [ lem : main ] , the analysis of @xcite mostly holds verbatim for higher - order finite element spaces .
in particular , the following counterpart to theorem 3.1 of @xcite holds .
[ t : opt ] assume that the bulk parameter @xmath159 and initial mesh size @xmath93 are sufficiently small .
assume that for some @xmath160 , . then there exists @xmath161 depending possibly on @xmath61 but independent of other essential quantities such that @xmath162 several remarks are in order . in @xcite optimality
is expressed with respect to the class @xmath163 where @xmath164 is the @xmath8-projection onto @xmath165 , the space of piecewise polynomials of degree @xmath166 subordinate to @xmath66 .
employing this class requires proving equivalence between @xmath167 and @xmath168 , that is , approximation by functions in @xmath165 is equivalent to approximation by gradients of functions in @xmath169 . in @xcite ,
the needed result is obtained by citing @xcite , in which the equivalence is shown up to data oscillation terms .
however , the necessary equivalence has recently been shown to hold on arbitrary meshes without data oscillation terms in @xcite , which could simplify the proof of proposition 3.1 of @xcite .
[ r : theta ] the dependence of @xmath93 and @xmath58 on various quantities is given in detail in @xcite .
the threshold condition for @xmath58 depends on the second equivalence constant in .
thus @xcite requires @xmath170^{-1}$ ] ( cf .
lemma 7.3 ) , where @xmath171 is independent of essential quantities ( especially cluster properties ) .
we only require @xmath172 , thus establishing that rate - optimal adaptive convergence holds with the marking parameter chosen independent of the cluster properties .
the analysis of @xcite however still indicates that the asymptotic nature of the optimality result may be more or less pronounced depending on cluster properties ( cf . ) .
finally , optimal approximation of eigenvalues follows as in ( * ? ? ?
* corollary 3.1 ) .
[ c : opt ] under the conditions of theorem [ t : opt ] there exists @xmath173 such that for @xmath174 , @xmath175 note that the constant @xmath176 above depends on properties of the eigenvalue cluster ; cf .
* corollary 3.1 ) for a precise expression .
to illustrate our result , consider the slit domain @xmath177 subdivided into @xmath178 squares , each of diameter @xmath179 .
the finite element spaces consist of continuous piecewise polynomials of degree @xmath180 , @xmath181 , or @xmath182 in each coordinate direction .
the adaptive algorithm is implemented within the _ deal.ii _
@xcite library .
we take @xmath183 and consider the cases @xmath184 .
our highest - precision computations indicate that the first twelve eigenvalues ( truncated to 6 decimal places ) are 10.147392 , 17.662596 , 17.662596 , 19.739208 , 26.101811 , 38.349159 , 38.349159 , 46.553966 , 49.149607 , 49.149607 , 49.348022 , and 49.348022 . in figure
[ f : estimators ] , we plot the error estimators @xmath185 against the total number of degrees of freedom for different values of marking parameters @xmath58 and cluster sizes @xmath55 .
we do not know the exact eigenfunctions and so can not plot actual errors .
however , in the case of the laplacian @xcite guarantees plain convergence ( without rates ) of afem for simple and multiple eigenvalues to the corresponding continuous eigenpairs .
these results rely on completely differently proof techniques than do ours , and adaptation to the case of clustered eigenvalues is straightforward .
the analysis of @xcite also guarantees that @xmath186 as @xmath187 , which yields reliability of @xmath52 for @xmath188 sufficiently large ( cf . ) .
it is thus meaningful to track @xmath52 instead of actual errors .
in addition , employing a cluster - independent marking parameter as suggested by our theory is reasonable also in the pre - asymptotic range as the plain convergence analysis of @xcite guarantees that the algorithm will eventually reach the asymptotic range
. optimal decay rates of @xmath189 are observed provided @xmath190 when @xmath191 and @xmath192 when @xmath180 . according to corollary [ c : opt ] ( @xmath193 ) ,
this indicates an optimal decay rate of @xmath194 for the error in approximating each eigenvalue in the cluster .
as guaranteed by our analysis , the range of @xmath58 for which optimality is recovered is not affected by the cluster properties .
in contrast , following @xcite as explained in remark [ r : theta ] leads to the restriction @xmath195^{-1}$ ] with @xmath171 independent of essential quantities . for our particular computations ,
this yields : @xmath196 thus following precisely the theory of @xcite would lead to a thousand - fold reduction in @xmath58 when moving from our first to our second computational example .
this would potentially require a massive increase in the number of afem iterations required in order to achieve a given error reduction .
we have demonstrated theoretically and confirmed computationally that this increase in computational expense is unnecessary . | proofs of convergence of adaptive finite element methods for approximating eigenvalues and eigenfunctions of linear elliptic problems have been given in a several recent papers .
a key step in establishing such results for multiple and clustered eigenvalues was provided by dai et .
al . in @xcite , who proved convergence and optimality of afem for eigenvalues of multiplicity greater than one .
there it was shown that a theoretical ( non - computable ) error estimator for which standard convergence proofs apply is equivalent to a standard computable estimator on sufficiently fine grids . in @xcite , gallistl used a similar tool to prove that a standard adaptive fem for controlling eigenvalue clusters for the laplacian using continuous piecewise linear finite element spaces converges with optimal rate . when considering either higher - order finite element spaces or non - constant diffusion coefficients , however , the arguments of @xcite and @xcite do not yield equivalence of the practical and theoretical estimators for clustered eigenvalues . in this note
we provide this missing key step , thus showing that standard adaptive fem for clustered eigenvalues employing elements of arbitrary polynomial degree converge with optimal rate .
we additionally establish that a key user - defined input parameter in the afem , the bulk marking parameter , may be chosen entirely independently of the properties of the target eigenvalue cluster .
all of these results assume a fineness condition on the initial mesh in order to ensure that the nonlinearity is sufficiently resolved .
eigenvalue problems , spectral computations , a posteriori error estimates , adaptivity , optimality 65n12 , 65n15 , 65n25 , 65n30 |
the na61/shine ( sps heavy ion and neutrino experiment ) experiment at the cern sps pursues a rich physics program in various fields @xcite .
first , precise hadron production measurements are performed for improving calculations of the neutrino flux in the t2k neutrino oscillation experiment @xcite , as well as for more reliable simulations of cosmic - ray air showers in the pierre auger and kascade experiments @xcite .
second , p+p , p+pb and nucleus+nucleus collisions will be studied extensively at sps energies .
this article presents first na61/shine results on charged pion spectra in p+c interactions at 31 gev / c which are needed for an accurate neutrino flux prediction in the t2k experiment .
the results are based on the data collected during the first na61/shine run in 2007 .
t2k is a long baseline neutrino experiment in japan , which uses a high intensity neutrino beam produced at j - parc .
it aims to precisely measure the @xmath0 appearance and @xmath1 disappearance @xcite . in order to generate the neutrino beam
a high intensity 30 gev ( kinetic energy ) proton beam impinging on a 90 cm long graphite target is used , where @xmath2 and k mesons decaying into ( anti)neutrinos are produced .
the neutrino fluxes and spectra are then measured both at the near detector complex , 280 m from the target , and by the super - kamiokande ( sk ) detector located 295 km away from the neutrino source and 2.5 degrees off - axis .
neutrino oscillations are probed by comparing the neutrino flux measured at sk to the predicted one . in order to predict the flux at sk
one uses the near detector measurements and extrapolates them to sk with the help of monte carlo simulations . up to now , these monte carlo predictions are based on hadron production models only . for more precise predictions , which would allow the reduction of systematic uncertainties to the level needed for the t2k physics goals , measurements of pion and kaon production off carbon targets are essential @xcite .
the purpose of the na61/shine measurements for t2k is to provide this information at exactly the proton extraction energy of the j - parc main ring synchrotron , namely 30 gev kinetic energy ( approximately 31 gev / c momentum ) .
presently , the t2k neutrino beam - line is set up to focus positively charged hadrons , in such a way that it produces a @xmath1 beam .
spectra of positively charged pions presented in this paper constitute directly an essential ingredient in the neutrino flux calculation .
the na61/shine experiment is a large acceptance hadron spectrometer in the north area h2 beam - line of the cern sps . the schematic layout is shown in fig .
[ fig : detector ] together with the overall dimensions .
the main components of the current detector were constructed and used by the na49 collaboration @xcite .
a set of scintillation and cherenkov counters as well as beam position detectors ( bpds ) upstream of the spectrometer provide timing reference , identification and position measurements of the incoming beam particles .
the main tracking devices of the spectrometer are large volume time projection chambers ( tpcs ) .
two of them , the vertex tpcs ( vtpc-1 and vtpc-2 in fig .
[ fig : detector ] ) , are located in a free gap of 100 cm between the upper and lower coils of the two superconducting dipole magnets .
their maximum combined bending power is 9 tm . in order to optimize the acceptance of the detector at 31 gev / c beam momentum , the magnetic field used during the 2007 data taking period
was set to a bending power of 1.14 tm .
two large tpcs ( mtpc - l and ) are positioned downstream of the magnets symmetrically to the beam line .
the tpcs are filled with ar : co@xmath3 gas mixtures in proportions 90:10 for vtpcs and 95:5 for mtpcs .
the particle identification capability of the tpcs based on measurements of the specific energy loss , @xmath4 , is augmented by time - of - flight measurements using time - of - flight ( tof ) detectors .
the tof - l and arrays of scintillator pixels have a time resolution of better than 90 ps @xcite .
before the 2007 run the experiment was upgraded with a new forward time - of - flight detector ( tof - f ) in order to extend the acceptance .
the tof - f consists of 64 scintillator bars with photomultiplier ( pmt ) readout at both ends resulting in a time resolution of about 115 ps .
the target under study is installed 80 cm in front of the vtpc-1 .
the results presented here were obtained with an isotropic graphite target of dimensions 2.5(w)@xmath52.5(h)@xmath52(l ) cm and with a density of @xmath6 g/@xmath7 .
the target thickness along the beam is equivalent to about 4% of a nuclear interaction length ( @xmath8 ) .
this section presents the procedures used for data analysis to extract pion cross sections .
crucial for this analysis is the identification of the produced pions . depending on the momentum interval ,
different approaches have been adopted , which lead also to different track selection criteria .
the calibrated @xmath4 distributions as a function of particle momentum for positively and negatively charged particles are presented in fig .
[ fig : dedx ] . ) and negatively ( @xmath9 ) charged particles as a function of momentum .
curves show parameterizations of the mean @xmath4 calculated for different particle species .
, title="fig : " ] ) and negatively ( @xmath9 ) charged particles as a function of momentum .
curves show parameterizations of the mean @xmath4 calculated for different particle species .
, title="fig : " ] the task is facilitated for the negatively charged pions , by the observation that more than 90% of primary negatively charged particles produced in p+c interactions at this energy are @xmath10 , and thus the analysis of @xmath10 spectra can also be carried out without additional particle identification . in the low momentum region
( less than about 1 gev / c ) , it is sufficient to distinguish pions from electrons / positrons , kaons and protons by means of particle identification via measurements of specific energy loss ( @xmath4 ) in the tpcs .
a reliable identification of @xmath11 mesons was not possible at momenta above 1 gev / c where the bethe - bloch ( bb ) curves for pions , kaons and protons cross each other . on the other hand , for @xmath12 mesons , where the contribution of @xmath13 and antiprotons is almost negligible , the @xmath4 analysis could be extended in momentum up to 3 gev / c allowing consistency checks with the other analysis methods in the region of overlap . .
the lines show the expected mass squared values for different particles .
_ left : _ example of two - dimensional @xmath14@xmath4 plots for positively charged particles in the momentum range 2 - 3 gev / c .
@xmath15 contour around fitted pion peaks are shown .
, title="fig : " ] .
the lines show the expected mass squared values for different particles .
_ left : _ example of two - dimensional @xmath14@xmath4 plots for positively charged particles in the momentum range 2 - 3 gev / c .
@xmath15 contour around fitted pion peaks are shown .
, title="fig : " ] high purity particle identification can be performed by combining the @xmath16 and @xmath4 information .
moreover , in the momentum range 1 - 4 gev / c , where @xmath4 bands for different particle species overlap , particle identification is in general only possible using the @xmath16 method , see figs .
[ fig : ftof_resolution ] .
the tof - f detector was designed to cover the necessary acceptance in momentum and polar angle required by the t2k experiment , although limited to particle momenta above about 0.8 gev / c .
indeed , three analysis methods were applied to obtain pion spectra : a. analysis of @xmath10 mesons via measurements of negatively charged particles ( _ @xmath17 analysis _ ) @xcite b. analysis of @xmath18 and @xmath10 mesons identified via @xmath4 measurements in the tpcs ( _ @xmath4 analysis at low momentum _ ) @xcite and c. analysis of @xmath18 and @xmath10 mesons identified via time - of - flight and @xmath4 measurements in the tof - f and tpcs , respectively ( _ @xmath19 analysis _ ) @xcite .
each analysis yields fully corrected pion spectra with independently calculated statistical and systematic errors .
the spectra were compared in overlapping phase - space domains to check their consistency .
complementary domains were combined to reach maximum acceptance .
the agreement between the spectra obtained by different methods is , in general , better than 10% .
note , that data points in the same ( @xmath20 , @xmath21 ) bin from different analysis methods are statistically correlated as they result from the analysis of the same data set . in order to obtain the final spectra consisting of statistically uncorrelated points the measurement with the smallest total error was selected .
the analysis was based on a sample of 521 k events selected from the total sample of 667 k of registered and reconstructed events .
this criterion essentially removes a contamination by interactions upstream of the target . for the event sample with the target removed
, the selection reduces the number of events from 46 k to 17 k.
this section presents results on inelastic and production cross sections as well as on differential spectra of @xmath18 and @xmath10 mesons in p+c interactions at 31 gev / c .
the total inelastic cross section is measured to be @xcite @xmath22 the production cross section was calculated from the inelastic cross section by subtracting the quasi - elastic contribution @xcite .
the result is @xcite : @xmath23 the production cross section is compared to previous measurements in fig .
[ fig : inelastic ] . ) mb . ]
the @xmath18 and @xmath10 spectra presented in this section refer to pions produced in strong and electromagnetic processes in p+c interactions at 31 gev / c .
the spectra are presented as a function of particle momentum in ten intervals of the polar angle .
both quantities are calculated in the laboratory system .
the chosen binning takes into account the available statistics of the 2007 data sample , detector acceptance and particle production kinematics .
the final spectra are plotted in figs .
[ pion_minus_model ] and [ pion_plus_model ] . for the purpose of a comparison of the data with model predictions
the spectra were normalized to the mean @xmath24 multiplicity in all production interactions by dividing by @xmath25 .
this avoids uncertainties due to the different treatment of quasi - elastic interactions in models as well as problems due to the absence of predictions for inclusive cross sections .
this work presents inelastic and production cross sections as well as positively and negatively charged pion spectra in p+c interactions at 31 gev / c on the 2007 thin ( 4% interaction length ) carbon target .
these data are essential for precise predictions of the neutrino flux for the t2k long baseline neutrino oscillation experiment in japan .
furthermore , they provide important input to improve hadron production models needed for the interpretation of air showers initiated by ultra high energy cosmic particles . a much larger data set with both the thin and the t2k replica carbon targets was recorded in 2009 and 2010 and is presently being analysed .
this will lead to results of higher precision for pions and extend the measurements to other hadron species such as charged kaons , protons , @xmath26 and @xmath27 .
analysis of the data collected with the t2k replica target during the 2007 run is in progress @xcite .
the new data will allow a further significant reduction of the uncertainties in the prediction of the neutrino flux in the t2k experiment .
this work was supported by the following funding agencies : the hungarian scientific research fund ( grants otka 68506 and 79840 ) , the polish ministry of science and higher education ( grants 667/n - cern/2010/0 , n n202 1267 36 , n n202 287838 ( pbp 2878/b / h03/2010/38 ) , dwm/57/t2k/2007 ) , the federal agency of education of the ministry of education and science of the russian federation ( grant rnp 2.2.2.2.1547 ) , the russian academy of sciences and the russian foundation for basic research ( grants 08 - 02 - 00018 and 09 - 02 - 00664 ) , the ministry of education , culture , sports , science and technology , japan , grant - in - aid for scientific research ( grants 18071005 , 19034011 , 19740162 , 20740160 and 20039012 ) , the toshiko yuasa lab .
( france - japan particle physics laboratory ) , the institut national de physique nuclaire et physique des particules ( in2p3 , france ) , the german research foundation ( grant ga 1480/2 - 1 ) , the swiss national science foundation ( investigator - driven projects and sinergia ) and the swiss state secretariat for education and research ( force grants ) .
+ the authors also wish to acknowledge the support provided by the collaborating institutions , in particular , the eth zurich ( research grant th-01 07 - 3 ) , the university of bern and the university of geneva . finally , it is a pleasure to thank the european organization for nuclear research for a strong support and hospitality and , in particular , the operating crews of the cern sps accelerator and beam lines who made the measurements possible .
nabgrall _ et al . _
[ na61/shine collaboration ] , `` measurements of cross sections and charged pion spectra in proton - carbon interactions at 31 gev / c , '' submitted to : phys.rev.c .
[ arxiv:1102.0983 [ hep - ex ] ] . | as neutrino long baseline experiments enter a new domain of precision , the careful study of systematic errors due to poor knowledge of production cross sections for pions and kaons require more dedicated measurements for precise neutrino flux predictions .
the cosmic ray experiments require dedicated hadron production measurements to tune simulation models used to describe air shower profiles . among other goals ,
the na61/shine ( sps heavy ion and neutrino experiment ) experiment at the cern sps aims at precision measurements ( 5% and below ) for both neutrino and cosmic ray experiments : those will improve the prediction of the neutrino flux for the t2k experiment at j - parc and the prediction of muon production in the propagation of air showers for the auger and kascade experiments .
na61/shine took data during a pilot run in 2007 and in 2009 and 2010 with different carbon targets .
the na61/shine set - up and spectra for positive and negative pions obtained with the 2007 thin ( 4% interaction length ) carbon target data are presented @xcite . |
piecewise - smooth systems are currently being utilized in a wide variety of fields to model physical systems involving a discontinuity or sudden change .
examples include electrical circuits with a switching component such as a diode or a transistor @xcite , vibro - impacting systems and systems with friction @xcite , and optimization in economics @xcite . in this paper
we study discrete - time , systems that are everywhere continuous .
a map @xmath0 is _ continuous _ if @xmath1 is everywhere continuous but non - differentiable on codimension - one boundaries called _ s_. the collision of an invariant set with a under parameter variation of ( [ eq : genmap ] ) may produce a bifurcation that can not occur in smooth systems .
when the invariant set is a fixed point and a bifurcation occurs , the bifurcation is called a _ _ .
border - collision bifurcations may be analogous to familiar smooth bifurcations , such as saddle - node and period - doubling bifurcations
. alternatively they may be very complex .
for instance a chaotic attractor may be generated at a , even if the map is only one - dimensional @xcite . in two or more dimensions ,
a complete classification of all possible dynamical behavior local to a is yet to be determined , see for instance @xcite . at a a fixed point , @xmath2 ,
lies on a . due to the lack of differentiability the stability multipliers associated with @xmath2 are not well - defined .
however , the locally divides phase space into regions where the map has different smooth components ; by continuity @xmath2 is a fixed point of each component and , restricted to each component , associated multipliers of @xmath2 are well - defined .
typically @xmath2 will be a hyperbolic fixed point of each component .
a special situation arises when @xmath2 is non - hyperbolic in at least one component . under the assumption that the occurs at a smooth point on a single , there are exactly two smooth map components ( half - maps ) , in some neighborhood .
consider the situation that @xmath2 is hyperbolic in one half - map but non - hyperbolic in the other half - map .
then there are three codimension - two cases to consider ; namely that on the non - hyperbolic side , @xmath2 has an associated multiplier 1 , a multiplier @xmath3 or a complex conjugate pair of multipliers , @xmath4 .
these correspond to the simultaneous occurrence of a with a saddle - node bifurcation , a period - doubling bifurcation and a neimark - sacker bifurcation , respectively . unlike a generic , in each codimension - two case linear terms of the half - maps are insufficient to describe all local dynamical behavior .
for the first case , generically a locus of saddle - node bifurcations emanates from the codimension - two , which in a two - parameter bifurcation diagram is a curve tangent to a locus of s @xcite ( see @xcite for the continuous - time analogue ) .
the third case may be very complicated and is left for future work .
the precise nature of local bifurcations may depend on the rationality of @xmath5 .
a recent analysis of these situations in a general map is given in @xcite .
the purpose of the present paper is to determine generic dynamical behavior of ( [ eq : genmap ] ) local to the codimension - two of the second case .
the coincidence of border - collision and period - doubling has been observed in real - world systems . recall that a corner collision in a filippov system provides an example of a in a map @xcite . in @xcite the authors describe the coincidence of a corner collision with period - doubling in a filippov model of a dc / dc power converter .
grazing - sliding ( which also corresponds to a , continuous poincar map , see @xcite ) and period - doubling have been seen to occur simultaneously in a model of a forced , dry - friction oscillator @xcite . in the next section
we introduce a piecewise form for ( [ eq : genmap ] ) suitable for the ensuing analysis and compute its fixed points . in [ sec : unfold ] we unfold the codimension - two situation for the case that the map is one - dimensional . here
all local dynamical behavior is determined .
we find there are essentially six distinct , generic , unfolding scenarios though each exhibit the same basic bifurcation curves .
when the map is of a dimension greater than one we prove that the same basic bifurcation curves exist but do not classify all possible local dynamics , [ sec : highd ] .
finally [ sec : conc ] gives conclusions .
we wish to explore local dynamical behavior near an arbitrary of ( [ eq : genmap ] ) .
we restrict ourselves to a neighborhood in both phase space and parameter space of the within which we assume the occurs on an isolated @xmath6 and that away from the , @xmath7 is @xmath8 .
then , via coordinate transformations similar to those given in @xcite , we may assume the occurs at the origin , @xmath9 , when a parameter , @xmath10 , is zero , and that to order @xmath11 the is simply the plane @xmath12 .
in this paper we are not concerned with effects due to a nonsmooth ( for studies of various systems involving a non - differentiable , see , for instance , @xcite ) .
for this reason we assume @xmath11 is sufficiently large to not affect local dynamics and for simplicity assume the is exactly the plane @xmath12 . for convenience
we denote the first component of the @xmath13-dimensional vector , @xmath14 , by @xmath15 , i.e. @xmath16 let @xmath17 be a second map parameter , independent of @xmath10 .
then we are interested in local dynamics of @xmath18 where @xmath19 is @xmath8 for @xmath20 . here
@xmath21 and @xmath22 and @xmath23 are @xmath24 matrices which , by the assumption of continuity of ( [ eq : pwsmap ] ) , are identical in their last @xmath25 columns . throughout this paper
we use the notation @xmath26 [ @xmath27 to denote terms that are order @xmath28 or larger [ larger than order @xmath28 ] in all variables and parameters of a given expression .
we refer to @xmath30 as the left half - map and @xmath31 as the right half - map .
let us first determine fixed points of ( [ eq : pwsmap ] ) . assume for the moment that for some @xmath17 , 1 is not an eigenvalue of each @xmath32 .
then near @xmath33 , each half - map , @xmath34 , has a unique fixed point @xmath35 recall that the adjugate of any matrix @xmath36 , @xmath37 , obeys @xmath38 @xcite . since @xmath22 and @xmath23 can differ in only the first column , it follows that the first row of @xmath39 and of @xmath40 are identical .
we denote this row by @xmath41 : @xmath42 multiplication of ( [ eq : xstar ] ) by @xmath43 on the left yields @xmath44 where @xmath45 denotes the first component of @xmath46 .
the point @xmath47 is a fixed point of the map ( [ eq : pwsmap ] ) and said to be _ admissible _ (
that is , a fixed point of ( [ eq : pwsmap ] ) ) whenever @xmath48 , otherwise it is said to be _
virtual_. similarly @xmath49 is admissible exactly when @xmath50 .
the condition @xmath51 ensures that the parameter @xmath10 `` unfolds '' the border - collision ; we will assume this nondegeneracy condition holds .
let us briefly review feigin s classification of s by the relative admissibility of fixed points and two - cycles ( refer to @xcite for further details ) .
if @xmath52 and @xmath53 are both nonsingular and ( [ eq : rhobnonzero ] ) is satisfied , for small @xmath10 each @xmath46 is admissible when @xmath54 and for one sign of @xmath10 .
let @xmath55 [ @xmath56 denote the number of real eigenvalues of @xmath32 that are greater than 1 [ less than @xmath3 ] . by ( [ eq : sstar ] ) ,
if @xmath58 is even , @xmath47 and @xmath49 are admissible for different signs of @xmath10 . in this case as @xmath10 is varied through zero a single fixed point effectively _
persists_. if instead @xmath58 is odd , @xmath47 and @xmath49 are admissible for the same sign of @xmath10 ; at @xmath54 the two fixed points collide and annihilate in a _ nonsmooth fold bifurcation_. furthermore if @xmath59 is nonsingular , @xmath3 is not an eigenvalue of @xmath60 and @xmath61 , and @xmath62 is odd , then a two - cycle of ( [ eq : pwsmap ] ) ( that collapses to the origin as @xmath63 ) is admissible for one sign of @xmath10 , whereas if @xmath62 is even , no such two - cycle exists . when @xmath54 , the origin is a fixed point of ( [ eq : pwsmap ] ) that lies on the , @xmath64 . for small @xmath65 , stability of , and dynamics local to , the fixed point , @xmath46 , are determined by the eigenvalues of @xmath32 . in nondegenerate situations the eigenvalues of the matrices @xmath32 completely determine dynamics local to the at @xmath54 .
a special situation arises whenever either @xmath60 or @xmath61 , has an eigenvalue that lies on the unit circle . without loss of generality
we may assume the former matrix has an eigenvalue on the unit circle when @xmath66 .
in this situation a local smooth bifurcation may occur for the left half - map , @xmath67 , at @xmath68 .
the nature of the smooth bifurcation is determined by nonlinear terms , consequently such terms are important to the .
the remainder of this paper is an analysis of the case that @xmath3 is an eigenvalue of @xmath69 .
we begin by treating the one - dimensional case , that is , @xmath70 . then the map ( [ eq : pwsmap ] )
may be written as @xmath71 we assume that the two smooth components of the map are at least @xmath72 ( i.e. @xmath73 ) and write them as @xmath74 the map ( [ eq:1dmappd ] ) has a at @xmath54 .
we assume that the codimension - two situation of interest occurs at @xmath68 , that is @xmath75 . in order for parameters
@xmath10 and @xmath17 to unfold the bifurcation therefore we require @xmath76 ( by ( [ eq : rhobnonzero ] ) ) and @xmath77 . for simplicity
we assume appropriate scalings upon @xmath10 and @xmath17 have been performed such that these values are both 1 as stated in the following theorem .
+ consider ( [ eq:1dmappd ] ) with ( [ eq:1dcoeffs ] ) and suppose 1 .
[ it : pdsing ] @xmath75 ( singularity condition ) , 2 .
[ it : pdndsw ] @xmath78 ( border - collision non - degeneracy condition ) , 3 .
[ it : pdtrans ] @xmath79 ( transversality condition ) .
let @xmath80 then there exists @xmath81 such that the left half - map restricted to the neighborhood @xmath82 has a unique fixed point given by a @xmath8 function , @xmath83 , and there exist unique , @xmath84 functions @xmath85 that satisfy @xmath86 such that 1 .
[ it : resh1 ] when @xmath87 , @xmath88 has an associated multiplier of @xmath3 , and if @xmath89 the curve @xmath90 corresponds to a locus of admissible , period - doubling bifurcations of @xmath47 , 2 .
[ it : resh2 ] the origin belongs to a period - two orbit of @xmath30 on @xmath91 that is admissible when @xmath87 , 3 . [ it : resfig ] if @xmath92 and @xmath89 , then fixed points and two - cycles of ( [ eq:1dmappd ] ) exist in the sectors shown in fig .
[ fig : bspd ] , 4 .
[ it : resncycles ] if @xmath93 or @xmath94 or @xmath95 , ( [ eq:1dmappd ] ) has no @xmath96-cycles for any @xmath97 in @xmath98 , 5 .
[ it : reschaos ] if @xmath99 and @xmath87 and @xmath100 , then ( [ eq:1dmappd ] ) exhibits chaos in @xmath98 .
[ th : c21dpd ] ) for different values of @xmath101 and @xmath102 , see theorem [ th : c21dpd ] .
blue curves correspond to loci of period - doubling bifurcations ; black curves correspond to border - collisions and are shown dashed when no topological change occurs at the border - collision .
insets show phase portraits with circles used to denote fixed points and triangles used to denote points on two - cycles .
these are blue when the solutions are stable and red otherwise .
the small vertical lines denote the .
chaotic dynamics occurs only for parameter values in the shaded regions ; indeed , periodic solutions of a period higher than two only occur in these regions .
the bifurcation diagram shown in fig .
[ fig : bifdiag1dex ] is taken along the dash - dot line segment of panel f. [ fig : bspd ] , width=566,height=604 ] ) when @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and there are no additional higher order terms .
the value of @xmath17 is fixed at @xmath108 .
the parameters considered correspond to the dash - dot line segment shown in fig . [ fig : bspd ] .
stable [ unstable ] fixed points and two - cycles are indicated by solid [ dashed ] curves .
a supercritical period - doubling bifurcation occurs at @xmath109 ; a of the period - doubled solution occurs at @xmath110 .
when @xmath111 there exists a chaotic attracting set . when @xmath94 there is no local attractor .
[ fig : bifdiag1dex ] , width=317,height=264 ] a proof of theorem [ th : c21dpd ] is given in appendix [ sec : proof1d ] . to facilitate the extension of theorem [ th : c21dpd ] to higher dimensions in [ sec :
highd ] , the theorem is stated in a manner that yields a partial result when the non - degeneracy condition on the nonlinear terms , @xmath89 , is not satisfied .
the sign of @xmath102 determines the criticality of the period - doubling bifurcations .
[ fig : bspd ] shows the six different bifurcation scenarios depending on the value of @xmath101 and the sign of @xmath102 that are predicted by the theorem . in each case the nature of the at @xmath54 for small @xmath112 may be determined by refering to the well - understood , one - dimensional , map @xmath113 see for instance @xcite .
for example , along the positive @xmath17-axis in fig .
[ fig : bspd]-a , we have @xmath114 and @xmath115 , thus by feigin s classification a single fixed point persists and a two - cycle is created that coexists with the left fixed point and is stable . in panels e and f , along the negative @xmath17-axis we have @xmath116 and @xmath117 . at points just to the left of the negative @xmath17-axis ,
both fixed points and a two - cycle are admissible and unstable .
forward orbits are attracted to a chaotic solution created at the at @xmath54 .
unstable high period orbits are also generated at the ; these are of even period . to further illustrate the unfolding , fig .
[ fig : bifdiag1dex ] shows a bifurcation diagram corresponding to a one - parameter slice of fig .
[ fig : bspd]-f .
the chaotic attracting set collapses to the origin at @xmath54 and to a two - cycle at @xmath118 .
two - cycles created in s at @xmath54 consist of one point in each half - plane . in contrast , two - cycles created in period - doubling bifurcations along @xmath90 lie entirely in the left half - plane . along @xmath91
these two orbits coincide and are given explicitly by @xmath119 .
the second iterate of ( [ eq:1dmappd ] ) can be transformed into a map of the form ( [ eq : pwsmap ] ) along @xmath91 , see ( [ eq : f2proofpd ] ) .
the resulting left and right period - two maps have multipliers @xmath120 and @xmath121 respectively . by feigin s classification ,
four - cycles are born at @xmath91 exactly when @xmath99 , i.e. in panels e and f of fig .
[ fig : bspd ] , though these are not shown .
we now analyze the codimension - two scenario in an arbitrary number of dimensions , i.e. @xmath122 .
recall that for smooth systems the concept of dimension reduction by center manifold analysis allows bifurcations in systems of any number of dimensions to be transformed to their low dimensional normal forms .
unfortunately this technique can not be applied here because ( [ eq : pwsmap ] ) is not differentiable in a neighborhood of the bifurcation .
however we can compute a center manifold of the @xmath8 , left half - map . assuming certain nondegeneracy conditions
, we will show that the restriction of the left half - map to this center manifold yields a one - dimensional map that has the same form as the left half - map of ( [ eq:1dmappd ] ) and satisfies the conditions ( [ it : pdsing])-([it : pdtrans ] ) of theorem [ th : c21dpd ] .
consequently the aspects of theorem [ th : c21dpd ] that incorporate only the left half - map extend immediately to the same scenario in higher dimensions . that is , there exists a curve @xmath90 along which period - doubling bifurcations occur and a curve @xmath91 along which the two - cycle created at the period - doubling bifurcation collides with the .
the drawback of this approach is that we fail to describe dynamical behavior of ( [ eq : pwsmap ] ) that occurs on both sides of the .
let us state our main result : + consider the piecewise-@xmath8 , continuous map ( [ eq : pwsmap ] ) with ( [ eq : fiform ] ) on @xmath123 and assume @xmath124 and @xmath125 .
suppose that near @xmath126 , @xmath127 has an eigenvalue @xmath128 with an associated eigenvector , @xmath129 .
in addition , suppose 1 . [
it : pdsing2 ] @xmath130 is of algebraic multiplicity one and @xmath69 has no other eigenvalues on the unit circle , 2 .
[ it : pdndsw2 ] @xmath131 ( where @xmath41 is given by ( [ eq : varrhodef ] ) ) , 3 .
[ it : pdtrans2 ] @xmath132 , 4 .
[ it : pdcenterman2 ] the first element of @xmath133 is nonzero , thus by scaling assume @xmath134 , 5 .
[ it : pdndrhm2 ] @xmath135 .
then , in the extended coordinate system @xmath136 , there exists a @xmath84 three - dimensional center manifold , @xmath137 , for the left half - system that passes through the origin and is not tangent to the at this point .
furthermore , @xmath138 , described in the coordinate system @xmath139 , is given by the left half - map of ( [ eq:1dmappd ] ) with `` hatted variables '' and the conditions ( [ it : pdsing])-([it : pdtrans ] ) of theorem [ th : c21dpd ] will be satisfied . if @xmath89 in theorem [ th : c21dpd ] ( where @xmath102 is computed for @xmath138 ) , then along a @xmath140 curve , @xmath141 , the unique fixed point of the left half - map near the origin undergoes a period - doubling bifurcation that is admissible when @xmath142 , and along a @xmath140 curve , @xmath143 , the period - doubled solution collides with the .
these two curves intersect and are tangent at @xmath144 .
moreover , a unique two - cycle consisting of one point on each side of the exists for small @xmath10 and @xmath17 .
this cycle is admissible exactly when @xmath145 and @xmath146 .
[ th : c2pd ] a proof is given in appendix [ sec : proofnd ] . conditions ( [ it : pdsing2])-([it : pdtrans2 ] ) are analogous to the first three conditions of theorem [ th : c21dpd ] . condition ( [ it : pdcenterman2 ] ) is a non - degeneracy condition that ensures the period - doubled cycles collide with the in the generic manner .
if this condition is not satisfied , a higher codimension situation that involves linear separation between @xmath147 and @xmath148 may arise .
finally condition ( [ it : pdndrhm2 ] ) is necessary to ensure that non - degenerate , two - cycles are created at @xmath54 .
theorem [ th : c2pd ] essentially predicts the same basic bifurcation structure as theorem [ th : c21dpd ] . the relative position of @xmath149 and @xmath150 and the criticality of the period - doubling bifurcation
are determined by the sign of @xmath102 . in one dimension
, @xmath102 is a simple function of coefficients of nonlinear terms of the left half - map ( [ eq : c0pd ] ) . to obtain the value of @xmath102 in higher dimensions ,
one may derive an expression for the restriction of the left half - map to the center manifold , then apply the one dimensional result .
this is done for an example below .
an explicit expression for @xmath102 in @xmath13 dimensions is probably too complicated to be of practical use . to illustrate theorem [ th : c2pd ]
consider the following map on @xmath151 using @xmath152 : @xmath153 this map is easily rewritten in the piecewise form ( [ eq : pwsmap ] ) with ( [ eq : fiform ] ) and as we will show ( [ eq : pdmapex ] ) satisfies the assumptions of theorem [ th : c2pd ] .
when @xmath54 the origin is a fixed point of ( [ eq : pdmapex ] ) which lies on the , @xmath64 .
consequently the @xmath17-axis corresponds to a locus of border - collision bifurcations .
[ fig : bspdex ] shows a two - dimensional bifurcation diagram of ( [ eq : pdmapex ] ) near @xmath68 .
period - doubling bifurcations occur along @xmath90 and the period - doubled solutions undergo border - collision along @xmath91 ; these curves are given explicitly by @xmath154 ( actually @xmath155 ) .
the curves are tangent at @xmath68 as predicted by theorem [ th : c2pd ] . near this point
the dynamics resemble fig .
[ fig : bspd]c although there is an additional unstable two - cycle present that does not undergo bifurcation at @xmath68 .
this two - cycle collides with the stable two - cycle born at @xmath54 for small @xmath156 along a locus of saddle - node bifurcations ( shown red in fig .
[ fig : bspdex ] ) .
the saddle - node locus emanates from the @xmath17-axis at @xmath157 which is where the stable two - cycle loses stability .
also a cusp bifurcation of period - two orbits occurs at @xmath158 .
a second simultaneous border - collision and period - doubling bifurcation occurs on the @xmath17-axis at @xmath159 .
as predicted by theorem [ th : c2pd ] a locus of period - doubling bifurcations and border - collision bifurcations of the period - doubled solution emanate from the codimension - two point and are tangent to one another here .
locally the bifurcation set resembles fig .
[ fig : bspd]a . map ( [ eq : pdmapex ] ) .
border - collision bifurcations of fixed points and two - cycles are indicated by black curves which are dashed when no bifurcation occurs in a topological sense .
blue curves correspond to period - doubling bifurcations of a fixed point ; red curves correspond to saddle - node bifurcations of a two - cycle . as in fig .
[ fig : bspd ] , schematics showing fixed points and two - cycles are included .
dynamics local to the codimension - two points @xmath126 and @xmath160 are described by theorem [ th : c2pd ] .
[ fig : bspdex ] , width=408,height=340 ] we now perform calculations to verify the details of theorem [ th : c2pd ] for the map ( [ eq : pdmapex ] ) near @xmath68 . here
$ ] , @xmath162^{\sf t}$ ] and @xmath163 , thus @xmath164 and therefore @xmath147 corresponds to admissible solutions when @xmath94 , matching fig .
[ fig : bspdex ] .
also , @xmath165 and @xmath166 , hence by the last statement of theorem [ th : c2pd ] , two - cycles born on @xmath150 with points on each side of the are admissible below @xmath150 , in agreement with fig .
[ fig : bspdex ] . using a series expansion
( see also the proof of theorem [ th : c2pd ] ) , one finds that the left half - map of ( [ eq : pdmapex ] ) restricted to the center manifold is given by @xmath167 here the coefficient @xmath168 , given in theorem [ th : c21dpd ] , is equal to @xmath169 . also @xmath170 .
thus @xmath171 and @xmath172 matching the explicit results given above .
a two - parameter bifurcation diagram provides a useful and concise picture of the behavior of a dynamical system near a codimension - two point .
loci or curves in such diagrams correspond to codimension - one phenomena .
the intersection of two or more bifurcation curves at a codimension - two point occurs in a manner that is generally determined by the corresponding codimension - one bifurcations .
we have shown that for a general , continuous map , loci of border - collision and period - doubling bifurcations of a single fixed point generically intersect non - tangentially ( unlike the analogous intersection of a and a saddle - node bifurcation @xcite ) .
moreover the period - doubled solution undergoes border - collision along a locus that emanates from the codimension - two point and is tangent to the period - doubling curve here .
we have determined all local dynamical phenomena that occurs when this scenario arises generically in a one - dimensional system .
in particular we have shown exactly when chaotic dynamics may be generated . in higher dimensions
s near the codimension - two point may in addition generate quasiperiodic solutions such as invariant circles or other more complicated orbits .
the theorem is proved in five steps .
first , the function @xmath149 is calculated by finding where @xmath47 has an associated multiplier of @xmath3 .
second , @xmath150 is calculated from an expression for @xmath173 .
third , an explicit computation of the , second iterate map allows the of the two - cycle that occurs along @xmath91 to be classified .
finally , in steps four and five we prove parts ( [ it : resncycles ] ) and ( [ it : reschaos ] ) respectively . since @xmath30 ( [ eq:1dcoeffs ] ) is @xmath8 and @xmath73 , we can write @xmath174 where we have used hypotheses of the theorem to simplify the coefficients .
* step 1 : * _ compute the function @xmath149 which corresponds to the existence of a fixed point of the left half - map with multiplier @xmath3 . _
+ by the implicit function theorem the left half - map has the fixed point @xmath175 which is a @xmath8 function of @xmath10 and @xmath17 and locally satisfies @xmath176 . by a second order expansion
, it is determined that the scalar coefficients , @xmath177 , have the values @xmath178 the multiplier associated with @xmath179 is found by substituting ( [ eq : xstarpd ] ) into @xmath180 ( see ( [ eq:1dcoeffs ] ) ) .
we obtain @xmath181 the implicit function theorem implies that @xmath182 is a @xmath84 function that locally satisfies @xmath183 , where @xmath184 is found by substituting ( [ eq : h1proofpd ] ) into ( [ eq : xstarmultpd ] ) , @xmath185 when @xmath89 , the existence of period - doubling bifurcations along @xmath90 for small @xmath10 is verified by checking the three conditions of the standard theorem , see for instance @xcite : * step 2 : * _ compute the function @xmath150 which corresponds to the existence of a period - two orbit of the left half - map at @xmath9 . _
+ to determine @xmath150 , we compute the @xmath8 function @xmath189 which may be written as @xmath190 where @xmath191 is @xmath84 . substitution of ( [ eq : coeffspd ] ) into ( [ eq : gpd ] ) produces @xmath192 the implicit function theorem implies that @xmath193 is a @xmath84 function that locally satisfies @xmath194 . by substituting ( [ eq : h2pd ] ) into ( [ eq : g2pd ] ) we deduce @xmath195 subtracting ( [ eq : l1pd ] ) from ( [ eq : l2pd ] ) produces ( after algebraic simplification ) @xmath196 which proves ( [ eq : h2pdth ] ) in the statement of the theorem .
* step 3 : * _ determine all two - cycles to verify the phase portraits in fig .
[ fig : bspd ] . _
+ for small , fixed @xmath87 , the two - cycle generated in a period - doubling bifurcation at @xmath90 undergoes a when it collides with the at @xmath91 .
the stability and relative admissibility of two - cycles emanating from this is found by determining a map of the form ( [ eq:1dmap ] ) that describes the bifurcation .
such a map may be obtained by computing the second iterate of ( [ eq:1dmappd ] ) and replacing @xmath17 with the new parameter @xmath197 which controls the border - collision . near @xmath198 , when @xmath87 period - two orbits are guaranteed to have one negative point , so we compute @xmath199 and @xmath200 which leads to @xmath201 a fixed point , @xmath202 , of the left half - map of ( [ eq : f2proofpd ] ) corresponds to a two - cycle of ( [ eq:1dmappd ] ) with both points negative - valued , when admissible . by ( [ eq : f2proofpd ] ) , if @xmath203 [ @xmath204 , then @xmath202 is unstable and admissible when @xmath205 [ stable and admissible when @xmath206 .
similarly , a fixed point , @xmath207 of the right half - map of ( [ eq : f2proofpd ] ) corresponds to a two - cycle of ( [ eq:1dmappd ] ) comprised of two points with different signs , when admissible . if @xmath208 , then @xmath207 is stable , otherwise it is unstable .
furthermore if @xmath209 [ @xmath210 , then @xmath207 is admissible when @xmath211 [ @xmath205 ] .
lastly , since @xmath213 , the fixed point , @xmath207 , is unique for small @xmath10 and @xmath17 . therefore the two - cycle created at @xmath54 collides with the at @xmath91 .
these statements verify all two - cycles shown in fig .
[ fig : bspd ] . * step 4 : * _ show that @xmath214 , ( [ eq:1dmappd ] ) , has no @xmath96-cycles with @xmath97 , verifying part ( [ it : resncycles ] ) . _
+ except when @xmath215 our proof is founded on the knowledge that a one - dimensional map can not have an @xmath96-cycle with @xmath97 contained in an interval on which the map is monotone . for the special case @xmath215 we used several additional logical arguments including the contraction mapping theorem .
* case ii : * @xmath215 .
+ first we construct an interval containing the origin on which @xmath214 is forward invariant .
since the two components of @xmath214 are differentiable , there exists @xmath219 such that @xmath220 $ ] , @xmath221
let @xmath222 , let @xmath223 $ ] and assume @xmath224 $ ] .
then @xmath225 because for any @xmath226 , we may assume @xmath227 when calculating a lower bound for @xmath228 : @xmath229 , and we may assume @xmath230 when calculating an upper bound : @xmath231 .
if @xmath232 then the image of any @xmath230 under @xmath214 is positive and since the right half - map is contracting and the slope of the left half - map is near @xmath3 , @xmath233 is a contraction on @xmath234 , indeed @xmath235 . by the contraction mapping theorem , iterates of @xmath233 approach a unique fixed point , which must be @xmath49
, therefore @xmath214 can not have an @xmath96-cycle with @xmath97 on @xmath98 ( [ eq : nbd ] ) .
so we now assume @xmath87 . recall that when @xmath91 , @xmath236 , i.e. @xmath237 , and when @xmath95 , @xmath238 . therefore if @xmath239 , see fig .
[ fig : a0rgen]-a , the interval @xmath240 $ ] is forward invariant . since @xmath214 is monotone on the interval @xmath241
\ ; , \label{eq : jpd}\ ] ] it contains no @xmath96-cycles with @xmath97 here . on the other hand if @xmath100 as in fig .
[ fig : a0rgen]-b , the forward orbit of any point in @xmath240 $ ] is either contained in this interval ( where @xmath214 is monotone ) or enters @xmath242 $ ] and approaches the stable two - cycle ( this may be shown formally by the contraction mapping theorem ) . in any case
, an @xmath96-cycle of @xmath214 with @xmath97 can not include a point in @xmath243 $ ] .
suppose for a contradiction @xmath214 has an @xmath96-cycle with @xmath97 , call it @xmath244 \setminus j$ ] .
we now show that the restriction of @xmath233 to this @xmath96-cycle is a contraction giving a contradiction according to the contraction mapping theorem .
specifically we will argue that for any @xmath245 and @xmath246 , @xmath247 there are several cases to consider .
for example , suppose that @xmath248 , @xmath249 and @xmath250 .
then @xmath251 thus @xmath252 .
also @xmath253 thus @xmath254 .
subtracting these inequalities leads to ( [ eq : contract49 ] ) .
the remaining cases can be shown similarly .
* case iii : * @xmath256 .
+ _ case iiia : _ first suppose that @xmath232 .
there exists @xmath257 such that whenever @xmath258 $ ] and @xmath259 $ ] , @xmath214 is decreasing on @xmath260 $ ] and increasing on @xmath261 $ ] . since @xmath232 ,
@xmath262 $ ] , @xmath263 .
thus any periodic solution of @xmath214 in @xmath264 $ ] must lie entirely in @xmath261 $ ] .
but @xmath214 is increasing on this interval thus has no @xmath96-cycles with @xmath97 here .
+ _ case iiib : _ now suppose @xmath87 and @xmath239 .
there exists @xmath265 such that whenever @xmath266 and @xmath267 $ ] , @xmath214 is decreasing on @xmath268 $ ] and increasing on @xmath269 $ ] , see fig .
[ fig : a0rgen]-a .
suppose for a contradiction , @xmath214 has an @xmath96-cycle with @xmath97 on @xmath270 $ ] .
such an @xmath96-cycle must enter both @xmath271 and @xmath272 $ ] , and so it includes a point @xmath273 $ ] with @xmath274 .
but @xmath228 lies in @xmath275 , ( [ eq : jpd ] ) , and since @xmath239 , @xmath276 , hence @xmath277 .
thus the forward orbit of @xmath14 can not return to @xmath272 $ ] which contradicts the assumption that @xmath14 belongs to an @xmath96-cycle .
+ _ case iiic : _ finally suppose @xmath87 , @xmath100 and @xmath278 . then there exists @xmath279 such that whenever @xmath280 , @xmath281 and @xmath282 $ ] , @xmath283 where the particular non - symmetric @xmath101-dependent bounds on the slopes have been chosen to simplify the subsequent analysis .
suppose for a contradiction @xmath214 has an @xmath96-cycle with @xmath97 in @xmath284 $ ] .
as before , such an @xmath96-cycle must include a point @xmath285 $ ] with @xmath286 .
again @xmath287 , but here @xmath100 , thus the forward orbit of @xmath14 may return to the right of the origin but must enter the interval @xmath288 $ ] , see fig .
[ fig : a0rgen]-b . notice @xmath289
. then by evaluating the first - order expansion of @xmath30 about @xmath290 at @xmath291 , we arrive at @xmath292 , therefore @xmath293
. then @xmath294 , @xmath295 .
thus @xmath296 is a contraction mapping and therefore iterates of @xmath214 that enter @xmath297 can not belong to an @xmath96-cycle with @xmath97 . +
finally let @xmath298 , then the result , ( [ eq:1dmappd ] ) , is proved . ) , @xmath233 , near @xmath9 when @xmath87 , @xmath100 and @xmath99 .
the slope of this map is approximately unity left of the origin and approximately @xmath299 right of the origin . as a consequence ,
see the text , the map exhibits chaos in @xmath300 $ ] . [
fig : f2pdschem ] , width=272,height=226 ] * step 5 : * _ prove the existence of chaos when @xmath87 , @xmath100 and @xmath99 . _ + we first construct a trapping set , @xmath301 , for @xmath233 .
let @xmath302 . since @xmath87 , by ( [ eq : f2proofpd ] ) , @xmath303 .
since @xmath100 , @xmath304
. therefore @xmath305 consequently , @xmath306 let @xmath307 \ ; , \nonumber\ ] ] it follows , see fig .
[ fig : f2pdschem ] , that @xmath308 , @xmath309 .
but @xmath310 , thus @xmath311 , i.e. @xmath301 is a trapping set for @xmath233 .
furthermore , @xmath312 , is an attracting set for @xmath233 .
let @xmath313 with @xmath314 .
we now show that @xmath308 , @xmath315 with @xmath316 such that @xmath317 .
suppose otherwise .
then whenever @xmath316 , @xmath318 and thus @xmath319 since @xmath233 is increasing on @xmath320 , it suffices to consider @xmath321 . by combining ( [ eq : tmin ] ) and
( [ eq : f2jx ] ) we find @xmath322 which is a contradiction .
now consider the map @xmath323 .
note that @xmath233 has one critical point on @xmath301 , namely 0 .
thus @xmath324 is a critical point of @xmath325 if @xmath326 for some @xmath316 .
consequently @xmath325 has at most @xmath327 critical points , @xmath328 , and @xmath325 is differentiable on @xmath301 except at each @xmath328 . for
all @xmath329 , @xmath330 , @xmath331 let @xmath332 be the number of iterates , @xmath333 , that are positive .
since left of zero the slope of @xmath233 is near 1 and right of zero the slope is near @xmath299 , we have @xmath334 for any @xmath330 , for sufficiently small @xmath335 .
consequently @xmath323 is _ piecewise expanding _
@xcite . by the theorem of li and yorke @xcite , @xmath325 is chaotic on @xmath301
. thus @xmath214 is chaotic on @xmath336 .
the theorem is proved in three steps . in the first two steps a center manifold , @xmath137 , for the left half - map , @xmath30 ,
is constructed and the lowest order terms of the restriction of @xmath30 to @xmath137 are calculated . in the third step the two - cycle that has points in each half - plane
is computed explicitly .
* step 1 : * _ compute @xmath137 .
_ + it is convenient to extend the map to @xmath337 by including the parameters in the extended phase space , @xmath338 = \left [ \begin{array}{c } f^{(l)}(x;\mu,\eta ) \\ \mu \\
\eta \end{array } \right ] \;.
\nonumber\ ] ] then @xmath339 \ ; , \nonumber\ ] ] has a three - dimensional center space @xmath340 , \left [ \begin{array}{c } \varphi \\ 1 \\ 0 \end{array } \right ] , e_{n+2 } \right\ } \ ; , \nonumber\ ] ] where @xmath341 notice @xmath342 since @xmath343 , the center manifold , @xmath137 , may be expressed locally in terms of @xmath15 , @xmath10 and @xmath17 .
by the center manifold theorem , there exists a @xmath84 function , @xmath344 , such that on @xmath137 @xmath345 where @xmath346 * step 2 : * _ determine an expression for @xmath30 on @xmath137 and verify conditions ( [ it : pdsing])-([it : pdtrans ] ) of theorem [ th : c21dpd ] . _ + on @xmath137 @xmath347 where @xmath348 denotes all terms of @xmath30 that are nonlinear in @xmath14 . with the given `` hatted '' variables ( [ eq : sppd ] )
satisfies condition ( [ it : pdndsw ] ) of theorem [ th : c21dpd ] because @xmath349 which with @xmath350 , in view of ( [ eq : varphiproofpd1 ] ) , produces @xmath351 a similar verification of ( [ it : pdsing ] ) and ( [ it : pdtrans ] ) of theorem [ th : c21dpd ] is now given . by scaling @xmath352
, we may assume that for small @xmath17 , @xmath353 . since ( [ eq : hpd ] ) lacks a linear @xmath17 term , @xmath354 and consequently @xmath355 thus @xmath356 as required .
* step 3 : * _ compute the two - cycle that has points in each half - plane to verify the final statement of the theorem . _
+ we have @xmath357 by applying the implicit function theorem to ( [ eq : flfrproof ] ) in view of condition ( [ it : pdndrhm2 ] ) of the theorem we see that the two - cycle with points in each half - plane exists and is unique for small @xmath10 and @xmath17 .
when admissible , the point of this cycle in the right half - plane is given by a @xmath8 function @xmath358 then @xmath359 since @xmath360 and @xmath22 and @xmath23 are identical in their last @xmath25 columns , @xmath361 and @xmath362 share the same first row .
thus @xmath363 by theorem [ th : c21dpd ] , there exists a @xmath140 curve , @xmath143 , along which , @xmath364 .
furthermore , when @xmath145 , the two - cycle is admissible along this curve . if @xmath365 , then @xmath366 which confirms the final statement of the theorem .
i. sushko and l. gardini .
center bifurcation for a two - dimensional piecewise linear map . in t.
puu and i. sushko , editors , _ business cycle dynamics : models and tools .
_ , pages 4978 .
springer - verlag , new york , 2006 .
p. kowalczyk , m. di bernardo , a.r .
champneys , s.j .
hogan , m. homer , p.t .
piiroinen , yu.a .
kuznetsov , and a. nordmark .
two - parameter discontinuity - induced bifurcations of limit cycles : classification and open problems . , 16(3):601629 , 2006 . | we unfold the codimension - two simultaneous occurrence of a and a period - doubling bifurcation for a general , continuous map .
we find that , with sufficient non - degeneracy conditions , a locus of period - doubling bifurcations emanates non - tangentially from a locus of s. the corresponding period - doubled solution undergoes a along a curve emanating from the codimension - two point and tangent to the period - doubling locus here . in the case
that the map is one - dimensional local dynamics are completely classified ; in particular , we give conditions that ensure chaos .
[ theorem]corollary [ theorem]lemma * piecewise - smooth , continuous maps are used to model nonsmooth physical situations with discrete - time inputs , and also arise as poincar maps in a variety of , systems of ordinary differential equations .
frequently a period - doubling bifurcation is the cause for a critical change in dynamics of a map producing a physically important period - doubled solution .
a special situation arises in a , continuous map if this bifurcation occurs at a non - differentiable point in phase space .
we show that soon after its creation , the period - doubled solution undergoes a . beyond this bifurcation
the period - doubled solution may persist but the solution may also immediately become chaotic . * |
due to the wave nature of electrons and the confined geometries in mesoscopic systems , the interplay between interference and interaction becomes one of the central issues in mesoscopic physics .
preservation of quantum coherence in electron transport through an interacting regime has been manifested in the observed kondo effect in semiconductor quantum dot systems@xcite , and more explicitly in the conductance ab oscillation in the interference experiment with a quantum dot embedded in one arm of an aharonov - bohm(ab ) interferometer@xcite .
recently , the fano resonance has attracted much research interest as another important interference effect in mesoscopic systems .
the fano effect was first proposed as a result of the interference between resonant and non - resonant processes in the field of atomic physics@xcite .
it is found to be a ubiquitous phenomenon observed in a large variety of experiments including neutron scattering , atomic photoionization , raman scattering , and optical absorption .
one recent progress is the observation of the fano resonances in condensed matter systems , including an impurity atom on metal surface@xcite , single - electron transistor@xcite , quantum dot in ab interferometer@xcite . in this paper
we show the fano effect , which can be manifested by gate voltage dependence of the linear conductance , is also important for the electron transport through double quantum dots(dqds ) in parallel configuration . for electron tunneling through quantum dots
, it is well known that the strong on - site coulomb interaction leads to the kondo effect at low temperatures , so that the coexistence of the fano resonance with the kondo effect is expected to yield interesting transport phenomena .
electron transport through dqds with series@xcite and parallel@xcite configurations have been realized in experiments , through which studies on the molecular states of the double dots and also the interference effect are carried out .
most of theoretical studies@xcite are devoted to electron transport through dqds connected in series , while relatively little attention is paid for the parallel configuration case especially for the system in the kondo regime@xcite . for the dqd system with a parallel coupling , interference effect should play an important role .
thus , in order to understand the role of the fano effect , it is essential to take into account the coherence of the whole system .
a model of the electron transport through a closed ab interferometer containing two single level quantum dots , which assumes the electrons transport through quantum dots are in full coherence , has been investigated in ref.[15].interesting phenomena such as flux - dependent level attraction and interference induced suppression of conductance have been found .
but the effect of one - site coulomb interaction and the interdot tunneling havent been considered .
ghost fano resonance has also been observed in the study of electron transport through parallel dqds with interdot tunneling but no on - site coulomb interaction@xcite . in this paper
we shall investigate electron transport through parallel dqds(schematically plotted in fig.1 ) with interdot tunneling and on - site coulomb interaction using the finite-@xmath0 slave boson mean field theory(sbmft ) approach developed by kotliar and ruckenstein@xcite .
this formulation reproduces the results derived from the well known gutzwiller variation wave function at zero temperature , and therefore is believed to be a powerful tool to study strong correlation effect of electron systems .
the finite - u sbmft has already been applied to investigate electron transport through single quantum dot@xcite , dqds in series in the kondo regime @xcite and persistent current in mesoscopic ring@xcite , and found to give good quantitative results for the kondo effect on the linear conductance .
electron transport through parallel dqds with interdot tunneling and on - site coulomb interation can be described by the following anderson impurity model : @xmath3 where @xmath4 denote annihilation(creation ) operators for electrons in the leads(@xmath5 ) , and @xmath6 those of the single level state in the i - th dot ( @xmath7 ) .
u is the intra - dot coulomb interaction between electrons , @xmath8 is the interdot tunnel coupling , and @xmath9 is the tunnel matrix element between lead @xmath10 and dot @xmath11 .
we consider the symmetric coupling case with @xmath12 , where @xmath13 is the hybridization strength between the @xmath11-th dot and the lead @xmath10 . in the finite-@xmath0 slave boson approach@xcite , a set of auxiliary bosons @xmath14 are introduced for each dot , which act as projection operators onto the empty , singly occupied(with spin up and spin down ) , and doubly occupied electron states on the quantum dot , respectively .
the fermion operators @xmath15 are replaced by @xmath16 , with @xmath17 . in order to eliminate unphysical states ,
the following constraint conditions are imposed : @xmath18 , and @xmath19 .
therefore , the hamiltonian ( 1 ) can be rewritten as the following effective hamiltonian in terms of the auxiliary boson @xmath14 and the pesudo - fermion operators @xmath20 : @xmath21 where the constraints are incorporated by the lagrange multipliers @xmath22 and @xmath23 . the first constraint can be interpreted as a completeness relation of the hilbert space in each dot , and the second one equates the two ways of counting the fermion occupancy of a given spin@xcite . in the framework of the finite - u sbmft , the slave boson operators @xmath24 and the parameter @xmath25 are replaced by real c numbers . in this paper , we only consider the spin degenerate case without external magnetic field , so that all parameters are independent of the electron spin .
we can neglect the spin index @xmath26 in the parameters hereafter .
thus in the mean field approximation , the effective hamiltonian is given as @xmath27 where @xmath28 and @xmath29 represents the renormalized tunnel coupling between quantum dots and the renormalized tunnel amplitude between @xmath11-th quantum dot and the lead @xmath10 , respectively .
@xmath30 and @xmath31 can be regarded as the wave function renormalization factors in the quantum dots .
@xmath32 is the renormalized dot energy level and @xmath33 $ ] is an energy constant . within this mean field
effective hamiltonian ( 3 ) the current formula through the dqds is given as@xcite @xmath34t(\omega)\;,\\\ ] ] where the transmission probability @xmath35 $ ] , and @xmath36 , with @xmath37 .
the retarded / advanced green s functions(gf ) @xmath38 have @xmath39 matrix structures , which account for the double dot structure of the system .
the matrix elements of the retarded gf are defined in time space as @xmath40 . by applying the equation of motion method@xcite
, one can obtain the retarded gf explicitly as @xmath41 the advanced gf is given by @xmath42^\dagger$ ] . substituting the retarded / advanced gf to the formula of transmission probability , one obtains @xmath43 ^ 2\over { [ ( \omega-{\tilde \epsilon}_1)(\omega-{\tilde \epsilon}_2)-{\tilde t_c}^2]^2+[{\tilde \gamma}_1(\omega-{\tilde \epsilon}_2)+{\tilde \gamma}_2(\omega-{\tilde \epsilon}_1)+2{\tilde t_c}\sqrt{{\tilde\gamma}_1{\tilde\gamma}_2 } ] ^2 } } \;.\\\ ] ] the conductance @xmath44 at the absolute zero temperature in the limit of zero bias voltage is given by @xmath45 .
it is noticed that the formula for the transmission probability and conductance is equivalent to that of the transport through non - interacting double qd system , except that , in this case , the dot levels @xmath46 , the coupling strength @xmath47 and @xmath48 are renormalized .
therefore , the electron transport through dqds is characterized by the parameters @xmath46 , @xmath47 and @xmath49 .
however , it should be noted that @xmath46 , @xmath47 and @xmath48 show strong dependence on the gate voltage applied to the quantum dots , hence the result of linear conductance is quite different from the non - interacting model . in the spin
degenerate case , we have ten unknown parameters @xmath50(i=1,2 ) in total to determine . from the constraints and the equation of motion of the slave boson operators in the effective hamiltonian , we obtain one set of self - consistent equations , which is a straightforward generalized form of the single dot case as discussed in ref.[18 ] . in this set of equations ,
the distribution gf of the quantum dots @xmath51 is involved , and its fourier transform is given by @xmath52g^a(\omega)$ ] .
we have solved the self - consistent equations numerically . in the following ,
we discuss the result of our calculation .
first , we consider two identical qds case : @xmath53 and @xmath54 . following eq.(5 )
, the transmission probability in this case has a breit - wigner resonance form , @xmath55 the retarded gf on each dot is also explicitly given as @xmath56\;.\\\ ] ] the spectral density in the @xmath11-th qd follows from the relation @xmath57 .
it shows that the spectral density is the sum of one lorentizan with the peak position located at @xmath58 and one dirac @xmath59 peak at @xmath60 , where @xmath61 and @xmath62 corresponds to energy of the bonding and the antibonding state of quantum dots , respectively .
the bonding state of dqds has level broadening @xmath63 due to its coupling with the leads .
the @xmath59 peak structure indicates that the antibonding state is totally decoupled from the leads .
therefore the electrons transport only through the channel of the bonding state , which gives a breit - wigner resonance form in the transmission .
in fig.2 we study the effect of interdot tunneling on the transmission probability and the local density of state of qd in the singly occupied regime . here
we take the hybridization strength as the energy unit @xmath64 , and @xmath65 .
fig.2(a ) shows that with increasing the interdot coupling @xmath8 , the line shape of breit - wigner resonance of transmission is preserved , while the center of the resonance shifts to higher energy .
thus , the value of transmission probability at zero frequency @xmath66 decreases , which , in turn , results in suppression of the linear conductance at zero bias voltage . for local density of state shown in fig.2(b ) , we see that the antibonding state energy is always nearby the fermi energy of the lead , whereas the center of spectral density contributed from the bonding state shifts to higher energy along with increasing @xmath8 .
the linear conductance @xmath44 as a function of the energy level @xmath67 of the qd at zero temperature is plotted in fig.3(a ) for several values of interdot tunneling @xmath8 . when there is no direct tunneling between two dots ( @xmath68 ) , the conductance reaches the unitary limit(@xmath2 ) in the kondo regime as expected . with increasing tunnel coupling @xmath8 ,
the conductance becomes suppressed and forms a plateau structure in the regime of singly occupied qd state .
when qds cross over to the doubly occupied state regime , the conductance increases to the maximum value @xmath2 .
the line shape of the linear conductance can be explained from the gate voltage dependence of the spectral density and the zero frequency transmission of qd . in fig.3(b ) we plot the bonding state energy @xmath61 , the antibonding state energy @xmath62 and the level broadening @xmath63 as functions of @xmath67 with @xmath69 .
one can see that in the singly occupied regime with decreasing @xmath67 , the antibonding state energy @xmath62 is fixed around the fermi energy of the leads(@xmath70 ) .
it indicates that @xmath71 , and the conductance @xmath72 } \approx { 1/ ( { \tilde t_c}^2/{\tilde\gamma}^2 + 1)}$ ] . for this identical quantum dot case ,
the value of @xmath73 is given by its bare value @xmath74 .
consequently , the conductance shows a plateau structure and the ratio of @xmath75 determines the height of the conductance plateau .
this is in agreement with the value of conductance at the plateau structure for different @xmath8 as shown in fig.3(a ) . with @xmath67 deceasing further , the qd state crosses over from the singly occupied to the doubly occupied regime , and
@xmath61 goes through from positive value to negative value . at the point , @xmath76
, we obtain the maximum conductance @xmath2 .
further down , the level broadening @xmath77 approaches zero and the dqd will be totally decoupled from the leads , thus the conductance becomes zero .
next , we consider dqd system with different dot levels , @xmath78 , and define @xmath79 and @xmath80 . for the sake of simplicity ,
we still assume @xmath81 . it is noted that in this case the renormalized hybridization strength @xmath82 . in fig.4
we plot the linear conductance as a function of average energy of the dot levels .
the parameters @xmath83 and @xmath84 are used . in the case of dqds without direct tunnel coupling @xmath68 , fig.4(a ) shows that the conductance curve has a narrow dip around the point @xmath85 . from the formula for transmission eq.(6 ) , we note that the conductance vanishes when the condition @xmath86 is satisfied .
the strictly zero transmission is a consequence of destructive quantum interference for electron transport through the parallel dqds , and it is absent for systems with dqds connected in series .
it is interesting to notice that only when the dqds with different energy levels @xmath87 , this characteristic of interference is revealed .
it originates from the fact that , in this case , both the bonding and antibonding state channels are involved in the transmission .
as the energy difference @xmath88 increases , the dip becomes more broadened . for non - zero interdot tunnel couplings as shown in fig.4(b ) , ( c ) and ( d ) , the conductance curves have asymmetric line shapes , which are typical for the fano resonance .
this results from the constructive and destructive interference processes for electrons transmitted through the channels of bonding and antibonding states .
it is noted that line broadening of the fano dip or peak depends on the value of dot level difference @xmath88 , which is similar to the non - interacting dqds case@xcite .
the effect of on - site interaction @xmath0 is to introduce strong renormalization of the dot levels and the hybridization strength , hence the center of the fano resonance and line broadening have nonlinear dependence on the interdot tunneling @xmath8 and the level difference @xmath88 .
it is interesting to notice the fano resonances obtained in this study have some similarity with the experiment results in ref.@xcite .
although their experiment is on electron transport through single qd , the coupling strength between quantum dot and the lead is strong and the kondo effect and multilevel of qd might be involved in the electron transport .
recently , b@xmath89sser et al.@xcite have studied the electron transport through multilevel quantum dots using exact - diagonalization techniques .
it is interesting to notice that they have also found conductance dip structure induced by interference effect as shown in fig.4(a ) .
actually , when @xmath68 , the model studied in our paper is equivalent to considering two levels in single quantum dot . for a dqd system with energy level difference
, the local density of state in @xmath90th qd@xmath91 is given by @xmath92 ^ 2\over { [ ( \omega-{\tilde \epsilon}_1)(\omega-{\tilde \epsilon}_2)-{\tilde t_c}^2]^2+[{\tilde \gamma}_1(\omega-{\tilde \epsilon}_2)+{\tilde \gamma}_2(\omega-{\tilde \epsilon}_1)+2{\tilde
t_c}\sqrt{{\tilde\gamma}_1{\tilde\gamma}_2 } ] ^2 } } \;.\\\ ] ] in fig.5 . we plot the local density of state in each dot .
the line shape of density of state can be regarded as a superposition of a fano line shape close to the antibonding state energy and a breit - wigner resonance around the bonding state energy(see ref.[16 ] for detail discussion ) .
the interference effect on the local density of state is manifested clearly as compared with that in fig.2(b ) .
in summary , we have studied the electron transport through dqds in parallel configuration with interdot tunneling in the kondo regime . the strong coulomb repulsion in the dots
is taken into account via the finite-@xmath0 slave boson technique .
the results of our calculation indicate several distinct features from the non - interacting model@xcite : the conductance shows plateau structure as a function of the dot level in the singly occupied regime ; without interdot tunneling @xmath68 , there is dip structure on the conductance plateau when the energy levels of two dots are different .
when @xmath93 , the conductance has fano resonance line shape on the conductance plateau as a function of the averaged dot level ; the energies of the bonding and antibonding states and the level broadening of the bonding state are strongly renormalized compared to the noninteracting model case .
for instance , the antibonding state energy is almost fixed around the fermi energy of the lead in the singly occupied region .
the results are also different from that of the dqds in series , in which the maximum conductance is achieved when the interdot tunneling @xmath94 and no fano resonance is observed@xcite .
the fano effect for parallel dqds originates from the interference effect for electron transport through the two channels of bonding and antibonding states of parallel dqds .
in one recent experiment , chen et al.@xcite have studied the kondo effect in parallel dqds system .
but the maximum conductance obtained in their experiment is only about @xmath95 by varying the gate voltage and interdot tunneling , so that we think the full coherent electron transport through dqds is nt achieved and the interference effect is not manifested .
one may expect further experiments on the parallel dqds system can observe the conductance plateau structure and also the fano resonance as discussed above .
d. goldhaber - gordon et al .
, nature(london)*391 * , 156(1998 ) ; s. m. cronenwett et al .
, science*281 * , 540(1998 ) ; w.g .
van der weil et al . , science*289 * , 2105(2000 ) .
a. yacoby et al . , phys .
lett.*74 * , 4047(1995 ) ; r. schuster et al . , nature(london)*385 * , 417(1997 ) ; yang ji et al . ,
science*290 * , 779(2000 ) .
u. fano , phys .
rev.*124 * , 1866(1961 ) .
v. madhavan , w. chen , t. jamneala , m. f. crommie , and n. s. wingreen , science*280 * , 567(1998 ) .
j. g@xmath96res , d. goldhaber - gordon , s. heemeyer , m.a .
kastner , h. shtrikman , d. mahalu , and u. meirav , phys .
rev . b*62 * , 2188(2000 ) ; i.g .
zacharia , d. goldhaber - gordon , g. granger , m.a .
kastner , y.b .
khavin , h. shtrikman , d. mahalu , and u. meirav , phys .
rev . b*64 * , 155311(2001 ) .
a. c. johnson , c. m. marcus , m. p. hanson , and a. c. gossard , phys .
lett.*93 * , 106803(2004 ) .
k. kobayashi , h. aikawa , s. katsumoto , and y. iye , phys .
* 88 * , 256806(2002 ) ; phys .
rev . b*68 * , 235304(2003 ) ; k. kobayashi , h. aikawa , a. sano , s. katsumoto , and y. iye , phys .
rev . b*70 * , 035319(2004 ) .
b. r. buka and p. stefaski , phys .
lett . * 86 * , 5128(2001 ) ; w. hofstetter , j. k@xmath97nig , and h. schoeller , phys . rev . lett . * 87 * , 156803(2001 ) .
r. h. blick et al .
rev . lett.*80 * , 4032(1998 ) ; g. schedelbeck et al .
, science*278 * , 1792(1997 ) ; t. h. oosterkamp , et al .
, nature(london)*395 * , 873(1998 ) ; h. jeong et al .
, science*293 * , 2221(2001 ) .
j. c. chen , a. m. chang , and m. r. melloch , phys .
lett.*92 * , 176801(2004 ) ; a. w. holleitner , r. h. blick , a. k. h@xmath98ttel , k. eberl , j.p .
kotthaus , science*297 * , 70(2002 ) ; a. w. holleitner , c. r. decker , h. qin , k. eberl , and r.h .
blick , phys .
lett . * 87 * , 256802(2001 ) .
a. georges and y. meir , phys .
rev . lett.*82 * , 3508(1999 ) .
r. aguado and d. c. langreth , phys .
lett . * 85 * , 1946(2000 ) .
r. l@xmath99pez , r. aguado , and g. platero , phys .
* , 136802(2002 ) .
b. r. buka and t. kostyrko , phys .
rev . b*70 * , 205333(2004 ) .
b. kubala and j. k@xmath97nig , phys .
rev . b*65 * , 245301(2002 ) .
m. l. l. de guevara , f. claro , and p. a. orellana , phys .
b*67 * , 195335(2003 ) .
g. kotliar and a. e. ruckenstein , phys .
* 57 * , 1362(1986 ) .
b. dong and x. l. lei , phys .
rev . b*63 * , 235306(2001 ) .
b. dong and x. l. lei , phys .
b*65*r241304(2002 ) g. h. ding and b. dong , phys .
b*67*,195327(2003 ) . h. haug and a. p. jauho , _ quantum kinetics in transport and optics of semiconductors_(springer - verlag , berlin , 1998 ) .
b@xmath98sser , g.b .
martins , k.a .
al - hassanieh , a. moreo , and e. dagotto , phys .
rev . b*70 * , 245303(2004 ) . and ( b ) the local density of state for the system with two identical quantum dots .
used parameters are @xmath100 . the interdot tunnel coupling @xmath8 are @xmath101(solid line ) , @xmath102(dashed line ) , @xmath103(dotted line ) .
( we take the energy unit as @xmath64 , and @xmath104 corresponds to the fermi energy of the leads ) . ] , and the interdot tunneling @xmath105(solid line ) , @xmath106(dashed line ) , @xmath102(dotted line ) , @xmath103(dash - dotted line ) .
( b ) the bonding state energy @xmath107(solid line ) , the antibonding state energy @xmath108(dashed line ) and the level broadening @xmath109 of the bonding state(dotted line ) for @xmath94 . ] | electron transport through parallel double quantum dot system with interdot tunneling and strong on - site coulomb interaction is studied in the kondo regime by using the finite-@xmath0 slave boson technique . for a system of quantum dots with
degenerate energy levels , the linear conductance reaches the unitary limit ( @xmath1 ) due to the kondo effect at low temperature when the interdot tunneling is absent . as the interdot tunneling amplitude increases , the conductance decreases in the singly occupied regime and a conductance plateau structure appears . in the crossover to the doubly occupied regime ,
the conductance increases to reach the maximum value of @xmath2 . for parallel double dots with different energy levels ,
we show that the interference effect plays an important role in the electron transport .
the linear conductance is shown to have an asymmetric line shape of the fano resonance as a function of gate voltage . |
this paper is concerned with time - reversal symmetry in dynamical systems with finite phase space , with applications to rational maps over finite fields .
the concept of reversibility originated in the theory of smooth maps and flows on manifolds @xcite . in this paper ,
a map @xmath1 is said to be _
reversible _ if it is the composition of two involutions @xmath2 and @xmath3 .
the involution @xmath2 conjugates the map to its inverse , namely @xmath4 ( for background information on reversibility and its generalization , see @xcite , and references therein . ) the nature of the fixed sets of @xmath2 and @xmath5 , respectively @xmath6 and @xmath7 , plays a key dynamical role in organizing the dynamics of a reversible map .
the reversibility property can be interpreted in purely algebraic terms , and in particular it applies to reversible algebraic maps @xmath1 of any dimension @xmath8 over any field @xmath9 , namely reversible maps defined by rational functions with coefficients in @xmath9 . if the field @xmath10 is a _ finite field _ with @xmath11 elements where @xmath11 is the power of a prime number @xmath12 we are led to the study of reversibility on a phase space with @xmath13 points .
( for background reference on finite fields , see @xcite . )
more meaningful is the case in which @xmath9 is an algebraic number field , e.g. , @xmath14 . the same map @xmath1
may then be reduced over infinitely many distinct finite fields , of increasing size .
we are concerned here with the case where the reductions of the component involutions @xmath2 and @xmath5 are involutory permutations of the phase space of @xmath13 points , so that the reduction of their composition @xmath1 yields a permutation of this space . in reduction ,
the fixed sets @xmath6 and @xmath7 will now be finite .
examples arising from dimension @xmath15 are the reversible planar polynomial automorphisms @xcite which include the well - known area - preserving hnon map @xmath16 , @xmath17 if @xmath18 , we find @xmath19 reversible polynomial automorphisms in any dimension provide other examples .
but certain rational maps in @xmath8 dimensions can induce permutations , e.g. , @xmath20 of @xcite @xmath21 if @xmath18 , with @xmath22 , the denominator in @xmath23 is non - vanishing and @xmath24 the representability of @xmath2 and @xmath5 over a finite field @xmath25 imposes some restrictions on the parameters present in the involutions .
for instance , if in , we take @xmath26 , say , we are led to consider the set @xmath27 of the prime divisors of the denominator of @xmath28 .
then the involution @xmath29 is representable over any finite field whose characteristic is not in @xmath27 , such as the field of integers modulo a prime @xmath30 .
additionally , one must verify that @xmath2 and @xmath5 remain involutions when reduced to a finite field .
this is an instance of the property of _ good reduction _ ( for a discussion of reduction in arithmetic dynamics , see @xcite ) .
representability assumed , the reduction of an algebraic map can be associated with interesting asymptotic phenomena , concerning the statistical behaviour of the periodic orbits in the large field limit . a main theme in this area of research is to ascertain whether such phenomena are _ universal , _ namely independent from the map , depending only on its structural properties e.g. , reversibility .
furthermore , in reduction , the original dimension @xmath8 of the underlying algebraic map ceases to be important as the finite phase space does not inherit the topology of the continuum @xmath31 or @xmath32 .
if the reduction of a reversible algebraic map is a permutation , it has been shown @xcite that over finite fields , reversibility manifests itself combinatorially .
in particular , the number of symmetric periodic orbits those invariant under @xmath2 equals @xmath33 strong experimental evidence @xcite suggests that the symmetric periodic orbits dominate the statistics over the asymmetric ones , so that asymptotically counts the number of periodic orbits in the permutation .
furthermore , numerical experiments @xcite suggest there exists an asymptotic ( large fields ) distribution of the periods of the orbits , given by @xmath34 ( in these experiments , ` large fields ' refers to the limit @xmath35 ; there are other asymptotic regimes , see remark ( 3 ) in section [ section : concludingremarks ] . ) the distribution ( [ eq : r ] ) describes the limiting probability of the set of points belonging to cycles with scaled period not exceeding @xmath36 .
this distribution is believed to be universal , within the class of reversible maps with a single time - reversal symmetry . by contrast ,
maps which are not reversible appear to behave like random permutations @xcite ; the associated distribution , also conjectured to be universal , is markedly different from that of reversible maps .
finally , there are asymptotic period distributions for integrable systems @xcite .
even though these distributions are map - dependent , they all feature a distinctive ` quantization ' of periods . with appropriate scaling ,
the allowed periods occur at the reciprocals of the natural numbers , resulting in step - like distribution functions .
these various asymptotic phenomena have led to the development of simple and effective tests for detecting integrability and reversibility in algebraic mappings @xcite . in this paper
we propose a combinatorial model for the orbits , over a finite field , of a map with a single family of reversing symmetries .
our model consists of the composition of two random involutions .
the size @xmath37 of the space and the numbers @xmath38 and @xmath39 of fixed points of each involution are the parameters of the model . by letting @xmath37 increase
, we build a sequence of probability spaces . under very general conditions , we derive the distribution ( [ eq : r ] ) found in @xcite .
we prove the following theorem , which appears in section [ section : randominvolutions ] as theorems [ theorem : distribution ] and [ theorem : asymmetriccycles ] .
* theorem a. * _ let @xmath40 be a pair of random involutions of a set @xmath41 with @xmath37 points , and let @xmath42 and @xmath43 .
let @xmath44 be the expectation value of the portion of @xmath41 occupied by cycles of @xmath45 with period less than @xmath46 , computed with respect to the uniform probability .
if , with increasing @xmath37 , @xmath47 and @xmath48 satisfy the conditions @xmath49 then for all @xmath50 , we have the limit @xmath51 , the distribution function in ( [ eq : r ] ) .
moreover , almost all points in @xmath41 belong to symmetric cycles . _
this result is formulated in terms of the sequence @xmath52 , and not of the two sequences individually .
the conditions on the growth rate of @xmath52 are quite mild , hence the wide scope of applicability of the theorem .
if @xmath47 and @xmath48 do not grow at the same rate , an interesting dynamical phenomenon occurs : asymptotically , almost all periodic orbits have even period ( corollary [ corollary : evenonly ] ) . for @xmath8-dimensional maps over the finite field @xmath25 , one has @xmath13 , while @xmath47 and @xmath48 will typically grow algebraically ( see remark 1 of section [ section : concludingremarks ] ) . as mentioned above ,
increasing the dimension does not introduce new difficulties , because the absence of a significant topological structure makes the combinatorial model effectively dimension - independent .
all that is needed is to keep track of the possible cardinalities of the fixed sets of the two involutions . in @xcite , experimental evidence from reductions of
the hnon map suggested that the probability of occurrence of several cycles with the same period followed a poisson law , with parameter depending exponentially on the ( rescaled ) period .
these findings motivate the study of the statistics of the occurrence of repeated periods in the random involutions model . in section [ section : repetitions ] we consider this question , and prove the following result .
* theorem b. * _ with the notation of theorem a , let @xmath53 be the probability that @xmath40 has @xmath54 cycles of period @xmath55 .
assume that the sequence @xmath56 has the ( possibly infinite ) limit @xmath57 and let @xmath58 then , if @xmath59 we have , as @xmath60 @xmath61 while if @xmath62 we have @xmath63 , where @xmath64 is kronecker s delta .
_ as it will appear in the proof of theorem b , in the case of constant @xmath65 , for convergence to a poisson distribution it is only required that @xmath66 , not the actual existence of a non - zero limit @xmath67 .
we conclude section 3 by showing the agreement between the prediction of theorem b with the actual repetitions @xmath53 for a small @xmath55 in the hnon map .
the latter are governed by the roots @xmath68 of certain polynomial functions of one variable , with a connection to the galois groups of these polynomials . in section [ section : concludingremarks ]
we discuss the question of the asymptotic ( large field ) growth rate of the parameters associated with a rational map over a finite field , namely the numbers @xmath47 and @xmath48 of fixed points of the involutions .
in this case @xmath37 is the power of a prime number .
we show that the growth conditions of theorem a on @xmath52 are easily satisfied , while @xmath69 in theorem b depends algebraically on @xmath37 , and all scenarios considered in the theorem can be realized .
finally , we also sketch how the random involution model might be extended to the case of reversible rational maps with singularities .
let @xmath37 be a positive integer , and let @xmath70 be integers in the range @xmath71 such that @xmath72 and @xmath73 are both even .
we consider ordered pairs @xmath74 of random involutions on a set @xmath41 of @xmath37 points , which fix @xmath47 and @xmath48 points , respectively .
thus @xmath42 , @xmath43 , and the cycle decomposition of @xmath2 consists of @xmath47 fixed points and @xmath75 two - cycles , and similarly for @xmath5 . for each pair @xmath40 of involutions
we consider the composition @xmath76 , which is a reversible permutation of @xmath41 .
let @xmath77 and @xmath78 be the sets of all involutions with the given parameters .
we regard the space @xmath79 as a probability space , with the uniform probability .
we define the random variable @xmath80 to be the fraction of the space @xmath41 occupied by the @xmath55-cycles of the map @xmath45 , namely @xmath81 for given @xmath37 , the sequence @xmath82 has only a finite number of non - zero terms .
the main object under study is the distribution function @xmath83 where @xmath84 is the floor ( greatest integer part ) function , @xmath85 is a scaling parameter to be specified below , and the average @xmath86 is computed with respect to the uniform probability on @xmath87 ( i.e. , @xmath88 ) . if @xmath89 , the sum is empty , and @xmath44 is defined to be zero .
the asymptotic parameter is @xmath37 .
we shall assume that @xmath90 and @xmath91 are positive integer sequences satisfying .
these sequences determine the sequence of probability spaces to be considered .
to get non trivial asymptotics , the scaling parameter @xmath85 must grow with @xmath37 at an appropriate rate . to this end
, we define the rational sequence @xmath92 which , due to ( [ eq : ghasymptotics ] ) , diverges to infinity .
the quantity @xmath93 is the ratio of the size of the phase space to the number of symmetric cycles cf
. equation ( [ eq : numsym ] ) .
its full significance will be clarified at the end of this section ( corollary [ corollary : z ] ) .
the main purpose of this section is to prove theorem a of the introduction , which is an amalgamation of theorems [ theorem : distribution ] and [ theorem : asymmetriccycles ] .
[ theorem : distribution ] the distribution function @xmath94 defined in ( [ eq : rn ] ) , with scaling sequence @xmath93 given by ( [ eq : z ] ) , admits the following limit @xmath95 it turns out that to establish theorem [ theorem : distribution ] it suffices to consider symmetric orbits only , because they provide the dominant contribution to the probability @xmath96 in equation ( [ eq : rn ] ) see theorem [ theorem : asymmetriccycles ] below .
accordingly , we separate out the contribution deriving from symmetric and asymmetric cycles as follows @xmath97 we begin with symmetric cycles of odd period @xmath98 .
it is known @xcite that a cycle is of this type if and only if it corresponds to an orbit of @xmath1 starting from @xmath6 and reaching @xmath7 for the first time after @xmath99 iterations . here and below , we follow the cycle dynamics keeping track of the action of @xmath2 and @xmath5 separately , which leads to a maximal sequence of @xmath100 points @xmath101 with the property that all @xmath102 are distinct .
we call such a sequence a _ @xmath99-arc_. in the case of odd period , we have the condition @xmath103 ( hence @xmath104 ) together with the requirement that the sequence contains @xmath98 distinct points ( see figure [ fig : arc ] )
. we must now determine the total number of @xmath99-arcs of this type generated by the elements of @xmath87 . throughout the paper ,
the underlined exponent will denote the falling factorial powers @xmath105 the information we require is contained in the following two lemmas .
[ lemma : e ] let @xmath106 be as in equation ( [ eq : egh ] )
. then @xmath107 the cardinality of the set of all involutions on @xmath37 points , with @xmath47 fixed points , is given by @xmath108 where the first term gives the number of ways of choosing the fixed points , and the second counts the number of ways of partitioning the remaining points into unordered pairs . multiplying the expression above with the corresponding expression for @xmath48 fixed points , we obtain @xmath109 as desired .
[ lemma : podd ] let @xmath110 be as above .
the following holds @xmath111 where @xmath112 let @xmath113 be the required @xmath99-arc ( see equation ( [ eq : arc ] ) and figure [ fig : arc ] ) .
the initial point @xmath114 belongs to @xmath6 , so that @xmath104 .
the first half of the cycle consists of the points @xmath115 , the second half of the points @xmath116 , giving @xmath117 distinct points in total .
hence the desired number of arcs is @xmath118 each @xmath99-arc occurs with multiplicity determined by the unconstrained action of the pairs of involutions on the rest of the space , which is given by @xmath119 thus the average value of @xmath120 is given by @xmath121 finally , the product formula for @xmath122 is obtained from lemma [ lemma : e ] with a straightforward computation . by analogy with equation ( [ eq : rn ] ) , we define the distribution function @xmath123 for odd symmetric cycles as @xmath124 we have [ theorem : distributionodd ] for any sequences @xmath70 satisfying ( [ eq : ghasymptotics ] ) , with @xmath85 as in ( [ eq : z ] ) , we have , as @xmath60 @xmath125 note that the limit distribution is symmetrical in @xmath47 and @xmath48 . to prove theorem [ theorem : distributionodd ] we need a lemma .
[ lemma : basic ] let @xmath126 be as in ( [ eq : ghasymptotics ] ) , let @xmath127 be a positive integer sequence , and let @xmath128 then , as @xmath60 @xmath129 \hskip 30pt\kappa=\frac{g+h}{n}.\ ] ] for fixed @xmath37 , @xmath130 and @xmath131 are fixed , and the quantity @xmath132 is the sum of an arithmetico - geometric progression , with value @xmath133\bigr\}.\ ] ] as @xmath60 , we have , from ( [ eq : ghasymptotics ] ) @xmath134 so we find @xmath135 \sim \frac{2}{\kappa^2}\bigl[1-e^{-m\kappa}(m\kappa+1)\bigr]\ ] ] as desired .
_ proof of theorem [ theorem : distributionodd ] . _ from lemma [ lemma : podd ] , we find , for @xmath136 @xmath137 now fix @xmath130 , with @xmath138 , and define @xmath139 for all @xmath99 such that @xmath140 , we have the inequalities @xmath141 which , together with ( [ eq : avgpodd ] ) , give @xmath142 summing over @xmath99 , we obtain the bounds @xmath143 where @xmath144 let now @xmath127 be a positive integer sequence such that @xmath145 then @xmath146 and applying lemma [ lemma : basic ] , we obtain , as @xmath60 @xmath147.\ ] ] we now fix a non - negative real number @xmath36 . for the convergence of the expression above ,
we specialize the sequence @xmath148 as follows @xmath149 where @xmath150 and @xmath85 were defined in equations ( [ eq : basic ] ) and ( [ eq : z ] ) , respectively . due to
( [ eq : ghasymptotics ] ) , @xmath151 , and for all sufficiently large @xmath37 , we have @xmath152 .
hence ( [ eq : rbounds ] ) is valid , and our result now follows from ( [ eq : rnodd ] ) and ( [ eq : asymptotics ] ) , noting that @xmath153 .
the sequence @xmath148 given in ( [ eq : m ] ) could be defined by the more general requirement that the sequence @xmath154 converge to a positive real number @xmath67 , where the limiting cases @xmath155 are excluded as they would lead to trivial distribution functions .
we have set @xmath156 , for normalization .
we remark that the choice of this constant does not affect the coefficient of the distribution function , which determines its limiting value for large arguments .
this fact will become relevant in the proof of theorem [ theorem : distribution ] below .
we now turn to even symmetric cycles , of period @xmath157 .
each cycle of this type corresponds to an orbit originating on @xmath6 ( or @xmath7 ) , and returning to it for the first time after @xmath99 iterations of @xmath1 .
the arc associated to such a cycle ( see equation ( [ eq : arc ] ) ) is a @xmath158-arc , with the property that @xmath114 and @xmath159 are in @xmath6 .
therefore @xmath160 and @xmath161 , and the total number of arcs of this type is equal to @xmath162 , where the denominator 2 prevents arcs beginning and ending on @xmath163 from being counted twice .
the denominator 2 cancels with the 2 in the period @xmath100 , so we obtain @xmath164 where @xmath165 putting together equation ( [ eq : peven ] ) and ( [ eq : ceeven ] ) , we obtain , for @xmath166 @xmath167 to be compared with equation ( [ eq : avgpodd ] ) . exchanging @xmath47 by @xmath48 in the formulae above gives the twin quantity @xmath168 for @xmath100-cycles with two points of @xmath7 . the distribution function @xmath169 for even symmetric cycles
is obtained by adding the contributions from the two fixed sets @xmath170 an analysis very similar to that of theorem [ theorem : distributionodd ] leads to the following result , whose proof we omit for the sake of brevity .
[ theorem : distributioneven ] for any sequences @xmath70 satisfying ( [ eq : ghasymptotics ] ) , with @xmath85 as in ( [ eq : z ] ) , we have , as @xmath60 @xmath171 proof of theorem [ theorem : distribution ] . from theorems [
theorem : distributionodd ] and [ theorem : distributioneven ] we deduce that the probability distribution associated to symmetric cycles is given by @xmath172 because @xmath173 this distribution function is properly normalized ( cf .
remarks following the proof of theorem [ theorem : distributionodd ] ) , and so it accounts for all cycles .
it is implicit from theorem [ theorem : distribution ] that , asymptotically , asymmetric cycles have zero probability .
asymmetric @xmath55-cycles of @xmath174 come in pairs , mapped to one another under @xmath2 .
they correspond to @xmath55-arcs of the type @xmath175 i.e. , they have @xmath176 distinct entries .
we can now determine the decay rate of the probability of asymmetric cycles .
[ theorem : asymmetriccycles ] consider a composition of random involutions on a set of @xmath37 points , whose fixed sets satisfy ( [ eq : ghasymptotics ] ) . as @xmath60 ,
the measure of the set of points that belong to asymmetric cycles is asymptotic to @xmath177 ; in particular , almost all cycles are symmetric .
recall from ( [ eq : decomposition ] ) that @xmath178 is the probability of finding an asymmetric @xmath55-cycle .
we compute its average over @xmath87 : @xmath179 from the above discussion , the right - hand side of this equation involves counting @xmath55-arcs like .
note that the period of the orbit is not present as a factor .
this is because it is absorbed by the possibility to create @xmath55-arcs that are one and the same from by cyclic permutations of the @xmath180 pairs ( @xmath55 possibilities ) and the switch @xmath181 together with reversing the order of the switched pairs ( @xmath182 possibilities ) .
let @xmath183 be the associated distribution function @xmath184 we now proceed as in the proof of theorem , with @xmath185 and @xmath85 given by ( [ eq : lambdaplusminus ] ) and ( [ eq : z ] ) , respectively .
summing the expression above over @xmath55 in the appropriate range gives the bounds @xmath186 where @xmath187 this time the bounds are geometric progressions ; as @xmath188 , we find @xmath189 as desired .
the theorems of this section have some immediate consequences .
consider the sequence @xmath93 , defined in ( [ eq : z ] ) .
its denominator @xmath190 is twice the number of symmetric cycles ( * ? ? ?
* proposition 1 ) , and from theorem [ theorem : distribution ] , asymptotically , the symmetric cycles have full measure .
these considerations give the following characterization of @xmath93 .
[ corollary : z ] asymptotically , the sequence @xmath93 defined in ( [ eq : z ] ) represents the average cycle length of the map @xmath45 .
an immediate consequence of theorems [ theorem : distributionodd ] and [ theorem : distributioneven ] is the following .
[ corollary : evenonly ] let @xmath70 be as above . if @xmath191 or @xmath192 , then almost all cycles of @xmath45 have even period .
this section first gives the proof of theorem b , stated in the introduction .
the asymptotic relations ( [ eq : ghasymptotics ] ) will be assumed to hold , together with the condition @xmath193 .
the latter does not entail loss of generality , as we may always exchange @xmath47 and @xmath48 , which amounts to considering the inverse mapping @xmath194 .
after the proof , we compare the results of theorem b with the mechanism that determines the repetition statistics for small period in the hnon map .
proof of theorem b. let @xmath53 be the probability that a randomly chosen @xmath195 has exactly @xmath54 cycles of period @xmath55 .
we wish to determine the asymptotic behaviour of @xmath196 ; due to theorem [ theorem : asymmetriccycles ] , it will suffice to consider symmetric cycles only .
we introduce some notation .
let @xmath197 be a @xmath55-cycle in @xmath41 , and let @xmath198 be any set of disjoint @xmath55-cycles .
we stipulate that the first point of each cycle belongs to some symmetry line ; then there are @xmath199 ways of choosing such cycles .
we define @xmath200 note that if the cycles of @xmath198 were not disjoint , then @xmath201 would be empty . given @xmath198 , with @xmath202
, we are interested in the number of pairs in * e * that support the cycles of @xmath198 , and that do not have any @xmath55-cycle in the complement @xmath203 .
the parameters of the involutions acting on @xmath203 are determined by the type of @xmath55-cycles being considered ( odd period symmetric , even period symmetric on @xmath6 , etc . ) .
accordingly , we ll consider the following sets of pairs of involutions @xmath204 whose cardinality is given by lemma [ lemma : e ] .
whenever we do nt require specialization , we ll denote any of the above sets by the common symbol @xmath205 .
the pairs of involutions in @xmath205 act on the space @xmath206 with @xmath207 elements .
let @xmath208 be the set of elements of @xmath205 that have no @xmath55-cycles at all . by the inclusion - exclusion principle ,
we find @xmath209 the number of non - zero summands is clearly finite .
to compute the inner sum for fixed @xmath55 and @xmath54 , we first choose @xmath8 distinct cycles , and this can be done in @xmath210 ways
. then we let the involutions act freely on the remaining space , which gives @xmath211 possibilities .
thus @xmath212 considering the equations ( [ eq : choosingcycles ] ) and ( [ eq : nt ] ) , we obtain @xmath213 where @xmath214 .
we specialize formula ( [ eq : mu ] ) to the case of odd period @xmath55 , using the notation @xmath215 . for this purpose ,
we consider the first expression @xmath87 in ( [ eq : es ] ) , to obtain @xmath216 inserting the values of @xmath217 from lemma [ lemma : e ] , we obtain , after some manipulation @xmath218 where @xmath219 with a similar argument we derive the probability @xmath220 for even period @xmath221 where @xmath222 as in the analysis of section [ section : randominvolutions ] , this expression takes into account cycles on both @xmath6 and @xmath7 , while the factor of 2 compensates for the double - counting that results from the presence of two points on each symmetry line .
we now consider the asymptotic behaviour of @xmath53 , for fixed @xmath54 .
let @xmath65 be given by ( [ eq : y ] ) , which we rewrite as @xmath223 for the purpose of scaling , the period @xmath55 will be allowed to vary with @xmath37 .
let us consider the expression @xmath224 which , for @xmath225 , provides the dominant contribution to the products @xmath226 , in both cases .
a straightforward calculation gives @xmath227\label{eq : bigo}.\end{aligned}\ ] ] in view of the bounds @xmath228 we compute @xmath229 \\ \pi(j , g , h)&=&e^{-(i+n)x}\left[1+o\left(\frac{n^2x^2}{g}\right)+ o\left(\frac{nxg}{n}\right)\right].\end{aligned}\ ] ] we obtain @xmath230 and hence @xmath231.\ ] ] this estimate is valid for both odd and even period .
next we consider the products @xmath232 in ( [ eq : ao ] ) and ( [ eq : ae ] ) .
recalling the definition of @xmath69 given in equation ( [ eq : f ] ) , and the fact that @xmath193 , we have the following asymptotic behaviour @xmath233,\qquad i+n>0 \label{eq : estimatea}\end{aligned}\ ] ] we must consider two cases .
_ case i : @xmath66._from ( [ eq : estimateb ] ) and ( [ eq : estimatea ] ) , we obtain , for both even and odd period @xmath234 \left [ 1 + o\left(\frac{nxg}{n}\right)+ o\left(\frac{n^2x^2}{g}\right ) \right]\nonumber\\ & = & \frac{(e^{-y})^i}{i!}\sum_{n\geq0}\frac{(-1)^ne^{-ny}}{n!}+ \frac{(e^{-y})^i}{i!}\sum_{n\geq0}\frac{(-1)^n}{n ! } \left [ o\left(\frac{n^2x}{g}\right ) + o\left(\frac{nxg}{n}\right ) \right]\nonumber\\ & = & \frac{(e^{-y})^i}{i ! } \left [ e^{\displaystyle-(e^{-y } ) } + o\left(\frac{x}{g}\right ) + o\left(\frac{xg}{n}\right)\label{eq : muerror } \right]\end{aligned}\ ] ] where we have used the identities @xmath235 now we adjust @xmath55 so that @xmath65 is constant .
this gives @xmath236 , and for both odd and even period the error terms in ( [ eq : muerror ] ) vanish .
we have convergence to a poisson distribution @xmath237 note that for the above result to hold , we have only required that @xmath66 , not the existence of a limit for @xmath69 .
if @xmath69 tends to a non - zero finite limit @xmath67 , there is no logarithmic term in the function @xmath238 cf .
( [ eq : ynew]) and it is possible to consider the probability @xmath53 without scaling the period @xmath55 . for fixed @xmath55 , we have @xmath239 and @xmath240 , and equation ( [ eq : muerror ] ) gives @xmath241 independent of @xmath55 .
_ case ii : @xmath242._from ( [ eq : estimatea ] ) we have @xmath243 ( kronecker s delta ) , and hence @xmath244 the proof of theorem b is complete .
we remark that theorem a can actually be derived from theorem b since the distribution @xmath245 follows from first principles as @xmath246 equivalently @xmath96 follows from @xmath247 note that the variable @xmath36 in is , asymptotically , the variable @xmath36 in @xmath245 . taking the result
gives @xmath248 since @xmath249 from , we have @xmath250 if @xmath98 is odd , we substitute for @xmath251 from and @xmath36 from to see @xmath252 comparing this to and noting there that @xmath253 shows that we will recover the result of theorem [ theorem : distributionodd ]
. a similar exercise on for @xmath157 using the appropriate @xmath251 from yields the result of theorem [ theorem : distributioneven ] and hence the distribution @xmath245 upon summation of odd and even contributions .
we have chosen to give the independent proof of theorem a in section [ section : randominvolutions ] as it is simple and direct and bypasses the calculation of @xmath53 .
the applicability of theorem b to algebraic maps over finite fields should be considered with care .
for fixed @xmath55 , the maximum number of @xmath55-cycles for an algebraic map is plainly finite , because the points in a cycle are solutions of a set of polynomial equations ; therefore , in the large field limit , one can not possibly expect convergence to ( [ eq : muasymptotic ] ) . under what circumstances
is then theorem b a model for such maps ? to articulate this question ,
let us consider a specific example , namely the hnon map ( [ eq : henon ] ) , with @xmath254 .
the set @xmath6 is the line @xmath255 , and so over the field @xmath256 , this map has parameters @xmath257 , hence @xmath258 in theorem b. let us consider symmetric cycles , which , due to theorem a , are expected to dominate the dynamics . when the fixed sets are one - dimensional , the study of symmetric orbits is greatly simplified , because the points in these orbits are roots of polynomial equations in one indeterminate ( as opposed to being points on varieties ) .
now , any symmetric odd cycle has a point @xmath259 on @xmath6 , and the values of @xmath36 for the period @xmath260 are found to be the roots of the irreducible polynomial @xmath261 thus the number of symmetric @xmath262-cycles over the field @xmath256 is equal to the number of linear factors in the factorization of @xmath263 modulo @xmath12 . in particular , the maximum number of 5-cycles is equal to 6 , the degree of @xmath263 . in general , polynomial factorization over a finite field is a highly irregular process , subject only to a slight deterministic constraint stickelberger s theorem which determines the parity of the number of factors @xcite .
significantly , this process is ruled by a probabilistic law cebotarev s density theorem which determines the probability associated with each factorization type ( number and degree of the factors ) in terms of properties of the galois group of the polynomial @xcite .
specifically , for each factorization with distinct roots , the degree of the factors defines an additive partition of the degree of the polynomial .
each partition in turn identifies uniquely a conjugacy class of the galois group , with the terms in the partition corresponding to cycle lengths in the cycle decompositions of the permutations in the class .
the probability of a factorization type is then given by the relative size of the corresponding class in the galois group . if the galois group of @xmath264 is the symmetric group , the largest possible one , then every factorization type does actually occur , and with positive density among all primes .
it can be verified that the galois group of @xmath263 is in fact @xmath265 , the set of all permutations of @xmath266 objects @xcite .
there are therefore 11 possible types of factorizations of @xmath263 , corresponding to the 11 additive partitions of the integer 6 .
we factor @xmath267 modulo a few consecutive primes @xmath268 these factorizations correspond to the partitions @xmath269 , respectively .
we see that , over the field @xmath256 , the hnon map has six symmetric 5-cycles for @xmath270 ( the smallest prime for which this happens ) , no 5-cycles at all for @xmath271 , and two 5-cycles for @xmath272 .
the degree of the polynomials @xmath273 grows exponentially with @xmath55 , and it is reasonable to expect that , typically , they will be irreducible , with large galois groups . for the symmetric group ,
the cebotarev s probability @xmath274 of occurrence of @xmath54 symmetric cycles of period @xmath55 is just the probability of having exactly @xmath54 fixed points in the appropriate symmetric group , namely @xmath275 in the large @xmath55 limit , @xmath274 converges to @xmath276 , independent of @xmath55 .
this agrees with the prediction of the random involution model .
it is instructive to compare the data ( [ eq : cebotarev ] ) from cebotarev s theorem with the prediction of theorem b , for the period @xmath260 .
we see that , even for such a low period , for @xmath277 the agreement is already reasonable .
thus , in the two - dimensional case , some statements on the probability of multiple occurrences of symmetric @xmath55-cycles translate into statements on the maximality of the galois groups of @xmath273 .
even though this formulation would allow one to re - cast some dynamical questions in algebraic terms , we note that problems related to galois groups of iterated polynomials are often quite difficult .
when the reduction over a finite field of a reversible algebraic map with a single family of reversing symmetries is a permutation , numerical experiments suggest the period distribution function is given by the period distribution function @xmath245 of . in this paper
, we have shown that @xmath245 corresponds to the expected period distribution function in a probability space of pairs of involutions characterized by the cardinalities of their fixed sets .
we find that the same probability space furnishes a poisson law for the probability of cycles with the same period , also in agreement with numerics . *
( 1 ) * to prove theorem a , only mild regularity conditions the existence of the limits ( [ eq : ghasymptotics ] ) were imposed on the behaviour of the sequences @xmath278 , @xmath279 , specifying the cardinalities of the fixed sets of the two involutions , as the size @xmath13 of the phase space increases ( with @xmath8 the dimensionality of the map ) .
similarly , theorem b involves checking the behaviour of the sequence @xmath69 in .
we point out that when a reversible map of a finite set is obtained by reducing an algebraic mapping to a finite field , the behaviour of such sequences is strongly constrained , resulting in algebraic growth .
a simple but significant case comprises maps for which the fixed set @xmath6 is an affine subspace of some linear space .
an example is given by two - dimensional reversible polynomial automorphisms over some field @xmath9 , where the existence of a normal form for the two involutions ensures that their fixed set is a single point or a line @xcite .
localization to the finite field @xmath25 gives a phase space with @xmath280 points , and since a line in the affine plane @xmath281 has @xmath11 points , we have an algebraic growth : @xmath282 . in dimension @xmath283 , each of @xmath6 and @xmath7 can have integer dimension ranging from 0 to @xmath284 , and if these sets are affine spaces , we obtain the sequence @xmath285 , with @xmath286 , @xmath287 .
the conditions are then met as long as one of @xmath288 or @xmath289 is nonzero , i.e. as long as one of @xmath6 and @xmath7 is at least one - dimensional .
furthermore , in this situation , we see from that @xmath290 or @xmath291 .
so either @xmath292 ( e.g. , when @xmath293 ) or @xmath294 algebraically or @xmath295 algebraically . more generally , the sets @xmath6 and @xmath7 are algebraic varieties over @xmath25 determined by @xmath8 polynomial equations , and we are interested in determining the number of points on these varieties .
the number @xmath296 of points on an @xmath288-dimensional irreducible algebraic variety in the @xmath8-dimensional projective space of degree @xmath283 over the field @xmath25 satisfies the hasse - weil bound @xcite @xmath297 .
for @xmath298 we obtain a square root bound , and we can also replace @xmath299 by the genus of the curve .
the result ensures that , in the large field limit , there is algebraic growth of the cardinality of fixed sets .
so the discussion of the previous paragraph applies concerning the validity of the asymptotic conditions ( [ eq : ghasymptotics ] ) and the behaviour of @xmath69 . * ( 2 ) * experiments show that algebraic maps without time - reversal symmetry behave quite differently from reversible ones , when represented on a finite field @xcite . in the absence of reversibility , the cycle distribution is found to be that of a random permutation , given by the identity function on the unit interval .
this distribution , obtained via the scaling @xmath300 , has led to a general conjecture ( * ? ? ? * conjecture 2 ) . * ( 3 ) * when a general rational reversible map is reduced over finite fields , we expect that points where the denominators vanish ( singularities ) will occur .
restricted to its periodic points , the reduced map is still a permutation .
numerically , we find the distribution of periods is still governed by @xmath245 , consistent with theorem a @xcite .
but nonperiodic orbits will also appear in the reduction of such a map .
we are presently investigating the extension of the random involution model to explain various aspects of the dynamics of the reduction to finite fields of general rational reversible maps @xcite .
we are also extending our numerical experiments to cover the asymptotic regime @xmath25 , @xmath301 , @xmath302 .
acknowledgements : this work was supported by the australian research council , via discovery project dp0774473 , and the australian academy of science .
jagr and fv would like to thank , respectively , the school of mathematical sciences at queen mary , university of london , and the school of mathematics and statistics at the university of new south wales , sydney , for their hospitality .
r. de vogelaere , on the structure of symmetric periodic solutions of conservative systems , with applications , in : _ contributions to the theory of nonlinear oscillations , vol . 4 _ , ed .
s. lefschetz , princeton university press , princeton ( 1958 ) , pp . | we study time - reversal symmetry in dynamical systems with finite phase space , with applications to birational maps reduced over finite fields . for a polynomial automorphism with a single family of reversing symmetries , a universal ( i.e. , map - independent ) distribution function
@xmath0 has been conjectured to exist , for the normalized cycle lengths of the reduced map in the large field limit @xcite .
we show that these statistics correspond to those of a composition of two random involutions , having an appropriate number of fixed points .
this model also explains the experimental observation that , asymptotically , almost all cycles are symmetrical , and that the probability of occurrence of repeated periods is governed by a poisson law . |
a piece of nonmagnetic , linear and isotropic dielectric material with frequency - dependent permittivity @xmath0 kept at some finite temperature @xmath1 generates and emits an electromagnetic field , which manifests itself as heat radiation , resulting from thermal and quantum mechanical fluctuations . besides the thermal far field , near - field phenomena associated with nonpropagating modes have recently attracted increasing attention @xcite .
possibly important effects have been revealed , such as the emergence of both temporal and spatial coherence in the near field of planar thermal sources due to surface waves @xcite .
in addition , the influence of these surface waves on radiative heat transfer and dispersion forces at the subwavelength scale has been investigated @xcite . in such studies
, one usually considers a simple half - space geometry , or two half - spaces separated by a narrow vacuum gap .
the characteristic length scale for absorption of heat radiation by the material is the skin depth @xmath2 evaluated at the dominant thermal frequency @xmath3 .
we use the notation @xmath4 , where @xmath5 is the velocity of light in vacuum ; @xmath6 is the relative permittivity , with @xmath7 denoting the permittivity of the vacuum .
thermal radiation generated inside the material can reach its surface only if it originates from its outermost layer with thickness on the order of @xmath8 .
hence , if the linear dimensions of a given dielectric are significantly larger than the skin depth , the emitted radiation preserves no information about the material s geometry . in that case , radiating and non - radiating components of the thermal electromagnetic field equal those emitted by a dielectric half - space @xcite . in this paper
we investigate a seminal example which shows that measurable effects occur when the above two length scales become comparable , so that the half - space model becomes inadequate .
we consider an infinite , planar dielectric slab and study the dependence of both its thermal far and near field on its thickness @xmath9 .
when @xmath9 is large compared to @xmath8 , the propagating radiation emitted by the slab is described by the planck
kirchhoff radiation law @xcite .
however , when @xmath9 is reduced below @xmath8 , two competing trends arise : on the one hand , multiple reflections of radiation inside the slab lead to a fabry perot - like enhancement of the field ; on the other , the radiating source volume is diminished .
we work out the consequences of this competition and show that , in particular , the near - field energy density close to the surface of a metal film can remain finite even when the thickness of that film becomes arbitrarily small , as a result of the emergence of a low - frequency surface plasmon polariton .
we proceed as follows : in section [ s_2 ] we briefly collect the required elements of fluctuational macroscopic electrodynamics @xcite , and outline in sec .
[ s_3 ] the construction of the classical dyadic green s functions for the dielectric slab ; these turn out to be significantly more complicated than the more often considered ones for a half - space , or for a vacuum gap between two half - spaces @xcite . although properties of propagating thermal radiation may also be obtained by more direct means @xcite , and numerical codes for investigating thermal radiation of layered structures do exist @xcite , the detailed analytical discussion of these green s functions is of its own intrinsic value .
green s functions for layered media appear in a variety of contexts , such as van der waals forces in multilayer systems @xcite , magnetic noise in conducting slabs @xcite , light scattering and control of spontaneous emission in planar cavities @xcite , or thermal spin flips in atoms chips @xcite , to name but a few .
unfortunately , the algebra involved in writing down such green s functions , though not difficult in principle , is vexatingly cumbersome .
here we resort to the formalism outlined in ref .
@xcite , based on the systematic use of vector wave functions @xcite , which combines versatility with transparency , and which allows us to treat both far - field and near - field effects on equal footing . with the help of the green s functions we then derive in sec .
[ s_4 ] general expressions for both the intensity of the slab s thermal radiation field and its energy density , deferring tedious mathematical details to the appendix [ a_1 ] .
section [ s_5 ] contains an extensive discussion of these results . for the sake of definiteness we concentrate on dielectric slabs effectuating a coupling between plasma - like electron motion and the photon field @xcite , such as metals . within the drude approach ,
their permittivity is given by @xmath10 \ ; , \label{eq : drude}\ ] ] where @xmath11 denotes the plasma frequency , and @xmath12 the relaxation time @xcite .
since the dominant thermal frequency amounts to @xmath13 s@xmath14 for @xmath15 k , and since typical relaxation times for metals are on the order of @xmath16 s , one can satisfy the inequality @xmath17 in the infrared , thus arriving at the hagen
rubens approximation @xcite @xmath18 for the relative permittivity .
for metals with comparatively short relaxation time , such as bismuth ( @xmath19 s at @xmath20 k ) , this approximation is quite good indeed .
however , a drude metallic state can also be achieved with conducting polymers ; for instance , hexafluorophosphate doped polypyrrole [ ppy(pf@xmath21 ) ] yields a plasma frequency in the far infrared at about @xmath22 s@xmath14 , combined with an anomalously long scattering time quantified as @xmath23 s in ref .
@xcite . hence , while we use this approximation ( [ eq : hr ] ) for deriving various analytical estimates , we rely on the full drude permittivity ( [ eq : drude ] ) in our numerical calculations . besides bismuth ,
for which @xmath24 at room temperature , we will also consider more typical metals , for which @xmath25 is two orders of magnitude larger .
we consider in subsection [ s_5b ] the radiative intensity emitted by thin metal films , and demonstrate that the opposing trends hinted at above result in an optimum film thickness which maximizes that intensity .
we then study in subsection [ s_5c ] the `` evanescent '' near - field energy density , and establish a somewhat counterintuitive result : while the contribution of the te modes to that density vanishes when the film thickness goes to zero , that of the tm modes does not , but remains finite and becomes universal , at least within the scope of the simple drude approach .
finally , we briefly spell out some experimental ramifications in sec .
as is customary , we consider the macroscopic electric and magnetic fields inside the dielectric material , @xmath26 and @xmath27 , obtained by averaging the microscopic fields over some appropriate volume @xcite , so that their small - scale , `` atomic '' fluctuations are smoothed out .
since these fields are described by real numbers , one has @xmath28 where @xmath29 denotes the complex conjugate of the preceding term .
a corresponding identity holds for @xmath27 .
hence , it suffices to restrict the temporal fourier transforms @xmath30 and @xmath31 to positive frequencies @xmath32 .
following rytov and co - workers @xcite , we describe the connection between the electromagnetic field and its sources by augmenting the dynamical macroscopic maxwell equations by fluctuating current fields . for nonmagnetic materials , characterized by the permeability @xmath33 of the vacuum ,
only an `` electric '' current is required , the frequency components @xmath34 of which are regarded as independent stochastic variables .
the resulting equations @xmath35 then adopt the status of langevin - type stochastic equations , with @xmath34 playing the role of a stochastic force . as in the theory of brownian motion , the correlation functions of these `` forces '' then are of central importance . as a consequence of the fluctuation - dissipation theorem
, they acquire the forms @xcite @xmath36 where angular brackets indicate an ensemble average , overbars denote complex conjugation , greek letters specify vector components , and @xmath37 is the mean thermal energy of a quantum mechanical harmonic oscillator with frequency @xmath32 , omitting the vacuum contribution . as usual , @xmath38 is the inverse temperature variable ; @xmath39 is the imaginary part of the material s permittivity @xmath40 .
since the stochastic maxwell equations are linear , there exists a linear relationship between the fluctuating sources and the generated fields , which we write as @xmath41 the dyadic kernels @xmath42 and @xmath43 are referred to as the classical electric and magnetic green s function , respectively .
once these kernels are known for the given geometry , the correlation functions ( [ eq : jjcor ] ) allow one to evaluate bilinear expressions of the fields such as @xmath44 which enter into the definition of basic observables , such as the poynting vector of the thermal radiation emitted by the material , or its energy density . in the following section , we outline the construction of such green s functions for a dielectric layer of finite thickness .
readers not interested in the mathematical details of this construction can proceed to sec .
the electric dyadic green s function obeys the inhomogeneous vector wave equation @xcite @xmath45 where @xmath46 is the square of the wave number inside a material with relative permittivity @xmath6 . besides , the green s function also has to satisfy the proper boundary conditions for the respective geometry . for nonmagnetic materials ,
as considered here , the magnetic green s function is then easily obtained from the electric one through the relation @xmath47 for constructing green s functions , one starts from solutions @xmath48 to the homogeneous vector wave equation @xmath49 assuming cylindrical symmetry in planes orthogonal to the @xmath50-axis , the appropriate solutions are provided by the vector wave functions @xcite @xmath51\cos\end{matrix}\bigr\}}(n \varphi ) \mathbf{e}_\rho - \frac{\partial j_n(\lambda \rho)}{\partial \rho } { \bigl\{\begin{matrix}\cos\\[-1.3mm]\sin\end{matrix}\bigr\}}(n\varphi ) \mathbf{e}_\varphi \right ) { { \rm e}}^{{{\rm i}}h z } \label{eq : m } \\
\mathbf{n}_{\pm n \lambda}(h ) & = & \left ( \frac{{{\rm i}}h}{k } \frac{\partial j_n(\lambda \rho)}{\partial \rho } { \bigl\{\begin{matrix}\cos\\[-1.3mm]\sin\end{matrix}\bigr\}}(n\varphi ) \mathbf{e}_\rho \mp \frac{{{\rm i}}h n}{k } \frac{j_n ( \lambda \rho)}{\rho } { \bigl\{\begin{matrix}\sin\\[-1.3mm]\cos\end{matrix}\bigr\}}(n \varphi ) \mathbf{e}_\varphi \right . \nonumber
& & \qquad \left .
+ \frac{\lambda^2}{k } j_n ( \lambda \rho ) { \bigl\{\begin{matrix}\cos\\[-1.3mm]\sin\end{matrix}\bigr\}}(n \varphi ) \mathbf{e}_z \right ) { { \rm e}}^{{{\rm i}}h z } \ ; .
\label{eq : n}\end{aligned}\ ] ] here , @xmath52 denotes an ordinary bessel function of order @xmath53 ; the upper ( lower ) trigonometric function goes with the respective upper ( lower ) sign .
the integer @xmath54 is a discrete mode index , whereas the real , wave number - like index @xmath55 is continuous .
this wave - number index refers to propagation orthogonal to the @xmath50-axis , and determines the wave number @xmath56 for propagation in @xmath50-direction through the relation @xmath57 observe that , in contrast to @xmath58 , this wave number @xmath56 generally is complex .
it appears as the argument of the above vector functions , whereas the cylindrical coordinates @xmath59 of @xmath60 are suppressed .
the functions @xmath61 are associated with ( @xmath62-polarized ) te modes , the functions @xmath63 with ( @xmath64-polarized ) tm modes .
besides solving the homogeneous wave equation ( [ eq : homwe ] ) , these functions ( [ eq : m ] ) and ( [ eq : n ] ) also satisfy the useful identities @xmath65 it should be pointed out that , besides the vector functions of the @xmath66 and @xmath67-type , there also exists a third type denoted @xmath68 .
these functions obey @xmath69 and are not needed in the present macroscopic approach @xcite .
we now consider the geometry depicted in fig .
[ f_1 ] : the layer between the infinite planes @xmath70 and @xmath71 is filled by a dielectric with permittivity @xmath72 ; in this layer we have stochastic source currents generating the electromagnetic field . for later convenience
, we do not yet assume at this point that the layer is embedded in a vacuum , but rather that the region @xmath73 be occupied by another dielectric with permittivity @xmath74 , while the half - space @xmath75 be characterized by still another permittivity @xmath76 .
this will allow us to apply the results obtained here to bulk materials covered by thin coatings @xcite . if there were no boundary conditions to respect at the interfaces , but the layer @xmath77 were extended to fill the entire three - dimensional space , the dyadic electric green s function with boundary conditions at @xmath78 specifying `` outgoing '' waves could directly be adapted from the literature @xcite : for @xmath79 , one then has @xmath80 whereas the signs of the arguments
@xmath81 of all four vector functions have to be reversed when @xmath82 . here and in the following ,
an unprimed vector function @xmath66 or @xmath67 always carries the coordinates of the observation point @xmath60 , whereas a primed function @xmath83 or @xmath84 refers to the source point @xmath85 .
the symbol @xmath86 indicates a dyadic ( exterior ) product .
finally , @xmath87 is a shorthand notation for @xmath88 . for the slab geometry specified in fig .
[ f_1 ] , this function @xmath89 has to be modified such that , apart from outgoing boundary conditions at @xmath78 , also the boundary conditions at the two interfaces can be implemented . to this end
, we divide the space into four zones : assuming the source to be located within the slab , so that @xmath90 , we refer to the region @xmath91 as zone @xmath92
. then @xmath93 is zone @xmath94 , while the remaining piece @xmath95 of the slab becomes zone @xmath96 , and the positive half space @xmath75 is dubbed zone @xmath97 .
since the tangential component of the electric field is continuous at the interfaces , one requires @xmath98 continuity of the magnetic field s tangential components leads to the further boundary conditions @xmath99 these requirements can be met with the following , piecewise ansatz : in zone @xmath92 one has only radiation transmitted through the left interface , directed towards @xmath100 .
specifying still unknown transmission amplitudes @xmath101 and @xmath102 for the te and tm modes , respectively , we therefore write the electric green s function @xmath103 with @xmath60 in this zone in the form @xmath104 in zone @xmath94 there is leftward - directed radiation ( with unit amplitude ) emitted by the source , but also rightward - moving radiation reflected ( with amplitudes @xmath105 and @xmath106 ) from the left interface , together with further , leftward - moving radiation reflected ( with amplitudes @xmath107 and @xmath108 ) from the right interface : @xmath109 in zone @xmath96 one has , _ mutatis mutandis _ , the same dynamics as in zone @xmath94 , giving @xmath110 whereas zone @xmath97 provides a mirror image of zone @xmath92 : @xmath111 of course , the above ansatz is characterized in mathematical terms by stating that a suitable solution of the homogeneous vector wave equation ( [ eq : homwe ] ) has been added to the particular solution ( [ eq : gps ] ) of the inhomogeneous equation ( [ eq : inhwe ] ) .
we are now left with eight unknowns @xmath112 and @xmath113 , which match the number of boundary conditions provided by eqs .
( [ eq : bce ] ) and ( [ eq : bcm ] ) , since these apply independently to both the te and the tm modes .
we concentrate on the radiation field in zone @xmath97 , and therefore evaluate the transmission coefficients @xmath114 and @xmath115 .
this procedure is quite cumbersome , but elementary , so we immediately proceed to the result : the contribution of the te modes to the radiation emitted by the slab into zone @xmath97 is determined by @xmath116 \ ; , \label{eq : te4 } \end{aligned}\ ] ] while that of the tm modes follows from @xmath117 \ ; . \label{eq : tm4 } \end{aligned}\ ] ] for ease of notation , we henceforth omit the mode indices `` @xmath118 '' from the vector wave functions , and use the determinants @xmath119 and @xmath120 inserting these results ( [ eq : te4 ] ) and ( [ eq : tm4 ] ) into the electric green s function ( [ eq : egf ] ) , the radiation field existing in the half - space @xmath75 is completely specified , since the magnetic green s function for that zone is immediately obtained from eq .
( [ eq : mgf ] ) , keeping in mind the relations ( [ eq : rmn ] ) .
as an application of the above formalism we evaluate the intensity of thermal radiation emitted by the slab into the half - space @xmath75 , as given by the @xmath50-component of the poynting vector , @xmath121 where @xmath122 denotes the levi - civita tensor . the calculation is not trivial and involves several algebraic manipulations also encountered when computing the field s energy density , so we collect the main steps in appendix [ a_1 ] .
because of translational symmetry in planes parallel to the interfaces , the result does not depend on the cylindrical coordinates @xmath123 : @xmath124 where @xmath125 is the imaginary part of the wave number @xmath126 , and the dimensionless transmission coefficients @xmath127 and @xmath128 are given by @xmath129 \label{eq : tte } \\
t_\parallel & = & \frac{4 { { \rm re}}(h_3 \overline{\varepsilon}_{{\rm r}3 } ) } { |d_\parallel|^2 |\varepsilon_{{\rm r}3}|^2 } \biggl [ { { \rm re}}(h_2\overline{\varepsilon}_{{\rm r}2 } ) a_\parallel + { { \rm im}}(h_2\overline{\varepsilon}_{{\rm r}2 } ) b_\parallel \biggr ] \ ; , \label{eq : ttm}\end{aligned}\ ] ] employing the relative permittivities @xmath130 , together with the auxiliary , real quantities @xmath131 and @xmath132 in the case of thick layers , where @xmath133 , one has @xmath134 so that one recovers the well - known coefficients which determine the radiation emitted by a bulk material with planar surface @xcite : @xmath135 the opposite limiting case of very thin layers , with @xmath136 and @xmath137 , is less obvious . after some tedious algebra , one finds @xmath138 with @xmath4 . if both @xmath139 and @xmath140 , so that the layer is surrounded by vacuum , this gives @xmath141 since according to eq .
( [ eq : defh ] ) one has @xmath142 for evanescent modes with @xmath143 , only propagating modes with @xmath144 contribute to the poynting vector .
it is then a simple matter to invoke eq .
( [ eq : poyntingres ] ) for computing the intensity of thermal radiation emitted by a very thin dielectric layer into the vacuum : @xmath145 as may have been expected , this intensity is proportional to the thickness @xmath9 of the source layer ; when that thickness goes to zero , there are no sources left and the intensity vanishes .
it is also noteworthy that the emitted intensity acquires substantial contributions from frequency intervals within which the layer s permittivity is close to zero @xcite . besides the intensity , a further quantity of interest is the energy density @xmath146 of the electromagnetic field .
in contrast to the radiative intensity , this quantity is sensitive also to evanescent modes .
focussing on dielectrics facing the vacuum , so that @xmath140 , we then have to evaluate @xmath147 a calculation which largely parallels the one outlined in appendix [ a_1 ] eventually leads to @xmath148 where @xmath149 is understood when computing the transmission coefficients ( [ eq : tte ] ) and ( [ eq : ttm ] ) .
in order to clarify the physical significance of the preceding formal results ( [ eq : poyntingres ] ) and ( [ eq : edensres ] ) , we now express the key quantities specifying the radiative properties of the dielectric layer , the transmission coefficients ( [ eq : tte ] ) and ( [ eq : ttm ] ) , in terms of the fresnel amplitude reflection coefficients . for radiation coming from a medium with permittivity @xmath74 and going into one with permittivity @xmath150 ,
these fresnel coefficients are given by @xcite @xmath151 for te modes , and @xmath152 for tm modes .
introduction of these quantities necessitates to distinguish explicitly between propagating and evanescent modes . in all of this section
we assume @xmath140 , and thus study thermal radiation emitted into the vacuum . then propagating modes
are characterized by @xmath153 , evanescent ones by @xmath143 . defining four functions @xmath154 where both the fresnel coefficients @xmath155 of the left interface and , hence , also the functions @xmath156 and @xmath157 themselves carry either the label `` @xmath158 '' or `` @xmath159 '' , then rearranging the r.h.s . of eqs .
( [ eq : tte ] ) and ( [ eq : ttm ] ) with @xmath149 , we find their equivalent form @xmath160 \label{eq : tpr}\ ] ] for propagating modes , whereas the corresponding expression for evanescent modes with @xmath161 is given by @xmath162 \ ; , \label{eq : tev}\ ] ] again for both types of polarization . in the case of thick layers , @xmath163 ,
these coefficients reduce to @xmath164 so that one correctly reobtains the familiar expressions describing thermal radiation emitted by a bulk dielectric @xcite . however , here we are interested in the radiative properties of dielectric films thinner than the skin depth of the material under consideration , so that there are two competing length scales .
this case requires substantially more care : provided @xmath165 and @xmath166 , the above functions @xmath156 and @xmath157 reduce to @xmath167 in linear approximation . for strongly evanescent modes with @xmath168
one has @xmath169 , and hence @xmath170 . in the case of propagating modes ,
a similar hierarchy can be established only if one specifies the film s permittivity @xmath171 . in the following discussion , we restrict ourselves to simple metals described by the drude approach , resulting in the permittivity ( [ eq : drude ] ) , or in its approximation ( [ eq : hr ] ) , provided the hagen
rubens condition @xmath172 can be met for all relevant frequencies .
given such a drude metal , the equality @xmath173 requires @xmath174 for thermal frequencies . by means of the approximation ( [ eq : fgthin ] ) , one then finds @xmath175 for sufficiently thin films , so that contributions proportional to @xmath157 may be neglected .
hence , we have @xmath176 for thin metallic films . in terms of the dimensionless variable @xmath177 ,
the dependence of both propagating and evanescent thermal radiation on the film thickness is therefore determined by the function @xmath178 ^ 2 + a''^2(1-\xi)^2 } \ ; , \label{eq : ftrans}\end{aligned}\ ] ] with @xmath179 .
from here onwards we focus on metallic films in vacuum , so that @xmath180 , and first estimate that film thickness @xmath181 which results in maximum far - field heat radiation due to propagating modes .
we consider only modes directed perpendicular to the interfaces , so that @xmath182 , and @xmath183 for both types of polarization .
the hagen rubens condition @xmath172 entails @xmath184 , or @xmath185 , implying both @xmath186 and @xmath187 . with @xmath188 one
deduces @xmath189 and @xmath190 , so that it suffices to maximize , instead of the function @xmath191 , its approximate version @xmath192 since @xmath190 , the maximum is located at @xmath193 giving @xmath194 for the optimum film thickness .
this characteristic length ( [ eq : woff ] ) coincides exactly with the so - called woltersdorff thickness , which quantifies that thickness of a metallic film which maximizes its absorption for frequencies @xmath32 in the hagen
rubens regime @xcite .
indeed , it has been demonstrated experimentally that metal films absorb more infrared radiation if they are made thinner ; for most metals , maximum absorptance is obtained for films less than @xmath195 m thick @xcite . in terms of the skin depth ( [ eq : skin ] ) , which takes the form @xmath196 for metals in the infrared @xcite , and using @xmath4 , the woltersdorff thickness can be expressed as @xmath197 the observation that a film of thickness @xmath181 also maximizes its emission reflects the fact that absorption balances emission in thermal equilibrium . in fig .
[ f_2 ] we display numerical data for the radiative intensity emitted at temperature @xmath15 k by a drude metal with parameters @xmath198 s@xmath14 and @xmath199 s corresponding to bismuth , as function of the film thickness @xmath9 ; in all our numerical calculations we employ the model permittivity ( [ eq : drude ] ) without the hagen rubens approximation .
although macroscopic electrodynamics will presumably start to fail for thicknesses below @xmath195 m , we also plot data for even smaller @xmath9 , in order to clearly bring out the asymptotic trend .
this example confirms the picture drawn so far : the intensity emitted by the film almost coincides with that emitted by the bulk material when @xmath9 exceeds the skin depth , but increases substantially when @xmath9 is made smaller , reaching a maximum which exceeds the bulk value by a factor of about @xmath200 at a thickness predicted neatly by the woltersdorff formula ( [ eq : woff ] ) , and then starting to decrease .
however , the regime of linear decrease described by eq .
( [ eq : plin ] ) is reached only for unrealistically small @xmath9 . for metals with more typical drude parameters at room temperature , such as silver ( @xmath201 s@xmath14 and @xmath202 s ) or aluminium ( @xmath203 s@xmath14 and @xmath204 s ) , the maximum is shifted to even lower @xmath9 , as witnessed by fig .
[ f_3 ] , so that only radiative intensity increasing with decreasing film thickness might be observable in such cases .
it is noteworthy that the intensity generated by a thin film can exceed the bulk limit by more than a factor of @xmath205 .
while the dependence of the radiative intensity on the film thickness thus conforms to expectation , the dependence of the `` evanescent '' energy density , at some distance @xmath50 from the film surface , on that thickness is more difficult to oversee .
this is related to the fact that for strongly evanescent modes with @xmath168 both fresnel cofficients ( [ eq : fte ] ) and ( [ eq : ftm ] ) differ strongly : one finds @xmath206 for @xmath207 , so that @xmath208 but @xmath209 giving @xmath210 for @xmath207 in the hagen rubens regime . hence , for strongly evanescent tm modes we have @xmath211 and @xmath212 , which are precisely the propositions which have enabled us to reduce the transmission function ( [ eq : ftrans ] ) to the simpler form ( [ eq : fpr ] ) for propagating modes .
hence , we can immediately adapt the result obtained in eq .
( [ eq : prmax ] ) : for evanescent tm modes with @xmath207 , the optimum film thickness maximizing the energy density close to the film s surface is given by @xmath213 which implies @xmath214 within the hagen rubens approximation ( [ eq : aprtm ] ) , using @xmath215 . in fig .
[ f_4 ] we show a plot of the transmission function ( [ eq : ftrans ] ) for evanescent tm modes , obtained for bismuth parameters with @xmath32 kept fixed at the dominant thermal frequency @xmath216 for @xmath15 k. also indicated is the locus of maximizing values in the @xmath58-@xmath9-plane , as predicted by the approximation ( [ eq : maxtm ] ) ; obviously this approximation works quite well . on the other hand , for evanescent te modes the approximation ( [ eq : aprte ] ) implies @xmath217 , so that the transmission function now is cast into the different form @xmath218 ^ 2 } \ ; , \ ] ] with its maximum located at @xmath219 this predicts @xmath220 as the optimum thickness for strongly evanescent te modes , which simplifies to @xmath221 in the hagen rubens regime .
again , we display in fig .
[ f_5 ] a plot of the function ( [ eq : ftrans ] ) , and indicate the forecast of the approximation ( [ eq : maxte ] ) for maximum transmission .
comparison of figs .
[ f_4 ] and [ f_5 ] reveals entirely opposite trends followed by both types of modes : whereas for tm modes maximum transmission occurs for comparatively small @xmath58 , unless @xmath222 is excessively low , the maximizing @xmath58 grows with @xmath222 in the case of te modes .
these opposite trends leave their imprints in the near - field energy density .
the energy density @xmath223 associated with evanescent modes of either type of polarization at a distance @xmath50 from the film s surface is obtained by integrating the spectral density @xmath224 characterizing a film of thickness @xmath9 in vacuum , @xmath225 according to eq .
( [ eq : edensres ] ) , we have @xmath226 with @xmath227 or @xmath228 as given by eq .
( [ eq : tev ] ) , and @xmath161 . since eq .
( [ eq : tthin ] ) states @xmath229 for thin films , and since @xmath230 does not depend on @xmath58 for strongly evanescent modes , the spectral density @xmath231 associated with tm modes is given essentially by the integrated product @xmath232 .
since the factor @xmath233 has a well - developed maximum at @xmath234 , the spectral density is maximized if the factor @xmath191 is adapted to that maximum . in view of eq .
( [ eq : dev ] ) , it follows that @xmath235 is that film thickness which maximizes the spectral density @xmath231 at a given distance @xmath50 .
this reasoning is confirmed in fig .
[ f_6 ] , which shows the full density @xmath236 , including the contribution from propagating modes , for @xmath237 m and some representative frequencies @xmath32 , again using the example of bismuth .
it is important to observe that this density is dominated by low frequencies in the limit of thin films ; the maximizing thickness is well captured by the approximate eq .
( [ eq : dmax ] ) .
in marked contrast , the density @xmath238 effectuated by te modes is maximized at a thickness which becomes larger with decreasing frequency , as depicted in fig .
[ f_7 ] . for calculating the near - field energy density @xmath223 we scale all wavenumbers by @xmath50 , obtaining dimensionless quantities such as @xmath239 .
the resulting integral over @xmath240 can then be evaluated to zeroth order in the small parameter @xmath241 , provided the error thus committed falls into a range of frequencies where it is suppressed by the bose einstein function @xmath242 .
this requirement is satisfied if @xmath243 , where @xmath244 is the characteristic thermal wavelength at temperature @xmath1 . with this proviso ,
we are led to @xmath245 resulting in the near - field approximation @xmath246 \ ; . \label{eq : ugen}\end{aligned}\ ] ] when the film thickness is still large compared to the distance from the film , so that @xmath247 , this formula simplifies considerably and yields @xmath248 which coincides with the expression for the energy density close to the surface of an infinitely thick layer , _
i.e. _ , of a bulk material : for distances small compared to the film thickness , the energy density contains no information about that thickness . for metal films
, the above result reduces to @xmath249 for te modes , and to @xmath250 for tm modes . in both these cases , the first expression on the respective r.h.s .
is valid in general , whereas the second one requires the hagen
rubens approximation .
thus , for rather short distances @xmath247 the total energy density is dominated by the tm modes , exhibiting the familiar @xmath251-divergence known from bulk materials @xcite .
in contrast , the other limiting case of distances large compared to the film thickness , @xmath252 , gives rise to a fairly counterintuitive feature .
the general expression ( [ eq : ugen ] ) then takes the form @xmath253 for te modes , eq .
( [ eq : rperpev ] ) guarantees that the denominator appearing here may be replaced by unity .
this simplification then gives @xmath254 where once again the second approximate equality hinges on the hagen rubens condition .
observe that this result differs from the previous eq .
( [ eq : hrten ] ) for the reverse situation only by the factor @xmath255 , which appears reasonable : the energy density caused by te modes at a fixed distance @xmath243 decreases when reducing the film thickness . however , the situation is more delicate when dealing with tm modes , for which eq .
( [ eq : aprtm ] ) enforces @xmath211 and @xmath256 , so that the integrand in eq .
( [ eq : sing ] ) acquires a small denominator .
we still may write @xmath257 and now have to make a further distinction : keeping in mind that the decisive @xmath240 are on the order of unity , we may set @xmath258 provided @xmath259 for all relevant frequencies @xmath260 . in view of eq .
( [ eq : aprtm ] ) , this requires @xmath261 it is interesting to observe that this condition validating the approximation ( [ eq : aprden ] ) can be cast into a particularly suggestive form involving only the woltersdorff thickness ( [ eq : woff ] ) and the characteristic thermal wavelength ( [ eq : lax ] ) : @xmath262 in terms of the maximizing thickness ( [ eq : dmax ] ) , this means nothing but @xmath263 given this , we find @xmath264 for films which are not too thin .
this expression differs from its counterpart ( [ eq : hrtmn ] ) by the factor @xmath265 , which is noteworthy : in the near - field regime , the energy density caused by tm modes increases substantially when reducing the thickness of the film , despite the loss of source volume .
the fact that this feature emerges only for the tm modes indicates that it is related to surface plasmon polaritons .
indeed , when the film thickness @xmath9 is sufficiently small , the plasmons associated with the two surfaces couple , splitting into one resonance with a frequency that converges to @xmath11 for @xmath266 , and a second resonance with a frequency that approaches zero @xcite .
this is exemplified in fig .
[ f_new ] , where we plot the local density of states @xmath267 , defined through the relation @xcite @xmath268 for a distance of @xmath269 m from bismuth films of various thicknesses .
the increase of the energy density @xmath270 with decreasing film thickness found in eq .
( [ eq : hrtmf ] ) occurs when the low - frequency surface plasmon polariton comes into the range of the thermal frequencies .
our findings are illustrated in fig .
[ f_8 ] , which shows the individual densities @xmath271 and @xmath272 at a distance @xmath237 m from the surface of a bismuth film , together with their sum , again as functions of the film thickness @xmath9 .
whereas the density caused by te modes decreases monotonically with decreasing @xmath9 , the one associated with tm modes actually increases , and approaches a finite value for @xmath266 .
these different trends obeyed by the two types of modes result in a non - monotonic dependence of the total energy density at @xmath50 on the thickness @xmath9 .
the limiting case of very thin films is reached when @xmath273 , or @xmath274 this limit may be hard to realize in practice , and fall outside the regime of validity of macroscopic electrodynamics for most materials , but it is nonetheless of conceptual interest .
it necessitates to approximate the denominator in the expression ( [ eq : sing ] ) in the form @xmath275 with none of the two terms on the r.h.s . being negligible against the other for all relevant frequencies . inserting ,
one is confronted with @xmath276 rescaling the @xmath240-integral according to @xmath277 we arrive at @xmath278 clearly , here the @xmath32-integral is dominated by low frequencies in the small-@xmath9-limit , as discussed before in the context of fig .
[ f_6 ] ; consequently , we are entitled to replace @xmath242 by @xmath279 .
then interchanging the order of integration , and using @xmath280 finally results in @xmath281 remarkably , the energy density associated with evanescent tm modes not only remains finite when approaching the ( formal ) limit of zero thickness , but it also becomes independent of the metal s parameters .
this universal feature stems from the fact that the low - frequency surface plasmon polariton resonance depicted in fig .
[ f_new ] converges to zero frequency for all drude materials . nonetheless , the width of this resonance depends on the relaxation time @xmath12 , so that the scale @xmath282 below which the universal behavior appears is proportional to @xmath283 .
as a consequence , metals such as gold and bismuth , which possess nearly identical plasma frequencies but substantially different relaxation times , reveal the above universality for rather different film ticknesses . as will be discussed in ref .
@xcite , the rise of the evanescent energy density expressed by eqs .
( [ eq : hrtmf ] ) and ( [ eq : hrtmuni ] ) also occurs with thin metal films coating a polar dielectric , but is lost when a thin metal film covers another metal .
figure [ f_9 ] shows a doubly logarithmic plot of the dependence of the energy density @xmath284 associated with te modes on the distance @xmath50 from the film , summing up evanescent and propagating contributions .
again we take drude parameters for bismuth at temperature @xmath15 k , and consider films of thickness @xmath285 m and @xmath286 m , together with an excessively thin model example with @xmath287 m. in the first two cases , one clearly recognizes the crossover from the @xmath288-behavior predicted for @xmath247 by the approximation ( [ eq : hrten ] ) to the @xmath289-dependence implied by eq .
( [ eq : hrtef ] ) for larger distances , @xmath252 .
as indicated by the arrow , @xmath284 decreases at fixed @xmath50 when @xmath9 is reduced .
figure [ f_10 ] displays the corresponding plot for tm modes . here
the data for @xmath285 m and @xmath290 m exhibit the change of slope from @xmath291 , as deduced from eq .
( [ eq : hrtmn ] ) for @xmath247 , to @xmath292 , as required by eq .
( [ eq : hrtmf ] ) for @xmath293 .
moreover , for the model case @xmath287 m one observes another crossover from that slope @xmath292 appearing when @xmath294 to the universal @xmath251-behavior found in eq .
( [ eq : hrtmuni ] ) in the opposite limit . finally , we display in fig .
[ f_11 ] the total energy densities for these examples . as a consequence of the opposing trends obeyed by the te and tm modes
, there is a well - developed range of distances around @xmath295 m where the total density varies non - monotonically with the film thickness .
a theoretical treatment of heat radiation and thermal near fields generated by thin dielectric slabs within the framework of macroscopic fluctuational electrodynamics hinges on two basic ingredients . on the one hand ,
maxwell s equations have to be solved with the boundary conditions imposed by the slab geometry ; on the other , the dielectric permittivity @xmath0 for the slab s material is required .
we have addressed the first problem by constructing the dyadic green s functions for the slab in sec .
[ s_3 ] , and stated general expressions for both the intensity of the heat radiation and its energy density , valid for any prescribed permittivity @xmath0 , in sec .
[ s_4 ] .
the restricion to metallic films , with dielectric response solely due to free particle - like electron motion , leads to an intricate competition of length scales . besides the geometrical thickness @xmath9 of such a film , and the skin depth ( [ eq : skin ] ) , the woltersdorff thickness @xmath181 defined in eq .
( [ eq : woff ] ) and the characteristic thermal wavelength @xmath296 come into play . as a consequence
, the dependence of the thermal near - field energy density on the distance @xmath50 from the film s surface is not characterized by a single exponent , but different exponents dominate in different regimes . in particular , we have shown that for @xmath297 the energy density is dominated by the universal contribution ( [ eq : hrtmuni ] ) brought about by tm modes .
our conclusions are based on the drude model ( [ eq : drude ] ) for the permittivity , regarding the plasma frequency @xmath11 and the relaxation time @xmath12 as fixed .
this does not hold exactly for electrical transport in ultrathin metallic films , since the resistivity of such a film increases with decreasing thickness @xcite , but we expect the main features of our discussion to persist at least qualitatively . with a view towards laboratory experiments , one might be interested not in metallic films in vacuum , but rather in thin metal coatings on a planar surface of another material .
this case has been anticipated in our general formulae ( [ eq : tpr ] ) and ( [ eq : tev ] ) , where the fresnel reflection coefficients for the left interface differ from those for the right one , but one still has to account for the bulk contribution @xcite .
qualitatively , however , the fabry perot - like effect brought about by thin metal coatings bears interesting possibilities of manipulating the intensity of thermal radiation , that is , the amount of energy transported per time from a hot body into the vacuum .
the insights obtained in this work are especially pertinent with regard to recent developmemts in scanning thermal microscopy @xcite .
it has been argued that the tip of a thermal microscope operating in ultrahigh vacuum is sensitive to the near - field energy density above the sample ; experiments along this direction are presently underway @xcite . near - field signatures offered by thin films , or by bulk materials coated with films of varying
, exactly specified thickness even down to that of atomic monolayers , and their actual observation in a suitable experiment could be of value for developing scanning thermal microscopy into a quantitative tool for materials science .
while we have formally extrapolated the realm of macroscopic electrodynamics down to even unrealistically small scales , it goes without saying that at some scale deviations from the macroscopic description will show up ; such deviations now have come within the scope of modern experimental set - ups @xcite .
in addition , one has to expect important corrections due to effects of non - local optical response @xcite . the pursuit of the question at precisely what length scale , and in precisely what manner
, these deviations start to manifest themselves is a quite important task in nanoscale thermal engineering . in this quest ,
the present study may serve as a reference .
b. acknowledges support from the studienstiftung des deutschen volkes .
we thank u. fleischmann - wischnath , o. huth , a. kittel , a. knbel , j. parisi , and f. rting for helpful discussions and criticism .
in this appendix we provide technical details required for the derivation of the expression ( [ eq : poyntingres ] ) for the intensity of heat radiation emitted by a dielectric layer .
computing the magnetic green s function @xmath298 according to eq .
( [ eq : mgf ] ) from the electric green s function ( [ eq : egf ] ) , and defining the convenient symbols @xmath299 one starts from the product of the two dyadics and integrates , obtaining @xmath300 \nonumber \\ & + & \frac{|k_2|^2}{|d_\parallel|^2 k_3 } \ , \mathbf{n}(h_3 ) \otimes
\overline{\mathbf{m}}(h_3 ) \cdot \nonumber \\ & & \quad \biggl [ |a_\parallel^{12}|^2 { { \rm e}}^{2 h_2 '' d } i_{\mathbf{n}}^{-- } - a_\parallel^{12 } \overline{b_\parallel^{12 } } { { \rm e}}^{- 2 { { \rm i}}h_2 ' d } i_{\mathbf{n}}^{-+ } - b_\parallel^{12 } \overline{a_\parallel^{12 } } { { \rm e}}^ { 2 { { \rm i}}h_2 ' d } i_{\mathbf{n}}^{+- } + |b_\parallel^{12}|^2 { { \rm e}}^{-2 h_2 '' d } i_{\mathbf{n}}^{++ } \biggr ] \biggr\ } \ ; .
\label{eq : long}\end{aligned}\ ] ] here we have introduced the source integrals @xmath301 where the integration extends over the dielectric layer labeled `` 2 '' , that is , over the slab @xmath77 occupied by the sources ( cf . fig .
[ f_1 ] ) ; the symbol @xmath302 is given by @xmath303 likewise , one has @xmath304 the emergence of the 8 different source integrals ( [ eq : im ] ) and ( [ eq : in ] ) is a characteristic complication brought about by the finite thickness of the slab ; it can be seen as a consequence of multiple reflections inside the slab . inserting the right hand sides of these integrals , the longish expression ( [ eq : long ] )
is reduced to @xmath305 \mathbf{m}(h_3 ) \otimes \overline{\mathbf{n}}(h_3 ) \nonumber \\ & + & \frac{|k_2|^2}{|d_\parallel|^2 k_3 } \ , \biggl [ \frac{a_\parallel}{h_2 '' } \frac{\lambda^2 + |h_2|^2}{|k_2|^2 } + \frac{b_\parallel}{h_2 ' } \frac{\lambda^2 - |h_2|^2}{|k_2|^2 } \biggr ] \mathbf{n}(h_3 ) \otimes \overline{\mathbf{m}}(h_3 ) \biggr\ } \ ; .\end{aligned}\ ] ] in the next step , we evaluate the vector product appearing in the definition of the poynting vector ( [ eq : poynting ] ) .
because of translational invariance in the planes orthogonal to the @xmath50-axis , it suffices to consider @xmath306 only . with the help of @xmath307 and @xmath308 and exploiting the elementary but important relations @xmath309 where @xmath4 , one is led to @xmath310 \nonumber \\ & + & \frac{h_3}{|d_\parallel|^2 \varepsilon_{r3 } } \ , \biggl [ \frac{{{\rm re}}(h_2\overline{\varepsilon}_{r2})}{h_2 ' h_2 '' } a_\parallel + \frac{{{\rm im}}(h_2\overline{\varepsilon}_{r2})}{h_2 ' h_2 '' } b_\parallel \bigg ] \biggr\ } \ ; , \label{eq : zwerg}\end{aligned}\ ] ] with the symbols @xmath311 , @xmath312 , @xmath313 , and @xmath314 as specified in eqs .
( [ eq : syma ] ) and ( [ eq : symb ] ) . inserting this expression ( [ eq : zwerg ] ) into eq .
( [ eq : poynting ] ) , adding the complex conjugate , and utilizing the identity @xmath315 for the imaginary part of the permittivity , one finally ends up with the result stated in eq .
( [ eq : poyntingres ] ) . | we study the properties of thermal radiation emitted by a thin dielectric slab , employing the framework of macroscopic fluctuational electrodynamics .
particular emphasis is given to the analytical construction of the required dyadic green s functions .
based on these , general expressions are derived for both the system s poynting vector , describing the intensity of propagating radiation , and its energy density , containing contributions from non - propagating modes which dominate the near - field regime .
an extensive discussion is then given for thin metal films .
it is shown that the radiative intensity is maximized for a certain film thickness , due to fabry perot - like multiple reflections inside the film .
the dependence of the near - field energy density on the distance from the film s surface is governed by an interplay of several length scales , and characterized by different exponents in different regimes .
in particular , this energy density remains finite even for arbitrarily thin films .
this unexpected feature is associated with the film s low - frequency surface plasmon polariton .
our results also serve as reference for current near - field experiments which search for deviations from the macroscopic approach . |
identifying the stars that harbor circumstellar disks within young clusters and associations is an essential step in using disks to study star and planet formation . for instance
, thorough surveys for disks provide well - defined samples of targets for detailed measurements of disk properties .
in addition , by comparing the prevalence of disks in different evolutionary stages within a given cluster and among clusters across a range of ages , one can estimate the lifetimes and clearing timescales of disks , which in turn constrain models of disk evolution and planet formation .
the best available tool for disk surveys has been mid - infrared ( ir ) imaging with the _
spitzer space telescope _
_ spitzer _ has offered the sensitivity to detect the photospheres of young brown dwarfs in nearby clusters ( @xmath10 pc ) , the field of view to efficiently map these clusters ( @xmath11 ) , the angular resolution to resolve most of their members ( @xmath12 ) , and a suite of mid - ir filters that could roughly classify the evolutionary stages of disks ( 3.624 ) .
although _ spitzer _ has focused on relatively compact clusters ( @xmath13 deg@xmath14 , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , it has also imaged more widely distributed populations in areas as large as @xmath1 deg@xmath14 @xcite .
the upper scorpius subgroup in the scorpius - centaurus ob association is an attractive target for disk surveys .
it contains one of the nearest and richest populations of young stars ( @xmath15165 pc , @xmath16 , * ? ? ?
* ) , making it possible to measure disk properties down to low stellar masses with good statistical accuracy .
perhaps most importantly , upper sco is at an age where both primordial disks and second - generation debris disks are abundant @xcite , and thus can potentially provide vital constraints on disk evolution .
most studies have adopted a mean age of 5 myr for upper sco @xcite , but @xcite recently derived a significantly older value of 11 myr , which affects the timescales of disk evolution derived from any census of disks in the region .
the ages of upper sco and its neighboring subgroups remain a subject of debate @xcite .
_ spitzer _ has been used to identify and classify disks in upper sco based on mid - ir photometry @xcite and to study the mineralogy of dust in those disks with mid - ir spectroscopy @xcite . in one of the most widely - cited results from the _ spitzer _ data
, @xcite found that , in a sample of @xmath17 members , nearly all of the stars more massive than the sun have lost their primordial disks while a significant fraction of the low - mass stars ( @xmath181/5 ) have retained them , indicating that disk lifetimes may be longer at lower stellar masses .
however , because upper sco is spread across a large area ( @xmath19 deg@xmath14 ) , _ spitzer _ imaged members individually rather than through a map of the association , and only a subset of the stellar population was observed . as a result ,
the potential of the association for constraining the properties of disks was not fully exploited .
the _ wide - field infrared survey explorer _ ( _ wise _ , * ? ? ? * ) has presented an opportunity to significantly improve the census of disks in upper sco . in 2010 ,
the satellite imaged the entire sky at 3.4 , 4.6 , 12 , and 22 , providing mid - ir photometry for the areas of upper sco that were not observed by _
spitzer_. although they have lower sensitivity and angular resolution than _ spitzer _ images , the data do detect every known member of upper sco ( that is resolved ) in the most sensitive bands , including the brown dwarfs .
photometry has already been used to search for disks around members of upper sco at b , a , and f types @xcite and mid- to late - m types @xcite . to obtain the most thorough census of disks in upper sco that is possible with available data
, we have compiled all photometry between 324 from both _ spitzer _ and for all known members of the association ( section [ sec : data ] ) .
we use these data to identify the members that exhibit significant mid - ir excesses ( section [ sec : exc ] ) and to estimate the evolutionary stages of the detected disks ( section [ sec : class ] ) .
we then measure the distribution of excesses as a function of spectral type , wavelength , disk class , and excess size and discuss the resulting implications for disk evolution ( section [ sec : global ] ) .
to characterize the population of disks in upper sco , we began by constructing a list of all published sources that are likely to be members of the association based on proper motions or diagnostics of youth .
we also have included new members that we have uncovered through an ongoing survey of the association .
the basis for inclusion in our membership catalog will be described in a separate study ( k. luhman , in preparation ) .
we have treated known binaries as single sources if they are unresolved by both the two micron all - sky survey ( 2mass , * ? ? ?
* ) and the united kingdom infrared telescope ( ukirt ) infrared deep sky survey ( ukidss , * ? ? ?
thus , secondaries that have been resolved only by high - resolution imaging with the _ hubble space telescope _ or adaptive optics do not receive separate entries in our catalog of members @xcite .
our resulting list of known members of upper sco contains 863 objects .
we present their names and spectral types in table [ tab : all ] .
our compilation of spectral classifications is not fully exhaustive ; additional measurements have been reported in the literature for some of the stars , particularly at earlier types , but they generally agree with the types in table [ tab : all ] . for our census of disks in upper sco , we make use of images at 3.6 , 4.5 , 5.8 , and 8.0 obtained with
_ spitzer _
s infrared array camera ( irac ; * ? ?
* ) and images at 24 obtained with the multiband imaging photometer for _ spitzer _ ( mips ; * ? ? ?
these bands are denoted as [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , [ 8.0 ] , and [ 24 ] .
the fields of view were @xmath20 and @xmath21 for irac and the 24 band of mips , respectively .
the cameras produced images with fwhm@xmath22@xmath23 from 3.6 to 8.0 and fwhm=@xmath24 at 24 . several previous studies have reported _ spitzer _ photometry for samples of members of upper sco @xcite . in this work , we consider all irac and mips 24 observations of all members in our catalog . these data were obtained through guaranteed time programs 19 ( j. houck ) , 40 , 58 ( g. rieke ) , 72 ( f. low ) , and 84 ( m. jura ) , director s discretionary time program 248 ( s. mohanty ) , legacy programs 139 , 173 , 177 ( n. evans ) , 148 ( m. meyer ) , and 30574 ( l. allen ) , and general observer programs 20069 ( j. carpenter ) , 20103 ( l. hillenbrand ) , 20146 ( k. gordon ) , 20435 ( r. jayawardhana ) , 20803 ( s. mohanty ) , 30765 ( k. stapelfeldt ) , 50025 ( p.
harvey ) , 50554 ( i. song ) , and 80187 ( k. luhman ) .
we have measured photometry from these images for the members of upper sco with the methods described by @xcite .
we present our measurements of _ spitzer _ photometry for upper sco in table [ tab : all ] .
some members were observed at multiple epochs , most of which were separated by a few days .
since none of these objects exhibit significant variability in these data , we report the mean of multiple measurements weighted by the inverse square of their flux errors . for each of the irac bands ,
the camera detected every member that appeared within the field of view and was resolved .
this was not the case for the 24 images from mips .
all of the non - detections in this band are indicated in table [ tab : all ] .
the typical detection limit is @xmath25\sim10 $ ] .
we have not measured 24 photometry for hip 80425 and hip 80338 since they are dominated by extended emission .
2mass j16105240 - 1937344 also may be significantly contaminated by extended emission at 24 .
we report a tentative measurement for this star , but it is excluded from the analysis in this paper .
@xcite noted that their 24 data for several additional stars were unreliable because of extended emission .
based on visual inspection of the images , we find that these stars are point sources that are sufficiently bright compared to spatial variations in the background emission so that our aperture photometry should be reasonably accurate .
this conclusion is supported by the fact that our 24 data for these stars agree with the fluxes expected from the photospheres .
irac and mips observed 381 and 442 members of upper sco , respectively .
a total of 484 members were observed by at least one of the cameras , corresponding to 56% of the known members .
irac and mips data have not been previously published for 117 and 124 of these sources , respectively , which includes 48 non - detections by mips .
in addition to the pointed _ spitzer _ observations , we utilize mid - ir photometry from the all - sky imaging survey by _ wise_. the bands of are centered at 3.4 , 4.6 , 12 , and 22 , which are denoted as w1 through w4 @xcite .
the angular resolution of the images is @xmath26 in the first three bands and @xmath27 in w4 .
we have retrieved the coordinates and profile - fit photometry of all objects in the preliminary release source catalog and the all - sky source catalog that are within a distance of @xmath28 from members of upper sco .
for w1 and w2 , most of the data agree within 5% between the two catalogs , and the catalogs exhibit comparable agreement with the _ spitzer _ data .
the width of the sequence of diskless stars in @xmath29 and @xmath30 as a function of spectral type is also similar for the two catalogs ( section [ sec : exc ] ) .
we adopt the w1 and w2 data from the all - sky source catalog since it was generated from a newer processing of the images .
however , we find that the sequences of diskless stars in @xmath31 and @xmath32 are noticeably tighter with the data from the preliminary release source catalog for the fainter stars , which implies better photometric accuracies in this catalog for w3 and w4 .
this result is supported by a comparison of w4 to our measurements of [ 24 ] ; the standard deviation of @xmath33 $ ] is 0.18 and 0.28 mag using the preliminary release and all - sky data , respectively .
therefore , we adopt the w3 and w4 measurements from the preliminary release source catalog , with the exception of 2mass j16045716 - 2104160 , 2mass j16133647 - 2327353 , and 2mass j16133688 - 2327298 , for which measurements are available from only the all - sky source catalog . on average , the data from the preliminary release source catalog for upper sco
are 0.03 mag brighter in w3 and 0.035 mag fainter in w4 than the data from the all - sky source catalog .
we have modified our catalog of data in several ways . since the w2 fluxes for saturated stars
are significantly overestimated ( 0.21 mag at w2@xmath34 , * ? ? ?
* ) , we have omitted these measurements from our tabulation .
these empty entries in the catalog are labeled with the flag err " .
biases of this kind are also present in the other bands for the brightest stars , but they are much smaller ( @xmath35 mag ) , so we retain these measurements .
although the all - sky source catalog contains data for antares , we label it as saturated in our catalog since it is too bright for reliable photometry . through visual inspection of the images ,
we have checked for members of upper sco whose photometry may be affected by the psf of another star . each band
in which this occurs is labeled with the flag bl " ( for blend " ) .
we also have inspected the images to check for unreliable detections and measurements that may be contaminated by extended emission . for the former ,
we have omitted the photometry and have added the flag false " to our catalog .
the photometry is labeled with the flag ext " if extended emission is suspected based on the visual inspection .
if measurements are known to be dominated by extended emission based on higher - resolution images from _ spitzer _ , they are omitted from the catalog .
some studies have excluded data that have high values of the reduced @xmath36 for the profile fitting in the catalog because this may indicate the presence of extended emission @xcite .
however , we find that most of the members of upper sco with high @xmath36 do not show evidence of extended emission in images from _ spitzer _ , when available .
therefore , we rely on the steps that we have described rather than the @xmath36 flag for identifying extended emission . finally , for the source in the catalog matched to each member , we have visually checked whether its position differs significantly ( @xmath37 ) among the four bands . by doing so
, we can identify erroneous matches in the catalog between one object that dominates at shorter wavelengths and a different object that dominates at longer wavelengths , as in the case of the components of a binary system in which only the secondary has a disk ( e.g. , 2mass j16101888 - 2502325 and 2mass j16101918 - 2502301 ) and stars and red galaxies with small projected separations ( e.g. , sco pms 17 , * ? ? ?
we have examined the images of all members that lack counterparts in the catalog .
several of the unmatched members are not detected because of their proximity to brighter stars .
we label their empty entries with the flag unres " . among the pairs separated by less than @xmath38 ( the fwhm for the shorter bands )
, we have attached bin " ( for binary " ) to the photometry of the brighter components to indicate a possible contribution from the fainter stars . as noted in section
[ sec : mem ] , many known close companions are not included in our catalog of members .
thus , they represent additional unresolved components in the data that are not marked in our catalog .
finally , through visual inspection of the images , we find that did detect four members that are absent from the source catalogs .
we found data for two of these stars , 2mass j16071750 - 1820348 and 2mass j16001730 - 2236504 , in the all - sky reject table .
we measured photometry for the remaining stars , 2mass j16090407 - 2417588 and 2mass j16081758 - 2348508 , by applying aperture photometry to the images .
we present our catalog of photometry for the known members of upper sco in table [ tab : all ] .
a dusty circumstellar disk within a few au of a star is brightest at ir wavelengths , becoming brighter than the star beyond @xmath39 . as a result
, one can detect the presence of a disk based on ir emission that is greater than that expected from a stellar photosphere .
we have applied this technique to the _ spitzer _ and photometry in upper sco to identify the members that have disks . to detect and measure excess emission for each member
, we have used the colors produced by the _
spitzer _ and bands relative to @xmath40 ( 2.2 ) .
we use @xmath40 for these colors because it is long enough in wavelength so that extinction is low while short enough in wavelength that it is usually dominated by the stellar photosphere in systems with disks . for most stars ,
we adopt @xmath41 from the 2mass point source catalog or @xmath40 from the seventh data release of ukidss , whose photometric systems are similar @xcite .
we use @xmath40 photometry from @xcite and @xcite for a few of the brightest members since their 2mass data have large uncertainties .
we focus on the _ spitzer _ bands [ 4.5 ] , [ 8.0 ] , and [ 24 ] and the bands w2 , w3 , and w4 for detecting excess emission
. we do not consider [ 3.6 ] and w1 since they should show the smallest excesses among the _
spitzer _ and bands .
the data at [ 5.8 ] is discussed only briefly since it is available for a relatively small number of members .
antares is excluded from our analysis since it is a red supergiant and it lacks data from _
spitzer _ or _
wise_. the color excess of a star is measured by comparing its observed color to the color expected for its stellar photosphere .
because of its richness and moderately advanced age , upper sco contains a large number of diskless stars , providing an accurate determination of the colors of young photospheres .
since photospheric colors can vary with spectral type , we plot the ir colors in upper sco as a function of spectral type in figure [ fig : pan ] .
the colors are not corrected for extinction , which is typically @xmath422 in this association .
we have indicated in figure [ fig : pan ] the known and suspected be stars @xcite and the candidates for disks in different evolutionary classes that are identified in section [ sec : class ] and in previous studies @xcite . for each color in figure
[ fig : pan ] , the data exhibit a narrow sequence of blue sources and a broader distribution of redder sources , corresponding to stellar photospheres and stars with disks , respectively .
we discuss the measurement of excesses from these colors in the remainder of this section .
the results of this analysis are summarized in table [ tab : all ] , where we indicate whether each member of upper sco has excess emission in each of the six _ spitzer _ and bands that we have considered .
we examine the data in [ 4.5 ] and w2 together since these bands have similar effective wavelengths ( 4.5 and 4.6 ) .
photometry in [ 4.5 ] and w2 is available for 359 and 817 members of upper sco , respectively . the stars that lack [ 4.5 ] were not observed by irac while the stars that lack w2 are too bright for good photometry ( section [ sec : wise ] ) or are blended with other sources .
thus , all members were detected in both bands as long as they were observed ( by irac ) and resolved .
26 members have [ 4.5 ] but not w2 , which consist of stars that are too bright for a reliable w2 measurement and companions that are resolved by irac but not _ wise_. as a result , 843 members have data in at least one of the two bands . in the diagrams of @xmath43
$ ] and @xmath30 versus spectral type in figure [ fig : pan ] , we have used the sequences of diskless stars to select boundaries for identifying stars that exhibit significant color excesses .
we use the same boundary for both colors since the data in [ 4.5 ] and w2 have an average offset of less than a few percent . however , the photospheric sequences are not well - defined at the latest spectral types because of the small number of members .
the bluest l - type members have colors near 1.2 , which is redder than the photospheric values of @xmath43 $ ] estimated by @xcite based on young l dwarfs in the solar neighborhood ( @xmath44 - 100 myr ) .
to better define the photospheric colors for l - type members of upper sco , we include in figure [ fig : pan ] data for members of taurus that are later than m9 and that are diskless based on photometry at longer wavelengths @xcite .
these data coincide with the bluest sources in upper sco and confirm the rapid increase in @xmath43 $ ] and @xmath30 beyond m9 .
the boundaries for @xmath43 $ ] and @xmath30 are defined by lines connecting ( b0 , 0.19 ) , ( k0 , 0.22 ) , ( m1.5 , 0.47 ) , and ( m8.5 , 0.87 ) . since the dependence on spectral type of these colors remains somewhat uncertain at m9l2 , we end the boundaries in figure [ fig : pan ] at m8.5 .
measurements of excesses at the latest types are also sensitive to uncertainties in spectral types because of the rapid increase in the photospheric colors .
nevertheless , we have inspected the data to check for members later than m8.5 that show obvious excesses .
[ lhj2006 ] j163919.15 - 253409.9 has an excess in w2 but not [ 4.5 ] .
this discrepancy is a reflection of the fact that w2 is brighter than [ 4.5 ] by 0.35 mag .
a similar offset is present between w1 and [ 3.6 ] , indicating that the source is variable .
[ lhj2007 ] j160918.69 - 222923.7 exhibits the same characteristics .
mid - ir variability of this size is more likely to occur in stars with disks @xcite .
therefore , we classify these two objects as having excesses in w2 .
we tentatively conclude that the remaining members later than m8 do not have excesses in [ 4.5 ] and w2 . to refine our identifications of excesses , we have checked for sources that are redder than the boundaries for [ 4.5 ] or w2 in figure [ fig : pan ] but have data at longer wavelengths that are consistent with a stellar photosphere .
all stars of this kind are only slightly above the excess thresholds in [ 4.5 ] and w2 .
therefore , we indicate that they lack [ 4.5 ] and w2 excesses in table [ tab : all ] .
these stars consist of 2mass j16072682 - 1855239 , 2mass j16104202 - 2101319 , 2mass j16133688 - 2327298 , 2mass j16152024 - 2333588 , 2mass j16203456 - 2430205 , 2mass j16204144 - 2425491 , 2mass j16370753 - 2432395 , sco pms 48 , hip 78977 , sco pms 42b , and hip 81455 .
the latter two stars may have small excesses in [ 24 ] ( section [ sec : exc24 ] ) , but they lack excess emission in w3 . several stars with excesses in [ 4.5 ] or w2 do not have photometry at longer wavelengths , and thus can not be examined in this way , because they were below the detection limit ( w3 , w4 , [ 24 ] ) or were not observed ( [ 5.8 ] , [ 8.0 ] , [ 24 ] ) .
we also have checked for members that have excesses in [ 4.5 ] but not w2 , or vice versa .
two objects of this kind were discussed earlier in this section in the context of the late - type members .
three additional stars show this discrepancy , consisting of 2mass j16090075 - 1908526 , 2mass j16095933 - 1800090 , and rx j1604.3 - 2130a .
as in the earlier cases , variability is the likely cause of the differences between @xmath43 $ ] and @xmath30 for these stars given that other bands indicate the presence of disks . for rx j1604.3 - 2130a ,
previous ir data have exhibited significant variability as well @xcite .
however , the variability in this star is rather unusual .
its @xmath45-[8.0]$ ] color and its spectrum from @xcite shortward 16 are consistent with a stellar photosphere , but the bands over this same wavelength range ( w1 , w2 , w3 ) show excess emission relative to @xmath41 and each other , which may be caused by changes in the structure of the inner disk @xcite .
photometry in [ 8.0 ] has been measured for 335 members of upper sco .
2mass j16075850 - 2039485 is excluded from the diagram of @xmath46 $ ] in figure [ fig : pan ] because its 8 flux is uncertain . as done for @xmath43 $ ] and @xmath30 , we have defined a threshold that varies with spectral type for identifying excesses in @xmath46 $ ] .
it is defined by lines connecting ( b0 , 0.2 ) , ( k0 , 0.32 ) , ( m0 , 0.45 ) , and ( m9 , 1.01 ) .
excesses from disks are larger at longer wavelengths , resulting in a greater separation between diskless and disk - bearing stars in @xmath46 $ ] compared to @xmath43 $ ] and @xmath30 . in all three of these colors , some of the early - type stars are 0.10.2 mag redder than the sequence of the bluest stars . most of these excesses can be attributed to extinction based on their lack of excesses at longer wavelengths and their red near - ir colors .
however , a few of these moderately red stars may have small amounts of emission from debris disks @xcite , even though they appear below our adopted boundaries for excesses in figure [ fig : pan ] .
finally , we have checked for discrepancies between apparent excesses in @xmath46 $ ] and data in other bands .
2mass j16210222 - 2358395 is slightly above our excess threshold for @xmath46 $ ] , but [ 8.0 ] does not show an excess relative to irac and wise data at shorter wavelengths .
therefore , the excess in @xmath46 $ ] may be due to stellar variability between the observations at @xmath41 and the mid - ir bands . in w3 , w4 , and
[ 24 ] , some members of upper sco are not detected or have large uncertainties in their photometry . for identifying excesses in these bands
, we consider only measurements with uncertainties less than 0.25 mag .
we also exclude stars that are flagged as extended .
these criteria are satisfied in w3 by 658 members .
our adopted threshold for excesses in @xmath31 is shown in figure [ fig : pan ] , which is defined by lines connecting ( b0 , 0.18 ) , ( g8 , 0.33 ) , and ( m9 , 1.52 ) .
for stars that are slightly above the boundary for excesses , we have checked whether excess emission is present in w4 or [ 24 ] .
hip 78977 lacks an excess in either band and 2mass j16233234 - 2523485 should have little if any excess in w4 based on its non - detection .
therefore , we conclude that these stars probably do not have excesses in w3 .
given the limits available at [ 24 ] and w4 , 2mass j16181618 - 2619080 , 2mass j16253672 - 2224285 , 2mass j16053077 - 2246200 , and 2mass j16044303 - 2318258 have colors of @xmath47 in @xmath32 and @xmath48 $ ] , indicating that they could have excess emission at w4 and [ 24 ] .
since we can not rule out the presence of excesses at longer wavelengths , we tentatively classify these stars as having excesses in w3 .
@xcite found that hip 79878 has excess emission in w3 , but it is slightly below our threshold for @xmath31 .
@xcite reported an excess in w3 for 2mass j16185037 - 2424319 , but we rejected this measurement as a false detection . we discuss the bands [ 24 ] and w4 together since they have similar effective wavelengths ( 23.7 and 22 ) . as with w3 , we consider only data with uncertainties less than 0.25 mag and that are not contaminated by extended emission .
the resulting sample contains 348 and 260 stars with photometry in [ 24 ] and w4 , respectively .
data are available in at least one of the two bands for 428 members . among the _ spitzer _ and bands that we are considering , [ 24 ] and w4 are the only ones in which debris disks are likely to produce noticeable excess emission ( section [ sec : class ] ) .
since these excesses can be arbitrarily small , we have defined thresholds in @xmath48 $ ] and @xmath32 that identify the smallest excesses that appear to be significant .
the threshold adopted for @xmath32 is higher than that for @xmath48 $ ] since the photometric uncertainties are larger in w4 than in [ 24 ] , as illustrated by the larger width in the sequence of photospheric colors for @xmath32 in figure [ fig : pan ] . as a result ,
some stars that show small excesses in @xmath48 $ ] are below the threshold for @xmath32 .
the thresholds are defined by ( b0 , 0.11 ) , ( k6 , 0.57 ) , and ( m9 , 1.4 ) for @xmath48 $ ] and ( b0 , 0.34 ) and ( m8 , 1.15 ) for @xmath32 .
most of our identifications of excesses in [ 24 ] and w4 agree with those from previous studies @xcite , with the following exceptions .
our [ 24 ] photometry for hip 81455 is 0.4 mag brighter than that measured by @xcite with the same data . as a result
, we find an an excess in this band while @xcite did not .
this star is partially resolved from another fainter star at a separation of @xmath49 , which was removed through psf subtraction prior to our aperture photometry .
we speculate that the psf photometry of @xcite may have been affected by this second star .
@xcite found that hip 78977 has an excess in @xmath48 $ ] , but it is slightly below our thresholds for both @xmath48 $ ] and @xmath32 . neither [ 24 ] nor w4 show an excess relative to w1 , w2 , and w3 while excesses do appear in @xmath30 and @xmath31 .
these discrepant colors are likely caused by variability of the star between the near- and mid - ir observations , and we conclude that it does not have excesses in [ 24 ] and w4 .
the following stars are above our [ 24 ] thresholds but slightly below those adopted by @xcite : hip 78702 , hip 78099 , hip 76633 , hip 79250 , and rx j1602.8 - 2401b .
@xcite used w4 to assess the presence of excess emission for hip 78265 , hip 80019 , hip 80311 , and hip 82319 , but we have rejected these data because they are false detections or too uncertain .
we have identified 22 new candidates in upper sco for weak excess emission in [ 24 ] or w4 , which may indicate the presence of debris disks or evolved transitional disks ( see section [ sec : class ] ) .
six of these stars were found to lack excesses in previous studies , as described above .
the remaining new candidates for weak excesses consist of hd 142506 , hip 80196 , 2mass j16020287 - 2236139 , 2mass j16050231 - 1941554 , 2mass j16052459 - 1954419 , 2mass j16061330 - 2212537 , 2mass j16080555 - 2218070 , 2mass j16094098 - 2217594 , 2mass j16103956 - 1916524 , 2mass j16104202 - 2101319 , 2mass j16114612 - 1907429 , 2mass j16124893 - 1800525 , 2mass j16230783 - 2300596 , rxj1614.6 - 1857 , sco pms 8b , and sco pms 42b .
the detection of a w4 excess for hd 142506 is tentative since the photometry may be affected by blending with another star .
similarly , 2mass j16235695 - 3312033 is above the threshold for an excess in @xmath32 , but it is blended with another star . because it lacks an excess in the higher - resolution images at [ 24 ] , we classify this star as lacking an excess in w4 . unlike the other _ spitzer _
bands that we consider , the [ 24 ] images did not detect all of the members of upper sco that they encompass . among the members observed in [ 24 ] ,
94 stars were not detected or have errors large enough to be excluded from our analysis .
three of these stars have early spectral types and lack data because they are projected against bright nebulosity or are dominated by extended emission .
the remaining 91 objects have m types , most of which are later than m3 .
a few of these stars have poor constraints on [ 24 ] because of bright nebulosity or the psf of another star , but most have limits of @xmath25\sim10 $ ] , indicating that these stars have colors of @xmath50 .
the detection limit from w4 is bright enough ( @xmath51 ) that the resulting limits on @xmath32 do not place useful constraints on the presence of disks for most members of upper sco that are not detected in this band .
multiple stages of evolution are usually present among the circumstellar disks in young stellar populations ( @xmath52 myr ) .
the evolution of a disk is largely characterized by a change in its structure , which is reflected in its ir spectral energy distribution ( sed ) . as a result
, one can use multi - band ir data like those we have compiled from _ spitzer _ and for upper sco to estimate the evolutionary stages of disks . in this section ,
we describe our adopted definitions for these stages and apply them to upper sco .
a variety of names and definitions have been devised for the evolutionary stages of disks .
we use the ones adopted by @xcite , which are as follows : a _ full disk _ is optically thick at ir wavelengths and has not experienced significant clearing as indicated by its sed .
a _ pre - transitional disk _ and a _ transitional disk _ have gaps and holes in their dust distributions , respectively , that are large enough to noticeably affect their seds ( @xmath53 au , @xmath54 au , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
_ evolved disk _ is becoming optically thin and does not have a large gap or hole @xcite .
evolved disks also have been named anemic , thin , weak , and homologously depleted @xcite . since we are examining a stellar population that includes relatively advanced stages of disk evolution , we also use the terms _ evolved transitional disk _ , which has optically thin primordial dust and a large inner hole @xcite , and _ debris disk _ , which consists of second - generation dust that is generated by collisions among planetesimals @xcite .
all of these stages except debris disks are considered _ primordial disks_. this definition of a primordial disk differs from the one that we adopted in @xcite , where was equivalent to a full disk as defined here . to define our observational criteria for the different evolutionary stages of disks , we begin by describing the qualitative features of their seds .
a full disk produces moderate - to - strong emission throughout the ir regime .
because it has a large inner hole , a transitional disk is faint at shorter ir wavelengths and is much brighter at longer wavelengths .
the emission from an evolved disk does not change abruptly with wavelength since it lacks a large hole or gap , and it is very weak since the disk is becoming optically thin .
an evolved transitional disk and a debris disk have similar seds , showing only weak emission at long ir wavelengths . to develop quantitative criteria for classifying disks , one approach has been to identify distinct populations in the ir seds of young stars and associate them with different evolutionary stages of disks .
populations of this kind were first found in taurus , where most stars have mid - ir colors that are either consistent with stellar photospheres , or are much redder .
the few sources within the gap between these two groups were described as in transition " @xcite .
similar gaps are present in the mid - ir colors of other star - forming regions as well ( * ? ? ?
* references therein ) . in some studies ,
the red edge of the gap in colors like @xmath46 $ ] has been adopted as the boundary between full disks and pre - transitional disks on the redward side , and evolved and transitional disks on the blueward side @xcite .
this threshold seems reasonable given that it roughly coincides with the colors produced by models of full disks with high degrees of dust settling @xcite .
distinctive populations of colors and supporting modeling also have been used to guide the identification of debris disks and evolved transitional disks @xcite . a different approach to classifying disks has been employed by @xcite .
they defined transitional disks to include disks with inner holes or gaps and disks that are optically thin , which correspond to transitional , pre - transitional , and evolved disks in our terminology . to identify disks of this kind , @xcite compared observed seds to the seds predicted by disk models .
when millimeter measurements were available , they also required that a disk have an estimated mass less than 0.001 @xmath55 in order for it to be considered optically thin . after applying these procedures , @xcite concluded that some of the disks previously classified as settled , optically thick disks ( i.e. , redward of the gap in ir colors described above ) are instead optically thin , and hence qualify as transitional disks as defined in their study . because their definition of transitional disks was broader and
their classification criteria encompassed redder disks , @xcite found a higher abundance of transitional disks in regions like taurus than some studies ( e.g. , * ? ? ? * ) .
of course , a comparison of the abundances of transitional disks , and the resulting evolutionary timescales ( section [ sec : histo ] ) , is not meaningful when the adopted definitions of transitional disks differ .
the classification scheme from @xcite has disadvantages .
transitional disks are not defined in the traditional manner ( disks with large holes ) , which can lead to confusion when classifications from different studies are compared . the observational criteria from @xcite rely on theoretical seds , which are subject to imperfections in the physical prescriptions of the models and degeneracies among the disk parameters .
the predicted seds also can vary from one set of adopted models to another .
for instance , as indicated above , some objects can be reproduced with either an optically thin disk or a settled optically thick disk , depending on the choice of models @xcite .
the use of disk mass as a criterion for identifying optically thin disks is also problematic .
the requisite millimeter data are unavailable for many members of young clusters and are incomplete at low disk masses , even in taurus @xcite .
estimates of disk masses and optical depths are sensitive to both the disk model and the opacities that are adopted @xcite .
in addition , @xcite selected a threshold of 0.001 @xmath55 as a lower mass limit for optically thick disks because nearly all optically thick disks in taurus with millimeter data have masses above this value
. however , 0.001 @xmath55 is only slightly above the detection limits for the millimeter observations in taurus ( @xmath56 @xmath4 , * ? ? ?
* ) , and hence does not represent a measurement of the minimum mass of optically thick disks .
deeper millimeter data are needed to reliably determine the distribution of disks masses down to low optical depths . in a critique of the disk classifications based on colors from @xcite
, @xcite noted that ux tau a has a disk with a large gap @xcite but did not satisfy the color criteria for a transitional disk from @xcite and instead was classified as a primordial disk ( or a full disk with our current terminology ) .
however , this classification was consistent with the definitions adopted by @xcite in which a disk with a gap was considered pre - transitional and a subset of primordial disks , whereas only disks with large holes were named transitional .
any set of classification criteria will exclude disks with holes or gaps below a certain size , so the absence of disks with gaps like ux tau a from our definition of transitional disk is not particularly worrisome .
we simply use the term transitional disk " to refer to a different range of structural evolution than @xcite . in summary
, we have selected a classification scheme for disks in which the evolutionary stages are defined by distinct structural characteristics @xcite .
the criteria for identifying members of these classes are based on patterns in the data for disks ( e.g. , gap in colors ) and can be applied to data that are available for large numbers of young stars and brown dwarfs , namely multi - band mid - ir photometry from _
spitzer _ and _
wise_. in @xcite , we used extinction - corrected mid - ir colors to estimate the evolutionary stages of disks in taurus and other regions . for upper sco , we modify this approach by considering instead the color excesses relative to photospheric colors ( e.g. , @xmath57 $ ] ) .
we have computed the excess in a given color for each member of upper sco by subtracting the color of a stellar photosphere in upper sco for the spectral type in question , as determined by a fit to the sequence of diskless stars in figure [ fig : pan ] . through this process
, the colors for all members are corrected for roughly the same level of extinction , which is the average extinction of the diskless members .
the resulting excesses for the colors from figure [ fig : pan ] are presented in figure [ fig : ex ] . the be stars
are omitted since they do not require classification .
we plot the excesses based on [ 4.5 ] and w2 together because of the similarity of these bands .
we do the same for [ 24 ] and w4 . for each object
, we show the data in [ 4.5 ] and [ 24 ] when available , and otherwise use w2 and w4 .
@xcite used @xmath58 $ ] , @xmath46 $ ] , and @xmath48 $ ] for their disk classifications . since relatively few members of upper sco have been observed in [ 5.8 ] , we begin our classifications with the other two colors ( and @xmath32 ) . in figure
[ fig : ex ] , we show our adopted thresholds in @xmath59)$ ] and @xmath57/w4)$ ] for distinguishing between full disks and later evolutionary stages .
these thresholds approximate those applied to @xmath46 $ ] and @xmath48 $ ] by @xcite .
we have used the objects classified in this way to reveal the boundary between full disks and later stages in @xmath60 vs. @xmath57/w4)$ ] , and we have used this boundary to classify stars that lack data in [ 8.0 ] . to classify the disks bluer than full disks ,
we apply the following criteria , which are similar to those from @xcite and @xcite : transitional disks have @xmath57/w4)>3.2 $ ] ; evolved disks have @xmath57/w4)<3.2 $ ] , @xmath59)>0.3 $ ] , and @xmath61 ; debris disks and evolved transitional disks have @xmath57/w4)<3.2 $ ] , @xmath59)<0.3 $ ] , and @xmath62 .
measurements of gas content are necessary for distinguishing between the latter two classes of disks since they exhibit similar mid - ir seds @xcite , but such data are unavailable for most disks in upper sco . for the early - type stars ,
the high abundance of these debris / evolved transitional candidates relative to full disks indicates that most are probably debris disks since evolved transitional disks are expected to be short - lived ( section [ sec : histo ] ) .
two exceptions to the above criteria are hip 79288 and hip 79977 .
both stars are redder than our thresholds for debris / evolved transitional disks , but we classify them as such based on previous studies @xcite .
we have attempted to classify the disks that lack photometry in [ 24 ] or w4 , and hence do not appear in figure [ fig : ex ] . for the disks of this kind that are detected in w3 , the excesses in this band are small enough to indicate that they are evolved or transitional . some of those disks are probably evolved rather than transitional based on the available constraints on @xmath57/w4)$ ] .
two of the disks without w3 , w4 , or [ 24 ] have data in [ 8.0 ] , both of which are full disks according to their values of @xmath59)$ ] . the remaining disks that lack data in [ 24 ] or w4 are detected only at @xmath63 , most of which are later than m6 .
we classify them as full disks for the purposes of this work , although some could be evolved disks . in table
[ tab : all ] , we present our disk classifications for all members of upper sco that show mid - ir excess emission . through our classifications , we arrive at 135 full disks , 28 evolved disks , 6 transitional disks , three disks that are evolved or transitional , and 63 debris / evolved transitional disks .
previous mid - ir studies have already identified many of these full disks @xcite , evolved and transitional disks @xcite , and debris / evolved transitional disks @xcite .
our new candidates for transitional disks consist of 2mass j16082733 - 2217292 , 2mass j16101888 - 2502325 , and 2mass j16151239 - 2420091 .
it is likely that a number of full disks have inner holes that are too small to be classified as transitional using our criteria , such as 2mass j16081566 - 2222199 , which has strong excesses in w3 and w4 but no excess in w2 ( see upper panel of figure [ fig : ex ] ) .
most of the candidates for evolved disks are newly identified as such in this work . in section [ sec : exc24 ] , we listed our new candidates for debris / evolved transitional disks and a few stars in this category that we classify differently than other studies .
we note that hip 81474 has been frequently treated as a herbig ae / be star , but it lacks a clear detection of line emission @xcite .
therefore , it is classified as debris / evolved transitional in this work .
the distribution of disk excesses as a function of parameters like wavelength , spectral type , cluster age , excess size , and disk class can provide insight into the evolution of disks . to characterize the distribution of excesses in upper sco , we begin by considering the fraction of members that exhibit excesses . in table [ tab : fex ] and figure [ fig : fex ] , we present the excess fraction as a function of spectral type for [ 4.5]/w2 , [ 8.0 ] , w3 , and [ 24]/w4 . to examine the evolution of primordial and debris disks separately ,
one excess fraction is computed for full , transitional , and evolved disks , and another is computed for debris and evolved transitional disks .
the latter disks do contain primordial dust , but they can not be distinguished from debris disks with available data , as discussed in the previous section . the images in w3 , w4 , and [ 24 ] were not sufficiently sensitive to detect all known members of upper sco .
the members that were too faint to be detected tend to have later spectral types or smaller excesses . as a result
, we can not compute meaningful excess fractions in these bands at the latest types .
therefore , the excess fractions for w3 and [ 24]/w4 in figure [ fig : fex ] are shown only for spectral types in which nearly all of the known members are detected .
for the joint [ 24]/w4 excess fraction , we adopt the [ 24 ] data when available , which have a high level of completeness down to a spectral type of m4 . among [ 24 ]
non - detections earlier than m4 , six stars were not subject to extended emission or blending with other stars , and hence have useful limits on their photometry .
the resulting constraints on @xmath48 $ ] indicate that these six stars have little , if any , excess emission , so we include them as non - excess stars in the [ 24]/w4 fraction . among stars that were not observed at [ 24 ] , the w4 images detected all but three stars earlier than m0 . therefore , we use the w4 data of members not observed at [ 24 ] and earlier than m0 in the [ 24]/w4 fraction .
the three stars earlier than m0 with non - detections in w4 have sufficient constraints on this band to indicate that excesses are probably not present , so we count them as non - excess stars .
it is unnecessary to impose such restrictions on the spectral types included in the excess fractions at [ 4.5 ] , w2 , and [ 8.0 ] since all known members were detected in these bands as long as they were observed ( by irac ) and not blended with another star .
the be stars are excluded from these excess fractions .
we first discuss the excess fractions for the full , transitional , and evolved disks ( i.e. , all primordial disks except for evolved transitional disks ) . for these disks ,
the excess fractions are very low at a roughly uniform level as a function of spectral type for b through g types , and then increase with later types from k through m. this trend was discovered by @xcite , supported by additional data from @xcite for late - m members , and now is reinforced further with better number statistics .
all of the bands in figure [ fig : fex ] exhibit a similar dependence on spectral type , except that the excess fractions are slightly higher at longer wavelengths due to the presence of a small number of transitional disks . as discussed by @xcite ,
these data indicate that disk lifetimes are longer for stars at lower masses , providing more time for planet formation .
we note that the excess fraction in [ 4.5]/w2 for the last bin in figure [ fig : fex ] at m8l2 ( @xmath64 @xmath4 ) is uncertain because of the small number of objects and the rapidly changing photospheric colors , which makes it difficult to reliably identify excesses .
in addition , since excesses in [ 4.5 ] are smaller at later types @xcite , the excess fraction in this band may significantly underestimate the disk fraction .
longer wavelengths are needed for more robust detections of disk excesses for brown dwarfs , but the members at m8l2 were too faint for w3 and w4 photometry and were not observed at [ 5.8 ] and [ 8.0 ] .
the disk fractions based on the data in figure [ fig : fex ] are compared to similar measurements in younger clusters by @xcite .
we now consider the excess fractions for the debris and evolved transitional disks .
as mentioned earlier , we can not distinguish between these two types of disks with the mid - ir data , but most of the candidates at earlier spectral types are probably debris disks .
it is unclear which class of disk dominates at later types . in figure
[ fig : fex ] , the excess fraction for this sample increases rapidly with longer wavelengths .
excesses in [ 4.5]/w2 are present for only a few stars , which may be some of the youngest debris disks in the association @xcite . the [ 24]/w4 fraction is highest for a stars ( @xmath65 ) , drops sharply for b stars ( @xmath66 ) , and decreases more gradually with later types , reaching 0.10.2 for g , k , and m stars .
these trends are consistent with those previously found in smaller samples of [ 24 ] data in upper sco by @xcite and @xcite .
those studies investigated various aspects of their [ 24 ] data , and their results apply to our new measurements as well .
for instance , @xcite found that the rapid decrease in [ 24 ] fraction from a to b stars is expected because of increased ejection of dust grains by radiation pressure ( see also * ? ? ? * ) .
the [ 24]/w4 excess fractions at later spectral types are more difficult to interpret since they contain an unknown contribution from evolved transitional disks . in addition
, many debris disks around late - type stars may be too cool to produce noticeable excesses in these bands .
@xcite and @xcite also examined the evolution of debris disks by comparing the [ 24 ] excess fractions among upper sco and other associations and clusters across a range of ages ( see also * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our data have improved the statistical constraints on excess fractions and sizes in upper sco , particularly at m types , but an analysis like that from @xcite and @xcite for m stars is not possible because sufficiently complete samples of [ 24 ] data are available for only a few regions .
the mid - ir colors of the members of a given star - forming region typically exhibit a gap between diskless stars and the bulk of the disk - bearing stars ( see section [ sec : criteria ] ) .
this trough in the distribution of excess sizes has been previously interpreted as evidence that once a disk has begun to clear to a certain degree , the remainder of the dust in the inner disk clears rapidly @xcite . in this section ,
we examine the constraints on the clearing timescale that are provided by the distribution of excesses in upper sco . because the contrast between a stellar photosphere and circumstellar disk increases at longer ir wavelengths , [ 24 ] and w4 offer the largest dynamic range in the sizes of excesses among the photometric bands that we have compiled for upper sco .
given that the [ 24 ] observations are more sensitive than those at w4 , we choose the former for constructing the distribution of excesses . to characterize how this distribution changes with cluster age , we compare the data for upper sco to measurements of [ 24 ] in taurus @xcite .
upper sco and taurus are the best available regions for such a comparison since they roughly bracket the ages of primordial disks ( @xmath13 and 10 myr ) , contain two of the richest and nearest samples of young stars , and have been thoroughly observed by _
spitzer_. in particular , few other regions have [ 24 ] measurements that reach the stellar photospheres of low - mass stars , which is essential for fully sampling excesses across all sizes .
since very few primordial disks are present among the early - type stars in upper sco , we exclude stars earlier than k0 from the distributions of @xmath48 $ ] excesses for upper sco and taurus .
we also omit stars later than m5 since most of the diskless stars at these types are below the detection limits of the [ 24 ] images .
roughly a dozen stars in each region at k5m5 are not detected in [ 24 ] and could have small excesses based on the limits on @xmath48 $ ] . as a result ,
the distributions of excesses that we construct are biased slightly against small excesses .
the distributions of @xmath57)$ ] for k0m5 members of taurus and upper sco are presented in figure [ fig : histo ] , excluding protostellar sources .
the two distributions exhibit similar troughs at @xmath57)\sim0.5$]3 .
most observations of other young clusters at [ 24 ] are less complete at low excesses or consider smaller numbers of stars , but they are generally consistent with gaps of this kind @xcite . meanwhile , among stars redward of the trough ( @xmath57)>3 $ ] ) , the average color is bluer in upper sco than in taurus .
these characteristics can be explained in the following manner .
the youngest systems have highly flared disks , and hence the largest ir excesses .
as they evolve ( e.g. , growth of grains , accretion onto the star ) , disks become more settled and produce smaller excesses .
after disks have become highly settled and have reached a certain level of initial clearing ( corresponding to @xmath57)\sim3 $ ] ) , they quickly clear their remaining inner disks , resulting in a scarcity of disks with small excesses . because star formation is ongoing in taurus but has ceased in upper sco , the distribution of excesses in the former is more heavily weighted to younger and redder disks . in other words ,
the youngest disks in upper sco have already evolved substantially , explaining its paucity of very red disks relative to taurus .
because the evolutionary stages of disks are not sampled in the same way by the differing star formation histories of taurus and upper sco , a comparison of the n(evolved+transitional)/n(full ) does not indicate the relative timescales of clearing between the two populations @xcite .
instead , one must compare the ratio of the number of evolved and transitional disks to the number of sources slightly redward of the trough , i.e. , the shape of the distribution of excesses at @xmath57)\lesssim3.5 $ ] .
this property is not tightly constrained in either region because of the small numbers of stars involved , but taurus and upper sco do not show any obvious differences . thus , there is no evidence to indicate that the clearing timescale changes from the age of taurus to the age of upper sco ( @xmath1310 myr ) . the data in figure [ fig : histo ]
do reveal a higher abundance of candidates for evolved transitional disks or debris disks in upper sco .
if the lifetime of the evolved transitional stage is similar to that of the transitional and evolved stages , then most of these candidates in upper sco are probably debris disks .
it is expected that upper sco would contain more debris disks than taurus given their relative ages . as discussed in section [ sec : criteria ] , the definition of transitional disks adopted by @xcite encompasses some of the objects redward of the gap in the mid - ir colors in regions like taurus and upper sco . as a result , they have derived higher abundances of transitional disks in young clusters , and hence a longer timescale for inner disk clearing , than found here and in other studies that use a more restrictive definition for transitional disks @xcite . of course , it is not meaningful to compare abundances of transitional disks and the resulting timescales among studies that do not define this evolutionary stage in the same manner .
we contend that the clearing timescales from @xcite are less useful and practical since they are based on a definition of transitional disks that is subject to uncertainties and degeneracies in model seds and that does not rely on distinct observational signatures ( i.e. , gap in colors ) .
@xcite suggested that the gap in mid - ir colors in taurus is not evidence of a short lifetime for the evolutionary stage of disks within this gap , contrary to the original interpretation @xcite .
instead , they argued that the gap is present because few members of taurus are old enough to have reached that stage , and that the diskless stars in taurus have lost their disks due to binary companions rather than advanced age . in @xcite , we noted that a rather large fraction of the taurus population is diskless ( 40% among non - protostars ) and that these stars are older than disk - bearing members , on average , based on their wider spatial distribution , demonstrating that a significant number of stars in taurus have aged sufficiently to fully clear their disks , and hence pass through the evolved and transitional stages .
@xcite reiterated the same points from their previous study , but also conceded that many of the diskless stars in taurus may indeed be older than their counterparts with disks .
here , we once again contend that if a significant number of stars have aged sufficiently to have fully cleared their disks , then the evolved and transitional stages should be fully sampled as well , and therefore their scarcity demonstrates that they are short - lived relative to full disks .
this conclusion is supported by the fact that older regions like upper sco exhibit the same gap in their colors ( figure [ fig : histo ] ; * ? ? ?
* references therein ) .
we have performed a census of the circumstellar disk population of the upper sco association ( @xmath0 myr ) using mid - ir images obtained with the _
spitzer space telescope _ and _ wise_. the results of this study are summarized as follows : 1 .
we have analyzed all images of known members of upper sco at 3.6 , 4.5 , 5.8 , 8.0 , and 24 that have been collected by _
spitzer _ to date . among the 863 known members , 484 have been observed in at least one of these bands .
we have presented a catalog of photometry and non - detections for these objects .
we have compiled all photometry at 3.4 , 4.6 , 12 , and 22 ( w1w4 ) for members of upper sco from the mission , thoroughly vetting these data for false detections , blends with neighboring stars , and extended emission .
all resolved , unblended members were detected by in at least 3.4 and 4.6 .
we have used colors computed between @xmath41 and six of the _ spitzer _ and bands to identify the members that exhibit excess emission from circumstellar disks .
we have estimated the evolutionary stages of the detected disks using the sizes of the excesses among these bands .
these stages are defined based on distinct structural characteristics that are apparent in the seds ( * ? ? ?
* references therein ) , and they consist of the following : optically thick disks without large holes or gaps ( _ full _ ) , optically thick disks with large holes ( _ transitional _ ) , optically thin disks without large holes or gaps ( _ evolved _ ) , optically thin disks with large holes ( _ evolved transitional _ ) , and disks of second - generation dust ( _ debris _ ) . through these classifications ,
we have found @xmath1 new candidates for transitional , evolved , and debris disks .
4 . based on our census of disks ,
the fraction of upper sco members with inner primordial disks is @xmath2% for b g stars ( @xmath3 @xmath4 ) and increases with later types , reaching @xmath5% at m5l0 ( @xmath670.2 @xmath4 ) .
this trend was first observed by @xcite , and is now reinforced with a sample that is larger and reaches lower stellar masses . the combination of these data and the latest age estimate for upper sco ( @xmath0 myr , * ? ? ?
* ) indicate that a significant fraction of primordial disks around low - mass stars survive for at least @xmath8 myr . 5 .
k0m5 members of upper sco show a dearth of disks with small excesses ( transitional and evolved disks ) .
when the differences in star formation histories of upper sco and taurus ( @xmath9 myr ) are taken into account , the distributions of excesses for both regions indicate a timescale for inner disk clearing that is much shorter than the typical lifetime of primordial disks .
we thank catherine espaillat , nuria calvet , and eric feigelson for comments on the manuscript .
k. l. was supported by grants ast-0544588 and nnx12ai58 g from the national science foundation and the nasa astrophysics data analysis program , respectively , and e. m. was supported by grant ast-1008908 from the national science foundation .
this work makes use of data from the _ spitzer space telescope _
, 2mass , _ wise _ , and the infrared processing and analysis center ( ipac ) infrared science archive ( irsa ) .
_ spitzer _ and irsa are operated by the jet propulsion laboratory ( jpl ) and the california institute of technology ( caltech ) under contract with nasa .
is a joint project of the university of california , los angeles , and jpl / caltech , funded by nasa .
2mass is a joint project of the university of massachusetts and the ipac / caltech , funded by nasa and the nsf .
the center for exoplanets and habitable worlds is supported by the pennsylvania state university , the eberly college of science , and the pennsylvania space grant consortium .
ll name & source name + othernames & other source names + sptype & spectral type + r_sptype & spectral type reference + adopt & adopted spectral type + 3.6mag & _ spitzer _ [ 3.6 ] band magnitude + e_3.6mag & error in [ 3.6 ] band magnitude + f_3.6mag & flag on [ 3.6 ] band magnitude + 4.5mag & _ spitzer _ [ 4.5 ] band magnitude + e_4.5mag & error in [ 4.5 ] band magnitude + f_4.5mag & flag on [ 4.5 ] band magnitude + 5.8mag & _ spitzer _ [ 5.8 ] band magnitude + e_5.8mag & error in [ 5.8 ] band magnitude + f_5.8mag & flag on [ 5.8 ] band magnitude + 8.0mag & _ spitzer _ [ 8.0 ] band magnitude + e_8.0mag & error in [ 8.0 ] band magnitude + f_8.0mag & flag on [ 8.0 ] band magnitude + 24mag & _ spitzer _ [ 24 ] band magnitude + e_24mag & error in [ 24 ] band magnitude + f_24mag & flag gon [ 24 ] band magnitude + w1mag & _ wise _ w1 band magnitude + e_w1mag & error in w1 band magnitude + f_w1mag & flag on w1 band magnitude + w2mag & _ wise _ w2 band magnitude + e_w2mag & error in w2 band magnitude + f_w2mag & flag on w2 band magnitude + w3mag & _ wise _ w3 band magnitude + e_w3mag & error in w3 band magnitude + f_w3mag & flag on w3 band magnitude + w4mag & _ wise _ w4 band magnitude + e_w4mag & error in w4 band magnitude + f_w4mag & flag on w4 band magnitude + exc4.5 & excess present in [ 4.5 ] ?
+ exc8.0 & excess present in [ 8.0 ] ? + exc24 & excess present in [ 24 ] ?
+ excw2 & excess present in w2 ?
+ excw3 & excess present in w3 ?
+ excw4 & excess present in w4 ? + disktype & disk type + llllll b0b8 & 2.818 & 0/25 = @xmath66 & 0/22 = @xmath68 & 0/25 = @xmath66 & 0/26 = @xmath66 + b8a6 & 1.82.8 & 0/45 = @xmath69 & 0/40 = @xmath69 & 1/46 = @xmath70 & 1/45 = @xmath70 + a6f4 & 1.51.8 & 0/20 = @xmath71 & 0/12 = @xmath72 & 1/21 = @xmath73 & 1/21 = @xmath73 + f4g2 & 1.41.5 & 0/18 = @xmath71 & 0/9 = @xmath74 & 0/18 = @xmath71 & 0/17 = @xmath75 + g2k0 & 1.31.4 & 1/25 = @xmath76 & 1/21 = @xmath73 & 1/25 = @xmath76 & 1/25 = @xmath76 + k0m0 & 0.71.3 & 6/67 = @xmath77 & 4/46 = @xmath78 & 8/68 = @xmath79 & 7/60 = @xmath79 + m0m4 & 0.20.7 & 35/231 = @xmath80 & 18/96 = @xmath81 & 41/210 = @xmath82 & 23/108
= @xmath83 + m4m8 & 0.0350.2 & 97/387 = @xmath84 & 22/84 = @xmath85 & & + m8l2 & 0.010.035 & 4/23 = @xmath86 & & & + b0b8 & 2.818 & 0/25 = @xmath66 & 0/22 = @xmath68 & 0/25 = @xmath66 & 0/26 = @xmath66 + b8a6 & 1.82.8 & 0/45 = @xmath69 & 0/40 = @xmath69 & 5/46 = @xmath87 & 20/45
= @xmath88 + a6f4 & 1.51.8 & 1/20 = @xmath89 & 0/12 = @xmath72 & 3/21 = @xmath90 & 7/21 = @xmath91 + f4g2 & 1.41.5 & 0/18 = @xmath71 & 0/9 = @xmath74 & 0/18 = @xmath71 & 3/17
= @xmath92 + g2k0 & 1.31.4 & 0/25 = @xmath66 & 0/21 = @xmath68 & 1/25 = @xmath76 & 4/25 = @xmath93 + k0m0 & 0.71.3 & 0/67 = @xmath94 & 0/46 = @xmath69 & 0/68 = @xmath94 & 8/60 = @xmath95 + m0m4 & 0.20.7 & 0/231 = @xmath96 & 0/96 = @xmath97 & 0/210 = @xmath96 & 13/108 = @xmath98 + m4m8 & 0.0350.2 & 0/387 = @xmath96 & 0/84 = @xmath97 & & + m8l2 & 0.010.035 & 0/23 = @xmath66 & & & + | we present photometry at 324 for all known members of the upper scorpius association ( @xmath0 myr ) based on all images of these objects obtained with the _
spitzer space telescope _ and the _ wide - field infrared survey explorer_. we have used these data to identify the members that exhibit excess emission from circumstellar disks and estimate the evolutionary stages of these disks . through this analysis , we have found @xmath1 new candidates for transitional , evolved , and debris disks .
the fraction of members harboring inner primordial disks is @xmath2% for b g stars ( @xmath3 @xmath4 ) and increases with later types to a value of @xmath5% at @xmath6m5 ( @xmath7 @xmath4 ) , in agreement with the results of previous disk surveys of smaller samples of upper sco members .
these data indicate that the lifetimes of disks are longer at lower stellar masses , and that a significant fraction of disks of low - mass stars survive for at least @xmath8 myr .
finally , we demonstrate that the distribution of excess sizes in upper sco and the much younger taurus star - forming region ( @xmath9 myr ) are consistent with the same , brief timescale for clearing of inner disks . |
boltzmann pondered on how to understand thermodynamics with the newton s equations ; his answer to this question along with gibbs theory have become the foundation of classical statistical mechanics @xcite . after quantum mechanics been fully formulated , many giants in physics discussed a similar issue : how to understand thermodynamics with the schrdinger equation @xcite . in a 1929 paper ,
von neumann provided an answer to this question by proving two inequalities in full rigor and without disorder assumptions " @xcite .
these two inequalities , which he called the quantum ergordic theorem and the quantum h - theorem , respectively , laid down a foundation for quantum statistical mechanics .
however , this work has been largely forgotten and apparently have never been mentioned in any modern textbook on quantum statistical mechanics @xcite .
there are discussions on why this work had almost been forgotten @xcite . in our opinion ,
one of the likely reasons is that von neumann introduced a rather unfamiliar concept , _ macroscopic operators _ , to prove his theorems .
it appears very hard to compute these commuting macroscopic operators , and related variables such as an entropy defined for a pure quantum state @xcite .
recently there have been renewed interests on the foundation of quantum statistical mechanics @xcite perhaps due to the remarkable progress in experimental realization of coherent quantum systems @xcite .
an important result achieved is an inequality proved by reimann @xcite and later modified by short _
et al.@xcite .
this inequality can be regarded as a different version of von neumann s quantum ergodic theorem .
the advantage of this new inequality is that every variable involved is well known and can be computed . for this reason , when we discuss quantum ergodic theorem , we refer to the inequality proved by reimann unless stated otherwise . _ according to the quantum ergodic theorem , an isolated quantum system starting with a far - from - equilibrium state will relax dynamically to an equilibrium state and stay there with very small fluctuations for almost all the time . to be more specific , for an isolated quantum system described by the wave function @xmath0
, it will relax to the following equilibrium state _ = _ k |c_k|^2 , where @xmath1 is the energy eigenstate of the system and @xmath2 s are the expansion coefficients of @xmath0 in term of these eigenstates .
the density matrix @xmath3 is regarded as the micro - canonical ensemble by von neumann @xcite .
it is different from the usual micro - canonical ensemble found in textbooks @xcite , where the coefficients @xmath2 s take an identical value within a narrow energy shell .
as @xmath4 s do not change with time , the micro - canonical density matrix @xmath3 is completely determined by the initial condition . by utilizing this fact and the supposition principle
, we were able to predict a new quantum state which is at equilibrium with multiple temperatures , challenging the conventional wisdom that an equilibrium state has only one temperature @xcite .
the quantum ergodic theorem holds only for quantum systems with no degenerate energy gaps .
mathematically , this condition is expressed as @xcite @xmath5 however , it is not clear at all how this condition of non - degenerate energy gap is related to the familiar classification of quantum systems by their integrability . for a general integrable system , this condition is not satisfied as quantum integrable systems have the poisson distribution of energy level spacing @xcite , which imply the existence of many degenerate eigen - energies . however , there are plenty of examples of integrable systems , which have no energy degeneracy at all .
in fact , there are already reports of dynamical relaxation in integrable systems @xcite .
the case for quantum chaotic systems is more complicated . as is well known , quantum chaotic systems have the wigner distribution of energy level spacing @xcite .
the wigner distribution has two prominent features : zero probability at zero energy level spacing and a peak at a finite energy level spacing .
the former feature means that there is little degeneracy , which is favorite for the condition ( [ condition ] ) being satisfied .
the latter implies that there are large number of energy level spacings around the peak value , which is clearly unfavorite for the condition ( [ condition ] ) being satisfied . as a result , it is not clear at all in the sense of mathematical rigor whether the quantum ergodic theorem and h - theorem hold for quantum chaotic systems or not .
we led by intuition tend to believe that the quantum ergodic theorem and h - theorem should hold for quantum chaotic systems .
von neumann believed that his theorems should hold when the condition ( [ condition ] ) is violated by
infrequent exceptions " @xcite .
this belief is now confirmed by short and farrelly @xcite . despite this progress , it is still not clear how these mathematical conditions are related to the integrability of a system . the main purpose of this paper is to demonstrate that the quantum ergordic theorem applies to quantum chaotic systems numerically with examples . in this paper
we study two quantum chaotic systems , the ripple billiard @xcite and the henon - heiles system @xcite .
our numerical simulation shows that both systems will indeed dynamically relax to an equilibrium state where the overall features of the wave function no longer change . for the ripple billiard system , where the successive energy - eigenstates can be computed
, the quantum ergodic inequality can be verified directly .
in addition , we define entropies for pure quantum states in the spirit of von neumann as it is not clear how to compute the entropy defined for a pure quantum state by him .
we find that these entropies will approach maximized values , offering another indication that the system is indeed equilibrating dynamically .
we have also analyzed the properties of the equilibrium state reached after the dynamical relaxation in quantum chaotic systems .
we find an interesting correspondence between the quantum equilibrium state and the classical micro - canonical ensemble .
we discuss the underlying mechanism with the correspondence between the quantum and the classical liouville equations . at the end , we consider the statistical properties of fluctuations in quantum chaotic systems and find a distinction between the distributions of quantum fluctuations and the classical fluctuations .
this paper is organized as follows . in section
we introduce the two quantum chaotic systems , ripple billiard and henon - heiles system . in section
we study the dynamical equilibration of quantum chaotic systems . in section
we numerically verify the quantum ergodic theorem in the ripple billiard system . in section
we discuss the quantum - classical correspondence for the equilibrium states in detail . in section
we discuss the fluctuation properties of the quantum systems and their corresponding classical systems . finally in section
we discuss the implications for many - body cases and summarize our results .
we focus on single particle quantum chaotic systems .
there are two main reasons for this choice .
first , single particle quantum systems are much less challenging numerically and easier to analyze .
second , according to random matrix theory , the statistics of energy level spacings of quantum chaotic systems only depends on the type of matrix of the system when its hamiltonian is expressed in an orthonormal basis @xcite .
this property has nothing to do with whether the system is single particle or many body .
according to the condition ( [ condition ] ) , the property of the eigen - energy spectrum is the most important factor determining whether the quantum system will equilibrate or not .
there are properties that exist only in many - body systems , for example , the correlations @xcite .
so far , no one has shown that the correlation plays any essential role in the equilibration process .
we also emphasize that we here consider isolated quantum systems and do not consider the dynamics of quantum systems under external driving @xcite .
it is well known that a classical system will behave very differently under different drivings .
this feature seems to be shared by quantum systems @xcite .
a single particle in a two - dimensional chaotic potential is described by the hamiltonian h = p^2/2m+v(x , y ) .
we choose the henon - heiles potential @xcite and the ripple billiard @xcite as two examples for our study .
the henon - heiles potential is given by @xmath6 .
the energy contour of henon - heiles potential is shown in fig .
[ contourh ] ; it has four special points : one stable point o@xmath7 , three saddle points a@xmath8 , b@xmath9 , c@xmath10 , where @xmath11 .
the classical orbits in the henon - heiles potential are chaotic when the energy is above @xmath12 and approach fully chaotic when the energy is close to @xmath13 with @xmath14 . for later use , we set @xmath15 .
billiard systems are a two dimensional area surrounded by infinite potential walls at the edges .
for the ripple billiard @xcite , as shown in fig .
[ contourr ] , the left and right edges are described by functions @xmath16 $ ] and the up and down edges are two straight lines at @xmath17 and @xmath18 .
the two geometrical parameters @xmath19 , @xmath20 control the shape of the billiard . when @xmath21 the ripple billiard is a square with width @xmath22 .
as @xmath19 increases from zero , it changes from an integrable system to a mixed , then to a fully chaotic system .
it becomes mixed again when @xmath19 becomes very large .
we have chosen for our computation @xmath23 , which corresponds to a fully chaotic case .
the two systems are chosen because each of them has its own advantages .
for the ripple billiard system , its successive eigenstates from the ground state up to the 3000th excited state can be computed numerically with great precision @xcite . to the best of our knowledge , for
all other studied quantum chaotic systems , the high - energy eigenstates can only be computed selectively @xcite .
the famous henon - heiles system is chosen to bring our study beyond billiards where @xmath24 is a constant of motion , allowing us to gain more insights into general systems .
in addition , we note that the henon - heiles system is beyond what is considered by von neumann and others as it has no bound eigenstates mathematically .
however , its resonant states can be regarded as bound states in our numerically studies , where the dynamical evolution lasts for a finite time and hard - wall boundaries are imposed at distance @xcite .
we numerically study the wave packet dynamics with the schrdinger equation for these two systems . in our numerical simulation
we set @xmath25 , @xmath26 .
the initial states are highly localized moving gaussian wave packet for both systems , ( , t=0)=(-^2(-)^2)(i/ ) [ initial_condition ] with @xmath27 , @xmath28 , and @xmath29 for the henon - heiles system ; @xmath30 , @xmath31 , and @xmath32 for the ripple billiard system . in both systems , a classical particle with the above initial position @xmath33 and momentum @xmath34 has a fully chaotic orbit , which we confirmed by computing the poincare section . and
( b ) @xmath35 in the ripple billiard .
the red lines are @xmath36 and @xmath36 .
the initial condition of @xmath37 is ( 5,0 ) .
@xmath38.,scaledwidth=48.0% ] the subsequent dynamical evolutions of these two gaussian wave packets are computed . for the ripple billiard
, the dynamical evolution can be found in ref .
@xcite ; for the henon - heiles system , the evolution is illustrated in ref .
both evolutions are very similar to each other .
here is a brief description : the smooth gaussian wave packet starts to spread out and gets diffracted by the potentials ; the interference between diffracted waves begins to make the wave packet appear more and more irregular ; eventually the wave packet spreads out over the classically allowed region rather uniformly with small speckles .
this overall feature will no longer change , signaling that the system has dynamically equilibrated . to illustrate this dynamical equilibration process
, we compute how the expectation of momentum changes with time for the ripple billiard system .
the results for both @xmath39 and @xmath40 are shown in fig .
[ momentum ] , where the momenta are seen to relax to equilibrium values after a short period of large fluctuations .
equilibration should be accompanied by a maximizing entropy .
von neumann was able to define an entropy for a pure quantum state and proved that this entropy will stay very close to its ensemble entropy almost all the time ( the quantum h - theorem ) @xcite .
however , there appears no viable procedure which one can use to compute this version of von neumann entropy . as an alternative
, we define an entropy in the spirit of von neumann , s_r=- dxdy , where @xmath41 is the long time average of @xmath42 @xcite . as shown in fig .
[ entropy ] , the entropy will increase with time with small fluctuations and eventually saturate to a maximized value .
the increasing entropies in fig .
[ entropy ] can be regarded as a spiritual " illustration of von neumann s quantum h - theorem . in summary ,
we have observed numerically that dynamical equilibration indeed occurs in both the henon - heiles system and the ripple billiard : an initially localized gaussian wave packet with low entropy will dynamically evolve into a quantum state with a maximized entropy , where the wave packet spreads out and looks irregular with speckles .
this is clearly consistent with both the quantum ergodic theorem and quantum h - theorem .
it is reasonable to expect that this kind of dynamical equilibration occurs in any quantum chaotic system .
meanwhile we note that the equilibration process for the henon - heiles system deserves more detailed study in the future .
as noted before , the henon - heiles system has no bound states . as a result , a gaussian wave packet will eventually leak out and spread out in the whole space , not confined to just the triangle area , beyond the tunneling time . in our numerical simulation , the equilibration time is clearly much shorter than the tunneling time
. it would be very interesting to investigate in which situations where the equilibration time becomes shorter than the tunneling time .
it might also be worthwhile to formulate mathematically the ergordic theorem for this kind of systems .
the mathematical expression of the quantum ergodic theorem is an inequality . for an arbitrary operator @xmath43 , this inequality reads @xcite @xmath44 where @xmath45 measures effectively how many energy eigenstates are occupied in the state @xmath46 .
the subscript @xmath47 in @xmath48 indicates a long time averaging .
we emphasize that this inequality is much stronger than the following approximation _ t(a _ ) , [ ergodic ] which can be readily proved for any quantum systems with no energy degeneracy .
the above approximation can still be true even when @xmath49 fluctuate greatly from @xmath50 as long as the positive large fluctuations cancel out the negative large fluctuations .
however , the fluctuations can not cancel each other in the inequality ( [ ineq ] ) ; this means that the inequality ( [ ineq ] ) dictates that the fluctuations are very small most of the time when @xmath51 .
so , the approximation relation ( [ ergodic ] ) coupled with the inequality ( [ ineq ] ) shows that the long time averaging is equivalent to ensemble averaging , essence of ergodicity , in all quantum systems that satisfy the condition ( [ condition ] ) . to test numerically the inequality ( [ ineq ] )
, one needs to compute the energy eigenstates @xmath1 successively up to a high energy value and find the expansion coefficients @xmath2 .
we are able to do it for the ripple billiard system .
the expansion coefficients of the initial state eq . ( [ initial_condition ] ) are computed and shown in fig .
[ decomposition ] , where @xmath2 s are grouped according the symmetry of the eigenstates .
both groups peak around 500th eigenstates and have a width around 300 . with the computed @xmath2 s , we find that the effective dimension @xmath52 is around 300 , satisfying the condition @xmath53 . without loss of generality , we choose to compute the left hand side ( l.h.s . ) of the inequality eq .
( [ ineq ] ) for momentum operator @xmath54 . by using the symmetry of the eigenstates
, one can readily show exactly that @xmath55 and @xmath56 .
the time evolution of momentum is already shown in fig . [ momentum ] .
we calculate the long time average between @xmath57 and @xmath58 .
the time unit @xmath38 is the period of the motion of a classical particle with the same initial momentum and position .
we find that @xmath59 and @xmath60 , very close to the ensemble average @xmath55 and @xmath56 , respectively . at the same time
, we find that & & _ t + & & = _ t 0.0039 . in the ripple billiard system for ( a )
even - even eigenstates and ( b ) odd - even eigenstates.,scaledwidth=45.0% ] for operator @xmath37 , the maximum value @xmath61 . since we have chosen @xmath62 and the wave function
is only none zero inside the billiard , we have @xmath63 , that is , @xmath64 is effectively the largest energy in the occupied hilbert space .
according to fig .
[ decomposition ] , the ocuppied hilbert space is roughly spanned by the eigenstates between the100th even - even(odd - even ) eigenstates and the 1000th even - even(odd - even ) eigenstates .
the upper bound can be estimated as the eigenvalue of the 1000th even - even(odd - even ) eigenstates , i.e. @xmath65 .
so , the relative fluctuation or the l.h.s . of eq . ( [ ineq ] ) is @xmath66 . as @xmath67
, we see that the inequality ( [ ineq ] ) is clearly satisfied .
moreover , our computation in fact shows that the right hand side ( r.h.s . ) of the inequality is about 30 times larger than the l.h.s .
this indicates that it is possible to improve the inequality , for example , replacing @xmath68 with the averaged value .
before we proceed further , we want to mention that the inequality for quantum ergodicity proved by von neumann @xcite has a different upper bound on the r.h.s .. however , his theoretical formalism relies on the introduction of macroscopic operators ( such as macroscopic position and momentum ) that commute with each other .
it appears not be a straightforward task to construct these macroscopic operators and compute them numerically . as a result , we did not compute von neumann s upper bound .
according to the quantum ergodic theorem , the equilibrium state is described by the density matrix @xmath3 , which can be regarded as the quantum micro - canonical ensemble .
we analyze this quantum equilibrium state and find that it possesses many features that resemble the classical micro - canonical ensemble .
this kind of quantum - classical correspondence has been studied and found in a spin system @xcite . to illustrate this quantum - classical correspondence
, we compare @xmath3 in both the real space and the momentum space with the classical micro - canonical ensemble _
c= ( h(,)-e ) where @xmath69 is the normalization factor and @xmath70 is the corresponding classical hamiltonian of the fully chaotic system .
here we choose the henon - heiles system to analyze because its probability distribution in phase space is more general than the billiard system where @xmath71 is a constant of motion .
for the quantum results , we calculate @xmath72 and @xmath73 by long - time averaging ; for the classical results , we calculate the probability distribution in real space and momentum space @xmath74 and @xmath75 from the micro - canonical ensemble @xmath76 by integration over @xmath77 and @xmath78 separately . for better comparison ,
we choose the following marginal distribution without loss of generality : for the real space , we integrate out the @xmath79 dimension to obtain the density distribution @xmath80 ; for the momentum space , we integrate out the angle variable to find the momentum distribution @xmath81 .
the results are shown in fig .
[ correspondence ] , where we see the quantum and classical distributions are consistent , except for some quantum tunneling effect indicated by the non - zero value of the quantum distribution in the classically forbidden region .
this correspondence also exists in phase space . in quantum mechanics
, the uncertainty relation does not allow the construction of a phase space in principle .
however , a kind of quasi - quantum distribution in phase space can be constructed with the husimi function @xmath82 @xcite .
this is to calculate the projection of the density operator on a gaussian wave packet @xmath83 centered around position @xmath78 and momentum @xmath77 h_r.p .
the husimi function can be understood as coarse grained phase space density with parameter @xmath84 controlling the coarse graining .
we have computed the husimi function @xmath85 for a density matrix @xmath86 at @xmath87 ( which is after equilibration ) and compared it to the classical ensemble @xmath88 . for easy illustration
, we use a 2-dimensional section in the 4-dimensional phase space of the henon - heiles system . without loss of generality , we choose the section at @xmath89 and the results are plotted in fig .
[ husimiplot ] .
we can see that the husimi function centers around the phase space where the classical density is non - zero , indicating the quantum - classical correspondence .
note that the quantum fluctuations in the phase space are much larger than the ones seen in fig .
[ correspondence ] .
the reason is that the results in fig .
[ correspondence ] are obtained after being averaged over time and integrated over a given dimension .
one possible understanding for this correspondence is through the quantum and classical liouville equations .
intuitively , one can think of the quantum wave - packet as an ensemble of particles with equivalent classical probability density in the phase space @xcite .
the time evolution of quantum density matrix operator @xmath90 is governed by the quantum liouville equation , @xmath91 $ ] ; the time evolution of the classical probability density @xmath92 is governed by the classical liouville equation , @xmath93_{\rm pb}$ ] .
these two time evolution equations have an identical algebraic structure , implying a possible quantum - classical correspondence in the equilibrium states .
moreover , there are numerical evidences for this correspondence in the studying of the time evolution of the ensemble average and quantum expectation value @xcite .
note that this quantum - classical correspondence is not implied in the quantum ergodic theorem @xcite .
+ we emphasize here that the quantum - classical correspondence discussed so far needs to be understood in the sense of typicality @xcite .
as dictated by the quantum ergodic theorem , the equilibrium state is completely determined by the initial expansion coefficient @xmath94 . for a typical initial state
, @xmath94 should have a distribution similar to what is shown in fig .
[ decomposition ] with a large effective dimension @xmath52 . for these states
, we expect that the quantum - classical correspondence hold
. however , for many atypical states , this quantum - classical correspondence may not hold .
for example , the ergodic inequality ( [ ineq ] ) holds for an energy eigenstate ; but energy eigenstate does not belong to the typicality class . here
we use the ripple billiard to illustrate this point as the energy eigenstates in this system can be computed successively @xcite .
we study the position space marginal distribution @xmath95 , and compare it with the classical result @xmath96 .
we compute @xmath97 first for the equilibrium state @xmath98 .
we see in fig .
[ finalstate ] that the quantum result ( black line ) are in good agreement with the classical result ( red line ) . for completeness
, we also show the density @xmath99 in fig .
[ finalstaten ] . for eigenstates ,
we choose to study the eigenstates that have relatively large expansion coefficients @xmath94 in the initial wave packet ( [ initial_condition ] ) .
we find that the results for them are similar . without loss of generality
, we choose the 493rd even - even eigenstates with coefficient @xmath100 as the example .
the marginal distribution @xmath97 for this eigenstate is shown in fig .
[ eigenfy ] . compared with the result for the equilibrium state in fig .
[ finalstate ] , we see a clear distinction : the eigenstate has larger fluctuations and bigger deviation from the classical distribution . in fact , this distinction is also seen in the density plot . as seen in fig .
[ eigen ] , the density distribution for the eigenstate has certain uneven patterns , similar to what have been found in scar " states @xcite ; these patterns are absent in the distribution of the equilibrium state ( fig .
[ finalstaten ] ) .
although eigenstates appear to be too special , their difference from a typical quantum state does indicate that not all quantum states that satisfy the inequality ( [ ineq ] ) can relax to an equilibrium state which has the quantum - classical correspondence .
in other words , all quantum states in a chaotic system may be classified into two groups : one has the quantum - classical correspondence and the other does not have .
it is not yet clear whether there are simple ways to separate them other than using direct computation as we did . in ref.@xcite , a remarkable correspondence between eigenstates and their classical counterparts
is found for the shape of eigenfunctions and the local spectral density of states .
this seems to indicates that when the properties not related to the complexity of a wave function is considered , the quantum - classical correspondence can still be found for eigenstates .
+ while the equilibrium density operator @xmath3 shares a good correspondence with the classical microcanonical ensemble @xmath88 , the operator @xmath90 does fluctuate around the equilibrium density operator , as shown in fig .
[ finalstaten ] .
this fluctuation has been shown to obey the exponential distribution in the ripple billiard system @xcite . here
we focus on the henon - heiles system to bring this exponential distribution beyond billiard systems . to quantitatively show the fluctuation
, we compare the probability density in both the real space and the momentum space , i.e. we compare @xmath101 with @xmath102 , @xmath103 .
numerically this is realized by calculating the density @xmath104 of @xmath105 small discrete region and calculate the distribution of the relative fluctuation @xmath106 . due to limited numerical precision
, we choose different @xmath105 in the real space and momentum space and the results are shown in fig .
[ exponential](a , b ) .
both distributions are found to be well fitted by exponential distribution . due to the different @xmath105 ,
the coefficients of the fitted exponential distributions are different in the real space and the momentum space .
this exponential distribution results from the randomness of wave function in a quantum chaotic system .
though the system is deterministic and evolves strictly with the schrdinger equation , the scattering by the potential wall bring the wave function to a state where the probability density and phases behave like random numbers .
here we demonstrate the randomness by computing the phase of the wave function and the normalized probability density correlation function @xmath107 in the real space after equilibrium in the henon - heiles system .
the results are shown in fig .
[ exponential](c , d ) .
we see that the phases vary rapidly in real space and the spatial correlation in real space decreases exponentially to zero within small distance compared to the characteristic length scale @xmath108 .
these features indicate certain randomness in the behavior of wave function in quantum chaotic systems . with this established randomness , the proof in ref .
@xcite with slight modification , can explain this exponential distribution ( see also the appendix ) .
note that this randomness of wave function in quantum chaotic system is consistent with the idea of typicality @xcite .
it also supports , though not directly , the assumption we made in eq .
( 2 ) and eq .
( 3 ) in our previously work @xcite .
this exponential fluctuation is a pure quantum phenomenon . to show the quantum nature we consider the classical correspondence of an ensemble of particles obeying the same initial gaussian distribution of position and momentum and calculate their time evolution .
their phase space distribution @xmath92 approaches the microcanonical distribution @xmath88 due to the ergodicity in the henon - heiles system with certain fluctuation . following the quantum case , we calculate the relative fluctuations @xmath109 and study their distributions .
the classical relative fluctuations is gaussian as indicated in fig .
[ classical ] .
this shows that the exponential distribution of relative fluctuations is of quantum nature and absent in the classical system . besides the difference from classical case , we note that this exponential distribution is also different from the porter - thomas distribution in eigenstates @xcite and the result now known as the berry s conjecture @xcite .
these results are only for eigenstates , which are real up to an overall phase for systems with time reversal symmetry .
the exponential distribution only exists for a complex wave function , for a generic quantum system , even if its all energy - eigenstates are real , its dynamical wave function is in general complex .
this distinction again shows that the general equilibrium statistical properties can not be understood simply by properties of single eigenstates .
superposition of numerous eigenstates is crucial .
we feel it very helpful to compare directly the exponential distribution and the the port - thomas distribution @xcite .
we have re - derived these two distributions within the same mathematical framework in the appendix .
in summary , we have studied equilibration of quantum chaotic systems in the framework suggested by von neumann in 1929 .
our study has examined various aspects of this issue , such as dynamical equilibration , quantum - classical correspondence between quantum equilibrium density operator @xmath3 and classical microcanonical ensemble @xmath88 , and the relative fluctuations around the equilibrium state .
all the results are illustrated with two single particle quantum chaotic systems .
we expect most of the results to hold in many - body quantum chaotic systems .
the dynamical equilibration should hold in many - body quantum chaotic systems because it is completely determined by the energy gap statistics , which is universal for all quantum chaotic systems .
the quantum - classical correspondence should also hold as it apparently originates from the similarity between quantum and classical liouville equations . however , the fluctuation properties in a many - body system will generally have gaussian form , different from the single - particle case . the only
except might be a bose - einstein condensate or a superconductor where the many - body system can be described by a single - variable wave function . in all other cases , the probability density at a given point @xmath103
can be expressed by n(_1)=d_2d_n n(_1,_2,_n ) , where @xmath105 is the number of particles in this system .
the fluctuation in @xmath110 should have the exponential distribution due to the quantum chaotic nature of the many - body system .
the integration over @xmath111 variables will produce a gaussian distribution due to the central limit theorem .
we note that one has to use many - body systems to discuss the canonical statistics for a subsystem of the large isolated system .
however , one can use the correspondence between @xmath3 and @xmath88 illustrated here with single - particle systems to show that a subsystem of the many - body system will behave in canonical ensemble statistics .
the derivation is similar to the work by srednicki @xcite and here we do not discuss further . although von neumann has laid down the basic theoretical framework for the dynamical equilibration in quantum systems in 1929 @xcite , we believe that the study of this fundamental issue has just started and many basic questions are still await to be answered . our study here and other related studies @xcite have indicated that the quantum ergodic theorem hold for a class of quantum systems larger than what is specified by the condition ( [ condition ] ) .
it is imperative to know how to relax the condition ( [ condition ] ) mathematically .
we are also not clear about the properties of the entropy defined for a pure state by von neumann .
we believe that these studies will not only lead to a better perspective on the foundation of quantum statistical mechanics but also produce new physics such as the quantum multi - temperature equilibrium state predicted in ref .
@xcite by us .
this work is supported by the nbrp of china ( 2012cb921300,2013cb921900 ) and the nsf of china ( 11274024,11128407 ) and the rfdp of china ( 20110001110091 ) .
35 k. huang , _ statistical mechanics _
, page 176 .
john wiley & sons , inc .
, 2 edition ( 1987 ) .
l. d. landau and e. m. lifshitz , _ statistical mechanics ( i ) _ , pergamon ( 1987 ) .
e. schrdinger , statistical thermodynamics , cambridge university press , cambridge , england ( 1952 ) . j. von neumann , z. phys .
57 , 30 ( 1929 ) ; [ eur . phys .
j. h * 35 * , 201 ( 2010 ) ] .
s. goldstein , j. l. lebowitz , r. tumulka , and n. zanghi european phys
. j. h * 35 * , 173 ( 2010 ) .
p. reimann , phys .
, 101:190403 ( 2008 ) .
p. reimann and m. kastner , new j. phys .
, 14:043020 ( 2012 ) .
short , new j. phys .
, 13(5):053009 ( 2011 ) .
short and t.c .
farrelly , new j. phys . , 14:013063 ( 2012 ) .
q. zhuang and b. wu , arxiv:1211.2956 ( 2012 ) .
h. xiong and b. wu , laser phys .
, 8:398404 ( 2011 ) .
j. gemmer , m. michel , and g. mahler , quantum thermodynamics , lecture notes in physics , pages 86126 .
verlag berlin heidelberg , 2 edition ( 2009 ) .
m. rigol , v. dunjko , and m. olshanii , nature , 452:854858 ( 2008 ) .
g. biroli , c. kollath , and a.m. luchli , phys .
, 105:250401 ( 2010 ) .
cassidy , c.w .
clark , and m. rigol , phys .
, 106:140405 ( 2011 ) .
m. srednicki , phys .
e , 50:888901 ( 1994 ) .
m. rigol and m. srednicki , phys .
, 108:110601 ( 2012 ) .
deutsch , phys .
a , 43:20462049 ( 1991 ) .
s. popescu , a.j .
short , and a. winter , nature phys .
, 2:754758 ( 2006 ) .
s. popescu , a.j .
short , and a. winter , arxiv : quant - ph/0511225 ( 2005 ) .
s. goldstein , j.l .
lebowitz , r. tumulka , and n. zangh , phys .
, 96:050403 ( 2006 ) .
h. dong , s. yang , x. f. liu , and c. p. sun , phys .
a * 76 * , 044104 ( 2007 ) .
wang , phys .
e * 86 * , 011115 ( 2012 ) .
p. reimann , phys .
, 99:160404 ( 2007 ) .
n. linden , s. popescu , a.j .
short , and a. winter , phys .
e , 79:061103 ( 2009 ) .
j. cho and m.s .
kim , phys .
lett . , 104:170402 ( 2010 ) .
ikeda , y. watanabe , and m. ueda , phys .
e , 84:021130 ( 2011 ) .
m. rigol , v. dunjko , v. yurovsky , and m. olshanii , phys .
, 98:050405 ( 2007 ) .
santos , a. polkovnikov , and m. rigol , phys .
e , 86:010102 ( 2012 ) .
yukalov , laser phys .
8 : 485507 ( 2011 ) .
yukalov , ann .
phys . , 327 : 253263 ( 2012 ) .
yukalov , phys .
a , 375 : 27972801 ( 2011 ) .
a. riera , c. gogolin , and j. eisert , phys .
108 : 080402 ( 2012 ) . c. gogolin , m.p .
mller , and j. eisert , phys .
lett . , 106(4 ) : 040401 ( 2011 ) .
k. ji and b.v .
fine , phys .
, 107:050401 ( 2011 ) .
fine and f. hantschel , arxiv : cond - mat/1010.4673 ( 2010 ) .
fine , phys .
e , 80:051130 ( 2009 ) .
f. borgonovi , i. guarneri , f. izrailev , and g. casati , phys .
a , 247:140144 ( 1998 ) .
flambaum , f.m .
izrailev , and g. casati , phys .
e , 54:21362139 ( 1996 ) . c. gogolin , arxiv : quant - ph/1003.5058 ( 2010 ) . c. ududec , n. wiebe , and j. emerson , phys .
* 111 * , 080403 ( 2013 )
. k. b. davis _ et al .
lett . , 75(22 ) , 39693973 ( 1995 ) .
_ y. shin _ et al . ,
lett . , 92 , 150401 ( 2004 ) .
_ w. yao , r.b .
liu , and l.j .
sham , phys .
b , 74 , 195301 ( 2006 ) .
n. zhao , z.y .
wang , and r.b .
liu , phys .
lett . , 106 , 217205 ( 2011 ) .
, nature commu . , 2 , 570 ( 2011 ) .
_ x .- d .
et al . , nature , 459 , 1105 ( 2009 ) ; w. yao and y. luo , epl , 92 17008 ( 2010 ) . _ h - j stckmann , quantum chaos , an introduction , chapter 2 , cambridge university press ( 1999 ) . l. f. santos , f. borgonovi , and f. m. izrailev , phys .
lett . * 108 * , 094102 ( 2012 ) .
l. f. santos , f. borgonovi , f. m. izrailev , phys .
e 85 , 036209 ( 2012 ) .
v. v. flambaum , g. f. gribakin , and f. m. izrailev , phys .
e * 53 * , 5729 ( 1996 ) .
m. wilkinson and e. j. austin , phys .
a * 46 * , 64 ( 1992 ) ; m. wilkinson and e. j. austin , j. phys .
a. * 28 * , 2277 ( 1995 ) .
w. li , l. e. reichl , and b. wu , phys .
e * 65 * , 056220 ( 2002 ) .
m. d. feit and j. a. fleck , j. chem .
phys . * 80 * , 2578 ( 1984 ) . b. li and m. robnik , j. phys .
a : math . gen .
* 27 * , 5509 ( 1994 ) .
r. a. pullen and a. r. edmonds , j. phys .
* 14 * l477 ( 1981 ) ; d engel , j main , and g wunner , j. phys . a : math .
* 31 * ( 1998 ) 6965 ; m. brack , r. k. bhaduri , j. law , and m. v. n. murthy , phys
lett . * 70 * , 568 ( 1993 ) ; d. w. noid and r. a. marcus , j. chem . phys . * 67 * , 559 ( 1977 ) .
s. b. lee and m. d. feit , phys .
e. , 47 : 4552 ( 1993 ) .
toscano _ et al . ,
a , 464(2094):15031524 ( 2008 ) .
_ j. emerson and l.e .
ballentine , phys .
a * 63 * , 052103 ( 2001 ) ; j. emerson and l. e. ballentine , phys .
e * 64 * , 026217 ( 2001 ) .
f. borgonovi , i. guarneri , and f. m. izrailev , phys .
e * 57 * , 5291 ( 1998 ) .
ballentine , y. yang , and j.p .
zibin , phys .
a , 50:28542859 ( 1994 ) .
ballentine , s.m .
mcrae , phys .
a , 58:17991809 ( 1998 ) .
j. l. lebowitz , phys .
today * 9 * , 32 ( 1993 ) .
heller , phys .
lett . , 53(16 ) : 15151518 ( 1984 ) . c. e. porter and r. g. thomas , phys .
, 104(2 ) : 483491 ( 1956 ) .
m. v. berry , j. phys .
a. * 10 * , 2083 ( 1977 ) .
here we give detailed derivations for the exponential distribution of equilibrium wave function @xcite and the port - thomas distribution of eigenstate wave function @xcite with the assumption that wave - function in quantum chaotic systems achieve the maximum randomness subject to the restriction of normalization and other symmetry requirements .
this assumption follows naturally from our numerical observations(fig .
[ exponential ] ( c ) , ( d ) ) and the spirit of random matrix theory , and was used by berry in his study of eigen - wave functions in chaotic systems @xcite it is important to note the difference between equilibrium wave functions and eigenstates wave functions is that equilibrium wave functions are generally complex and eigenstates wave functions are real up to an overall phase for systems with time reversal symmetry .
this key difference leads to the different distributions . for simplicity
we restrict our derivation for real space wave functions in chaotic billiards .
extension to other systems and momentum space can be achieved ( however , without strict mathematical rigorousness ) by renormalizing densities with averaged values . for a wave function @xmath112 ,
we discretize the system area into @xmath105 equal infinitesimal pieces with their centers located at @xmath113 and denote @xmath114 , where @xmath43 is the total area of the system .
it follows from normalization that @xmath115 .
the notation here is the same with ref.@xcite . for a complex wave function in dynamical evolution @xmath116
is complex ; for an eigenfunction @xmath116 is real . in the following
we derive the distribution of @xmath117 for both cases .
note that the following proof is given without consideration of the discrete symmetries : ( 1 ) reflection symmetry initial state(eqn .
[ initial_condition ] ) in ripple billiard system ; ( 2 ) reflection symmetries in eigenstates of ripple billiard system , and ( 3 ) the reflection symmetries in eigenstates of henon - heiles system .
this is because these symmetries can be treated by simply assuming full randomness of a wave function in a smaller area while changing the normalization , respectively , by @xmath118(eq .
[ initial_condition ] ) ) , @xmath119 ( ripple billiard ) , or @xmath120(henon - heiles ) . [ [ proof - of - exponential - distribution - of - equilibrium - wave - function ] ] proof of exponential distribution of equilibrium wave function @xcite ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ consider @xmath105 complex numbers @xmath116 that satisfies normalization condition @xmath115 and each complex number is equivalent to another .
suppose the @xmath105 complex numbers are fully random subject to the normalization condition .
this implies that each state @xmath121 is of equal possibility on the hypersphere @xmath122 .
so the distribution of the amplitude of one complex number @xmath123 at @xmath124 is let @xmath125 .
the denominator equals the surface area of @xmath126-dimensional hypersphere .
the numerator includes a factor of a @xmath127-dimensional hypersphere with radius @xmath128 .
recall the @xmath126-dimensional hypersphere of radius @xmath129 has an area of @xmath130 .
we have p()&= & + & = & ( n-1)(1-)^n-3/2 . let @xmath131 , where @xmath132 is the probability density at @xmath78 .
we find that for an eigenstate wave function , everything is the same except that all the @xmath116 are real rather than complex .
so , we have @xmath134 , and similarly the distribution becomes p()=. let @xmath135 .
the denominator equals the area of the @xmath105-dimensional hypersphere with unit radius .
the numerator includes a factor of the area of a @xmath136-dimensional hypersphere with radius @xmath128 .
this leads to p()&= & + & = & ( ) ^n-2 with @xmath131 , we arrive at the porter - thomas distribution in the limit of large @xmath105 , p(n ) = e^-n where we have used @xmath137 . | quantum ergordic theorem for a large class of quantum systems was proved by von neumann [ z. phys . *
57 * , 30 ( 1929 ) ] and again by reimann [ phys .
rev .
lett .
* 101 * , 190403 ( 2008 ) ] in a more practical and well - defined form . however , it is not clear whether the theorem applies to quantum chaotic systems . with the rigorous proof still elusive ,
we illustrate and verify this theorem for quantum chaotic systems with examples .
our numerical results show that a quantum chaotic system with an initial low - entropy state will dynamically relax to a high - entropy state and reach equilibrium .
the quantum equilibrium state reached after dynamical relaxation bears a remarkable resemblance to the classical micro - canonical ensemble .
however , the fluctuations around equilibrium are distinct : the quantum fluctuations are exponential while the classical fluctuations are gaussian . |
recently wang , zheng , yao , and langley@xcite ( wzyl ) have verified by experiment an effect , which is indeed the sagnac effect seen by a uniformly moving observer .
wzyl s experiment actually shows that what matters is that light moves along a closed circuit and that the observer is in motion with respect to the circuit .
the observer needs not rotate for the effect to appear .
the sagnac effect has been the object of innumerable papers@xcite@xcite and has stimulated an almost one century long debate on its meaning and interpretation .
most of this debate is rather surprising since the effect is perfectly well described and understood in the framework of special relativity ( sr ) .
a simple way to describe and derive the effect in sr consists in a purely geometrical approach
. in the traditional configuration one has to do with a uniformly rotating observer ( angular velocity @xmath0 ) and two light rays obliged to stay on a circle ( radius @xmath1 ) by an optical fiber or a cylindrical mirror .
drawing a space - time diagram of the situation , one has world - lines drawn on a cylinder@xcite , like in fig .
[ fig.1 ] .
the graph is even clearer when one considers that the cylinder is a flat surface , i.e. a minkowski two - dimensional space .
we are then allowed to cut the cylinder along a generatrix , and to open it on a plane leading to fig .
[ fig.2 ] .
2 actually displays the same situation as fig .
[ fig.1 ] .
the cyclical periodicity implies that @xmath2 and @xmath3 coincide with @xmath4 .
@xmath5 is proportional to the sagnac delay .
a simple calculation@xcite gives the result@xmath6 in the first order approximation this is traditionally written as@xmath7and one says that the proper time difference is proportional to the area @xmath8 encircled by the beam .
actually , considering geometry and the fact that lengths are involved , it would be more appropriate to write ( [ sagnac2 ] ) in the form ( first order)@xmath9where @xmath10 is the length of the path as seen by an inertial observer at rest with respect to the axis of the system , and @xmath11 is the velocity of the observer in the initial direction of both the co - moving and counter - moving rays .
( [ sagnac3 ] ) is the one written by wzyl .
formula ( [ sagnac2 ] ) has been considered as universal , however it is the product of a fortuitous coincidence when light and observer are both rotating along the same circumference .
let us now consider the more general situation depicted in fig .
3 . now there is an observer coinciding with the source / receiver of light and steadily moving along the straight line between @xmath1 and @xmath12 .
the line joining @xmath12 and @xmath1 may be thought of as an optical fiber or a mixed path including mirrors and fibers , guiding light beams in both directions ( from @xmath13 to @xmath1 and viceversa ) .
light emitted by @xmath4 forwardly enters the guide in @xmath12 and exits in @xmath1 , aiming at @xmath4 .
light sent backwardly does the reverse .
a simple , almost trivial , argument shows that light emitted in both directions in a single burst is received back at two different instants .
the proper time difference measured by @xmath4 will be:@xmath14 again the formula coincides with the one written by wzyl . now it is evident that no connection exists with rotations and with the area enclosed in the path of light . what matters are the length of the contour and the speed of the observer , which in this case is an inertial one .
of course @xmath15 is parallel to the light rays in @xmath16 .
using the geometrical approach the result is rather obvious . in space - time
the world lines are drawn again on a cylinder .
the only difference is that on fig .
1 the cylinder had a circular cross section , now the cross section is rather irregular . in any case
the cylinder is a flat bidimensional minkowski surface and , when opening in a plane , the graph is exactly like in fig .
the accidental introduction of the area @xmath8 in ( [ sagnac2 ] ) suggested to some people@xcite an interpretation of the sagnac effect ( at least for matter waves ) as a sort of aharonov - bohm effect .
this is definitely not the case .
a relevant part of the discussion of the sagnac effect concerned the problem of the one- or two - ways velocity of light@xcite . using a closed circuit is in some sense a way to measure a `` one - way '' velocity of light and no doubt that naively dividing the length of the path by the time of flight measured by the observer two different results are obtained .
this is however no problem : \a ) everything is perfectly consistent with sr and all the basic assumptions in it ( including the constancy of the speed of light for all inertial observers ) ; \b ) the observer knows that the difference in times is due to the relative motion with respect to the physical trail of light .
some years ago selleri(see for instance@xcite ) , considering the problem of the speed of light on a rotating disk remarked that letting the radius of the disk go to infinity and the angular velocity to zero in such a way to leave the peripheral speed constant , one would smoothly pass from rotational to uniform translational motion .
yet there would be a discontinuity in the speed of light forward and backward .
here we have seen that the uniform motion of the observer together with a closed path for light rays does indeed produce different times of flight for light with no need for rotations and infinitely sized disks .
selleri s gedanken experiment is in fact impossible , because the times of flight whose difference should remain constant , would become infinite , then unmeasurable .
looking to fig .
[ fig.3 ] it is clear that it is indeed easy to measure ( and this is what wzyl have done ) a discontinuity in the times of flight backward and forward , without so challenging in any way the sr .
of course if one measures a two - ways time of flight , i.e. light going back and forth along the same path , no discontinuity is found .
we have shown that the result obtained by wzyl is indeed general and sheds a new light on the interpretation of the sagnac effect .
actually this effect is due to the closeness of the path followed by light and to the relative motion of the observer with respect to the physical system obliging the beam to bend and come back to the observer .
if this is correct the sagnac effect has nothing to do with rotations and accelerated motion , rather it is a peculiarity of relative motion relevant to the old debate about the one- or two - ways measurement of the speed of light . | the sagnac effect for a uniformly moving observer is discussed . starting from a recent measurement
we show that the sagnac effect is in fact not a consequence of the rotation of the observer , but simply of its , even inertial , motion with respect to the device resending light towards the ( moving ) source . |
the simple and accurate solution of some problems involving the particle density distributions uses the nuclear effective surface ( es ) approximation@xcite .
it exploits the property of saturation of the nuclear matter and a narrow diffuse - edge region in finite nuclei .
the es is defined as the location of points of the density gradient maximum .
the coordinate system related locally to the es is specified by a distance @xmath1 from the given point to the surface and tangent coordinate @xmath2 ( see fig . 1 ) .
the variational condition of the nuclear energy minimum at fixed other integrals of motion within the local energy density theory , in particular , the extended thomas - fermi ( etf ) approach@xcite is simplified much in the @xmath3 coordinates for any deformations by using expansion in small parameter @xmath4 for heavy enough nuclei ( @xmath5 is the diffuse edge thickness of the nucleus , and @xmath6 its mean curvature radius ) .
the accuracy of the es approximation in the etf approach was checked@xcite by comparing results of the hartree - fock ( hf ) and etf theories based on skyrme forces@xcite without spin - orbit and asymmetry terms . within the es approximation a rather reasonable agreement of the calculations with the experimental data on a mean particle - number dependence on the excitation energies and reduced transition probabilities of the low - lying collective states of non - magic nuclei was found@xcite . in the present work ,
we extend the es approach@xcite taking into account the spin - orbit and the asymmetry effects in nuclei .
fig . 1 : es and local @xmath3 coordinates .
the profile function @xmath7 in cylindrical @xmath8 coordinates is shown schematically by thick solid curve @xcite ; @xmath5 is the parameter of the nuclear diffuse edge .
enden we begin with the nuclear energy @xmath9 within a local energy functional approach@xcite : energytot e&= & , ( _ + , _ - ) - b_v _ + + ^2 _ + + ( 1-x ) _ + + & + & _ + + ( a + b _
+ + ) ( _ + ) ^2 + a_- ( _ -)^2 , where @xmath10 $ ] is the energy density as a function of the isoscalar @xmath11 and isovector @xmath12 particle densities .
it overlaps approximately most of the realistic skyrme forces@xcite .
@xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 is the coulomb potential and @xmath19 is its average up to a small exchange component @xcite . as usually , @xmath20 of ( 1 ) contains the volume , and the surface terms without and with the gradient density terms @xcite , @xmath21 = 16 mev is the separation energy per particle and @xmath22=60 mev is the symmetry energy constant of the nuclear matter .
the semiclassical @xmath23 corrections appear through @xmath24 in the etf kinetic energy density@xcite , @xmath25 is the nucleon mass . in ( [ energytot ] ) , we have neglected relatively small isovector ( spin - orbit and semiclassical ) corrections .
the isoscalar surface energy density part , independent of the density gradient terms , is determined by the function @xmath26 which satisfies the saturation condition : satcond _ + ( ) = 0 , _ + ( ) /_+ = 0 , where @xmath27 0.16 @xmath28 is the density of the infinite nuclear matter , @xmath29 is constant independent of @xmath30 . for the isovector component one
has epsminus _ -=(x^2 - _ -^2/_+^2 ) - .
the spin - orbit gradient terms in ( [ energytot ] ) are defined with a constant : @xmath31 , @xmath32=100 - 130 mev @xmath33 ( see refs .
5,7 ) . from the condition of the energy @xmath9 ( [ energytot ] ) minimum together with the constraints for the fixed particle number @xmath30 , neutron excess @xmath34 , and deformation @xmath35 of the nucleus@xcite : constraintsatq a = _ + ( ) , n - z = _ - ( ) , q = _ + ( ) q ( ) , one arrives at the variational lagrange equations : lagrangeqt -_+ -_q q = 0 , - _ - = 0 .
here , @xmath36 , @xmath37 and @xmath38 are the corresponding lagrange multipliers where @xmath36 and @xmath37 are the isoscalar and isovector chemical potentials , respectively .
isivden _ in the nuclear volume _ , up to the second order in @xmath39 one gets@xcite : epsvol _ + ( _ + ) = ( _ + - ) ^2 , ( w)==(1-w)^2 , w=. where @xmath40 is the incompressibility of the infinite nuclear matter . from the lagrange equations ( [ lagrangeqt ] )
one finds for the volume densities @xmath41 : rhominusvol _
+ ^(v ) ( 1 + ) , _
-^(v ) ( x+ ) .
small finite - size corrections of the order of @xmath42 ( @xmath43 is a mean es curvature ) are determined by the surface components of the corresponding chemical potentials : lambdat _ tot^(+ ) & = & _ + + b_v -^2 - + _ q q()~ah ~a / r ~a^-1/3 , + _
tot^(-)&=&_- -b_sym x+ ~ah .
_ for the dimensionless isoscalar density
_ @xmath44 ( [ epsvol ] ) , from the first equation ( [ lagrangeqt ] ) , _ up to the leading order _ in @xmath45 one obtains the ordinary first - order differential equation : yeq0plus = -w , x= , a= , where @xmath46 , @xmath47 . by differentiating equation ( [ yeq0plus ] )
one finds the boundary condition from the definition of the es : @xmath48 at @xmath49 ( @xmath50 ) , boundcond ( w_0 -w_0 ^ 2 - ) ( w_0 ) + w_0(w_0 + w_0 ^ 2 + ) ( ) _ w = w_0=0 , together with the condition of the exponentially vanishing the density outside the nucleus : @xmath51 . solving the problem ( [ yeq0plus ] ) , ( [ boundcond ] ) , one arrives at the solution in the inverse form @xmath52 : ysolplus x=-_w_0^w . with the quadratic approximation ( [ epsvol ] ) for @xmath53 one gets the analytical solutions in terms of the algebraic , trigonometric and logarithmic functions . for @xmath54 it simplifies to @xmath55 $ ] for @xmath56 and zero for @xmath57 outside the nucleus@xcite . in fig . 2 ( left )
the influence of the semiclassical correction to @xmath44 is shown by comparing the @xmath58 ( dashed line ) and the `` exact '' ( thin solid line ) cases .
this correction is small everywhere , besides the quantum tail outside the nucleus for @xmath59 .
almost the same results one obtains for the skm@xmath60 and sly7 forces@xcite .
one should also notice a rather big effect of the spin - orbit interaction as compared to the simplest analytical solution at @xmath54 .
we found also a good convergence of the expansion of the @xmath53 in powers of @xmath61 in the density solution ( [ ysolplus ] ) by comparing the exact numerical function@xcite @xmath53 to its approximate solution ( [ epsvol ] ) .
the agreement is within a precision of the line thickness .
2 ( right ) presents a weak sensitivity of the isoscalar solution @xmath44 ( [ ysolplus ] ) on the different skyrme forces with ( [ epsvol ] ) for @xmath53 .
fig . 2 : density @xmath44 ( [ ysolplus ] ) ( left ) as a function of @xmath62 and its comparison ( right ) for several skyrme forces@xcite ( @xmath635 - 7,230a and 230b in slyn ) for @xmath53 ( [ epsvol ] ) .
_ for the isovector density up to the leading order _ in @xmath45 , after simple transformations one finds the equation and the boundary condition in the form yeq0minus = c_sym , w_-(1 ) = 1 , w_-= , where @xmath64 . up to the leading order of the es approximation in @xmath45 , one obtains the analytical solution through the expansion in powers of @xmath61 , ysolminus w_- = w , u(w)= .
the dependence of the dimensionless isovector density @xmath65 ( [ ysolminus ] ) on the semiclassical and spin - orbit effects versus the corresponding results for the density @xmath44 ( [ ysolplus ] ) are shown in fig .
3 ( left ) .
a weak sensitivity of the dimensionless isovector density @xmath66 on the choice of the skyrme forces is seen in fig .
3 ( right ) .
esldm _ for more exact isoscalar particle density _ we calculate the main terms of higher order in the parameter @xmath45 in the first equation ( [ lagrangeqt ] ) . integrating this equation over the es in normal - to - surface @xmath1 direction and using the equation ( [ yeq0plus ] ) up to the leading order in @xmath45 , one arrives at the differential equation boundcondpplus p_+|_es & = & p_s^(+),p_+= _ tot^(+)= \{(_+ - ) + + } ^(v),p_s^(+)=2 _ + h. this equation can be considered with respect to the unknown sought profile shape @xmath7 of the es in the cylindrical coordinates @xmath8 with the symmetry axis @xmath67 ( see fig .
1 ) through the curvature @xmath68 in terms of the main es curvature radii@xcite curvature r_1=l y ( ) , r_2=-l^3/(^2 y/^2 ) , = ^1/2 . fig . 3 : isovector density @xmath65 ( [ ysolminus ] ) compared to the isoscalar one @xmath44 ( [ ysolplus ] ) as a function of @xmath62 within the approximation ( [ epsvol ] ) to @xmath53 calculated for the same sly7 forces ( left ) and @xmath65 for several skyrme forces@xcite ( right ) .
( [ boundcondpplus ] ) is associated with the macroscopic boundary condition@xcite with the isoscalar capilliary surface pressure @xmath69 which is proportional to the surface tension coefficient @xmath70 : sigmaplus
_ + 2_-^ ( a + b _
+ + ) ( ) ^2 .
_ for more exact isovector particle density _ similarly one obtains the isovector macroscopic boundary condition@xcite , boundcondpminus p_-|_es= p_s^(-),p_-= _
tot^(-)= x^(v ) , with the isovector surface pressure sigmaminus p_s^(-)=2 _ - h , _ - 2 a_- _
-^ ( ) ^2 , where @xmath71 is the isovector tension coefficient .
surfent the nuclear energy @xmath72 ( [ energytot ] ) in the es approximation is split into the volume , @xmath73 , and the surface terms : eves e_s = s= ( b_s^(+ ) + b_s^(- ) ) s/4r_0 ^ 2,=_+ + _ -,b_s^()= 4 r_0 ^ 2 _ , where @xmath74 is the surface area of the es .
the energy @xmath75 ( [ eves ] ) is determined by the sum of the isoscalar @xmath76 and isovector @xmath77 surface energy constants .
these constants are proportional to the same tension coefficients @xmath78 which appear in ( [ sigmaplus ] ) and ( [ sigmaminus ] ) and expressed through the surface pressures in ( [ boundcondpplus ] ) and ( [ sigmaminus ] ) , respectively , bsplus b_s^(+)= _ 0 ^ 1 w , bsminus b_s^(-)= 108 _ - x^2 _ 0 ^ 1 w \{-
w u(w)}^2 , where @xmath79 and see ( [ ysolminus ] ) for @xmath80 .
simple expressions for constants ( [ bsplus ] ) and ( [ bsminus ] ) in terms of the algebraic and trigonometric functions can be easily obtained by calculating explicitly the integrals over @xmath81 in the above equations with @xmath53 taken from eq.(6 ) .
neglecting relatively small spin - orbit terms and semiclassical corrections one finds the approximate relationship between the isovector and the isoscalar energy constants , @xmath82 . in table
1 the analytical , @xmath83 with the approximated ( [ epsvol ] ) for @xmath53 , the numerical , @xmath84 with the exact @xmath53 , and @xmath76 from ref.7 are shown for all skyrme forces . one can see a very good agreement between all these calculations , besides of siii .
modula of the isovector constants for the lyon skyrme forces slyn@xcite are much larger than for other ones .
the precision of the spin - orbit and semiclassical terms of the isovector energy density part ( [ energytot ] ) is not enough accurate for all considered skyrme interactions as the isovector surface tension @xmath71 which appears in the surface energy ( [ eves ] ) becomes inconsistent with that of the capilliary isovector pressure ( [ boundcondpminus ] ) , ( [ sigmaminus ] ) .
these terms can be improved by fitting to more detailed experimental information . _
the @xmath85-stability line _ is determined by the equivalence of the neutron and proton chemical potentials , @xmath86 , and @xmath87 , according to ( [ lambdat ] ) , ( [ boundcondpminus ] ) and ( [ sigmaminus ] ) . with _ the finite - size correction _ , one obtains @xmath88 , where @xmath89 is the leading term@xcite .
[ cols="^,^,^,^,^,^,^,^,^",options="header " , ] table 1 : the isoscalar ( [ bsplus ] ) and isovector ( [ bsminus ] ) energy surface constants @xmath90 with skyrme parameters @xcite .
concl the asymmetry and spin - orbit terms of the energy density within the etf with skyrme forces were taken into account analytically by using expansion in @xmath91 of the es at any deformation .
we derived the ordinary first - order equations for the isoscalar and isovector particle densities giving simple analytical solutions . when higher order terms are taken into account one gets equations for the moving es in terms of the macroscopic boundary conditions .
expressions for the isoscalar and isovector tension coefficients @xmath78 in the surface energy were found as those of the macroscopic capilliary pressures of the fermi liquid edge in these boundary conditions .
a simple approximate relation of @xmath92 to @xmath93 and the finite - size es correction to the @xmath85-stability condition were obtained .
our approach might be helpful as a macroscopic part of the nuclear collective dynamical macro - micromodels @xcite based on the etf ( or ldm ) for a `` macro '' component@xcite , and the semi - microscopic approach of the strutinsky shell correction method@xcite for study of the low - lying collective excitations and fission processes .
authors thank profs .
v.m . kolomietz , k. pomorski , h.j .
krappe , p. ring , v.o .
nesterenko , p. danielewicz , j. kvasil , and g. colo for many useful discussions . | isoscalar and isovector particle densities are derived analytically by using the approximation of a sharp edged nucleus within the local energy density approach with the proton - neutron asymmetry and spin - orbit effects .
equations for the effective nuclear - surface shapes as collective variables are derived up to the higher order corrections in the form of the macroscopic boundary conditions .
the analytical expressions for the isoscalar and isovector tension coefficients of the nuclear surface binding energy and the finite - size corrections to the @xmath0 stability line are obtained . |
the methods of statistical physics have shown to be very fruitful in physics , but in the last decades they also have become increasingly important in interdisciplinary research . for example , the _ master equation _ has found many applications in thermodynamics @xcite , chemical kinetics @xcite , laser theory @xcite and biology @xcite .
moreover , weidlich and haag have successfully introduced it for the description of social processes @xcite like opinion formation @xcite , migration @xcite , agglomeration @xcite and settlement processes @xcite .
another kind of wide - spread equations are the boltzmann equations , which have been developed for the description of the kinetics of gases @xcite and of chemical reactions @xcite .
however , boltzmann-_like equations _
@xcite play also an important role for quantitative models in the social sciences : it turns out ( cf .
[ s2.2 ] ) that the _ logistic equation _ for the description of limited growth processes @xcite , the socalled _ gravity model _ for spatial exchange processes @xcite , and the _ game dynamical equations _ modelling competition and cooperation processes @xcite are special cases of boltzmann - like equations .
moreover , boltzmann - like models have recently been suggested for avoidance processes of pedestrians @xcite and for attitude formation by direct pair interactions of individuals occuring in discussions @xcite . in this paper
we shall show that boltzmann - like equations and boltzmann - fokker - planck equations @xcite are suited as a foundation of quantitative behavioral models .
for this purpose , we shall proceed in the following way : in section [ s2 ] the boltzmann - like equations will be introduced and applied to the description of behavioral changes .
the model includes _ spontaneous _ ( or _ externally induced _ ) behavioral changes and behavioral changes by _ pair interactions _ of individuals .
these changes are described by _
transition rates_. they reflect the results of mental and psychical processes , which could be simulated with help of osgood and tannenbaum s _ congruity principle _ @xcite , heider s _ balance theory _ @xcite or festinger
s _ dissonance theory _ @xcite . however , it is sufficient for our model to determine the transition rates empirically ( sect .
the _ ansatz _ used for the transition rates distinguishes _ imitative _ and _ avoidance processes _ , and assumes _ utility maximization _ of the individuals ( sect .
[ s2.1 ] ) .
it is shown , that the resulting boltzmann - like model for imitative processes implies as special cases many generally accepted theoretical approaches in the social sciences ( sect .
[ s2.2 ] ) . in section [ s3 ] a consequent mathematical formulation related to an idea of lewin @xcite is developed , according to which the behavior of individuals is guided by a _
social field_. this formulation is achieved by a kramers - moyal expansion of the boltzmann - like equations leading to a kind of _ diffusion equations _ : the socalled boltzmann - fokker - planck _ equations _ @xcite . in these equations
the most probable behavioral change is given by a vectorial quantity that can be interpreted as _
social force _
[ s3.1 ] ) .
the social force results from external influences ( the environment ) as well as from individual interactions . in special cases
the social force is the gradient of a potential .
this potential reflects the public opinion , social norms and trends , and will be called the _ social field_. by _ diffusion coefficients _ an individual variation of the behavior ( the `` freedom of will '' ) is taken into account . in section [ s4 ] representative cases
are illustrated by computer simulations .
the boltzmann - fokker - planck modell for the behavior of individuals under the influence of a social field shows some analogies with the physical model for the behavior of electrons in an electric field ( e.g. of an atomic nucleus ) @xcite ( cf .
hartree s _ selfconsistent field ansatz _ @xcite ) .
especially , individuals and electrons influence the concrete form of the _ effective _ social resp . electric field .
however , the behavior of electrons is governed by a _
different _ equation : the schrdinger _
equation_. in physics , the boltzmann - fokker - planck equations can be used for the description of _ diffusion _ processes @xcite .
let us consider a _ system _ consisting of a great number @xmath0 of _ subsystems_. these subsystems are in one _ state _ @xmath1 of several possible states combined in the set @xmath2 .
due to _ fluctuations _ one can not expect a deterministic theory for the temporal change @xmath3 of the state @xmath4 to be realistic .
however , one can construct a _
stochastic _ model for the change of the _ probability distribution _
@xmath5 of states @xmath4 within the given system ( @xmath6 , @xmath7 ) . by introducing an index
@xmath8 we may distinguish @xmath9 different _ types _ @xmath8 of subsystems . if @xmath10 denotes the number of subsystems of type @xmath8 , we have @xmath11 , and the following relation holds : @xmath12 our goal is now to find a suitable equation for the probability distribution @xmath13 of states for subsystems of type @xmath8 ( @xmath14 , @xmath15 ) . if we neglect memory effects ( cf .
[ s6.1 ] ) , the desired equation has the form of a _ master equation _
@xcite : @xmath16 \ , .
\label{boltz}\ ] ] @xmath17 is the _ effective transition rate _ from state @xmath1 to @xmath18 and takes into account the fluctuations .
restricting the model to spontaneous ( or externally induced ) transitions and transitions due to pair interactions , we have @xcite : @xmath19 @xmath20 describes the rate of spontaneous ( resp .
externally induced ) transitions from @xmath1 to @xmath18 for subsystems of type @xmath8 .
@xmath21 is the transition rate for two subsystems of types @xmath8 and @xmath22 to change their states from @xmath1 and @xmath23 to @xmath18 and @xmath24 due to pair interactions . inserting ( [ effrate ] ) into ( [ boltz ] ) , we now obtain the socalled boltzmann-_like equations _
@xcite [ boltzlike ] @xmath25 \\ & + & \sum_{b=1}^a \sum_{{x}'\in \omega } \sum_{{y}\in \omega } \sum_{{y}'\in \omega } w_{ab}({\mbox{\boldmath $ x$}},{\mbox{\boldmath $ y$}}'|{\mbox{\boldmath $ x$}}',{\mbox{\boldmath $ y$}};t ) p_b({\mbox{\boldmath
$ y$}},t)p_a({\mbox{\boldmath $ x$}}',t ) \nonumber \\ & - & \sum_{b=1}^a \sum_{{x}'\in \omega } \sum_{{y}\in \omega } \sum_{{y}'\in \omega } w_{ab}({\mbox{\boldmath $ x$}}',{\mbox{\boldmath $ y$}}'|{\mbox{\boldmath $ x$}},{\mbox{\boldmath $ y$}};t ) p_b({\mbox{\boldmath $ y$}},t)p_a({\mbox{\boldmath $ x$}},t)\end{aligned}\ ] ] with @xmath26 obviously , ( [ boltzlike]b ) depends nonlinearly on the probability distributions @xmath13 , which is due to the interaction processes . neglecting spontaneous transitions ( i.e. , @xmath27 ) the boltzmann - like equations agree with the boltzmann equations , that originally have been developed for the description of the kinetics of gases @xcite .
a more detailed discussion can be found in @xcite . in order to apply the boltzmann - like equations to behavioral changes we have now to take the following specifications given by table 1 ( cf .
@xcite ) : .specification of the notions used in statistical physics for an application to behavioral models . [ cols="<,<",options="header " , ] it is possible to generalize the resulting behavioral model to simultaneous interactions of an arbitrary number of individuals ( i.e. , higher order interactions ) @xcite .
however , in most cases behavioral changes are dominated by pair interactions .
many of the phenomena occuring in social interaction processes can already be understood by the discussion of pair interactions . in the following we have to find a concrete form of the effective transition rates [ eff ] @xmath28 @xmath29 ( cf .
( [ effrate ] ) ) that is suitable for the description of behavioral changes .
@xmath30 is the rate of pair interactions @xmath31 where an individual of subpopulation @xmath8 changes the behavior from @xmath1 to @xmath18 under the influence of an individual of subpopulation @xmath22 showing the behavior @xmath23 .
there are only two important kinds of social pair interactions : [ ints ] @xmath32 obviously , the interpretation of the above kinds of pair interactions is the following : * the interactions ( [ ints]a ) describe _ imitative processes _ ( processes of persuasion ) , that means , the tendency to take over the behavior @xmath18 of another individual . * the interactions ( [ ints]b )
describe _ avoidance processes _ , where an individual changes the behavior when meeting another individual showing the same behavior @xmath1
. processes of this kind are known as aversive behavior , defiant behavior or snob effect .
the corresponding transition rates are of the general form [ totrate ] @xmath33 where the term ( [ totrate]a ) describes imitative processes and the term ( [ totrate]b ) describes avoidance processes .
@xmath34 has the meaning of the kronecker function . by inserting ( [ totrate ] ) into ( [ eff ] )
we arrive at the following general form of the effective transition rates : [ concrates ] @xmath35 \ , .\ ] ] for behavioral models one often assumes @xmath36 in ( [ concrates ] ) , * @xmath37 is a measure for the rate of spontaneous ( or externally induced ) behavioral changes within subpopulation @xmath8 . *
@xmath38 [ resp .
@xmath39 is the _ readiness _ for an individual of subpopulation @xmath8 to change the behavior from @xmath1 to @xmath18 spontaneously [ resp . in pair interactions ] .
* @xmath40 is the _ interaction rate _ of an individual of subpopulation @xmath8 with individuals of subpopulation @xmath22 .
* @xmath41 is a measure for the frequency of imitative processes .
* @xmath42 is a measure for the frequency of avoidance processes . a more detailled discussion of the different kinds of interaction processes and of _ ansatz _ ( [ concrates ] ) is given in @xcite . for @xmath43
we take the quite general form [ util ] @xmath44 with @xmath45 ( cf .
then , the readiness @xmath43 for an individual of subpopulation @xmath8 to change the behavior from @xmath1 to @xmath18 will be the greater , * the greater the _ difference _ of the _ utilities _ @xmath46 of behaviors @xmath18 and @xmath1 is , * the smaller the _ incompatibility ( `` distance '' ) _
@xmath47 between the behaviors @xmath1 and @xmath18 is .
similar to ( [ util ] ) we use @xmath48 and , therefore , allow the utility function @xmath49 for spontaneous ( or externally induced ) behavioral changes to differ from the utility function @xmath50 for behavioral changes in pair interactions .
_ ansatz _ ( [ util ] ) is related to the _ multinomial logit model _
@xcite , and assumes _ utility maximization _ with incomplete information about the exact utility of a behavioral change from @xmath1 to @xmath18 , which is , therefore , estimated and stochastically varying ( cf .
@xcite ) .
computer simulations of the boltzmann - like equations ( [ boltz ] ) , ( [ concrates ] ) , ( [ util ] ) are discussed and illustrated in @xcite ( cf . also sect .
[ s4 ] ) .
the boltzmann - like equations ( [ boltz ] ) , ( [ concrates ] ) include a variety of special cases , which have become very important in the social sciences : * the _ logistic equation _
@xcite describes limited growth processes .
let us consider the situation of two behaviors @xmath51 ( i.e. , @xmath52 ) and one subpopulation ( @xmath53 ) .
@xmath54 may , for example , have the meaning to apply a certain strategy , and @xmath55 not to do so .
if only imitative processes @xmath56 and processes of spontaneous replacement @xmath57 are considered , one arrives at the _ logistic equation _ @xmath58 * the _ gravity model _
@xcite describes processes of exchange between different places @xmath1 .
it results for @xmath59 , @xmath60 , @xmath61 , and @xmath53 : @xmath62 p({\mbox{\boldmath $ x$}},t)p({\mbox{\boldmath $ x$}}',t ) \ , .\ ] ] here , we have dropped the index @xmath8 because of @xmath63 .
@xmath5 is the probability of being at place @xmath1
. the absolute rate of exchange from @xmath1 to @xmath18 is proportional to the probabilities @xmath5 and @xmath64 at the places @xmath1 and @xmath18 .
@xmath65 is often chosen as a function of the metric distance @xmath66 between @xmath1 and @xmath18 : @xmath67 . *
the _ behavioral model _ of weidlich and haag @xcite assumes spontaneous transitions due to _ indirect interactions _ , which are , for example , induced by the media ( tv , radio , or newspapers ) .
we obtain this model for @xmath68 and @xmath69 @xmath70 is the _ preference _ of subpopulation @xmath8 for behavior @xmath1 .
@xmath71 are _ coupling parameters _ describing the influence of the behaviorial distribution within subpopulation @xmath22 on the behavior of subpoplation @xmath8 . for @xmath72 , @xmath71
reflects the _ social pressure _ of behavioral majorities .
* the _ game dynamical equations _
@xcite result for @xmath73 , @xmath61 , and @xmath74 ( cf .
their explicit form is + [ game ] + @xmath75 \\ & + & \nu_{aa}(t ) p_a({\mbox{\boldmath $ x$}},t ) \big [ e_a({\mbox{\boldmath $ x$}},t ) - \langle e_a \rangle \big ] \ , .\end{aligned}\ ] ] + whereas ( [ game]a ) again describes spontaneous behavioral changes ( _ `` mutations '' _ , innovations ) , ( [ game]b ) reflects competition processes leading to a _ `` selection '' _ of behaviors with a _ success _ @xmath76 that exceeds the _ average success _
@xmath77 the success @xmath76 is connected with the socalled _ payoff matrices _
@xmath78 by @xmath79 @xcite .
@xmath80 means the success of behavior @xmath1 with respect to the environment .
+ since the game dynamical equations ( [ game ] ) agree with the _ selection mutation equations _ @xcite they are not only a powerful tool in social sciences and economy @xcite , but also in evolutionary biology @xcite .
we shall now assume the set @xmath2 of possible behaviors to build a _ continuous _ space .
the @xmath81 dimensions of this space correspond to different characteristic _ aspects _ of the considered behaviors . in the continuous formulation ,
the sums in ( [ boltz ] ) , ( [ effrate ] ) have to be replaced by integrals : [ cont ] @xmath82 \nonumber \\ & = & \int d^n x ' \ , \big [ w^a[{\mbox{\boldmath $ x$}}'|{\mbox{\boldmath $ x$}}-{\mbox{\boldmath $ x$}}';t ] p_a({\mbox{\boldmath $ x$}}-{\mbox{\boldmath $ x$}}',t ) - w^a[{\mbox{\boldmath $ x$}}'|{\mbox{\boldmath $ x$}};t ] p_a({\mbox{\boldmath $ x$}},t ) \big ] \
, , \end{aligned}\ ] ] where @xmath83 : = w^a({\mbox{\boldmath $ x$}}'|{\mbox{\boldmath $ x$}};t ) : = w_a({\mbox{\boldmath $ x$}}'|{\mbox{\boldmath $ x$}};t ) + \sum_{b=1}^a \int\limits_\omega d^n y \int\limits_\omega d^n y ' \ , n_b \ , \widetilde{w}_{ab } ( { \mbox{\boldmath $ x$}}',{\mbox{\boldmath $ y$}}'|{\mbox{\boldmath $ x$}},{\mbox{\boldmath $ y$}};t ) p_b({\mbox{\boldmath $ y$}},t ) \ , .\ ] ] a reformulation of the boltzmann - like equations ( [ cont ] ) via a kramers - moyal expansion @xcite ( second order taylor approximation ) leads to a kind of _ diffusion equations _ : the socalled boltzmann - fokker - planck _ equations
_ @xcite [ bfp ]
@xmath84 + \frac{1}{2 } \sum_{i , j=1}^n \frac{\partial}{\partial x_{i}}\frac{\partial } { \partial x_{j } } \big[q_{a i j}({\mbox{\boldmath $ x$}},t ) p_a({\mbox{\boldmath $ x$}},t)\big ] \label{bfp}\ ] ] with the effective _ drift coefficients _ @xmath85 and the effective _ diffusion coefficients _ contains additional terms due to another derivation of ( [ bfp ] ) .
however , they make no contributions , since they result in vanishing surface integrals ( cf .
@xcite ) . ]
@xmath86 whereas the drift coefficients @xmath87 govern the systematic change of the distribution @xmath13 , the diffusion coefficients @xmath88 describe the spread of the distribution @xmath13 due to fluctuations resulting from the individual variation of behavioral changes . for _ ansatz _ ( [ concrates ] )
, the effective drift and diffusion coefficients can be splitted into contributions due to spontaneous ( or externally induced ) transitions ( @xmath89 ) , imitative processes ( @xmath90 ) , and avoidance processes ( @xmath91 ) : [ sum ] @xmath92 where @xmath93 and @xmath94 the behavioral changes induced by the _ environment _ are included in @xmath95 and @xmath96 .
the boltzmann - fokker - planck equations ( [ bfp ] ) are equivalent to the stochastic equations ( langevin equations ) [ langevin ] @xmath97 with @xmath98 g_{ajk}({\mbox{\boldmath $ x$}},t ) \label{drift}\ ] ] and @xmath99 ( cf .
@xcite ) . for an individual of subpopulation
@xmath8 the vector @xmath100 with the components @xmath101 describes the contribution to the change of behavior @xmath1 that is caused by behavioral fluctuations @xmath102 ( which are assumed to be delta - correlated and gaussian @xcite ) . since the diffusion coefficients @xmath88 and the coefficients @xmath103 are usually small quantities , we have @xmath104 ( cf .
( [ drift ] ) ) , and ( [ langevin ] ) can be put into the form @xmath105 whereas the fluctuation term describes individual behavioral variations , the vectorial quantity @xmath106 drives the systematic change of the behavior @xmath4 of individuals of subpopulation @xmath8 .
therefore , it is justified to denote @xmath106 as _ social force _ acting on individuals of subpopulation @xmath8 .
the social force influences the behavior of the individuals , but , conversely , due to interactions , the behavior of the individuals also influences the social force via the behavioral distributions @xmath13 ( cf .
( [ cont]b ) , ( [ effdr ] ) ) .
that means , @xmath107 is a function of the social processes within the given population . under the integrability conditions @xmath108
there exists a time - dependent _ potential _
@xmath109 so that the social force is given by its gradient @xmath110 : @xmath111 the potential @xmath112 can be understood as _ social field_. it reflects the social influences and interactions relevant for behavioral changes : the public opinion , trends , social norms , etc . clearly , the social force is no force obeying the newtonian laws of mechanics
. instead , the social force @xmath106 is a vectorial quantity with the following properties : * @xmath106 drives the temporal change @xmath3 of another vectorial quantity : the behavior @xmath4 of an individual of subpopulation @xmath8 . *
the component @xmath113 r^a({\mbox{\boldmath $ x$}}'|{\mbox{\boldmath $ x$}};t)\ ] ] of the social force @xmath106 describes the reaction of subpopulation @xmath8 on the behavioral distribution within subpopulation @xmath22 and usually differs from @xmath114 , which describes the influence of subpopulation @xmath8 on subpopulation @xmath22 . * neglecting fluctuations , the behavior @xmath4 does not change if @xmath106 vanishes .
@xmath115 corresponds to an _ extremum _ of the social field @xmath112 , because it means @xmath116 we can now formulate our results in the following form related to lewin s _ `` field theory '' _ @xcite : * let us assume that an individuals objective is to behave in an optimal way with respect to the social field , that means , he or she tends to a behavior corresponding to a _
minimum _ of the social field .
* if the behavior @xmath1 does not agree with a minimum of the social field @xmath112 this evokes a _
psychical tension ( force ) _
@xmath117 that is given by the gradient of the social field @xmath112 . *
the psychical tension @xmath106 is a vectorial quantity that induces a behavioral change according to @xmath118 * the behavioral change @xmath3 drives the behavior @xmath4 towards a minimum @xmath119 of the social field @xmath112 .
* when the behavior has reached a minimum @xmath119 of the social field @xmath112 , it holds @xmath120 and , therefore , @xmath115 , that means , the psychical tension vanishes . *
if the psychical tension @xmath106 vanishes , except for fluctuations no behavioral changes take place in accordance with ( [ accor ] ) . in the special case , where an individual s objective is the behavior @xmath119
, one would expect behavioral changes according to @xmath121 which corresponds to a social field @xmath122 with a minimum at @xmath119 .
examples for this case are discussed in @xcite .
note , that the social fields @xmath112 of different subpopulations
@xmath8 usually have _ different _ minima @xmath119 .
that means , individuals of different subpopulations @xmath8 will normally feel different psychical tensions @xmath106 .
this shows the psychical tension @xmath106 to be a _
`` subjective '' _ quantity .
in the following , the boltzmann - fokker - planck equations for behavioral changes will be illustrated by representative computer simulations .
we shall examine the case of @xmath123 subpopulations , and situations for which the interesting aspect of the individual behavior can be described by a certain _ position _
@xmath124 $ ] on a one - dimensional continuous scale ( i.e. , @xmath125 , @xmath126 ) .
then , the integrability conditions ( [ intcond ] ) are automatically fulfilled , and the social field @xmath127 is well - defined .
the parameter @xmath128 can be chosen arbitrarily .
we will take for @xmath128 the value that shifts the absolute minimum of @xmath129 to zero , that means , @xmath130 * since we will restrict the simulations to the case of imitative or avoidance processes , the shape of the social field @xmath129 changes with time only due to changes of the probability distributions @xmath131 ( cf .
( [ sum ] ) ) , that means , due to behavioral changes of the individuals ( see figures [ fi1 ] to [ fi6 ] ) . in the one - dimensional case
one can find the formal _ stationary solution _
@xmath132 which we expect to be approached in the limit of large times @xmath133 . due to the dependence of @xmath134 and @xmath135 on @xmath136 , equations ( [ abh ] ) are only _ implicit _ equations
. however , from ( [ abh ] ) we can derive the following conclusions : * if the diffusion coefficients are constant ( @xmath137 ) , ( [ abh ] ) simplifies to @xmath138 \right ) \
, , \ ] ] that means , the stationary solution @xmath136 is completely determined by the social field @xmath139 .
especially , @xmath136 has its maxima at the positions @xmath140 , where the social field @xmath139 has its minima ( see fig .
[ fi1 ] ) .
the diffusion constant @xmath141 regulates the width of the behavioral distribution @xmath136 . * if the diffusion coefficients @xmath135 are varying functions of the position @xmath142 , the behavioral distribution @xmath136 is also influenced by the concrete form of @xmath135 . from ( [ abh ] )
one expects high behavioral probabilities @xmath136 where the diffusion coefficients @xmath135 are small ( see fig .
[ fi2 ] , where the probability distribution @xmath143 can not be explained solely by the social field @xmath144 ) . *
since the stationary solution @xmath136 depends on both , @xmath134 and @xmath135 , different combinations of @xmath134 and @xmath135 can lead to the same probability distribution @xmath136 ( see fig .
[ fi4 ] in the limit of large times ) . for the following simulations , we shall use the _ ansatz _ [ ans ] @xmath145 for the readiness @xmath146 to change from @xmath142 to @xmath147 ( cf .
( [ util ] ) ) . with the utility function @xmath148
subpopulation @xmath8 prefers behavior @xmath149 .
@xmath150 means the _ indifference _ of subpopulation @xmath8 with respect to variations of the position @xmath142 .
moreover , we take @xmath151 where @xmath152 can be interpreted as measure for the _ range of interaction_. according to ( [ dist ] ) , the rate of behavioral changes is the smaller the greater they are
. only small changes of the position ( i.e. , between neighboring positions ) contribute with an appreciable rate .
let @xmath153 be the degree of _ sympathy _ which individuals of subpopulation @xmath8 feel towards individuals of subpopulation @xmath22 .
then , one expects the following : whereas the frequency @xmath41 of imitative processes will be increasing with @xmath153 , the frequency @xmath42 of avoidance processes will be decreasing with @xmath153 .
this functional relationship can , for example , be described by @xmath154 with @xmath155 @xmath156 is a measure for the frequency of imitative processes within subpopulation @xmath8 , @xmath157 a measure for the frequency of avoidance processes .
if we assume the sympathy between individuals of the same subpopulation to be be maximal , we have @xmath158 . in the following simulations of imitative processes we assume the prefered positions to be @xmath161 and @xmath162 . with @xmath163 the individuals of subpopulation @xmath63 like the individuals of subpopulation @xmath164 , but not the other way round .
that means , subpopulation 2 influences subpopulation 1 , but not vice versa .
one could say , the individuals of subpopulation 2 act as _
as expected , in both behavioral distributions @xmath131 there appears a maximum around the prefered behavior @xmath149 .
in addition , due to imitative processes of subpopulation 1 , a second maximum of @xmath165 develops around the prefered behavior @xmath166 of the trendsetters .
this second maximum is small , if the indifference @xmath167 of subpopulation 1 with respect to variations of the position @xmath142 is low ( see fig .
[ fi1 ] ) . for high values of the indifference @xmath167 even the _ majority
_ of individuals of subpopulation 1 imitates the behavior of the trendsetters ( see fig .
[ fi2 ] ) !
we shall now consider the case @xmath168 for which the subpopulations influence each other mutually with equal strengths .
if the indifference @xmath150 with respect to changes of the position @xmath142 is small in both subpopulations @xmath8 , _ each _ probability distribution @xmath131 has _ two _ maxima .
the higher maximum is located around the prefered position @xmath149 .
a second maximum can be found around the position prefered in the _ other _ subpopulation .
it is the higher , the greater the indifference @xmath150 is ( see fig .
[ fi3 ] ) . however , if @xmath150 exceeds a certain value in at least one subpopulation , only _
one _ maximum develops in each behavioral distribution @xmath131 ! despite the fact , that the social fields @xmath129 and diffusion coefficients @xmath169 of the subpopulations @xmath8 are different because of their different prefered positions @xmath149 ( and different utility functions @xmath170 ) , the behavioral distributions @xmath131 agree after some time ! especially , the maxima @xmath140 of the distributions @xmath131 are located at the _ same _ position @xmath171 in both subpopulations .
@xmath171 is nearer to the position @xmath149 of the subpopulation @xmath8 with the lower indifference @xmath150 ( see fig .
[ fi4 ] ) . for the simulation of avoidance processes we assume with @xmath174 and @xmath175 that both subpopulations nearly prefer the same behavior .
figure [ fi5 ] shows the case , where the individuals of different subpopulations dislike each other : @xmath176 this corresponds to a mutual influence of one subpopulation on the respective other .
the computational results prove : * the individuals avoid behaviors which can be found in the other subpopulation . *
the subpopulation @xmath63 with the lower indifference @xmath177 is distributed around the prefered behavior @xmath178 and pushes away the other subpopulation ! despite the fact that the initial behavioral distribution @xmath179 agrees in both subpopulations , there is nearly no overlapping of @xmath165 and @xmath180 after some time .
this is a typical example for _ polarization phenomena _ in the society . in figure [ fi6 ]
, we assume that the individuals of subpopulation 2 like the individuals of subpopulation 1 and , therefore , do not react on the behaviors in subpopulation 1 with avoidance processes : @xmath181 as a consequence , @xmath180 remains unchanged with time , whereas @xmath165 drifts away from the prefered behavior @xmath178 due to avoidance processes .
surprisingly , the polarization effect is much smaller than in figure [ fi5 ] !
the distributions @xmath165 and @xmath180 overlap considerably .
this is , because the slope of @xmath180 is smaller than in figure [ fi5 ] ( and remains constant ) . as a consequence
, the probability for an individual of subpopulation 1 to meet a disliked individual of subpopulation 2 with the same behavior @xmath142 can hardly be decreased by a small behavioral change .
one may conclude , that polarization effects ( which often lead to an escalation ) can be reduced , if individuals do not return dislike by dislike .
for practical purposes one has , of course , to determine the model parameters from empirical data
. therefore , let us assume to know empirically the distribution functions @xmath182 , [ the interaction rates @xmath183 and the effective transition rates @xmath184 ( @xmath185 ) for a couple of times @xmath186 .
the corresponding effective social fields @xmath187 and diffusion coefficients @xmath188 are , then , easily obtained as [ empi ] @xmath189 with @xmath190 and @xmath191 much more difficult is the determination of the utility functions @xmath192 ,
@xmath193 , the distance functions @xmath194 , and the rates @xmath195 , @xmath196 , @xmath197 .
this can be done by numerical minimization of the _ error function _
@xmath198 p_a^{\rm e}({\mbox{\boldmath $ x$}},t_l ) \right\}^2 \ , , \label{error}\ ] ] for example with the method of _ steepest descent _ @xcite . in ( [ error ] ) ,
we have introduced the abbreviation @xmath199
\mbox{e}^{u^a({x}',t_l ) - u^a({x},t_l ) } \ , .\ ] ] it turns out ( cf .
@xcite ) , that the rates @xmath200 have to be taken constant during the minimization process ( e.g. , @xmath201 ) , whereas the parameters @xmath202 , @xmath203 , @xmath204 and @xmath205 are to be varied . for @xmath206 one inserts [ dista ] @xmath207 with @xmath208 ^ 2 \ !
+ w_{\rm e}^a({\mbox{\boldmath $ x$}}|{\mbox{\boldmath $ x$}}';t_l ) g_a({\mbox{\boldmath $ x$}},{\mbox{\boldmath $ x$}}';t_l ) \big [ p_a^{\rm e}({\mbox{\boldmath $ x$}}',t_l ) \big]^2\ ] ] and @xmath209 ^ 2 \
! + \big [ g_a({\mbox{\boldmath $ x$}},{\mbox{\boldmath $ x$}}';t_l ) p_a^{\rm e}({\mbox{\boldmath $ x$}}',t_l ) \big]^2 \ , .\ ] ] ( [ dista ] ) follows from the minimum condition for @xmath210 ( cf .
@xcite ) .
since @xmath211 may have a couple of minima due to its nonlinearity , suitable start parameters have to be taken . especially , the numerically determined rates @xmath212 and @xmath213 have to be non - negative . if @xmath211 is minimal for the parameters @xmath202 , @xmath203 , @xmath210 , @xmath200 , @xmath212 and @xmath213 ,
this is ( as can easily be checked ) also true for the scaled parameters @xmath214 in order to obtain unique results we put @xmath215 and @xmath216 which leads to @xmath217 and @xmath218 @xmath219 and @xmath220 are _ mean utilities _ , whereas @xmath221 is a kind of _ unit of distance_. the distances @xmath222 are suitable quantities for _ multidimensional scaling _ @xcite .
they reflect the `` psychical structure '' ( psychical topology ) of individuals of subpopulation @xmath8 , since they determine which behaviors are more or less related ( compatible ) . by the dependence on @xmath8 , @xmath222 distinguishes different psychical structures resulting in different types @xmath8 of behavior and , therefore , different `` characters '' .
in this article , a behavioral model has been proposed that incorporates in a consistent way many models of social theory : the diffusion models , the multinomial logit model , lewin s field theory , the logistic equation , the gravity model , the weidlich - haag model , and the game dynamical equations .
this very general model opens new perspectives concerning a theoretical description and understanding of behavioral changes , since it is formulated fully mathematically .
it takes into account spontaneous ( or externally induced ) behavioral changes and behavioral changes due to pair interactions
. two important kinds of pair interactions have been distinguished : imitative processes and avoidance processes .
the model turns out to be suitable for computational simulations , but it can also be applied to concrete empirical data . the formulation of the model in the previous sections has neglected _ memory effects _ that may also influence behavioral changes . however , memory effects can be easily included by generalizing the boltzmann - like equations to @xmath225 \\ & & \qquad \quad \displaystyle + \frac{1}{2 } \sum_{i , j=1}^n \frac{\partial}{\partial x_{i}}\frac{\partial } { \partial x_{j } } \big[q_{a i j}^{t - t'}({\mbox{\boldmath $ x$}},t ' ) p_a({\mbox{\boldmath $ x$}},t')\big ] \bigg\ } \end{array } \label{membfp}\ ] ] obviously , in these formulas there only appears an additional integration over past times @xmath229 @xcite .
the influence of the past results in a dependence of @xmath230 , @xmath231 , and @xmath232 on @xmath233
. the boltzmann - like equations ( [ boltzlike ] ) resp . the boltzmann - fokker - planck equations ( [ bfp ] ) used in the previous sections result from ( [ memb ] ) resp .
( [ membfp ] ) in the markovian limit @xmath234 of short memory ( where @xmath235 is the dirac delta function ) .
the boltzmann - like equations ( [ boltzlike ] ) can also be used for the description of chemical reactions , where the states @xmath1 denote the different sorts of molecules ( or atoms ) , and @xmath8 distinguishes different isotopes or conformeres .
imitative and avoidance processes correspond in chemistry to self - activatory and self - inhibitory reactions .
although the concrete transition rates will be different from ( [ concrates ] ) , ( [ util ] ) in detail , there may be found analogous results for chemical reactions .
note , that the arrhenius formula for the rate of chemical reactions @xcite can be put into a form similar to ( [ util ] ) @xcite .
+ this work has been financially supported by the _
volkswagen stiftung _ and the _ deutsche forschungsgemeinschaft _ ( sfb 230 ) . the author is grateful to prof .
w. weidlich and dr .
r. reiner for valuable discussions and commenting on the manuscript .
99 r. zwanzig . on the identity of three generalized master equations .
_ physica _ * 30 * , 11091123 , 1964 . i. oppenheim , k. e. schuler and g. h. weiss , eds .
_ stochastic processes in chemical physics : the master equation_. mit press , cambridge , mass . , 1977 . h. haken .
_ laser theory_. springer , berlin , 1984 .
l. arnold and r. lefever , eds . _ stochastic nonlinear systems in physics , chemistry and biology_. springer , berlin , 1981
. w. weidlich & g. haag .
_ concepts and models of a quantitative sociology .
the dynamics of interacting populations_. springer , berlin , 1983 .
w. weidlich .
physics and social science the approach of synergetics . _ physics reports _ * 204 * , 1163 , 1991
. w. weidlich . the use of statistical models in sociology .
_ collective phenomena _ * 1 * , 51 - 59 , 1972 . w. weidlich & g. haag , eds . _
interregional migration_. springer , berlin , 1988 .
w. weidlich & g. haag . a dynamic phase transition model for spatial agglomeration processes
. _ journal of regional science _ * 27*(4 ) , 529569 , 1987 .
w. weidlich & m. munz . settlement formation , part i : a dynamic theory .
_ annals of regional science _ * 24 * , 83106 , 1990 . l. boltzmann .
_ lectures on gas theory_. university of california , berkeley , 1964 .
f. wilkinson _ chemical kinetics and reaction mechanisms_. van nostrand reinhold co. , new york , 1980 .
d. helbing .
interrelations between stochastic equations for systems with pair interactions . _
physica a _ * 181 * , 2952 , 1992 .
_ studies in human biology_. williams & wilkins , baltimore , 1924 .
p. f. verhulst .
bruxelles _ * 18 * , 1 , 1845 . g. k. zipf . the p1p2/d hypothesis on the intercity movement of persons
. _ american sociological review _ * 11 * , 677686 , 1946 . j. hofbauer & k. sigmund .
_ the theory of evolution and dynamical systems_. cambridge university press , cambridge , 1988 .
p. schuster , k. sigmund , j. hofbauer , r. wolff , r. gottlieb & ph .
selfregulation of behavior in animal societies i iii .
cybern . _ * 40 * , 125 , 1981 .
d. helbing .
_ stochastische methoden , nichtlineare dynamik und quantitative modelle sozialer prozesse_. phd thesis , university of stuttgart , 1992 , _ submitted to _
oldenbourg publishers , mnchen .
d. helbing . a fluid dynamic model for the movement of pedestrians .
submitted to _
complex systems_. d. helbing . a mathematical model for attitude formation by pair interactions .
_ behavioral science _ * 37 * , 190214 , 1992 .
e. osgood & p. h. tannenbaum .
the principle of congruity in the prediction of attitude change .
_ psychological review _ * 62 * , 4255 , 1955 . f. heider .
attitudes and cognitive organization . _ journal of psychology _ * 21 * , 107112 , 1946 . l. festinger . _
a theory of cognitive dissonance_. row & peterson , evanston , il , 1957 .
_ field theory in social science_. harper & brothers , new york , 1951 .
d. r. hartree .
cambridge philos .
* 24 * , 111 , 1928 .
d. montgomery .
brownian motion from boltzmann s equation . _
the physics of fluids _ * 14*(10 ) , 20882090 , 1971 .
d. helbing . a mathematical model for behavioral changes by pair interactions and its relation to game theory .
_ angewandte sozialforschung _ * 17 * ( 3/4 ) , 179194 , 1991/92 .
a. domencich & d. mcfadden .
_ urban travel demand .
a behavioral analysis _ , pp .
north - holland , amsterdam , 1975 .
j. de d. ortzar & l. g. willumsen .
_ modelling transport_. wiley , chichester , 1990 .
d. helbing .
stochastic and boltzmann - like models for behavioral changes , and their relation to game theory .
_ physica a _ , 1993 ( in press ) .
r. axelrod .
_ the evolution of cooperation_. basic books , new york , 1984
. j. von neumann & o. morgenstern .
_ theory of games and economic behavior_. princeton university press , princeton , 1944 .
the selforganization of matter and the evolution of biological macromolecules .
_ naturwissenschaften _ * 58 * , 465 , 1971 .
r. a. fisher . _
the genetical theory of natural selection_. oxford university press , oxford , 1930 .
m. eigen & p. schuster .
_ the hypercycle_. springer , berlin , 1979 .
r. feistel & w. ebeling .
_ evolution of complex systems_. kluwer academic , dordrecht , 1989 .
h. a. kramers .
_ physica _ * 7 * , 284 , 1940
. j. e. moyal .
_ j. r. stat .
soc . _ * 11 * , 151210 , 1949 .
d. helbing . a mathematical model for the behavior of pedestrians .
_ behavioral science _ * 36 * , 298310 , 1991 . g. e. forsythe , m. a. malcolm & c. b. moler .
_ computer methods for mathematical computations_. prentice hall , englewood cliffs , n.j .
, 1977 . j. b. kruskal & m. wish .
_ multidimensional scaling_. sage , beverly hills , 1978 .
f. w. young & r. m. hamer .
_ multidimensional scaling : history , theory , and applications_. lawrence erlbaum associates , hillsdale , n.j . , 1987 .
c. w. gardiner .
_ handbook of stochastic methods_. springer , berlin , 2nd edition , 1985 . | + it is shown , that the boltzmann - like equations allow the formulation of a very general model for behavioral changes .
this model takes into account spontaneous ( or externally induced ) behavioral changes and behavioral changes by pair interactions . as most important social pair interactions imitative and avoidance processes are distinguished .
the resulting model turns out to include as special cases many theoretical concepts of the social sciences . +
a kramers - moyal expansion of the boltzmann - like equations leads to the boltzmann - fokker - planck equations , which allows the introduction of `` social forces '' and `` social fields '' .
a social field reflects the influence of the public opinion , social norms and trends on behaviorial changes .
it is not only given by external factors ( the environment ) but also by the interactions of the individuals .
variations of the individual behavior are taken into account by diffusion coefficients .
@twoside -32pt |
strong replica consistency is an essential property for replication - based fault tolerant distributed systems .
it can be achieved via a number of different techniques . in this paper , we investigate the challenges in achieving integrity - preserving strong replica consistency and present our solutions for state - machine based byzantine fault tolerant systems @xcite . while it is widely known that strong replica consistency can also be achieved through the systematic - checkpointing technique @xcite for nondeterministic applications in the benign fault model , it is generally regarded as too expensive and it is not suitable for byzantine fault tolerance .
the state - machine based approach is one of the fundamental techniques in building fault tolerant systems @xcite . in this approach ,
replicas are assumed to be either deterministic or rendered - deterministic .
there has been a large body of work on how to render replicas deterministic in the presence of replica nondeterminism under the benign fault model ( _ e.g.,@xmath0 _ @xcite ) . however , when the replicas can be subject to byzantine faults , which is the case for many internet - based systems , most of the previous work is no longer effective .
furthermore , the determinism ( or rendered - determinism ) of the replicas is often considered harmful from the security perspective ( _ e.g.,@xmath0_with replication , an adversary can compromise any of the replicas to obtain confidential information @xcite ) and for many applications , their integrity is strongly dependent on the randomness of some of their internal operations ( _ e.g.,@xmath0_random numbers are used for unique identifier generation in transactional systems and for shuffling cards in online poker games , and if the randomness is taken away by a deterministic algorithm to ensure replica consistency , the identifiers or the hands of cards can be made predictable , which can easily lead to exploit @xcite ) .
this calls for new approaches towards achieving strong replica consistency while preserving the randomness of each replica s operations . in this paper
, we present two alternative approaches towards our goal . the first one is based on byzantine agreement @xcite ( referred to as the ba - algorithm in this paper ) and the other on a threshold coin - tossing scheme @xcite ( referred to as the ct - algorithm ) .
both approaches rely on a collective determination for decisions involving randomness , and the determination is based on the contributions made by a set of replicas ( at least one of which must be correct ) , to avoid the problems mentioned above .
they differ mainly by how the collective determination is carried out . in the ba - algorithm ,
the replicas first reach a byzantine agreement on the set of contributions from replicas , and then apply a deterministic algorithm ( for all practical purposes , the bitwise exclusive - or operation @xcite ) to compute the final random value .
the ct - algorithm uses the threshold coin - tossing scheme introduced in @xcite to derive the final random value , without the need of a byzantine agreement step .
even though the ct - algorithm saves on communication cost , it does incur significant computation overhead due to the cpu - intensive exponentiation calculations .
consequently , as we will show in section [ implsec ] , the ba - algorithm performs the best in a local - area network ( lan ) environment , where the ct - algorithm is more appropriate for the wide - area network ( wan ) environment where message passing is expensive .
furthermore , to ensure the freshness of the random numbers generated , the replicas using the ba - algorithm should have access to high entropy sources ( which is relatively easy to satisfy ) and the replicas should be able to refresh their key shares periodically in the ct - algorithm . for the latter , we envisage that a proactive threshold signature scheme could be used @xcite . however , the discussion of proactive threshold signature techniques is out of the scope of this paper . to summarize ,
we make the following research contributions in this paper : * we point out the danger and pitfalls of controlling replica randomness for the purpose of ensuring replica consistency . removing randomness from replica operations ( when it is needed ) could seriously compromise the system integrity .
* we propose the use of collective determination of random numbers contributed from replicas , as a practical way to reconcile the requirement of strong replica consistency and the preservation of replica randomness .
* we present a light - weight , byzantine agreement based algorithm to carry out the collective determination .
the ba - algorithm only introduces two additional communication steps because the byzantine agreement for the collective determination of random numbers can be integrated into that for message total ordering , as needed by the state - machine replication .
the ba - algorithm is particularly suited for byzantine fault tolerant systems operating in the lan environment , or where replicas are connected by high - speed low - latency networks .
* we further present an algorithm that uses the threshold coin - tossing scheme @xcite as an alternative method for collective determination of random numbers .
the coin - tossing scheme is introduced in @xcite as an instrumental mechanism for a group of replicas to reach byzantine agreement in asynchronous systems . to the best of our knowledge ,
our work is the first to show its usefulness in helping to ensure strong replica consistency without compromising the system integrity .
* we conduct extensive experiments , in both a lan testbed and an emulated wan environment , to thoroughly characterize the performance of the two approaches .
in this section , we introduce the system model for our work , and the practical byzantine fault tolerance algorithm ( bft algorithm , for short ) developed by castro and liskov @xcite as necessary background information .
byzantine fault tolerance refers to the capability of a system to tolerate byzantine faults .
it can be achieved by replicating the server and by ensuring that all server replicas reach an agreement on the total ordering of clients requests despite the existence of byzantine faulty replicas and clients .
such an agreement is often referred to as byzantine agreement @xcite . in recent several years , a number of efficient byzantine agreement algorithms @xcite have been proposed . in this work ,
we focus on the bft algorithm and use the same system model as that in @xcite .
the bft algorithm operates in an asynchronous distributed environment .
the safety property of the algorithm , _
i.e.,@xmath0_all correct replicas agree on the total ordering of requests , is ensured without any assumption of synchrony .
however , to guarantee liveness , _ i.e.,@xmath0_for the algorithm to make progress towards the byzantine agreement , certain synchrony is needed .
basically , it is assumed that the message transmission and processing delay has an asymptotic upper bound .
this bound is dynamically explored in the algorithm in that each time a view change occurs , the timeout for the new view is doubled .
the bft algorithm is executed by a set of @xmath1 replicas to tolerate up to @xmath2 byzantine faulty replicas .
one of the replicas is designated as the primary while the rest are backups .
each replica is assigned a unique i d @xmath3 , where @xmath3 varies from @xmath4 to @xmath5 . for view @xmath6 , the replica whose i d @xmath3 satisfies @xmath7 would serve as the primary
the view starts from 0 .
for each view change , the view number is increased by one and a new primary is selected .
the normal operation of the bft algorithm involves three phases . during the pre - prepare phase , the primary multicasts a pre - prepare message containing the client s request , the current view and a sequence number assigned to the request to all backups . a backup verifies the request and the ordering information .
if the backup accepts the pre - prepare message , it multicasts a prepare message containing the ordering information and the digest of the request being ordered .
this starts the prepare phase .
a replica waits until it has collected @xmath8 matching prepare messages from different replicas , and the pre - prepare message , before it multicasts a commit message to other replicas , which starts the commit phase .
the commit phase ends when a replica has collected @xmath9 matching commit messages from different replicas ( possibly including the one sent or would have been sent by itself ) . at this point ,
the request message has been totally ordered and it is ready to be delivered to the server application once all previous requests have been delivered .
all messages exchanged among the replicas , and those between the replicas and the clients are protected by an authenticator @xcite ( for multicast messages ) , or by a message authentication code ( mac ) ( for point - to - point communications ) .
an authenticator is formed by a number of macs , one for each target of the multicast .
we assume that the replicas and the clients each has a public / private key pair , and the public keys are known to everyone .
these keys are used to generate symmetric keys needed to produce / verify authenticators and macs . to ensure freshness ,
the symmetric keys are periodically refreshed by the mechanism described in @xcite .
we assume that the adversaries have limited computing power so that they can not break the security mechanisms described above .
furthermore , we assume that a faulty replica can not transmit the confidential state , such as the random numbers collectively determined , to its colluding clients in real time .
this can be achieved by using an application - level gateway , or a privacy firewall as described by yin et al.@xcite , to filter out illegal replies .
a compromised replica may , however , replace a high entropy source to which it retrieves random numbers with a deterministic algorithm , and convey such an algorithm via out - of - band or covert channels to its colluding clients .
in this section , we analyze a few well - known approaches possibly be used to ensure replica consistency in the presence of replica randomness .
we show that they are not robust against byzantine faulty replicas and clients . for replicas that use a pseudo - random number generator ,
they can be easily rendered deterministic by ensuring that they use the same seed value to initialize the generator .
one might attempt to use the sequence number assigned to the request as the seed . even though this approach is perhaps the most economical way to render replicas deterministic ( since no extra communication step is needed and no extra information is to be included in the control messages for total ordering of requests ) , it virtually takes the randomness away from the fault tolerant systems . in the presence of byzantine clients
, the vulnerability can be exploited to compromise the integrity of the system .
for example , a byzantine faulty client in an online poker game can simply try out different integer values as the seed to the pseudo - random generator ( if it is known to the client ) to guess the hands of the cards in the dealer and compare with the ones it has gotten .
the client can then place its bets accordingly and gain unfair advantage .
a seemingly more robust approach is to use the timestamp as the seed to the pseudo - random number generator .
as shown in @xcite , the use of timestamp does not offer more robustness to the system because it can also be guessed by byzantine faulty clients .
furthermore , the use of timestamp imposes serious challenges in asynchronous distributed systems because of the requirement that all replicas must use the same timestamp to seed the pseudo - random number generator . in @xcite
, a mechanism is proposed to handle this problem by asking the primary to piggyback its timestamp , to be used by backups as well , with the pre - prepare message . however , the issue is that the backups have very limited ways of verifying the timestamp proposed ( other than that the timestamp must be monotonically increasing ) without resorting to strong synchrony assumptions ( such as bounds on processing and message passing ) . the only option remaining seems to be the use of a truly random number to seed the pseudo - random number generator ( or to obtain random numbers entirely from a high entropy source ) .
we note that the elegant mechanism described in @xcite can not be used in this case because backups have no means to verify whether the number proposed by the primary is taken from a high - entropy source , or is generated according to a deterministic algorithm .
if the latter is the case , the byzantine faulty primary could continue colluding with byzantine faulty clients without being detected .
therefore , we believe the most effective way in countering such threats is to collectively determine the random number , based on the contributions from a set of replicas so that byzantine faulty replicas can not influence the final outcome .
the set size depends on the algorithms used , as we will show in the next two sections , but it must be greater than the number of faulty replicas tolerated ( @xmath2 ) by the system .
the normal operation of the ba - algorithm is illustrated in figure [ bafig ] . as can be seen , the collective - determination mechanism is seamlessly integrated into the original bft algorithm . on ordering a request ,
the primary determines the order of the request ( _ i.e.,@xmath0_assigns a sequence number to the request ) , and queries the application for the type of operation associated with the request .
if the operation involves with a random number as input , the primary activates the mechanism for the ba - algorithm .
the primary then obtains its share of random number by extracting from its own entropy source , and piggybacks the share with the pre - prepare message multicast to all backups .
the pre - prepare message has the form @xmath10pre - prepare,@xmath11@xmath12@xmath13 , where @xmath6 is the view number , @xmath14 is the sequence number assigned to the request , @xmath15 is the digest of the request , @xmath16 is the random number generated by the primary , and @xmath13 is the authenticator for the message . on receiving the pre - prepare message , a backup performs the usual chores such as the verification of the authenticator before it accepts the message .
it also checks if the request will indeed trigger a randomized operation , to prevent a faulty primary from putting unnecessary loads on correct replicas ( which could lead to a denial of service attack ) .
if the pre - prepare message is acceptable , the replica creates a pre - prepare certificate for storing the relevant information , generates a share of random number from its entropy source , and multicasts to all replicas a pp - update message , in the form @xmath10pp - update,@xmath17@xmath12@xmath18 , where @xmath3 is the sending replica identifier , @xmath19 is the random number contributed by replica @xmath3 .
when the primary has collected @xmath8 pp - update messages , it combines the random numbers received according to a deterministic algorithm ( referred to as the entropy combination step in figure [ bafig ] ) , and builds a pp - update message with slightly different content than those sent by backups . in the pp - update message sent by the primary ,
the @xmath19 component is replaced by a set of @xmath9 tuples containing the random numbers contributed by replicas ( possibly including its own share ) , @xmath20 .
each tuple has the form @xmath10@xmath21@xmath12 .
the replica identifier is included in the tuple to ease the verification of the set at backups . on receiving a pp - update message , a backup accepts the message and stores the message in its data structure provided that the message has a correct authenticator , it is in view @xmath6 and it has accepted a pre - prepare message to order the request with the digest @xmath15 and sequence number @xmath14 . a backup proceeds to the entropy combination step only if ( 1 ) it has accepted a pp - update message from the primary , and ( 2 ) @xmath8 pp - update messages sent by the replicas referenced in the set @xmath20 .
the backup requests a retransmission from the primary for any missing pp - update message .
after the entropy combination step is completed , a backup multicasts a prepare message in the form @xmath10prepare@xmath22@xmath12@xmath23 , where @xmath24 is the digest of the request concatenated by the combined random number . when a replica has completed the entropy combination step , and it has collected @xmath8 valid prepare messages from different replicas ( possibly including the message sent or would have been sent by itself ) , it multicasts to all replicas a commit message in the form @xmath10commit@xmath22@xmath12@xmath23 . when a replica receives @xmath9 valid commit messages , it decides on the sequence number and the collectively determined random number . at the time of delivery to the application , both the request and the random number
are passed to the application . in figure
[ bafig ] , the duration of the entropy extraction and combination steps have been intentionally exaggerated for clarify . in practice , the entropy combination can be achieved by applying a bitwise exclusive - or operation on the set of random numbers collected , which is very fast .
the cost of entropy extraction depends on the scheme used .
some schemes , such as the truerand method @xcite , allows very prompt entropy extraction .
truerand works by gathering the underlying randomness from a computer by measuring the drift between the system clock and the interrupts - generation rate on the processor .
the normal operation of the ct - algorithm is shown in figure [ ctfig ] .
the ct - algorithm is the same as the bft algorithm in the first two phases ( _ i.e.,@xmath0_pre - prepare and prepare phases ) .
the commit phase is modified by incorporating threshold coin - tossing operations .
most existing @xmath25 threshold signature schemes @xcite can be used for the ct - algorithm , where @xmath26 is the threshold number of signature shares needed to produce the group signature , and @xmath27 is the total number of players ( _ i.e.,@xmath0_replicas in our case ) participating the threshold signing . in most @xmath25 threshold signature schemes ,
a correct group signature can be derived by combining shares from @xmath28 players , where @xmath29 is the maximum number of corrupted players tolerated .
some schemes , such as the rsa - based scheme in @xcite , allow the flexibility of using up to @xmath30 as the minimum number of shares required to produce the group signature .
since @xmath27 in our work , @xmath26 can be set as high as @xmath9 .
this property offers additional protection against byzantine faulty replicas @xcite . at the beginning of the commit phase
, each replica generates its share of threshold signature by signing @xmath10@xmath31@xmath12 using its private key share , where @xmath15 is the digest of the request message and @xmath14 is the sequence number assigned to the request .
this operation is referred to as the share - generation step in figure [ ctfig ] .
the signature share is piggybacked with the commit message , in the form @xmath10commit@xmath32@xmath12@xmath23 , where @xmath33 is the replica @xmath3 s share of threshold signature .
when a replica has collected @xmath8@xmath34@xmath35 valid commit messages from different replicas , it executes the shares - combination step by combining @xmath26 threshold signature shares piggybacked with the commit messages .
after the shares have been combined into a group signature , it is mapped into a random number , first by hashing the group signature with a secure hash function ( _ e.g.,@xmath0_sha1 ) , and then by taking the first group of most significant bits from the hash according to the type of numbers needed ,
_ e.g.,@xmath0_32bits .
the random number will be delivered together with the request to the application , when all previous requests have been delivered .
in this section , we provide an informal argument on the correctness of our two algorithms .
the correctness criteria for the algorithms are : * all correct replicas deliver the same random number to the application together with the associated request , and * the random number is secure ( _ i.e.,@xmath0_it is truly random ) in the presence of up to @xmath2 byzantine faulty replicas .
we first argue for the ba - algorithm .
c1 is guaranteed by the use of byzantine agreement algorithm .
c2 is ensured by the collection of @xmath9 shares contributed by different replicas , and by a sound entropy combination algorithm ( _ e.g.,@xmath0_by using the bitwise exclusive - or operation on the set to produce the combined random number ) . by collecting @xmath9 contributions
, it is guaranteed that at least @xmath36 of them are from correct replicas , so faulty replicas can not completely control the set .
shares are all that needed for this purpose . however , collecting more shares is more robust in cases when some correct replicas use low - entropy sources .
this is analogous to the benefit of shoup s threshold signature scheme @xcite . ]
the entropy combination algorithm ensures that the combined random number is secure as long as at least one share is secure .
the bitwise exclusive - or operation could be used to combine the set and it is provably secure for this purpose @xcite .
therefore , the ba - algorithm satisfies both c1 and c2 .
next we argue for the ct - algorithm .
c1 is guaranteed by the following fact : ( 1 ) the same message ( @xmath10@xmath31@xmath12 ) is signed by all correct replicas , according to the ct - algorithm .
( 2 ) the threshold signature algorithm guarantees the production of the same group signature by combining @xmath26 shares .
different replicas could obtain different set of @xmath26 shares and yet they all lead to the same group signature .
( 3 ) the same secure hash function is used to hash the group signature .
c2 is guaranteed by the threshold signature algorithm . for the threshold signature algorithm used in our implementation
, its security is ensured by the random oracle model @xcite .
therefore , the ct - algorithm is correct as well .
this completes our proof .
the ba - algorithm and the ct - algorithm have been implemented and incorporated into a java - based bft framework .
the java - based bft framework is developed in house and it is ported from the c++ based bft framework of castro and liskov @xcite . due to space limitation , the details of the framework implementation is omitted .
the ct - algorithm uses shoup s threshold signature scheme @xcite , implemented by steve weis and made available at sourceforge @xcite .
the development and test platform consists of a group of dell sc440 servers each is equipped with a pentiumd processor of 2.8ghz and 1 gb of ram running suse 10.2 linux .
the nodes are connected via a 100mbps lan . as we noted earlier ,
the wan experiments are emulated by introducing artificial delays in communication , without injecting message loss .
to character the cost of the two algorithms , we use an echo application with fixed 1kb - long requests and replies .
the server is replicated at four nodes , and hence , @xmath37 in all our measurements .
up to 12 concurrent clients are launched across the remaining nodes ( at most one client per node ) .
each client issues consecutive requests without any think time . for the ct - algorithm
, we vary a number of parameters , including the threshold value and the key length .
we also experiment with certain optimizations .
for all measurements , the end - to - end latency is measured at the client and the throughput is measured at the replicas . the java system.nanotime ( ) api
is used for all timing - related measurements .
we first report the mean execution latency of basic cryptographic operations involved in the ba - algorithm and the ct - algorithm because such information is beneficial to the understanding of the behaviors we observe .
the latency cost is obtained when running a single client and 4 server replicas in the lan testbed .
the results are summarized in table [ cryptocost ] . as can be seen , the threshold signature operations are quite expensive , and it is impractical to use a key as large as 1024bit - long .
.execution time for basic cryptographic operations involved with our algorithms .
the data shown for ct signing is for a single share . [
cols="<,<,<",options="header " , ] without any optimization ( and without fault ) , an end - to - end remote call from a client to the replicated server using the original bft algorithm involves a total of 4 authenticator generation operations ( @xmath38 ) , 5 authenticator verification operations ( @xmath39 ) ( one does not need to verify the message sent by itself ) , 1 mac generation operation ( @xmath40 ) and 2 mac verification operation ( @xmath41 ) on the critical execution path ( _ i.e.,@xmath0_@xmath42 for request sending and receiving , @xmath42 for the pre - prepare phase , @xmath42 for the prepare phase , @xmath43 for the commit phase , and @xmath44 for the reply sending and receiving ) .
the ba - algorithm introduces two additional communication steps and 2 @xmath38 and 3 @xmath39 on the critical path .
the ct - algorithm does not require any additional communication step , but introduces 1 threshold signing operation ( @xmath45 ) and 1 operation for threshold shares verification and combination ( @xmath46 ) . from this analysis ,
the minimum end - to - end latency achievable using the ba - algorithm is @xmath47 ( a replica can proceed to the next step as soon as it receives 1 valid prepare message from other replica in the prepare phase , and 2 valid commit messages from other replicas in the commit phase , and the client can proceed to deliver the reply as soon as it has gotten 2 consistent replies ) .
similarly , the minimum latency using the ct - algorithm is @xmath48 .
based on the values given in table [ cryptocost ] , @xmath49 and @xmath50 for @xmath51 and 64bit - long key .
the minimum overhead incurred by the ba - algorithm is @xmath52 and that by the ct - algorithm is @xmath53 for @xmath51 and 64bit - long key .
figure [ lanperf ] shows the summary of the experimental results obtained in the lan testbed .
the end - to - end latency ( plotted in log - scale ) measured at a single client under various configurations is shown in figure [ lanperf](a ) . as a reference ,
the latency for the bft system without the additional mechanisms described in this paper is shown as `` base '' . in the figure ,
the result for the ba - algorithm is shown as `` ba '' , and the results for the ct - algorithm with different parameter settings are labeled as ct#-i , where # is the @xmath26 value , and @xmath3 is the key length . as can be seen , only if a very short key is used , the ct - algorithm incurs significant overhead .
furthermore , the observed end - to - end latency results are in - line with the analysis provided in the previous subsection .
the throughput measurement results shown in figure [ lanperf](b ) are consistent with those in the end - to - end latency measurements .
the results labeled with `` no batching '' are obtained for the original ct - algorithm described in section [ ctsec ] , _
i.e.,@xmath0_one coin - tossing operation ( _ i.e.,@xmath0_threshold share signing , combination and verification of @xmath26 shares ) is used for _ every _ request . those labeled with `` with batching '' are measured when the requests are batched ( for total ordering , they all share the same sequence number @xcite ) and only one coin - tossing operation is used for the entire batch of requests . as can be seen from figure [ lanperf](b )
, the gain in throughput is significant with the batching optimization .
however , if sharing the same random number among several requests is a concern , this optimization must be disabled . for the ba - algorithm ,
the communication steps for reaching a byzantine agreement on the set of random numbers are automatically batched together with that for requests total - ordering . batching the byzantine agreement for
a set of random numbers does not seem to introduce any vulnerability .
the additional optimization of one set of entropy extraction and combination per batch of requests does not have any noticeable performance benefit .
therefore , it is advised that this further optimization not to be considered in practice due to possible security concerns .
figure [ lanperf](c ) shows the end - to - end latency as a function of the load on the system in the presence of concurrent clients .
we use the system throughput as a metric for the system load because it better reflects the actual load on the system than the number of clients .
it is also useful to compare with the results in the wan experiments . as can be seen , for the ct - algorithm , without the batching optimization , the latency increases very sharply with the load , due to the cpu intensive threshold signature computations .
the results for the ct - algorithm with keys larger than 64bits are omitted in figure [ lanperf](b ) and ( c ) to avoid cluttering .
the throughput is significantly lower and the end - to - end latency is much higher than those of the ba - algorithm in these configurations , especially when the load is high .
the experimental results obtained in an emulated wan environment are shown in figure [ wanperf ] .
the observed metrics and the parameters used are identical to those in the lan experiments . as can be seen in figure [ wanperf](a ) , the end - to - end latency as perceived by a single client is similar for the ba - algorithm and the ct - algorithm with a key size up to 256bits ( for either @xmath51 or @xmath54 ) .
this can be easily understood because the end - to - end latency is dominated by the communication delays , as indicated by the end - to - end latency for the base system included in the figure .
figure [ wanperf](b ) shows part of the measurement results on system throughput under different number of concurrent clients . to avoid cluttering ,
only the results for @xmath51 and key sizes of up to 256bits are shown .
the throughput for the base system is included as a reference .
as can be seen , when batching for the coin - tossing operation is enabled , the ct - algorithm with short - to - medium sized keys out - performs the ba - algorithm .
when batching is disabled , however , the ba - algorithm performs better unless very small key is used for the ct - algorithm .
the end - to - end latency results shown in figure [ wanperf](c ) confirm the trend .
how to ensure strong replica consistency in the presence of replica nondeterminism has been of research interest for a long time , especially for fault tolerant systems using the benign fault model @xcite .
however , while the importance of the use of good random numbers has long been recognized in building secure systems @xcite , we have yet to see substantial research work on how to preserve the randomized operations necessary to ensure the system integrity in a fault tolerant system . for the type of systems where the use of random numbers is crucial to their service integrity ,
the benign fault model is obviously inadequate and the byzantine fault model must be employed if fault tolerance is required . in the recent several years
, significant progress has been made towards building practical byzantine fault tolerant systems , as shown in the series of seminal papers such as @xcite .
this makes it possible to address the problem of reconciliation of the requirement of strong replica consistency and the preservation of each replica s randomness for real - world applications that requires both high availability and high degree of security .
we believe the work presented in this paper is an important step towards solving this challenging problem .
we should note that some form of replica nondeterminism ( in particular , replica nondeterminism related to timestamp operations ) has been studied in the context byzantine fault tolerant systems @xcite .
however , we have argued in previous sections that the existing approach is vulnerable to the presence of colluding byzantine faulty replicas and clients . the main idea of this work , _
i.e.,@xmath0_collective determination of random values based on the contributions made by the replicas , is borrowed from the design principles for secure communication protocols @xcite . however , the application of this principle in solving the strong replica consistency problem is novel .
the ct - algorithm is inspired by the work of cachin , kursawe and shoup @xcite , in particular , the idea of exploiting threshold signature techniques for agreement .
however , we have adapted this idea to solve a totally different problem , _
i.e.,@xmath0_it is used towards reaching integrity - preserving strong replica consistency .
furthermore , we carefully studied what to sign for each request so that the final random number obtained is not vulnerable to attacks .
in this paper , we presented our work on reconciling the requirement of strong replica consistency and the desire of maintaining each replica s individual randomness .
based on the central idea of collective determination of random values needed by the applications for their service integrity , we designed and implemented two algorithms . the first one , the ba - algorithm , is based on reaching a byzantine agreement on a set of random number shares provided by @xmath9 replicas . the second one , the ct - algorithm , is based on threshold signature techniques .
we thoroughly characterized the performance of the two algorithms in both a lan testbed and an emulated wan environment .
we show that the ba - algorithm in general out - performs the ct - algorithm in most cases except in wan operations under relatively light load .
furthermore , the overhead incurred by the ba - algorithm with respect to the base bft system is relatively small , making it possible for practical use .
future research work will focus on the threshold key share refreshment issue for the ct - algorithm . to ensure long - term robustness of the system
, the key shares must be proactively refreshed periodically .
otherwise , the random numbers generated this way may age over time , which may open the door for attacks . the threshold signature algorithm used in this work @xcite does not have built - in mechanism for key share refreshment .
we will explore other threshold signature algorithms that offer this capability @xcite .
j. slember and p. narasimhan . living with nondeterminism in replicated middleware applications . in _ proceedings of the acm / ifip / usenix 7th international middleware conference _ , pages 81100 , melbourne , australia , 2006 .
j. yin , j .-
martin , a. venkataramani , l. alvisi , and m. dahlin . separating agreement from execution for byzantine fault tolerant services . in _ proceedings of the acm symposium on operating systems principles _ , pages 253267 , bolton landing , ny , usa , 2003 . | strong replica consistency is often achieved by writing deterministic applications , or by using a variety of mechanisms to render replicas deterministic .
there exists a large body of work on how to render replicas deterministic under the benign fault model .
however , when replicas can be subject to malicious faults , most of the previous work is no longer effective .
furthermore , the determinism of the replicas is often considered harmful from the security perspective and for many applications , their integrity strongly depends on the randomness of some of their internal operations .
this calls for new approaches towards achieving replica consistency while preserving the replica randomness . in this paper
, we present two such approaches .
one is based on byzantine agreement and the other on threshold coin - tossing .
each approach has its strength and weaknesses .
we compare the performance of the two approaches and outline their respective best use scenarios .
* keywords * : replica consistency , byzantine fault tolerance , middleware , threshold signature , coin - tossing |
the robertson - seymour graph minors project has shown that minor - closed classes of graphs can be described by finitely many forbidden minors . excluding a small number of minors can give graph classes with interesting properties .
the first such result was wagner s demonstration @xcite that planar graphs are precisely the graphs that are @xmath14- and @xmath15-minor - free .
excluding certain special classes of graphs as minors seems to give close connections to other graph properties .
one of the most important open problems at present is hadwiger s conjecture , which relates excluded complete graph minors to chromatic number .
our interest is in excluding complete bipartite graphs as minors .
together with connectivity conditions , and possibly other assumptions , graphs with no @xmath16 as a minor can be shown to have interesting properties relating to toughness , hamiltonicity , and other traversability properties .
the simplest result of this kind follows from a well - known consequence of wagner s characterization of planar graphs .
this consequence says that @xmath6-connected @xmath17-minor - free graphs are outerplanar or @xmath5 ; hence , they are hamiltonian .
for some recent examples of this type of result , involving toughness , circumference , and spanning trees of bounded degree , see .
our work was originally motivated by trying to find forbidden minor conditions to make @xmath1-connected planar graphs , or @xmath1-connected graphs more generally , hamiltonian . in examining the hamiltonicity of @xmath1-connected @xmath0-minor - free graphs
we were led to a complete picture of their structure , which we then extended to @xmath0-minor - free graphs in general . using this ,
we show in section [ consequences ] that @xmath1-connected @xmath0-minor - free graphs are hamiltonian , and that @xmath6-connected @xmath0-minor - free graphs have hamilton paths . for @xmath0-minor - free graphs , or @xmath18-minor - free graphs in general
, there are a number of previous results .
dieng and gavoille ( see dieng s thesis @xcite ) showed that every @xmath6-connected @xmath0-minor - free graph contains two vertices whose removal leaves the graph outerplanar .
streib and young @xcite used dieng and gavoille s result to show that the dimension of the minor poset of a connected graph @xmath19 with no @xmath0 minor is polynomial in @xmath20 .
chen et al .
@xcite proved that @xmath6-connected @xmath18-minor - free graphs have a cycle of length at least @xmath21 .
myers @xcite proved that a @xmath18-minor - free graph @xmath19 with @xmath22 satisfies @xmath23 ; more recently chudnovsky , reed and seymour @xcite showed that this is valid for all @xmath24 , and provided stronger bounds for @xmath6- , @xmath1- and @xmath25-connected graphs .
our results improve their bound for @xmath1-connected graphs when @xmath26 .
an unpublished paper of ding @xcite proposes that @xmath18-minor - free graphs can be built from slight variations of outerplanar graphs and graphs of bounded order by adding ` strips ' and ` fans ' using an operation that is a variant of a @xmath6-sum ( and which corresponds to the idea of replacing subdividable sets of edges that is used later in this paper ) .
ding s result involves subgraphs that have @xmath0 minors , and so not all aspects of his structure can be present in the case of @xmath0-minor - free graphs ; our results illuminate the extent to which ding s structure still holds . as part of our work
we use rooted minors , where particular vertices of @xmath19 must correspond to certain vertices of @xmath27 when we find @xmath27 as a minor in @xmath19 .
for example , robertson and seymour @xcite characterized all @xmath1-connected graphs that have no @xmath17 minor rooted at the three vertices on one side of the bipartition .
fabila - monroy and wood @xcite characterized graphs with no @xmath5 minor rooted at all four vertices .
demasi @xcite characterized all @xmath1-connected planar graphs with no @xmath0 minor rooted at the four vertices on one side of the bipartition . in this paper
we characterize all graphs with no @xmath8 minor rooted at two vertices on one side of the bipartition .
this result is useful not only here , but also in the authors proof that @xmath1-connected @xmath28-minor - free planar graphs are hamiltonian ( see @xcite ) .
we begin with some definitions and notation .
all graphs are simple .
we use ` @xmath29 ' to denote set difference and deletion of vertices from a graph , ` @xmath30 ' to denote deletion of edges , ` @xmath31 ' to denote contraction of edges , and ` @xmath32 ' to denote both addition of edges and join of graphs .
since we work with simple graphs , when we contract an edge any parallel edges formed are reduced to a single edge .
a graph @xmath27 is a _
minor _ of a graph @xmath19 if @xmath27 is isomorphic to a graph formed from @xmath19 by contracting and deleting edges of @xmath19 and deleting isolated vertices of @xmath19 .
we delete multiple edges and loops , so all minors are simple . another way to think of
a @xmath33-vertex minor @xmath27 of @xmath19 is as a collection of disjoint subsets of the vertices of @xmath19 , @xmath34 where each @xmath35 corresponds to a vertex @xmath36 of @xmath27 , where @xmath37 $ ] ( the subgraph of @xmath19 induced by the vertex set @xmath35 ) is connected for @xmath38 , and for each edge @xmath39 there is at least one edge between @xmath35 and @xmath40 in @xmath19 .
we call this a _ model _ of @xmath27 in @xmath19
. we will often identify minors in graphs by describing the sets @xmath34 .
the set @xmath35 is known as the _ branch set _ of @xmath36 , and may be thought of as the set of vertices in @xmath19 that contracts to @xmath36 in @xmath27 .
suppose we are given @xmath41 , @xmath42 , and a bijection @xmath43 .
we say that a model of @xmath27 in @xmath19 is a _ minor rooted at @xmath44 in @xmath19 and at @xmath45 in @xmath27 by @xmath46 _ if each @xmath47 belongs to the branch set of @xmath48 .
if the symmetric group on @xmath45 is a subgroup of the automorphism group of @xmath27 ( as it will be in our case ) then the exact bijection @xmath46 between @xmath44 and @xmath45 does not matter .
a graph is _ @xmath27-minor - free _ if it does not contain @xmath27 as a minor .
@xmath33-separation _ in a graph @xmath19 is a pair @xmath49 of edge - disjoint subgraphs of @xmath19 with @xmath50 , @xmath51 , @xmath52 , and @xmath53 .
suppose @xmath18 has bipartition @xmath54 .
let @xmath55 and @xmath56 be the branch sets of @xmath57 and @xmath58 in a model of @xmath18 in a graph @xmath19 .
suppose @xmath59 is the branch set of @xmath60 for some @xmath61 .
then there is a path @xmath62 , @xmath63 , with @xmath64 , @xmath65 , and @xmath66 for @xmath67 .
let @xmath68 and let @xmath69 .
we can replace @xmath59 with @xmath70 and @xmath71 with @xmath72 and still have a model of @xmath18 ( possibly using fewer vertices of @xmath19 than before ) . hence without loss of generality we may assume that the branch set of each vertex @xmath60 , @xmath73 , contains a single vertex @xmath74 .
let @xmath75 .
we say @xmath76 represents a _ standard @xmath18 minor_. observe that @xmath19 contains a @xmath18 minor if and only if @xmath19 contains a standard @xmath18 minor .
note that the standard model also applies to @xmath18 minors rooted at two vertices corresponding to @xmath57 and @xmath58 .
a _ wheel _ is a graph @xmath77 with @xmath78 .
a vertex of degree @xmath79 in @xmath80 is a _
hub _ and its incident edges are _ spokes _ while the remaining edges form a cycle called the _
rim_. in @xmath81 every vertex is a hub and every edge is both a spoke and a rim edge , but in @xmath80 for @xmath82 there is a unique hub and the edges are partitioned into spokes and rim edges .
note that we identify wheels by their number of vertices , rather than their number of spokes .
a graph is _ outerplanar _ if it has an _ outerplane embedding _ , an embedding in the plane with every vertex on the outer face .
in the next section , we define a class of graphs and describe several small examples which together make up all @xmath1-connected @xmath0-minor - free graphs .
we begin with @xmath1-connected graphs because all @xmath83-connected graphs on at least six vertices have a @xmath0 minor .
this is obvious for complete graphs . otherwise , a pair of nonadjacent vertices and the four internally disjoint paths between them guaranteed by menger s theorem yield a @xmath0 minor . in section
[ 2connected ] we extend the characterization to @xmath6-connected graphs .
the generalization to all graphs follows because a graph that is not @xmath6-connected is @xmath0-minor - free if and only if each of its blocks is @xmath0-minor - free .
section [ consequences ] presents applications of our characterization to hamiltonicity , topological properties , counting , and edge bounds .
all graphs @xmath19 with @xmath84 are trivially @xmath0-minor - free ; the @xmath1-connected ones are @xmath14 , @xmath85 , @xmath86 , and @xmath87 . for @xmath88 , first we define a class of graphs and identify those that are @xmath1-connected and @xmath0-minor - free .
we then look at some small graphs that do not fit into this class .
finally , we show that every @xmath1-connected @xmath0-minor - free graph is one of these we have described . for @xmath90 and @xmath91 ,
let @xmath92 consist of a spanning path @xmath93 , which we call the _ spine _ , and edges @xmath94 for @xmath95 and @xmath96 for @xmath97 .
the graph @xmath98 is @xmath99 ; we call @xmath100 the _ plus edge_. all graphs @xmath101 are planar . the graph
@xmath102 is a wheel @xmath80 with hub @xmath103 .
examples are shown in figure [ fig : g_p , r , sm ] . since @xmath104
we often assume @xmath105 .
[ fig : g_p , r , sm ] in the following three lemmas we first determine when a graph @xmath89 is @xmath1-connected , and then when it is @xmath0-minor - free . for @xmath106
, @xmath107 is @xmath1-connected if and only if ( i ) @xmath108 , @xmath109 , and the plus edge is present ( or symmetrically @xmath110 , @xmath111 , and the plus edge is present ) or ( ii ) @xmath112 and @xmath113 .
[ lem : g3conn ] assume that @xmath114 . to prove the forward direction ,
assume @xmath19 is @xmath1-connected and first suppose @xmath108 .
if the plus edge is not present , then @xmath115 has degree @xmath6 and @xmath116 is a @xmath6-cut .
similarly if @xmath117 , then @xmath118 has degree @xmath6 and @xmath119 is a @xmath6-cut .
next suppose @xmath112 . if @xmath120 , then there is necessarily a degree @xmath6 vertex @xmath36 with @xmath121 and hence a @xmath6-cut in @xmath19 .
to prove the reverse direction , assume ( i ) or ( ii ) .
if ( i ) holds , @xmath19 is a wheel , which is @xmath1-connected , so we may assume that ( ii ) holds .
to show @xmath1-connectedness we find three internally disjoint paths between each possible pair of vertices . for @xmath115 and @xmath103
we have paths @xmath122 , @xmath123 , and @xmath124 ( where possibly @xmath125 ) .
next suppose that only one of @xmath115 and @xmath103 is in the considered pair , say @xmath115 without loss of generality .
first consider @xmath115 and @xmath36 where @xmath126 .
when @xmath127 , then the three disjoint paths are @xmath128 , @xmath129 , and @xmath130 .
when @xmath131 , then @xmath132 and @xmath133 and the three disjoint paths are @xmath134 , @xmath135 , and @xmath129 .
now consider @xmath115 and @xmath36 where @xmath136 .
then the three disjoint paths are @xmath128 , @xmath137 , and @xmath138 .
finally consider @xmath36 and @xmath139 where @xmath140 and @xmath141 .
if @xmath36 and @xmath139 are both adjacent to the same end vertex , say @xmath115 , where @xmath142 , then the three disjoint paths are @xmath143 , @xmath144 , and @xmath145 .
otherwise the three disjoint paths are @xmath143 , @xmath146 , and @xmath147 . for @xmath148 , @xmath149 is @xmath0-minor - free if and only if @xmath150 .
[ lem : gk24 ] to prove the forward direction , suppose @xmath151 .
then there are vertices @xmath36 and @xmath152 such that both @xmath115 and @xmath103 are adjacent to both @xmath36 and @xmath152 and @xmath153 . then there is a standard @xmath0 minor @xmath76 in @xmath19 : let @xmath154 , @xmath155 , and @xmath156 . now suppose that @xmath150 .
we claim that if @xmath19 has a standard @xmath0 minor @xmath76 , then @xmath64 and @xmath157 ( or vice versa ) .
the graph @xmath158 is outerplanar and thus has no @xmath17 minor . therefore ,
if @xmath19 has a @xmath0 minor , then it must include @xmath115 .
we can not have @xmath159 because then the outerplanar graph @xmath158 would have a @xmath17 minor . by symmetry , @xmath103
must also be included in the minor and @xmath160 .
if @xmath161 , then @xmath162 has a @xmath163 minor , but @xmath162 is a path and there is no @xmath163 minor in a path .
the only remaining possibility is @xmath64 and @xmath164 ( or vice versa ) .
let @xmath165 denote the set of neighbors of @xmath166 .
let @xmath167 and @xmath168 , which intersect only if @xmath169 .
suppose @xmath19 has a standard @xmath0 minor @xmath76 .
then by the claim proved in the previous paragraph , @xmath64 and @xmath157 .
we consider the makeup of @xmath44 .
suppose @xmath170 , in that order along the spine .
since @xmath171 separates @xmath172 and @xmath103 , and @xmath157 , we can not have @xmath71 adjacent to @xmath172 , which is a contradiction .
thus @xmath173 .
symmetrically , @xmath174 .
we must have @xmath175 and @xmath176 in the order @xmath177 along the spine .
since @xmath157 , there must be a @xmath178-path in @xmath179 , and hence @xmath180 .
then @xmath181 is a cutvertex separating @xmath103 and @xmath172 in @xmath179 , so @xmath182 .
now there must also be a @xmath183-path in @xmath184 but no such path exists .
thus there is no @xmath0 minor .
define @xmath185 to be the set of ( labeled ) graphs of the form @xmath101 that are both @xmath1-connected and @xmath0-minor - free .
of the four @xmath1-connected graphs on fewer than six vertices , three are planar , and all three belong to @xmath185 : @xmath186 , @xmath187 , and @xmath188 . from this and lemmas [
lem : g3conn ] and [ lem : gk24 ] we get @xmath189 let @xmath190 denote the class of all graphs isomorphic to a graph in @xmath185 . note that graphs in @xmath185 are @xmath1-sums of two wheels , a fact we will see in more detail later on . there are some isomorphisms between graphs in @xmath185 and also symmetries within certain graphs of the class .
let @xmath191 be the involution with @xmath192 for @xmath193 .
then @xmath194 provides the isomorphism ( in both directions ) between @xmath101 and @xmath195 that we have already noted ; if @xmath196 it is an automorphism .
the graph @xmath197 is isomorphic to @xmath80 , with @xmath103 as a hub .
it has the obvious symmetries .
[ fig : isomorph ] define @xmath198 to be the involution fixing @xmath199 and @xmath103 and with @xmath200 for @xmath201 .
then @xmath202 is an automorphism of @xmath203 , an isomorphism ( in both directions ) between @xmath204 and @xmath205 , and an automorphism of @xmath206 .
the case @xmath207 is illustrated in figure [ fig : isomorph ] , where @xmath208 corresponds to reflection about a vertical axis .
the graph without the dashed edges @xmath209 and @xmath210 is @xmath211 . with the edge @xmath209 , the graph is @xmath212 and with @xmath210 ,
the graph is @xmath213 . with both edges @xmath209 and @xmath210 , the graph is @xmath214 . in general
@xmath202 maps the spine @xmath215 to the path @xmath216 .
for @xmath101 with @xmath217 we call @xmath218 the _ second spine_. when @xmath219 we have a similar involution @xmath220 , and the path @xmath221 can be regarded as an extra spine .
when @xmath222 , @xmath223 is the image of @xmath218 under the automorphism @xmath194 . finally , besides some obvious special symmetries when @xmath224 or @xmath25 , @xmath225 is vertex - transitive and is isomorphic to the triangular prism .
these symmetries and isomorphisms will be important later , particularly in section 3 when we discuss which edges of @xmath226 can be subdivided without creating a @xmath0 minor .
up to isomorphism the class @xmath185 contains one @xmath83-vertex graph and @xmath2 @xmath3-vertex graphs for each @xmath4 .
we now examine the effect of deleting or contracting a single edge of a graph in @xmath185 .
[ lem : gdeledge ] suppose @xmath227 and @xmath228 .
the following are equivalent .
\(i ) @xmath229 .
\(ii ) @xmath230 is @xmath1-connected .
\(iii ) @xmath19 is not a wheel and either @xmath231 is a plus edge , or @xmath232 and @xmath233 .
clearly ( iii ) @xmath234 ( i ) @xmath234 ( ii ) .
if ( iii ) does not hold then @xmath230 has at least one vertex of degree @xmath6 , so ( ii ) does not hold ; thus ( ii ) @xmath234 ( iii ) .
.[tbl : contract ] : contracting an edge @xmath231 in @xmath235 with @xmath236 [ cols="<,<,^,^",options="header " , ] all remaining small graphs are covered by table [ table ] . verifying these results is straightforward ; complete proofs may be found in ( * ? ? ?
* section 5.2 ) .
these results were also confirmed by computer ( the program may be obtained from the first author ) .
the dashed edges in the table indicate edges present in one graph but not the other .
for example , in the row for @xmath237 and @xmath238 , the dashed edge is present in @xmath238 but not @xmath237 .
the maximal subdividable sets of edges for the nine small cases not in @xmath185 as well as @xmath239 , @xmath240 , @xmath241 , @xmath225 , @xmath242 and @xmath243 are listed in table [ table ] .
[ lem : subsmall ] as mentioned earlier , a graph @xmath19 is @xmath0-minor - free if and only if each of its blocks is @xmath0-minor - free , so our overall result can now be stated as follows .
a graph is @xmath0-minor - free if and only if each of its blocks is described by theorem [ lem:2connclass ] , where for theorem [ lem:2connclass ] ( iii ) , the @xmath1-connected graphs are given in theorem [ thm : main24 ] and the subdividable sets are described in theorem [ lem : subg ] and lemma [ lem : subsmall ] .
[ thm : combin ]
our characterization has a number of consequences .
first , as mentioned in the introduction , we are interested in hamiltonian properties of @xmath0-minor - free graphs .
\(ii ) a @xmath6-connected graph described by theorem [ lem:2connclass](ii ) is planar and has no closed spanning trail ; the simplest example is @xmath17 .
( it is also possible to construct examples using theorem [ lem:2connclass](iii ) . )
\(iii ) define a _
hamilton base _ in a graph to be a hamilton path extended by a new edge at one or both ends , i.e. , a trail of the form @xmath246 , @xmath247 , or @xmath248 , where @xmath249 is a hamilton path .
if @xmath59 is a hamilton base in a graph @xmath250 and @xmath251 is obtained from @xmath250 by subdividing the elements of a subset of @xmath252 arbitrarily many times , we observe that @xmath251 has a hamilton path .
now consider a @xmath6-connected @xmath0-minor - free graph @xmath19 .
if @xmath19 is described by theorem [ lem:2connclass](i ) then @xmath19 is hamiltonian .
suppose @xmath19 is described by theorem [ lem:2connclass](iii ) , as constructed from a @xmath1-connected @xmath0-minor - free graph @xmath250 by replacing each edge of a subdividable set @xmath253 by an @xmath13-outerplanar graph
. then @xmath19 has a spanning subgraph @xmath251 which is obtained from @xmath250 by subdividing each edge of @xmath44 some number of times .
if @xmath250 has a hamilton base containing @xmath44 , then @xmath251 , and hence @xmath19 , has a hamilton path .
so we just need to verify that each maximal subdividable set of edges in @xmath250 is contained in a hamilton base .
each subdividable set from theorem [ lem : subg ] itself forms a hamilton base , and it is not difficult to show that the subdividable sets from lemma [ lem : subsmall ] ( table [ table ] ) are contained in hamilton bases ; we omit the details .
finally , if @xmath19 is described by theorem [ lem:2connclass](ii ) then @xmath19 has a spanning subgraph @xmath251 obtained by subdividing edges of @xmath254 , and @xmath17 has a hamilton base containing all its edges , so a similar argument applies .
consider the structure of @xmath19 as described in theorem [ lem:2connclass ] .
if ( i ) holds no vertices need to be deleted , and if ( ii ) holds then one of @xmath9 or @xmath10 can be deleted . to verify the result
when ( iii ) holds , it suffices to show that for every @xmath1-connected @xmath0-minor - free @xmath250 and every maximal subdividable set of edges @xmath259 in @xmath250 , there is @xmath260 with @xmath261 ( @xmath257 if @xmath250 is planar ) so that @xmath262 has an outerplane embedding with all remaining edges of @xmath259 ( those not incident with @xmath263 ) on the outer face .
if @xmath264 is covered by theorem [ lem : subg ] then @xmath265 always works .
the result must be checked for the small graphs in table [ table ] .
the @xmath1-connected graphs described in theorem [ thm : main24 ] are planar or minors of @xmath238 , and it is not difficult to find toroidal and projective - planar embeddings of @xmath238 . for connectivity @xmath6 the graphs
@xmath19 constructed in theorem [ lem:2connclass ] are either planar or have the same genus as some @xmath1-connected @xmath0-minor - free graph @xmath250 .
note that corollary [ cor : genus ] does not follow from dieng and gavoille s result , corollary [ cor : dg ] , since a result of mohar @xcite implies that graphs which become outerplanar after deleting two vertices can have arbitrarily high ( orientable ) genus .
fourth , our result shows that the number of @xmath1-connected @xmath0-minor - free graphs grows only linearly .
for @xmath268 the only such @xmath3-vertex graphs are those in @xmath190 , and there are only @xmath2 nonisomorphic such graphs .
although we have not done so , it should also be possible to deduce counting results for @xmath6-connected @xmath0-minor - free graphs from our characterization .
finally , chudnovsky , reed and seymour @xcite showed that the number of edges in a @xmath1-connected @xmath18-minor - free graph is at most @xmath269 .
they provide examples to show that this is in a sense best possible for @xmath270 .
theorem [ thm : main24 ] shows that this can be improved when @xmath26 . using theorem [ lem:2connclass ]
we can also obtain a result for @xmath6-connected @xmath0-minor - free graphs .
we omit the straightforward proofs , which use the fact that an @xmath3-vertex outerplanar graph has at most @xmath271 edges .
\(i ) every @xmath1-connected @xmath0-minor - free @xmath3-vertex graph with @xmath272 has at most @xmath273 edges , and such graphs with @xmath273 edges exist for all @xmath274 .
( @xmath14 has @xmath275 edges , and @xmath276 has @xmath277 edges . )
\(ii ) every @xmath6-connected @xmath0-minor - free @xmath3-vertex graph with @xmath278 has at most @xmath277 edges , and such graphs with @xmath277 edges exist for all @xmath279 .
( @xmath14 has @xmath275 edges . )
ruy fabila - monroy and david r. wood , rooted @xmath5-minors , no . 2 ( 2013 ) # p64 ( 19 pages ) .
emily a. marshall , hamiltonicity and structure of classes of minor - free graphs , ph.d .
dissertation , vanderbilt university , may 2014 .
noah streib and stephen j. young , dimension and structure for a poset of graph minors .
( downloaded 22 may 2014 ) k. wagner , ber eine eigenschaft der ebenen komplexe , 114 ( 1937 ) 570590 . | we provide a complete structural characterization of @xmath0-minor - free graphs .
the @xmath1-connected @xmath0-minor - free graphs consist of nine small graphs on at most eight vertices , together with a family of planar graphs that contains @xmath2 nonisomorphic graphs of order @xmath3 for each @xmath4 as well as @xmath5 . to describe the @xmath6-connected @xmath0-minor - free graphs we use _
@xmath7-outerplanar _ graphs , graphs embeddable in the plane with a hamilton @xmath7-path so that all other edges lie on one side of this path .
we show that , subject to an appropriate connectivity condition , @xmath7-outerplanar graphs are precisely the graphs that have no rooted @xmath8 minor where @xmath9 and @xmath10 correspond to the two vertices on one side of the bipartition of @xmath8 .
each @xmath6-connected @xmath0-minor - free graph is then ( i ) outerplanar , ( ii ) the union of three @xmath7-outerplanar graphs and possibly the edge @xmath7 , or ( iii ) obtained from a @xmath1-connected @xmath0-minor - free graph by replacing each edge @xmath11 in a set @xmath12 satisfying a certain condition by an @xmath13-outerplanar graph . from our characterization
it follows that a @xmath0-minor - free graph has a hamilton cycle if it is @xmath1-connected and a hamilton path if it is @xmath6-connected .
also , every @xmath6-connected @xmath0-minor - free graph is either planar , or else toroidal and projective - planar . |
astrophysical gravitational waves ( gws ) affect long - term timing measurements of radio pulsars @xcite .
gw - induced metric perturbations at the earth cause variations in pulse arrival times that differ between pulsars only by geometric factors .
hence , a specific gw signal may be directly detected in contemporaneous timing measurements of multiple pulsars .
three such ` pulsar timing arrays ' ( ptas ; e.g. , * ? ? ?
* ) are currently in operation : the european pulsar timing array ( epta ; * ? ? ?
* ) , the north american nanohertz observatory for gravitational waves ( nanograv ; * ? ? ?
* ) , and the parkes pulsar timing array ( ppta ; * ? ? ?
these groups also share data as part of the international pulsar timing array ( ipta ; * ? ? ?
* ) consortium .
currently , at least 50 millisecond pulsars are observed every @xmath5 weeks , and data sets stretching for @xmath6yr exist for 34 of these pulsars @xcite .
together , these timescales imply that ptas are sensitive to gws in the frequency band @xmath7hz , which is complementary to other gw detection experiments .
the best - studied sources of gws within the pta frequency band are binary supermassive black holes ( smbhs ) .
stellar- or gas - dynamical evidence exists for smbhs at the centres of 87 nearby galaxies at the time of writing @xcite , with masses @xmath8 ranging between @xmath9 .
phenomenological models of the buildup of the cosmological mass density in smbhs during luminous quasar phases for redshifts @xmath10 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) suggest short quasar lifetimes .
this , considered together with local correlations between @xmath8 and , for example , galaxy bulge mass @xmath11 ( e.g. , * ? ? ? * ) , suggests that all massive galaxies ( @xmath12 ) which formed since the @xmath13 peak of quasar activity host smbhs ( see also * ? ? ?
* ) . in the context of hierarchical structure formation
, mergers are integral to the formation histories of massive galaxies , and evidence for interacting galaxies is seen across most of cosmic time @xcite .
multiple smbhs are expected to be found in galaxy merger products .
indeed , pairs of active galactic nuclei ( agn ) are observed in galaxies in the late stages of mergers @xcite , with projected separations as small as 7pc @xcite .
the central smbhs in a pair of merging galaxies are likely to sink in the merger remnant potential well through dynamical friction and form a bound binary ( e.g. , * ? ? ?
* ; * ? ? ?
dynamical friction becomes inefficient once stars within the binary orbit are ejected , and the slingshot scattering of stars on radial orbits @xcite or friction against circumbinary gas @xcite is required to drive further orbital decay . at a binary component separation of @xmath14pc ,
energy and angular momentum losses to gravitational wave ( gw ) emission can lead to smbh - smbh coalescence within a hubble time ( e.g. , * ? ? ?
a few candidate binary smbhs with such separations have been identified ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the existence of a large cosmological population of binary smbhs is thus inferred , some of which emit gws in the pta frequency band . if the orbits of all binary smbhs radiating gws in the pta frequency band are circular and evolving under gw emission alone , the summed gw signals from all binaries together have the expected spectral form ( e.g. , * ? ? ?
* ) : @xmath15 here , @xmath16 is the gw characteristic strain per logarithmic frequency unit , @xmath17 is the gw frequency at the earth , @xmath18 and @xmath19 is the characteristic spectral amplitude at @xmath20 .
the summed signals from binary smbhs may be collectively modelled as a gw background ( gwb ) .
gws from individual binaries may be detectable as continuous - wave ( cw ) sources ( e.g. , * ? ? ?
* ; * ? ? ?
coalescing pairs of smbhs also emit gw ` memory ' bursts @xcite , which are abrupt , propagating metric changes that also affect pulsar timing measurements ( e.g. , * ? ? ?
* ) . of these three types of gw signal , searches for
a gwb provide the best constraints on models for the binary smbh population @xcite and are perhaps the closest to yielding a successful detection @xcite .
@xcite used the first ppta data release @xcite to show that @xmath21 with 95% confidence .
@xcite suggest that the nanograv pta is likely to be able to detect a gwb with @xmath22 by around the year 2020 .
constraints on gws from individual binary smbhs with the current ppta data set showed that binaries with component masses @xmath23 and with separations of less than 0.02pc are unlikely to exist within 30mpc of the earth @xcite .
no search for memory bursts with pta data has so far been published .
wang et al . (
submitted ) describe an unsuccessful search for memory bursts using current ppta data .
predictions for gw signals from binary smbhs are based either on physical models for galaxy formation and evolution which predict the binary smbh population , or on directly observed quantities such as the galaxy merger rate and smbh - galaxy scaling relations .
the former predictions typically combine dark matter halo merger rates in the cold dark matter paradigm , analytic or numerical estimates of galaxy merger and binary smbh formation timescales , prescriptions for the cosmic evolution of the galaxy and smbh population and the assumption of gw - driven binary orbital evolution @xcite .
once tuned to reproduce observables such as local smbh - galaxy relations , the galaxy stellar mass function ( gsmf ) and colour distribution , and the quasar luminosity function , these models result in estimates of @xmath19 in the range @xmath24 .
the exact value depends on , for example , the assumed cosmological parameters , galaxy merger timescales and the specific models for smbh formation and growth . models for the binary smbh population which more directly incorporate observational information @xcite result in similar values of @xmath19 .
while various studies suggest that individual binary smbhs may be viable cw sources of gws for ptas @xcite as well as viable sources of memory bursts @xcite , quantitative predictions of source counts have only been calculated for cw sources using theoretical galaxy formation models @xcite . in this paper
, we adopt an observations - based approach towards modelling the binary smbh population , in order to predict the range of gw signals in the pta frequency band . a similar study by @xcite , hereafter s13 , combined a plethora of observational estimates of the merger rate of massive galaxies , the gsmf and local smbh - galaxy relations to derive a range of possible gwb amplitudes .
however , some of the observational quantities included in the s13 study do not represent the best current knowledge .
furthermore , some relevant uncertainties , such as in the possible redshift evolution of the smbh - galaxy relations , were not accounted for by s13 , and no predictions for individual gw sources were made . besides addressing these issues , this paper builds on previous studies in the following ways : * we quantify the impacts of different observational uncertainties on the amplitude of the gwb generated by binary smbhs .
we focus in particular on aspects of our model for the binary smbh population for which little observational information currently exists . *
we highlight the redshifts , masses and types of merging galaxies which result in binary smbhs which dominate the gwb amplitude . *
we provide new , observations - based predictions for the counts of individual gw sources .
we further present the first estimates for the expected numbers of detectable individual gw sources , given different pta configurations , that are robust with respect to pulsar parameter fitting . in 2
, we outline our model , and present our results in 3 .
we state the key implications of this work for ptas in 4 . finally , we discuss these results in 5 , and summarise our conclusions in 6 . throughout this work
, we adopt a concordance cosmology based on results from the _ planck _ satellite @xcite , including @xmath25kms@xmath26mpc@xmath26 , @xmath27 and @xmath28 .
the cosmological population of binary smbhs emitting gws in the pta frequency band can be characterised using the smbh - smbh coalescence rate .
the numbers of binary smbhs in different orbits are related to the coalescence rate through a continuity equation @xcite that includes assumptions about the rate of binary smbh orbital evolution .
we assume that the smbh - smbh coalescence rate is equivalent to the galaxy merger rate .
this is justified because massive galaxy mergers are typically completed within a few galaxy dynamical times , whereas the timescales for two smbhs to form a gravitationally - bound binary and then coalesce through losses of energy and angular momentum to their environments and gws are much shorter ( e.g. , * ? ? ?
* ; * ? ? ?
in this work , we neglect systems of more than two gravitationally - interacting smbhs resulting from multiple galaxy mergers , because we expect these to be rare for the high mass ratio ( @xmath29 mergers between massive ( @xmath12 ) galaxies that we consider .
we further assume that each galaxy contains a central smbh with a mass related to the galaxy bulge mass .
we use measured quantities to determine the all - sky coalescence rate of pairs of smbhs . for each quantity
, we define a fiducial prescription , and also describe the possible ranges over which the prescription can vary . similarly to s13 , we express the galaxy merger rate as @xmath30 where @xmath31 is the number of mergers between two galaxies of combined stellar mass @xmath32 , @xmath33 is the ratio between the smaller and larger galaxy stellar masses and @xmath34 is the cosmological redshift .
the merger rate , @xmath35 , is defined as the number of mergers per units @xmath36 , @xmath33 , @xmath34 and observer time @xmath37 . in equation
[ eq : c7e3 ] , @xmath38 is the number of galaxies across the entire sky with a given @xmath36 at a given @xmath34 .
this is related to the standard gsmf , @xmath39 , as @xmath40 where @xmath41 is the sky - integrated comoving volume shell between redshifts @xmath34 and @xmath42 . in equation [ eq :
c7e4 ] , @xmath43 is the probability density function for a galaxy merger event with mass @xmath36 at redshift @xmath34 having a mass ratio @xmath33 , @xmath44 is the average proper time between major mergers for a galaxy with a mass @xmath36 at redshift @xmath34 and @xmath45 . also , @xmath46 is the number of mergers , @xmath47 , per unit proper time , @xmath48 , for a single galaxy with a mass @xmath36 at redshift @xmath34 . in order to convert galaxy stellar masses to bulge masses ( @xmath11 ) we distinguish between quiescent , red - sequence early - type galaxies and star - forming , blue - cloud late - type galaxies .
we write the total gsmf as a sum of the gsmfs of early- ( @xmath49 ) and late - type ( @xmath50 ) galaxies : @xmath51 we relate @xmath36 to @xmath11 for early- and late - type galaxies using a scheme described in 2.1.3 .
to convert between @xmath11 and the central smbh masses ( @xmath8 ) we use the widely - known @xmath52 relation @xcite .
in contrast to s13 , we express this relation as @xmath53 where @xmath54 denotes a normal probability density function with centre @xmath55 and variance @xmath56 and @xmath57 , @xmath58 and the intrinsic scatter , @xmath59 , are observationally - determined constants .
it is important to account for intrinsic scatter in the @xmath52 relation when inferring the smbh mass function from the bulge mass function ( e.g. , * ? ? ?
* ) , because to not do so would lead to the smbh mass function being underestimated .
observational estimates of @xmath60 require knowledge of the fraction of galaxies , @xmath61 , within a mass - complete sample at a given redshift that are undergoing mergers , and the proper time @xmath62 during which merger events can be observationally identified ( for a review , see * ? ? ?
* ) . then , @xmath63 . in this work ,
we focus on major mergers with stellar mass ratios @xmath64 , because these systems are likely to dominate the gw signal ( e.g. , * ? ? ?
* s13 ) .
we consider three recent measurements of @xmath61 for major mergers at different redshifts in wide - area galaxy surveys , which are largely complete for galaxy stellar masses @xmath12 .
these three studies fit their data to the function @xmath65 , where @xmath66 and @xmath67 are free parameters .
* @xcite used structural analyses of concentration , asymmetry and clumpiness ( the ` cas ' parameters ) to identify systems in the process of merging among @xmath68 galaxies in the cosmos and extended groth strip surveys with @xmath12 at @xmath69 .
this technique is sensitive to major mergers in particular , with mass ratios @xmath70 @xcite .
@xcite found @xmath71 .
* @xcite counted galaxy pairs with projected separations between 5h@xmath26kpc and 20h@xmath26kpc from the cosmos survey to estimate @xmath61 for @xmath72 and @xmath73 .
they scaled their results to include galaxy pairs for all @xmath64 using the argument that @xmath61 is inversely proportional to the logarithm of minimum mass ratio of the observed galaxy pair sample .
@xcite found @xmath74 . *
both the above works may suffer from incorporating small galaxy samples at low redshifts .
this issue was addressed by @xcite using a large sample of galaxy pairs from the galaxy and mass assembly ( gama ) survey in the redshift interval @xmath75 . when standardised to the same projected separation , galaxy mass and mass ratio windows as @xcite
, they found a substantially higher value of @xmath61 at these redshifts . by combining their results with all recent measurements of @xmath61 at redshifts up to 1.2 , and normalising to the same projected pair separations of @xcite
, @xcite found @xmath76 .
other works have estimated @xmath61 with varying levels of accuracy .
s13 included results from the galaxy pair studies of @xcite , @xcite and @xcite .
however , @xcite and @xcite had significantly smaller pair samples than were utilised by either @xcite or @xcite , and @xcite only considered major mergers of galaxies with @xmath77 . in the absence of observational estimates of the galaxy merger timescale ( @xmath62 ) used in calculating @xmath60 from different measurements of @xmath61 , we make use of theoretical predictions . however , the range of possible predictions spans a factor of three .
@xcite used a mock galaxy catalogue from a semi - analytic model implemented within the millennium simulation @xcite to estimate @xmath62 for galaxies with different masses at different stages of merging assuming circular galaxy orbits and angular momentum loss through dynamical friction .
however , a suite of hydrodynamic simulations of galaxy mergers conducted by @xcite and @xcite , hereafter collectively l08 , resulted in significantly shorter merger timescales .
while some authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ) use the estimates of @xcite to calculate @xmath60 , others ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) argue that these estimates are incorrect , at least for major mergers ( see footnote 15 of * ? ? ?
analyses of cosmological hydrodynamical simulations combining dark matter and baryonic components suggest that the merger timescales assumed in semi - analytic models of galaxy formation and used by @xcite are overestimated for major mergers @xcite . furthermore , l08 presented estimates of @xmath62 specifically calibrated to the cas technique of @xcite using mock galaxy images . here , as a fiducial case
, we only use the estimates of @xmath62 from l08 , specific to estimates of @xmath61 from both galaxy pair counts and cas analyses .
these merger timescales were averaged over both field and cluster environments , and hence account for environmental dependencies . however , the simulation suite of l08 was not large enough to reveal significant mass- or redshift - dependence of @xmath62 .
hence , we consider it possible that the weak dependencies on these quantities identified by @xcite may be present .
we therefore have the three following estimates of the times between galaxy mergers : @xmath78 based on the work of @xcite , @xcite and @xcite respectively . while we consider each of equations [ eq : c7e7]@xmath4[eq : c7e9 ] to be equally possible , we choose equation [ eq : c7e7 ] @xcite as a fiducial prescription . the possible mass- and redshift - dependence of @xmath60 is given by the factor @xmath79 @xcite ; we further consider it equally likely that this factor is present or absent , while choosing its absence as fiducial .
together , there are then six different possibilities for @xmath60 that we consider , each with observational uncertainties .
we also demonstrate the effects on the gw signal from binary smbhs of using systematically larger values of @xmath62 that are consistent with @xcite .
the fitting formulae in equations [ eq : c7e7]@xmath4[eq : c7e9 ] are consistent with results at higher redshifts @xcite .
we hence adopt these equations for @xmath80 , and also assume @xmath81 @xcite .
uncertainties in @xmath60 for @xmath82 do not significantly affect our predictions for gw signals from binary smbhs , because , as we demonstrate , it appears that these signals are dominated by contributions from binary smbhs at lower redshifts .
we use the latest measurements of the gsmf for @xmath80 in the range @xmath83 based on the cosmos / ultravista catalogue @xcite . @xcite
present gsmfs for quiescent ( early - type ) and star - forming ( late - type ) galaxies , which were identified using a colour cut .
utilising uv to mid - ir galaxy photometry , with improved sensitivity and sky - coverage over previous compilations , these authors provide the most accurate determinations of the early- and late - type gsmfs currently available .
however , we still need to account for a selection of systematic errors .
@xcite use redshift , luminosity and mass measurements obtained through spectral energy distribution analyses . assuming galaxy magnitude measurements of sufficient accuracy , systematic errors in the photometric redshifts and galaxy
stellar mass measurements are dominated by how the stellar populations are modelled ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
systematic errors in stellar mass measurements can lead to errors in the gsmf of greater than 0.6dex @xcite .
@xcite present five separate determinations of the gsmfs of early- and late - type galaxies using different choices for the stellar population synthesis model and star formation history , as well as expanded possibilities for galaxy metallicities and dust attenuation laws .
we assume that each of these five gsmf determinations , for which schechter function fits are given in table 3 of @xcite , are equally likely to be correct , but choose the default gsmf of @xcite , given in their table 1 , as fiducial .
the method of colour selection used to identify early- and late - type galaxies adds further systematic uncertainty to the gsmf estimates .
for example , @xcite showed that edge - on dusty spiral galaxies are in fact the reddest among the galaxy population , and that more than a third of a red - sequence sample of galaxies could be actively star - forming objects . while @xcite suggest further simple morphological selections based on galaxy light concentrations to mitigate these effects , these data were not available in the cosmos / ultravista catalogue .
instead , we use a crude estimation of the uncertainty range of the gsmf caused by the colour selection from @xcite , who presented gsmfs determined for significantly different colour cuts to their fiducial scheme ( their table 4 ) .
we consider this entire range of variability in the gsmf to be possible .
we also demonstrate the effects on the resulting gw signal of possible contamination of colour - selected early - type galaxy samples with late - type galaxies in an extreme scenario by also performing our calculations with the early - type gsmf reduced by @xmath84 .
the scheme we use to relate @xmath36 to @xmath11 for different types of galaxies is summarised as follows .
* of late - type galaxies with @xmath12 , less than 10% have no bulge component @xcite ; in this work , we assume a conservative value of 10% . of the others ,
@xmath85 is in the range @xmath86 @xcite . * early - type galaxies with @xmath12 consist of a significant fraction that are best modelled with both bulges and disks , which are identified with the s0 ( lenticular ) galaxy population @xcite .
these galaxies have values of @xmath85 which are approximately log - normally distributed with mean 0.7 and log - deviation @xmath87dex . a mild correlation between @xmath85 and @xmath36
may be present for s0 galaxies @xcite , which we neglect in this work . * for @xmath88 , approximately 75% of early - type galaxies are s0s and 25% are true ellipticals @xcite .
these fractions change to 55% and 45% respectively for larger stellar masses . while these results are quite approximate , and only derived for a low - redshift ( @xmath89 ) galaxy sample , we adopt them as a fiducial scheme for relating @xmath36 to @xmath11 for @xmath80 .
this scheme is roughly consistent with that used by s13 . in the same way as accounting for scatter in the @xmath52 relation raises the inferred smbh mass function ( e.g. , * ? ? ?
* ) , the bulge mass function inferred from the gsmf will be raised given scatter in relating @xmath36 to @xmath11 .
scatter in the @xmath90 relations can be simply combined with the scatter in the @xmath52 relation by modifying equation [ eq : c7e6 ] as follows : @xmath91,\ ] ] where we assume @xmath92 for both early- and late - type galaxies . in summary ,
the function @xmath93 in our fiducial model is defined by @xmath94 we demonstrate the effects of possible errors in the fraction of early - type galaxies which are ellipticals by considering cases where this fraction is reduced and increased by 50% . despite intense interest in evincing the @xmath52 relation over the last 15 years
, the form of the relation remains uncertain @xcite .
@xcite argue that the @xmath52 relation is well modelled by a single power law for all galaxies containing classical bulges which include ellipticals , s0s and spirals with bulges displaying steep central light gradients .
however , @xcite find , using an extended version of the galaxy sample of @xcite and independent measurements of @xmath11 , that two power laws are required , with a break at @xmath95 .
a physical distinction between the two power laws was identified by splitting the sample into ` cusp ' galaxies with steep power - law central light gradients and galaxies where ` cores ' , or light - deficits with respect to a cusp , are present .
cusp galaxies are typically of lower masses than core galaxies , and were found by @xcite to have a steeper log - linear @xmath52 relation than core galaxies . while we the consider the @xmath52 relations of @xcite and @xcite equally likely , we choose the simpler relation of @xcite as a fiducial case . in equation [ eq : c7e10 ] ,
@xcite find @xmath96 , @xmath97 and @xmath98 .
@xcite instead find @xmath99 and @xmath100 for @xmath101 and @xmath102 and @xmath103 for @xmath104 .
as @xcite do not estimate the intrinsic scatter , we assume @xmath98 for the entire range of @xmath11 . we do not consider estimates of the @xmath52 relation made substantially prior to @xcite and @xcite .
previous estimates are thought to be incorrect because of systematic errors in smbh and bulge mass estimates , the absence of recently - measured smbh masses in brightest cluster galaxies , and the presence of galaxies without classical bulges in samples used to fit the relations ( for details , see * ? ? ?
various authors infer modest redshift evolution in the @xmath52 relation such that the typical ratio @xmath105 may be up to a factor of @xmath106 larger at @xmath107 than the local value ( * ? ? ?
* and references therein ) .
this can be approximately represented by letting @xmath108 with @xmath109 and @xmath110 as above . as a fiducial case ,
however , we assume the conservative value of @xmath111 . in this paper , we assume that all binary smbhs are in circular orbits that evolve only under losses of energy and angular momentum to gws .
while the effects of binary smbh environments and non - zero orbital eccentricities could modify the gw characteristic strain spectrum from the form in equation 1 at frequencies up to @xmath112hz at the earth , these effects are highly uncertain ( * ? ? ?
* and references therein ) . for frequencies @xmath113hz within the pta band ( e.g. , at @xmath114 )
, the characteristic strain spectrum does indeed take the form of equation ( 1 ) , because the orbits of all binaries radiating gws at these frequencies are likely to have circularised because of gw - driven evolution .
our assumption allows for direct comparison with the majority of studies on this topic @xcite , and for the gwb spectrum to be characterised by a single amplitude ( @xmath19 ) . a circular binary smbh radiates monochromatic gws at twice its orbital frequency .
we use standard expressions from the literature for the rms gw strain amplitude , @xmath115 ( e.g. , equation 7 of * ? ? ?
* ) radiated by a circular binary , and the rms gw - induced sinusoidal variations to the pulse times of arrival ( toas ) from radio pulsars , @xmath116 ( e.g. , equation 20 of * ? ? ?
both @xmath115 and @xmath116 are averaged over all binary orientation parameters .
following @xcite , we approximate the strain amplitude of a memory burst from a coalescing binary smbh as @xmath117 where @xmath118 is the reduced mass of the coalescing binary system . to calculate the gwb amplitude for a population of binary smbhs , consider a multivariate density function , @xmath119 , for the observed binary smbh coalescence rate , @xmath120 , in terms of a @xmath121-component parameter vector @xmath122 with components @xmath123 indexed by an integer @xmath124 : @xmath125}{\partial x_{i}}.\ ] ] following , e.g. , @xcite , @xmath19 is given by @xmath126^{1/2},\ ] ] here , @xmath127 for the domains of @xmath37 and @xmath17 under consideration .
mergers between galaxies containing bulges with masses @xmath36 and @xmath128 come in nine types , because the galaxies with each mass may be either elliptical , s0 or late - type . in each case
, a different prescription is required to identify the bulge masses of the merging galaxies , and hence the masses of the smbhs in the merging galaxies .
consider a merger between a galaxy of type @xmath124 , with mass @xmath36 , and a galaxy of type @xmath129 , with mass @xmath128 , where @xmath124 and @xmath129 each denote either an elliptical , s0 or late - type galaxy .
the fraction of cases where this merger will occur is given by @xmath130 , where @xmath131 is given by equation [ eq : c7e5 ] and the mass functions are evaluated at a mass @xmath128 . for early - type galaxies , @xmath132
is specified according to the fractions of ellipticals and s0s at different masses , and for late - type galaxies , @xmath132 is simply the fraction which contain bulges .
the smbh masses corresponding to the galaxies of types @xmath124 and @xmath129 , @xmath133 and @xmath134 respectively , are described by the probability density function in equation [ eq : c7e10 ] for @xmath11 given by @xmath135 and @xmath136 respectively .
hence , in order to calculate @xmath19 , we combine equations [ eq : c7e3 ] , [ eq : c7e4 ] , [ eq : c7e5 ] , [ eq : c7e10 ] and [ eq : c7e17 ] as follows : @xmath137 we evaluate this integral numerically by summing over the integrand in bins of @xmath138 , @xmath34 , @xmath139 , @xmath140 , and @xmath141 . to determine the predicted numbers of cw sources , we count the numbers of individual binary smbhs in each bin with different values of @xmath115 radiating gws at @xmath20 within a nominal bandwidth of @xmath142 .
we also record the rate of memory bursts in each bin with corresponding amplitudes @xmath143 .
these latter operations are equivalent to numerically evaluating the conditional densities of gw sources in terms of @xmath115 and @xmath143 .
equation [ eq : c7e18 ] builds on the approach of s13 in two ways .
first , we account for the effects of intrinsic scatter in the @xmath52 relation and in relating @xmath36 to @xmath11 .
we also attempt to match the numbers of galaxy mergers of different types to the measured gsmfs , rather than assuming the same galaxy pair fractions for all types of mergers .
we first calculated @xmath19 using equation [ eq : c7e18 ] given the fiducial prescriptions for @xmath144 , the gsmf , the scheme relating @xmath36 and @xmath11 and the @xmath52 relation , as detailed in sections [email protected] .
the resulting fiducial value for @xmath19 was @xmath145 .
we then identified the possible ranges of @xmath19 consistent with the observational uncertainties in each of @xmath144 , the gsmf and the @xmath52 relation alone .
this was accomplished by generating 600 realisations of @xmath19 with the parameters of a single one of these quantities randomised and with the other terms in equation [ eq : c7e18 ] held fixed at their fiducial values .
the process was then repeated with randomisation individually in the other two quantities .
histograms of the resulting three samples of realisations of @xmath19 are shown in the middle three panels of figure [ fig : c7f1 ] , along with the fiducial value of @xmath19 ( as a vertical dashed line ) . while the possible ranges of @xmath19 given observational uncertainties in @xmath144 and the gsmf are roughly equivalent , observational uncertainty in the @xmath52 relation results in a slightly larger range of possible @xmath19 values .
we also considered the effects of adopting four modifications to the fiducial model relating to parameters for which current observational constraints are poor .
these modifications , which were introduced in subsections [email protected] , result in the following values of @xmath19 : 1 .
when we decrease the early - type gsmf by a factor of @xmath84 to simulate an extreme case of contamination of colour - selected early - type galaxy samples by late - type galaxies ( e.g. , edge - on spirals ) , we obtain @xmath22 .
this represents a decrease of 0.12dex over the fiducial model .
when we adopt the massive galaxy merger timescale from @xcite with the associated mass- and redshift - dependence , rather than from numerical simulations of galaxy mergers ( l08 ) , we obtain @xmath146 .
this represents a decrease of 0.24dex over the fiducial model .
3 . when we introduce a redshift - dependent normalisation , @xmath57 , of the @xmath52 relation with @xmath109 such that the normalisation is a factor of three greater at @xmath147 , we obtain @xmath148 .
this is an increase of 0.14dex over the fiducial model .
4 . finally ,
when we explore the effects of either reducing or increasing the fraction of early - type galaxies which are ellipticals by 50% , we obtain an associated variation in @xmath19 of 5% ( 0.02dex ) .
the first three modifications are clearly significant , as compared to the fourth : we depict the resulting values of @xmath19 in the top panel of figure [ fig : c7f1 ] . while the effects of modifications ( @xmath124 ) and ( @xmath149 ) are comparable in magnitude , adopting the galaxy merger timescales of @xcite makes a large difference to the prediction of @xmath19 .
this is expected , because the @xcite merger timescales are roughly a factor of three longer than those of l08 . indeed ,
modification ( @xmath150 ) results in a value of @xmath19 that is lower than the 2.5% percentile of the distributions of @xmath19 values given the three observational uncertainties considered so far . as an illustration of the full range of possible values of the gwb amplitude given the uncertainties considered in @xmath144 , the gsmf and the @xmath52 relation combined with modifications @xmath151 listed above
, we generated a new sample of 600 realisations of @xmath19 . in this case
, we simultaneously randomised over @xmath144 , the gsmf and the @xmath52 relation as described above , and also ( _ i _ ) decreased the early - type gsmf by a factor uniformly drawn from the interval @xmath152 $ ] , ( _ ii _ ) set the galaxy merger timescale at a value uniformly drawn between the predictions of l08 and @xcite ( neglecting any mass- or redshift- dependence ) , and ( _ iii _ ) set the redshift - evolution index @xmath153 of the normalisation of the @xmath52 relation to a number uniformly drawn from the interval @xmath154 $ ] .
a histogram of the resulting sample of realisations of @xmath19 is shown in the bottom panel of figure [ fig : c7f1 ] , labelled ` all ' . the long tail to lower values of @xmath19 , which is not reflected in the other histograms ,
is caused specifically by the inclusion of uncertainties in the galaxy merger timescale and in the early - type gsmf .
the magnitudes of these effects on @xmath19 are indicated by the arrows at the top of figure [ fig : c7f1 ] .
the 95% confidence interval on @xmath19 , considering all uncertainties , is @xmath0 .
we next compare our results for @xmath19 with earlier predictions . in figure
[ fig : c7f2 ] , we again show the histogram of realisations of @xmath19 corresponding to randomisation over all uncertainties , as well as the values of @xmath19 corresponding to modifications ( @xmath124 ) to ( @xmath149 ) listed above . above these
, we show the 68% confidence intervals on @xmath19 from four recent , independent models for the binary smbh population ( s13 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the predictions that we consider all account for the most recent determinations of the @xmath52 relation @xcite .
the range of possible values of @xmath19 predicted by s13 is consistent with ( albeit somewhat broader than ) the range we predict given all uncertainties that we consider in this paper .
both the present work and s13 attempt to synthesise all uncertainties in quantities relevant to characterising the smbh - smbh coalescence rate , and use the same underlying model assumptions to predict the gwb amplitude .
our range of predictions is less extended than that of s13 because of the greater uncertainty assumed by s13 in the gsmf and the galaxy merger rate .
were obtained . ]
a semi - analytic approach @xcite was used by @xcite to predict smbh - smbh coalescence rates within the millennium simulation @xcite , coupled with prescriptions for binary smbh orbital evolution in stellar environments @xcite .
the results of @xcite indicate that the characteristic strain spectrum may be attenuated relative to the case of circular binary orbits and gw - driven evolution at frequencies @xmath155hz .
however , the 68% confidence interval on the characteristic strain spectral amplitude at a frequency of @xmath20 is consistent with the range of values of @xmath19 we find .
the prediction of @xcite is derived from hydrodynamic numerical galaxy formation simulations in cluster and field environments , but may be biased relative to semi - analytic galaxy formation models implemented in large - volume numerical dark matter simulations because of the specific choice of overdense and underdense regions to study .
however , the prediction of @xcite naturally includes a particiularly sophisticated treatment of galaxy merger timescales .
@xcite suggest a model for the binary smbh population which includes the assumption that all evolution in the early - type gsmf at @xmath72 is driven by galaxy mergers ; however , their predicted gwb amplitude appears to be inconsistent with current pta constraints @xcite .
this model would necessarily include a shorter galaxy merger timescale than that predicted by l08 in order to maintain consistency with the observed numbers of merging galaxies .
overall , besides the study of @xcite , it is encouraging that different models appear to agree on the amplitude of the characteristic strain spectrum from binary smbhs .
in particular , the upper ends of most predicted ranges of @xmath19 all appear to be consistent . in figure [
fig : c7f2 ] , we also depict the best existing 95% confidence upper limit on @xmath19 from @xcite as a shaded region .
some realisations of @xmath19 given observational uncertainties in our model are inconsistent with this upper limit .
however , the upper limit is generally consistent with our model given all uncertainties . in figure
[ fig : c7f3 ] , we plot the values of @xmath19 predicted by the fiducial model in different ranges of @xmath36 ( left panel ) and @xmath34 ( right panel ) . we also show the 95% confidence intervals on these values given all uncertainties we consider .
the galaxy mass ranges correspond to the values of @xmath36 of the larger galaxies in mergers .
the dominant contributions to the gwb are from binary smbhs formed in mergers involving galaxies with @xmath156 , and from binary smbhs at redshifts @xmath157 .
the confidence intervals that we provide further suggest that contributions to the gwb from outside these ranges are not significant .. ] finally , binary smbhs created in mergers involving at least one late - type galaxy correspond to @xmath158 , whereas mergers involving only early - type galaxies correspond to @xmath159 .
hence , within our model , the gwb is likely to be dominated by galaxy mergers involving only early - type galaxies ( s0s and ellipticals ) . in the process of evaluating equation [ eq : c7e18 ] , we also calculated the numbers of individual binary smbhs that produce monochromatic ( cw ) gw signals , along with the numbers of gw memory bursts emitted during smbh - smbh coalescence events .
we counted individual binaries emitting gws at frequencies @xmath114 in a frequency bin of width @xmath142 and evaluated the numbers of binaries with different gw strain amplitudes @xmath115 .
these results are shown in the left panel of figure 4 for the fiducial model as well as for two variations to the fiducial model ( modifications ( ii ) and ( iii ) listed above ) .
we also show results for the fiducial model while restricting the source counts to binaries at redshifts @xmath72 and with the more massive progenitor galaxy mass @xmath77 .
the restricted source counts are identical to the full source counts for @xmath160 . from @xcite ,
the characteristic amplitude of the sinusoidal toa variations induced by a binary smbh with strain amplitude @xmath115 at @xmath114 , over a 10yr observation , is @xmath161ns .
scaling these cw source counts to other gw frequencies is non - trivial .
the gw strain amplitude of a binary smbh radiating at a frequency @xmath17 can be expressed as @xmath162 , where @xmath163 is the strain amplitude radiated by that binary at a frequency @xmath20 .
furthermore , the total number of binaries per unit frequency radiating gws at a frequency @xmath17 is related to the number of binaries per unit frequency radiating gws at @xmath20 by the factor @xmath164 , assuming gw - driven binary orbital evolution .
then , the number of binaries per unit frequency emitting gws at or above a strain amplitude of @xmath115 , at a frequency @xmath17 , may be written as @xmath165 .
for example , while the fiducial model predicts @xmath166@xmath167 cw sources with @xmath168 in a frequency bin of width @xmath142 at @xmath114 , this prediction changes to @xmath1660.1 sources at @xmath169 with @xmath168 in the same frequency bin width .
we can hence directly compare our predicted cw source counts with the work of @xcite .
these authors considered a wide variety of smbh growth scenarios within the framework of a semi - analytic model for galaxy formation @xcite implemented in the millennium simulation results @xcite .
we directly compare predictions for the number of binary smbhs inducing toa variations with characteristic amplitudes @xmath170ns . for consistency
, we consider an observation time span of @xmath171yr and gw frequencies @xmath172hz , and integrate over the number of sources per unit frequency with @xmath170ns in the range @xmath173hz ( integrating to higher frequencies does not significantly alter our results ) .
we neglect the issue of whether these signals are resolvable given the presence of a gwb .
we predict 0.6 cw sources with @xmath170ns for our fiducial model , 0.1 cw sources for a pessimistic model assuming the galaxy merger timescales of @xcite , and 1.2 cw sources for our optimistic model with significant redshift - evolution in the @xmath52 relation .
@xcite predict between 0.05 and 3 such sources ( their figure 3 ) , which is consistent with our results .
we also predicted the numbers of binary smbh coalescence events per observed year at or above a given gw memory burst amplitude , @xmath143 ( see equation 12 ) for @xmath174 .
the results are shown in the right panel of figure 4 , again for the fiducial model and two variations to this model .
we also again show results for the fiducial model with the restrictions of @xmath72 and @xmath77 ; for @xmath175 , the restrictions make no significant difference . in summary ,
the expected numbers of individual gw sources predicted by our empirical binary smbh model are small .
at most @xmath176 cw source is expected to induce toa variations with characteristic amplitudes @xmath177ns over a 5yr observation time span .
also , approximately one gw memory burst with @xmath178 is expected every 1000yr .
the future sensitivities of ptas to the gwb are the subjects of ongoing research ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for example , future pulsar observing systems and cadences , new pulsar discoveries , the effects of the interstellar medium and pulsar timing noise characteristics , all of which significantly affect pta sensitivities , are difficult to forecast because of a lack of quantitative , predictive models .
an idealised treatment of the problem by @xcite suggests that , for the nanograv collaboration , a gwb with amplitude @xmath22 may be detectable before the year 2020 .
we note that @xcite assumed that the gwb characteristic strain spectrum has the power law form given in equation [ eq : c7e1 ] .
we find in this paper that the gwb amplitude is likely to be in the range @xmath0 with 95% confidence . if the gwb amplitude were to lie in the upper part of this range , as is expected given the more commonly preferred major galaxy merger timescale ( l08 ) , we suggest that detecting a gwb from binary smbhs is indeed an attainable , short - term goal for ptas .
values given all uncertainties , while assuming the l08 galaxy merger timescales , we find @xmath179 with 95% confidence . ] what can pta upper limits on or detections of the gwb reveal about the determinants of the gwb amplitude
? the gwb may be parameterised by a single number , @xmath19 ( at least at gw frequencies @xmath180hz ; * ? ? ?
* ) , the value of which is dependent on myriad quantities .
useful information can be gleaned if one of these quantities is particularly unconstrained otherwise .
for example , if we remain agnostic with respect to the galaxy merger timescale , a particular value of this timescale would correspond to a range of possible gwb amplitudes given our knowledge of all the other determinants of @xmath19 .
then , a pta constraint on @xmath19 would correspond to a constraint on the galaxy merger timescale , given the assumptions inherent in our model . through such exercises , ptas could directly impact our understanding of galaxy and smbh growth , in a more general sense than by testing specific gwb models using pta data .
we leave a demonstration of such techniques for future work .
future pta observations with planned telescopes such as the five hundred metre aperture spherical telescope ( fast , li , nan & pan 2013 ) and the square kilometre array ( ska , * ? ? ?
* ) may include up to 100 pulsars with timing noise standard deviations of @xmath166100ns @xcite .
@xcite constructed theoretical pta sensitivity curves using simulated data sets with both 100 arbitrarily - located pulsars or 17 pulsars at the locations of the best - timed pulsars observed by the nanograv collaboration , in all cases with timing noise standard deviations of 100ns and 5yr observation times .
these sensitivity curves , shown in their figure 5 , represent the values of @xmath115 at different frequencies at which the probability of a false detection was less than @xmath181 in 95% of realisations of their simulated data sets .
importantly , the sensitivity curves were averaged over all source positions and orientations , and account for pulsar parameter fitting .
we predict the numbers of detectable sources for ptas with these sensitivity curves by evaluating the following integral : @xmath182}{df}df,\ ] ] where @xmath183 is the sensitivity curve and @xmath184 is the predicted number of sources with strain amplitudes @xmath185 per unit frequency at a frequency @xmath17 .
the sensitivities of ptas to cw sources are generally poor for frequencies @xmath186hz and few sources are expected at these frequencies . using our predictions for the numbers of cw sources ,
we evaluate @xmath184 by scaling the predictions as described in 3.2 .
then , for the fiducial model and for the two sensitivity curves of @xcite corresponding to their coherent @xmath187-statistic , we obtain predictions of 0.07 and 1.3 detectable sources for the 17- and 100-pulsar cases respectively . for the restricted fiducial model , corresponding only to sources with @xmath72 and @xmath77 these reduce marginally to 0.06 and 1 source respectively . for the optimistic case with strong redshift - evolution of the @xmath52 relation ,
we obtain predictions of 0.2 and 2.8 detectable sources for the 17- and 100-pulsar cases respectively .
in contrast , the current ppta sensitivity curve produced by @xcite corresponds to @xmath188 detectable sources .
` noise ' caused by the summed gw signal from the binary smbh population will further increase the difficulty of detecting individual binaries ( e.g. , * ? ? ?
* ; * ? ? ?
a pta data set with 20 pulsars timed with a precision of 100ns for 10yr is sensitive to memory bursts with amplitudes @xmath189 over @xmath190 of the data span @xcite .
as the sensitivity of such an idealised pta to memory bursts scales roughly as the square root of the number of pulsars @xcite , a pta with 100 pulsars timed with 100ns precision for 10yr may be sensitive to memory bursts with @xmath191 .
however , our model suggests that only @xmath166@xmath192 bursts with @xmath189 and @xmath166@xmath193 bursts with @xmath191 are expected over 10yr .
thus , under the model presented here , gw memory bursts from coalescing binary smbhs do not represent viable sources for ptas .
our predictions for the gwb amplitude , @xmath19 , are conservative within their respective scenarios , for a number of reasons .
( _ i _ ) we do not account for minor galaxy mergers with stellar mass ratios @xmath194 , or for mergers where the more massive galaxy has a mass @xmath195 .
( _ ii _ ) we do not consider the possibility of gas accretion onto smbhs prior to coalescence during galaxy mergers ( e.g. , * ? ? ? * ) , which would raise the smbh masses and hence the emitted gw amplitudes ( e.g. , * ? ? ? * ) .
( _ iii _ ) the most massive galaxies are typically found in cluster environments , where times between galaxy mergers may be shorter ( cf .
* ) , implying a higher merger rate for these galaxies and hence a higher gw signal .
however , we do not expect the inclusion of these factors to significantly affect our predicted gwb amplitudes .
we reiterate that the effects of interactions between binary smbhs and their environments are unlikely to affect the predictions for the gwb amplitude at frequencies @xmath180hz @xcite , such as at @xmath20 .
this is because the orbital evolution of binary smbhs radiating gws at these frequencies is expected to be predominantly gw - driven , which further leads to the circularisation of the orbits . of all sources of uncertainty we consider in predicting the gwb amplitude given relevant observational quantities , the choice of galaxy merger timescale dominates the range of possible gwb amplitudes .
furthermore , the merger timescale may be even more uncertain than the range spanned by the predictions we consider ( l08 ; * ? ? ?
the simulations of l08 were conducted only for mergers between gas - rich disk galaxies , some of which contained small bulges , whereas we find that the gwb is likely dominated by binary smbhs formed in mergers solely between early - type galaxies .
further theoretical studies of galaxy merger timescales for early - type systems are clearly required in order to better predict the gwb amplitude .
the dominance of low - redshift ( @xmath196 ) early - type major galaxy mergers of massive ( @xmath156 ) galaxies in determining the gwb amplitude is a further important consequence of our work for both theoretical and observational studies of galaxy mergers aimed at informing pta research .
the other significant source of uncertainty in our predictions is in the @xmath52 relation , both in its local form and in its possible redshift - evolution .
in contrast to uncertainty in the galaxy merger timescale , it is likely that this uncertainty will only be resolved through further observations which significantly expand the sample of known smbh masses .
promisingly , @xcite report that hundreds of smbh mass measurements may be possible with the atacama large millimetre array ( alma ) .
under what circumstances could the gwb amplitude lie outside the range we predict given all uncertainties that we consider ?
the predicted range of gwb amplitudes , @xmath0 , encompasses all purely observational uncertainties , as well as uncertainty ranges that we set for other quantities for which observational constraints are poor , such as the galaxy merger timescale .
it may be possible that these latter ranges are incorrect .
furthermore , not all galaxies may host a central smbh , as we have assumed .
the interaction between a binary smbh and a third smbh would likely cause the least massive smbh to be ejected ( e.g. , * ? ? ?
* ) , lowering the number of coalescing smbhs .
if not every massive galaxy at @xmath197 formed with a central smbh , the gwb amplitude would again be lowered . it may also be possible that binary smbhs do not always coalesce on timescales less than the times between galaxy mergers .
the presence of a few strong gw emitters among the binary smbh population implies that some excess , non - gaussian scatter will be present in the gw signals produced by this population .
the magnitude of this excess scatter in @xmath19 depends on exactly how many binary smbh systems contribute significantly to the gwb . using a semi - analytic galaxy formation model implemented in the millennium simulation @xcite , @xcite suggested that the statistics of toa variations induced by gws from binary smbhs are mildly non - gaussian for frequencies @xmath198 because of appreciable contributions to the squared characteristic strain spectrum , @xmath199 , from individual binaries at every gw frequency .
figure 5 shows the number of binary smbhs in our fiducial model corresponding to galaxy mergers with primary stellar masses greater than or equal to a given @xmath36 ( top ) , as well as the fractions of @xmath200 contributed by these binaries ( bottom ) .
we show in particular binaries radiating at a gw frequency of @xmath20 in a frequency bin of width @xmath142 .
our fiducial model suggests that the contributions of individual gw sources to @xmath199 are lower than estimated by @xcite .
for example , the modelling in @xcite found that one source contributed @xmath201 of @xmath199 at a frequency of @xmath202 in a frequency bin of width @xmath203 ( their figure 2 ) .
in contrast , our empirical modelling in this paper suggests that the strongest @xmath204 sources in such a frequency bin contribute @xmath201 of @xmath205 .
this work and @xcite clearly predict different numbers of the most massive binary smbhs .
while this discrepancy will only be resolved with gw observations , we point out that the schechter functions for the gsmfs that we use under - predict observed galaxy counts at the highest masses and the lowest redshifts @xcite .
hence , it is possible that our model under - represents the contributions of the most massive binary smbhs to the total gw signal .
differing typical galaxy merger mass ratios in cluster and field environments ( e.g. , * ? ? ?
* ) are a further complicating factor .
in this paper , we predicted the strength of the gwb from binary smbhs and the occurrence of individual binary smbh gw sources .
our approach was to use a selection of recent observational estimates for the average times between major mergers for galaxies with @xmath12 and @xmath80 and for the gsmfs of early- and late - type galaxies in this mass and redshift range .
we combined these quantities with empirical relations between galaxy and bulge stellar masses and between bulge and smbh masses .
we find that while current ptas are unlikely to be sensitive to individual binary smbhs , a pta consisting of @xmath166100 pulsars timed with @xmath166100ns precision for 5yr will be sensitive to up to @xmath106 binary smbhs .
such a pta may be achievable with the ska @xcite , but is possibly beyond the capabilities of fast @xcite .
even such a pta will , however , have a less than 0.1% chance of detecting a gw memory burst from a coalescing binary smbh .
thus , we conclude that while individual binary smbhs may be detectable with a pta based on next - generation radio telescopes , memory bursts from coalescing smbhs are not likely to be detectable with any envisaged pta .
we predict that the characteristic strain amplitude of the gwb lies in the range @xmath0 with 95% confidence , accounting for a variety of uncertainties .
the upper end of the predicted amplitude range is equivalent to the best published 95% confidence upper limit on the gwb amplitude @xcite .
this reinforces the conclusion of @xcite that some models for the binary smbh population that are consistent with current electromagnetic observations are already inconsistent with pta constraints on the gwb . the dominant uncertainty in predicting the gwb amplitude appears to be caused by differences in theoretical predictions for the major merger timescale of massive galaxies .
higher values within our predicted range for @xmath19 correspond to the more commonly preferred choice of galaxy merger timescale ( l08 ) ; gwb amplitudes @xmath206 are within the sensitivity ranges of current and future ptas .
we strongly urge further work on quantifying the galaxy merger timescale , in particular for the mergers between massive early - type galaxies at redshifts @xmath207 which are likely to host the dominant contributors to the gwb .
the other significant uncertainty in our predictions is in the local form and possible redshift - evolution of the @xmath52 relation .
pta upper limits on or detections of the gwb may be able to meaningfully improve our knowledge of such otherwise poorly constrained facets of the formation and evolution of galaxies and smbhs .
the authors thank alberto sesana for useful discussions and justin ellis for sharing results on predicted sensitivity curves .
the authors also acknowledge the comments of the anonymous referee , which helped to significantly improve the manuscript .
v.r . is a recipient of a john stocker postgraduate scholarship from the science and industry endowment fund and j.s.b.w .
acknowledges an australian research council laureate fellowship .
gh is supported by an australian research council future fellowship .
this work was performed on the swinstar supercomputer at the swinburne university of technology .
j. m. , papovich c. , faber s. m. , ferguson h. c. , grogin n. , guo y. , kocevski d. , koekemoer a. m. , lee k .- s .
, mcintosh d. , momcheva i. , rudnick g. , saintonge a. , tran k .- v . , van der wel a. , willmer c. , 2013 , apj , 773 , 154 r. m. , ravi v. , coles w. a. , hobbs g. , keith m. j. , manchester r. n. , wyithe j. s. b. , bailes m. , bhat n. d. r. , burke - spolaor s. , khoo j. , levin y. , oslowski s. , sarkissian j. m. , van straten w. , verbiest j. p. w. , want j .- b . , 2013 , science , 342 , 334 d. r. b. , hobbs g. b. , jenet f. a. , verbiest j. p. w. , wen z. l. , manchester r. n. , coles w. a. , van straten w. , bailes m. , bhat n. d. r. , burke - spolaor s. , champion d. j. , hotan a. w. , sarkissian j. m. , 2010 , mnras , 407 , 669 | large - area sky surveys show that massive galaxies undergo at least one major merger in a hubble time .
ongoing pulsar timing array ( pta ) experiments are aimed at measuring the gravitational wave ( gw ) emission from binary supermassive black holes ( smbhs ) at the centres of galaxy merger remnants . in this paper , using the latest observational estimates for a range of galaxy properties and scaling relations , we predict the amplitude of the gw background generated by the binary smbh population .
we also predict the numbers of individual binary smbh gw sources .
we predict the characteristic strain amplitude of the gw background to lie in the range @xmath0 at a frequency of @xmath1 , with 95% confidence .
higher values within this range , which correspond to the more commonly preferred choice of galaxy merger timescale , will fall within the expected sensitivity ranges of existing pta projects in the next few years .
in contrast , we find that a pta consisting of at least 100 pulsars observed with next - generation radio telescopes will be required to detect continuous - wave gws from binary smbhs .
we further suggest that gw memory bursts from coalescing smbh pairs are not viable sources for ptas .
both the gw background and individual gw source counts are dominated by binaries formed in mergers between early - type galaxies of masses @xmath2 at redshifts @xmath3 .
uncertainties in the galaxy merger timescale and the smbh mass @xmath4 galaxy bulge mass relation dominate the uncertainty in our predictions .
[ firstpage ] black hole physics galaxies : evolution gravitational waves pulsars : general |
the tully - fisher relation @xcite is one of the strongest empirical correlations in extragalactic astronomy .
in addition to being a useful distance indicator ( e.g. , * ? ? ?
* ) , the physical basis of this relation is of fundamental importance . in the context of @xmath0cdm , the tully - fisher relation places tight constraints on galaxy formation models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in a broader context , it provides a useful test of theories that seek to modify gravity in order to obviate the need for dark matter ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
these should make specific and testable predictions for the tully - fisher relation .
the tully - fisher relation was originally posed as an empirical relation between optical luminosity and the width of the 21 cm line .
these are presumably proxies for more physical quantities .
the 21 cm line - width is a proxy for rotation velocity ( e.g. , * ? ? ?
* ; * ? ? ?
* ) which is widely presumed to be set by the dark matter halo .
luminosity is a proxy for stellar mass , which is presumably related to the total mass of the system .
thus a relation like tully - fisher seems natural , with the mass of the parent halo specifying both luminosity and rotation velocity .
luminosity is not a perfect proxy for mass .
the stellar mass - to - light ratio can , and presumably does , vary with galaxy type .
this leads to tully - fisher relations with different slopes and scatter depending on the band - pass used to measure the luminosity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. moreover , low luminosity galaxies tend to fall below the extrapolation of the fit to bright galaxies @xcite . yet all these observed relations presumably stem from a single underlying physical relation .
@xcite and @xcite suggested that baryonic mass is a more fundamental quantity than luminosity . for
any given stellar population , different mass - to - light ratios will be produced in different bands by the same mass of stars .
more importantly , physics should not distinguish between mass in stars and mass in other forms . adding the observed gas mass to the stellar mass results in a baryonic tully - fisher relation ( btfr ) that is linear ( in log space ) over many decades in mass @xcite .
the btfr appears to be the fundamental physical relation underpinning the empirical tully - fisher relation . as such ,
the true form of the btfr is of obvious importance . in this paper ,
i calibrate the btfr with gas rich galaxies .
the masses of gas rich galaxies can be more accurately measured than those of the star dominated galaxies that traditionally define the tully - fisher relation because they are less affected by the systematic uncertainty in the stellar mass - to - light ratio @xcite .
this results in a purely empirical calibration with a minimum of assumptions . in 2 ,
i discuss the data required construct the btfr , assemble samples of gas rich galaxies , explore the statistics of the data , and calibrate the relation . in
3 i examine the implications of the relation for @xmath0cdm and mond .
i define a feedback efficacy parameter @xmath4 to quantify the mapping between observations and theoretical dark halo parameters .
i estimate the value of the mond acceleration constant @xmath5 from the empirical btfr
. a summary of the conclusions are given in 4 .
in order to construct the btfr , it is necessary to estimate the baryonic mass of a galaxy on the one hand , and its rotation velocity on the other .
the baryonic mass is the sum of all observed components , stars and gas : @xmath6 combining this with measurements of the rotation velocity , one obtains the btfr : @xmath7 remarkably , this simple formula suffices to describe the data for rotating galaxies .
there is no indication of any need for a third parameter such as radius or surface brightness @xcite : the tully - fisher relation is the optimal projection of the fundamental plane of disk galaxies .
the stellar mass of a galaxy is determined from its measured luminosity and an estimate of its stellar - mass - light ratio @xmath8 : @xmath9 the mass - to - light ratio is commonly obtained from stellar population models ( e.g. , * ? ? ?
* ; * ? ? ?
these predict correlations between @xmath8 and color so that the observed color is used to estimate @xmath8 .
the use of a stellar mass - to - light ratio estimated from stellar population models causes a significant systematic uncertainty in the stellar mass @xcite , which is particularly sensitive to the imf .
depending on the choice of model , one can derive btfrs with slopes anywhere in the range of @xmath10 to 4 @xcite .
this is a basic limitation that is hard to avoid in samples of spiral galaxies where most of the baryonic mass is in the form of stars .
this can be overcome by use of samples of galaxies whose baryonic mass is dominated by gas rather than stars @xcite . in such gas
rich galaxies , the uncertainty in stellar mass is reduced to a minor contributor to the total error budget .
nevertheless , i am careful to employ a self - consistent population model to all galaxies employed in the analysis ( table [ datatable ] ) .
galaxies with @xmath11 can be found among very late type ( sd and sm ) spirals and irregulars with rotation velocities @xmath12 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . with enough such galaxies ,
it is possible to obtain an absolute calibration of the btfr that is effectively independent of the stellar mass estimator @xcite . here
i describe gas mass estimates for galaxies used in the analysis .
neutral atomic ( hi ) gas usually dominates the gas component of disk galaxies .
the mass of hi follows directly from the physics of the spin - flip transition of hydrogen , the cosmic hydrogen fraction , and the distance to each galaxy .
there is no uncertain mass - to - light ratio as with stars .
the gas mass is estimated by @xmath13 where @xmath14 is the observed atomic gas mass .
the factor @xmath15 accounts for the cosmic hydrogen fraction and forms of gas other than hi . since the majority of objects of interest here are low luminosity late type galaxies , a primordial hydrogen fraction is assumed ( @xmath16 ) rather than the commonly assumed solar value ( @xmath17 ) .
this difference is less than the typical uncertainty in @xmath14 , but does make the numbers here differ slightly from those in the source data .
after atomic hydrogen , molecular gas is the next significant gas component .
molecular gas persists in being difficult to detect in late type , low surface brightness galaxies . where detected , it appears to contain relatively little of the interstellar gas mass @xcite .
such direct measurements of molecular gas are not available for most galaxies . to estimate the molecular gas mass
, we might resort to scaling relations for @xmath18 such that @xmath19 @xcite showed that @xmath18 depends on morphological type , albeit with large scatter . from their data , @xcite derived the scaling relation @xmath20 .
the vast majority of of gas rich galaxies are very late types ( @xmath21 ) , so the typical correction is small ( @xmath22 ) .
we can use the recent observations of things galaxies @xcite to provide an independent estimate of @xmath18 .
@xcite use the observed ( low ) star formation in dwarf irregular galaxies to infer the amount of molecular gas present .
this maps smoothly into the observed molecular content of brighter galaxies , with all disks following a relation between surface densities such that @xmath23 ( their equation 33 ) .
this is a relation between local surface densities , while a relation between global masses is desired here . the atomic gas is typically spread over a much larger area than the stars
, so we can not simply assume that the ratio of surface densities is the same as the ratio of masses .
however , the size of the stellar and molecular disk are usually similar @xcite . assuming that they are the same gives @xmath24 in this case
, the amount of molecular mass depends on the stellar mass rather than the atomic gas mass and morphological type as in equation ( [ etaeqn ] ) . in order to apply equation ( [ mh2mhi ] ) ,
we need an estimate of the average hi surface density .
this is almost always @xmath25 in low surface brightness galaxies @xcite . in the prototypical case
ddo 154 , @xmath26 at all radii @xcite . taking this value as typical implies that molecular gas is perhaps 9% of the stellar mass of these systems . for galaxies with less than half of their mass in the form of stars
, we again infer that the molecular gas makes up @xmath22 of the baryonic mass . for either of these estimators ,
the molecular gas content of hi - dominated late type galaxies is small .
so small , in fact , that it is less than the uncertainty in the measurement of the hi mass in most cases . this would seem to be a good working definition of negligible , so rather than make an uncertain correction for the molecular gas mass , we neglect it entirely . in effect , we assume a constant @xmath27 . we restrict our baryonic mass budget to the components that are actually detected , as discussed above .
it is of course possible that other reservoirs of baryons exist in some undetected form .
there is , however , no direct evidence for baryons in any form that is comparable in mass to the stars and atomic gas . one conceivable reservoir of baryonic mass is hot gas in the halo .
such a component is expected in some galaxy formation scenarios .
while some hot gas is certainly present , it appears that its density is too low to make a substantial contribution here @xcite .
another suggestion is that some of the dark matter could be in the form of undetected cold molecular gas in galaxy disks @xcite .
@xcite showed that a marginal improvement in the scatter of the btfr of @xcite could be obtained by treating @xmath15 as a free parameter , and suggested that the best fit value ( @xmath28 ) implied additional dark baryonic mass in the disk . in more accurate data ,
the scatter is consistent with being caused entirely by observational uncertainties @xcite , nullifying this motivation for dark baryons . in sum
, there is no compelling evidence for substantial reservoirs of baryons in disk galaxies in any form other than those that are directly detected .
i therefore consider only the physically motivated value of @xmath27 .
obviously , a different result would be obtained if we allowed @xmath15 to include hypothetical mass components ( cf .
* ; * ? ? ?
. the characteristic circular velocity of a disk galaxy can be measured with resolved rotation curves , or approximated by line - widths .
the latter is a flux - weighted integral over the velocity field , and lacks the clear physical meaning of a rotation curve .
it is , however , readily obtained from single disk 21 cm observations , whereas obtaining and analyzing interferometric data cubes is considerably more effort intensive ( e.g. , * ? ? ?
* ; * ? ? ?
when a full rotation curve is available , one can make different choices about where to measure the circular velocity .
common choices include the maximum observed velocity ( @xmath29 , e.g. , * ? ? ?
* ) , that in the outermost measured regions where rotation curves tend towards becoming approximately flat ( @xmath30 , e.g. , * ? ? ?
* ; * ? ? ?
* ) or the velocity at some particular optical radius ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the gas rich galaxies of greatest interest here typically have gradually rising rotation curves that slowly bend over toward becoming flat without ever having an intermediate peak . for this common morphology , all these measures of velocity are very nearly equivalent
. in particular , @xmath31 , and the velocity measured at a particular optical radius is also the same unless that radius is chosen to be very small ( see * ? ? ?
indeed , it is not particularly meaningful to define an optical radius for gas rich galaxies , as the scale length of the baryonic mass can be considerably larger than that of the stars alone @xcite . for quantitative analysis ,
i restrict consideration to cases where @xmath30 is explicitly measured , simply adopting the measured velocity from each source . for comparison purposes ,
i also show line - width ( @xmath32 ) data , but do not include these data in the calibration of the btfr as line - widths and rotation velocities are comparable but not identical measures . in order to make comparisons , i assume
@xmath33 for gas dominated galaxies with line - width measurements on the presumption that they would show rising rotation curves like those observed for comparable galaxies . for the star dominated galaxies ,
i assume @xmath34 .
this qualitatively accounts for the fact that bright galaxies have rotation curves that peak before falling gradually toward flatness @xcite .
the quantitative factor is chosen to reconcile the btfr fit to @xmath35 ( * ? ? ? *
@xmath36 ) with that fit to @xmath30 ( * ? ? ?
* @xmath37 ) .
this is only done to facilitate comparison between independent samples . to account for non - circular motions ,
i accept whatever prescription was used by each source . in general
, these corrections turn out to be rather small ( cf . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , as the typical velocity dispersion of low mass galaxies is @xmath38 @xcite .
since the correction to the circular velocity occurs in quadrature , it is only somewhat important only in the slowest rotators .
there exist dynamically cold , rotating disks down to at least @xmath39 , and perhaps as low as @xmath40 @xcite .
there is an enormous amount of data in the literature , but only a finite amount of it pertains to gas dominated galaxies with @xmath11 that also have all the required information for constructing the btfr .
i describe here the accumulated data to date .
i also include associated samples of star dominated galaxies .
this is done strictly for comparison as the uncertainties for star dominated galaxies are dominated by the choice of stellar imf .
this is the systematic uncertainty that we eliminate by making use of gas dominated galaxies . for gas rich galaxies ,
the major sources of uncertainty are the 21 cm flux measurement and the distance to each galaxy .
the uncertainty in the measurement of @xmath30 is usually much smaller than that in mass ( table [ datatable ] ) , provided that the inclination is reasonably well constrained .
cases where this condition might not be satisfied are discussed and , if suspect , excluded from fitting .
the data of @xcite , @xcite , and @xcite constitute the sample of gas rich galaxies for which all the necessary information is available to construct the btfr .
for these galaxies , stellar mass is , by definition , a minority of the baryonic mass . nevertheless , we take care to estimate stellar masses self - consistently . for specificity
, we use their observed @xmath41 colors and the population model of @xcite with a kroupa imf @xcite .
other plausible choices make no difference to the result because the gas mass dominates @xcite .
several other samples are included for comparison @xcite .
since these data are not used in the analysis , we make no effort to reconcile them to the same population model and simply adopt the stellar mass estimated by each source . despite the minor differences in the methods used to estimate stellar mass , these data are remarkably consistent with one another .
[ [ section ] ] @xcite : + + + + + + + + data for remarkably slowly rotating galaxies are presented and analyzed by @xcite .
the corrections for non - circular motions are substantial in some cases . however , the most significant difference in the analysis here is that we determine the gas mass as discussed in [ gasmasssect ] , and we select galaxies for consistency in their inclination determinations .
it is not uncommon for the optical portions of gas rich galaxies to be much smaller than their dominant hi disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and for the outer hi isophotes to have a different shape than the optical isophotes @xcite . if the stars suffer an oval distortion ( bar ) , then an inclination based on optical isophotes will be systematically overestimated @xcite .
indeed , @xcite observe that the optically inferred inclinations often exceed those inferred from hi .
one must be wary that beam smearing might make the hi isophotes rounder than they should be , potentially causing an underestimate of inclination . to avoid either effect
, i demand consistency between optical and hi inclination determinations .
specifically , i require that @xmath42 be within 12% of @xmath43 .
the restriction is placed on @xmath44 because this is what matters to the accuracy with which @xmath30 can be measured .
we are concerned with the velocity at large radii , so the hi inclination is adopted ( as in * ? ? ?
* ) ; the difference @xmath45 is treated as an additional source of error .
the particular threshold of 12% is determined by trial and error , balancing the desire for accuracy with the need to retain at least some objects . with this criterion ,
16 objects are retained from table 1 of @xcite and 12 are rejected .
requiring any greater accuracy eliminates nearly the entire sample .
ugc 5456 is excluded for being marginally star dominated .
three objects are also in the sample of @xcite .
[ [ section-1 ] ] @xcite : + + + + + + + + data for gas rich galaxies are selected from the literature with an emphasis on data quality .
@xcite required resolved rotation curves that were observed to flatten sufficiently that the logarithmic slope became @xmath46 in their outer parts ( see * ? ? ? * from which much of this sample originates ) .
this requirement prevents the data from skewing to low @xmath30 due to the inclusion of galaxies whose rotation curves are still rising rapidly at the last measured point .
lccccr ddo 210 & 0.94 & @xmath47 & @xmath48 & @xmath49 & 1,2 + cam b & 3.34 & @xmath50 & @xmath51 & @xmath52 & 1,2 + ugc 8215 & 4.5 & @xmath53 & @xmath54 & @xmath55 & 1 + ddo 183 & 3.24 & @xmath56 & @xmath57 & @xmath58 & 1 + ugc 8833 & 3.2 &
@xmath59 & @xmath60 & @xmath61 & 1 + d564 - 8 & 6.5 & @xmath62 & @xmath63 & @xmath64 & 3 + ddo 181 & 3.1 & @xmath65 & @xmath66 & @xmath67 & 1 + p51659 & 3.6 & @xmath68 & @xmath69 & @xmath70 & 1 + kk98 246 & 7.83 & @xmath71 & @xmath72 & @xmath73 & 1 + ugca 92 & 3.01 & @xmath74 & @xmath75 & @xmath76 & 1 + d512 - 2 & 14.1 & @xmath77 & @xmath78 & @xmath79 & 3 + ugca 444 & 0.95 & @xmath80 & @xmath81 & @xmath82 & 2 + kk98 251 & 5.6 & @xmath80 & @xmath83 & @xmath84 & 1 + ugc 7242 & 5.4 & @xmath85 & @xmath86 & @xmath87 & 1 + ugc 6145 & 7.4 & @xmath88 & @xmath89 & @xmath67 & 1 + ngc 3741 & 3.0 & @xmath90 & @xmath57 & @xmath91 & 1,2 + d500 - 3 & 18.5 & @xmath92 & @xmath93 & @xmath94 & 3 + d631 - 7 & 5.5 & @xmath95 & @xmath96 & @xmath97 & 3 + ddo 168 & 4.3 & @xmath98 & @xmath99 & @xmath100 & 2 + kkh 11 & 3.0 & @xmath101 & @xmath102 & @xmath70 & 1 + ugc 8550 & 5.1 & @xmath103 & @xmath104 & @xmath105 & 2 + d575 - 2 & 12.2 & @xmath106 & @xmath107 & @xmath108 & 3 + ugc 4115 & 7.5 & @xmath109 & @xmath110 & @xmath111 & 1 + ugc 3851 & 3.2 & @xmath112 & @xmath113 & @xmath114 & 2 + ugc 9211 & 12.6 & @xmath115 & @xmath116 & @xmath117 & 2 + ngc 3109 & 1.3 & @xmath118 & @xmath119 & @xmath120 & 2 + ugc 8055 & 17.4 & @xmath121 & @xmath122 & @xmath123 & 1 + d500 - 2 & 17.9 & @xmath124 & @xmath125 & @xmath126 & 3 + ic 2574 & 4.0 & @xmath127 & @xmath128 & @xmath129 & 2 + ugc 6818 & 18.6 & @xmath130 & @xmath131 & @xmath132 & 2 + ugc 4499 & 13.0 & @xmath133 & @xmath134 & @xmath135 & 2 + ngc 1560 & 3.45 & @xmath136 & @xmath137 & @xmath138 & 2 + ugc 8490 & 4.65 & @xmath139 & @xmath140 & @xmath141 & 2 + ugc 5721 & 6.5 & @xmath142 & @xmath143 & @xmath144 & 2 + f565-v2 & 48 . & @xmath145 & @xmath146 & @xmath147 & 2 + f571-v1 & 79 .
& @xmath148 & @xmath149 & @xmath150 & 2 + ic 2233 & 10.4 & @xmath151 & @xmath152 & @xmath153 & 2 + ngc 2915 & 3.78 & @xmath154 & @xmath155 & @xmath156 & 2 + ngc 5585 & 5.7 & @xmath157 & @xmath158 & @xmath159 & 2 + ugc 3711 & 7.9 & @xmath160 & @xmath161 & @xmath162 & 2 + ugc 6983 & 18.6 & @xmath163 & @xmath164 & @xmath165 & 2 + f563-v2 & 61 . &
@xmath166 & @xmath167 & @xmath168 & 2 + f568 - 1 & 85 . & @xmath169 & @xmath170 & @xmath171 & 2 + f568 - 3 & 77 . & @xmath172 & @xmath173 & @xmath174 & 2 + f568-v1 & 80 . & @xmath175 & @xmath176 & @xmath177 & 2 + ngc 2403 & 3.18 & @xmath178 & @xmath179 & @xmath180 & 2 + ngc 3198 & 14.5 & @xmath181 & @xmath182 & @xmath183 & 2 + [ datatable ] [ [ section-2 ] ] @xcite : + + + + + + + + new wsrt observations of 11 very gas rich galaxies selected from the sample of @xcite and @xcite are used to measure both @xmath32 and @xmath30 . of
the 11 galaxies observed , 6 show clear evidence of rotation .
interestingly , these all have hi profiles that are double - horned or at least show some structure .
galaxies for which rotation is not positively detected tend to have single - horned profiles . @xcite
discuss two methods for measuring the outer circular velocity that are largely consistent .
i adopt for @xmath30 their values measured from the position - velocity diagram ( @xmath184 from their table 3 ) . for d575 - 2 , i estimate @xmath185 based on the outermost detected 21 cm emission in their data .
[ [ section-3 ] ] @xcite : + + + + + + + + the data for star dominated galaxies with @xmath30 measured from extended hi rotation curves are shown for comparison .
they are not used in the analysis here . for illustration ,
stellar masses adopt the @xmath186 mond mass - to - light ratios .
the btfr fit to these data [ @xmath187 can be tested by extrapolation to the much lower circular velocities of the new gas rich galaxies .
the three data sets of @xcite , @xcite , and @xcite are the most interesting for our purposes here .
they compose a sample of gas dominated galaxies with resolved rotation curve measurements .
these data are listed in table [ datatable ] and are used in the analysis .
the data for these galaxies is less precise than those for bright star dominated galaxies .
however , they are more accurate , as their masses are not sensitive to systematic uncertainties in the imf and stellar population mass - to - light ratios . samples of galaxies with line - width data are used for comparison purposes only .
we shall see that line - width data give a systematically shallower slope and greater scatter for the btfr than do rotation curve data .
this appears to be a limitation of the line - width as a proxy for the rotation velocity . aside from this detail ,
the data present a consistent overall picture .
[ [ section-4 ] ] @xcite : + + + + + + + + gas rich galaxies were selected from the sample of @xcite and @xcite to have inclinations @xmath188 .
stellar masses were estimated from @xmath189-band data assuming a constant mass - to - light ratio ( @xmath190 ) .
we retain this original prescription , making only the correction to stellar mass discussed by @xcite .
data for star dominated galaxies with @xmath191-band photometry were taken from @xcite . a constant @xmath191-band mass - to - light ratio of @xmath192 was assumed .
these data are included here for comparison to more recent data but are not used in the analysis .
[ [ section-5 ] ] @xcite : + + + + + + + + new data are presented for galaxies with @xmath11 .
detailed stellar population modeling has been performed to determine the stellar mass ( table 11 of * ? ? ?
the same modeling procedure has also been applied to the star dominated sample of @xcite .
we adopt the stellar mass determinations based on @xmath191-band photometry - band data are used to estimate the stellar mass for the bright galaxies of @xcite makes a difference to the slope of the btfr : @xmath10 in the optical and @xmath193 in the @xmath191-band .
this exemplifies the systematic uncertainty that gas dominated galaxies save us from . ] for these galaxies ( table 4 of * ? ? ?
* ) to facilitate direct comparison with the data of @xcite employed by @xcite .
all of the bright galaxies of @xcite have nicely double horned line profiles with well measured line - widths .
this is not the case for the new gas rich galaxies presented by @xcite , for which the hi profiles are all narrow and single - horned ( gurovich , private communication ) .
such profiles can be poor indicators of @xmath30 @xcite .
[ btffancyseb ] shows the data . over all
, the agreement between the various data is good .
this is true even for the star dominated galaxies , despite the differences in the population modeling .
the spiral galaxies of @xcite fall in the midst of the data of @xcite ( inset of fig .
[ btffancyseb ] ) with indistinguishable slope and normalization .
the chief difference is the lower scatter in the data of @xcite .
this presumably results from the more accurate distances to the individual galaxies based on hst observations @xcite .
this is expected : as the data improve , the scatter is reduced .
indeed , essentially all of the scatter can be accounted for by observational uncertainty and the expected variation in stellar mass - to - light ratios @xcite .
the scatter is larger for lower luminosity galaxies .
this is also expected : it is more challenging observationally to obtain precise data for faint dwarfs than it is for bright spirals .
nevertheless , the data for the gas rich dwarfs appear to be largely consistent .
moreover , the scatter is reduced as we progress from line - widths to resolved measures of the rotation velocity ( left vs. right panels in fig .
[ btffancyseb ] ) . the improved accuracy with which the rotation velocity is measured
is reflected in reduced scatter .
there are some outliers among the gas rich dwarfs , especially when @xmath32 is used as the velocity estimator .
the two most dramatic outliers from the sample of @xcite are ugc 3755 and kk98 - 65 , which fall far to the left of the main correlation .
these are rejected by the requirement of consistent optical and hi inclinations as are a number of galaxies that do lie along the btfr in fig .
[ btffancyseb ] when the hi inclination is used .
the outliers from @xcite are those objects for which rotation is not clearly detected .
( d575 - 5 is the most discrepant case ) .
three objects from the sample of @xcite stand out from the rest of the data : eso 085-g 088 , am 0433 - 654 , and hipass j1424 - 16b .
eso 085-g 088 has a large velocity error , so its status as an outlier is only marginal .
the other two objects clearly deviate from the rest of the data , including the bulk of the data of @xcite . since we do not have resolved measures of rotation for these objects , it is possible that they , like the single - peaked hi profiles of @xcite , would not show clear evidence of rotation if so observed .
another possibility is that the optical inclinations used by @xcite are systematically in error as discussed in [ rcsamples ] .
lacking resolved hi observations , we can not make the same consistency check that is possible with the data of @xcite .
the inclination of am 0433 - 654 is @xmath194 and that of hipass j1424 - 16b is @xmath195 .
a shift of @xmath196 dex in @xmath32 would reconcile these galaxies with the other data .
if inclination is responsible , the corresponding inclinations would be @xmath197 for am 0433 - 654 and @xmath198 for hipass j1424 - 16b . from experience with such objects ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , this is well within the realm of possibility .
indeed , it would be surprising if something like this did not happen with only optical inclinations to work with . over all ,
the data paint a consistent picture ( fig .
[ btffancyseb ] ) despite the inevitable concern over the accuracy of a few individual measurements .
the btfr is continuous and linear in log space over five decades in mass .
there is a hint of curvature when @xmath32 is used as the rotation velocity measure , but it is not statistically significant .
there is not even a hint of curvature when resolved rotation curves are available to measure @xmath30 .
we expect the accuracy of the data to improve as we move from integral measures like the line - width to physically meaningful rotation velocities measured from resolved hi data cubes .
this expectation is realized in the reduced scatter seen in the right panel of fig .
[ btffancyseb ] .
we can quantify this with ordinary least - square fits to the data .
if we apply the maximum likelihood method of @xcite to fit the new line - width data ( i.e. , all the data in the left panel of fig .
[ btffancyseb ] except those of * ? ? ?
* ) , we obtain a slope @xmath199 and an intercept @xmath200 . if we perform the same exercise with the rotation curve data in the right panel of fig .
[ btffancyseb ] , we obtain @xmath201 and @xmath202 .
these formal fits have rather different slopes even though the chief difference is in the scatter .
the method of @xcite accounts for uncertainties in both velocity and mass , and can optionally estimate the amount of intrinsic scatter in a sample .
accounting for the stated observational errors ( e.g. , * ? ? ?
* ) implies an intrinsic scatter in the line - width data of @xmath203 dex .
that is , there is 0.2 dex of scatter not explained by observational uncertainty . in contrast , this number is zero for the rotation curve data in the right hand panel : observational errors suffice to explain all the scatter , leaving no room for intrinsic scatter .
this difference in scatter implies some source of variance in line - width measures that is perhaps not accounted for in the stated uncertainties , and not likely to be meaningful to the physics we are interested in here .
at least part of the reason for the difference in fitted slopes is physical : @xmath32 is not a perfect proxy for @xmath30 .
bright galaxies typically have peaked rotation curves with @xmath204 .
line - width measurements do not resolve where emission originates , and are thus sensitive to the maximum velocity rather than that at larger radius .
this immediately skews the slope to shallower values . a similar if less pronounced effect can be seen in optical rotation curves that are resolved but not particularly extended @xcite . for the physical interpretation of the btfr
, we wish to have as precise a probe as possible of the quantity of interest . for @xmath0cdm
, this would be the virial velocity of the dark matter halo .
in the case of mond , we need a measure that probes the low acceleration regime . in both cases
, @xmath30 provides a better measure than @xmath32 .
the slope of the btfr has been addressed by many authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and is typically found to be between @xmath10 and 4 .
most of this variation is due to the prescription adopted for estimating the mass - to - light ratio @xcite , which is by far the strongest systematic effect .
this can be remedied by selecting gas dominated galaxies as described in [ stellarmasses ] .
@xcite found a slope @xmath205 with a bisector fit while @xcite found @xmath206 from their gas dominated galaxies ( their table 12 ) .
these results appear to be contradictory . repeating the analysis of @xcite
, i obtain the same result . however , simply looking at the data , it is not obvious that they are in conflict .
part of the difference is simply the difference in velocity estimators , as line - widths give shallower slopes than resolved rotation velocities .
their sample is also small , so is easily influenced by a few outliers as discussed above ( [ datacomp ] ) .
this difference between @xmath207 and 4 is precisely the difference between the nominally anticipated slopes in @xmath0cdm and mond .
the correct answer is therefore of considerable importance and lends itself to controversy .
bear in mind that the data themselves are quite consistent ( fig .
[ btffancyseb ] ) , so any controversy is in the fitting and choice of velocity estimator , not in the data .
the line - width is a convenient proxy for the rotation velocity when resolved rotation curves are unavailable , but not a substitute for them when predictive theories are being tested . to illustrate the difference between slopes of 3 and 4
, we rotate the btfr into a new coordinate system ( @xmath208,@xmath209 ) parallel to it such that @xmath208 measures the position of a galaxy along the btfr and @xmath209 its deviation perpendicular to it .
that is , @xmath210 where @xmath211 is the slope of the btfr . in fig .
[ rotate ] we show the data rotated into the frames defined by the fits of @xcite with slope @xmath212 and @xcite with @xmath213 .
a slope of 4 does a good job of describing the bulk of the data : there is no systematic deviation from the dotted line in the lower panel of fig .
[ rotate ] .
in contrast , there is an obvious residual from a slope of 3 in the the top panel .
the data show a clear preference for the steeper slope .
[ rotate ] also illustrates the systematic caused by different mass - to - light ratio estimates for star dominated galaxies .
the galaxies from @xcite are shown twice , once with masses based on @xmath191-band data ( table 4 of * ? ? ?
* ) and again with the average of @xmath214 and @xmath191 band stellar masses ( table 5 of * ? ? ?
* ) . the hybrid @xmath214-@xmath191 masses are nicely consistent with a slope of 3 , while @xmath191 alone prefers a slope of 4 .
while one must choose which band to trust for bright galaxies , this choice is rendered irrelevant for gas dominated galaxies .
we therefore concentrate the rest of our analysis on the gas rich galaxy sample described in [ rcsamples ] and [ stellarmasses ] . taking the 47 gas rich galaxies with reasonably trustworthy data ( table [ datatable ] ) ,
the best fit slope is @xmath215 with intercept @xmath216 .
various methods ( forward , reverse , and maximum likelihood ) all give the same result . of these data ,
the velocity measurements of @xcite are the most challenging : only three of their objects meet the quality criteria of @xcite . excluding the data of @xcite but retaining those of @xcite and @xcite gives @xmath217 from a forward fit , @xmath218 from a reverse fit , and a bisector slope @xcite of @xmath219 .
the corresponding intercepts are @xmath220 , 1.82 , and 1.71 , respectively , with uncertainty @xmath221 .
the slight difference that occurs when also fitting the data of @xcite may result from a slight skew to low velocities , and testifies to the difficulty in obtaining quality data for very slow rotators ( see also * ? ? ?
the skew effect becomes more pronounced if we relax the selection criterion that requires consistent optical and hi inclination determinations .
irrespective of these details , these fits are all consistent with one another ( fig .
[ btfgasfit ] ) , and none yield a slope that is meaningfully different from 4 . an interesting question we can pose to the gas rich galaxy data is what the normalization of the btfr is with slope fixed to 4 .
the gas rich galaxies provide an absolute calibration of the btfr that is very nearly independent of the estimator we use for the stellar mass - to - light ratio . by fixing the slope
, we can sharpen our estimate of the intercept , and also examine higher order moments of the distribution .
in addition to the scatter perpendicular to the btfr , @xmath222 , we can also examine the skew @xmath223 and kurtosis @xmath224 .
table [ fittab ] shows these statistics for several combinations of the rotation curve samples .
we start with the sample of @xcite as their sample was constructed for this exercise . adding the galaxies of @xcite decreases the intercept slightly . adding the data of @xcite raises again , back near to the value found by (
* @xmath225 ) .
table [ fittab ] gives quantitative estimates for the intercept .
for each combination of samples both a weighted and unweighted estimate is given .
the weighted estimate comes from minimizing @xmath226 by adjusting @xmath227 with fixed slope : we minimize the orthogonal deviations .
the unweighted estimate is made by finding the centroid of the distribution in the orthogonal deviations @xmath209 of the rotated data and requiring the centroid to lie at @xmath228 .
the latter method gives no heed to the accuracy of the data , but is less sensitive to the influence of outliers whose formal uncertainty may understate systematic effects .
these methods give indistinguishable results .
lrccccccc s & 28 & 1.04 & 1.71 & 1.65 & @xmath2290.07 & @xmath2290.14 & 1.18 & 1.36 + st & 34 & 0.92 & 1.67 & 1.64 & @xmath2290.01 & 0.01 & 1.30 & 1.39 + stb & 47 & 0.99 & 1.69 & 1.71 & @xmath2290.13 & @xmath2290.10 & 1.24 & 1.18 [ fittab ] all combinations of the various samples give consistent estimates for the intercept . to be specific , we adopt the weighted estimate from the combination of the stark - trachternach sample : @xmath230 this is the midpoint of the various determinations and has the smallest @xmath231 .
this value is of course very similar to the unweighted value ( @xmath232 ) obtained from the @xcite sample alone @xcite .
it is also similar to the value ( @xmath233 ) found by @xcite from star - dominated galaxies , but has the virtue of being completely empirical with little sensitivity to how stellar masses are estimated .
it appears that at this juncture , systematic effects and the necessary assumptions are as limiting as data quality .
it is traditional to assume @xmath234 which is appropriate for solar metallicity atomic gas , but is slightly larger than the primordial 1.33 taken here for these little evolved , gas rich , low mass galaxies .
the choice of @xmath234 is not important to the star dominated galaxies in @xcite , but we note that adopting it here would cause a change in mass of at most 5% ( for purely gaseous galaxies ) that would drive @xmath235 .
this is at the level of our best guess in the mass of molecular mass . at this level
the systematic uncertainty in the imf that plagues stellar mass estimates also begins to play a role even in gas dominated galaxies . given these considerations , any value in the range 45
@xmath233 is plausible , with values much outside this range quickly becoming less plausible . if we use the btfr to estimate stellar masses @xcite , then more massive star dominated galaxies are consistent with a kroupa imf .
grossly different stellar masses would imply a break in the slope of the relation around @xmath236 where the galaxy population transitions from frequent to infrequent gas domination .
the scatter in our rotated coordinates is @xmath237 with only the third digit varying from sample to sample . in terms of mass
this is 0.24 dex .
the bulk of this uncertainty ( 0.20 dex ) is provided by the error in the baryonic mass .
velocity uncertainties are less important in most samples , and only rival the uncertainty in mass in the lowest mass systems .
together , the uncertainties in mass and rotation velocity suffice to account for essentially all of the observed scatter .
a robust estimator of the dispersion of a sample is the median absolute deviation ( mad ) .
the dispersion of the various samples about the btfr as measured by their mad is shown in fig .
[ scatter ] .
also shown is the median measurement uncertainty @xmath238 . for data that scatter about the relation simply because of observational uncertainty , @xmath239 1.48mad as
n becomes large presuming there are no systematic uncertainties that have not been accounted for .
[ scatter ] shows that indeed @xmath240 1.48mad in most cases .
this leaves little , if any , room for intrinsic scatter in the btfr .
indeed , the only times when scatter measured by the mad significantly exceeds that expected from measurement errors are in the cases where we have identified systematic uncertainties that are not included in the error estimate : in the subset of the sample of @xcite where no signature of rotation has been detected , and in the sample of @xcite where the optical and hi inclination estimates differ greatly .
had we been unaware of these sources of systematic error , we might infer some finite intrinsic scatter .
instead , it appears that including less accurate data merely increases the scatter as one would expect .
the btfr of gas rich galaxies appears to have essentially zero intrinsic scatter .
the same conclusion follows from noting that @xmath226 is close to unity ( table [ fittab ] ) .
a finite intrinsic width to the relation would cause @xmath241 @xcite .
we can place an upper limit on the scatter by assuming the data are gaussian ( a dubious but conservative assumption for this purpose ) and subtracting the measurement error in quadrature from 1.48mad .
this gives a limit on the intrinsic scatter of @xmath242 dex in mass for the @xcite sample and @xmath243 for the trimmed sample of @xcite .
essentially zero intrinsic scatter is inferred by this method for the gas rich data of @xcite or the star dominated galaxies of @xcite . for the data of @xcite , the expected error exceeds the observed scatter .
this might indicate that the uncertainties ( especially in distance ) have been overestimated , but the sample is small so we should not expect the assumption of a normal distribution to hold . indeed , that is generally true here , which mitigates against there being much intrinsic scatter .
it is common in extragalactic astronomy for correlations between physical quantities to exhibit lots of scatter . in the btfr
we seem to have a relation with little or no intrinsic scatter .
there are some possible couple of caveats to this .
first , it is conceivable that the uncertainties in the data have been overestimated .
second , selection effects might limit the sample to only those objects that obey the btfr .
overestimation of the uncertainties seems unlikely . that @xmath244 is a sign not only that the model is a reasonable description of the data , but also that the uncertainties have been neither over nor underestimated .
even if the data were arbitrarily accurate , so that all of the observed scatter is intrinsic , the intrinsic width of the relation is still small : the observed 0.24 dex .
if anything , this would appear to confirm that the disks of the selected galaxies are well behaved dynamically cold rotators .
if it were otherwise , e.g. , if there were substantial non - circular motions in the disks of these galaxies , then this would be a substantial source of intrinsic scatter and the observed scatter would be much larger @xcite .
it is more difficult to address the issue of potential selection effects .
however , in the time since @xcite , the persistent surprise has been how tight the tully - fisher relation remains when challenged with data for new types of galaxies that we would naively expect to deviate from it . since this exercise
has now been performed by many different observers for many different types of galaxies , it seems unlikely that selection effects play a significant role in suppressing the scatter of the btfr .
we select gas rich galaxies here , but it is already well established that star dominated galaxies obey the btfr with little scatter ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the only further selection made here is on data quality . when lower quality data are admitted , the scatter increases ( fig .
[ btffancyseb ] and [ scatter ] ) , just as one would expect .
there is no clear evidence for intrinsic scatter beyond that induced by observational uncertainty and the systematic effects inevitable in astronomical data .
it is impossible to estimate the scatter that might result from including objects that are not already included .
we have restricted our discussion to rotating galaxies , since these are the objects that define the tully - fisher relation .
it is not obvious that it should apply to non - rotating galaxies , though the faber - jackson relation for elliptical galaxies provides an obvious analog .
dwarf spheroidal galaxies fall remarkably close to the btfr with a simple estimate of the equivalent rotation velocity ( @xmath245 : * ? ? ?
* ) though the ultrafaint dwarfs fall systematically below it @xcite .
even objects that have no apparent business adhering to to the btfr either do so , or come surprisingly close to doing so ( e.g. , the hi ring in leo ) . among rotating disk galaxies
one would expect differences in the baryonic mass distribution to have some impact on the btfr , just because @xmath246 . and
yet disk dominated spirals , bulge dominated early types , diffuse low surface brightness galaxies , and dwarf irregulars all adhere to the same btfr in spite of their palpably different mass distributions @xcite .
rather than finding greater scatter as we expand our samples beyond the sc i galaxies traditionally utilized in distance scale studies , we find that rotating galaxies of all types adhere to the same btfr .
it therefore seems unlikely that there remains some large , unknown population of galaxies in rotational equilibrium that would induce substantial scatter in it .
a persistent problem with interpreting astronomical data is whether gaussian statistics actually apply .
one reason for this is that systematic errors often dominate random ones .
we can check whether this might be an issue here by examining higher order statistics like the skew ( lopsidedness ) and kurtosis ( pointiness ) of the distribution in the deviations @xmath209 perpendicular to the btfr .
if the data are well behaved ( i.e. , distributed normally ) , the skew and kurtosis should be small . for all three combinations of samples in table [ fittab ] , both the skew and the kurtosis are small : @xmath247 and @xmath248 .
these samples are consistent with normal distributions as they should be if random errors dominate over systematic effects .
this is especially true for the combination of the @xcite and @xcite samples , which have @xmath249 . for comparison ,
the sample of @xcite has @xmath250 and @xmath251 around their best fit . if we trim the three outlying galaxies discussed in
[ datacomp ] , the skew is reduced ( @xmath252 ) but the kurtosis grows ( @xmath253 ) .
the negative kurtosis means that the distribution is flat - topped : it does not have a well - defined central peak as it should if the data simply scatter around the correct relation .
these data are not distributed normally .
this undermines the basic assumption of the least square fitting method , so the apparently straightforward fit of @xcite yields a misleading constraint on the slope .
understanding the physical basis of the btfr is central to one of the most fundamental issues presently confronting extragalactic astronomy .
the btfr provides a simple test that in principle can distinguish between @xmath0cdm and mond @xcite .
here we discuss the interpretation of the btrf in terms of each paradigm in turn .
the interpretation of the btfr in @xmath0cdm depends on how we relate dark and baryonic mass .
different assumptions can plausibly be made , leading to rather different interpretations .
here we start from general considerations , later proceeding to specific models .
it is conventional in cosmology to refer to structures by the density contrast they represent with respect to the critical density of the universe .
the mass enclosed within a radius encompassing the over - density @xmath254 is @xmath255 with @xmath256 , this becomes @xmath257 by the same token , the circular velocity of a tracer particle at @xmath258 is @xmath259 .
consequently , @xmath260 in @xmath0cdm , the density contrast @xmath261 marks the virial extent of a dark matter halo @xcite . for @xmath262 , @xmath263
this includes all mass , dark and baryonic , that reside within the radius @xmath264 .
dark matter halos thus obey a scaling relation reminiscent of the tully - fisher relation ( e.g. , * ? ? ?
this is not the same as the observed tully - fisher relation , as both mass and velocity refer to a radius @xmath264 that is much larger than what is observed . in order to map these theoretical quantities into the observed baryonic mass @xmath265 and circular velocity @xmath30
, we need to introduce some factors : @xmath266 and @xmath267 the total mass of baryons available within @xmath264 is @xmath268 , where @xmath269 is the cosmic baryon fraction ( @xmath270 : * ? ? ?
the factor @xmath271 represents the fraction of these available baryons that contribute to the observed baryonic mass .
similarly , the observed velocity need not be identical to @xmath272 ; hence the factor @xmath273 .
the btfr would follow from equation ( [ darktf ] ) with a slope of @xmath274 and a normalization @xmath275 if @xmath276 .
this is shown as the dashed line in fig .
[ btfcdm ] .
had this matched the data , it would provide a very satisfactory interpretation @xcite .
instead , it exceeds the data , leading us to consider models with @xmath277 and @xmath278 .
equations [ darktf]-[fveqn ] to provide a generic constraint on @xmath0cdm galaxy formation models .
these typically invoke feedback to suppress the cold baryon content of disks so that @xmath277 .
we define a feedback efficacy parameter @xmath279 when @xmath280 , @xmath4 is just the inverse of @xmath271 , so a galaxy with @xmath281 has lost 90% of its baryons , and one with @xmath282 has lost 99% of them . by
we mean that they are not part of the disk of stars and cold gas that contribute to @xmath265 in the btfr .
it is irrelevant whether these baryons have been physically expelled as envisioned in many feedback models , or remain in the halo in some dark , unobserved form .
@xmath4 merely tells us how effective feedback ( or any other mechanism ) has been in suppressing the inclusion of the available baryons into the observed disk .
in addition to variations in@xmath271 , the mapping of the observed to virial velocity @xmath273 can also play a role .
this ratio can easily vary with mass as more massive disks boost @xmath273 by engendering more adiabatic contraction in their host dark matter halos than low mass disks ( e.g. , * ? ? ?
my own experience with modeling adiabatic contraction @xcite leads me to expect this to be a modest effect , most likely in the range @xmath283 .
very recently , @xcite have combined dynamical and gravitational lensing measurements to estimate @xmath284 for galaxies with @xmath285 .
this represents the upper end of the btfr ( roughly @xmath286 ) . in order to help with the observed variation in @xmath4
, @xmath273 would need to be substantially less in lower mass galaxies .
while these often have rising rotation curves ( so possibly @xmath287 ) , flatness of the rotation curve is an explicit selection criterion of the @xcite sample . for these galaxies
, @xmath288 seems the best guess , as their disks are not massive enough to substantially alter their dark matter halos to yield @xmath289 , nor does it seem likely that the rotation curve will resume rising after being observed to flatten out so as to yield @xmath287 . for the very low mass galaxies of @xcite ,
@xmath287 is certainly possible , but it seems unlikely to be tiny . if @xmath290 at @xmath291 and 1.3 at @xmath292 , the total variation in @xmath4 is a factor of 3 .
this does not , by itself , suffice to explain the data . equation [ eqn : efficacy ] maps between the parameters of the dark matter halo and those of the disk galaxy it hosts .
it is completely general , encompassing what needs to happen to connect theory with observational reality . in order to be consistent with the data presented here
, models must meet the requirement @xmath293 this constraint applies to models of disk galaxies over the range @xmath294 . in applying this constraint
, care should be taken to use consistent definitions of total mass ( @xmath295 with @xmath296 ) and rotation velocity .
the term involving @xmath254 is a simple shift in normalization that translates to other common definitions of halo mass ( e.g. , @xmath297 instead of 100 ) .
the constraint holds irrespective of the particular choice of definition for halo mass .
equation [ finetuning ] is a scaling between @xmath4 and rotation velocity . that the efficacy of feedback scales with velocity is not surprising , as the shallower potential wells of low mass halos should be less effective at retaining baryons .
the efficacy of feedback should also depend on the physics driving it . for low mass systems like those under consideration here
, feedback is often imagined to be driven by star formation and subsequent supernova events . in this case , we might expect some correlation between @xmath4 and measures of star formation .
[ efficacy ] shows @xmath4 as a function of past average star formation rate and gas mass fraction .
the gas fraction @xmath298 while to obtain the past average star formation rate we assume all galaxies have an age of 12 gyr so that @xmath299/(12 gyr ) .
both of these quantities correlate with @xmath4 , albeit with substantial scatter at low @xmath300 and high @xmath301 .
the curious thing about these correlations is that feedback has been most effective in removing or suppressing baryons in galaxies that have experienced the least star formation and have the most cold gas remaining .
this would seem to be the opposite of what one would expect from a picture in which intense star formation results in winds that drive out baryons .
the effect is not subtle .
galaxies with @xmath302 have @xmath303 ; those with @xmath304 have @xmath305 .
it appears that the vast majority of baryons associated with the halos of these galaxies have been ejected or otherwise suppressed .
a more remarkable thing is that @xmath4 scales so precisely with @xmath30 . in order not to impart scatter to the btfr
, there must be very little scatter in equation [ finetuning ] .
i would expect feedback from supernovae to be a chaotic process : the lack of scatter in @xmath4 seems unnatural .
a halo of a given circular velocity could easily host galaxies of very different @xmath4 .
apparently , they do not all halos of the same @xmath30 have basically the same @xmath4 .
the tight relation between @xmath30 and @xmath4 provides strong constraints on models of galaxy formation .
indeed , it seems to me that it constitutes a serious fine - tuning problem .
one can readily envision models that obtain equation [ finetuning ] as a mean scaling relation ( e.g. , * ? ? ?
. however , doing so with so little scatter is a much tougher requirement .
i would expect @xmath271 and @xmath273 to fill out some distribution of values ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , even for halos of fixed mass .
there is no obvious reason why these distributions should be either narrow or closely coupled .
yet they must be in order to avoid introducing substantial scatter into the btfr .
stochastic feedback is not the only expected source of scatter in the btfr .
the scatter in the concentration circular velocity relation for dark matter halos is another . without some tuning @xcite ,
this alone exceeds the scatter allowed by the error budget .
another expected source of scatter is the triaxiality of dark matter halos ( e.g. , * ? ? ?
if the gravitational potential is non - symmetric , then the observed velocity depends on the viewing angle with respect to the principle axis of the potential .
otherwise identical galaxies will appear to have difference @xmath30 depending on how they happen to be oriented on the sky .
a significant degree of non - axisymmetry is clearly excluded @xcite .
this constraint might be lessened by the rounding of dark matter halos by the formation of the baryonic disk , but the disk needs to be heavy enough to have an impact .
@xcite find that disks need to be at least half maximal to have a significant impact on their halos .
the low surface brightness disks that make up the majority of the gas rich galaxy sample typically fall a factor of two below this level @xcite .
there are other factors though could , and probably should , contribute to the scatter @xcite .
it is difficult to accommodate all the expected sources of scatter in the error budget .
this holds even if we have _ overestimated _ the observational uncertainties , simply because the observed scatter is small .
the small intrinsic scatter in the btfr appears to be a serious challenge for @xmath0cdm .
this is hardly the only problem faced by the theory ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , so the persistent failure to detect wimps is worrisome .
the @xcite has excluded much of the mass cross section parameter space where wimps were expected to reside @xcite . given the persistence of dynamical puzzles ( e.g. ,
* ; * ? ? ?
* ) , one might worry that non - baryonic dark matter does not exist after all .
the generic constraints discussed in the previous section may or may not be adequately addressed by specific models .
there are very many relevant models in the literature .
it is well beyond the scope of this paper to review all of them .
nevertheless , it is useful to examine several illustrative cases .
@xcite built a model following the same logic outlined above .
they define a parameter @xmath306 that is the fraction of the total mass in the disk .
this is analogous to @xmath271 , such that @xmath307 when @xmath308 .
the resulting tully - fisher relation ( their equation 4 ) is shown as lines in the first panel of fig .
[ btfcdm ] for @xmath309 ( as they suggest ) and for @xmath310 ( as suggested by * ? ? ? * ) .
the higher value gives a tolerable approximation to high mass spirals while the lower value is appropriate to gas dominated dwarfs .
a single universal value , as often assumed in galaxy formation modeling , does not work simply because the slope of the btfr is steeper than the @xmath0cdm halo mass rotation velocity relation .
more recent models include semi - analytic treatments ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and fully numerical results ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these are shown in the remaining panels of fig .
[ btfcdm ] .
all invoke feedback in some way , but the prescription for feedback is specific to each so some come closer to matching the constraint on @xmath4 than others .
the model of @xcite exhibits a substantial scatter and dependence on the distribution of stellar mass .
this is natural : the distribution of baryonic mass should matter @xcite .
this model matches the high mass end of the observed btfr reasonably well , but over - predicts the mass in galaxies with @xmath311 .
nevertheless , this model , like that of @xcite , provides an honest example of what we most naturally expect from @xmath0cdm : a substantial amount of scatter about a relation with a slope of approximately 3 .
the model of @xcite also indicates a shift in normalization of the relation for galaxies residing in sub - halos relative to that of primary galaxies ( their fig .
this shift is already apparent at relatively large @xmath312 .
there is no evidence of this in the data , which includes both primary galaxies and dwarfs that are sometimes isolated and sometimes group members ( and hence sub - halos ) .
for example , kk98 - 251 is a dwarf irregular satellite of ngc 6946 , but both lie on the same btfr with no indication of a shift in normalization .
current data for the ultrafaint dwarf satellites of the milky way do place them below the btfr , but this shift occurs at much smaller scales ( @xmath313 * ? ? ?
* ; * ? ? ?
* ) . the model of @xcite exhibits significant curvature .
it matches the data well at @xmath314 but falls below it at both high and low @xmath30 .
there is no evidence in the data for curvature in the btfr of rotating galaxies .
it is of course conceivable that curvature may appear beyond the bounds of the current data , and may already be observed if one considers dwarf spheroidals along with dwarf irregulars @xcite .
this occurs at a significantly lower velocity scale ( @xmath315 ) than in the model of ( * ? ? ?
* @xmath316 ) . the model of @xcite fares
better , being in rather good agreement with the bulk of the data for gas rich galaxies . like most @xmath0cdm models , its slope is relatively shallow ( @xmath317 ; their table 3 ) , but this is only problematic at the high mass end where the model under - predicts the mass of star dominated galaxies .
this might be remedied by adopting a different imf , but the change would have to be substantial : a factor of 3 or so in stellar mass .
the scatter is respectably small .
most recently , detailed numerical simulations have produced model galaxies consistent with the btfr @xcite .
these are shown in the final panel of fig .
[ btfcdm ] . here
we plot the velocity measured at the peak of the baryonic rotation curve @xmath318 rather than @xmath30 because this is closer to what is reported by @xcite . even though there is not a great deal of difference between these two measures of rotation velocity , the already good agreement between the models and the data shown by @xcite improves even further when @xmath318 is utilized .
numerical simulations of this sort remain computationally expensive .
@xcite present eight model galaxies , which at present is a large number for such models @xcite . of these eight ,
seven are very nicely consistent with the observed btfr . as they discuss
, the other case is very discrepant .
obviously , no conclusions can be drawn from one case in eight .
but it does raise some obvious and important questions .
what is the intrinsic scatter of the btfr ? how often we should expect to see grossly discrepant cases ?
would we recognize these objects as normal disk galaxies ?
resolution also remains an issue in numerical modeling .
the model galaxies of @xcite are all in the star dominated regime .
only the one model galaxy of @xcite resides in the regime of gas dominated galaxies where the comparison is most clean .
comparison with the millenium simulation suggests that simulated galaxies might well follow the observed slope and normalization of the btfr down to low rotation velocities ( white & guo 2011 , private communication ) .
the critical issue of the intrinsic scatter that should be exhibited by low mass model galaxies in such simulations remains open .
detailed numerical simulations appear to have finally reached the point where they can produce quantitatively realistic model galaxies .
this is clearly a great success .
it remains to be seen if they can provide a satisfactory explanation for the small intrinsic scatter in the btfr .
if they can , then the question becomes whether feedback prescriptions that accomplish this feat correspond to how feedback works in nature . among alternatives to dark matter , one , mond @xcite , makes strong _ a priori _ predictions about the btfr , which is a consequence of the form of the force law in mond .
this deviates from purely newtonian at small accelerations , @xmath319 , where @xmath5 is the one new parameter introduced by the theory .
the value of @xmath5 must be constrained by observation ( @xmath320 : * ? ? ?
* ) but once specified is constant . in the deep modified regime , @xmath321 , the effective acceleration @xmath322 , where @xmath323 is the newtonian acceleration calculated for the observed baryonic mass in the usual way . for circular motion around a point source , we can equate the centripetal acceleration to this effective force to obtain @xmath324 since both @xmath5 and @xmath325 are constants , and all mass is baryonic , one recognizes the btfr , @xmath326 .
there are several consequences of equation [ eqn : mondtf ] @xcite .
the tully - fisher relation is absolute , being a direct consequence of the force law .
there should be no intrinsic scatter .
the normalization is set by constants of nature .
contrary to the case in newtonian dynamics , the circular velocity does not depend on the radial extent of the mass distribution , so there is no dependence on surface brightness or scale length .
this naturally explains the lack of observed size or surface brightness residuals from the tully - fisher relation @xcite .
indeed , @xcite specifically predicted that low surface brightness galaxies would share the same tully - fisher relation , with the same normalization , as high surface brightness spirals .
this prediction was subsequently confirmed @xcite .
the data for gas rich galaxies now provides the opportunity to test another _ a priori _ prediction .
the circular velocities of the gas rich dwarfs considered here are measured .
their baryonic masses are observed .
their location on the btfr is fixed .
consequently , the gas rich dwarfs provide a test of the mond prediction with _ zero _ free parameters .
the results of this test is shown in fig .
[ btfmond ] . even though the value of @xmath5 was set by early fits to bright spirals ( @xmath327 : * ? ? ?
* ) , equation [ eqn : mondtf ] fits the data for much lower mass galaxies well .
the gas rich dwarfs follow the prediction of mond with no fitting whatsoever @xcite .
there was no guarantee that mond would pass the test posed by gas rich galaxies .
a tully - fisher type relation is built into the mond force law , but its slope is fixed and its normalization is constrained by star dominated galaxies . the data for gas rich galaxies could have fallen anywhere in the btfr plane .
they might have exhibited the slope of 3 that is more natural to @xmath0cdm .
instead , they fall along the one and only line permitted in mond . the data for gas rich galaxies are consistent with existing estimates of the acceleration constant @xmath5 @xcite .
we can reverse the question , and ask what the best value of @xmath5 is according to these data .
the best fit intercept of the btfr already contains this information , and the corresponding value of @xmath5 is given in table [ fittab ] .
there are some subtle issues to consider in calibrating @xmath5 .
this has previously been done by fits to full rotation curves ( e.g. , * ? ? ?
* ; * ? ? ?
these contain more information than simply the outer circular velocity , but require an additional assumption .
in particular , a form for the function that interpolates between the newtonian and mond regimes must be assumed .
for an assumed form , the best fit mass - to - light ratio must be found simultaneously with the best fit value of @xmath5 . there is not much range allowed in either of these parameters , but there can be covariance between them , so it is challenging to disentangle them uniquely . here
, the data are not so precise , and we make no use of the information in the shape of the rotation curve , only its outer value . however , this approach has the virtue of making a minimum of assumptions and using mass estimates that are completely independent of the theory .
strictly speaking , equation [ eqn : mondtf ] applies at infinite distance from an isolated point mass .
real galaxies are not point masses , and @xmath30 is measured at a finite radius .
consequently , the intercept of the empirical btfr is related to the acceleration constant of mond by @xmath328 where @xmath329 is a factor of order unity that accounts for the finite size and non - spherical geometry of galaxies .
@xcite estimated @xmath330 from purely geometrical considerations thin disks rotate faster than the equivalent spherical mass distribution . the sample discussed by @xcite
has mond fits performed for @xmath327 and a separate fit for the btfr yielding @xmath37 .
this gives an empirical estimate of @xmath331 which i adopt here .
the gas rich galaxy data thus give @xmath332 the uncertainty here includes the formal uncertainty in @xmath333 ( equation [ bestfita ] ) plus an allowance for 20% uncertainty in @xmath329 .
it is conceivable that there could be a small systematic offset in the mean value of @xmath329 between gas and star dominated galaxies .
detailed rotation curve fits to some of the same galaxies suggest @xmath334 @xcite .
this becomes @xmath335 if we exercise the same quality control on the data as applied here .
this gives some idea of the systematic uncertainty and the impact of the best - fit stellar mass - to - light ratio when it is treated as a fitting parameter along with @xmath5 .
however , data quality remains the paramount issue .
the btfr is clearly a success of mond , despite its many other drawbacks .
further tests are possible , as adherence to the btfr does not guarantee that it is possible to obtain an acceptable fit to the entire rotation curve of each galaxy . while mond is known to do quite well in some gas rich galaxies ( e.g. , * ? ? ?
* ) , it may have problems in others ( e.g. , * ? ? ?
* where again the issue appears to be with the inclination ) .
it is particularly challenging to fit the full rotation curve of ngc 3198 @xcite , the most massive galaxy to meet the gas rich criterion imposed here .
perhaps this suffices to falsify mond , but we should not be so eager to disbelieve mond as a theory that we ignore the remarkable empirical virtues of the simple formula suggested by @xcite .
there are other suggestions for modifying gravitational theory besides mond .
conformal weyl gravity @xcite and mog @xcite offer two examples .
if these theories can fit rotation curves @xcite , then they will also fit the btfr @xcite .
however , i am not aware of a clear _ a priori _ prediction of this phenomenon by any theory other than mond .
the baryonic mass velocity relation provides an important test of @xmath0cdm and alternatives like mond .
i have assembled recent data for gas rich rotating galaxies in order to test these theories . with @xmath11 , the location of these objects in the baryonic mass rotation velocity plane
is effectively measured directly from observations without the usual systematic uncertainty in the stellar mass - to - light ratio .
a long standing prediction @xcite of mond is that rotating galaxies will fall on a single mass velocity relation with slope 4 : @xmath326 .
this prediction is realized in multiple independent data sets .
gas rich galaxies fall where predicted by mond with no free parameters .
there are not many predictions in extragalactic astronomy that fare so well over a quarter century after their publication . in @xmath0cdm ,
the naive expectation of a of tully - fisher type relation with a slope around 3 fails to predict the location of gas rich galaxies in the plane of the btfr . only the most recent detailed numerical models ( e.g. , * ? ? ?
* ; * ? ? ?
* ) succeed in reproducing the observed phenomenology .
this is certainly progress .
however , a physical understanding for why galaxy formation in the context of @xmath0cdm should pick out the particular phenomenology predicted _ a priori _ by mond remains wanting . | the baryonic tully - fisher relation ( btfr ) is an empirical relation between baryonic mass and rotation velocity in disk galaxies .
it provides tests of galaxy formation models in @xmath0cdm and of alternative theories like mond .
observations of gas rich galaxies provide a measure of the slope and normalization of the btfr that is more accurate ( if less precise ) than that provided by star dominated spirals , as their masses are insensitive to the details of stellar population modeling .
recent independent data for such galaxies are consistent with @xmath1 with @xmath2 .
this is equivalent to mond with @xmath3 .
the scatter in the data is consistent with being due entirely to observational uncertainties .
it is unclear why the physics of galaxy formation in @xmath0cdm happens to pick out the relation predicted by mond .
we introduce a feedback efficacy parameter @xmath4 to relate halo properties to those of the galaxies they host .
@xmath4 correlates with star formation rate and gas fraction in the sense that galaxies that have experienced the least star formation have been most impacted by feedback . |
the loss of gas that signifies the end of star formation in an embedded proto - cluster may also cause the young cluster to dissolve @xcite .
the energy output from all the massive stars typically exceeds the total binding energy of the embedded star cluster .
the survivability of the cluster therefore depends on the efficiency with which the radiative , thermal and mechanical energy of the stellar outflows couple to the inter - cluster gas ( and as long as this energy is not carried away by outflows , * ? ? ?
the time scale on which gas is removed from the embedded cluster ( and any spatial dependence of this process ) affects the mass and number of stars in the surviving cluster , but also its degree of mass segregation and the density profile .
the parameters that strongly affect the outcome of the out - gassing phase are the star formation efficiency ( sfe , which by itself could depend on feedback processes , * ? ? ?
* ; * ? ? ?
* ) , the efficiency of the radiative coupling and the population of the most massive stars in the proto cluster .
traditionally the critical value for the star formation efficiency is taken to be @xmath2 on the basis of virial arguments ( e.g. * ? ? ?
* ; * ? ? ?
* ) , but later work showed that sfes as low as 0.1 can still lead to bound clusters , albeit with a reduced mass @xcite . observed star formation efficiencies fall into the range 0.05 - 0.3 @xcite .
it is therefore expected that clusters must experience considerable structural disruption before settling . from an observational point of view , the number of embedded clusters is too high with respect to the number of observed gas - free clusters . within the solar neighborhood
the observed distribution of cluster ages suggests that more than 95% of embedded clusters dissolve into the field within 100 myr @xcite .
the disruption of young clusters by the loss of the remnant gas provides a natural explanation , although alternatively the young clusters may be disrupted by tidal forces @xcite . in this scenario
it is not the internal evolution of the cluster that disrupts the structure , but tidal shocks from passing gas clouds ( see also * ? ? ?
in order to find out what the relative contribution is from these two processes we need to understand the effects of gas loss quantitatively .
ultimately the modeling of the formation of cluster populations is needed to interpret the observations as they play a pivotal role in galaxy evolution studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . without understanding early cluster evolution
we can not relate the young cluster populations ( which are easily observed to large distances ) to the older stars and the stellar population in the field .
contrary to stars , clusters are sensitive to the galactic environment and if we would understand the starting conditions of the cluster population we could search for the signatures of later galactic evolutionary events such as galaxy mergers and major gas inflows . the expulsion of gas from young clusters has been studied by applying a time - dependent background potential within which the n - body system is integrated @xcite . the time scale for out - gassing
can then be controlled rather straightforwardly by the decay time scale of the background potential .
this parametrization allows the study of different mass loss scenarios without the complications of modeling gas dynamics and feedback .
@xcite examined the evolution of a cluster both using the external potential approximation and for a limited set of combined n - body/ sph simulations ( using simplified feedback prescriptions ) .
they found general agreement between the simulations with a dynamic gas component and a static gas potential , and found that relatively long timescales of gas loss are necessary for clusters to survive . to reconcile observed low sfes with their simulations ,
they point out that although the global sfe may be low in a star forming region , locally the sfe may be much higher and allow for bound clusters ( also * ? ? ?
furthermore they proposed sub - virial stellar velocity dispersions as an alternative possibility ( see also * ? ? ?
@xcite conducted a grid of simulations in which they varied the sfe , gas expulsion timescale and external tidal field .
they confirmed that clusters with a sfe as low as 10% could survive , although they found a strong dependence on the timescale ( parametrized by an e - folding time ) on which the gas potential was removed .
surviving clusters had expanded drastically ; for sfes of around 25% this was a factor 3 - 4 , nearly independent of the gas loss timescale .
the velocity distributions became strongly radially anisotropic in the cases where the gas was removed on timescales much shorter than a crossing time .
it has become clear from observations and simulations that young stars start out in a structured , clumpy , state @xcite .
recent work examining hierarchical cluster formation @xcite suggests that sub - clustering increases the survivability .
if clusters tend to be born sub - virial , they can survive even the loss of large fractions of intercluster gas , in which case the sfe may not be a good indicator of survivability .
@xcite ( see also * ? ? ?
* ) found that the local gas fraction immediately prior to gas loss is a better diagnostic for the cluster survival than the global sfe . by studying sub - clusters in hydrodynamical simulations of collapsing molecular clouds where sink particles represent stars
, @xcite found that the environment around @xmath3 clumps has a gas content of only 10%-30% .
the relative lack of gas resulted from the accretion of gas onto stars ( which depletes the gas reservoir ) and the shrinkage of the virialized sub clumps ( because the dynamics of these clumps decouples from the gas dynamics ) .
for such low gas fractions no major disruption is expected when feedback clears the gas . on the larger scales gas fractions
are still considerably higher .
@xcite argues that a similar effect will also cause the larger scales to become gas poor , if sufficient time elapses before the onset of feedback .
however , during this delay the different clumps would merge and relax , evolving from the highly structured state towards a smooth density distribution ( this was also the case for the simulations for @xcite , where this happened within 3 myr ) .
the conditions at this stage would determine the outcome with regard to the final survival , and by this stage the local depletion of gas by accretion may no longer be present .
the above studies have expanded considerably on the simple picture where clusters dissolution is determined solely or mainly by the single sfe parameter .
clusters seem to have a number of ways in which they can resist the disruptive effect of the loss of their gas content : disruption is less or absent if the gas loss occurs on timescales longer than the crossing time or if the gas loss is delayed enough that shrinking due to relaxation processes can occur . on the other hand , if the stellar population of the cluster forms sub - virialized , some of the stars in the cluster can evolve into low energy orbits and these will not be ejected as easily , even if the sfe is low . hence , to constrain the disruption process of clusters , we need to employ numerical simulations to unravel various complex physical processes at work .
the determination of the gas loss timescale and the spatial distribution requires careful modeling of the response of the parent gas cloud to the mechanical luminosity of the young stars using realistic stellar evolution models ( the feedback from radiation , stellar winds and supernovae will be neither constant nor instantaneous ) coupled to a hydrodynamic code . following the evolution of the stellar component
requires that the stellar dynamics be resolved in high detail , including two - body relaxation processes . for this an n - body code for collisional gravitational dynamics
is necessary .
the relaxation processes also require that a realistic initial mass function ( imf ) is adopted , which is also required for the stellar evolution modeling .
the stellar dynamics and hydrodynamics code must be mutually coupled as they feel each other s gravity . here
we will present simulations that were conducted using the newly developed astrophysical multi - purpose software environment ( amuse ; * ? ? ?
* ; * ? ? ?
* ) that couple stellar dynamics of young clusters with the gas dynamics using realistic feedback physics . using these simulations we will explore idealized but realistic scenarios of gas loss and analyze the trends with sfe , feedback strength and the dependence on the initial distribution
. we will start by presenting the amuse package , the solvers and initial conditions of the runs we conducted in section [ sec_method ] .
the results are presented in section [ sec_results ] and discussed in [ sec_disc ] .
we summarize our main conclusions in section [ sec_concl ] .
we study the evolution of embedded star clusters using a hybrid simulation environment .
our models combine the effects of the self - gravity of the stars , their nuclear evolution and the hydrodynamics of the gas .
the latter has contributions of the primordial gas content and of the gas liberated by the stars in stellar winds and supernovae .
the various simulation ingredients are coupled self - consistently using amuse @xcite . in the code , a script in the amuse framework , we couple a collisional n - body code , a smoothed particles hydrodynamics ( sph ) code and a stellar evolution code together to form a complete description of the physics that govern the evolution of an embedded star cluster .
the integration of the equations of motion of the stars was conducted using the fourth - order predictor - corrector hermite n - body solver phigrape @xcite .
the dynamics of the gas was resolved using the sph code gadget-2 @xcite . for the gravitational coupling between the stars and the self - gravity of the gas we used the gravitational treecode octgrav @xcite .
the stars in our simulations were evolved using parametrized stellar evolution tracks ( using the sse code , * ? ? ?
each of the numerical simulation codes is embedded in the amuse framework , and the code that realizes the self - consistent coupling between each of the physical sub - domains turns out to be a relatively straightforward script . within the amuse environment the individual sub - solvers are referred to as _ community codes_. the algorithmes used by these community codes remain the same as described by their respective authors . the codes are written in different languages and have a wide range of programming styles . because the adopted community codes are untouched , we will not further explain their operations , but simply refer the reader to the appropriate literature .
the amuse framework that stitches each of the community codes together is one of the major ingredients that enables us to perform a consistent coupling between the various scales and physical domains .
therefore we give a short overview of the framework in the following section .
the amuse environment that we use to conduct the simulations is a package that allows astrophysical codes from different domains to be combined to conduct numerical experiments ( see pelupessy et al .
2011 for a more complete description ) .
amuse is a development of the multi - physics and multi - scale software environment ( muse , * ? ? ?
* ) and is freely available for download .
the fundamental idea of amuse is the abstraction of the functionality of simulation codes into physically motivated interfaces that hide the complexity and numerical implementation of the codes .
amuse presents the user with building blocks that can be combined into applications and numerical experiments . the binding language that stitches the codes together
is python , as the focus in these high level interactions is not so much performance ( the computational cost being concentrated in the component codes ) but on algorithmic flexibility and ease of programming to allow rapid prototyping .
an amuse application consists , roughly speaking , of a _ user script _ , one or more _ community modules _ and the _ community code _
( fig [ fig_interface ] ) .
the user script specifies the initial conditions and selects the simulation codes .
the relationships between the community codes define the solver and for our case we construct a combined n - body / hydrodynamic solver using a direct n - body code and an sph solver .
the setup and communication with a community code is handled by the community module .
this consists of an mpi based communication interface with the code as well as unit handling facilities and an object oriented data model .
the heart of an amuse application consists of the solvers employed : here these are the gadget-2 sph code @xcite , the phigrape hermite n - body code @xcite , the tree gravity code octgrav @xcite and the stellar evolution code sse @xcite . an overview of the calling sequence of our combined solver is given in figure [ fig_int ] .
it consists of a combined integrator for coupled gas / gravitational dynamics systems and a feedback prescription for mechanical energy input .
the bridge integrator @xcite provides a semi - symplectic mapping for the gravitational evolution in cases were the dynamics of a system can be split in different regimes .
a typical application would be a dense star cluster in a galaxy where the internal dynamics of the former evolves on a relatively short timescale compared to the dynamics of the host galaxy .
a similar idea was applied by @xcite by splitting the gravitational and hydrodynamic evolution operators for simulating gas - rich galaxy mergers .
they implemented the algorithm in a single monolithic code , whereas we adopt the concept of operator splitting within amuse to couple different codes . in our case
we will also employ a split between the gravitational evolution operator of the stellar component ( which is governed by collisional dynamics ) and the hydrodynamic evolution operator ( governed by sph dynamics ) .
the bridge formulation is derived by applying the splitting method developed in planetary dynamics @xcite to the hamiltonian @xmath4 of a system of particles which consists of subsystems a and b by dividing the hamiltonian into three parts @xmath5 where @xmath6 and @xmath7 are the hamiltonians of subsystems a and b respectively and the cross terms are collected in @xmath8 .
the formal time evolution operator of the system can then be written as ( @xmath9 is the hamiltonian vector field corresponding to a hamiltonian @xmath10 ) : @xmath11 the @xmath12 operator represents momentum kicks because @xmath8 depends only on the position of the particles .
the operator @xmath13 splits into secular evolution operators , because @xmath6 and @xmath7 are independent .
the evolution of the total system over a time step @xmath14 can subsequently be calculated from the mutual forces exerted by systems a on b and by kicking the momenta for time @xmath15 , after which the two systems are evolved in isolation for a time @xmath14 , after which the time step is finished by another mutual kick .
this procedure is beneficial if , for example , the time step allowed by the mutual interactions is considerably longer than that required by the internal dynamics of the component systems , or if the gravitational dynamics of the two sub - systems need different solving strategies .
within amuse we view the various gravitational and hydrodynamical codes as implementations of the time evolution operators @xmath16 ( for this the interface specification provides an evolve function ) .
the momentum kicks are easy and reasonably fast to implement within the framework in python once the forces are known , and for this the gravitational dynamics interface provides utility functions to query the gravitational forces at arbitrary positions .
the gas dynamical sph evolution can also be derived from a hamiltonian formalism and thus the above directly carries over to a split between purely gravitational and sph particles .
the application of a bridge type integrator to sph is computationally efficient if the time step requirement for the hydrodynamics is more stringent than for the gravitational dynamics ( which generally is the case , * ? ? ?
our application within amuse has the additional benefit that we can easily swap in and out different gravitational solvers and sph codes , even if they use completely different algorithms ( such as in our case a direct hermite code and a tree based sph integrator ) . for validation ,
the resulting combined gravitational - hydrodynamical solver was checked in problems which could also be calculated using a monolithic sph code . for this a number of static and dynamic test runs were conducted and the results where verified to be consistent with runs conducted purely with gadget-2 .
first , in order to check that the alternation of hydrodynamic and gravitational evolution operators does not introduce artifacts in cases where gravitational and hydrodynamic forces are in equilibrium , we conducted static tests where the ability of the code to maintain a plummer sphere in equilibrium was examined .
tests where conducted using different particle numbers ( 10k-200k ) and gas fractions ( 50% and 100% ) .
as a dynamic test we conducted the _
evrard _ @xcite collapse tests with the combined amuse solver .
both the dynamic and static tests showed satisfactory results compared to a conventional integrated code @xcite . in our calculations
we include feedback from stellar wind and supernovae .
radiative processes are not taken into account self - consistently , even though the ionizing radiation from massive stars can drive violent outflows @xcite ; the relative strength of the radiative feedback is expected to be of the same order of magnitude as the mechanical feedback from stellar winds @xcite , which is included in our calculations . given these uncertainties ( and the fact that we do not explicitly include radiative feedback ) we have chosen to simplify the modeling by taking an adiabatic equation of state and not solving the thermal and chemical balance of the gas .
we parametrize the energy budget in the absence of radiative losses with a coupling strength parameter @xmath17 , which is the fraction of the total supernova and wind energy output that ends up as thermal energy in the ism .
the relative strength of the feedback is of the order of a few percent , the rest is radiated away @xcite . for the actual implementation of the feedback within our model we calculate the mechanical luminosity @xmath18 of the stars , from parametrized mass loss rates and terminal wind velocities @xcite .
the mass liberated in the stellar wind of the stars is subsequently injected into the cluster medium .
all gas particles have the same mass @xmath19 throughout the simulation .
the gas particles are released at rest relative to the star releasing them in a sphere with a radius equal to the smoothing length of the gas particles around the star , and the stellar mass is reduced accordingly .
a consequence of this prescription is that each star loses mass in discrete steps of at least @xmath19 .
for the runs presented here the @xmath19 varies between @xmath20 . as a consequence
the stellar wind feedback will exhibit some discreteness noise if the mass loss rate of a star is lower than @xmath21 ( note that the feedback will still be approximately isotropic if a low number of particles is released due to the sph smoothing ) .
we verified that the results of our simulations with respect to the stellar response converges and therefore are not affected by this discretization .
for the most active wind phases ( @xmath22 ) the feedback prescription leads to 10s-100s of particles released per timestep and a smooth wind ( see section [ sec_results ] ) .
the internal energy @xmath23 of the gas lost by the star is calculated from the total energy output of the star @xmath24 since the last discrete mass loss event : @xmath25 here @xmath26 is the dynamical mass and @xmath27 the mass calculated by the stellar evolution code ( so reduced by the mass loss ) . in case that a star experiences a supernova explosion we inject @xmath28 in the escaping particles by added this amount of energy to @xmath24 .
the feedback efficiency parameter @xmath29 that accounts for the radiative losses is a free parameter , which we varied between @xmath30 and @xmath31 . in order to track the rapid evolution caused by feedback
we adopt a maximum time step for the gas particles of @xmath32 years and after supernova events the maximum hydrodynamic time step is reduced to 20 years for 30k years in order to prevent time - stepping artefacts ( a more elegant fix is described in * ? ? ?
* ; * ? ? ?
* but our approach has only a minor effect on the total running time ) .
we summarize the various numerical parameters in table [ table_par ] . [ cols="<,^,^,^,^,^,^,^ " , ] the simulated clusters in table[table_ic ] are composed of a mixture of gas and stars ( with @xmath33 ) distributed in a plummer sphere .
stellar - stellar gravitational interactions are smoothed ( plummer smoothing ) with a smoothing length @xmath34 and the interaction between gas particles and star particles are smoothed ( using the sph spline kernel ) with the gas particle smoothing length .
the stellar masses are assigned using a salpeter imf between 0.1 and 100 @xmath35 , with the additional constraint that the most massive star in each cluster has a mass close to the median maximum mass for a cluster of this size ( @xmath36 ) .
this additional requirement is consistent with the expected maximum mass for small clusters @xcite .
the masses of the stars are assigned independently of the position in the cluster ( constant sfe ) , so no initial mass segregation is imposed .
stellar collisions are not taken into account and during the simulation no gas accretion unto existing stars or new star formation is allowed .
note that any accretion of gas onto the stars would be very limited since we start from a smooth gas distribution and no cooling is included .
feedback induced star formation can also not occur in our approach , but is not very likely in the systems we simulate ( see also * ? ? ?
* ) . in the series of calculations indicated with the letter `` a '' in table[table_ic ] we explore the effect of the formation efficiency , the initial scale length and the feedback efficiency .
we run models with a sfe of 0.05 , 0.3 and 0.5 .
the initial plummer radius of the initial conditions is either 0.1 , 0.5 or 1.0pc .
simulations with a feedback efficiency of @xmath30 and @xmath31 are performed . for small clusters
the number of high mass stars can vary quite substantially between different realizations of the imf , and we perform the model a2 and a5 with realizations of the imf generated from different random seeds to examine this effect .
we have performed additional simulations in which we varied the number of gas particles to tests that the results are independent of the resolution of the gas dynamics .
we also run a series of models in which the stars are dynamically cold ( these models with a sub - virialized velocity dispersion are designated as model set `` b '' in table [ table_ic ] ) .
this choice is motivated by the fact that the initial stellar velocity dispersion comes from turbulent velocities in the parent cloud @xcite .
the stellar component may form with a lower velocity dispersion than that required to be dynamically stable if the turbulence in the parent gas cloud has a scale length smaller than the jeans length or if the gas cloud is ( partially ) pressure supported or partially supported by magnetic fields @xcite .
for this set of simulations we decrease the stellar velocities to @xmath37 of the virialized values .
in figure [ fig_maps ] we show the stellar and gas distribution of our ` baseline ' run a2 and its low feedback equivalent , run a5 . at four moments in time
we plot the stars and a cut through the @xmath38 plane of the gas density distribution .
looking at the first a2 frame ( at 0.96 myr ) we see the early stages where stellar winds blow bouyant bubbles that rise in the potential of the star cluster .
as the mechanical luminosity increases these bubbles grow until they start blowing away sizable fractions of the cluster medium and a free - flowing wind develops ( 4.37 myr frame ) . the strong feedback proceeds to unbind most of the gas of the cluster . at approximately 9.5 myr
the cluster ism has been blown away - the gas that is visible in this frame originates from the strong agb wind of the heaviest progenitor ( 21 @xmath35 ) . the supernova that explodes at 9.54
myr leaves the cluster devoid of gas . for the a5 simulation ( @xmath30 ) ( both simulations use the same initial realizations for the imf and stellar positions )
the initial wind stages proceed less violently , with smaller bubbles ( 0.96 myr ) , while a free - flowing wind does not develop ( compare the frame at 4.37 myr ) until just before the supernova .
the main difference between the a2 and a5 run is that most of the cluster gas remains present in the latter case , until the first supernova , at which stage the remaining gas is blown away .
just before the supernova the cluster is much more compact than in the a2 run .
the corresponding evolution of the lagrangian radii and core radius and central gas density ( of models a1-a10 ) is shown in figure [ fig_lr ] and we plot in figure [ fig_bf ] the ( instantaneous ) bound fraction of the gas and star component . for run a2 , in the early phase ( before 8 myr ) the gas is lost by stellar winds . from the beginning of the simulation
the gas starts to boil off vigorously under the influence of the stellar winds .
the stellar distribution reacts by expanding . within 3 myr
the core radius increases a factor @xmath39 ( within this time approx @xmath40 of the gas is lost ) and stays more or less constant thereafter .
this is also visible in the maps of fig .
[ fig_maps ] , where after the 4.37 myr frame the stellar distribution remains qualitatively similar . fig .
[ fig_bf ] also shows that when the first supernova goes off , the gas in the cluster has already been completely removed .
later supernovae affect the stellar distribution very little .
if we look at the bound fraction of stars shown in figure [ fig_bf ] we see that @xmath41 of the stars remain bound at the end of the simulation .
the escape of stars happens mostly at the time of early gas loss , but with a noticeable delay .
note that the bound fraction plotted here and for the other runs is an upper limit because this fraction is determined from the instantaneously bound stars ( so at a particular point in time ) and does not account for further unbinding by the escape of marginally unbound mass ( this can account for part of the observed delay ) .
comparing runs with different sfe ( models a1,a2 and a3 ) we see that in the case of low sfe ( run a1 ) most of the gas is retained up until the moment of the first supernova ( see fig .
[ fig_bf ] ) .
the gas completely unbinds because of the supernova at 9.54 myr , and the cluster is essentially dissolved immediately at that point .
the high sfe case a3 is very similar to the a2 case
. the initial gas loss may be somewhat steeper , but in this case ( fig .
[ fig_bf ] ) the loss of stars of the cluster proceeds more slowly .
looking at the evolution before the first sn , it is clear that the greater gas mass ( @xmath42 more ) gives the a1 run greater resilience to the disturbance from stellar winds .
considering the escape of stars this leads to counteracting effects : when the fraction of gas is large , the loss of gas can more easily disturb the stellar distribution .
however this loss happens initially much more slowly . in the a2 and a3 runs on the other hand ,
the gas starts to be blown away right from the beginning , but the fraction of gas is small enough not to cause major disruption for the stellar distribution .
when the supernova explodes , the gas of the a1 run is instantaneously removed , leading to the quick dissolution of most of the cluster . on the other hand , about 70% of the stars
is retained in the a3 run .
if we look at the runs with the lower effective feedback strength ( a4 , a5 and a6 ) , we see that in the sfe @xmath43 case ( a4 ) gas is still present after the first and second supernova .
the stellar component reacts completely differently for the a4 run ( compared with a1 ) : in this case @xmath44 of the stars remain bound initially , where only @xmath45 remains in the a1 run . in the plot of the lagrangian and core radii of run a4 ( fig .
[ fig_lr ] ) the core radius initially increases after the supernova , but then grows smaller again , as the cluster relaxes - markedly different from the continuing expansion in the a1 case . for the other two runs ( a5 and a6 ) , the lower effective feedback strength mainly results in a much less violent boil off compared with the high @xmath29 cases ( e.g. a2 vs a5 ) .
gas is still present at the time of the first supernova . for the stellar distribution
this means that initially the stellar content remains intact ( in particular the a5 run shows an almost constant bound fraction for the stellar component up until 9.5 myr ) .
the final bound stellar fractions and lagrangian radii are qualitatively similar .
the reason for this is that in both cases the total feedback energy is sufficiently large to unbind the cluster gas eventually .
much of the extra energy in the simulations with high feedback is carried away by the escaping gas . in the high feedback case gas loss by stellar winds
is much more important , whereas in the low feedback case gas is driven out by supernovae .
the final stellar distribution of the final cluster is little affected by the feedback efficiency over the range explored here , unless the gas fraction is very high . for the low mass clusters we examine here different realizations of the stellar density distribution and the imf will have very different masses and orbits for the heavy stars ( and thus in the number , timing and location of supernovae ) just due to low number statistics ( the number of stars larger than @xmath46 is @xmath47 ) .
we have run additional realizations of the a2 and a5 simulations to quantify this effect .
for the a2 runs the range in final stellar bound fractions is @xmath48 , with the final core radii lying between 0.6 and 1.2 parsec .
for the a5 runs we find similar ranges , @xmath49 and @xmath50 parsec . however
if we compare the runs in more detail we see considerable differences .
two examples of this are given in figures [ fig_lr2 ] ( lagrangian and core radii ) and [ fig_bf2 ] ( bound fractions ) .
the a2c run quickly loses @xmath51 of its mass within 1.5 myr , after which the gas loss slows down momentarily .
this results in a stellar distribution that quickly expands , but then contracts again until the remaining gas is expelled .
the a5d run seems to follows a similar pattern to the a5 run , if one compares the core radius .
however if one looks more closely to the inner lagrangian radii the a5d run is very different from the a5 run . in the latter case the ( e.g. ) @xmath45 lagrangian radius hovers around 0.2 parsec after 10 myr , while in the a5d run it expands more than @xmath52 further , to @xmath53 parsec .
the time evolution of the lagrangian radii and the bound fractions of the models with sub - virialized velocity dispersion are plotted in figures [ fig_lrb ] and [ fig_bfb ] .
figure [ fig_lrb ] shows that the stellar distributions quickly shrink compared to the virialized cases : e.g. for b1 the 0.5 lagrangian radius drops roughly a factor 2 within less than 0.5 myr .
this collapse of the stellar distribution can increase the survival of the stellar cluster , because the collapse increases the local sfe @xcite .
the b1 model shows this effect , because compared with the a1 model it shows a much less drastic expansion of the core radius ( fig .
[ fig_lrb ] ) , although in the end the cluster still becomes completely unbound ( fig .
[ fig_bfb ] ) .
however , surprisingly , if we look at the b2 model with a lower @xmath29 ( which is the sub - virialized version of the a4 run ) we see exactly the opposite : in this case the cluster dissolves quickly , where as one may expect on the basis of the a1 and a4 run that the b2 run would dissolve much more slowly
. runs b3 and b4 show a similar pattern .
the reason for this becomes clear if we examine figure [ fig_fs ] , where we have plotted the stellar mass fraction ( the ratio of stellar mass to total mass ) within the stellar half mass radius as function of time up to the time of the first supernova .
the dotted lines here show the a1 and a4 runs , while the drawn lines are the b1 and b2 runs .
the b1 and b2 runs initially show an increase of a factor of 2 in stellar fraction ( which is smaller than expected from the contraction of the half mass radius shown in fig .
[ fig_lrb ] , indicating the gas distribution also contracts ) .
after that they both show a slow increase , but crucially the b1 run shows a stronger increase to around @xmath54 vs the b2 run .
although the increase is modest , it is probably the cause of the difference between the b1 and b2 run .
this is consistent with the @xcite results , which show a very sensitive dependence of the bound fraction on sfe around a value of 0.25 .
note that figure [ fig_fs ] shows that the stellar fraction of the a2 run is almost the initial value ( sfe@xmath43 ) . in this case
the survival of the cluster after the first supernova is not due to a high stellar mass fraction , but due to the fact that not all gas is expelled ( fig .
[ fig_bf ] ) .
the effective sfe hence increases to sfe@xmath55 ( measured after 1 myr ) if one takes into account the gas that is not expelled .
hence the b1 and a2 run show an increased survivability due to completely different mechanisms : the b1 stellar component clears its environment of gas slowly through stellar winds and adapts to the slowly changing potential , while the stellar component of the a2 run is held together after the first supernova by remnant gas ( in spite of this only being a very minor fraction of the initial gas component )
. visual inspection of the maps such as those in figure [ fig_maps ] shows that the resulting remnant cluster are mass segregated . a useful way to quantify the mass segregation
was presented by @xcite - they plotted the number fraction enclosed at given radii @xmath56 against the enclosed mass fraction at those radii ( @xmath57 ) .
for mass segregated clusters ( e.g. the pleiades cluster in their paper ) @xmath57 will lie above @xmath56 ( an example of this procedure for our simulation a2 is given in fig .
[ fig_gini ] ) .
a measure of the mass segregation can then be given by calculating the area between the graph of @xmath57 and line of equality , which converse & stahler labeled the gini coefficient @xmath58 for the cluster .
we have plotted @xmath58 for various runs as a function of time in figure [ fig_gt ] .
looking at run a5 ( solid red line ) we see that in the beginning @xmath58 rises quite quickly , within 5 myr ; after that @xmath58 is more or less constant , @xmath59 . at the end of the simulations
the cluster is hence in a mass segregated state .
a similar picture emerges for run a6 ( higher sfe , dotted line in fig .
[ fig_gt ] ) and for the higher feedback run ( a2 , dashed line ) . in the latter case @xmath58 plateaus at a somewhat lower value .
the other two drawn lines in figure [ fig_gt ] show the runs with a higher and lower initial half mass radius , runs a10 and a8 .
as is expected in these cases the initial rise in @xmath58 is slower ( a10 ) and faster ( a8 ) respectively . in all cases
the final @xmath58 settles around @xmath60 .
it is interesting to note that @xcite tried to model the pleiades cluster in some detail by running pure n - body models , and they could not explain the observed @xmath58 unless they started their simulations with a nonzero gini coefficient of at least @xmath60 .
although our models do not try to specifically reproduce the pleiades cluster , this seems consistent with figure [ fig_gt ] : @xcite started their simulation after the dispersion of gas and our models suggest that the nonzero @xmath58 is a remnant of the fast evolution before gas dispersal .
after gas dispersal the cluster is left in a much more dispersed state ( the models of @xcite had virial radii @xmath61 pc ) , with a @xmath58 close to its present value .
this suggests that the mass segregation in open clusters such as the pleiades is not a result of the secular evolution of its present configuration , but a remnant of its embedded evolution stage .
the fact that mass segregation in our simulations occurs early in the embedded stage of the cluster evolution also has a consequence for the stellar population of the remnant cluster . in figure [ fig_cbf ]
we plot the number of stars heavier than a given mass which end up in the remnant cluster ( i.e. are bound ) as a function of star mass .
the panels in figure [ fig_cbf ] show that in most cases where there results a bound cluster , the heavier stars are preferentially retained ( 11 out of 15 runs show this pattern ) .
a number of models do not show this pattern , especially a7 and a8 .
note these clusters have drastically expanded .
note also that while the a7 and a8 runs do not show strong preferential retention of heavy stars , they do show mass segregation .
it was previously noted @xcite that simple virial theorem arguments underestimate the size of clusters after gas clearing , because there are always a number of stars in low velocity orbits . the preferential retention of heavy stars can be understood by noting that the mass segregation during the initial evolutionary stages happens because of the partitioning of stars in equal energy orbits : the heavy stars end up in low velocity orbits that are much more difficult to unbind .
the heavy stars that are more easily retained may skew the imfs that are derived from intermediate age clusters ( as long as they are not observed in the initial stage ) and this may result in an observable signature .
note that this could be tested for the pleiades cluster , which is already known to have an excess of heavy stars in the center ( * ? ? ?
* e.g. ) .
loosely bound associations can be distinguished from ` true ' stellar clusters using limited observables by considering the ratio @xmath62 of age @xmath63 and the crossing time @xmath64 @xcite : @xmath65 where the @xmath64 is derived from @xmath66 where the effective radius @xmath67 and mass @xmath68 are easily estimated observable quantities .
we determined the time evolution of @xmath62 using eq .
[ eq_tcross ] for our simulations by measuring the @xmath67 and @xmath68 .
the resulting @xmath62 are plotted in figure [ fig_pi ] .
note that we plot the @xmath62 for the case where we identify the cluster as being composed of all stars initially present or alternatively where only the bound stars are taken into account .
the evolution of the @xmath62 of the clusters in figure [ fig_pi ] can be understood as follows : in the beginning @xmath62 increases linearly but as the cluster is disturbed the @xmath62 ratio is `` reset , '' as the @xmath67 increases and the mass decreases .
the tracks of our clusters end mostly with either a @xmath69 ( bound case ) or with a very low @xmath62 in which case they should be dissolved ( and no longer recognized as clusters ) .
this is consistent with a possible explanation for the distribution of @xmath62 as a function of mass noted by @xcite .
they found that for young stellar associations both low and high @xmath62 values are present . while older stellar associations were exclusively found with @xmath70 ( and identified as proper clusters ) .
we performed simulations of relatively small clusters of 1000 stars with gas .
our simulations self - consistently take the dynamical evolution of the stars , their internal nuclear evolution and the dynamics of the embedded gas and the gas liberated by stellar evolution into account .
each of the physical domains , stellar dynamics , stellar evolution and hydrodynamics , is resolved with well tested numerical solvers which have been developed independently .
the coupling between the domain - specific solvers is realized with the amuse framework , which enables us to couple codes written in a wide variety of languages in a homogeneous , transparent and self - consistent fashion .
the final code has the form of a relatively simple python script , which is general in its purpose of simulating embedded star clusters .
our model clusters start with a mix of stars and gas in virial equilibrium at a hypothetical moment we call @xmath71 , at which all stars are on the zero - age main - sequence .
the distribution of gas and stars was taken identical with a star - formation efficiency to control the relative mass fraction of the gas with respect to the stars .
the feedback strength is parametrized using a parameter @xmath29 .
we constrain the initial mass function to follow a salpeter imf with a minimum mass of @xmath72 and an upper limit of @xmath73 , but we fixed the maximum mass of a cluster star to @xmath74 in order to prevent random fluctuations from the coincidental generation of a more massive star , because it turns out that the most massive star in our cluster dominates the out - gassing by its efficient segregation to the cluster center , the strong stellar wind and the moment the star experiences a supernova . in our simulations
a star formation efficiency sfe@xmath75 results in a bound cluster , which is consistent with earlier simulations ( e.g. * ? ? ?
however , a more detailed analysis of our simulation results reveals that the process is considerably more complicated .
it turns out that the qualitative result with respect to final bound fraction can be approximated reasonably well with the parametrization of the sfe , but that the entire process is much richer .
@xcite found that the most important parameters affecting the final cluster were the sfe and the gas loss timescale .
our simulations convincingly demonstrate that the simple parametrization in terms of the sfe and/or gas loss timescale is a gross oversimplification of the process of stellar - supported out - gassing of the young cluster .
for example for the low feedback a4 run the initial supernova is not enough to disperse all cluster gas .
after the first supernova @xmath76 of the gas remains bound . from the standpoint of the dynamics of the stars
this remnant gas increases the effective sfe .
hence the cluster initially relaxes ( between 10 - 20 myr ) with a @xmath77 bound fraction . comparing with the baumgardt & kroupa results ( figure 1 )
, we see that this is a reasonable value if we consider their sfe@xmath78 curves : at these sfe a bound fraction of @xmath44 is quite possible ( even not accounting for any radial dependence of the gas loss ) .
further supernova events blow away the remaining gas , but do so only after the cluster has adapted itself dynamically . on the other hand ,
the b1 run shows if the gas loss occurs more gradually , the cluster can also adapt to gas loss ( albeit in our simulations not enough to give rise to a bound cluster ) .
the time scale on which the gas is expelled can not be described as a simple function , but depends on the distribution of the masses and positions of the stars , when they start blowing a strong stellar wind and the details regarding the moment of the supernovae ( and the adopted feedback strength ) . considering the wealth of variation in cluster parameters that result from this complex process
the question remains why young star clusters show a relatively narrow range of properties , whereas we would naively expect a much wider range of cluster radii , and masses . at this moment
we are unaware what key parameter may play a moderating role in the range of observables for young star clusters .
note that we did not take into account external limitations , such as the galactic tidal fields or the environment of the parent cloud . in our simulations we do not model the stellar feedback fully self - consistently .
the main uncertainties in our prescription concern the uncertainty in the efficiency of the coupling of the mechanical luminosity , the constraints of the numerical method and the lack of radiative feedback in our models .
in future simulations we expect to deal with this latter point by resolving the radiative feedback using a radiative transfer code , but for now this remains part of the uncertainty encoded in the free parameter for the feedback strength we introduced in our models .
this efficiency with which the output of stars couples to the surrounding gas may also depend on the mass of the star itself @xcite , and on the state of the gas : the density , temperature but also on its clumpiness ( an aspect we have not modeled ) .
we do demonstrate , however , that the core radii and mass of the surviving cluster does not depend strongly on this efficiency parameter . on the numerical side
, it is well known that sph shows relatively poor performance in the regime of explosive shock waves : for example the peak densities are generally smoothed out .
this could be important for the energy budget of the feedback , as the cooling happens mainly in these dense regions .
on the other hand , the speed and post - shock states are well reproduced @xcite in sph , so with the limitations we imposed on the details of the thermal evolution ( the fact that we did not try to account self - consistently for the energy budget of the feedback ) this has probably not a major impact .
in future work , amr employing godunov - type solvers for the gas dynamics may be employed to improve this aspect .
this would necessitate very high resolution simulations to reach a converged solution and the inclusion of much more physics to follow the ( non - equilibrium ) heating and cooling processes in the gas .
* we find that a sfe@xmath79 is necessary to result in bound clusters . *
the actual dispersion of clusters at sfe@xmath43 may occur gradual or sudden depending on the details of feedback . before eventual dissolution , stable phases in the cluster evolution
may occur if not all gas is expelled . contributing to this
is the increase in the effective sfe from the point of view of the stellar dynamics if feedback does not blow away all gas .
* we find that statistical variations can have a considerable effect on the internal structure of the remnant cluster .
* we find different modes of gas loss depending on the feedback coupling strength adopted .
a high coupling strength results in a gradual loss of gas on timescales of @xmath80 myr . for the low feedback coupling stellar winds are not able to disperse the gas : the timescale of gas loss in this case is very short ( due to supernova explosions ) .
* more detailed modeling is necessary to put constraints on the energy budget and thermal evolution of the gas , especially for the early phases ( where we do not take into account radiative feedback ) .
* during the embedded phase mass segregation can develop which persists after gas has been removed and the cluster has become less dense .
this suggests that the observed mass segregation of open clusters such as the pleiades is a remnant rather than the result of later secular evolution .
* mass segregation followed by partial dissolution of the cluster tends to lead to an excess of heavy stars in the cluster .
this is due to the fact that low velocity orbits are preferentially retained and the energy equipartition that forms the basis of mass segregation means that heavier stars preferentially populate the low velocity orbits . *
the dependence of the observed distribution of the ratio of the age of an association and its crossing time ( @xmath62 ) is consistent with the `` resetting of the @xmath62 clock '' we find in our models .
the short crossing times of young dense clusters increase due to stirring effect of mass loss .
unbound clusters are seen to have low @xmath62 values .
these two effects together mean that clusters with both @xmath81 and @xmath82 are expected to be present as long as the gas dispersal and mass loss of young stars happens ( before 30 myr ) .
this is consistent with what is observed ( although more precise modeling should determine whether the observed distributions can be matched ) .
the fact that we find mass segregation and the preferential retention of high mass stars has consequences for the interpretation of cluster observations .
in fact , observed clusters like the pleiades show mass segregation .
this mass segregation is usually attributed to the secular evolution in the pure n - body regime after the short lived embedded phase . for the pleiades primordial mass segregation
is necessary @xcite .
our models are able to reproduce the expected amount of initial mass segregation for the pleiades cluster ( although our models are not specifically build to reproduce the pleaides ) .
the fact that high mass stars are better able to withstand the disruptive effect of the loss of the natal cloud may lead to an excess number of unbound low mass stars in the vicinity of young clusters , which may be testable by expanding the search for cluster members to include kinematically unbound stars of the right age .
we have demonstrated that it is important to consider both the embedded phase evolution and the gas clearing phase of cluster evolution together .
future studies will aim to include radiative feedback effects and should also take into account more realistic initial conditions for the cluster and gas distribution . | we study the evolution of embedded clusters .
the equations of motion of the stars in the cluster are solved by direct n - body integration while taking the effects of stellar evolution and the hydrodynamics of the natal gas content into account .
the gravity of the stars and the surrounding gas are coupled self consistently to allow the realistic dynamical evolution of the cluster .
while the equations of motion are solved , a stellar evolution code keeps track of the changes in stellar mass , luminosity and radius . the gas liberated by the stellar winds and supernovae deposits mass and energy into the gas reservoir in which the cluster is embedded .
we examine cluster models with 1000 stars , but we varied the star formation efficiency ( between 0.05 - 0.5 ) , cluster radius ( 0.1 - 1.0 parsec ) , the degree of virial support of the initial population of stars ( 0 - 100% ) and the strength of the feedback .
we find that an initial star fraction @xmath0 is necessary for cluster survival .
survival is more likely if gas is not blown out violently by a supernova and if the cluster has time to approach virial equilibrium during out - gassing . while the cluster is embedded , dynamical friction drives early and efficient mass segregation in the cluster .
stars of @xmath1 are preferentially retained , at the cost of the loss of less massive stars .
we conclude that the degree of mass segregation in open clusters such as the pleiades is not the result of secular evolution but a remnant of its embedded stage .
star clusters star formation numerical simulations : hydrodynamic , n - body |
since the discovery of the current accelerated expansion of the universe @xcite , many efforts have been made to explain the physical essence behind this phenomenon of acceleration @xcite . within the framework of the general relativity , the acceleration can be attributed to the existence of a mysterious negative pressure component named phenomenally as dark energy . a positive cosmological constant , which fits to the observations
well , is currently still the simplest candidate for dark energy .
however , it is plagued with the fine - tuning problem and the coincidence problem @xcite . according to the holographic principle @xcite that the number of degrees of freedom of a physical system scales with the area of its boundary rather than its volume .
this might indicate that degrees of freedom are overcounted in the calculation of the vacuum energy in effective field theory .
the author in @xcite suggested that the ultraviolet ( uv ) cutoff @xmath22 in the effective field theory should be related to the infrared ( ir ) cutoff @xmath23 due to the limit set by the formation of a black hole , i.e. , in terms of the natural units , @xmath24 where @xmath25 with @xmath26 the newton s constant .
such relation leads to a dramatic depletion of quantum states , which results in a much small vacuum energy density , @xmath27 when the ir cutoff @xmath23 is comparable to the current hubble radius , the fine - tuning problem of the cosmological constant would not exist @xcite .
some interesting studies on cosmology with holography have been carried out @xcite . in ref.@xcite , we have shown that when the ir cutoff @xmath23 is characterized by the total comoving horizon of the universe , the inflation @xcite in early universe makes the vacuum energy given in eq.([vac ] ) behave as a cosmological constant without the fine - tuning problem as well as the coincidence problem , which provided an interesting insight on the cosmological constant as the simplest and best candidate for the dark energy . in alternative ,
we have found in @xcite that when the ir cutoff @xmath23 is taken to be a conformal - age - like length , the dark energy density given by the vacuum energy in eq.([vac ] ) is almost insensitive to the inflation in early universe . in the spirit of the holographic principle @xcite ,
some interesting holographic dark energy models in which the dark energy density is assumed to scale as @xmath28 with @xmath23 being some characteristic length scale of the universe have been proposed and studied by many groups @xcite .
particularly , the first holographic dark energy model ( hde ) was proposed in @xcite by identifying @xmath23 with the radius of the future event horizon , i.e. @xmath29 , here @xmath30 is the scale factor of the universe at cosmic time @xmath31 .
alternatively , the total comoving horizon @xmath32 including the very large primordial part generated by inflation was taken in @xcite as the ir cutoff to establish the so - called @xmath21hde model , as a consequence , the holographic dark energy behaves almost like a cosmological constant . on the other hand , a conformal - age - like length @xmath33 motivated from the four dimensional spacetime volume at cosmic time @xmath31 of the flat frw universe was adopted in @xcite to build the holographic type dark energy model(chde ) .
based on different physical origin , the age @xmath34 of the universe is chosen as the ir cutoff to construct the agegraphic dark energy model(ade ) in @xcite , its improved new version(nade ) was proposed with replacing the age of the universe by the conformal age @xmath35 of the universe @xcite . from the above considerations ,
it is interesting to investigate a general class of holographic type dark energy models with the characteristic length scale given by the following form @xmath36 here we will focus on the case with @xmath15 , @xmath37 being the integers . obviously , the ade model @xcite corresponds to the case @xmath38 ; the @xmath21hde model @xcite corresponds to @xmath39 , while the nade model @xcite corresponds to the same case but without considering a primordial part generated by inflation@xcite ; the chde model @xcite corresponds to @xmath40 . in this note
, we are going to investigate in detail a general class of holographic type dark energy models characterized by the ir cutoff given in eq.([l0 ] ) .
our main purpose is to see the possible holographic type dark energy candidates for various cases of @xmath41 by requiring the corresponding models to be self - consistent and also consistent with the expansion history of the universe . in particular
, we will show which cases are more favorable .
in sec.ii , we first present a description on the general class of holographic type dark energy models , and then we make a detail investigation on various models and find out the possible candidates for the various choices @xmath41 ; in sec.iii , we will mainly focus on the single - parameter models similar to the @xmath7cdm model and perform the best - fit analysis by using recent cosmological observations ; some concluding remarks and discussions are given in sec.iv .
let us begin with the general characteristic length scale of the universe @xmath42 where @xmath43 is the hubble parameter and @xmath44 denotes the derivative with respect to the cosmic time @xmath31 .
the corresponding holographic type dark energy density is parameterized by @xmath45 where @xmath46 is a positive constant parameter .
the fractional energy density is defined by @xmath47 for simplicity , let us consider a flat friedmann - robertson - walker ( frw ) universe containing the holographic type dark energy and ambient constituent with constant equation of state ( eos ) @xmath2 .
the friedmann equation is given by @xmath48 or in fractional energy densities @xmath49 where @xmath50 .
if there is no direct energy interchange , each energy component is conservative respectively , which results in conservation equations given by @xmath51 with @xmath52 . by using eqs.([l ] ) ,
( [ rho ] ) , ( [ frho ] ) and ( [ ceq ] ) , the eos of the holographic type dark energy is found to be @xmath53 cosmic acceleration requires that @xmath54 recently , which indicates @xmath55 or equivalently @xmath56 as @xmath15 is taken to be integer under our present consideration .
the conservation of the ambient constituent with constant @xmath2 leads to @xmath57 , where @xmath58 is a constant coefficient . combining with the friedmann equation and the definition of fractional energy densities
, we have @xmath59 by using eqs.([l ] ) and ( [ frho ] ) , we get @xmath60 substituting eq.([ha ] ) into above equation and taking derivative with respect to @xmath61 in both sides , we arrive at the differential equation of motion for @xmath62 @xmath63 obviously , under the transformation @xmath64 , with taking @xmath65 as the present scale factor of the universe , the energy density eq.([rho ] ) , the fractional energy density eq.([frho ] ) , the eos eq.([wde ] ) and the differential equation eq.([ode ] ) are all invariant .
namely , performing such transformation , all expressions keep the same , so we can set @xmath66 . from now on
, we adopt that the parameter @xmath46 has absorbed a factor @xmath67 and set @xmath66 .
note that the subscript``0 '' always indicates the present value of the corresponding quantity .
considering that our universe has successively experienced the inflation during which @xmath68 ( which indicates quasi - de sitter expansion ) , the radiation - dominated epoch during which @xmath69 , and the matter - dominated epoch during which @xmath70 before turning to accelerated expansion recently , we are actually able to approximately study the behaviors of @xmath23 , thus the fractional density @xmath62 in the early universe directly . here , we will simply ignore the effect due to the transition from one ambient - constituent - dominated epoch to another ambient - constituent - dominated epoch .
let us define @xmath71 where the subscript @xmath72 denotes the beginning of some epoch under consideration . obviously , we have @xmath73
when the constituent with constant @xmath2 dominates the universe from @xmath74 , we have approximately @xmath75 from fridemann equations , which results in the following consequences @xmath76 noticing that the approximation in the limit @xmath77 is due to the fact that @xmath78 and @xmath79 when @xmath80 , then @xmath81 all the time in the early universe . from eqs.([lw ] ) and ( [ lw2 ] ) , we have @xmath82 for the constituent with constant @xmath2 dominating the universe . during the inflation ,
the hubble parameter @xmath83 is constant approximately and the universe expands exponentially .
those terms with the factor @xmath84 will soon become negligible , we have @xmath85 approximately , which also results in @xmath86 at the end of the inflation . approximately , for the radiation dominated epoch , the initial value @xmath87 is at the order of @xmath88 . referring to eq.([lwneq ] ) for @xmath81 and reminding that the expansion of the universe , we have @xmath89 for radiation - dominated epoch during which @xmath90 as well .
similar result holds during the matter - dominated era . by using @xmath73 and eq.([frho ] ) , we get the approximate fraction of dark energy in early universe , i.e. @xmath91 where @xmath68 during inflation , @xmath90 in radiation - dominated epoch and @xmath92 in matter - dominated epoch respectively . for self - consistency , we need @xmath93 to ensure that @xmath94 when @xmath95 , thus the ambient matter dominated the early universe .
it is not difficult to prove that for @xmath93 , eq.([frhom ] ) is the approximate solution of the differential equation of @xmath62 under the limit @xmath96 when @xmath95 consistently . for the case @xmath97 , from eqs.([lw ] ) and ( [ lw2 ] ) , we simply have @xmath98 during inflation . while the approximate solution in matter- or radiation - dominated epoch is also given by eq.([frhom ] ) .
thus @xmath93 is also required to be self - consistent .
let us now pay attention to the case @xmath99 .
in this case , we have @xmath100 at the matter - dominated epoch , which is the exact solution of the differential equation . referring to the eos of dark energy eq.([wde ] ) , we get @xmath101 which means that the holographic type dark energy tracks the dominated component and never dominates .
this is of course unacceptable . moreover ,
unless the parameter @xmath46 is small enough , the faction of dark energy would be too large in early universe to be consistent with primordial nucleosynthesis ( bbn ) @xcite .
while by referring to the eos eq.([wde ] ) , a tiny @xmath46 would make the present eos of dark energy @xmath102 severely deviating from @xmath103 and be inconsistent with recent observations @xcite . for the case @xmath104 ,
then @xmath105 holds all the time in the early universe . from eqs.([lw ] ) , ( [ lw2 ] ) and eq.([lwneq ] ) , we have in early universe @xmath106 combining with @xmath73 and @xmath75 , we get the fraction of dark energy in early universe when @xmath107 @xmath108 as it is analyzed above that @xmath109 is required by recent cosmic acceleration , we then have @xmath94 when @xmath95 consistently for @xmath110 here .
the scaling property of @xmath62 with respect to @xmath61 can also be resulted from the differential equation of @xmath62 under the limit @xmath111 when @xmath112 .
this is because in light of the above equation the last factor in the differential equation eq.([ode ] ) is found to be @xmath113 which is negligible small . for the cases @xmath114 and @xmath115
, we can also obtain consistently eq.([frhom2 ] ) for @xmath56 during the inflation by following the same argument for the case @xmath104 . while for the radiation- and matter - dominated era , more attention is needed .
let us rewrite @xmath116 in the following way , @xmath117 where subscripts @xmath118 and @xmath119 denote the beginning and the end of inflation respectively . from eq.([lw2 ] ) , it is not difficult to find that the second term @xmath120 is approximately at order of @xmath121 for the cases @xmath114 and @xmath115 , while the first term is given by @xmath122 in order to solve the horizon problem @xcite , the inflation is required to last enough time to make @xmath123 moreover , due to the fact that @xmath124 has been growing as @xmath125 until recent cosmic acceleration ( from then on , @xmath126 ) , we generally have @xmath127 when @xmath128 .
thus for @xmath129 , we yield @xmath130 which indicates that for the cases @xmath114 and @xmath115 , the first term in eq.([tl12 ] ) is much larger than the second term .
thus we approximately have @xmath131 in the radiation- and matter - dominated epoch
. therefore , eq.([frhom2 ] ) also holds at these two epoches for @xmath109 and @xmath114 or @xmath115 .
it is noticed that for the case @xmath132 and @xmath4 the characteristic scale @xmath23 will be dominated by the primordial part generated in inflation with referring to eqs.([l ] ) , ( [ lw2 ] ) , ( [ neq2 ] ) , i.e. @xmath133 in the radiation- and matter - dominated epoch .
then the dark energy behaves almost like a cosmological constant during these two epoches with @xmath134 correspondingly , the fractional energy density of dark energy scales as @xmath135 the @xmath21hde model corresponding to the case @xmath39 has been investigated in @xcite . in the cases with @xmath4 ,
the observation constraints on the parameter @xmath46 is very weak and the models seemly have only one effective parameter , i.e. the ratio @xmath136 . in the @xmath21hde model , the parameter @xmath46 can take value in a normal order @xcite . while for the case @xmath137 , the parameter @xmath46 seems to take very large value due to the very large factor in @xmath23 , i.e.
@xmath138 , in order to make the ratio @xmath139 in the right order . in summary , @xmath1
is required by the recent cosmic accelerated expansion . for @xmath1 and @xmath4 , the fraction of dark energy density scales as @xmath140 with a tiny proportionality coefficient in the early universe
when @xmath95 .
since the proportionality coefficient can not be determined only by the parameter @xmath46 , there are in general two model parameters .
they may be chosen to be the parameter @xmath46 and the present fraction of dark energy @xmath141 . however ,
if @xmath132 , the holographic type dark energy would behave almost like a cosmological constant and there seems to be only one effective parameter @xmath141 .
since the total energy density @xmath142 approximately when the constituent with constant @xmath2 dominates the universe , we have @xmath143 .
thus @xmath3 in the early universe . for @xmath144
, we have @xmath8 in the early universe when @xmath95 . for @xmath145
, we also have @xmath146 in the matter- or radiation - dominated epoch , while @xmath147 during inflation .
it is seen that for all cases @xmath6 the fraction of dark energy can be ignored naturally in early universe when @xmath95 as long as the parameter @xmath46 takes a normal order value . moreover , due to such analytic feature , we can use the approximate solution at some @xmath148 in matter - dominated epoch as the initial condition to solve the equation of motion for @xmath62 . noticing that
once @xmath46 is given , the present fractional energy density @xmath149 can be obtained .
so the model is a single - parameter model like the @xmath7cdm model . substituting this approximate solution to the eos of dark energy eq.([wde ] ) , we get @xmath150 in the ambient - constituent - dominated epoch . referring to eq.([wde ] ) , @xmath151 will transit to @xmath152 due to the expansion of the universe .
obviously , for models with @xmath17 , the eos of the dark energy in the inflation , radiation- and matter - dominated epoch are @xmath153 , @xmath154 and @xmath155 respectively , and transit to @xmath152 eventually .
therefore , the eos of dark energy in such kinds of models transit from @xmath18 to @xmath19 and back to @xmath18 during the universe expansion .
the choices @xmath99 are abandoned because the dark energy can not dominate the universe forever and there might be too large fraction of dark energy in early universe . while the choices @xmath10 are forbidden by the self - consistent requirement that @xmath156 when @xmath112 .
we are going to mainly focus on the single - parameter models corresponding to the cases @xmath157 . in next section
, we will perform the best - fit analyses on such kinds of models by using recent observations . as the @xmath21hde model
@xcite corresponding to the case @xmath20 has only one effective parameter , we shall also take it into consideration for a comparison .
in this section , we are going to mainly focus on the single - parameter models corresponding to the cases @xmath157 and perform best - fit analyses on them by using recent cosmological observations including the union2.1 compilation of 580 supernova ia ( snia ) data @xcite , the parameter @xmath158 from bao measurements@xcite and the shift parameter @xmath159 from cmb measurements @xcite . the observational data and analysis method are given in appendix a.
as the cosmological observations mainly come from the epoch when the fraction of radiation energy is tiny , we then simply consider a flat friedmann - robertson - walker ( frw ) universe containing only dark energy and matter with @xmath92 .
using @xmath160 with @xmath161 the redshift , we can rewrite eq.([ode ] ) as @xmath162 and the approximate solution for the cases @xmath157 in the matter - dominated epoch is given by @xmath163 which can be chosen as an approximate solution at some @xmath164 .
namely , we may take @xmath165 as the initial condition to solve the differential equation eq.([odez ] ) .
the solution depends weakly on the choice of @xmath164 in a wide range , since @xmath62 is tiny and scales as @xmath166 when @xmath167 . here ,
we simply take @xmath168 at which matter dominates the universe well .
as the @xmath21hde model @xcite corresponding to the case @xmath20 behaves as a single effective parameter model , we would like to take it into consideration for a comparison . reminding that there are actually two parameters in this model
, we may choose @xmath46 and @xmath169 as two parameters .
taking @xmath170 as the boundary condition , we are able to solve the differential equation eq.([odez ] ) . from the friedmann equation and conservative equations , we have @xmath171 equivalently , @xmath172 substituting the results of @xmath173 and @xmath174 by solving eq.([odez ] ) into eq .([ez ] ) , the function @xmath175 can be obtained . of course
, it is impossible to perform the best - fit analyses on all single - parameter models corresponding to the cases @xmath157 .
however , we can still learn something from the best - fit analyses on a sample of models .
here , we are going to focus on the models with subjecting to @xmath176 as well as @xmath177 and to figure out some general results . the @xmath21hde model corresponding to the case @xmath20 is also included . in table
[ [ fit1 ] ] , we present the best - fit @xmath178 results by using only the union2.1 compilation of 580 supernova ia ( snia ) data @xcite . for comparison
, we also fit the @xmath7cdm model to the same observational data , and find that the minimal @xmath179 for the best fit parameter @xmath180 .
obviously , the @xmath7cdm model fits to the snia data best .
it is interesting to note that the best - fit result of the @xmath21hde model is the same as the @xmath7cdm model .
actually , the @xmath21hde model reduces to the @xmath7cdm model when model parameter @xmath181 @xcite . from table
[ [ fit1 ] ] , when focusing on the single - parameter models with @xmath157 , we see that models with @xmath182 have much smaller best - fit @xmath178 functions than models with @xmath183 for the given @xmath15 . also for the given @xmath16 , the best - fit @xmath178 function increases with @xmath15 .
[ fit1 ] .
the minimum of @xmath178 by using only the union2.1 snia data ; for comparison , @xmath179 . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] the @xmath184 functions and corresponding likelihoods of models by using snia+bao+cmb data sets.,title="fig:",width=158,height=283 ] the @xmath184 functions and corresponding likelihoods of models by using snia+bao+cmb data sets.,title="fig:",width=158,height=283 ] the @xmath184 functions and corresponding likelihoods of models by using snia+bao+cmb data sets.,title="fig:",width=158,height=283 ] the @xmath184 functions and corresponding likelihoods of models by using snia+bao+cmb data sets.,title="fig:",width=158,height=283 ] probability contours at 68.3% and 95.4% confidence levels of @xmath21hde model with @xmath20 ; the best - fit values are @xmath185 and @xmath186 by using snia+bao+cmb data sets.,width=491,height=188 ] it is noticed that the single - parameter models characterized by the conformal - age - like length scale with @xmath182 seem to be more favored , we shall pay more attention to those models .
the best - fit results for those models with some corresponding quantities are summarized in table [ [ fit2 ] ] .
we can also see that the present eos of dark energy @xmath102 deviating from @xmath103 become much more for larger @xmath15 . among the single - parameter models @xmath187 the present fraction of matter @xmath188 is increased with larger @xmath15 .
it is interesting to note that the parameter @xmath46 is compatible with @xmath189 in table [ [ fit2 ] ] .
this can be enlightened by the eos given in eq.([wde ] ) .
the present eos of dark energy is @xmath190 .
as @xmath191 , it leads to @xmath46 being compatible with @xmath189 in order to get @xmath102 deviating less from @xmath103 which seems to be more favored by the observations .
the complimentary data sets from different cosmological observations often constrain dark energy models better .
therefore , we would like to perform joint analyses on the same sample of models by using the union2.1 compilation of 580 supernova ia ( snia ) data @xcite , the parameter @xmath158 from bao measurements @xcite and the shift parameter @xmath159 from cmb measurements @xcite . in table
[ [ fit3 ] ] , we show the best - fit @xmath178 results for those models . for the given @xmath15 , we again conclude that models with @xmath182 are more favored . for the given @xmath16 , the best - fit @xmath178 function increases generally with @xmath15 except that the best - fit @xmath178 of the model with @xmath192 is larger than the model with @xmath193 .
this is mainly because the model with @xmath192 favors much smaller fraction of matter and goes more against the bao and cmb observations which are more sensitive to the present fraction of matter than the snia data .
the best - fit analysis indicates that the models characterized by the conformal - age - like length @xmath194 with small @xmath15 are more favored from the cosmological observations . in table
[ fit4 ] , we present the best - fit results of models with @xmath182 as well as @xmath195 at the @xmath196 confidence level by using snia + bao+cmb data sets . for the single - parameter model , @xmath197
is used to constrain the model parameter at the @xmath196 confidence level , while @xmath198 is used for the two - parameter model . the likelihoods or probability contours of these models
are plotted in figure [ fig1 ] and [ fig2 ] . from table
[ fit4 ] , we find that the present eos of the dark energy @xmath102 is around @xmath103 for model with @xmath192 , while for models with @xmath199 the present eos of dark energy are significantly below @xmath103 at two @xmath200 level .
it is also noticed that the eos of dark energy in the @xmath21hde model has @xmath201 and slightly deviates from @xmath103 . in table
[ fit4 ] , just as in table [ fit2 ] , one notices again that the model parameter @xmath46 is around @xmath202 in those four single - parameter models .
it seems to hold for models with @xmath203 . by referring to eq.([frhom ] ) , the holographic type dark energy in such single - parameter models characterized by the conformal - age - like length @xmath194 with small positive @xmath15 can be ignored in early universe as @xmath204 when @xmath205 , and the conformal - age - like length is given by @xmath206 .
we have investigated a general class of holographic type dark energy models described by the characteristic length scale @xmath0 with integers @xmath15 , @xmath37 .
it has been shown that the recent cosmic accelerated expansion requires @xmath1 . for @xmath1 and @xmath4
, we have the fraction of dark energy density scales as @xmath207 with a tiny proportionality coefficient and @xmath3 in the early universe when @xmath95 .
there are in general two model parameters which can be chosen as the parameter @xmath46 and the present fraction of dark energy @xmath141 . for @xmath132 and @xmath4 ,
the holographic type dark energy has been found to behave like a cosmological constant and there is actually only one effective parameter @xmath141 . for @xmath208
, we have @xmath8 and @xmath5 in the early universe when @xmath95 except that @xmath209 during inflation for @xmath97 . for those cases ,
the fraction of dark energy can be ignored in early universe when @xmath95 as long as the parameter @xmath46 takes a normal value . moreover , due to the analytic feature in early universe , the models with @xmath210 have been found to be single - parameter models like the @xmath7cdm model .
particularly , the eos of the dark energy in the models with @xmath203 have been shown to transit from @xmath19 in the radiation- and matter - dominated epoch to @xmath18 eventually .
it has been demonstrated that the choices @xmath99 should be abandoned as the dark energy can not dominate the universe forever and there might be too large fraction of dark energy in early universe , while the choices @xmath10 must be forbidden from the self - consistent requirement that @xmath11 when @xmath112 .
it is worth to point out that although the energy density of the dark energy in those self - consistent models grows or falls more slowly than the energy density of radiation and matter , the dark energy can not dominate the early universe due to the much smaller initial density , e.g. for the models with @xmath203 , we have @xmath211 at the end of the inflation at @xmath212 .
but the dark energy will eventually dominate the universe and be responsible for the recent cosmic acceleration .
interestingly , for models with @xmath208 , the pre - inflation part of the characteristic length @xmath23 is redshifted by the inflation which results in that @xmath213 is determined by the model parameter @xmath46 and the inflation approximately .
this means that the coincident problem of dark energy might be solved by the inflation naturally in these models . the model with @xmath214 and the single - parameter models corresponding to cases @xmath12
have been studied by using the recent cosmological observations .
it has been shown that the five models with @xmath214 , @xmath215 , @xmath216 , @xmath217 and @xmath40 fit to observations well . in the case
@xmath214 , the characteristic length scale @xmath23 is dominated by the primordial part generated by the inflation , which resulted in small and almost constant dark energy density @xcite .
while for the single - parameter models characterized by the conformal - age - like length @xmath194 with @xmath218 , it can be seen that the characteristic length scale behaves as @xmath14 in early universe , and the best - fit analysis leads the model parameter @xmath46 to be @xmath202 .
thus , the holographic type dark energy in such single - parameter models characterized by the conformal - age - like length @xmath194 with small positive @xmath15 can be ignored in early universe as @xmath204 when @xmath205 .
the best - fit analysis has shown that the models characterized by the conformal - age - like length @xmath194 with small @xmath219 are more favored from the cosmological observations .
we would like to thank r.g .
cai , m. li , y. ling , j.x .
lu for useful discussions .
the author ( z.p.h ) would like to thank m. q. huang and m. zhong for their helpful support .
this work is supported in part by the national basic research program of china ( 973 program ) under grants no .
2010cb833000 ; the national nature science foundation of china ( nsfc ) under grants no .
10975170 , 10975184 , 10947016 .
in this appendix , we present the method used for the best - fit analysis on the observational data including type ia supernovae ( snia ) , baryon acoustic oscillations ( bao ) , cosmic microwave background ( cmb ) radiation . the model parameters yielding a minimal @xmath222 and a maximal @xmath223 will be favored by the observations . in the following , we present the calculation for the various @xmath224 of each observational data set .
we consider the supernova cosmology project ( scp ) union2.1 compilation @xcite , which compiles the distance modulus @xmath225 of 580 supernovae .
the theoretical distance modulus is defined as @xmath226 in which the parameter @xmath227 is a nuisance parameter but is independent of the data .
the hubble - free luminosity distance is given by @xmath228 with @xmath229 .
the @xmath178 for the snia data is given by @xmath230 ^ 2\over \sigma_i^2 } ~ , \label{ochisn}\ ] ] where @xmath225 and @xmath231 are the observed value and the corresponding 1@xmath232 error of distance modulus for each supernova , respectively .
to reduce the effect of nuisance parameter @xmath227 @xcite , one can expand @xmath233 with respect to @xmath234 as follows @xmath235 where @xmath236 ^ 2\over \sigma_i^2 } ~,\ ] ] @xmath237 @xmath238 evidently , @xmath239 has a minimum as @xmath240 which is independent of @xmath234 . since @xmath241
, we will adopt @xmath242 in our best - fit analysis . from the distribution of sdss luminous red galaxies ,
the quantity @xmath158 via the measurement of the bao peak is defined as @xcite @xmath243 at the redshift @xmath244 .
the sdss bao measurement @xcite gives @xmath245 , where the scalar spectral index is taken to be @xmath246 from the wmap7 measurement @xcite .
the @xmath178 of the bao data is given by : @xmath247 the shift parameter @xmath159 is defined by @xcite @xmath248 where the redshift of the recombination @xmath249 wmap7 @xcite .
the shift parameter @xmath159 , which relates the angular diameter distance to the last scattering surface , the comoving size of the sound horizon at @xmath250 and the angular scale of the first acoustic peak in cmb power spectrum of temperature , has been measured to be @xmath251 @xcite . the @xmath184 of the cmb data is given by : @xmath252 p. horava and d. minic , phys .
lett . *
85 * , 1610 ( 2000 ) [ hep - th/0001145 ] .
s. d. thomas , phys .
* 89 * , 081301 ( 2002 ) .
w. fischler and l. susskind , hep - th/9806039 .
r. bousso , jhep * 9907 * , 004 ( 1999 ) [ hep - th/9905177 ] . k. a. olive , g. steigman and t. p. walker , phys .
rept . * 333 * , 389 ( 2000 ) [ arxiv : astro - ph/9905320 ] .
n. suzuki , d. rubin , c. lidman , g. aldering , r. amanullah , k. barbary , l. f. barrientos and j. botyanszki _ et al .
_ , astrophys .
j. * 746 * , 85 ( 2012 ) [ arxiv:1105.3470 [ astro-ph.co ] ] .
d. j. eisenstein _ et al .
_ [ sdss collaboration ] , astrophys . j. * 633 * , 560 ( 2005 ) [ arxiv : astro - ph/0501171 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys .
j. suppl .
* 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
j. r. bond , g. efstathiou and m. tegmark , mon . not .
soc . * 291 * , l33 ( 1997 ) [ arxiv : astro - ph/9702100 ] . | we present a detail analysis on a general class of holographic type dark energy models characterized by the length scale @xmath0 .
we show that @xmath1 is required by the recent cosmic accelerated expansion of universe . in the early universe dominated by the constituent with constant equation of state @xmath2
, we have @xmath3 for @xmath1 and @xmath4 , and @xmath5 for @xmath6 .
the models with @xmath6 become single - parameter models like the @xmath7cdm model due to the analytic feature @xmath8 at radiation- and matter - dominated epoch .
whereas the cases @xmath9 should be abandoned as the dark energy can not dominate the universe forever and there might be too large fraction of dark energy in early universe , and the cases @xmath10 are forbidden by the self - consistent requirement @xmath11 in the early universe .
thus a detailed study on the single - parameter models corresponding to cases @xmath12 is carried out by using recent observations .
the best - fit analysis indicates that the conformal - age - like models with @xmath13 , i.e. @xmath14 in early universe , are more favored and also the models with smaller @xmath15 for the given @xmath16 are found to fit the observations better .
the equation of state of the dark energy in models with @xmath17 transits from @xmath18 during inflation to @xmath19 in radiation- and matter - dominated epoch , and then back to @xmath18 eventually .
the best - fit result of the case @xmath20 which is so - called @xmath21hde model proposed in @xcite is the most favorable model and compatible with the @xmath7cdm model . |
in our previous paper @xcite the points mentioned in the abstract were outlined only briefly , using order - of - magnitude arguments and with very little explanation , due to space limitations . here
we present a more detailed explanation and analysis of the prompt thermal x - ray emission and x - ray afterglow , as well as a calculation of the early optical - uv emission , and show that some claims made in recent publications @xcite , according to which the observations are inconsistent with the massive wind interpretation , are not valid .
grb060218 was unique mainly in two respects : it showed a strong thermal x - ray emission accompanying the prompt non - thermal emission , and a strong optical - uv emission at early , @xmath0 d , time .
we show here that these features , as well as the x - ray afterglow , can all be explained by a model in which a radiation mediated shock propagates through a compact progenitor star into a massive wind .
we have shown in a separate paper @xcite that the prompt non - thermal x - ray emission can also be explained by this model . as detailed below , the prompt thermal x - ray emission
can be explained as shock breakout at a radius of @xmath4 cm , which requires a thomson optical depth ( of the plasma ahead of the shock ) of @xmath5 .
breakout may occur at @xmath4 cm if this is the stellar progenitor radius .
however , since the progenitor is likely to be smaller , we suggested the possibility of it being surrounded by an optically thick wind .
another possibility is a pre - explosion ejection of a small mass , @xmath6 , shell .
here we shall adopt the wind interpretation , since it allows one to explain also the x - ray afterglow .
so far , no other quantitative physical models have been worked out for the thermal x - ray emission , nor for the x - ray afterglow . @xcite and @xcite have suggested that the afterglow x - ray emission is due to extended activity of the source , for which there is no model or explanation .
@xcite suggest , for explaining the x - ray afterglow , an ansatz consisting of the ad - hoc existence of electrons with some prescribed energy distribution , which is different at different times to account for the observations , without a model for the dynamics of the plasma or for the electron energy distribution .
we note that the radio afterglow of grb060218 discussed by @xcite is difficult to explain within the context of the current model . indeed , as pointed out by @xcite and by @xcite
, it is difficult to explain the radio afterglow and the x - ray afterglow of grb060218 as due to emission from a single shock wave .
these authors have thus chosen to construct models that account for the radio emission only , attributing the x - ray afterglow to a continued activity of the source of an unexplained nature , and not accounting for the prompt x - ray emission and for the early optical - uv emission .
we adopt a different approach , showing that the prompt ( thermal and non - thermal ) x - ray emission , the early optical - uv emission , and the late x - ray afterglow can all be explained within the context of the same model .
we argue that it is the radio afterglow , rather than all the other components , which remains unexplained , and which should be attributed to a different component . since the radio emission carries only a negligible fraction of the energy , and given the large anisotropy of the explosion , it is not difficult to imagine the existence of such an additional low energy component . in
[ sec : model ] we discuss the observations and our model . in
[ sec : compare ] we compare our analysis with those of other authors . our results and conclusions are discussed in [ sec : discussion ] .
possibly the most distinguishing feature of grb060218 is the strong thermal x - ray emission accompanying the prompt non - thermal emission .
the temperature of the thermal x - ray photons observed up to @xmath7 s is @xmath8 kev @xcite .
the integrated flux of the black - body x - ray component , @xmath9 ( see figure [ fig : flux ] ) , corresponds ( using @xmath10 mpc and after correcting for the flux that falls outside the xrt band ) to a thermal x - ray energy of @xmath11 erg .
this is only an approximate estimate , and the total thermal energy may be somewhat larger , due to the gap in xrt observations between @xmath7 s and @xmath12 s. we consider here a model where this radiation is due to a shock breakout " . .
_ lower panel _ : the uvot light curve
. filled circles of different colors ( on - line version ) represent different uvot filters ( from bottom to top between @xmath13 s ) : @xmath14 ( centered at 544 nm ) ; @xmath15 ( 439 nm ) , @xmath16 ( 345 nm ) , uvw1 ( 251 nm ) ; uvm1 ( 217 nm ) and uvw2 ( 188 nm ) .
specific fluxes have been multiplied by their fwhm widths ( 75 , 98 , 88 , 70 , 51 and 76 nm , respectively ) .
, width=245 ] supernova ( sn ) shock waves become radiation - mediated when propagating through the stellar envelope ( see * ? ? ?
* for review ) .
the energy density behind the shock is dominated by radiation , and the mechanism that converts kinetic energy to thermal energy at the shock transition is compton scattering .
the optical depth of the shock transition layer is @xmath17 , where @xmath18 is the shock velocity .
this thickness is determined by the requirement that the time it takes a fluid element to flow through the transition layer , @xmath19 , should be comparable to the diffusion time of photons across this layer , @xmath20 . as the shock propagates outward , the ( thomson ) optical depth of the plasma lying ahead of the shock decreases , and when this optical depth becomes comparable to @xmath21 , compton scattering
can no longer maintain the shock .
at this point , radiation escapes ahead of the shock , producing the shock breakout flash @xcite , and the shock becomes collisional ( or collisionless , see * ? ? ? * ) .
hereafter we use the term `` shock breakout '' to denote the event of the transition from radiation to collisional ( collisionless ) shock mediation , accompanied by the emission of radiation .
if the optical depth of the wind surrounding the progenitor star is small , shock breakout will take place as the shock approaches the stellar surface .
it is for this reason customary to identify shock breakout with the emergence of the shock from the stellar surface .
however , if the optical depth of the wind is large , @xmath22 , breakout would occur once the shock reaches a radius where the wind optical depth drops to @xmath21 . as argued in @xcite , in order to obtain a breakout flash with energy of @xmath23 erg and temperature @xmath8 kev , breakout must occur at a radius of @xmath24 cm , and the shock must be mildly relativistic ( which implies @xmath25 ) . since the progenitor star is presumably smaller than this ( e.g. if it is a wolf - rayet star ) , a value of @xmath26 at @xmath24 cm may be obtained either by assuming that the progenitor is surrounded by a dense wind , or that an outer shell of the star was ejected to this radius prior to the grb explosion .
the mass of the shell required to obtain @xmath25 is only @xmath27 , where @xmath28 is the thomson opacity for ionized he . for
the calculation below we adopt a density profile @xmath29 , as would be expected for the wind model .
the results obtained for the breakout radius , shock velocity and plasma density at this radius are not sensitive , however , to the details of the density profile shape .
we consider therefore a shock wave of velocity @xmath30 driven into the wind surrounding the progenitor , whose 4-velocity is @xmath31 where @xmath32 is the lorentz factor .
let us first derive the energy and temperature of the post - shock radiation .
the post - shock temperature @xmath33 is related to the post - shock pressure by @xmath34 , and the observed temperature is given by @xmath35 , where @xmath36 and @xmath37 is the post - shock plasma velocity ( @xmath38 ) . for a strong radiation dominated shock @xmath39 , where @xmath40 is the pre - shock density and @xmath41 is a factor of order unity .
as we show below , the shock is required to be mildly relativistic , @xmath42 a few , for which we may approximate @xmath43 ( for @xmath44 , @xmath45 ) , and @xmath46 ( @xmath47 for @xmath44 ) .
this defines the observed temperature , @xmath48 , to be @xmath49 ( for @xmath44 , the numerical coefficient is @xmath50 instead of 1 ) .
the energy carried by the radiation may be estimated by noting that the energy density of the radiation is given , in the observer frame , by @xmath51 , and that the thickness of the shocked plasma shell may be approximated as @xmath52 , where @xmath53 is the shock radius .
this yields @xmath54 ( for @xmath44 , the numerical coefficient is @xmath55 instead of 0.5 ) . for a wind density profile ,
the optical depth of the wind at radius @xmath53 is simply @xmath56 . expressing the density in terms of @xmath57 and using eqs .
( [ eq : t ] ) and ( [ eq : e ] ) we find that the shock velocity at breakout is @xmath58 and that breakout occurs at a radius @xmath59^{1/7}\left(\frac{\kappa}{\tau}\right)^{2/7 } \nonumber\\ & = & 7.8\times10^{12}\left(\frac{t}{0.17\rm kev}\right)^{-4/7}e_{\rm th,49}^{3/7}\,{\rm cm},\end{aligned}\ ] ] where @xmath60 erg and we have used @xmath28 and @xmath61 in the numerical evaluations . note that the results depends only weakly on the exact values of @xmath62 and @xmath63 .
( [ eq : u ] ) and ( [ eq : r ] ) imply that the breakout of the shock takes place at a radius @xmath64 cm , and that the shock is mildly relativistic at breakout , @xmath65 i.e. @xmath66 .
these results are in agreement with the order of magnitude estimates given in @xcite . since the shock is found to be mildly relativistic , using @xmath61 at shock breakout is justified . in order to produce the shock that is driven into the wind , the explosion is thus required to produce a mildly relativistic shell , @xmath67 , with an energy of a few @xmath2 erg .
this is remarkably similar to the case of grb980425/sn1998bw , for which the ejection of a shell of energy @xmath68 erg and velocity @xmath67 was inferred from x - ray @xcite and radio @xcite observations .
the wind density at the breakout radius , @xmath69 , is also in agreement with our results in @xcite .
it is straightforward to verify that the energy density is dominated by radiation , @xmath70 . for a wind velocity of @xmath71
it corresponds to a mass loss rate of few @xmath72 .
the relevant mass loss here is that occurring within a day or less of the explosion .
the data currently available on wind mass losses suggesting @xmath73 refer to much longer timescales , for stars well before any explosion @xcite .
physically , it is however quite plausible that the mass loss increases considerably as the evolution of the core rapidly approaches the final collapse , accompanied by a rapid increase in the luminosity and the envelope expansion rate .
note that if both the shell and the wind were spherically symmetric , the characteristic timescale would be @xmath74 s ( for the inferred post shock velocity , @xmath75 , the effects of relativistic beaming are not significant ) , while the observed timescale of the thermal x - ray emission is @xmath76 s @xcite . however , an anisotropic shell ejection is a natural expectation in a core collapse grb , since strong rotation is a requisite to make the jet ( e.g. @xcite ) . even
normal " core collapse sn simulations show strong rotation - related anisotropy in the expanding gas ( e.g. @xcite ) .
thus , the semi - relativistic outer shell ejected is likely to be anisotropic , either due to an anisotropic explosion , or due to being driven by a jet .
this , as well as an anisotropic wind profile caused by rotation @xcite should lead to significant departures from sphericity in the shock propagation .
anisotropy is in fact a prediction of this model , which is supported by the detection of linear polarization @xcite . in an anisotropic shock , however , the timescale is no longer the naive spherical @xmath77 value , but is rather given by the sideways pattern expansion timescale , which depends on the angular velocity profile of the anisotropic shell ( e.g. at larger angles the shock emerges later due a decreasing velocity profile or due to an increasing wind density away from the symmetry axis , etc . ) .
if the wind shock breakout interpretation is adopted , the subsequent interaction of the ejected relativistic shell with the wind is expected to produce an x - ray afterglow @xcite .
the electrons accelerated to high energy in the collisionless shock driven into the wind emit x - ray synchrotron radiation , and may also inverse - compton scatter optical - uv sn photons to the x - ray band .
as the shock wave driven into the wind expands , it heats an increasing amount of mass to a high temperature and the initial energy of the ejected shell is transferred to the shocked wind .
the shell begins to decelerate beyond a radius at which the shocked wind energy becomes comparable to the initial shell energy .
this occurs at a time @xcite @xmath78 here , @xmath79 erg is the kinetic energy of the shell and @xmath80 for the wind density and shell velocity inferred in [ sec : breakout ] , @xmath66 and @xmath81 , @xmath82 s. at @xmath83 the energy is carried by shocked wind plasma , and is continuously transferred to a larger mass of newly shocked parts of the wind .
thus , the rate at which energy is transferred to accelerated electrons is @xmath84 , where @xmath85 is the fraction of the post - shock thermal energy carried by electrons .
if the electrons cool on a time scale shorter than the expansion time , @xmath86 , this would lead to a bolometric luminosity @xmath87 .
the fraction of this luminosity emitted as x - rays depends on the electron energy distribution . for electrons accelerated to a power - law distribution in energy , @xmath88 , @xmath89 , where @xmath90 and @xmath91 are the maximum and minimum electron lorentz factor .
the more detailed analysis given in @xcite of the synchrotron emission of shock accelerated electrons gives ( * ? ? ?
( 6 ) and ( 7 ) ) @xmath92 for @xmath93 here , @xmath94 day , @xmath95 and @xmath96 are the fractions of the post - shock thermal energy carried by electrons and magnetic field and @xmath97 is the cooling frequency , the frequency of synchrotron photons emitted by electrons for which the synchrotron cooling time equals @xmath98 ( higher energy electrons cool faster and emit higher energy photons ) .
( [ eq : synch ] ) implies an x - ray luminosity in swift s xrt band , 0.310 kev , of @xmath99 let us consider next the inverse - compton scattering of sn optical - uv photons by the shock - heated electrons .
the inverse - compton luminosity emitted by electrons with lorentz factor @xmath100 is given by @xmath101 , where @xmath102 is the thomson optical depth of these electrons and @xmath103 is the sn luminosity .
the lowest energy ic photons are produced by the lowest energy , thermal , electrons , @xmath104 ev , where @xmath105 is the lorentz factor of the lowest energy electrons .
@xmath105 is determined from the post - shock thermal energy density , @xmath106\rho v_s^2\approx\rho v_s^2 $ ] where @xmath18 is the shock velocity and @xmath107 ( @xmath108 is the adiabatic index ) , through @xmath109 . here
, @xmath110 is a dimensionless constant , the value of which depends on the exact form of the electron energy distribution . for a power - law distribution , @xmath88 , @xmath111 .
the shock velocity may be inferred from the shock radius , which is given by ( * ? ? ?
2 ) @xmath112 using @xmath113 we find @xmath114 and @xmath115 thus , on a time scale of a days ic scattering by the lowest energy ( thermal ) electrons is expected to contribute to the x - ray flux .
it is important to note here that the electrons producing the x - ray synchrotron flux , eq .
( [ eq : lx ] ) , are of much higher energy , @xmath116 , and lie at the high energy part of the accelerated electron energy distribution . the thomson optical depth of the thermal electrons is approximately given by @xmath117 , where @xmath118 is the wind mass accumulated up to @xmath53 . assuming that the thermal electrons do not lose all their energy by ic scattering on a time scale shorter than the expansion time @xmath98 , the ic luminosity produced by the thermal electrons is @xmath119 the ic luminosity given by eq .
( [ eq : l_ic ] ) may exceed , depending on the values of @xmath110 and @xmath120 , the luminosity given by eq .
( [ eq : lx ] ) .
@xmath121 is not , of course , a valid result , since @xmath122 given in eq .
( [ eq : lx ] ) is the luminosity obtained assuming that the electrons lose to radiation all the energy they gained from the shock .
the ic luminosity is thus limited by @xmath122 , and @xmath121 simply implies that the thermal electrons lose all their energy to ic scattering on a short time scale .
let us compare the predicted x - ray afterglow with the observed one .
the observed x - ray afterglow , following the prompt emission which ends at @xmath123 s , is well approximated by ( see figure [ fig : flux ] ) @xmath124 , which corresponds to @xmath125 .
this is in excellent agreement with the predictions of eqs .
( [ eq : lx ] ) and ( [ eq : l_ic ] ) for @xmath126 few @xmath2 erg .
the fact that the energy of the shock driven into the wind inferred from the x - ray afterglow , @xmath127 , is comparable to that of the thermal x - rays , @xmath128 , supports our model , in which both are due to the same shock driven into the wind . at early time , @xmath129 day , the emission is expected to be dominated by the synchrotron component , and for the power - law energy distribution of electrons , @xmath88 , the x - ray spectrum is expected to follow @xmath130 . on a time scale of a few days
, the emission is expected to be dominated by ic scattering of thermal sn photons by thermal shock electrons . at this stage ,
the x - ray spectrum may become steeper , reflecting the energy distribution of the lowest energy electrons heated by the shock .
these results are consistent with the @xmath130 x - ray spectrum measured at @xmath131 day , and with the indication , based on an xmm - newton observation , that at a later time , @xmath132 day , the spectrum is steeper , @xmath133 @xcite .
the following point should be made here .
as the shock speed @xmath134 decreases with time , shells ejected with velocities lower than that of the fast , @xmath135 , shell may `` catch up '' with the shock ( @xmath136 for @xmath137 and @xmath138 ) .
this may lead to increase with time of the shock kinetic energy @xmath127 , and thus to a modification of the x - ray light curve .
the fact that the x - ray luminosity decreases roughly as @xmath139 up to @xmath140 d implies that the energy in shells ejected with @xmath141 is not much larger than that of the fast shell , i.e. not much larger than @xmath142 erg . as the sn
shock propagates through the stellar envelope , it heats it to @xmath143 kev .
as the envelope expands , the photosphere propagates into the envelope , and we see deeper shells with lower temperatures . we derive here a simple analytic model for the photospheric radius and temperature , based on which we can derive approximately the flux and temperature of the escaping radiation .
we assume that the density of the stellar envelope near the stellar surface is given by @xmath144 where @xmath145 ( @xmath146 is the stellar radius ) , and that the the plasma ahead of the photosphere is highly ionized he , so that @xmath147 .
the velocity of the shock as it propagates through the envelope is approximately given by @xcite @xmath148{}&0.8\left(\frac{e_{ej}}{m_{ej}}\right)^{1/2 } \left(\frac{m_{ej}}{\rho_{1/2 } r_*^3}\right)^{0.2}\delta^{-0.2n},\end{aligned}\ ] ] where @xmath149 is the mass of the ejected envelope , and @xmath150 is the energy deposited in the envelope . since the shock is radiation dominated , the energy density behind the shock is given by @xmath151 , @xmath152 after the shock breaks through the envelope , the envelope begins to expand and cool .
it is useful do label the shells with lagrangian coordinates , defining @xmath153 as the ( time independent ) mass that lies ahead of a shell originally located at @xmath154 .
@xcite show that the final velocity ( after acceleration due to the adiabatic expansion ) of each shocked shell , @xmath155 , is related to the shock velocity approximately by @xmath156 $ ] .
after significant expansion , @xmath157 , the radius of each shell is given by @xmath158 .
given @xmath159 it is straightforward to derive the time dependent shell density .
the shell s ( time dependent ) temperature is then determined by @xmath160 , which holds for adiabatic expansion . after some tedious algebra we find that , for @xmath157
, the photosphere is located at @xmath161 here , @xmath162 erg and @xmath163 is the ratio between @xmath164 and the average envelope density @xmath165 .
although @xmath166 depends on the structure of the progenitor star far from the surface , where eq .
( [ eq : rho0 ] ) no longer holds , the results are very insensitive to its value . using eqs .
( [ eq : vs ] ) we find that the radius of the photosphere is @xmath167 and using eq .
( [ eq : t0 ] ) we find that the temperature of the photosphere is @xmath168 here ,
@xmath169 cm .
@xcite find @xmath170 cm and @xmath171 ev at @xmath172 s. this is clearly consistent with the cooling envelope interpretation .
we note that at @xmath172 s the emission originates from a shell of mass @xmath173 , in excellent agreement with our rough estimate in @xcite . at the time @xmath174
s optical photons are in the rayleigh - jeans tail , and the flux is approximately given by @xmath175 , where @xmath176 is the distance to grb060218 .
using eqs .
( [ eq : r_photo ] ) and ( [ eq : t_photo ] ) we find @xmath177 where @xmath178 s. this is consistent with the optical flux observed at early time , see figure [ fig : flux ] ( note that our derivation holds only for @xmath179 , i.e. for @xmath180 s ) . a more detailed study of the optical - uv early light could provide interesting constraints on the size and density distribution near the surface of the progenitor star .
such a detailed analysis would require taking into account effects neglected here ( e.g. , photon diffusion , which may be important on @xmath181 day time scale , anisotropy , etc . ) and is beyond the scope of the current manuscript .
note that we have neglected the effects of photon diffusion in the above derivation .
we do not expect diffusion to play an important role , due to the following argument .
the size of a region around @xmath159 over which diffusion has a significant effect is @xmath182 .
thus , the value of @xmath183 up to which diffusion affects the radiation field significantly , @xmath184 , is determined by @xmath185 , which gives @xmath186 this implies that photon diffusion is not expected to significantly modify the predicted light curve ( applying , e.g. , the approximate self - similar diffusion wave solutions of * ? ? ?
* to our density and velocity profiles yields a luminosity that differs by @xmath187 from that derived neglecting diffusion ) . for the massive wind discussed in our model ,
the synchrotron self - absorption optical depth is very large at radio frequencies .
the characteristic frequency of synchrotron photons emitted by the lowest energy , @xmath188 , electrons is @xmath189 , where the magnetic field is given by @xmath190 . using eqs .
( [ eq : rs ] ) and ( [ eq : gamma_m ] ) we have @xmath191 on a timescale of a few days we expect therefore the radio frequency , @xmath192 ghz , to be in the range @xmath193 ( see eq .
( [ eq : nu_c ] ) ) , i.e. we expect the lorentz factor of electrons dominating the emission and absorption of radio waves to be in the range @xmath194 ( where @xmath195 is the lorentz factor of electrons with cooling time comparable to the expansion time ) . in this case , the synchrotron self - absorption optical depth is given by @xmath196n_e(\gamma_\nu)\delta$ ] , where @xmath197 is the number density of electrons with lorentz factor @xmath198 and @xmath199 is the thickness of the shocked wind shell . for a power - law distribution , @xmath200 , we have @xmath201 , and using @xmath202 for the electron column density we have @xmath203 this large optical depth leads to a strong suppression of the radio synchrotron flux , compared to the x - ray synchrotron flux given in eq .
( [ eq : synch ] ) , @xmath204 the flux ratio measured at few to 20 days is @xmath205 @xcite .
thus , as noted in the introduction , the observed radio flux is higher than predicted by this model , and should be explained as a different component ( see also [ sec : fan06 ] ) .
@xcite argue that the shock breakout interpretation is not valid , since the optical - uv emission at few @xmath206 s is higher than the extrapolation to low frequencies of the planck spectrum which fits the thermal x - ray emission at the same time . as we have pointed out in @xcite , and as explained in detail in
[ sec : model ] , the thermal x - ray emission and the early optical - uv emission originate from different regions .
the thermal x - rays originate from a ( compressed ) wind shell of mass @xmath207 , while the optical - uv emission originates from the outer shells of the ( expanding ) star . at @xmath208
s the optical - uv radiation is emitted from a shell at a `` depth '' of @xmath209 from the stellar edge .
the optical - uv and the thermal x - rays need not correspond to the same planck spectrum . as we have explained in @xcite , and in more detail in
[ sec : model ] , if the explosion had been isotropic , the thermal x - ray emission should have disappeared altogether on a time scale of few hundred seconds , at which time @xcite compare the x - ray and uv emission .
the fact that the thermal x - ray emission is observed over a few thousand seconds can be accounted for by assuming an anisotropic explosion .
this assumption , made explicitly both in @xcite and here , is ignored by @xcite , whose criticism is framed in the context of a spherical model , which we never assumed .
concerning their own model , in their preferred explanation for the properties of this burst @xcite , unlike in our model , do not include any dynamics in their scenario , and adopt different ad - hoc parameters ( e.g. for the electron distribution ) at different times to account for the observations .
@xcite argues that in order to explain the temperature and the total energy of the blackbody component observed in grb 060218 by the shock breakout , the progenitor wr star has to have an unrealistically large core radius ... larger than @xmath210 " .
the results of @xcite are in fact consistent with our analysis .
the core radius " , @xmath211 , adopted by this author refers not to the hydrostatic core radius of the star , but rather to the radius at which the optical depth equals 20 . in non - lte modeling of the winds of wr - stars ,
@xmath211 is located within the wind , typically near the sonic point .
in fact , in some cases the wind velocity is already supersonic at @xmath211 ( e.g. hamann & koesterke 1998 ) .
@xmath211 is much larger than the hydrostatic core radius @xmath212 obtained in evolutionary calculations ( e.g. schaerer & maeder 1992 ) . as shown by nugis & lamers ( 2002 )
, @xmath213 varies with the spectral type of wr stars , increasing from about 2 for early types to about 20 for late types ( see table 5 of nugis & lamers 2002 ) .
since li ( 2006 ) considers in his calculations models where the ratio between the wind photospheric radius @xmath214 and @xmath211 is @xmath215 ( using his notation , for @xmath216=5 the choice of @xmath217 to @xmath218 corresponds to @xmath215 , see his fig .
3 ) , his statement that @xmath219 is required implies that a wind with a photospheric radius of @xmath220 cm to @xmath221 cm is needed to account for the thermal x - ray emission of grb 060218 as shock breakout .
this is consistent with our eq .
( [ eq : r ] ) .
it is important to emphasize that the relevant radius is not @xmath211 , but rather the wind photospheric radius , @xmath214 , where @xmath61 .
it is the photospheric radius which needs to be large in order to allow sufficiently large shock breakout energy .
@xmath214 is larger than @xmath211 by a factor of 210 , depending on the model adopted for the wind velocity profile .
nugis & lamers 2002 present late type models with @xmath222 , and a large fraction of the stars analyzed in hamann & koesterke ( 1998 ) have @xmath223 ( see their table 2 ) .
@xmath214 larger than @xmath210 may therefore be obtained for stars with large mass loss rates . it should be emphasized that even if the wind photospheric radius required for grb 060218 had turned out to be larger by a factor of a few than the largest @xmath214 of known wr stars , this would not have been a strong argument against the wind interpretation of grb 060218 , since grb progenitors may well be more extreme than normal wr stars .
clearly , not all wr stars end their lives as grbs .
moreover , it should be remembered that practically nothing is known about the mass loss on a day time scale prior to the explosion , which determines the wind density at the relevant radii .
@xcite raised different criticisms in different versions of their paper . in the first version , astro - ph/0604016v1
, they argued that the presence of a massive wind would result in strong optical emission at early time , which is inconsistent with observations .
while this claim was retracted in subsequent versions of their manuscript , it may be worthwhile to clarify this issue here .
the synchrotron emission given in eq .
( [ eq : synch ] ) holds for frequencies above the cooling frequency , @xmath224 .
since the cooling frequency is low for a massive wind , see equation ( [ eq : nu_c ] ) , we expect in this model a similar x - ray and optical - uv luminosity , @xmath225 ( corresponding to a flux of @xmath226 in the @xmath227 band of swift s uvot ) .
this predicted flux exceeds the observed flux at @xmath228 s , hence the statement of @xcite .
however , as pointed out in [ sec : wind - shell ] , eq .
( [ eq : synch ] ) holds only for times @xmath229 s ( and the luminosity is lower at earlier times ) .
moreover , at @xmath228 s the large x - ray luminosity , @xmath230 , would suppress the synchrotron emission of the shock accelerated electrons . as explained in
[ sec : wind - shell ] , the cooling time of shock accelerated electrons due to synchrotron emission is short compared to the shock expansion time . at early times , @xmath228 s , the ratio of the magnetic energy density , @xmath231 , to the x - ray energy density , @xmath232 , is @xmath233 , which implies that the ic cooling time is much shorter than the synchrotron cooling time , suppressing the synchrotron emission in the optical band .
second , @xcite argue ( see also * ? ? ?
* ) that radio observations rule out the existence of a massive wind .
as mentioned in the introduction , indeed the radio and the x - ray afterglow are difficult to explain in the framework of a single shock .
however , the radio observations alone can not be used to rule out the existence of a massive wind , since radio observations do not allow one to determine the explosion parameters .
this is illustrated by the fact that estimates for the kinetic energy based on modeling the radio data alone range from @xmath234 erg to @xmath235 erg , and that the ambient medium density estimates range from @xmath236 to @xmath237 @xcite .
finally , @xcite argue that in the presence of a massive wind , the radio flux would be higher than observed . as explained in [ sec : radio ] , our problem is quite the opposite : our model flux would be too low to account for the observed radio emission . @xcite and
@xcite have chosen to construct models of grb060218 which concentrate on accounting for the radio emission , while attributing the x - ray afterglow to a continued activity of the source of an unexplained nature , and de - emphasizing the importance of prompt x - ray emission and the early optical - uv emission .
however , the radio emission represents a negligible fraction of the total energy . here , and in @xcite
, we have adopted a different approach , which is that the prompt ( thermal and non - thermal ) x - ray emission , the early optical - uv emission , and the late x - ray afterglow can all be explained within the context of the same model , which is based on the energetics of the early phases of the explosion .
we argue that it is the radio afterglow , rather than all the other components , which plays a lesser role , and which may be attributed to a different component .
given the very low energy of the radio emission and the the large anisotropy expected in the explosion , such an additional low energy radio component is imaginable , whose role is unlikely to be important in determining the characteristics of the early high energy emission .
we have discussed a comprehensive model of the early x - ray and optical / uv behavior of the grb060218/sn2006aj system , which provides the quantitative justification for the interpretation outlined in @xcite , as well as a number of additional points .
the most exciting features of this event were that it showed a strong thermal x - ray component as well as a strong optical / uv component in its early phases , at @xmath0 day , transitioning later to a more conventional x - ray and optical afterglow , and a radio afterglow .
we have shown that the early x - ray / o / uv behavior can be understood in terms of a mildly relativistic radiation - mediated shock which breaks out of a ( thomson ) optically thick wind produced by the progenitor star , leading to the observed thermal x - rays ( see [ sec : breakout ] ) .
the early optical / uv behavior arises as the shocked stellar envelope expands to larger radii ( [ sec : uv ] ) , and the x - ray afterglow arises from synchrotron and inverse - compton emission of electrons accelerated by the propagation of the shock further into the wind ( [ sec : wind - shell ] ) .
a detailed analysis of the optical - uv emission may therefore provide stringent constraints on the progenitor star .
the thermal x - ray emission requires a mildly relativistic shock , @xmath238 , carrying @xmath239 erg , driven into a massive wind characterized by a mass loss rate of a few @xmath240 for a wind velocity of @xmath241 .
the later x - ray afterglow is consistent with emission due to the propagation of this shock into the wind beyond the breakout radius , where the shock becomes collisionless .
this situation is very similar to that of grb980425/sn1998bw , for which the x - ray and radio afterglow are interpreted as emission from a shock driven into a ( much less massive ) wind by the ejection of a shell of energy @xmath68 erg and velocity @xmath67 ( * ? ? ?
* and references therein ) .
the early optical - uv emission is consistent with the expansion of the outer part , @xmath3 , of the stellar envelope , which was heated to a high temperature by the radiation dominated shock as it accelerates in its propagation towards the stellar edge .
it is important to note , however , that the acceleration of the shock near the stellar surface is not sufficient to account for the large energy , @xmath239 erg , deposited in the mildly relativistic , @xmath238 , component , which is required to have a mass of @xmath242 .
for the parameters inferred from the sn2006aj light curve , @xmath243 erg and @xmath244 @xcite , the energy predicted to by carried by @xmath245 is less than @xmath246 erg ( see fig .
6 of * ? ? ?
this suggests that the mildly relativistic component is driven not ( only ) by the spherical sn shock propagating through the envelope , but possibly by a more relativistic component of the the explosion , e.g. a relativistic jet propagating through the star .
an important factor in our interpretation of the early x - ray emission is the anisotropy of this shock , which leads to a characteristic timescale of the thermal x - ray emission controlled by the sideways pattern expansion speed , rather than by a simple radial line of sight velocity .
a significant anisotropy of the mildly relativistic shock component is expected for various reasons , e.g. as a result of being driven by a jet , or due to propagation through an anisotropic envelope and into an anisotropic wind caused by fast progenitor rotation , etc .
the simplest explanation for the prompt gamma - ray behavior may be that it is due to a relativistic jet , which can contribute to the anisotropy of the mildly relativistic shock propagating through the stellar envelope and the mildly relativistic ejected shell . in this case , the contribution of the relativistic jet to the late ( @xmath247 day ) x - ray and optical afterglow is not the dominant effect ( c.f . * ? ? ?
* ) . however , a possible explanation of the prompt non - thermal emission , both in grb980425 and grb060218 , may be that it is due to repeated ic scattering at breakout , as suggested in @xcite ( c.f .
in this case there may be no highly relativistic jet- it may either be mildly relativistic to begin with , or it may have been relativistic but it was choked , only its mildly relativistic bow shock emerging from the star . if our interpretation is correct , the importance of the early thermal x - ray component is that it represents the first detection ever of the breakout of a supernova shock from the effective photosphere of the progenitor star . since this is a grb - related supernova ,
a strong stellar wind can be expected , which results in the breakout photospheric radius being in the wind , rather than in the outer atmosphere of the star .
this provides a potentially valuable tool for investigating the physical conditions , mass loss and composition of the long grb progenitors in the last few days and hours of their evolution prior to their core collapse . | we show that the prompt and afterglow x - ray emission of grb060218 , as well as its early ( @xmath0 d ) optical - uv emission , can be explained by a model in which a radiation- mediated shock propagates through a compact progenitor star into a dense wind .
the prompt thermal x - ray emission is produced in this model as the mildly relativistic shock , @xmath1 carrying few @xmath2 erg , reaches the wind ( thomson ) photosphere , where the post - shock thermal radiation is released and the shock becomes collisionless . adopting this interpretation of the thermal x - ray emission , a subsequent x - ray afterglow is predicted , due to synchrotron emission and inverse - compton scattering of sn uv photons by electrons accelerated in the collisionless shock . early optical - uv emission is also predicted , due to the cooling of the outer @xmath3 envelope of the star , which was heated to high temperature during shock passage . the observed x - ray afterglow and the early optical - uv emission are both consistent with those expected in this model .
detailed analysis of the early optical - uv emission may provide detailed constraints on the density distribution near the stellar surface . |
the main purpose of this article is to introduce a new notion of generalized symplectic handles , which can be attached to contact @xmath0-manifolds along certain transverse links and pre - lagrangian tori .
we begin by stating in [ sec : results ] the essential definitions and explaining some existence results for non - exact symplectic cobordisms that follow from the handle construction . as easy applications ,
these results imply new and substantially simplified proofs of several recent results of the author and collaborators on obstructions to symplectic fillings , contact type embeddings and embeddings of partially planar domains .
the main results concerning the handle construction itself will be explained in [ sec : handles ] , together with a simple application to embedded contact homology and a conjectured generalization . in [ sec : consequences ] we discuss further applications and examples , providing a unified framework for reproving several important previous results of eliashberg , gay , etnyre and others involving the existence of symplectic cobordisms . the hard work is then undertaken in
[ sec : details ] , of which the first several sections construct the symplectic handles described in [ sec : handles ] , [ sec : proofs ] completes the proofs of the results stated in the introduction , and [ sec : holomorphic ] discusses the construction of a holomorphic foliation in the cobordisms , which we expect should have interesting applications in embedded contact homology and/or symplectic field theory .
the writing of this article benefited substantially from discussions with john etnyre , david gay , paolo ghiggini , michael hutchings , klaus niederkrger , andrs stipsicz , cliff taubes and jeremy van horn - morris .
several of these conversations took place at the march 2010 msri workshop _
symplectic and contact topology and dynamics : puzzles and horizons_. in topology , an oriented cobordism from one closed oriented manifold @xmath4 to another @xmath5 is a compact oriented manifold @xmath6 such that @xmath7 . if @xmath6 has dimension @xmath8 and also carries a symplectic structure @xmath9 , then it is natural to consider the case where @xmath10 is * symplectically convex * at @xmath5 and * concave * at @xmath4 : this means there exists a vector field @xmath11 near @xmath12 which points transversely outward at @xmath5 and inward at @xmath4 , and is a * liouville * vector field , i.e. @xmath13 . in this case
the @xmath14-form @xmath15 is a primitive of @xmath9 and its restriction to each boundary component @xmath16 is a ( positive ) * contact form * , meaning it satisfies @xmath17 .
the induced * contact structure * on @xmath16 is the ( co - oriented ) hyperplane field @xmath18 , and up to isotopy it depends only on the symplectic structure near the boundary .
we thus call @xmath10 a * ( strong ) symplectic cobordism * from @xmath19 to @xmath20 .
when such a cobordism exists , we say that @xmath19 is * symplectically cobordant * to @xmath20 and write @xmath21 if @xmath22 also extends to a global primitive of @xmath9 , or equivalently , @xmath11 extends to a global liouville field on @xmath6 , then we call @xmath10 an * exact symplectic cobordism * from @xmath19 to @xmath20 and write @xmath23 when @xmath24 , it is also interesting to consider a much weaker notion : without assuming that @xmath9 is exact near @xmath12 , we call @xmath10 a * weak symplectic cobordism * from @xmath19 to @xmath20 if @xmath25 are any two positive co - oriented ( and hence also oriented ) contact structures such that @xmath26 .
we then say that @xmath9 * dominates * the contact structures on both boundary components .
it is a standard fact that strong symplectic cobordisms can always be glued together along contactomorphic boundary components , thus the relations @xmath27 and @xmath28 define preorders on the contact category , i.e. they are reflexive and transitive .
they are neither symmetric nor antisymmetric , as is clear from some simple examples that we shall recall in a moment .
regarding the empty set as a trivial example of a contact manifold , we say that @xmath29 is * strongly * ( or * exactly * ) * fillable * if @xmath30 ( or @xmath31 respectively ) .
a @xmath0-dimensional contact manifold @xmath29 is likewise * weakly fillable * if there is a weak symplectic cobordism from @xmath32 to @xmath29 .
for example , the tight @xmath0-sphere @xmath33 is exactly fillable , as it is the convex boundary of the unit @xmath34-ball with its standard symplectic structure .
there are many known examples of contact @xmath0-manifolds that are not fillable by these various definitions : the original such result , that the so - called _ overtwisted _ contact manifolds are not weakly fillable , was proved by gromov @xcite and eliashberg @xcite .
in contrast , etnyre and honda @xcite showed that every contact @xmath0-manifold admits a * symplectic cap * , meaning it satisfies @xmath35 , although @xmath36 is never true ( due to stokes theorem ) .
there is an obvious obstruction to the relation @xmath37 whenever @xmath19 is strongly fillable but @xmath20 is not , e.g. @xmath19 can not be the tight @xmath0-sphere if @xmath20 is overtwisted . put another way , symplectic cobordisms imply filling obstructions , as @xmath29 can not be fillable if it is cobordant to anything overtwisted .
the following open question may be viewed as a test case for the existence of subtler obstructions to symplectic cobordisms .
[ question : cob ] is every contact @xmath0-manifold @xmath29 that is not strongly fillable also symplectically cobordant to some overtwisted contact manifold @xmath38 ?
the answer to the corresponding question for exact cobordisms is _ no _ : by an argument originally due to hofer @xcite , @xmath39 implies that every reeb vector field on @xmath29 admits a contractible periodic orbit , yet there are simple examples of contact manifolds without contractible orbits that are known to be non - fillable , e.g. all of the tight @xmath0-tori other than the standard one .
more generally , it has recently become clear that overtwistedness is only the first level in an infinite hierarchy of filling obstructions called _
planar @xmath40-torsion _ for integers @xmath41 , cf .
a contact manifold is overtwisted if and only if it has planar @xmath42-torsion , and there are many examples which are tight or have no giroux torsion but have planar @xmath40-torsion for some @xmath43 , and are thus not strongly fillable . the aforementioned argument of hofer
then generalizes to define an algebraic filling obstruction @xcite that lives in symplectic field theory and sometimes also gives obstructions to exact cobordisms from @xmath40-torsion to @xmath44-torsion .
our first main result says that no such obstructions exist for non - exact cobordisms , thus hinting at a possibly positive answer to question [ question : cob ] .
for a candidate counterexample . ] [ thm : overtwisted ] every closed contact @xmath0-manifold with planar torsion admits a ( non - exact ) symplectic cobordism to an overtwisted contact manifold .
this of course yields a new and comparatively low - tech proof of the fact , proved first in @xcite , that planar torsion obstructs strong fillings .
it also generalizes a result proved by david gay in @xcite , that any contact manifold with giroux torsion at least two is cobordant to something overtwisted ; as shown in @xcite , positive giroux torsion implies planar @xmath14-torsion ( cf .
[ sec : gay ] ) . by a result of etnyre and honda @xcite , every connected
overtwisted contact manifold admits a stein cobordism to any other connected contact @xmath0-manifold , and gay @xcite showed that the word `` connected '' can be removed from this statement at the cost of dropping the stein condition .
we thus have the following consequence : [ cor : equivalence ] every closed contact @xmath0-manifold with planar torsion is symplectically cobordant to every other closed contact @xmath0-manifold .
it should be emphasized that due to the obstructions mentioned above , corollary [ cor : equivalence ] is not true for _ exact _ cobordisms , not even if the positive boundary is required to be connected .
in fact , there is no known example of an exact cobordism from anything tight to anything overtwisted , and many examples that are tight but non - fillable ( e.g. the @xmath0-tori with positive giroux torsion ) certainly do not admit such cobordisms .
there is also a version of theorem [ thm : overtwisted ] that implies the more general obstruction to weak fillings proved in @xcite .
recall ( see definitions [ defn : planar ] and [ defn : planartorsion ] ) that for a given closed @xmath45-form @xmath46 on a contact @xmath0-manifold @xmath29 , we say that @xmath29 has _ @xmath46-separating _ planar torsion if it contains a planar torsion domain in which a certain set of embedded @xmath45-tori @xmath47 all satisfy @xmath48 if this is true for all closed @xmath45-forms @xmath46 , then @xmath29 is said to have _ fully separating _
planar torsion .
[ thm : weak ] suppose @xmath29 is a closed contact @xmath0-manifold with @xmath46-separating planar torsion for some closed @xmath45-form @xmath46 on @xmath49 with @xmath50
. then there exists a weak symplectic cobordism @xmath10 from @xmath29 to an overtwisted contact manifold , with @xmath51 . using a darboux - type normal form near the boundary
, weak symplectic cobordisms can be glued together along contactomorphic boundary components of opposite sign whenever the restrictions of the symplectic forms on the boundaries match ( see lemma [ lemma : gluing ] ) .
thus if @xmath29 has @xmath46-separating planar @xmath40-torsion and admits a weak filling @xmath10 with @xmath52 = [ \omega ] \in h^2_{{\operatorname{dr}}}(m)$ ] , then theorem [ thm : weak ] yields a weak filling of an overtwisted contact manifold , and hence a contradiction due to the well known theorem of gromov @xcite and eliashberg @xcite .
we thus obtain a much simplified proof of the following result , which was proved in @xcite by a direct holomorphic curve argument and also follows from a computation of the twisted ech contact invariant in @xcite .
if @xmath29 has @xmath46-separating planar torsion for some closed @xmath45-form @xmath46 on @xmath49 , then it does not admit any weak filling @xmath10 with @xmath52 = [ \omega ] \in h^2_{{\operatorname{dr}}}(m)$ ] .
in particular , if @xmath29 has fully separating planar torsion then it is not weakly fillable .
we now state some related results that also apply to fillable contact manifolds .
the aforementioned existence result of @xcite for symplectic caps was generalized independently by eliashberg @xcite and etnyre @xcite to weak cobordisms : they showed namely that for any @xmath29 with a closed @xmath45-form @xmath46 that dominates @xmath53 , there is a symplectic cap @xmath10 with @xmath54 and @xmath51 .
our next result concerns a large class of contact manifolds for which this cap may be assumed to have a certain very restrictive property .
[ thm : caps ] suppose @xmath29 is a contact @xmath0-manifold containing an @xmath46-separating partially planar domain @xmath55 ( see definition [ defn : planar ] ) for some closed @xmath45-form @xmath46 on @xmath49 with @xmath50 .
then @xmath29 admits a symplectic cap @xmath10 such that @xmath51 and there exists a symplectically embedded @xmath45-sphere @xmath56 with vanishing self - intersection number .
as the work of mcduff @xcite makes clear , symplectic manifolds that contain symplectic @xmath42-spheres are quite special , and for instance any closed symplectic manifold obtained by gluing the cap from theorem [ thm : caps ] to a filling of @xmath29 must be rational or ruled .
an easy adaptation of the main result in @xcite also provides the following consequence , which was proved using much harder punctured holomorphic curve arguments in @xcite : [ cor : separating ] suppose @xmath29 contains an @xmath46-separating partially planar domain for some closed @xmath45-form @xmath46 on @xmath49 . if @xmath10 is a closed symplectic @xmath34-manifold and @xmath49 admits an embedding @xmath57 such that @xmath58 and @xmath59 = [ \omega ] \in h^2_{{\operatorname{dr}}}(m)$ ] , then @xmath60 separates @xmath6 .
since planar torsion domains are also partially planar domains , this implies that planar torsion is actually an obstruction to contact type embeddings into closed symplectic manifolds , not just symplectic fillings .
note that examples of contact manifolds that admit non - separating embeddings arise from special types of symplectic fillings : we shall say that @xmath29 is ( strongly or weakly ) * hyper - fillable * if there is a connected ( strong or weak ) filling @xmath10 whose boundary is the disjoint union of @xmath29 with an arbitrary non - empty contact manifold .
put another way , @xmath29 admits a connected * semi - filling * with disconnected boundary .
given such a filling , one can always attach a symplectic @xmath14-handle to connect distinct boundary components and then cap off the boundary to realize @xmath29 as a non - separating contact hypersurface .
various examples of contact manifolds that are or are not hyper - fillable have been known for many years : * the tight @xmath0-sphere @xmath33 is not weakly hyper - fillable , by arguments due to gromov @xcite , eliashberg @xcite and mcduff @xcite .
etnyre @xcite extended this result to all planar contact manifolds . *
mcduff @xcite showed that for any riemann surface @xmath61 of genus at least @xmath45 , the unit cotangent bundle @xmath62 with its canonical contact structure is strongly hyper - fillable .
further examples were found by geiges @xcite . *
giroux @xcite showed that every tight contact structure on @xmath63 is weakly hyper - fillable . however , none of them are strongly hyper - fillable , due to a result of the author s in @xcite .
all of the negative results just mentioned can be viewed as special cases of corollary [ cor : separating ] , and so can the closely related result in @xcite , that partially planar contact manifolds never admit non - separating contact type embeddings .
observe that any contact manifold cobordant to one for which corollary [ cor : separating ] holds also can not be hyper - fillable : in particular this shows that not every contact @xmath0-manifold is cobordant to @xmath33 . we are thus led to an analogue of question [ question : cob ] that also applies to fillable contact manifolds : [ question : s3 ] does every closed and connected contact @xmath0-manifold @xmath29 that is not strongly hyper - fillable admit a connected symplectic cobordism to the tight @xmath0-sphere @xmath33 ?
note that it s important to assume the cobordism is connected in this question , as technically @xmath64 is always satisfied by taking the disjoint union of a cap for @xmath29 with a filling of @xmath33 .
the answer is again clearly no for exact cobordisms , as a variation on hofer s argument from @xcite also shows that @xmath29 must always admit contractible reeb orbits if @xmath65 .
the following result provides some evidence for a positive answer in the non - exact case , though it is not quite as strong as one might have hoped .
[ thm : tos3 ] suppose @xmath29 is a connected contact @xmath0-manifold containing a partially planar domain which either has more than one irreducible subdomain or has nonempty binding .
then @xmath29 admits a connected symplectic cobordism to @xmath33 , and hence also to every connected strongly fillable contact @xmath0-manifold .
the conditions of theorem [ thm : tos3 ] hold in particular for all planar contact manifolds , and in fact a stronger version can be stated since the fully separating condition is always satisfied .
we will show in [ sec : etnyre ] that this implies etnyre s planarity obstruction from @xcite .
[ thm : planar ] suppose @xmath29 is a connected planar contact @xmath0-manifold .
then there exists a compact connected @xmath34-manifold @xmath6 with @xmath66 , with the property that for every closed @xmath45-form @xmath46 on @xmath49 with @xmath50 , @xmath6 admits a symplectic structure @xmath9 such that @xmath51 and @xmath10 is a weak symplectic cobordism from @xmath29 to @xmath33 .
[ remark : counterexamples ? ] let us describe some contact manifolds that could conceivably furnish negative answers to questions [ question : cob ] and [ question : s3 ] .
first , consider the standard @xmath0-torus @xmath67 with the contact structure defined in below , and divide it by the @xmath68-action induced by the contact involution @xmath69 the quotient @xmath70 then inherits a contact structure @xmath53 , which is supported by a summed open book with empty binding , one interface torus @xmath71 and fibration @xmath72 ) = \begin{cases } \phi & \text { for $ 0 < \eta < 1/2$,}\\ -\phi & \text { for $ 1/2 < \eta < 1$. } \end{cases}\ ] ] since the pages are cylinders , @xmath73 is a partially planar domain , so corollary [ cor : separating ] implies that it is not strongly hyper - fillable .
( note that @xmath73 _ is _ stein fillable , as it can be constructed from the stein fillable torus @xmath67 by a sequence of contact @xmath74-surgeries along legendrian curves in the pre - lagrangian fibers @xmath75 . )
theorem [ thm : tos3 ] however does not apply , as there is only one irreducible subdomain and no binding .
it is not clear whether @xmath76 .
relatedly , pick a loop @xmath77 transverse to the pages and define @xmath78 to be the manifold obtained from @xmath70 by a @xmath42-surgery along @xmath79 relative to the page framing . then @xmath78 naturally carries a summed open book with one irreducible subdomain , one interface torus and one binding circle ; the pages are pairs of pants .
if @xmath80 denotes the supported contact structure on @xmath78 , a holomorphic curve argument in @xcite shows that @xmath81 is not strongly fillable , but it also does not appear to have planar torsion ( at least we do not see it from this construction ) since the planar piece has no boundary .
it is not clear whether @xmath81 is cobordant to anything overtwisted , nor whether its ech contact invariant vanishes ( cf . the discussion accompanying corollary [ cor : ech ] below ) .
the cobordisms of the previous section are constructed by repeated application of two handle attaching constructions that we shall now describe .
the handles we will work with take the form @xmath82
\times s^1,\ ] ] where in each case @xmath61 is a compact oriented surface with boundary . in the first case
, we shall attach @xmath83 to the neighborhood of a transverse link , and in the second case , @xmath84 \times s^1 $ ] is attached to the neighborhood of a disjoint union of pre - lagrangian tori .
it is important however to understand that these constructions are not truly _ local _ , as the attaching requires neighborhoods that are in some sense sufficiently large .
this condition on the neighborhoods is most easily stated in the language of ( possibly blown up and summed ) open book decompositions that is not necessarily the only natural setting in which these operations make sense , but it is the first that comes to mind .
we have derived considerable inspiration from the symplectic capping technique introduced by eliashberg in @xcite .
the goal of eliashberg s construction was somewhat different , namely to embed any weak symplectic filling into a closed symplectic manifold , but it can also be used to construct symplectic cobordisms between contact manifolds with supporting open books that are related to each other by capping off binding components .
indeed , eliashberg s
capping construction works as follows : 1 . given @xmath29 with a supporting open book @xmath85 ,
attach @xmath45-handles to @xmath86\times m$ ] at @xmath87 along each component of @xmath88 via the page framing .
this transforms @xmath49 by a @xmath42-surgery along each binding component , producing a new @xmath0-manifold @xmath78 with a fibration @xmath89 whose fibers are the closed surfaces obtained by capping the original pages with disks .
any symplectic structure on @xmath86\times m$ ] dominating @xmath53 can then be extended over the handles so that the fibers of @xmath89 are symplectic .
2 . cap off the boundary of the cobordism above by presenting @xmath78 as the boundary of a lefschetz fibration over the disk with closed fibers .
the first step can be generalized by observing that if we choose to attach @xmath45-handles along some but _ not all _ components of the binding , then the new manifold @xmath78 inherits an open book decomposition @xmath90 obtained from @xmath91 by _ capping off _ the corresponding boundary components of the pages ( cf .
@xcite ) , and we will show that the symplectic structure can always be arranged so as to produce a weak symplectic cobordism from @xmath29 to @xmath81 , where @xmath80 is supported by @xmath92 . under some additional topological assumptions one can actually arrange the weak cobordism to be strong ; this variation on eliashberg
s construction has already been worked out in detail by gay and stipsicz @xcite . to generalize further
, one can also imagine replacing the usual @xmath45-handle @xmath93 by @xmath1 for any compact orientable surface @xmath61 .
we shall carry out this generalization below , though the reader may prefer to pretend @xmath94 on first reading , and this suffices for most of the applications we will discuss .
the key to our construction will be to combine the above brand of handle attachment with a `` blown up '' version , in which a round handle is attached to @xmath29 along a @xmath45-torus that can be thought of as a blown up binding circle .
this is most naturally described in the language of _ blown up summed open books _ , a generalization of open book decompositions that was introduced in @xcite and will be reviewed in more detail in
[ sec : busobd ] .
rougly speaking , a blown up summed open book on a compact @xmath0-manifold @xmath49 , possibly with boundary , defines a fibration @xmath95 where the _ binding _ @xmath96 is an oriented link and the _ interface _
@xmath97 is a set of disjoint @xmath45-tori , and the connected components of the fibers , which intersect @xmath98 transversely , are called _
pages_. an ordinary open book is the special case where @xmath99 , and in general we allow any or all of @xmath88 , @xmath100 and @xmath98 to be empty , so there may be closed pages . as with ordinary open books
there is a natural notion of contact structures being _ supported _ by a blown up summed open book , in which case binding components become positively transverse links and interface and boundary components become pre - lagrangian tori .
such a contact structure exists and is unique up to deformation unless the pages are closed .
suppose @xmath29 is a closed contact @xmath0-manifold containing a compact @xmath0-dimensional submanifold @xmath55 , possibly with boundary , which carries a blown up summed open book @xmath101 that supports @xmath102 and has nonempty binding . pick a set of binding components @xmath103 each of which comes with a natural framing determined by the pages adjacent to @xmath104 , called the * page framing*. for each @xmath105 , we identify a tubular neighborhood @xmath106 of @xmath107 with the oriented solid torus @xmath108 via this framing so that @xmath109 with the correct orientation and the fibration @xmath91 takes the form @xmath110 on @xmath111 , where @xmath112 denote polar coordinates on the disk , normalized so that @xmath113 .
assign to @xmath114 its natural orientation as the boundary of @xmath115 and denote by @xmath116 the distinguished positively oriented homology basis for which @xmath117 is a meridian and @xmath118 is the longitude determined by the page framing .
denote @xmath119 .
now pick a compact , connected and oriented surface @xmath61 with @xmath120 boundary components @xmath121 and choose an orientation preserving diffeomorphism of each @xmath122 to @xmath123 , thus defining a coordinate @xmath124 for @xmath122 . using this
, we define new compact oriented manifolds @xmath125 by gluing in @xmath126 via orientation reversing diffeomorphisms @xmath127 that take the form @xmath128 in the chosen coordinates . on the level of homology , the map @xmath129 identifies @xmath130 $ ] with @xmath118 and @xmath131 $ ] for @xmath132 with @xmath117 . in the special case @xmath133 , the operation just defined is simply a dehn surgery along a binding component @xmath134 with framing @xmath42 relative to the page framing .
the fibration @xmath135 extends smoothly over @xmath126 as the projection to the second factor , thus @xmath136 inherits from @xmath101 a natural blown up summed open book @xmath137 , with binding @xmath138 , interface @xmath100 and pages that are obtained from the pages of @xmath101 by attaching @xmath139 , gluing @xmath140 to the boundary component adjacent to @xmath107 .
we say that @xmath137 is obtained from @xmath101 by * @xmath61-capping surgery along @xmath141*. if @xmath137 does not have closed pages , then it supports a contact structure @xmath80 on @xmath136 which can be assumed to match @xmath53 outside the region of surgery , and thus extends to @xmath78 .
the @xmath61-capping surgery can also be defined by attaching a generalized version of a @xmath34-dimensional @xmath45-handle : define @xmath142 with boundary @xmath143 the above identifications of the neighborhoods @xmath115 with @xmath108 yield an identification of @xmath144 with @xmath145 , which we use to attach @xmath146 to the trivial cobordism @xmath86 \times m$ ] by gluing @xmath147 to @xmath148 , defining @xmath149 \times m ) \cup_{{{\mathcal n}}(b_0 ) } { \mathcal{h}}_\sigma,\ ] ] which after smoothing the corners has boundary @xmath150 we will refer to the oriented submanifolds @xmath151 \times b_0 ) \cup_{b_0 } ( - \sigma \times \{0\ } ) \subset w\ ] ] and @xmath152 for an arbitrary interior point @xmath153 as the * core * and * co - core * respectively .
note that @xmath154 , @xmath155 and @xmath156 , where @xmath157 denotes the algebraic count of intersections .
the following generalizes results in @xcite and @xcite .
[ thm:2handles ] suppose @xmath9 is a symplectic form on @xmath86 \times m$ ] with @xmath158 , and let @xmath6 denote the handle cobordism defined in , after smoothing corners . then after a symplectic deformation of @xmath9 away from @xmath159 , @xmath9 can be extended symplectically over @xmath6 so that it is positive on @xmath160 , @xmath161 and the pages of @xmath137 . moreover , if the latter pages are not closed , then @xmath9 also dominates a supported contact structure @xmath80 on @xmath78 , thus defining a weak symplectic cobordism from @xmath29 to @xmath81
. we will refer to the cobordism @xmath10 of theorem [ thm:2handles ] henceforward as a * @xmath61-capping cobordism*. in general it is a weak cobordism , but under certain conditions that depend only on the topology of the setup , it can also be made strong .
recall the standard fact ( cf . @xcite*prop .
4.1 ) that whenever @xmath10 has a boundary component @xmath49 on which @xmath9 dominates a positive contact structure @xmath53 and is exact , @xmath9 can be deformed in a collar neighborhood to make @xmath49 convex , with @xmath53 as the induced contact structure . in
[ sec : cohomology ] we will use routine mayer - vietoris arguments to characterize the situations in which this trick can be applied to the above construction .
[ thm:2handlecohomology ] let @xmath10 be the weak symplectic cobordism of theorem [ thm:2handles ] , and choose a real @xmath14-cycle @xmath162 in @xmath163 such that @xmath164 \in h_1(m ; { { \mathbb r}})$ ] is poincar dual to the restriction of @xmath9 to @xmath159 .
then there is a number @xmath165 such that @xmath166 ) = [ 0,1 ] \times [ h ] + c\ , [ { \mathcal{k}'}_\sigma ] \in h_2(w,{\partial}w ; { { \mathbb r}}),\ ] ] where @xmath167 denotes the poincar - lefschetz duality isomorphism . in particular ,
if @xmath168 is concave then the following conditions are equivalent : 1 .
@xmath9 is exact .
2 . @xmath169 = 0 \in h_2(w,{\partial}w ; { { \mathbb r}})$ ] .
3 . @xmath170 + \ldots + [ \gamma_n]$ ] is not torsion in @xmath171 .
moreover , assuming again that @xmath159 is concave , the following conditions are also equivalent : 1 .
@xmath10 can be arranged to be a strong symplectic cobordism from @xmath29 to @xmath81 .
2 . @xmath172 = 0 \in h_1(m ' ; { { \mathbb r}})$ ] .
@xmath173 is not torsion in @xmath174 , where @xmath118 denote the longitudes on @xmath114 determined by the page framing .
it should be emphasized that the above theorem assumes @xmath61 is _
connected_. the case where @xmath61 is disconnected is equivalent to performing multiple surgery operations in succession , but the statement of theorem [ thm:2handlecohomology ] would then become more complicated . for the case
@xmath94 , if @xmath134 denotes the binding component where @xmath42-surgery is performed , then theorem [ thm:2handlecohomology ] means that @xmath9 will be exact on @xmath6 if and only if @xmath104 is not torsion in @xmath171 , and @xmath10 can be made into a strong cobordism if and only if @xmath104 has no nullhomologous cover whose page framing matches its seifert framing .
an equivalent condition is assumed in @xcite , which only constructs strong cobordisms .
[ remark : notexact ] though @xmath9 in the above construction is sometimes an exact symplectic form , @xmath10 is _ never _ an exact cobordism , i.e. it does not admit a global primitive that restricts to suitable contact forms on both boundary components .
this follows immediately from the observation that the core @xmath175 is a symplectic submanifold whose oriented boundary is a _ negatively transverse _ link in @xmath29 , hence if @xmath176 and @xmath177 defines a contact form on @xmath29 with the proper co - orientation , then @xmath178 a contradiction .
a similar remark applies to the round handle cobordism considered in theorems [ thm : roundhandles ] and [ thm : roundhandlecohomology ] below .
the non - exactness of @xmath10 is important because there are examples in which it is known that no exact cobordism from @xmath29 to @xmath81 exists ( see [ sec : nonexact ] ) . to describe the blown up version of these results , we continue with the same setup as above and choose a set of interface tori , @xmath179 together with an orientation for each @xmath180
there is then a distinguished positively oriented homology basis @xmath181 where @xmath118 is represented by some oriented boundary component of a page adjacent to @xmath182 , and @xmath117 is represented by a closed leaf of the characteristic foliation defined on @xmath182 by @xmath53 .
choose tubular neighborhoods @xmath183 of @xmath182 and identify them with @xmath184 \times s^1 $ ] to define positively oriented coordinates @xmath185 in which @xmath186 $ ] and @xmath187 $ ] .
we may then assume that for every @xmath188 the loop @xmath189 is legendrian , and the fibration @xmath91 takes the form @xmath190 denote the two oriented boundary components of @xmath191 by @xmath192 where we define the oriented tori @xmath193 with corresponding homology bases @xmath194 such that @xmath195 denote the union of all the neighborhoods @xmath191 by @xmath196 . then writing two identical copies of @xmath61 as @xmath197 and choosing a positively oriented coordinate @xmath124 for each boundary component @xmath198
, we construct new compact oriented manifolds @xmath199 from @xmath49 and @xmath200 respectively by gluing along orientation reversing diffeomorphisms @xmath201 that take the form @xmath202 in the chosen coordinates .
thus in homology , @xmath203 \in h_1({\partial}_j \sigma_\pm \times s^1)$ ] is identified with @xmath204 and @xmath205 \in h_1({\partial}_j\sigma_\pm \times s^1)$ ] with @xmath206 .
once again the fibration @xmath135 extends smoothly over the glued in region @xmath207 as the projection to @xmath123 , so @xmath208 inherits from @xmath101 a natural blown up summed open book @xmath137 , with interface @xmath209 , binding @xmath88 and fibers that are obtained from the fibers of @xmath101 by attaching @xmath210 along the boundary components adjacent to @xmath211 .
we say that @xmath137 is obtained by * @xmath61-decoupling surgery along @xmath211*. the choice of the term _ decoupling _ is easiest to justify in the special case @xmath94 : then the surgery cuts open @xmath49 along @xmath47 and glues in two solid tori that cap off the corresponding boundary components of the pages .
even if @xmath49 is connected , @xmath78 may in general be disconnected , and there is a ( possibly empty ) component @xmath212 defined as the union of all the closed pages of @xmath137 .
denote @xmath213 , so that @xmath214 on @xmath215 there is a contact structure @xmath80 which matches @xmath53 away from the region of surgery and is supported by @xmath137 in @xmath216 .
the above surgery corresponds topologically to the attachment of a * round handle * : denote the annulus by @xmath217 \times s^1\ ] ] and define @xmath218 with boundary @xmath219 where we identify the two connected components of @xmath220 with @xmath221 via the orientation preserving maps @xmath222 using the identifications of the neighborhoods @xmath191 with @xmath223 \times s^1 $ ] chosen above , we can identify @xmath224 \times s^1\ ] ] with @xmath196 and use this to attach @xmath225 to @xmath86 \times m$ ] by gluing @xmath226 to @xmath227 , defining an oriented cobordism @xmath228 \times m ) \cup_{{{\mathcal n}}({{\mathcal i}}_0 ) } { \widehat{\mathcal{h}}}_\sigma\ ] ] with boundary @xmath229 .
use the coordinates @xmath230 \times s^1 $ ] on each @xmath231 to define an oriented link @xmath232 as the union of all the loops @xmath233 then the core and co - core respectively can be defined as oriented submanifolds by @xmath234 \times \widehat{b}_0 ) \cup_{\widehat{b}_0 } ( -\sigma \times \{(0,0)\ } ) \subset w\ ] ] and @xmath235 for an arbitrary interior point @xmath153 .
we have @xmath236 , @xmath237 and @xmath238 .
[ thm : roundhandles ] suppose @xmath9 is a symplectic form on @xmath86 \times m$ ] which satisfies @xmath158 and @xmath239 and @xmath6 denotes the round handle cobordism of . then after a symplectic deformation away from @xmath159 , @xmath9 can be extended symplectically over @xmath6 so that it is positive on @xmath240 , @xmath241 and the pages of @xmath137 , and @xmath9 dominates a supported contact structure @xmath80 on @xmath215 . in particular , after capping @xmath242 by attaching a lefschetz fibration over the disk as in @xcite , this defines a weak symplectic cobordism from @xmath29 to @xmath243
. we will refer to @xmath10 in this construction from now on as a * @xmath61-decoupling cobordism*. the homological condition is clearly not removable since the @xmath45-cycles @xmath244 $ ] both become nullhomologous in @xmath78 .
note that here the chosen orientations of the tori @xmath182 play a role , i.e. they can not in general be chosen arbitrarily unless @xmath9 is exact .
no such issue arose in theorem [ thm:2handles ] because @xmath9 is always exact on a neighborhood of a binding circle .
this is the reason why the `` @xmath46-separating '' condition is needed for many of the results in [ sec : results ] , and there are easy examples to show that those theorems are not true without it ( cf . remark [ remark : needomegaseparating ] ) . for the analogue of theorem [ thm:2handlecohomology ] in this context
, we shall restrict for simplicity to the case where @xmath245 vanishes for every @xmath180 . note that in this case , the poincar dual of @xmath246 can be represented by a real @xmath14-cycle in @xmath247 .
[ thm : roundhandlecohomology ] suppose @xmath248 for each @xmath249 , @xmath10 is the weak cobordism constructed by theorem [ thm : roundhandles ] , and @xmath162 is a real @xmath14-cycle in @xmath247 with @xmath164 \in h_1(m ; { { \mathbb r}})$ ] poincar dual to the restriction of @xmath9 to @xmath159
. then one can choose @xmath9 such that there is a number @xmath165 with @xmath166 ) = [ 0,1 ] \times [ h ] + c\ , [ { \widehat{\mathcal{k}}'}_\sigma ] \in h_2(w,{\partial}w ; { { \mathbb r}}).\ ] ] in particular , if @xmath168 is concave then the following conditions are equivalent : 1 .
@xmath9 is exact .
2 . @xmath250 = 0 \in h_2(w,{\partial}w ; { { \mathbb r}})$ ] .
3 . there are no integers @xmath251 with @xmath252 and @xmath253 such that the homology class @xmath254 is trivial in @xmath171 . moreover ,
if @xmath159 is concave and @xmath255 , then the following conditions are also equivalent : 1 .
@xmath10 can be arranged to be a strong symplectic cobordism from @xmath29 to @xmath81 .
2 . @xmath256 = 0 \in h_1(m ' ; { { \mathbb r}})$ ] .
3 . there are no integers @xmath257 with @xmath258 and @xmath259 such that @xmath260 finally , if @xmath242 and @xmath215 are both nonempty , assume the labels are chosen so that @xmath261 and @xmath262 , and consider the weak cobordism @xmath263 from @xmath29 to @xmath243 obtained by capping off @xmath242 with a symplectic lefschetz fibration @xmath264 as in @xcite .
the following conditions are then equivalent : 1 .
@xmath10 can be arranged to be a strong symplectic cobordism from @xmath29 to @xmath243 .
2 . @xmath265 = 0 \in h_1(m'_{\text{convex } } ; { { \mathbb r}})$ ] .
3 . there are no integers @xmath251 with @xmath252 and @xmath266 such that the homology class @xmath267 is trivial in @xmath268 .
though they never form exact cobordisms ( cf .
remark [ remark : notexact ] ) , the capping and decoupling cobordisms do have a nice property in common with exact cobordisms : under favorable circumstances , they induce natural maps on embedded contact homology ( cf .
recall that for a closed contact @xmath0-manifold @xmath29 and homology class @xmath269 , @xmath270 is the homology of a chain complex generated by sets of reeb orbits with multiplicities whose homology classes add up to @xmath162 , with a differential counting embedded index @xmath14 holomorphic curves with positive and negative cylindrical ends in the symplectization @xmath271 .
similarly , counting embedded index @xmath45 holomorphic curves through a generic point in @xmath49 yields the so - called @xmath272-map , @xmath273 the * ech contact invariant * @xmath274 is the homology class represented by the `` empty orbit set '' .
it is equivalent via an isomorphism of taubes @xcite to a corresponding invariant in seiberg - witten theory , and conjecturally also to the ozsvth - szab contact invariant @xcite ; like those invariants , its vanishing gives an obstruction to strong symplectic fillings , and a version with twisted coefficients also obstructs weak fillings .
technically the definitions of @xmath270 and @xmath275 depend not just on @xmath53 but also on a choice of contact form and almost complex structure .
however , taubes isomorphism to seiberg - witten floer homology implies that they are actually independent of these choices , thus we are safe in writing @xmath270 without explicitly mentioning the extra data . an argument due to eliashberg shows that @xmath276 whenever @xmath29 is overtwisted , and a much more general computation in @xcite established the same result whenever @xmath29 has planar @xmath40-torsion for any @xmath41 .
it is therefore natural to ask whether the latter result can be recovered as a consequence of theorem [ thm : overtwisted ] .
this is highly non - obvious , and we will not attempt to prove it here in full generality : the trouble is that non - exact cobordisms do not in general give rise to well behaved maps on ech . recently however , hutchings @xcite has shown that such maps do exist for the case @xmath277 whenever @xmath10 is a strong cobordism with an exact symplectic form
hutchings refers to such cobordisms as _ weakly exact_. an easy corollary is thus the following : [ cor : ech ] suppose @xmath10 is a strong symplectic cobordism from @xmath19 to @xmath20 arising from theorem [ thm:2handles ] or [ thm : roundhandles ] such that @xmath9 is exact .
then there is a @xmath272-equivariant map @xmath278 that takes @xmath279 to @xmath280 . for the example of a @xmath45-handle cobordism constructed from an ordinary open book decomposition ,
the analogue of corollary [ cor : ech ] in heegaard floer homology has been established by john baldwin @xcite .
let us now discuss a conjectural generalization of corollary [ cor : ech ] which could remove the condition on @xmath9 .
recall that for any closed @xmath45-form @xmath46 on @xmath49 , one can define ech with * twisted coefficients * in the group ring @xmath281 $ ] , which we shall abbreviate by @xmath282\big).\ ] ] here the differential keeps track of the homology classes in @xmath283 of the holomorphic curves being counted , see @xcite .
the @xmath272-map can again be defined as a degree @xmath284 map on @xmath285 , and the twisted contact invariant @xmath286 is again the homology class in @xmath287 generated by the empty orbit set .
the vanishing results in @xcite give convincing evidence that a more general version of the map in corollary [ cor : ech ] should exist , in particular with the following consequence : [ conj : c ] suppose @xmath10 is a @xmath61-capping or @xmath61-decoupling cobordism from @xmath19 to @xmath20 , and write @xmath288 . then : 1 .
if @xmath289 vanishes , then so does @xmath290 .
if @xmath289 is in the image of the map @xmath291 on @xmath292 for some @xmath43 , then @xmath290 is in the image of @xmath291 on @xmath293 . the first part of the conjecture would reduce both the untwisted and twisted vanishing results in @xcite to the fact , proved essentially by eliashberg in the appendix of @xcite , that the fully twisted contact invariant vanishes for every overtwisted contact manifold .
the second part is related to another result proved in @xcite , namely the twisted ech version of the planarity obstruction of oszvth - stipsicz - szab @xcite in heegaard floer homology : if @xmath29 is planar , then @xmath286 is in the image of @xmath291 for all @xmath40 and all @xmath46 .
if the conjecture holds , then this fact follows from theorem [ thm : planar ] and the computation of @xmath294 . the obvious way to try to prove conjecture [ conj
: c ] would be by constructing a @xmath272-equivariant map @xmath295 which takes @xmath289 to @xmath290 . due to the non - exactness of @xmath9 and a resulting lack of energy bounds
, it seems unlikely that such a map would exist in general , but a more probable scenario is to obtain a map @xmath296 where @xmath297 is a novikov completion of @xmath298 $ ] , and we take advantage of the natural inclusions @xmath299 to define the ech of @xmath300 with coefficients in @xmath297 . in cases
where @xmath5 has connected components with closed leaves , one would expect this map to involve also the periodic floer homology ( cf .
@xcite ) of the resulting mapping tori . defining such a map would require a slightly more careful construction of the weak cobordism @xmath10 , such that both boundary components inherit stable hamiltonian structures which can be used to attach cylindrical ends and define reasonable moduli spaces of finite energy punctured holomorphic curves .
this can always be done due to a construction in @xcite , which shows that suitable stable hamiltonian structures exist for any desired cohomology class on the boundary .
it is probably also useful to observe that for an intelligent choice of data , the holomorphic curves in @xmath10 with no positive ends can be enumerated precisely : we will show in proposition [ prop : nopositive ] that all of them arise from the symplectic core of the handle .
we shall now give some concrete examples of capping and decoupling cobordisms and survey a few more applications , including new proofs of several known results and one or two new ones . in @xcite ,
holomorphic curve arguments were used to show that planar torsion is a filling obstruction , but theorems [ thm : overtwisted ] and [ thm : weak ] make these proofs much easier by using essentially `` soft '' methods to reduce them to the well - known result of gromov - eliashberg that overtwisted contact manifolds are not weakly fillable .
this does not of course make everything elementary , as the gromov - eliashberg theorem still requires some technology the original proof used a `` bishop family '' of holomorphic disks with totally real boundary , and these days one can instead use punctured holomorphic curves , seiberg - witten theory or heegaard floer homology if preferred . while this technological overhead is probably not removable , we can use a decoupling cobordism to simplify the level of technology a tiny bit : namely we can reduce it to the following standard fact whose proof requires only _ closed _ holomorphic spheres , e.g. the methods used in @xcite .
[ lemma : spheres ] if @xmath10 is a connected weak filling of a nonempty contact manifold @xmath29 , then it contains no embedded symplectic sphere with vanishing self - intersection .
this lemma follows essentially from mcduff s results @xcite , but by today s standards it is also easy to prove on its own : if one chooses a compatible almost complex structure to make the boundary @xmath3-convex and the embedded symplectic sphere @xmath3-holomorphic , then vanishing self - intersection implies that the latter lives in a smooth @xmath45-dimensional moduli space of holomorphic spheres that foliate @xmath6 ( except at finitely many nodal singularities ) . this forces some leaf of the foliation to hit the boundary tangentially , thus contradicting @xmath3-convexity .
every weakly fillable contact manifold is tight . a schematic diagram of the proof is shown in figure [ fig : overtwisted ] .
suppose @xmath10 is a weak filling of @xmath29 and the latter is overtwisted .
then @xmath29 contains a planar @xmath42-torsion domain - torsion relies on eliashberg s classification of overtwisted contact structures @xcite , quite a large result in itself .
the original `` bishop disk '' argument of gromov and eliashberg had the advantage of not requiring this . ]
@xmath200 , whose planar piece @xmath301 is a solid torus with disk - like pages , attached along an interface torus @xmath302 to another subdomain whose pages are not disks . since @xmath303 = 0 \in h_2(m)$ ] , @xmath304 and we can attach a @xmath305-decoupling cobordism along @xmath47 , producing a larger symplectic manifold @xmath306 whose boundary has two connected components @xmath307 of which the latter carries a contact structure @xmath80 dominated by @xmath9 . the component @xmath242 has closed sphere - like pages , and is thus the trivial symplectic fibration @xmath308 . after capping @xmath242 by a symplectic fibration @xmath309 , we then obtain a weak filling of @xmath243 containing a symplectic sphere with vanishing self - intersection , contradicting lemma [ lemma : spheres ] . a proof of the gromov - eliashberg theorem using a @xmath305-decoupling cobordism .
the left shows the effect on the pages when a round handle @xmath310 is attached to a planar @xmath42-torsion domain .
the right shows the resulting cobordism and consequent weak filling which furnishes a contradiction to lemma [ lemma : spheres ] . ]
a related argument appears in @xcite , using the fact that overtwisted contact manifolds always have giroux torsion ; see also [ sec : gay ] below .
let @xmath312 with coordinates @xmath313 and define for @xmath314 the contact structure @xmath315.\ ] ] these contact structures are all tight , but eliashberg showed in @xcite that they are not strongly fillable for @xmath316 , which follows from the fact that disjoint unions of multiple copies of @xmath33 are not fillable , together with the following : for any @xmath314 , @xmath317 is symplectically cobordant to the disjoint union of @xmath318 copies of the tight @xmath0-sphere .
the torus @xmath317 admits a supporting summed open book decomposition with @xmath8 irreducible subdomains @xmath319 each having cylindrical pages and trivial monodromy , attached to each other along @xmath8 interface tori @xmath320 such that @xmath321 for @xmath322 .
attaching round handles @xmath323 along every second interface torus @xmath324 yields a weak symplectic cobordism to the disjoint union of @xmath318 copies of the tight @xmath325 ( figure [ fig : t3 ] ) .
the latter is also supported by an open book with cylindrical pages and trivial monodromy , so we can attach a @xmath45-handle @xmath93 along one binding component to create a weak cobordism to the tight @xmath326 .
the resulting weak cobordism from @xmath63 to @xmath311 can be deformed to a strong cobordism since the symplectic form is necessarily exact near @xmath311 .
[ remark : needomegaseparating ] note that @xmath317 is always _ weakly _
fillable @xcite , and indeed , the above cobordism can not be attached to any weak filling @xmath10 of @xmath317 for which @xmath327 .
this shows that the homological condition in theorem [ thm : roundhandles ] can not be removed .
the torus @xmath328 can be constructed out of four irreducible subdomains containing cylindrical pages with trivial monodromy .
attaching one @xmath305-decoupling cobordism yields an overtwisted @xmath325 , and one can then attach a second one to obtain a disjoint union of two copies of the tight @xmath325 . ]
recall that a contact manifold @xmath29 is said to have * giroux torsion * @xmath329 if it admits a contact embedding of @xmath330 \times t^2 , \xi_n)$ ] , where @xmath331 is given by ; we write @xmath332 if there are no such embeddings and @xmath333 if they exist for arbitrarily large @xmath318
. every contact manifold with positive giroux torsion also has planar @xmath14-torsion ( see @xcite ) , thus as a special case of theorem [ thm : overtwisted ] , every @xmath29 with @xmath334 is symplectically cobordant to something overtwisted ; this was proved by david gay in @xcite for @xmath335 .
a concrete picture of this cobordism and the one that is constructed explicitly in @xcite for the case @xmath335 are _ weak _ cobordisms , not strong in general .
as david gay has pointed out to me , these can always be turned into strong cobordisms by attaching additional @xmath45-handles to make the positive boundary an overtwisted rational homology sphere ( see the proof of theorem [ thm : overtwisted ] in [ sec : proofs ] ) . ]
is shown in figure [ fig : giroux ] : namely , if @xmath334 , then @xmath49 contains a domain @xmath86 \times t^2 \cong m_0 \subset m$ ] on which @xmath53 is supported by a blown up summed open book with three irreducible subdomains @xmath336 each with cylindrical pages attached to each other in a chain along two interface tori @xmath337 . attaching a round handle @xmath323 along
@xmath338 produces a weak symplectic cobordism to a new contact manifold @xmath78 , containing the disconnected domain @xmath208 shown in figure [ fig : giroux ] : in particular @xmath4 and @xmath339 are each transformed into subdomains @xmath340 and @xmath341 with disk - like pages .
now @xmath342 contains an overtwisted disk ; indeed , it is a planar @xmath42-torsion domain .
observe that this construction can also be used to show that @xmath29 is not _ weakly _
fillable if the torsion domain separates @xmath49 , as then @xmath343 for any symplectic form @xmath9 arising from a weak filling . attaching a @xmath305-decoupling cobordism to @xmath29 with giroux torsion at least @xmath14 yields an overtwisted contact manifold @xmath81 . attaching one more yields the disjoint union of a contact manifold with a trivial symplectic @xmath344-fibration over @xmath123 .
] gay s proof in @xcite that giroux torsion obstructs strong filling did not directly use the above cobordism , but proved instead that @xmath29 with @xmath334 admits a symplectic cobordism to some non - empty contact manifold such that the cobordism itself contains a symplectic sphere with vanishing self - intersection
gay s argument then used gauge theory to derive a contradiction if @xmath29 has a filling , but one can just as well use lemma [ lemma : spheres ] above .
a close relative of gay s cobordism construction is easily obtained from the above picture : attaching round handles @xmath323 along both @xmath338 and @xmath345 , the top of the cobordism contains a connected component with closed sphere - like pages ( the top picture in figure [ fig : giroux ] ) , which can be capped by @xmath346 to produce a cobordism that contains symplectic spheres of self - intersection number @xmath42 .
gromov s theorem @xcite on the non - existence of exact lagrangians in @xmath349 provides perhaps the original example of a pair of contact manifolds that are strongly but not exactly cobordant : indeed , viewing @xmath67 as the boundary of any lagrangian torus in the standard strong filling of the tight @xmath0-sphere @xmath33 , we obtain @xmath350 the nonexistence of the exact cobordism here can also be proved by the argument of hofer @xcite mentioned in the introduction , which implies that if @xmath29 admits a reeb vector field with no contractible periodic orbit , then @xmath351 whenever either @xmath81 is overtwisted or @xmath352 .
together with theorem [ thm : overtwisted ] , this implies that for any @xmath38 overtwisted and @xmath316 , @xmath353 a subtler obstruction to exact cobordisms is defined in joint work of the author with janko latschev @xcite via symplectic field theory , leading to the following example . for any integer @xmath354 , suppose @xmath61 is a closed , connected and oriented surface of genus @xmath355 , and @xmath356 is a multicurve consisting of @xmath40 disjoint embedded loops which divide @xmath61 into exactly two connected components @xmath357 such that @xmath358 has genus @xmath42 and @xmath359 has genus @xmath360 . by a construction due to lutz @xcite , the product @xmath361 then admits a unique ( up to isotopy ) @xmath123-invariant contact structure @xmath362 such that the convex surfaces @xmath363 have dividing set @xmath364 .
the contact manifold @xmath365 then has planar @xmath44-torsion , as the two subsets @xmath366 can be regarded as the irreducible subdomains of a supporting summed open book with pages @xmath367 , so we view @xmath368 as the planar piece and @xmath369 as the padding ( see definition [ defn : planartorsion ] ) . in particular
, @xmath365 is overtwisted if and only if @xmath370 , and for @xmath371 it has a reeb vector field with no contractible periodic orbits .
it turns out in fact that each increment of @xmath40 contains an obstruction to exact fillings that is invisible in the non - exact case .
[ thm : janko ] if @xmath372 then for any @xmath355 and @xmath373 , @xmath374 the nonexistence of the exact cobordism is a result of @xcite .
the existence of the non - exact cobordism follows immediately from corollary [ cor : equivalence ] , but in certain cases one can construct it much more explicitly as in figure [ fig : mk ] . in particular ,
@xmath365 is supported by a summed open book consisting of the two irreducible subdomains @xmath366 with pages @xmath367 attached along @xmath40 interface tori .
attaching @xmath310 along one of the interface tori gives a weak @xmath305-decoupling cobordism to @xmath375 .
theorem [ thm : roundhandlecohomology ] then implies that this can be deformed to a strong cobordism , as the restriction of the symplectic form to @xmath376 is poincar dual to a multiple of the boundary of the co - core ; the latter consists of two loops of the form @xmath377 with opposite orientations and is thus nullhomologous .
an explicit cobordism from @xmath378 to @xmath379 as in the proof of theorem [ thm : janko ] can be realized as a @xmath305-decoupling cobordism . ]
any compact , connected and oriented surface @xmath61 with boundary , together with a diffeomorphism @xmath380 fixing the boundary , determines a contact @xmath0-manifold @xmath381 , namely the one supported by the open book decomposition with page @xmath61 and monodromy @xmath382 .
recall that the mapping class of the monodromy map @xmath382 is said to be * reducible * if it has a representative that preserves some multicurve @xmath356 such that no component of @xmath383 is a disk or an annulus .
consider the simple case in which @xmath382 preserves each individual connected component @xmath384 and also preserves its orientation ( note that this is always true for some iterate of @xmath382 ) . in this case we may assume after a suitable isotopy that @xmath382 is the identity on a neighborhood of @xmath385 , so that for some open annular neighborhood @xmath386 of each curve @xmath384 , @xmath387 contains a thickened torus region @xmath388 on which the open book decomposition is the projection to the first factor .
let @xmath389 denote the union of all the neighborhoods @xmath390 and define the possibly disconnected compact surface @xmath391 with boundary ; then @xmath382 restricts to this surface as an orientation preserving diffeomorphism @xmath392 that preserves each connected component and equals the identity near @xmath393 .
denote the connected components of @xmath394 by @xmath395 and the corresponding restrictions of @xmath396 by @xmath397 for @xmath398 . since each @xmath399 necessarily has nonempty boundary , each gives rise to a connected contact manifold @xmath400 .
[ thm : reducible ] given a reducible monodromy map @xmath380 as described above and a closed @xmath45-form @xmath46 on @xmath401 which dominates the respective contact structures , there exists a weak symplectic cobordism @xmath10 from @xmath402 such that @xmath9 matches @xmath46 at the negative boundary .
if @xmath403 , then one can also arrange so that the negative boundary is concave and the restriction of @xmath9 to the positive boundary is poincar dual to a positive multiple of @xmath404 \in h_1(m_\varphi;{{\mathbb r}}),\ ] ] where the summation is over the connected components of @xmath364 and @xmath405 denotes an arbitrarily chosen point .
the cobordism is a stack of @xmath406-capping cobordisms , constructed by attaching handles of the form @xmath407
\times s^1 \times { { \mathbb d}}$ ] via theorem [ thm:2handles ] along all pairs of binding circles in @xmath408 that correspond to the same curve in @xmath364 .
the co - core of each of these handles is a disk with boundary of the form @xmath409 , thus the cohomology class of @xmath9 at the positive boundary follows immediately from theorem [ thm:2handlecohomology ] .
if the contact manifolds @xmath400 for @xmath410 are all weakly fillable , then so is @xmath411 .
[ remark : johnjeremy ] john baldwin @xcite has observed that topologically , the cobordism of theorem [ thm : reducible ] can also be obtained by performing boundary connected sums on the pages and then using @xmath305-capping cobordisms to remove extra boundary components ; in @xcite this is used to deduce a relation between the ozsvth - szab contact invariants of ( @xmath412 ) and the pieces @xmath413 .
additionally , jeremy van horn - morris and john etnyre have pointed out to me that if one also assumes every component of @xmath383 to intersect @xmath414 , then one can replace the weak cobordism of theorem [ thm : reducible ] with a stein cobordism .
this does not appear to be possible if any component of @xmath383 has its full boundary in @xmath364 .
let us say that a connected contact @xmath0-manifold @xmath29 is * maximally cobordant to @xmath326 * if there exists a compact connected @xmath34-manifold @xmath6 with @xmath66 such that for every closed @xmath45-form @xmath46 on @xmath49 with @xmath50 , there is a symplectic form @xmath9 on @xmath6 with @xmath51 defining a weak symplectic cobordism from @xmath29 to @xmath33 .
theorem [ thm : planar ] says that every planar contact manifold is maximally cobordant to @xmath326 .
it turns out that this suffices to give an alternative proof of the planarity obstruction in @xcite .
[ thm : etnyre ] suppose @xmath29 is maximally cobordant to @xmath326 .
then every weak semi - filling of @xmath29 has connected boundary and negative definite intersection form .
let @xmath415 be the compact @xmath34-manifold with @xmath416 guaranteed by the assumption , and suppose @xmath417 is a weak filling of @xmath418 , where @xmath81 is some other contact manifold , possibly empty .
if @xmath419 is defined by gluing these two along @xmath49 , then by assumption @xmath9 can be extended over @xmath415 so that @xmath10 becomes a weak filling of @xmath420 , implying that @xmath78 must be empty since @xmath33 is not weakly hyper - fillable .
now @xmath9 is exact near @xmath421 , so without loss of generality we may assume @xmath10 is a strong filling of @xmath33 .
we claim that the map induced on homology @xmath422 by the inclusion @xmath423 is injective . indeed ,
if @xmath424 satisfies @xmath425 , then obviously @xmath426 is also nonzero and thus @xmath427 .
if @xmath428 but @xmath424 is nontrivial , we can pick any closed @xmath45-form @xmath429 on @xmath430 with @xmath431 and replace @xmath9 by @xmath432 for any @xmath433 sufficiently small so that @xmath434 remains a weak filling of @xmath29 .
then @xmath432 also extends over @xmath415 , so that the above argument goes through again to prove that @xmath435 is nontrivial .
finally , we use the fact that the strong fillings of @xmath33 have been classified : by a result of gromov @xcite and eliashberg @xcite , @xmath6 is necessarily diffeomorphic to a symplectic blow - up of the @xmath34-ball , i.e. @xmath436 since the latter has a negative definite intersection form and @xmath437 is injective , the result follows . our proof of theorem [ thm : planar ] combined with conjecture [ conj : c ] would also reprove the algebraic planarity obstruction established in @xcite , which is the twisted ech version of a heegaard floer theoretic result by due to oszvth , stipsicz and szab @xcite . note that the condition of being maximally cobordant to @xmath326 does not require @xmath29 to be fillable .
it is also not clear whether there can exist non - planar contact manifolds that also satisfy this condition ; the author is unaware of any known invariants that would be able to detect this distinction .
[ question : planar ] is there a non - planar contact @xmath0-manifold which is maximally cobordant to @xmath326 ?
the filling obstruction known as planar torsion was introduced in @xcite with mainly holomorphic curves as motivation , as it provides the most general setting known so far in which the existence and uniqueness of certain embedded holomorphic curves leads to a vanishing result for the ech contact invariant . in light of our cobordism construction , however , one can now provide an alternative motivation for the definition in purely symplectic topological terms .
the first step is to understand what kinds of blown up summed open books automatically support overtwisted contact structures : using eliashberg s classification theorem @xcite and giroux s criterion ( cf .
@xcite ) , this naturally leads to the notion of planar @xmath42-torsion .
then a more general blown up summed open book defines a planar @xmath40-torsion domain for some @xmath354 if and only if it can be transformed into a planar @xmath42-torsion domain by a sequence of @xmath305-capping and @xmath305-decoupling surgeries ; this is the essence of proposition [ prop : ktok-1 ] proved below .
from this perspective , the definition of planar torsion and the crucial role played by blown up summed open books seem completely natural .
more generally , the partially planar domains are precisely the blown up summed open books for which a sequence of @xmath305-capping and @xmath305-decoupling cobordisms can be used to construct a symplectic cap that contains a symplectic @xmath42-sphere . as far
as the author is aware , almost all existing uniqueness or classification results for symplectic fillings ( e.g. @xcite ) apply to contact manifolds that admit caps of this type .
however , it does not always suffice to construct an appropriate cap and then apply mcduff s results @xcite : e.g. the classification of strong fillings of planar contact manifolds in terms of lefschetz fibrations @xcite truly relies on _ punctured _ holomorphic curves , as there is no obvious way to produce a lefschetz fibration with bounded fibers out of a family of holomorphic spheres in a cap . finally
, we remark that while theorems [ thm : overtwisted ] and [ thm : weak ] substantially simplify the proof that planar torsion is a filling obstruction , they do not reproduce all of the results in @xcite : in particular the technology of embedded contact homology is not yet far enough along to deduce the vanishing of the contact invariant from a non - exact cobordism .
moreover , a proof using capping and decoupling cobordisms simplifies the technology needed but does not remove it , as a simplified version of the very same technology is required to prove the gromov - eliashberg theorem ( cf .
[ sec : gromoveliashberg ] ) . from the author s own perspective , the idea for constructing symplectic cobordisms out of these types of handles would never have emerged without a holomorphic curve picture in the background ( cf .
figure [ fig : jhol ] ) , and as we will discuss in [ sec : holomorphic ] , after one has constructed the symplectic structure , it is practically no extra effort to add a foliation by embedded @xmath3-holomorphic curves which reproduces the @xmath3-holomorphic blown up open books of @xcite on both boundary components .
the moral is that whether one prefers to prove non - fillability results by direct holomorphic curve arguments or by constructing cobordisms to reduce them to previously known results , it is essentially the same thing : neither proof would be possible without the other .
the plan for proving the main results is as follows .
we begin in [ sec : busobd ] by reviewing the fundamental definitions involving blown up summed open books and planar torsion , culminating with the ( more or less obvious ) observation that one can always use capping or decoupling surgery to decrease the order of a planar torsion domain . in [ sec : model ] , we introduce a useful concrete model for a blown up summed open book and its supported contact structure .
this is used in [ sec : modelcobordism ] to write down a model of a weak @xmath61-decoupling cobordism , and minor modifications explained in [ sec : notround ] yield a similar model for the @xmath61-capping cobordism .
it then remains to show that any given symplectic form on @xmath86 \times m$ ] can be deformed so as to attach smoothly to these model cobordisms ; we show this in [ sec : deformation ] .
we address in [ sec : cohomology ] the essentially cohomological question of when the weak cobordism can be made strong , and when its symplectic form is exact . with these ingredients all in place ,
the proofs of the main results are completed in [ sec : proofs ] . finally ,
[ sec : holomorphic ] gives a brief discussion of the existence and uniqueness of holomorphic curves in the cobordisms we ve constructed .
assume @xmath49 is a compact oriented @xmath0-manifold , possibly with boundary , the latter consisting of a union of @xmath45-tori .
a * blown up summed open book * @xmath101 on @xmath49 can be described via the following data . 1 . an oriented link @xmath96 , called the * binding*. 2 .
a disjoint union of @xmath45-tori @xmath97 , called the * interface*. 3 . for each interface
torus @xmath438 a distinguished basis @xmath439 of @xmath440 , where @xmath441 is defined only up to sign .
4 . for each boundary torus
@xmath442 a distinguished basis @xmath439 of @xmath440 . 5 .
a fibration @xmath443 whose restriction to @xmath98 is a submersion .
the distinguished homology classes @xmath444 associated to each torus @xmath445 are called * longitudes * and * meridians * respectively , and the oriented connected components of the fibers @xmath446 are called * pages*. we assume moreover that the fibration @xmath91 can be expressed in the following normal forms near the components of @xmath447 . as in an ordinary open book decomposition
, each binding circle @xmath134 has a neighborhood admitting coordinates @xmath448 , where @xmath112 are polar coordinates on the disk ( normalized so that @xmath113 ) , such that @xmath449 and @xmath450 near an interface torus @xmath438 , we can find a neighborhood with coordinates @xmath451 \times s^1 $ ] such that @xmath452 with @xmath439 matching the natural basis of @xmath453 , and @xmath454 a neighborhood of a boundary torus @xmath442 similarly admits coordinates @xmath455 \times s^1 $ ] with @xmath456 and @xmath457 observe that unlike the normal form , the map is well defined at @xmath458 , since there are no polar coordinates and hence no coordinate singularity .
the above conditions imply that the closure of each page is a smoothly _ immersed _
surface , whose boundary components are each embedded submanifolds of @xmath88 , @xmath100 or @xmath98 , and in the last two cases homologous to the distinguished longitude @xmath22 .
the `` generic '' page has an embedded closure , but in isolated cases there may be pairs of boundary components that are identical as oriented @xmath14-dimensional submanifolds in @xmath100 . in general ,
any or all of @xmath88 , @xmath100 and @xmath98 may be empty , and @xmath49 may also be disconnected .
if @xmath459 we have simply a fibration @xmath460 whose fibers are closed oriented surfaces . if @xmath461 but @xmath462 and @xmath49 is connected , we have an ordinary open book .
we say that @xmath101 is * irreducible * if the fibers @xmath446 are connected , i.e. there is only one @xmath123-parametrized family of pages .
more generally , any blown up summed open book can be presented uniquely as a union of * irreducible subdomains * @xmath463 which each inherit irreducible blown up summed open books and are attached together along boundary tori ( which become interface tori in @xmath49 ) . the notion of a contact structure supported by an open book generalizes in a natural way : we say that a contact structure @xmath53 on @xmath49 is * supported * by @xmath101 if it is the kernel of a * giroux form * , a contact form whose reeb vector field is everywhere positively transverse to the pages and positively tangent to their boundaries , and which induces a characteristic foliation on @xmath464 with closed leaves parallel to the distinguished meridians .
a giroux form exists and is unique up to homotopy through giroux forms on any connected manifold with a blown up summed open book , except in the case where the pages are closed , i.e. @xmath459 .
the binding is then a positively transverse link , and the interface and boundary are disjoint unions of pre - lagrangian tori .
[ defn : planar ] an irreducible blown up summed open book is called * planar * if its pages have genus zero .
an arbitrary blown up summed open book is then called * partially planar * if its interior contains a planar irreducible subdomain , which we call a * planar piece*. a * partially planar domain * is a contact @xmath0-manifold @xmath29 , possibly with boundary , together with a supporting blown up summed open book that is partially planar . for a given closed @xmath45-form @xmath46 on @xmath49 , and a partially planar domain @xmath29 with planar piece @xmath465 , we say that @xmath29 is * @xmath46-separating * if @xmath466 for all interface tori @xmath47 of @xmath49 that lie in @xmath467 , and * fully separating * if this is true for all @xmath46 .
[ defn : symmetric ] a blown up summed open book is called * symmetric * if it has empty boundary , all its pages are diffeomorphic and it contains exactly two irreducible subdomains @xmath468 each of which has empty binding and interface .
[ defn : planartorsion ] for any integer @xmath41 , an @xmath46-separating partially planar domain @xmath29 with planar piece @xmath465 is called an * @xmath46-separating planar @xmath40-torsion domain * if it satisfies the following conditions : we say that a contact manifold @xmath29 with closed @xmath45-form @xmath46 has * @xmath46-separating planar @xmath40-torsion * if it contains an @xmath46-separating planar @xmath40-torsion domain .
if this is true for all closed @xmath45-forms @xmath46 on @xmath49 , then we say @xmath29 has * fully separating * planar @xmath40-torsion .
it was shown in @xcite that a contact manifold is overtwisted if and only if it has planar @xmath42-torsion , which is always fully separating since the interface then intersects the planar piece only at its boundary , a single nullhomologous torus .
the proofs of theorems [ thm : overtwisted ] and [ thm : weak ] thus rest on the following easy consequence of the preceeding definitions .
[ prop : ktok-1 ] if @xmath49 is a planar @xmath40-torsion domain for some @xmath354 , then it contains a binding circle @xmath104 or interface torus @xmath47 in its planar piece such that the following is true .
let @xmath78 denote the manifold with corresponding blown up summed open book obtained from @xmath49 by @xmath305-capping surgery along @xmath104 or @xmath305-decoupling surgery along @xmath47 respetively .
then some connected component of @xmath78 is a planar @xmath79-torsion domain for some @xmath472 . by assumption
, @xmath49 contains a planar piece @xmath467 with nonempty boundary , and if @xmath473 denotes a boundary component , then the pages in @xmath467 have exactly one boundary component adjacent to @xmath474 .
the pages in @xmath467 have @xmath470 boundary components , and without loss of generality we may assume no other irreducible subdomain in the interior of @xmath49 has planar pages with fewer boundary components than this .
since @xmath354 , these pages have at least one boundary component adjacent to some binding circle @xmath104 or interface torus @xmath47 distinct from @xmath474 . performing @xmath305-capping sugery to remove @xmath104 or @xmath305-decoupling surgery to remove @xmath47
produces a new manifold @xmath475 containing a planar irreducible subdomain @xmath476 whose pages have @xmath79 boundary components where @xmath79 is either @xmath40 or @xmath477 ; the latter can only result from a decoupling surgery along @xmath478 .
thus @xmath475 is a planar @xmath479-torsion domain unless it is symmetric .
the latter would mean @xmath480 , hence also @xmath481 , and @xmath482 is also irreducible and has planar pages with @xmath79 boundary components .
this can not arise from capping surgery along a binding circle or decoupling surgery along a torus in the interior of @xmath467 , as we assumed all planar pages in the interior of @xmath49 outside of @xmath467 to have at least @xmath483 boundary components .
the only remaining possibility would be decoupling surgery along @xmath484 , but then symmetry of @xmath475 would imply that @xmath49 must also have been symmetric , hence a contradiction .
assume @xmath485 is a compact contact @xmath0-manifold , possibly with boundary , supported by a blown up summed open book @xmath101 with binding @xmath88 , interface @xmath100 and fibration @xmath486 we assume that each connected component of @xmath200 contains at least one component of @xmath487 , so that @xmath101 will support a contact structure everywhere .
it will be useful to identify this with the following generalization of the notion of an _ abstract _ open book ( cf .
the closure of a fiber @xmath488 is the image of some compact oriented surface @xmath489 with boundary under an immersion @xmath490 which is an embedding on the interior .
the monodromy of the fibration then determines ( up to isotopy ) a diffeomorphism @xmath491 which preserves connected components and is the identity in a neighborhood of the boundary , and we define the mapping torus @xmath492 with @xmath493 for all @xmath494 , @xmath495 .
denote by @xmath496 the natural fibration .
let us label the connected components of @xmath497 by @xmath498 and for each @xmath499 choose an open collar neighborhood @xmath500 of @xmath501 on which @xmath502 is the identity .
denote the union of all these neighborhoods by @xmath503 .
now for each @xmath499 , choose positively oriented coordinates @xmath504 for some @xmath505 .
these neighborhoods give rise to corresponding collar neighborhoods of @xmath506 , @xmath507 which can be identified with @xmath508 via the coordinates @xmath185 .
the index set @xmath509 comes with an obvious partition @xmath510 where @xmath511 there is also a free @xmath68-action on @xmath512 defined via an involution @xmath513 such that @xmath514 if and only if @xmath515 and @xmath516 lie in the same connected component of @xmath100 . now define for each @xmath517 the domain @xmath518 \times s^1 & \text { if $ i \in i_{{\mathcal i}}$,}\\ s^1 \times [ 0,1 ] \times s^1 & \text { if $ i \in i_{\partial}$ , } \end{cases}\ ] ] and denote by @xmath185 the natural coordinates on @xmath519 , where for @xmath520 we view @xmath112 as polar coordinates on the disk with the angle normalized to take values in @xmath521 .
denote the subsets @xmath522 by @xmath523 the chosen coordinates on the neighborhoods @xmath524 then determine a gluing map @xmath525 which takes @xmath524 to @xmath519 , and we use this to define a new compact and oriented manifold , possibly with boundary , @xmath526 where the equivalence relation identifies @xmath527 for @xmath528 with @xmath529 .
this naturally contains @xmath530 and @xmath531 as submanifolds , and the fibration @xmath532 can be extended over @xmath533 so that it matches the canonical @xmath534-coordinate on @xmath519 wherever @xmath535 .
now @xmath200 can be identified with @xmath536 via a diffeomorphism that maps @xmath88 to @xmath530 and @xmath100 to @xmath531 , and transforms the fibration @xmath537 to @xmath534 .
a supported contact structure on @xmath536 can be defined as follows .
first , define a smooth @xmath14-form of the form @xmath538 where @xmath539 \to { { \mathbb r}}$ ] are smooth functions chosen to have the following properties : 1 . as @xmath540 moves from @xmath42 to @xmath14
, @xmath541 defines a path through the first quadrant from @xmath542 to @xmath543 .
2 . @xmath544 is contact on @xmath545 .
3 . @xmath546 for @xmath547 $ ] .
4 . @xmath548 for @xmath549 $ ] , for some positive number @xmath550 .
@xmath551 for @xmath552 .
[ remark : aremark ] the contact condition is satisfied if and only if @xmath553 , except at @xmath530 , where the coordinate singularity changes the condition to @xmath554 .
one consequence is that @xmath555 for @xmath556 , hence @xmath557 .
the assumption that @xmath544 is a _
@xmath14-form imposes some additional conditions , namely for @xmath520 , @xmath558 and @xmath559 must define smooth functions at the origin in @xmath560 ( in polar coordinates ) , and for @xmath528 , @xmath561 and @xmath562 can be extended smoothly over @xmath407 $ ] such that @xmath563 in particular this implies @xmath564 for @xmath565 $ ] .
we will assume these conditions are always satisfied without further comment .
the co - oriented distribution @xmath566 is a confoliation on @xmath536 , which is integrable on the mapping torus @xmath567 and outside of this is a positive contact structure . to perturb it to a global contact structure ,
choose a @xmath14-form @xmath568 on @xmath489 which satisfies @xmath569 and takes the form @xmath570 on @xmath571 . by a simple interpolation trick ( cf .
@xcite ) , @xmath568 can be used to construct a @xmath14-form @xmath572 on @xmath573 that satisfies @xmath574 choosing @xmath433 sufficiently small , we can bring @xmath575 sufficiently @xmath576-close to @xmath577 on @xmath573 so that @xmath578
. then a contact form that equals @xmath544 near @xmath579 can be defined by @xmath580 \ } \subset { { \mathcal n}}_i$ , } \\
\lambda_0 & \text { on $ \ { \rho \le r'\ } \subset { { \mathcal n}}_i$ , } \end{cases}\ ] ] where the fact that @xmath557 allows us for @xmath433 sufficiently small to choose smooth functions @xmath581 \to { { \mathbb r}}$ ] satisfying note that for @xmath528 , @xmath588 also extends naturally over @xmath407 $ ] with @xmath589 .
all contact forms that one can construct in this way are homotopic to each other through families of contact forms , so the resulting contact structure @xmath590 is uniquely determined up to isotopy .
moreover , it is easy to check that the reeb vector field determined by @xmath591 is everywhere positively transverse to the pages : in particular , @xmath591 is a giroux form for the blown up summed open book we ve constructed on @xmath536 , thus @xmath592 is contactomorphic to @xmath485 .
assume now that the manifold @xmath200 from the previous section is embedded into a closed contact @xmath0-manifold @xmath29 such that @xmath53 is an extension of the contact structure that was given on @xmath200 .
without loss of generality , we can identify @xmath485 with the abstract model @xmath592 , and assume in particular that @xmath544 and @xmath591 are @xmath14-forms on @xmath49 which restrict on @xmath200 to the models constructed above , and on a neighborhood of @xmath593 define matching contact forms whose kernel is @xmath53 .
our goal in this section is to construct a weak symplectic cobordism that realizes a @xmath61-decoupling surgery along some set of oriented interface tori @xmath594 the chosen orientation of each @xmath182 splits a tubular neighborhood @xmath183 of @xmath182 naturally into positive and negative parts @xmath595 whose intersection is @xmath182 . to simplify notation in the following , let us assume these neighborhoods are chosen and the page boundary components @xmath596 are ordered so that for each @xmath597 , @xmath598 \times s^1 \quad \text { and } \quad { { \mathcal n}}_+(t_j ) = s^1 \times [ 0,1 ] \times s^1.\ ] ] we will fix on @xmath191 the standard coordinates @xmath185 of @xmath599 , and assume all the functions chosen to define @xmath544 and @xmath591 are the same for all of these neighborhoods , so we can write @xmath600 for @xmath410 .
assume @xmath601 is a closed @xmath45-form on @xmath49 that satisfies the cohomological condition . after adding an exact @xmath45-form , we may then assume @xmath601 satisfies @xmath602 for each @xmath410 , where @xmath603 are constants satisfying @xmath604 since @xmath605 everywhere on @xmath49
, we can define an exact symplectic form on the trivial cobordism @xmath86 \times m$ ] as follows : fix any smooth , strictly increasing function @xmath606 \to { { \mathbb r}}$ ] with @xmath607 and @xmath608 uniformly small , and set @xmath609 if @xmath610 is sufficiently small then @xmath611 is symplectic and restricts positively to both @xmath53 and the pages of @xmath101 , everywhere on @xmath86 \times m$ ] .
now if @xmath612 is a sufficiently large constant , then the @xmath45-form @xmath613 also has these properties . in the following
we shall always assume @xmath614 is arbitrarily large whenever convenient . to construct a cobordism corresponding to the round handle attachment
, we shall first `` dig a hole '' in the trivial cobordism @xmath86\times m$ ] near each of the tori @xmath615 . in order to find nice coordinates near the boundary of the hole
, it will be useful to consider the vector field @xmath616 on @xmath86 \times { { \mathcal n}}(t_j)$ ] defined by the condition @xmath617 [ lemma : xtheta ] the vector field @xmath616 is locally hamiltonian with respect to @xmath618 and takes the form @xmath619 for some smooth functions @xmath620 \times [ 1,1 ] \to { { \mathbb r}}$ ] with the following properties : by a direct computation , @xmath616 takes the form with @xmath628 and @xmath88 satisfying the linear system @xmath629 \\
\varphi'(t ) g(\rho ) & \left [ \varphi(t ) + 1 \right ] g'(\rho ) \end{pmatrix } \begin{pmatrix } a(t,\rho ) \\
b(t,\rho ) \end{pmatrix } = \begin{pmatrix } 1 \\ 0 \end{pmatrix } .\ ] ] the determinant @xmath630 of this matrix is always negative since the contact condition requires @xmath631 for @xmath632 , and for @xmath621 $ ] we have @xmath633 , @xmath634 and @xmath635 . the general solution for @xmath628 and @xmath88 can thus be written as @xmath636 g'(\rho ) \\ - \varphi'(t ) g(\rho ) \end{pmatrix}.\ ] ] the stated conditions on @xmath637 and @xmath638 then follow immediately from the conditions we ve placed on @xmath639 , @xmath640 , @xmath641 and @xmath382 . due to the lemma , we can choose a smoothly embedded curve @xmath645 \to [ 1/2,1 ] \times [ -1,1 ] : \tau \mapsto ( t(\tau),\rho(\tau))\ ] ] that is everywhere transverse to the vector field and also satisfies @xmath646 and @xmath647 near @xmath648 ( see figure [ fig : hole ] ) . writing the annulus as @xmath649 \times s^1 $ ] , use the curve just chosen to define an embedding @xmath650 \times { { \mathcal n}}(t_j ) : ( \theta,\tau,\phi ) \mapsto ( t(\tau),\theta,\rho(\tau),\phi),\ ] ] which traces out a smooth hypersurface @xmath651 \times { { \mathcal n}}(t_j)$ ] that meets @xmath87 transversely at the pair of tori @xmath652 .
denote by @xmath653 \times m\ ] ] the interior of the component of @xmath330 \times m ) \setminus h_{t_j}$ ] that contains @xmath615 .
observe that by construction , @xmath654 lies entirely within @xmath655 \times { { \mathcal n}}(t_j)$ ] , and the locally hamiltonian vector field @xmath616 points transversely outward at @xmath656 .
thus for sufficiently small @xmath657 , we can use the flow @xmath658 of @xmath616 to parametrize a neighborhood of @xmath659 in @xmath660 by an embedding @xmath661 \times s^1 \times { { \mathbb a}}&\hookrightarrow [ 0,1 ] \times m \\ ( \sigma,\theta,\tau,\phi ) & \mapsto \varphi_{x_\theta}^{\sigma-1}(\psi(\theta , \tau,\phi ) ) .
\end{split}\ ] ] [ lemma : pullback ] we have @xmath662 where @xmath663 is an @xmath123-invariant @xmath14-form on @xmath664 that satisfies @xmath665 \ , d\phi \quad\text { near $ \{\tau = \pm 1\ } = { \partial}(s^1 \times { { \mathbb a}})$},\ ] ] and @xmath666 everywhere . in @xmath86 \times { { \mathcal n}}(t)$ ]
we can write @xmath667 , where @xmath668 then defining @xmath669 on @xmath664 , we have @xmath670 and can write @xmath663 explicitly near @xmath648 by plugging in @xmath671 , @xmath672 , @xmath673 , @xmath674 and @xmath675 , hence @xmath676 \ , d\phi \end{split}\ ] ] as desired . since @xmath677 is invariant under the @xmath123-action by translation of @xmath678 , @xmath663 is also @xmath123-invariant .
the claim @xmath666 is a consequence of the fact that @xmath659 is transverse to the vector field @xmath616 , which is @xmath611-dual to @xmath679 : indeed , ignoring combinatorial factors we find @xmath680 it follows that @xmath681 is positive since this is obviously true near @xmath648 .
the formula now follows from the fact that @xmath682 and @xmath616 has a symplectic flow .
since @xmath683 acts trivially on the coordinates @xmath534 and @xmath678 , the lemma also yields a formula for the pullback of @xmath618 , namely @xmath685 + c_j\ , d\phi \wedge d\theta.\ ] ] for each @xmath410 , denote by @xmath686 \times { { \mathcal n}}(t_j)$ ] the image of the map @xmath683 as constructed above ( figure [ fig : hole2 ] ) : @xmath687 thus inherits negatively oriented coordinates @xmath688 \times s^1 \times [ -1,1 ] \times s^1 $ ] in which @xmath618 has the form given in .
we are now ready to write down a smooth model of the round handle attachment . as in [ sec : intro ] , assume @xmath61 is a compact , connected and oriented surface with @xmath120 boundary components @xmath689 near each component @xmath122 , identify a collar neighborhood @xmath690 with @xmath691 \times s^1 $ ] and denote the resulting oriented coordinates by @xmath692 .
then denote the union of all the subsets @xmath654 by @xmath693 and define the cobordism @xmath694 \times m ) \setminus { { \mathcal u}}_{{{\mathcal i}}_0}\right ) \cup ( -\sigma \times { { \mathbb a}})\ ] ] by removing @xmath693 from @xmath86 \times m$ ] and replacing it by the handle @xmath695 , gluing @xmath696 to @xmath687 via the natural identification of the coordinates @xmath697 .
this yields a smooth @xmath34-manifold with two boundary components @xmath698 where we identify @xmath49 with @xmath159 and write @xmath699 using the identification @xmath700 defined in .
the oriented surfaces @xmath701 now glue together smoothly with the fibers @xmath702 in @xmath247 to form the pages of the natural blown up summed open book @xmath137 on @xmath78 obtained from @xmath101 by @xmath61-decoupling surgery along @xmath211 .
it remains to define a suitable symplectic form on @xmath703 that matches near @xmath704 and is positive on these pages .
[ lemma : extendingomega0 ] there exists a symplectic form on @xmath695 that matches near @xmath704 and is positive on the oriented surfaces @xmath705 for any @xmath706 and @xmath707 for any @xmath708 , and makes @xmath709 and @xmath710 into symplectically orthogonal symplectic subspaces everywhere along @xmath711 .
we will use a standard deformation trick to simplify on each of the regions @xmath696 so that it can be extended as a split symplectic form .
choose a @xmath14-form @xmath712 on @xmath406 with @xmath713 and lift it in the obvious way to @xmath664 .
since @xmath714 > 0\ ] ] and @xmath663 has no @xmath715-term near @xmath716 , we can also arrange for @xmath712 to match @xmath663 on a neighborhood of @xmath716 .
next choose a smooth cutoff function @xmath717 \to [ 0,1]$ ] that satisfies @xmath718 near @xmath719 and @xmath720 near @xmath721 , and use this to define a smooth function @xmath722 $ ] which satisfies @xmath723 we observe that the expression @xmath724 now gives a well - defined @xmath14-form on @xmath703 by lifting @xmath712 from @xmath406 to @xmath703 and @xmath663 from @xmath664 to @xmath725 \times s^1 \times { { \mathbb a}}= { { \mathcal v}}_j \times { { \mathbb a}}$ ] in the obvious ways .
choose also a smooth function @xmath726 \to [ 1,\infty)$ ] satisfying @xmath727 and @xmath728 near @xmath729 , and a @xmath14-form @xmath441 on @xmath61 such that @xmath730 a suitable symplectic form on @xmath703 can then be defined by @xmath731 by construction , @xmath732 matches near @xmath704 , while near @xmath711 and outside of the regions @xmath696 it takes the split form @xmath733 which is symplectic and makes each of the spaces spanned by @xmath734 and @xmath735 into symplectic subspaces which are symplectically orthogonal to each other . to test whether @xmath732 is symplectic on @xmath696 ,
we compute @xmath736 \\ & + \beta \beta ' \ , d\sigma \wedge ( \eta - \eta_0 ) \wedge \left [ \beta\ , d\eta + ( 1 - \beta)\ , d\eta_0 \right ] .
\end{split}\ ] ] the first term is always nonzero since @xmath737 and @xmath738 are both positive .
the whole expression is therefore nonzero whenever either @xmath739 or @xmath740 is sufficiently large , and we are free to choose @xmath502 so that it increases fast in the region where @xmath741 is not constant .
this choice also ensures @xmath742 everywhere on @xmath696 . to find a symplectic extension of over @xmath703 , choose now a closed @xmath14-form @xmath743 on @xmath61 which takes the form @xmath744 near each boundary component @xmath122 ; this is possible due to the homological condition .
then if @xmath732 denotes the extension of @xmath745 given by lemma [ lemma : extendingomega0 ] , we extend as @xmath746 whenever @xmath614 is sufficiently large , lemma [ lemma : extendingomega0 ] implies that this form is also symplectic and restricts positively to the surfaces @xmath707 and @xmath705 if @xmath153 lies outside a neighborhood of the boundary .
this implies that it is positive on the pages of @xmath137 , as well as on the core @xmath747 \times \widehat{b}_0 ) \cup ( - \sigma \times \{(0,0)\ } ) \subset w,\ ] ] and co - core @xmath748 to summarize : we have constructed a smooth cobordism @xmath6 with symplectic form @xmath749 that matches @xmath618 near @xmath750 and is positive on the core and co - core and on the pages of the induced blown up summed open book at the other boundary component @xmath78 .
an appropriate confoliation @xmath14-form @xmath751 can now be defined on @xmath78 by @xmath752 where we use @xmath534 to denote the natural @xmath123-coordinate on @xmath753 .
the distribution @xmath754 is then tangent to the pages on the glued in region , hence @xmath755 .
it follows that on any connected component of @xmath78 that does not contain closed pages , @xmath756 has a perturbation to a contact structure @xmath80 that is supported by @xmath137 and dominated by @xmath749 .
[ remark : coreislarge ] it will be useful later to observe that @xmath757 is not only positive but can be assumed to be arbitrarily large .
in fact it _ must _ in general be large due to the deformation trick used in the proof of lemma [ lemma : extendingomega0 ] . [
remark : nokappa ] if the constants @xmath758 all vanish , i.e. @xmath759 on @xmath196 , then one can choose the @xmath14-form @xmath743 in the above construction to be identically zero .
this has the useful consequence that for any @xmath760 $ ] and any closed embedded loop @xmath761 outside a neighborhood of @xmath414 , the torus @xmath762 is lagrangian . more generally ,
if @xmath761 is any properly embedded compact @xmath14-dimensional submanifold transverse to @xmath414 , then @xmath763 indeed , with @xmath764 it is equivalent to show that the integral of @xmath765 vanishes , and using we find @xmath766 since @xmath441 vanishes on the @xmath123-factor in @xmath767 . since @xmath663 is @xmath123-invariant on @xmath664 , this integral does nt depend on the position of any point in @xmath768 but only on the algebraic count of these points , which is zero , thus @xmath769 the above construction works essentially the same way for the handle @xmath1 , so we will be content to briefly summarize the differences . here
we pick binding components @xmath770 and denote the corresponding solid torus neighborhoods by @xmath771 with coordinates @xmath185 , viewing @xmath112 as polar coordinates , and denote the union of these neighborhoods by @xmath144 .
the model symplectic form @xmath618 on the trivial cobordism @xmath86 \times m$ ] is again defined via and , with the difference that since every closed @xmath45-form on @xmath115 is exact , we can assume ( after adding an exact @xmath45-form ) that @xmath601 _ vanishes _ on all of these neighborhoods .
the role of @xmath659 is now played by a hypersurface @xmath772 \times { { \mathcal n}}(\gamma_j)\ ] ] parametrized by an embedding @xmath773 \times m,\ ] ] thus defining a similar set of coordinates @xmath774 on @xmath775 , where @xmath776 are now _ polar _ coordinates on @xmath305 .
we can again arrange @xmath775 to be transverse to the vector field @xmath616 , defined exactly as before , and then use its flow to parametrize a neighborhood of @xmath775 in the region @xmath777 that it bounds via a map @xmath661 \times s^1 & \times { { \mathbb d}}\hookrightarrow [ 0,1 ] \times { { \mathcal n}}(\gamma_j ) : ( \sigma,\theta,\tau,\phi ) \mapsto \varphi^{\sigma-1}_{x_\theta}(\psi(\theta,\phi,\tau ) ) \end{split}\ ] ] for which @xmath745 again takes the form @xmath778 for some @xmath14-form @xmath663 on @xmath108 that satisfies @xmath779 \
, d\phi\ ] ] near @xmath780 and @xmath666 everywhere .
denote the image of @xmath683 corresponding to each @xmath107 by @xmath687 , with negatively oriented coordinates @xmath781 \times s^1 \times { { \mathbb d}}$ ] .
writing the union of the regions @xmath782 as @xmath783 , the smooth cobordism is then defined by @xmath784 \times m ) \setminus { { \mathcal u}}_{b_0 } ) \cup ( -\sigma \times { { \mathbb d}}),\ ] ] where @xmath1 is glued in by identifying @xmath785 with @xmath687 so that the coordinates match .
this has boundary @xmath229 , where @xmath750 and @xmath786 hence the glued in region @xmath126 carries the coordinates @xmath787 near its boundary . choosing a @xmath14-form @xmath712 on @xmath305 that matches @xmath663 near @xmath788 and satisfies @xmath713 , the interpolation trick can again be used to deform @xmath611 in a collar neighborhood of @xmath83 so that it admits a symplectic extension over the rest of @xmath1 in the form @xmath789 .
the resulting form @xmath790 is symplectic everywhere on @xmath6 and is also positive on the pages of @xmath137 at @xmath78 if @xmath614 is sufficiently large , as well as on the core @xmath791 \times b_0 ) \cup ( -\sigma \times \{0\ } ) \subset w\ ] ] and the co - core @xmath792 for an appropriate choice of @xmath153 .
the confoliation @xmath14-form extends smoothly over @xmath126 as @xmath793 , so that @xmath749 is also positive on @xmath754 and thus dominates any contact form obtained as a small perturbation . to apply the constructions of the previous sections in proving theorems [ thm:2handles ] and [ thm : roundhandles ]
, we must show that any given symplectic form @xmath9 on @xmath86 \times m$ ] dominating @xmath53 can be deformed away from @xmath159 to reproduce the model @xmath794 where @xmath606 \to { { \mathbb r}}$ ] is a smooth function with @xmath795 , @xmath607 and @xmath610 small , @xmath601 is some closed @xmath45-form on @xmath49 in the appropriate cohomology class , and @xmath612 is a constant that we can assume to be as large as necessary . the following application of a standard moser deformation argument ( cf .
@xcite*lemma 2.3 ) will be useful .
[ lemma : moser ] suppose @xmath10 is a symplectic @xmath34-manifold , @xmath49 is a closed oriented @xmath0-manifold with an embedding @xmath796 and @xmath22 is a @xmath14-form on @xmath49 that satisfies @xmath797 .
then for sufficiently small @xmath433 , @xmath798 extends to an embedding @xmath799 such that @xmath800 and @xmath801 .
observe that if @xmath802 is an oriented hypersurface in a symplectic @xmath34-manifold @xmath10 with a positive contact structure @xmath53 , then @xmath9 dominates @xmath53 if and only if it satisfies @xmath803 for every contact form @xmath22 on @xmath29 . using the obvious variants of lemma [ lemma : moser ]
when the hypersurface is a positive or negative boundary component of @xmath6 , we obtain the following useful consequence : [ lemma : gluing ] suppose @xmath29 is a closed contact @xmath0-manifold and @xmath804 \times m , \omega_-)$ ] and @xmath805 are two symplectic manifolds such that the restrictions of @xmath806 and @xmath807 to @xmath159 define the same @xmath45-form @xmath46 on @xmath49 , with @xmath50 .
then for any small @xmath433 , @xmath808 admits a symplectic form which matches @xmath807 on @xmath809 and @xmath806 on @xmath810 \times m$ ] .
[ prop : deformation ] suppose @xmath29 is any closed contact @xmath0-manifold with contact form @xmath22 , @xmath544 is a @xmath14-form on @xmath49 satisfying @xmath811 , @xmath9 is a symplectic form on @xmath86 \times m$ ] with @xmath158 , and @xmath601 is a closed @xmath45-form on @xmath49 with @xmath812 = [ \omega|_{tm } ] \in h^2_{{\operatorname{dr}}}(m)$ ] .
then for any @xmath813 and sufficiently large @xmath612 , there exists a symplectic form @xmath814 on @xmath86 \times m$ ] that matches @xmath9 on a neighborhood of @xmath159 and takes the form on @xmath815 \times m$ ] . by lemma [ lemma : moser ]
we can assume without loss of generality that @xmath9 has the form @xmath816 near @xmath159 , where @xmath46 is the closed @xmath45-form on @xmath49 defined as the restriction of @xmath9 to @xmath159 .
the proof now proceeds in two steps , of which the first is to put the symplectic structure @xmath618 of into a slightly simpler form via a coordinate change near @xmath159 .
define the @xmath14-form @xmath817 on @xmath86 \times m$ ] and write @xmath818 , so @xmath819 .
let @xmath820 denote the vector field that is @xmath618-dual to @xmath821 , i.e. @xmath822 . for @xmath614 sufficiently large , @xmath820 is then a small perturbation of the vector field that is @xmath611-dual to @xmath823 , which is a liouville ( with respect to @xmath611 ) vector field positively transverse to @xmath159 since @xmath824 is contact .
hence we may assume @xmath820 is also positively transverse to @xmath159 and use its flow @xmath825 to define an embedding @xmath826 \times m : ( t , m ) \mapsto \varphi_v^t(m)\ ] ] for @xmath433 sufficiently small .
if @xmath827 denotes the reeb vector field determined by @xmath22 , along @xmath159 we then have @xmath828 and @xmath829 hence @xmath830 matches the symplectic form @xmath831 pointwise at @xmath159 , and another moser deformation argument thus allows us to isotope the embedding @xmath502 so that @xmath830 takes this form on some neighborhood of @xmath159 .
equivalently , this means @xmath618 admits a deformation to a new symplectic form @xmath749 which takes the form @xmath832 on an arbitrarily small neighborhood of @xmath159 and matches the original @xmath618 outside a slightly larger neighborhood . for step two ,
we show that the given @xmath9 can be deformed outside a small neighborhood of @xmath159 to a new symplectic form @xmath814 that matches outside a slightly larger neighborhood .
indeed , choose a constant @xmath833 large enough so that @xmath834 and since @xmath46 and @xmath601 are cohomologous by assumption , choose a @xmath14-form @xmath663 on @xmath49 such that @xmath835 . for some @xmath657 small , choose a cutoff function @xmath836 that equals @xmath42 near @xmath837 and @xmath14 near @xmath838 , and define @xmath839 with @xmath840 \to [ 0,\infty)$ ] a smooth function satisfying if @xmath639 is chosen to increase sufficiently fast , then @xmath814 is symplectic ( cf .
@xcite*lemma 2.9 ) , and this can always be arranged if @xmath612 is made sufficiently large .
this depends in particular on the fact that the @xmath45-forms @xmath46 and @xmath844 are both positive on @xmath53 .
the restrictions of @xmath814 and @xmath749 to the hypersurface @xmath845 now match , thus the two can be glued together smoothly by lemma [ lemma : gluing ] .
assume first that @xmath846 \times m ) \cup { \mathcal{h}}_\sigma$ ] is a @xmath61-capping cobordism , with @xmath847 attached along a neighborhood @xmath144 of @xmath848 .
write @xmath849 and @xmath850 due to [ sec : deformation ] , we may assume without loss of generality that @xmath46 has the form @xmath851 where @xmath612 is arbitrarily large , @xmath22 is the usual contact form on @xmath49 and @xmath601 vanishes on @xmath144 . the decomposition of @xmath6 into @xmath86 \times m$ ] and @xmath146 , which intersect at @xmath148 , gives rise to the mayer - vietoris sequence , @xmath852 in which @xmath853 , @xmath854 and @xmath855 .
thus there is an isomorphism @xmath856 denote by @xmath857 and @xmath858 the natural inclusions . then @xmath859 )
= [ { \partial}_j\sigma ] \in h_1(\sigma ) = h_1({\mathcal{h}}_\sigma)$ ] , so since @xmath61 is connected , @xmath860 is isomorphic to @xmath861 and is generated by @xmath170 + \ldots + [ \gamma_n]$ ] .
it follows that the second summand in consists of all integer multiples of @xmath170 + \ldots + [ \gamma_n]$ ] which are also in @xmath862 , i.e. it is isomorphic to @xmath861 if @xmath170 + \ldots + [ \gamma_n]$ ] is torsion in @xmath171 , and is otherwise trivial . in the former case ,
let @xmath863 be the smallest number for which @xmath864 + \ldots + [ \gamma_n ] ) = 0 \in h_1(m)$ ] , and construct a cycle @xmath865 in the form @xmath866,\ ] ] where @xmath867 is any @xmath45-chain in @xmath159 with @xmath868 + \ldots + [ \gamma_n])$ ] and @xmath175 is the core .
the isomorphism implies that everything in @xmath869 is an element of @xmath870 plus an integer multiple of .
let @xmath162 denote a real @xmath14-cycle in @xmath163 such that @xmath164 = { \operatorname{pd}}([\omega ] ) \in h_1(m;{{\mathbb r}})$ ] ; note that this is always possible since @xmath46 is necessarily exact on @xmath144 .
the product @xmath86 \times h$ ] then represents a relative homology class in @xmath871 .
it suffices to show that for every @xmath873 , the evaluation of @xmath9 on @xmath628 matches the intersection product @xmath874 \times [ h ] + c [ { \mathcal{k}'}_\sigma ] \right).\ ] ] for any @xmath875 this is immediately clear since @xmath876,\ ] ] where the latter is the intersection product in @xmath49 , and @xmath628 does not intersect anything in the handle . by , either the image of @xmath877 is the entirety of @xmath869 or there is one more generator @xmath878 $ ] . for the latter we have @xmath879 and @xmath880 \times [ h ] + c [ { \mathcal{k}'}_\sigma ] \right ) = c_m \bullet [ h ] + k_0 c,\ ] ] so is satisfied if and only if @xmath881 \right).\ ] ] this is positive without loss of generality since @xmath9 can be defined to make @xmath882 arbitrarily large ( cf .
remark [ remark : coreislarge ] ) .
the above argument also shows that if @xmath168 is concave , then @xmath9 can never be exact if @xmath170 + \ldots + [ \gamma_n ] \in h_1(m)$ ] is torsion , even without assuming @xmath882 to be arbitrarily large .
indeed , in this case we have @xmath883 for a contact form @xmath22 on @xmath29 , and @xmath164 = 0 $ ] , hence @xmath884 and @xmath885 on the other hand if @xmath170 + \ldots + [ \gamma_n ] \in h_1(m)$ ] is not torsion , then @xmath870 generates everything in @xmath869 , so @xmath886 always vanishes since @xmath46 is exact .
this proves the first half of theorem [ thm:2handlecohomology ] .
we also conclude from the above that if @xmath168 is concave , then there is a constant @xmath165 such that @xmath887 ) = c [ { \partial}{\mathcal{k}'}_\sigma ] \in h_1(m ' ; { { \mathbb r}}),\ ] ] so the second half of the theorem is proved by showing that @xmath172 = 0 \in h_1(m ' ; { { \mathbb r}})$ ] if and only if the stated homological condition on @xmath888 is satisfied . writing @xmath889 , we obtain the mayer - vietoris sequence @xmath890 where @xmath891 , with each component @xmath114 carrying the two distinguished generators @xmath892 defined in [ sec : intro ] .
denote the inclusions @xmath893 and @xmath894 .
then @xmath895 \in h_1(\sigma \times s^1)$ ] and @xmath896 \in h_1(\sigma \times s^1)$ ] , so @xmath897 consists of all classes of the form @xmath898 with @xmath899 and @xmath266 .
now , @xmath172 $ ] is represented by the cycle @xmath900 , and it vanishes in @xmath901 if and only if @xmath902 = 0\ ] ] for every @xmath903 .
this is true if and only if the image of the map @xmath904 in the above sequence contains only cycles of the form @xmath905 . in light of the above description of @xmath897 ,
this is true if and only if @xmath906 for all @xmath907 .
this completes the proof of theorem [ thm:2handlecohomology ] .
the proof of theorem [ thm : roundhandlecohomology ] proceeds similarly : assume @xmath846 \times m ) \cup { \widehat{\mathcal{h}}}_\sigma$ ] is a @xmath61-decoupling cobordism , with @xmath908 attached along a neighborhood @xmath196 of @xmath909 , write @xmath229 , @xmath910 and @xmath911 .
we again assume @xmath46 takes the form of , and also impose the extra condition @xmath912 in this case we can find a real @xmath14-cycle @xmath162 in @xmath913 that represents @xmath914 ) \in h_1(m;{{\mathbb r}})$ ] . without changing the cohomology class or the symplectic properties of @xmath9 , we can then also assume that @xmath601 is supported in a tubular neighborhood of the cycle @xmath162 .
recall from
[ sec : intro ] that each oriented torus @xmath180 comes with a distinguished homology basis @xmath915 , where @xmath118 is a boundary component of a page and @xmath117 is represented by a legendrian loop in @xmath182 .
this also gives rise to bases @xmath916 of @xmath917 , where the orientation of @xmath918 is reversed compared with @xmath117 . for @xmath846 \times m ) \cup { \widehat{\mathcal{h}}}_\sigma$ ] we have the mayer - vietoris sequence @xmath919 and resulting isomorphism @xmath920 denote the generator of @xmath921 by @xmath922 $ ] , which can also naturally be regarded as a primitive class in @xmath923 . then writing the inclusions as @xmath924 and @xmath925
, we have @xmath926 $ ] and @xmath927 $ ] , hence @xmath897 consists of all classes of the form @xmath898 for @xmath899 with @xmath266 . for any @xmath928 of this form which is also nullhomologous in @xmath49 , we can form a cycle @xmath929 as follows . first
choose a @xmath45-chain @xmath867 in @xmath159 with @xmath930 .
choose also a @xmath14-chain @xmath79 in @xmath61 with boundary in @xmath414 such that the @xmath45-chain @xmath931 in @xmath225 has boundary @xmath932 which is always possible since @xmath266 .
we can represent @xmath79 by a properly immersed submanifold in @xmath61 so that by remark [ remark : nokappa ] , @xmath933 . now extend @xmath931 to a @xmath45-chain in @xmath6 with boundary in @xmath159 by attaching trivial cylinders over the appropriate covers of legendrian representatives of @xmath117 .
since these cylinders are lagrangian , this construction yields an immersed submanifold @xmath934 which satisfies @xmath935 and @xmath936 , with @xmath937 = - \sum_j m_j \mu_j \in h_1({{\mathcal i}}_0)$ ] .
we define @xmath929 by @xmath938 + k[{\widehat{\mathcal{k}}}_\sigma].\ ] ] the goal is again to prove @xmath939 \times [ h ] + c [ { \widehat{\mathcal{k}}'}_\sigma ] \right)\ ] ] for every @xmath873 , and it is again immediate if @xmath875 .
it is also clear for @xmath940 , as @xmath941 is generated by classes of the form @xmath942 $ ] for @xmath943 , hence both sides of vanish ( see remark [ remark : nokappa ] ) .
the rest of @xmath869 is generated by classes of the form @xmath944 defined in , for which @xmath945 in light of .
similarly , @xmath946 does not intersect either @xmath86 \times h$ ] or @xmath241 , thus @xmath947 \times [ h ] + c [ { \widehat{\mathcal{k}}'}_\sigma ] \right ) = c_m \bullet [ h ] + kc,\ ] ] and is thus satisfied if and only if @xmath948 \right),\ ] ] which is positive if @xmath949 is made sufficiently large . to see that this formula for @xmath950 does nt depend on any choices , observe that if @xmath46 is exact , then @xmath277 and @xmath951 , so @xmath952 = c \int_{{\partial}c_m } \lambda\ ] ] is proportional to @xmath40 , as the integral of @xmath22 vanishes on all the meridians @xmath117 .
when @xmath46 is not exact but equals @xmath844 with @xmath601 supported in a tubular neighborhood of @xmath162 , we can find a real homology class @xmath953 with @xmath954 = c_m \bullet [ h]$ ] and thus define a real @xmath45-chain @xmath955 with @xmath956 and @xmath957 = 0 $ ]
. then up to relative homology , @xmath958 can be represented by a real linear combination of immersed surfaces that have no geometric intersection with @xmath162 , hence @xmath959 .
now since @xmath960 $ ] , @xmath952 = \int_{c_m ' } \omega = c \int_{c_m ' } d\lambda = c \int_{{\partial}c_m } \lambda,\ ] ] and this is again proportional to @xmath40 . if @xmath168 is concave , then writing @xmath277 and @xmath883 gives @xmath961 for any cycle @xmath962 with @xmath266 that is nullhomologous in @xmath49 .
since @xmath963 is also positively proportional to @xmath40 , this proves that @xmath9 is exact if and only if there is no such nullhomologous cycle @xmath364 with @xmath252 .
moreover , @xmath964 ) = c [ { \partial}{\widehat{\mathcal{k}}'}_\sigma ] \in h_1(m';{{\mathbb r}})$ ] for some @xmath165 , so it remains to characterize the situations where this homology class vanishes .
write @xmath965 and consider the resulting mayer - vietoris sequence @xmath966 where @xmath967 is freely generated by the @xmath968 cycles @xmath969 .
denote the inclusions @xmath970 and @xmath971 , where the latter maps @xmath972 into @xmath753 .
then @xmath973 \in h_1(\sigma_\pm \times s^1)$ ] and @xmath974 \in h_1(\sigma_\pm \times s^1)$ ] , thus @xmath897 consists of all classes of the form @xmath975 with @xmath976 satisfying @xmath259 .
the co - core @xmath241 has two boundary components , one generating each of the cycles @xmath977 , which we will denote by @xmath922_\pm \in h_1(m')$ ] .
thus @xmath256 $ ] vanishes in @xmath901 if and only if @xmath978_+ + [ s^1]_- ) = 0\ ] ] for every @xmath903 .
this is true if and only if the image of the map @xmath979 in the above sequence contains only cycles of the form with @xmath980 , meaning that cycles of this form with @xmath981 are never trivial in @xmath268 .
we ve now characterized the cases in which @xmath982 is globally exact on @xmath78 ; of course this never happens if @xmath983 since the latter then contains closed pages on which @xmath982 is positive .
if both @xmath215 and @xmath242 are nonempty , then the interesting question is when @xmath982 will be exact on @xmath215 , which is the case if and only if @xmath984 = 0 \in h_1(m'_{\text{convex } } ; { { \mathbb r}}).\ ] ] assuming the labels are chosen so that @xmath261 and @xmath262 , @xmath265 $ ] is now represented by @xmath985 , and a repeat of the above argument shows that this cycle vanishes if and only if @xmath967 contains no cycle of the form which vanishes in @xmath268 and has @xmath986 .
we observe however that in this situation , @xmath211 must separate @xmath49 so that each @xmath987 lies in a different connected component of @xmath988 from each @xmath989 , hence a cycle of the form vanishes in @xmath268 if and only if both the plus and minus parts vanish .
our condition is thus reduced to the nonexistence of a cycle @xmath267 with @xmath907 that vanishes in @xmath268 . to prove theorem [ thm : weak ] , suppose @xmath29 contains an @xmath46-separating planar @xmath40-torsion domain @xmath200 for some closed @xmath45-form @xmath46 with @xmath50 and an integer @xmath354 .
then @xmath466 for every interface torus @xmath47 in @xmath200 that lies in the planar piece , so we are free to remove any such torus by attaching a @xmath305-decoupling cobordism whose symplectic structure matches @xmath46 at @xmath49 . by proposition [ prop : ktok-1 ] , one can find a binding component @xmath104 or interface torus @xmath47 such that if @xmath10 with @xmath229 denotes the result of attaching the corresponding @xmath305-capping or @xmath305-decoupling cobordism respectively , then @xmath78 contains a planar torsion domain of order either @xmath477 or @xmath990 .
writing @xmath911 , the latter is also @xmath982-separating since near each of the remaining interface tori , which lie outside the region of surgery , @xmath9 is still cohomologous to the original @xmath46 .
the process can therefore be repeated until the manifold at the top has planar @xmath42-torsion , meaning it is overtwisted .
theorem [ thm : overtwisted ] is essentially the special case of theorem [ thm : weak ] for which we assume @xmath46 is exact to start with , except that the above argument actually gives a _ weak _ symplectic cobordism @xmath10 from @xmath29 to some overtwisted @xmath38 , where we can assume @xmath991 is concave and @xmath992 is not necessarily convex , but @xmath993 .
this can now be turned into a strong cobordism by the following trick which was suggested to me by david gay : first , observe that if @xmath994 is a rational homology sphere , then @xmath9 is exact near @xmath994 and can thus be deformed to make @xmath994 convex .
otherwise , take any knot @xmath995 that is not torsion in @xmath996 , and assume without loss of generality ( e.g. by performing a full lutz twist somewhere disjoint from @xmath997 and appealing to eliashberg s classification theorem @xcite ) that @xmath997 is disjoint from some overtwisted disk . then after a @xmath576-small perturbation to make @xmath997 legendrian
, one can attach a symplectic @xmath45-handle along @xmath997 to so that the new positive boundary becomes an overtwisted contact manifold @xmath998 with @xmath999 . repeating this process enough times , the positive boundary eventually becomes a rational homology sphere , so that the weak cobordism can be deformed to a strong one .
suppose @xmath29 contains an @xmath46-separating partially planar domain @xmath55 with planar piece @xmath1000 , where @xmath46 is a closed @xmath45-form on @xmath49 satisfying @xmath50 . then for every binding circle or interface torus in @xmath301 , we can attach @xmath305-capping or @xmath305-decoupling cobordisms to produce a symplectic manifold @xmath10 with @xmath229 , @xmath51 and @xmath1001 where @xmath242 contains a component that is a symplectic @xmath344-fibration over @xmath123 , and @xmath215 carries a contact structure @xmath80 with @xmath1002 .
the desired cap is then obtained from @xmath10 after capping @xmath1003 via @xcite or @xcite .
note that since @xmath1004 , any weak cobordism from @xmath29 to @xmath33 that is concave at @xmath49 can be deformed to a strong cobordism , so it suffices to prove that @xmath33 can be obtained from @xmath29 by a finite sequence of ( generally weak ) capping and decoupling cobordisms .
suppose @xmath55 is a partially planar domain .
if it is also a planar torsion domain then the result already follows from corollary [ cor : equivalence ] , thus assume not .
if @xmath200 has only one irreducible subdomain with nonempty binding , we can remove binding components by @xmath305-capping cobordisms and interface tori by @xmath305-decoupling cobordisms until the planar piece has exactly one binding component left and no interface or boundary , which means it is the tight @xmath326 .
the desired cobordism can then be obtained by capping any additional components that may remain at the end of this process .
if @xmath200 has more than one irreducible subdomain but does not have planar torsion , then it must be symmetric ( see definition [ defn : symmetric ] ) .
this means that @xmath1005 , the binding and boundary are empty and the interface tori divide @xmath49 into exactly two irreducible subdomains that have diffeomorphic planar pages .
then we can remove interface tori by @xmath305-decoupling cobordisms until exactly one remains , and the resulting contact manifold is the tight @xmath1006 .
the latter is cobordant to @xmath326 by a @xmath305-capping cobordism that removes one binding component from the supporting open book with cylindrical pages and trivial monodromy .
theorem [ thm : planar ] follows essentially by the same argument since every planar open book is also a fully twisted partially planar domain .
we only need to add that the topological construction of the cobordism by attaching @xmath45-handles along binding components does not depend on the choice of @xmath45-form @xmath46 on @xmath49 , which after the deformation carried out in [ sec : deformation ] , always looks the same on a large tubular neighborhood of the binding . for applications to embedded contact homology and symplectic field theory among other things , it may be quite helpful to observe that the cobordism @xmath10 generally admits not only a symplectic structure but also a foliation by @xmath3-holomorphic curves .
we do nt plan to pursue this here in full detail , but we shall give a sketch of the general picture .
for simplicity , we consider only the @xmath61-decoupling cobordism along @xmath909 in the case where the negative boundary component @xmath29 is concave , so we can arrange @xmath9 near @xmath159 to have the form @xmath1007 as in [ sec : modelcobordism ] , where @xmath544 and @xmath22 are the confoliation @xmath14-form and contact form respectively that were constructed in [ sec : model ] . these have the following convenient properties : together with the obvious fact that @xmath1009 is closed , these properties mean that the pair @xmath1010 is a * stable hamiltonian structure * , to which we associate the co - oriented distribution @xmath1011 and positively transverse vector field @xmath1012 on @xmath49 such that @xmath1013 similarly , writing @xmath911 and recalling the confoliation @xmath14-form @xmath751 defined on @xmath78 by , the pair @xmath1014 form a stable hamiltonian structure on @xmath78 , and we define the corresponding distribution @xmath1015 and vector field @xmath1016 on @xmath78 such that @xmath1017 in fact , with a little more effort one can show that @xmath751 admits a perturbation to a contact form @xmath1018 on @xmath78 such that , without loss of generality , @xmath1019 and @xmath1020 for some smooth function @xmath1021 that satisfies @xmath1022 and is constant outside a neighborhood of the boundary of the co - core , @xmath237 , hence @xmath1016 is colinear with the reeb vector field determined by @xmath1018 .
there is now a collar neighborhood @xmath1023 \times m ' \subset w$ ] of @xmath78 on which @xmath9 takes the form @xmath1024 , thus we can attach positive and negative cylindrical ends to define the * completion * of @xmath10 , @xmath1025 \times m \right ) \cup \left ( [ 0,\infty ) \times m ' \right)\ ] ] and extend @xmath9 symplectically so that it takes the form @xmath1026 on @xmath1027 \times m$ ] and @xmath1028 on @xmath1029 for suitable choices of functions @xmath382 and @xmath502 .
equivalently , @xmath1030 can be constructed directly from the symplectization of @xmath29 as follows : extend the function @xmath1031 to @xmath1032 so that defines a symplectic form on @xmath271 , and extend the `` hole '' @xmath693 defined in
[ sec : modelcobordism ] to a hole in @xmath271 by including the interior of the region @xmath1033 ; let @xmath1034 denote the extended hole . then @xmath1030 can be obtained by removing @xmath1035 from @xmath1036 and replacing it with the completion of the round handle , @xmath1037 extending the symplectic form over @xmath1038 by an adaptation of the argument in [ sec : modelcobordism ] .
an almost complex structure @xmath3 on @xmath1030 is now * admissible * if it is @xmath9-compatible on @xmath6 and is compatible with the stable hamiltonian structures on both cylindrical ends , meaning it is @xmath1032-invariant , restricts to a complex structure on the respective distributions @xmath577 and @xmath756 defining the correct orientations , and maps the unit vector @xmath1039 in the @xmath1032-direction to the vector field @xmath1012 or @xmath1016 .
it was shown in @xcite that for a suitable choice of almost complex structure @xmath1040 on @xmath271 compatible with @xmath1010 , the pages of the blown up summed open book @xmath101 in @xmath200 admit lifts to embedded @xmath1040-holomorphic curves in @xmath271 which match the fibers of the mapping torus @xmath567 outside of the neighborhoods @xmath519 of @xmath487 , have positive cylindrical ends approaching closed orbits of @xmath1012 in @xmath487 and satisfy a suitable finite energy condition .
we can now define an admissible almost complex structure on @xmath1030 which matches @xmath1040 outside of @xmath1035 and is @xmath9-compatible on @xmath1038 .
the @xmath1040-holomorphic curves in @xmath1041 can be extended into @xmath1038 as symplectic surfaces that are diffeomorphic to @xmath61 and foliate @xmath1038 , thus we can extend @xmath1040 into the handle so that it is @xmath9-compatible and these surfaces become @xmath1040-holomorphic . in doing this , we can also make the natural completion of the core @xmath1042 \times \widehat{b}_0)\ ] ] @xmath1040-holomorphic , as well as its translations under the @xmath123-action by translating the local @xmath534-coordinates , and the completion of the co - core @xmath1043 the result is a foliation of @xmath1044 ( or at least the region outside of @xmath1045 ) by finite energy @xmath1040-holomorphic curves .
we summarize this construction as follows ( see figure [ fig : jhol ] ) .
[ prop : jhol ] one can choose an admissible almost complex structure @xmath1040 on the completion @xmath1030 of a @xmath61-decoupling cobordism @xmath10 , such that there exists a foliation @xmath1046 by embedded @xmath1040-holomorphic curves with the following properties : 1 . in each cylindrical end , the leaves of @xmath1046 match the holomorphic lifts of the pages of @xmath101 and @xmath137 constructed in @xcite .
2 . the completed core @xmath1047 and all its @xmath123-translations are leaves of @xmath1046 .
3 . the trivial cylinders over periodic orbits in @xmath88 , @xmath1048 and @xmath209 are all leaves of @xmath1046 .
all other leaves of @xmath1046 have only positive cylindrical ends asymptotic to orbits in @xmath1049 , and they are homotopic in the moduli space to the holomorphic lifts of the pages of @xmath137 in @xmath1050 . 5 . the completed co - core @xmath1051 is also @xmath1040-holomorphic and intersects the leaves of @xmath1046 transversely . the @xmath1040-holomorphic foliation described by proposition [ prop : jhol ] for a @xmath61-decoupling cobordism ,
not including the cylindrical ends .
the picture shows the @xmath1052 and @xmath540 coordinates near various binding , interface and boundary components , including an interface torus @xmath47 where a handle @xmath703 has been attached .
the circles at the ends of leaves in this region represent capping by @xmath61 .
the core @xmath1047 is shown as the vertical leaf directly in the center , which emerges from @xmath1053 and is capped off in the handle.,width=480 ] in considering the behavior of holomorphic curve invariants under symplectic cobordisms , a special role is typically played by curves that have no positive ends such curves can only exist in non - exact cobordisms .
one useful application of the foliation constructed above is that we can now characterize all such curves precisely : [ prop : nopositive ] suppose @xmath1054 is a finite energy @xmath1040-holomorphic curve that is connected , somewhere injective and has no positive ends .
then @xmath1055 is a leaf of @xmath1046 , specifically it is an @xmath123-translation of the core @xmath1047 .
there are no curves without positive ends outside the region of surgery since here the symplectic form is exact , thus we may assume @xmath1055 intersects both the handle and its complement . if @xmath1055 is a leaf of @xmath1046 then it must be an @xmath123-translation of the core , as all other leaves have positive ends .
if it is not a leaf of @xmath1046 then it has a positive intersection with some leaf @xmath1056 , and without loss of generality we may suppose that @xmath1056 has only positive ends
. then @xmath1056 is homotopic in the moduli space to a holomorphic lift of a page of @xmath137 , which we may assume lives in the region @xmath1057 for an arbitrarily large number @xmath165 and thus can not intersect @xmath1055 .
this is a contradiction , due to positivity of intersections . to apply these constructions to ech , or to symplectic field theory for that matter , one must perturb @xmath577 and @xmath756 to contact structures and perturb @xmath1040 with them .
the @xmath1040-holomorphic leaves of @xmath1046 will not generally behave well under this perturbation : a leaf with only positive ends for instance , if it has genus @xmath640 , will have fredholm index @xmath1058 and thus disappears under any generic perturbation of the data unless @xmath1059 .
proposition [ prop : nopositive ] , however , should still hold under a sufficiently small perturbation , because for any sequence @xmath1060 of perturbed almost complex structures converging to @xmath1040 , a sequence of @xmath1060-holomorphic curves should converge in the sense of @xcite to a @xmath1040-holomorphic building , and proposition [ prop : nopositive ] determines what this building can look like .
this is a variation on the uniqueness argument used in @xcite to prove vanishing of the ech contact invariant : higher genus holomorphic curves do not generically exist , but they remain useful for proving that no other curves can exist either . | we show that the pre - order defined on the category of contact manifolds by arbitrary symplectic cobordisms is considerably less rigid than its counterparts for exact or stein cobordisms : in particular , we exhibit large new classes of contact @xmath0-manifolds which are symplectically cobordant to something overtwisted , or to the tight @xmath0-sphere , or which admit symplectic caps containing symplectically embedded spheres with vanishing self - intersection .
these constructions imply new and simplified proofs of several recent results involving fillability , planarity and non - separating contact type embeddings .
the cobordisms are built from symplectic handles of the form @xmath1 and @xmath2 \times s^1 $ ] , which have symplectic cores and can be attached to contact @xmath0-manifolds along sufficiently large neighborhoods of transverse links and pre - lagrangian tori .
we also sketch a construction of @xmath3-holomorphic foliations in these cobordisms and formulate a conjecture regarding maps induced on embedded contact homology with twisted coefficients . |
the dicke model ( dm)@xcite describes an ensemble artificial two - level atoms coupling with a cavity device . it has been attracted considerable attentions recently , mainly due to the fact that the dicke model is closely related to many recent interesting fields in quantum optics and condensed matter physics , such as the superradiant behavior by an ensemble of quantum dots @xcite and bose - einstein condensates @xcite , coupled arrays of optical cavities used to simulate and study the behavior of strongly correlated systems@xcite , and superconducting charge qubits@xcite . it is known from the previous studies@xcite that the full dm undergoes the second - order quantum phase transition @xcite . as claimed in ref .
@xcite , a sequence of instabilities , so - called quantum - phase - like transitions , is involved in the problem of an ensemble of two - level atoms system interacting with a bosonic field in the rotating - wave approximation ( rwa)@xcite . as addressed ref .
@xcite , the absence of field @xmath1 term from the minimal coupling hamiltonian in the approximation of the dm leads to the possibility of the instability . in the presence of @xmath1 term ,
the classical thermodynamic properties have been studied previously @xcite .
whether the instabilities disappear when the interaction of the bosonic field @xmath1 term is taken into account is a long - standing issue and remains very controversial to date@xcite .
it is known that quantum critical phenomena exhibits deep relations to the berry phase ( bp ) @xcite and the fidelity @xcite .
the bp has been extensively studied by the geometric time evolution of a quantum system , providing means to detect the quantum effects and critical behavior , such as quantum jumps and collapse @xcite .
a recent proposal is to use the fidelity in identifying the quantum phase transition@xcite . as a consequence of the dramatic changes in the structure of the ground states , the fidelity should drop at critical points . in our opinion
, the quantum - phase - like transitions might also be studied in terms of the bp and the gs fidelity . in this paper
, we calculate the ground state bp and fidelity in the rwa dm with and without @xmath1 term to quantify phenomena of the quantum - phase - like transitions in finite system . without @xmath1 , a exact solution to the dm
is given explicitly .
we solve the rwa dm with an additional @xmath1 term by a exact diagonalization in the fock space of the bosonic operators .
the paper is organized as follows . in sec.ii
, we review the rwa dm and the hamiltonian with the @xmath1 term to obtain exact solutions respectively . in sec.iii , we study the instability of the rwa dm by measuring the bp and the ground state fidelity
. the behaviors of these two quantities as a function of the interaction strength of the field for the rwa dm with @xmath1 term are also evaluated .
finally , we present the conclusion in sec.iv .
let us consider dm of @xmath0 two - level atoms with energy level @xmath2 , interacting with a single - mode bosonic field with the frequency @xmath3 . in the rwa dm ,
ignoring the counter - rotating term , the corresponding hamiltonian is given by @xmath4 where @xmath5 and @xmath6 are the photonic creation and annihilation operators , @xmath7 denotes the collective spin-@xmath8 atomic operators , @xmath9 is the atom - field coupling strength , and @xmath10is set unity
. motivated by the exact technique of the jaynes - cammings model with rwa , we present a detailed numerical diagonalization procedures to solve a set of closed equations to the dm with rwa , which was also briefly discussed in ref . @xcite . since the hamiltonian ( [ h2 ] ) commutes with the total excitation number operator @xmath11 , the subspace of the hilbert space consists of a sum of subspaces labeled by different number of excitations @xmath12 .
the hilbert space of the collective algebra is spanned by the kets @xmath13 . by adapting schwinger s representation of spin in terms of harmonic oscillators
@xcite , @xmath14 can be expressed as @xmath15 , which is a dike state of @xmath16 spin - up atoms and @xmath17 spin - down atoms , @xmath18 . in this work , @xmath19 takes its maximal value @xmath20 .
@xmath21 is also known as the eigenstates of @xmath22 and @xmath23 with @xmath24 .
the action of the corresponding raising and lowering operators on this state gives @xmath25 @xmath26 . in the subspace of @xmath27 excitations the wave function is supposed as @xmath28 where @xmath29 s are coefficients , @xmath30 is a fock state of the bosonic field with an alterative number @xmath31 , ranging from @xmath32 to infinity .
@xmath31 is equal to @xmath32 in the weak coupling regions corresponding to @xmath33 excitations and then increases with the coupling parameter @xmath34 .
when the number of excitations @xmath12 is larger than the number of atoms @xmath0 , i.e. @xmath35 , the ground state of @xmath36 lies in the subspace spanned by @xmath37 vectors . in this way , the hamiltonian is expressed by a tridiagonal @xmath38 matrix . from eq .
( [ wave function ] ) we obtain the exact expression of the @xmath39-th row of the schrdinger equation @xmath40c_m\end{aligned}\ ] ] where @xmath41 note that the above equations are closed and the set of linear equations for @xmath42s takes a tridiagonal form .
solutions for a given @xmath31 are readily obtained through gaussian elimination and back substitution
. finally the chosen @xmath31 corresponds to the lowest energy among eigenvalues of the solutions for a fixed coupling strength @xmath34 .
it is interesting to find that the excitation number @xmath12 is added step by step and keeps a constant in a coupling parameter interval @xmath43 $ ] , where @xmath44 is a quantum - phase - like transition point , as shown in fig .
[ exitation ] .
the first transition point is denoted as @xmath45 .
the sensitive quantities like the ground state bp and the fidelity will be calculates to quantify the discontinuities , so called instability , in the finite dm with rwa .
versus atom - field coupling parameter @xmath34 for different numbers of atoms @xmath46.,width=302 ] it is interesting to discuss the effect of the interacting bosonic field in the atom - field system . as the interactions vector potential @xmath48 , caused by the longitudinal part of the bosonic field ,
are taken into account , the hamiltonian in the rwa dm can be evaluated with an additional term @xmath1 @xcite.the additional term @xmath49 has been discussed classically about thermodynamic properties by rzaewski et al @xcite . to extensively quantify the contributions of the @xmath1 term , we employ the quantum information tools such as the bp and the ground state fidelity to detect the quantum - phase - like transitions . in terms of the bosonic operators ,
the @xmath1 term is given by @xmath50 , where @xmath51 is the interacting strength of the bosonic field .
the overall hamiltonian of the ensemble of two - level atoms interacting with the bosonic field is expressed as @xmath52+\varepsilon ( a^{\dagger}+a)^{2}\ ] ] in order to obtain the numerical exact solution , we perform a standard bogoliubov transformation by introducing bosonic annihilation ( creation ) operator @xmath53 , such that @xmath54 and @xmath55 . after substituting @xmath6 , @xmath5 into eq .
( [ a^2 ] ) the total hamiltonian is diagonalized as @xmath56\end{aligned}\ ] ] where @xmath57 note that a counter - rotating term is included in the modified hamiltonian ( [ ha ] ) , which may plays an essential role in the following discussion .
because the @xmath1 term breaks the gauge invariance of the hamiltonian ( [ h2 ] ) in the dm with rwa , it was argued in ref .
@xcite the instability would be then eliminated .
we now consider the wave functions of the total hamiltonian with @xmath0 atoms , which are of the form @xmath58 is the maximum photonic number in the artificially truncated fock space , and @xmath59 are coefficients .
@xmath60 is a fock state with @xmath39 photons .
@xmath61 is a dicke state in schwinger s representation of spin with @xmath17 atoms in excited state .
the @xmath39-th row of the schrdinger equation reads @xmath62d_{nm}\nonumber \\ & + & \frac{\lambda\mu}{\sqrt{n}}\sqrt{(m+1)(n - n+1)n}d_{n-1,m+1}\nonumber \\
& + & \frac{\lambda\mu}{\sqrt{n}}\sqrt{m(n+1)(n - n)}d_{n+1,m-1}\nonumber \\ & -&\frac{\lambda\nu}{\sqrt{n}}\sqrt{(m+1)(n+1)(n - n)}d_{n+1,m+1}\nonumber \\ & -&\frac{\lambda\nu}{\sqrt{n}}\sqrt{m(n - n+1)n}d_{n-1,m-1}\end{aligned}\ ] ] the eigenvalues and eigenfunctions can be obtained numerically by diagonalizing a @xmath63 matrix .
the bp and the fidelity can be calculated through these eigenfunctions . to be complete
, we also briefly review the contribution of the @xmath1 term in the dm model without rwa , which yields some unimportant corrections .
the hamiltonian of the full dm with the @xmath1 term reads @xmath64 with a rotation around an @xmath65 axis by an angle @xmath66 and the same bogoliubov transformation , the modified hamiltonian @xmath67 is rewritten as @xmath68 therefore , @xmath1 term in the dm model without rwa would not change the nature of the phase transition , except that the position of the critical point is shifted .
of the rwa dm as a function of the coupling parameter @xmath69 for different numbers of atoms @xmath70 .
the inset shows a discontinuous picture of instability for @xmath71 atoms .
, width=302 ] berry s adiabatic geometric phase describes a phase factor of the wave functions in a time - dependent quantum system .
the interesting paths of evolution for generating a bp are those for which the ground state of the system can evolve around a region of criticality .
we first measure a nontrivial bp circulating a region of `` criticality '' corresponding to a abrupt change .
the bp @xmath72 generated after the system undergoing the time - dependent unitary transformation @xmath73 $ ] , varying the angle @xmath74 adiabatically from @xmath33 to @xmath75 , can be evaluated as a function of the atom - field coupling parameter @xmath34 @xmath76 where @xmath77 is the ground - state wave function of hamiltonian ( [ h2 ] ) of the rwa dm . as shown in fig .
[ berry phase ] , the first phase - transition - like occurs at the `` critical '' value of the coupling parameter @xmath78 for arbitrary atom number @xmath0 , which recover the result in the thermodynamical limit @xcite .
the average bp @xmath79 is equal to @xmath80 at @xmath81 and increases abruptly at discontinuous `` critical '' points when @xmath82 .
note that the plateau is formed clearly for @xmath83 , the width of the plateau becomes narrower and narrower with the increasing @xmath0 , which are quite different from the phenomenon of the quantum phase transition in the full dm @xcite . a clear picture of the instability in the ground state of the rwa dm is given in terms of the bp with @xmath84 atoms shown in the inset of fig .
[ berry phase ] .
the effect of decoherence of the driving field on adiabatic evolutions of spin and quantized modes system has been investigated @xcite . in the fully quantized context
we need a procedure capable of generating an analogous phase change in the state of the field .
the bp @xmath85 is obtained in terms of the bosonic operator by the phase shift unitary operator @xmath86 $ ] , where @xmath87 is the number of bosons in the field .
changing the angle @xmath74 slowly from @xmath33 to @xmath75 the ground state @xmath85 is given by @xmath88 we now have a general expression for the bp @xmath85 related to the photonic number , which is driven by fields .
we plot behaviors of @xmath89 in units of @xmath75 as a function of the atom - field coupling parameter @xmath34 for different number of atoms @xmath0 in fig .
[ phonon ] .
the bp @xmath89 is @xmath33 in the weak coupling region for @xmath90 and then increases discontinuously as @xmath34 increases . as shown in fig .
[ phonon ] , when @xmath0 increases the interval of the
critical " values @xmath34 become smaller , leading to the curve with more steps .
the inset of fig .
[ phonon ] shows that there actually exist many phase - transition - like critical " points beyond @xmath91 for large @xmath84 . .i.e .
the average phonon number , in units of @xmath75 of the rwa dm as a function of the coupling parameter @xmath34 for different number of atoms @xmath0 .
the inset gives a discontinuous picture of instability for @xmath71 atoms .
[ phonon],width=302 ] an increasing interest has been drawn in the role of the ground state fidelity in detecting the quantum phase transitions for various many - body system , with a narrow drop at the transition point .
below we propose to use this quantum tool to identify the quantum - phase - like transitions , where the gs fidelity drops to @xmath33 in the rwa dm .
it is defined as the overlap between two ground states @xmath92 and @xmath93 @xcite , where @xmath94 is a tiny perturbation parameter , i.e. @xmath95 note that @xmath96 is a function of both @xmath34 and @xmath94 .
based on the normalized and orthogonalized wave function in eq.([wave function ] ) for the rwa dm , the ground state fidelity can be derived analytically @xmath97 , and then can be simplified as @xmath98 at each transition point in rwa dm , the alternative number @xmath31 is changed abruptly and @xmath96 is then equal to @xmath33 . beyond `` critical '' points ,
@xmath96 should be constant @xmath99 .
the energy gap @xmath100 between the first excited and ground state energies tends to @xmath33 at each `` critical '' point , as shown in fig .
we attribute this level crossing to the fact that the gs wave functions at the different sides of each transition point are orthogonal .
the ground state fidelity drops to @xmath33 at each `` critical '' point , exhibiting similar critical singularity of the first - order quantum phase transition @xcite .
it is also obvious that the instabilities increase with a increasing number of atoms . and energy gap @xmath100 between the energies of the first excited and ground state in the rwa dm as a function of the coupling parameter of @xmath101 for different number of atoms @xmath102.,width=302 ] the average berry phase ( a ) @xmath103 , ( b ) @xmath104 in units of @xmath75 , as well as ( c ) ground state fidelity @xmath105 of the rwa dm with @xmath1 term versus @xmath34 for different interacting strengthes of the field @xmath106 for @xmath107 . , title="fig:",width=264 ] the average berry phase ( a ) @xmath103 , ( b ) @xmath104 in units of @xmath75 , as well as ( c ) ground state fidelity @xmath105 of the rwa dm with @xmath1 term versus @xmath34 for different interacting strengthes of the field @xmath106 for @xmath107 .
, title="fig:",width=264 ] the average berry phase ( a ) @xmath103 , ( b ) @xmath104 in units of @xmath75 , as well as ( c ) ground state fidelity @xmath105 of the rwa dm with @xmath1 term versus @xmath34 for different interacting strengthes of the field @xmath106 for @xmath107 .
, title="fig:",width=264 ] ground state fidelity @xmath105 ( a ) , berry phase @xmath103 ( b ) and its first derivative @xmath108 ( c ) of the rwa dm with @xmath1 term as a function of the coupling of @xmath109 for different interacting strengthes of the field @xmath110 . , width=302 ] we next turn to study the rwa dm with @xmath1 term by using the above quantum information tools . by manse of the general bp formula eqs .
( [ bp1 ] ) and ( [ bp2 ] ) , the ground state bp of the rwa dm with @xmath1 term can be evaluated as a function of the atom - field coupling @xmath34 and the interaction strength of the field @xmath51 , i. e. , @xmath111 where @xmath112 is the ground state of the hamiltonian ( [ a^2 ] ) .
we plot the average bp @xmath103 in fig .
[ n4](a ) and @xmath104 in units of @xmath75 in fig .
[ n4](b ) as a function of the coupling @xmath34 for different interaction strengthes @xmath106 for @xmath107 atoms .
we can observe that , for both @xmath103 and @xmath104 , the abrupt jump occurs at the same coupling parameter @xmath34 for the same interaction strengthes @xmath51 .
we denote the coupling parameter @xmath34 where the first abrupt jump occurs as a characteristic parameter @xmath113 .
it is interesting to observe that the characteristic @xmath113 increases with the interaction strength @xmath51 .
the stability claimed in ref .
@xcite is only found in the smooth curves for @xmath114 .
thus , there still exist quantum - phase - like transitions when the interaction of the bosonic field @xmath51 is strong . according to the ground state wave function @xmath112 ,
the fidelity of the rwa dm with @xmath1 term can be also calculated @xmath115 .
the numerical results for the different @xmath51 with @xmath107 atoms are exhibited in fig .
[ n4](c ) .
the singularities at the `` critical '' points for @xmath106 are demonstrated by a sudden drop of @xmath105 .
it is clearly shown that the characteristic @xmath113 moves towards the right regime with the increasing @xmath51 , providing the evidence of the quantum - phase - like transitions even for a strong interaction of the field . to show the instabilities more obviously
, for a more wide range of interacting strengthes @xmath110 , we plot the ground state fidelity @xmath105 , berry phase @xmath103 and its first derivative @xmath116 of the rwa dm with @xmath1 term as a function of the scaled coupling parameter @xmath109 together in fig .
[ n4fgap ] .
when the interaction strength of the field increases the fidelity @xmath96 still drops to @xmath33 at the characteristic @xmath113 , where the first derivative @xmath117 also changes abruptly .
this is another piece of evidence that the instability of the rwa dm does not vanish in the rwa dm including the @xmath1 term .
it is illustrated that the contribution of the @xmath1 term does not eliminates the instability of the rwa dm , contrarily to the previous studies by rzaewski et al.@xcite .
for strong interaction strength of the bosonic field @xmath51 , a characteristic parameter @xmath113 becomes larger than @xmath78 in the absence of @xmath1 term .
a sequence of the ground state stability reported previously only appears for @xmath114 .
in summary , we have investigated the instability of the rwa dm by quantum information tools such as the bp and the ground state fidelity .
an obvious discontinuous behaviors of these quantities with finite @xmath0 atoms are observed .
it is demonstrated that the quantum - phase - like transitions occur beyond the characteristic @xmath113 for strong interaction of fields .
we propose that the instability would not be eliminated by involving the @xmath1 term of the dm with rwa .
previous observed instability may be limited to the coupling regime @xmath114 .
it should be pointed out that the quantum information tools are very sensitive quantities to detect quantum phase ( or like ) transition .
this work was supported by national natural science foundation of china , pcsirt ( grant no .
irt0754 ) in university in china , national basic research program of china ( grant no .
2009cb929104 ) , zhejiang provincial natural science foundation under grant no .
z7080203 , and program for innovative research team in zhejiang normal university .
r. h. dicke , phys . rev . * 93 * , 99(1954 ) .
m. scheibner et al .
, nature phys . * 3 * , 106(2007 ) .
d. schneble et al . , science * 300 * , 475 ( 2003 ) .
m. j. hartmann et al . , nature phys . * 2 * , 849(2006 ) ; a. d. greentree et al .
* 2 * , 856(2006 ) .
j. q. you , and f.nori , phys . today .
* 58(11 ) * , 42(2005 ) .
a. m. zagoskin et al . , phys .
lett . * 97 * , 077001(2006 ) . k. hepp and e. lieb , ann .
phys . , * 76 * , 360(1973 ) . c. emary and t. brandes , phys .
e * 67 * , 066203(2003 ) ; phys .
lett . * 90 * , 044101(2003 ) . q. h. chen , y. y. zhang , t. liu , and k. l. wang , phys .
a. * 78 * , 051801(r)(2008 ) .
s. sachdev , quantum phase transitions ( cambridge university press , cambridge , england , 2000 ) .
l. m. narducci et al . , collective phenomena 1,113(1973 ) ; phys .
a. * 8 * , 1892(1973 ) . v. buek , m. orszag , and m. roko , phys . rev . lett . * 94 * , 163601(2005 ) . b. v. thompson , j. phys .
a. * 10 * , 89(1977 ) .
b. v. thompson , j. phys .
a. * 10 * , l179(1977 ) .
k. rzaewski , and k. wdkiewicz , phys .
lett . * 96 * , 089301 ( 2006 ) .
k. rzaewski , and k. wdkiewicz , phys .
rev . a. * 43 * , 593 ( 1991 ) .
k. rzaewski , k. wdkiewicz , and w. zakowicz , phys .
lett . * 35 * , 432(1975 ) .
k. rzaewski , and k. wdkiewicz , phys .
a. * 13 * , 1967(1976 ) .
m. v. berry , proc .
london , ser .
a , 392,45 ( 1984 ) .
i. f. guridi , a. carollo , s. bose , and v. vedral , phys .
* 89 * , 220404(2002 ) .
a. carollo , i. f. guridi , m. f. santos , and v. vedral , phys .
* 92 * , 020402(2004 ) .
f. plastina , g. liberti , and a. carollo , europhys .
. h. t. quan , z. song , x. f. liu , p. zanardi , and c. p. sun , phys .
* 96 * , 140604 ( 2006 ) . w. l. you , y. w. li , and s. j. gu , phys .
b. * 76 * , 022101 ( 2007 ) ; s. j. gu , h. m. kwok , w. q. ning , and h. q. lin , ibid . * 77 * , 245109(2007 ) .
m. cozzini , r. ionicioiu , and p. zanardi , phys .
b * 76 * , 104420 ( 2007 ) .
m. f. yang , phys .
b. * 76 * , 180403 ( 2007 ) .
s. chen , l. wang , s. j. gu , and y. wang , phys .
e. * 76 * , 061108 ( 2007 ) .
s. chen , l. wang , y. j. hao , and y. p. wang , phys .
a * 77 * , 032111 ( 2008 ) . h.
q. zhou , r. orus , g. vidal , phys .
100 * , 080601 ( 2008 ) .
t. liu , y. y. zhang , q. h. chen , and k. l. wang , arxiv : 0812.0321 .
s. dusuel , and j. vidal , phys .
lett . * 93 * , 237204(2004 ) .
a. c. m. carollo , and j. k. pachos , phys .
lett . * 95 * , 157203(2005 ) . s. j. zhu , phys . rev .
* 96 * , 077206(2006 ) .
j. schwinger , l. c. biedenharn , and h. vandam , in quantum theory of angular momentum , new york : academic , pp .
229 - 279 ( 1965 ) .
j. j. sakurai , modern quantum mechanics , reading , ma : addison - wesley , ( 1994 ) .
g. chen , j. q. li , and j. q. liang , phys .
a. * 74 * , 054101(2006 ) . | the instability , so - called the quantum - phase - like transition , in the dicke model with a rotating - wave approximation for finite @xmath0 atoms is investigated in terms of the berry phase and the fidelity .
it can be marked by the discontinuous behavior of these quantities as a function of the atom - field coupling parameter .
involving an additional field @xmath1 term , it is observed that the instability is not eliminated beyond the characteristic atom - field coupling parameter even for strong interaction of the bosonic fields , contrarily to the previous studies . |
astronomical masers are variable phenomenon seen in a high mass star - forming region .
interstellar h@xmath0o masers ( @xmath11 = @xmath12@xmath13 ) at 22.23508@xmath14ghz , which appear at the earliest phase of formation , are highly variable in particular .
they are often excited in strong shocks on a working surface between protostellar outflow and dense envelope ( e.g. , @xcite ) .
outflow properties such as size , velocity and morphology , predicted from interferometric observations of h@xmath0o masers vary source to source ( e.g. , collimated jet @xcite , wide angle flow @xcite , expanding shell @xcite , equatorial flow @xcite or combination of them @xcite ) .
it is possibly related to a number of factors such as stellar mass , evolutional stage , geometry of surrounding envelope and driving mechanism of host outflow .
several single - dish based studies have reported sub - yr scale maser flares in some cases ( e.g. , @xcite ) .
this time - sale is comparable with or even shorter than that of episodic radio jet , which traces a root of protostellar outflow ( e.g. , @xcite and reference therein ) , and indicates that some portion of h@xmath0o masers contains valuable information about an innermost outflow .
exact relation between an outflow activity and maser variability is still unclear , but once it is set up , h@xmath0o maser will be an excellent tool to survey dynamic outflow activities in small scale , since its bright emission allows us easy and frequent * monitoring even with * a small size radio telescope . in this paper
, we report on the new vlbi and single - dish study for intermittent flare activities of h@xmath0o masers in g353.273 + 0.641 ( hereafter g353 ) .
g353 is a strong 22 ghz h@xmath0o maser site in the southern high mass star - forming region ngc6357 @xcite .
the source distance is 1.7 kpc from the sun @xcite .
multi - epoch atca observation has been reported in @xcite ( hereafter cp08 ) .
class ii ch@xmath15oh maser emission ( @xmath16 = 5@xmath176@xmath18 a@xmath19 ) at 6.668519 ghz is also associated with this source ( cp08 ) , and hence , the host young stellar object ( yso ) is identified as a high mass yso ( e.g. , @xcite ) .
cp08 has also suggested that g353 is still in the pre - ultra compact ( uc ) h*ii * region phase , i.e. , high mass protostellar phase , based on the absence of any detectable oh masers ( e.g. , @xcite ) .
g353 has been classified as dominant blue - shifted h@xmath0o maser in cp08 .
that is , almost all flux is concentrated on blue - shifted emission , despite of very broad velocity range of @xmath6 100 km s@xmath4 with respect to the systemic velocity of -5 km s@xmath4 .
they have argued that this type of masers can be caused by well - collimated jet aligned close to the line of sight .
there are a few maser sources which show similar blue - shift dominance ( e.g. , @xcite ; cp08 ; @xcite ) .
some of them suggest acceleration of outflowing materials in the jet .
the statistical analysis in @xcite indicates that such a blue - shift dominance is a characteristic of h@xmath0o masers at the earliest evolutionary stage of star - formation .
although the signature of yr - scale variability of the h@xmath0o maser has been pointed out by cp08 , their small dataset is still insufficient for detailed discussion .
our vlbi and single - dish monitoring with vera ( vlbi exploration of radio astrometry ) and the hokkaido university tomakomai 11-m radio telescope in japan , which has been started from november 2008 , has provided more tight - interval dataset and detected intermittent flare activities accompanied by spatial changes of the shock front traced by maser distribution .
cccccccc epoch&date & synthesised beam & pa & 1-@xmath20 noise@xmath21 & dn@xmath22@xmath23 / dn@xmath24 & positional error & comments ' '' '' + & & ( mas @xmath1 mas ) & ( @xmath25 ) & ( jy beam@xmath4 ) & ( @xmath20 ) & ( mas @xmath1 mas ) & for @xmath26 + 1&2008/7/1 & 3.16 @xmath1 0.74 & -17.1 & 0.48 & 13 / 117 & 0.15 @xmath1 0.49 &
+ 2&2008/11/18 & 3.38 @xmath1 0.78 & -21.7 & 0.30 & 6 /49 & 0.92 @xmath1 2.15 & @xmath5 1000 k at og + 3&2009/2/6 & 2.63 @xmath1 0.91 & -14.1 & 0.37 & 14 / 351 & 0.06 @xmath1 0.24 &
+ 4&2009/5/18 & 2.57 @xmath1 0.88 & -13.2 & 0.30 & 16 / 1122 & 0.04 @xmath1 0.16 & + 5&2009/9/16 & 2.86 @xmath1 0.82 & -12.3 & 0.37 & 10 / 19 & 0.11 @xmath1 0.48 &
+ 6&2009/11/5 & 2.79 @xmath1 0.79 & -13.3 & 0.16 & 15 / 338 & 0.23 @xmath1 0.98 &
+ 7&2010/1/21 & 3.15 @xmath1 0.85 & -19.5 & 0.21 & 15 / 72 & 0.09 @xmath1 0.25 & @xmath5 2000 k at og + 8&2010/4/21 & 2.59 @xmath1 0.75 & -14.1 & 0.40 & 12 / 184 & 0.10 @xmath1 0.41 & @xmath5 1500 k at is + + + +
vlbi monitoring of h@xmath0o masers with vera @xcite has been performed eight epochs from november 2008 to april 2010 and is still on - going . each observation was made in vera s dual - beam mode in which targeted maser source and phase calibrator were observed simultaneously ( @xcite ) .
left - handed circular polarised signals were quantised at 2-bit sampling and filtered with the vera digital filter unit @xcite , after that , data were recorded onto magnetic tapes at a data rate of 1024 mbps .
the data correlation was performed with the mitaka fx correlator @xcite .
total bandwidths assigned for calibrators and the maser are 240 mhz ( 16 ifs ) and 8 mhz ( 1 if ) , respectively
. whole velocity coverage of 108 km s@xmath4 were divided into 512 spectral channels yielding final spectral resolution of 0.21 km s@xmath4 .
special calibration methods for vera s dual - beam instrument have already been described in @xcite .
we used j1717 - 3342 @xcite as a phase calibrator .
this source is separated from g353 by 1.83 at a position angle of -108@xmath14 east of north .
j1717 - 3342 has been enough bright ( @xmath5 0.7 jy ) and shown no significant structure during two years .
we also observed nrao530 ( = j1733 - 1304 : @xcite ) as a delay and bandpass calibrator every 120 minutes .
we have not observed any flux calibrator and antenna gains have been corrected based on measured system noise temperatures ( @xmath27 ) of each station .
typical @xmath27 values were @xmath5 350 k for mizusawa and iriki stations and @xmath5 500 k for ogasawara and ishigaki stations in the case of normal weather condition .
total on - source time is about 2.5 hour in the first five epochs and 5 hour in the last three epochs .
we note that the observations were time - shared with another maser
calibrator pair , g5.89 - 0.39 j1755 - 2232 , in the first five epochs @xcite .
this causes relatively short on - source time in the former five epochs , but total hour angle range ( i.e. , achieved @xmath28 coverage ) is almost same as the latter three epochs .
data reduction was carried out using nrao astronomical imaging processing system ( aips ) package .
we adopted standard reduction procedure for vera data ( see @xcite in details ) , where we have first determined absolute coordinates of several maser spots in 2-beam phase referencing analysis , and then , performed self - calibration using the brightest maser spot in each epoch .
we finally searched all maser spots within [email protected] area with 7-@xmath20 detection limit . here , the term maser spot indicates a maser component seen in a single velocity channel , and hereafter , the term maser feature means physical gas clump which is consist of several maser spots detected in successive velocity channels and closely located each other .
table 1 summarised all the vera observations .
it contains observing dates , synthesised beam sizes in milli - arcsecond ( mas ) , beam position angles ( pa ) east of north , typical noise levels and dynamic ranges of synthesised images for the brightest maser spot , estimated errors of absolute positions ( see section 3.1 ) and brief comments , if any , about @xmath27 value .
single - dish monitoring has been done using the hokkaido university tomakomai 11-m radio telescope @xcite .
observing interval is about 1 3 times per month except for the receiver maintenance session during july to october 2009 .
the beam width ( fwhm ) and aperture efficiency is 4.2 and 49 % at 22 ghz , respectively .
these values have been slightly improved compared with that mentioned in @xcite because of the upgrade in optics .
the pointing accuracy is better than 30@xmath14 in each of azimuth and elevation direction .
observations were performed in the position switching method .
we received single left circular polarisation with 16-mhz bandwidth ( 216 km s@xmath4 in velocity ) and it was divided into 2048 spectral channels .
an intensity scale was calibrated by a chopper wheel method every 10 minutes .
observed data were reduced with the newstar software package developed in nobeyama radio observatory .
initial 2048 spectral channels were binned up every two channels , in order to obtain the same spectral resolution as vera data .
typical rms noise level is @xmath5 3 jy in winter and increases up to 6 25 jy in the summer season ,
+ figure 1 shows the spatial and velocity distribution of h@xmath0o maser spots in g353 .
all detected maser spots in eight epochs are exhibited in this figure , since several maser features have significant velocity width and spatial elongation .
detailed parameters of identified maser features are listed in table 2 .
the @xmath29 , @xmath30 axes and colour show the right ascension offset ( @xmath31@xmath1cos@xmath32 ) , declination offset ( @xmath33 ) and @xmath7 , respectively .
the coordinate origin is ( @xmath34,@xmath32)@xmath35 = ( @xmath36 , -341514.905 ) , which is the absolute position of the brightest maser spot in the second * vlbi * epoch .
this position is consistent with cp08 position within the range of errors .
estimated errors of our position measurements are listed in table 1 .
we have evaluated three error sources , i.e. , thermal noise in the phase - referenced images , errors of baseline vectors and atmospheric phase fluctuations in the same way as @xcite .
the atmospheric zenith delay residuals ( @xmath5 0.7 cm on average ) are determined by the image optimizing method ( @xcite , 2008 ) .
the largest atmospheric residuals ( @xmath5 3 cm ) was found in the second vlbi epoch , where estimated error reached 1 mas in @xmath29 and 2 mas in @xmath30 directions .
the relative position uncertainty between each maser spot is , on the other hand , typically 30 @xmath37as in @xmath29 and 100 @xmath37as in @xmath30 directions , based on thermal noise in the self - calibrated images .
all maser spots have been found within 200 mas @xmath1 100 mas area .
this corresponds to 350 au @xmath1 200 au at the source distance of 1.7 kpc .
we have detected successive maser alignment , which looks like a filament , in the central 50 mas @xmath1 50 mas region .
this filamentary structure is pronounced only in the flare - up phase tracing shock front ( see next subsection ) .
total @xmath7 coverage of maser emissions in vera observations is -121 to -5 km s@xmath4 , and almost same as the single - dish data .
this is well consistent with the blue - shifted part of cp08 spectrum .
the red end of -5 km s@xmath4 is same as the systemic velocity .
extremely red ( + 87 km s@xmath4 ) component reported in cp08 is out of our observing band .
large portion of flux is concentrated in the range of -80 to -40 km s@xmath4 .
spatially , there is no clear gradient of @xmath7 for features in this dominant velocity range .
the features out of this range roughly split into east ( blue ) and west ( red ) part of g353 .
the most blue and red components are separated only 100 mas ( 200 au ) along se nw direction .
the wide range of velocity within a very compact region readily dismisses the possibility of disc maser which has been proposed in a few cases ( e.g. , @xcite ) .
it also indicates relatively pole - on geometry of the disc - jet system as suggested in cp08 .
expected young age of the driving source may also be related to such a concentrated distribution , however , observed proper motions ( see next section ) comparable to the @xmath7 values still support moderately pole - on geometry even in this case .
we have detected strong maser flare in the second vlbi epoch .
time variation of single - dish spectrum after the flare is displayed in figure 2 .
there are two velocity components which show significant brightening .
one is the component of -53 @xmath6 7 km s@xmath4 ( hereafter c50 ) and another is the component of -73 @xmath6 7 km s@xmath4 ( hereafter c70 ) .
almost all emissions disappeared in the quiet phase .
figure 3 presents the light curves for both flared components during 950 days from jan 1st , 2008 .
the vertical axis indicates peak flux density for single - dish data , and sum of the flux densities of maser spots detected in the peak velocity channel for vlbi data .
downward arrows show 3-@xmath20 upper limit for c70 .
the plotted data correspond to column 5 of table 3 .
only c50 had been flared at first and c70 started to increase its flux about 100 days after observed c50 peak .
sign of second flare is detected only for c50 after the 8th vlbi epoch ( day 841 ) .
the systematic increase of the 3-@xmath20 upper limits in the latter part of c70 is not a signature of second flare , but because of bad weather condition in summer season .
a variation pattern of h@xmath0o maser is , in general , combination of long - term , global pattern and short - term fluctuation ( e.g. , @xcite ) . smoothly varying
light curve indicates that global variation pattern is dominant in c50 rather than random fluctuation during our monitoring .
although large dip was caught in day 473 for c70 , same global pattern was identifiable , i.e. , the flare - up continued for @xmath5 150 days and rapid decaying .
the dip in c70 is roughly factor of three weaker than the peak flux density and explainable by short - term fluctuation .
figure 4 presents time variation of the prominent part of the maser alignment that contains most flux of c50 .
this successive alignment had been appeared after the first flare .
the maximum spatial scale of the alignment is roughly 50 mas ( @xmath5 90 au ) that is almost typical value for an individual maser filament in an outflow shock ( e.g. , @xcite ; @xcite ) .
a number written in upper right part of each panel shows relative day from the 1st vlbi epoch ( day 183 ) .
the alignment was initially vertical and dramatically changed to be horizontal during the flare phase ( 2nd 4th epoch ) .
all maser features in this region completely disappeared in the 7th vlbi epoch .
another maser alignment appeared in the final vlbi epoch , tracing the second flare .
because of non - negligible accelerations of @xmath7 , the identification of individual maser feature in these alignments is difficult ( see section 3.3 ) .
we , thus , estimated averaged proper motions of these alignment , where we averaged the positions of all maser spots in each alignment and calculated @xmath38 and @xmath39 for vertical and horizontal alignments , respectively . here ,
@xmath38 and @xmath39 show the linear proper motion in each directions .
such a motion perpendicular to a maser alignment should trace bulk motion of shock front , if we consider the standard model that h@xmath0o masers are excited in a thin sheet of dense gas compressed by outflow shock @xcite .
estimated @xmath38 and @xmath39 are 7.8 mas yr@xmath4 from east to west and 7.6 mas yr@xmath4 from north to south , respectively .
these averaged proper motions correspond to @xmath5 60 km s@xmath4 at 1.7 kpc and comparable with their @xmath7 values of -50 -70 km s@xmath4 . [ cols="^,^,^,^,^,^ " , ] ccccccepoch & @xmath7 & & total width@xmath23 & integrated flux + & ( km s@xmath4 ) & @xmath29 & @xmath30 & ( km s@xmath4 ) & ( jy km s@xmath4 ) + & -44.56 & 0.10 ( 0.01 ) & -8.42 ( 0.02 ) & 0.63 & 0.73 ( 0.02 ) + & -48.56 & -2.83 ( 0.01 ) & -3.57 ( 0.03 ) & 1.48 & 2.76 ( 0.08 ) + & -48.77 & -3.86 ( 0.02 ) & 27.88 ( 0.04 ) & 1.47 & 0.81 ( 0.05 ) + & -49.61 & -0.01 ( 0.03 ) & -8.01 ( 0.05 ) & 0.84 & 0.77 ( 0.03 ) + & -52.77 & 10.05 ( 0.01 ) & -12.93 ( 0.02 ) & 1.90 & 2.33 ( 0.07 ) + & -51.72 & 13.01 ( 0.02 ) & -13.39 ( 0.05 ) & 1.05 & 0.34 ( 0.02 ) + & -53.83 & 2.62 ( 0.01 ) & -9.81 ( 0.03 ) & 1.69 & 2.06 ( 0.08 ) + & -53.2 & 9.93 ( 0.01 ) & -12.67 ( 0.01 ) & 1.05 & 1.36 ( 0.03 ) + & -55.09 & 5.47 ( 0.01 ) & -11.64 ( 0.02 ) & 1.05 & 1.32 ( 0.04 ) + & -54.88 & 7.36 ( 0.03 ) & -12.12 ( 0.03 ) & 1.26 & 0.94 ( 0.06 ) + & -56.14 & 7.10 ( 0.02 ) & -11.70 ( 0.02 ) & 1.05 & 0.35 ( 0.02 ) + & -56.14 & 8.04 ( 0.03 ) & -11.61 ( 0.03 ) & 1.69 & 0.46 ( 0.03 ) + & -65.62 & 107.80 ( 0.02 ) & 47.85 ( 0.05 ) & 0.63 & 0.05 ( 0.00 ) + & -69.84 & 33.25 ( 0.02 ) & 40.05 ( 0.05 ) & 0.63 & 0.05 ( 0.00 ) + & -71.94 & 32.86 ( 0.00 ) & 41.11 ( 0.01 ) & 2.95 & 3.10 ( 0.04 ) + & -74.47 & 33.31 ( 0.00 ) & 41.60 ( 0.00 ) & 2.74 & 14.07 ( 0.03 ) + & -95.96 & 96.44 ( 0.01 ) & 12.48 ( 0.03 ) & 0.84 & 0.08 ( 0.00 ) + & -120.18 & 27.84 ( 0.01 ) & 8.38 ( 0.02 ) & 1.69 & 0.39 ( 0.01 ) + 5 & -43.74 & 83.18 ( 0.03 ) & 58.33 ( 0.08 ) & 0.42 & 1.01 ( 0.13 ) + & -49.43 & -4.83 ( 0.01 ) & -5.81 ( 0.02 ) & 2.11 & 11.75 ( 0.34 ) + & -54.7 & 12.57 ( 0.02 ) & -15.34 ( 0.06 ) & 0.42 & 0.24 ( 0.03 ) + & -56.6 & 9.12 ( 0.03 ) & -14.84 ( 0.05 ) & 1.05 & 0.77 ( 0.07 ) + & -75.14 & 33.00 ( 0.01 ) & 41.03 ( 0.02 ) & 1.69 & 4.67 ( 0.16 ) + 6 & -24.13 & -12.84 ( 0.04 ) & 34.28 ( 0.14 ) & 0.21 & 0.25 ( 0.05 ) + & -42.88 & -14.26 ( 0.01 ) & 1.07 ( 0.06 ) & 0.84 & 1.36 ( 0.09 ) + & -43.72 & 83.85 ( 0.01 ) & 58.7 ( 0.03 ) & 0.63 & 0.41 ( 0.03 ) + & -45.41 & -5.21 ( 0.04 ) & -5.68 ( 0.08 ) & 0.21 & 0.04 ( 0.00 ) + & -47.94 & 10.58 ( 0.06 ) & -25.2 ( 0.09 ) & 0.21 & 0.05 ( 0.01 ) + & -49.41 & -4.78 ( 0.00 ) & -5.68 ( 0.01 ) & 2.53 & 7.10 ( 0.09 ) + & -48.78 & -6.67 ( 0.01 ) & 1.54 ( 0.04 ) & 0.42 & 0.08 ( 0.01 ) + & -57.42 & 9.65 ( 0.01 ) & -14.76 ( 0.02 ) & 1.26 & 0.54 ( 0.02 ) + & -67.32 & 72.03 ( 0.01 ) & 68.13 ( 0.02 ) & 0.84 & 0.23 ( 0.01 ) + & -75.96 & 33.86 ( 0.00 ) & 41.62 ( 0.00 ) & 1.90 & 4.07 ( 0.04 ) + & -107.35 & 92.64 ( 0.01 ) & 38.53 ( 0.02 ) & 1.05 & 0.39 ( 0.01 ) + 7 & -50.31 & -5.23 ( 0.05 ) & 23.23 ( 0.11 ) & 0.63 & 0.94 ( 0.15 ) + & -56.63 & 52.02 ( 0.03 ) & 41.85 ( 0.06 ) & 0.63 & 0.74 ( 0.07 ) + & -67.16 & 71.42 ( 0.01 ) & 69.19 ( 0.02 ) & 1.05 & 3.96 ( 0.09 ) + & -77.27 & 33.30 ( 0.03 ) & 42.95 ( 0.07 ) & 0.63 & 0.22 ( 0.02 ) + 8 & -38.38 & 11.95 ( 0.02 ) & -18.89 ( 0.06 ) & 0.21 & 0.96 ( 0.09 ) + & -47.22 & -6.98 ( 0.01 ) & -2.75 ( 0.03 ) & 1.26 & 6.14 ( 0.26 ) + & -47.43 & -25.62 ( 0.02 ) & 12.09 ( 0.13 ) & 0.21 & 0.14 ( 0.03 ) + & -47.43 & -20.70 ( 0.03 ) & 11.09 ( 0.13 ) & 0.21 & 0.11 ( 0.03 ) + & -48.49 & -5.75 ( 0.00 ) & -4.57 ( 0.01 ) & 2.11 & 24.19 ( 0.30 ) + & -51.65 & -2.10 ( 0.01 ) & -3.84 ( 0.03 ) & 1.26 & 1.15 ( 0.05 ) + & -55.86 & 20.96 ( 0.04 ) & -16.42 ( 0.08 ) & 0.42 & 0.21 ( 0.02 ) + & -58.39 & 5.95 ( 0.03 ) & -12.31 ( 0.10 ) & 0.21 & 0.03 ( 0.00 ) + & -67.02 & 73.96 ( 0.02 ) & 69.02 ( 0.06 ) & 0.42 & 0.08 ( 0.01 ) + + + + we note that these averaged motions are absolute motions respect to the referenced calibrator , and hence , include both of the solar motion relative to the lsr and the systemic motion of g353 .
the solar motion has already been subtracted from averaged motions above .
we used the solar motion estimated from hipparcos data ( 0.3 mas yr@xmath4 for @xmath29 and -1.1 mas yr@xmath4 for @xmath30 direction . ;
the systemic motion of g353 is expected to be negligible since rotating motion of g353 should be almost parallel to the sun , if we assume that g353 is moving along flat galactic rotation .
but if g353 has significant deviation from flat rotation , possible contribution can be up to 15 30 km s@xmath4 ( e.g. , @xcite ) . however , actual systemic motion seems to be enough small because no constant and systemic velocity vector has been seen for maser features in g353 .
variations of @xmath7 during our monitoring is plotted in figure 5 .
plotted @xmath7 values are flux density weighted means of relevant velocity channels in c50 and c70 ( see table 3 ) .
associated error bars show standard deviations from estimated means .
these are good indicator of how maser emissions concentrate in the velocity domain . filled squares and triangles correspond to single - dish and vlbi data , respectively .
we also distinguish different peak flux densities by colour coding .
linear acceleration is evident in c70 during all detected epochs .
+ + for c50 ( top ) and c70 ( bottom ) .
the definitions of horizontal axis and markers are same as figure 4 .
the colour scale shows the peak flux densities .
error bars indicate standard deviations from the weighted means , which are measure of flux concentrations in the velocity domain .
, title="fig : " ] + systematic acceleration is also seen for c50 only in the flare - up phase ( day 300 600 ) , although these data points have relatively larger deviations compared to c70 .
linearly fitted acceleration is -5.4 @xmath6 1.2 km
s@xmath4 yr@xmath4 .
this is consistent with the value of -4.8 @xmath6 0.5 km s@xmath4 yr@xmath4 for c70 which is isolated from other components in both of velocity and spatial domains .
the fact that two independent components have shown same acceleration suggests that it is not apparent acceleration discussed in @xcite , but real one caused by a momentum supply from a protostellar outflow or wind as is the case of w75n @xcite .
in addition to the acceleration , two discontinuous jumps of @xmath7 occurred in c50 . the first jump
was seen at the initial flare event ( @xmath5 day 200 ) and the other one was synchronised with the run - out of the horizontal alignment ( @xmath5 day 600 ) .
the initial jump clearly trace the flare - up of the maser alignment and the second jump is caused by attenuation and extinction of higher velocity ( @xmath5 -56 km s@xmath4 ) components in c50 .
@xcite has attributed one possible cause of long - term , grobal variation of h@xmath0o masers to a variation of outflow activity itself .
this seems to be applicable for g353 , because it is difficult to fully attribute the systematic change of the shock front to internal fluctuations of coherent path .
synchronised acceleration of c50 and c70 , which are separated @xmath5 100 au in space and 20 km s@xmath4 in velocity , implies that subsequent momentum supply accelerates entire part of masing gas sheet , keeping or gradually losing coherence
. if this is the case , intermittent maser flare of g353 is simply interpretable as episodic launching of an outflow .
expected launching time - scale is only @xmath5 2 yr based on the interval between two flares in this case .
this is well comparable with the typical time - scale of several radio jets @xcite .
ccccccday & @xmath40 & @xmath41 & total width & peak flux density & integrated flux + & ( km s@xmath4 ) & ( km s@xmath4 ) & ( km s@xmath4 ) & ( jy ) & ( jy km s@xmath4 ) + + 426 & -53.20 & 1.63 & 8.40 & 174.67 ( 3.74 ) & 617.91 ( 1.74 ) + 434 & -53.49 & 1.48 & 5.67 & 138.75 ( 8.01 ) & 489.77 ( 3.67 ) + 437 & -53.43 & 1.48 & 5.67 & 122.79 ( 6.54 ) & 412.05 ( 3.00 ) + 442 & -54.00 & 1.70 & 5.88 & 62.63 ( 5.27 ) & 319.77 ( 2.40 ) + 473 & -54.94 & 1.40 & 3.99 & 64.57 ( 8.01 ) & 204.19 ( 3.67 ) + 500 & -54.13 & 1.67 & 4.62 & 68.64 ( 6.14 ) & 215.81 ( 2.80 ) + 501 & -53.86 & 1.77 & 3.78 & 76.39 ( 7.41 ) & 212.60 ( 3.41 ) + 551 & -55.36 & 0.85 & 0.42 & 40.73 ( 11.95 ) & 93.08 ( 5.48 ) + 671 & -48.86 & 0.27 & 0.63 & 24.30 ( 5.74 ) & 33.59 ( 2.60 ) + 686 & -50.97 & 3.49 & 1.05 & 28.38 ( 4.74 ) & 43.13 ( 2.14 ) + 702 & -48.81 & 0.56 & 1.89 & 14.56 ( 2.87 ) & 37.99 ( 1.34 ) + 716 & -48.64 & 0.55 & 1.68 & 12.29 ( 2.27 ) & 32.92 ( 1.07 ) + 729 & -50.00 & 2.96 & 1.26 & 10.88 ( 2.47 ) & 24.57 ( 1.14 ) + 744 & -48.81 & 0.74 & 0.84 & 10.55 ( 2.80 ) & 20.57 ( 1.27 ) + 759 & -48.71 & 0.66 & 1.47 & 15.62 ( 2.47 ) & 36.79 ( 1.14 ) + 780 & -49.53 & 3.24 & 1.89 & 17.76 ( 3.27 ) & 48.41 ( 1.47 ) + 796 & -48.44 & 0.31 & 0.84 & 23.44 ( 5.14 ) & 66.10 ( 2.40 ) + 832 & -49.34 & 2.65 & 5.67 & 96.08 ( 2.80 ) & 204.39 ( 1.34 ) + 834 & -49.15 & 2.30 & 6.09 & 101.63 ( 3.27 ) & 211.07 ( 1.47 ) + 839 & -49.35 & 2.79 & 6.72 & 114.65 ( 3.14 ) & 250.66 ( 1.47 ) + 852 & -48.68 & 2.06 & 4.41 & 108.64 ( 4.27 ) & 235.30 ( 1.94 ) + 874 & -48.11 & 1.37 & 2.94 & 151.04 ( 8.15 ) & 283.25 ( 3.74 ) + 893 & -48.27 & 1.82 & 3.15 & 218.41 ( 9.15 ) & 368.71 ( 4.21 ) + 899 & -47.95 & 0.46 & 1.89 & 125.33 ( 11.62 ) & 210.26 ( 5.34 ) + 901 & -47.98 & 0.48 & 1.89 & 154.31 ( 15.96 ) & 275.03 ( 7.34 ) + 906 & -47.82 & 0.56 & 2.31 & 153.64 ( 11.42 ) & 272.76 ( 5.27 ) + 912 & -47.93 & 0.31 & 1.05 & 142.02 ( 26.71 ) & 234.97 ( 12.29 ) + 913 & -47.88 & 0.43 & 1.68 & 189.97 ( 17.09 ) & 310.15 ( 7.81 ) + + 183 & -56.65 & 3.57 & 2.52 & 23.50 ( 0.99 ) & 24.00 ( 0.56 ) + 323 & -51.24 & 1.70 & 12.81 & 716.30 ( 17.01 ) & 2224.24 ( 14.28 ) + 402 & -52.93 & 1.15 & 14.07 & 632.37 ( 2.86 ) & 969.45 ( 3.29 ) + 503 & -53.13 & 2.50 & 9.24 & 39.42 ( 1.39 ) & 122.53 ( 1.51 ) + 624 & -51.08 & 2.83 & 4.20 & 12.80 ( 0.94 ) & 21.95 ( 0.12 ) + 674 & -50.30 & 3.13 & 4.62 & 11.26 ( 0.40 ) & 23.36 ( 0.05 ) + 757 & -54.61 & 2.99 & 1.26 & 2.17 ( 0.37 ) & 2.10 ( 0.07 ) + 841 & -48.94 & 2.19 & 6.51 & 66.42 ( 0.63 ) & 87.55 ( 0.12 ) + + 426 & -73.79 & 1.24 & 4.20 & 106.50 ( 3.74 ) & 203.39 ( 1.94 ) + 434 & -74.13 & 0.71 & 2.94 & 122.39 ( 8.01 ) & 233.3 ( 4.14 ) + 437 & -74.12 & 0.75 & 3.15 & 117.25 ( 6.54 ) & 249.53 ( 3.34 ) + 442 & -73.88 & 0.87 & 3.36 & 82.73 ( 5.27 ) & 202.32 ( 2.67 ) + 473 & -74.26 & 0.41 & 1.47 & 49.74 ( 8.01 ) & 95.68 ( 4.14 ) + 500 & -74.45 & 0.34 & 1.47 & 151.44 ( 6.14 ) & 168.47 ( 3.14 ) + 501 & -74.31 & 0.72 & 1.89 & 155.98 ( 7.41 ) & 195.24 ( 3.81 ) + 551 & & & & @xmath42 35.86 ( 11.95 ) & + 671 & -75.95 & 0.20 & 0.63 & 23.44 ( 5.74 ) & 35.12 ( 2.94 ) + 686 & -76.28 & 0.25 & 0.84 & 41.87 ( 4.74 ) & 55.75 ( 2.40 ) + 702 & -76.72 & 0.24 & 0.84 & 19.16 ( 2.87 ) & 11.02 ( 1.47 ) + 716 & -76.72 & 0.25 & 0.84 & 9.68 ( 2.27 ) & 3.54 ( 1.14 ) + 729 & -76.81 & 0.11 & 0.21 & 7.34 ( 2.47 ) & 7.41 ( 1.27 ) + 744 & & & & @xmath42 8.41 ( 2.80 ) & + 759 & & & & @xmath42 7.41 ( 2.47 ) & + 780 & & & & @xmath42 9.82 ( 3.27 ) & + 796 & & & & @xmath42 15.42 ( 5.14 ) & + 832 & & & & @xmath42 8.41 ( 2.80 ) & + 834 & & & & @xmath42 9.82 ( 3.27 ) & + 839 & & & & @xmath42 9.41 ( 3.14 ) & + 852 & & & & @xmath42 12.82 ( 4.27 ) & + ccccccday & @xmath40 & @xmath41 & total width & peak flux density & integrated flux + & ( km s@xmath4 ) & ( km s@xmath4 ) & ( km s@xmath4 ) & ( jy ) & ( jy km s@xmath4 ) + + 874 & & & & @xmath42 24.44 ( 8.15 ) & + 893 & & & & @xmath42 27.44 ( 9.15 ) & + 899 & & & & @xmath42 34.85 ( 11.62 ) & + 901 & & & & @xmath42 47.88 ( 15.96 ) & + 906 & & & & @xmath42 34.25 ( 11.42 ) & + 912 & & & & @xmath42 80.13 ( 26.71 ) & + 913 & & & & @xmath42 51.28 ( 17.09 ) &
+ + 323 & -69.44 & 1.40 & 4.41 & 16.60 ( 0.68 ) & 38.03 ( 0.51 ) + 402 & -71.92 & 1.67 & 7.35 & 18.90 ( 0.22 ) & 65.93 ( 0.28 ) + 503 & -74.04 & 1.14 & 6.51 & 165.00 ( 0.26 ) & 175.37 ( 0.44 ) + 624 & -75.41 & 0.38 & 1.89 & 18.60 ( 0.80 ) & 19.77 ( 0.23 ) + 674 & -75.99 & 0.38 & 2.10 & 28.80 ( 0.15 ) & 24.79 ( 0.23 ) + 757 & -77.32 & 0.25 & 0.63 & 2.04 ( 0.34 ) & 1.08 ( 0.11 ) + + + + the characteristic velocity of @xmath5 500 km s@xmath4 , which is estimated from the spatial scale of maser distribution ( @xmath5 200 au ) divided by the time - scale ( @xmath5 2yr ) , is also consistent with the proper motion velocity of radio jets associated with several high mass ysos ( e.g. , @xcite ) .
thus , observed variation can be caused by standard radio jet in high mass star - forming region .
the relation between the expected velocity of a jet and the 3d velocity of masers ( @xmath5 100 km s@xmath4 ) is roughly same as the case of cepa hw2 region @xcite , where velocities of radio jet and relatively wide angle maser flow is 500 and 70 km s@xmath4 , respectively .
g353 is thought to be a possible candidate of high mass yso with a radio jet activity through these reasons .
the momentum rate of maser flow is estimated as 1.1 @xmath1 10@xmath8 @xmath43 km s@xmath4 yr@xmath4 based on the acceleration of @xmath5 5 km s@xmath4 yr@xmath4 . here
, we assumed that the masing gas sheet has uniform thickness of 1 au , i.e. , typical size of maser feature .
total area of the sheet is evaluated as 200 @xmath1 200 au@xmath2 from the extent of maser distribution . adopted
number density of molecular hydrogen is 10@xmath44 cm@xmath8 that is also typical value for a h@xmath0o maser feature .
we note that assumed volume is , and hence , the momentum rate is only a lower limit , because a mass fraction of masing gas sheet to total shocked gas should be small @xcite .
estimated lower limit is clearly larger than a typical values of low mass star - formation ( @xmath5 10@xmath46 @xmath43 km s@xmath4 yr@xmath4 ; @xcite ) and consistent with that of high mass star - formation ( e.g. , @xcite for comparison ) .
if we adopt a jet velocity of 500 km s@xmath4 , it is immediately converted to the outflow rate of 2.2 @xmath1 10@xmath47 m@xmath9 yr@xmath4 .
we finally get the lower - limit accretion rate of 10@xmath46
10@xmath48 m@xmath9 yr@xmath4 , where we have adopted typical fraction between outflow rate over accretion rate ( 1 10 % ) in a case of mhd driven flow ( e.g. , @xcite ) .
although we have considered pure mhd jet ( e.g. , @xcite ) here , strong radiation from the host yso should also play an important role for outflow launching in high mass star - fomation .
if the host yso is enough young and an accretion rate is enough high , accretion luminosity is thought to be significant or even dominant in total flux @xcite .
outflow variations may directly reflect variations of accretion rate onto the protostellar surface in this case .
if this is true , variation pattern of the h@xmath0o maser can be important clue for accretion mechanism at the innermost part .
further discussion about this requires exact stellar parameters in ir wavelength .
another explanation for intermittent maser flare is a precessing jet , where we can observe maser emission only in a certain fraction of precession period because of beaming effect .
if simple binary interaction which has rotating period of 2 yr is considered , expected binary separation is over 3 au with a total mass of 10 m@xmath9 .
by contrast , if we assume escape velocity at he launching point as typical velocity of a jet , expected launching radius is estimated as follows , @xmath49 here @xmath50 and @xmath51 indicate mass of the driving source and velocity of launched jet , respectively .
it is well smaller than the expected binary separation and this scenario seems to be possible .
high - resolution image of molecular lobe obtained with upcoming atacama large millimeter / submillimeter array ( alma ) will be helpful to distinguish two scenarios .
direct detection of precessing trajectory of a jet is also the best way to test them .
this can be achieved by high - sensitivity observation of radio continuum or ir line emissions , as the case of iras16547 - 4247 ( @xcite ) .
the h@xmath0o maser distribution is completely inside the error circle of the position of class ii ch@xmath15oh maser reported by cp08 .
since there is no detectable oh maser emission ( e.g. , cp08 ; @xcite ) , these masers are thought to be excited by identical high mass protostellar object . if the h@xmath0o masers in g353 is actually excited by a radio jet , such a close location ( @xmath42 a few hundred au ) of two types of masers and the radio jet is also similar to the cases of several high mass protostellar objects . besides cepa
hw2 @xcite , @xmath10 20126 + 4104 is the best example of this kind of source ( @xcite ) .
h@xmath0o masers in @xmath10 20126 + 4104 are obviously excited in front of the radio knot which emerges from the nearly edge - on disc ( e.g. , @xcite ) . the intriguing feature of @xmath10 20126 + 4104 is asymmetry of the radio knot and h@xmath0o maser in @xcite . in spite of overall bipolar structure of outflow verified by radio and ir line observations @xcite , the radio knot and the most of h@xmath0o masers have been detected only in blue - shifted side .
the non - negligible blue - shift dominance of h@xmath0o maser is also evident in long - term single - dish study @xcite .
their light curve , furthermore , has shown weak intermittent variation of 2 3 yr scale alike g353 .
these facts strongly support the interpretation that g353 is a candidate of high mass protostellar object with intermittent maser activity caused by a protostellar jet , as proposed in cp08 .
if this is the case , g353 can be a good target of direct imaging of the accretion disc , because of the possible pole - on geometry of the disc - jet system ( e.g. , @xcite ) .
our vlbi and single - dish monitoring of 22 ghz h@xmath0o maser emission from the high mass young stellar object g353.273 + 0.641 with vera and tomakamai 11-m radio telescope has detected two times of maser flares .
the eight epochs of vlbi data have revealed that these flares have been accompanied by structural change of the prominent shock front traced by h@xmath0o maser alignments .
we have detected only blue - shifted maser emissions and all maser features have been distributed within very small area ( 200 @xmath1 200 au@xmath2 ) in spite of wide velocity range ( @xmath3 100 km s@xmath4 ) .
these facts suggest relatively pole - on geometry of the disc - jet system .
the light curve shows notably intermittent variation .
this suggests a possibility that the h@xmath0o masers in g353.273 + 0.641 are excited by episodic radio jet .
the interval of two flares is about two years and well comparable to typical time - scale of a radio jet . the characteristic velocity of 500 km s@xmath4 , which is estimated from the time and spatial scale , is also consistent with the velocity of the radio jets associated with several high mass young stellar objects .
these facts also suggest that the h@xmath0o maser in g353.273 + 0.641 is excited in a strong shock of a radio jet
. the precessing jet may be also able to explain intermittent maser flare and this will be tested by high - resolution imaging of molecular lobe or direct detection of precessing trajectory of a jet .
two strong velocity components of c50 ( -53 @xmath6 7 km s@xmath4 ) and c70 ( -73 @xmath6 7 km s@xmath4 ) have shown synchronised linear acceleration of the flux weighted @xmath7 values ( @xmath5 -5 km
s@xmath4 yr@xmath4 ) during the flare phase .
the acceleration can be converted to the lower - limit momentum rate of 1.1 @xmath1 10@xmath8 m@xmath9 km s@xmath4 yr@xmath4 . if we use a jet velocity of 500 km s@xmath4 , expected lower limit of accretion rate is 10@xmath46
10@xmath48 m@xmath9 yr@xmath4 in the case of mhd driven jet .
the co - existence of class ii ch@xmath15oh maser at 6.7 ghz and absence of any detectable oh masers , as pointed - out in cp08 , indicate that the driving source is a high mass protostellar object . if h@xmath0o masers are actually excited by a radio jet , the close location ( @xmath42 a few hundred au ) of two types of masers and the radio jet is also similar to the cases of several high mass protostellar objects .
@xmath10 20126 + 4104 especially has h@xmath0o masers excited by the radio jet and quite resembles g353.273 + 0.641 in blue - shift dominant spectra and short - term ( 2 3 yr ) intermittent variation of the maser .
we conclude that g353 is a candidate of high mass protostellar object alike @xmath10 20126 + 4104 , as proposed in cp08 .
the possible pole - on geometry of disc - jet system can be suitable for direct imaging of the accretion disc in this case .
we would like to thank all the member of vera and hokkaido university for their assistance in observations and data analyses .
we also thank the referee for helpful comments .
this work was financially supported by the grant - in - aid for the japan society for the promotion of science fellows ( k. m. ) .
99 anglada g. , 1996 , aspc , 93 , 2 arce h. g. , sargent a. i. , 2006 , apj , 646 , 1070 arce h. g. , shepherd d. , gueth f. , lee c .- f . , bachiller r. , rosen a. , beuther h. , 2007 , in reipurth b. , jewitt d. , keil k. , eds , protostars and planets v. univ . of arizona press , tucson , p. 245
baba j. , asaki y. , makino j. , miyoshi m. , saitoh t. r. , wada k. , 2009 , apj , 706 , 471 brand j. , cesaroni r. , comoretto g. , felli m. , palagi f. , palla f. , valdettaro r. , 2003 , a&a , 407 , 573 breen , s. l. , caswell j. l. , ellingsen , s. p. , phillips , c. j. , 2010 , mnras , 406 , 1487 brooks , k. j. , garay , g. , mardones , d. , bronfman , l. , 2003 , apj , 594 , l131 caswell j. l. , 1997 , mnras , 209 , 203 caswell j. l. , 2004 , mnras , 351 , 279 caswell j. l. , phillips c. j. , 2008 , mnras , 386 , 1521 ( cp08 ) caswell j. l. , breen s. l. , 2010 , mnras , 407 , 2599 caratti o garatti a. , froebrich d. , eisleffel j. , giannini t. , nisini , b. , 2008 , a&a , 485 , 137 cesaroni r. , neri r. , olmi l. , testi l. , walmsley c. m. , hofner p. , 2005 , a&a , 434 , 1039 curiel s. et al . , 2006 , apj , 638 , 878 dehnen w. , binney j. j. , 1998 , mnras , 298 , 387 edris k. a. , fuller g. a. , cohen r. j. , etoka s. , 2005 , a&a , 434 , 213 elitzur m. , hollenbach d. j. , mckee c. f. , 1989 , apj , 346 , 983 felli m. et al .
, 2007 , a&a , 476 , 373 furuya r. s. , kitamura y.,wootten h. a. , claussen m. j. , saito m. , marvel k. b. , kawabe r. 2000 , apj , 542 , l135 hofner p. , cesaroni r. , olmi l. , rodrguez l. f. , mart j. , araya e. , 2007 , a&a , 465 , 197 honma m. et al . , 2003 ,
pasj , 55 , l57 honma m. , et al . , 2007 , pasj , 59 , 889 honma m. , tamura y. , reid m. j. , 2008 , pasj , 60 , 951 hosokawa t. , yorke h. , omukai k. , 2010 , apj , 721 , 478 hunter t. r. , taylor g. b. , felli m. , tofani g. , 1994 , a&a , 284 , 215 iguchi s. , kurayama t. , kawaguchi n. , kawakami k. , 2005 , pasj , 57,259 imai h. , omodaka t. , hirota t. , umemoto t. , sorai k. , kondo t. , 2006 , pasj , 58 , 883 kawaguchi n. , sasao t. , manabe s. , 2000 , in proc .
spie , 4015 , radio telescopes , ed . h. r. butcher ( washington : spie ) , 544 kobayashi h. et al . , 2008 , iau symp .
248 , 148 kraus s. et al . , 2010 , nat , 466 , 339 ma c. et al . , 1998 , aj , 116 , 516 machida m. n. , inutsuka s. , matsumoto t. , 2006 , apj , 647 , l151 machida m. n. , inutsuka s .-
i . , matsumoto t. , 2008 , apj , 676 , 1088 mart , j. , rodrguez l. f. , reipurth b. , 1998 , apj , 502 , 337 minier v. , ellingsen s. p. , norris r. p. , booth r. s. , 2003 , a&a , 403 , 1095 moscadelli l. , cesaroni r. , rioja m. j. , 2000 , a&a , 360 , 663 moscadelli l. , cesaroni r. , rioja m. j. , dodson r. , reid m. j. , 2011 , a&a , 526 , a66 motogi k. et al . , 2008 , mnras , 390 , 523 motogi k. , sorai k. , habe a. , honma m. , kobayashi h. , sato k. , 2011 , pasj , 63 , 31 nakagawa a. et al . , 2008 , pasj , 60 , 1013 neckel t. , 1978 , a&a , 69 , 51 pelletier g. , pudritz r. e. , 1992 , apj , 394 , 117 petrov l. , kovalev y. y. , fomalont e. b. , gordon d. , 2006 , apj , 131 , 1872 reid m. j. et al . , 2009 , apj , 700 , 137 rodrguez l. f. , garay g. , brooks k. j. , mardones d. , 2005 , apj , 626 , 953 sakellis s. , taylor m. i. , taylor k. n. r. , vaile k. a. , han t. d. , 1984 , pasp , 96 , 543 shepherd d. s. , borders t. , claussen m. , shirley y. , kurtz s. , 2004 , apj , 614 , 211 shibata k. m. , kameno s. , inoue m. , kobayashi h. 1998 , in asp conf .
144 , iau colloq .
164 , radio emission from galactic and extragalactic compact sources , ed .
j. a. zensus , g. b. taylor , & j. m. wrobel ( san francisco : asp ) , 413 sorai k. et al . , 2008 , pasj , 60 , 1285 su
liu s .- y . , chen h .-
r . , zhang q. , cesaroni r. , 2007 , apj , 671 , 571 torrelles j. m. et al . , 2003 , apj , 598 , l115 torrelles j. m. et al . , 2011 , mnras , 410 , 627 torstensson k. j. e. , van langevelde h. j. , vlemmings w. h. t. , bourke , s. , 2011 , a&a , 526 , a38 trinidad m. a. , torrelles j. m. , rodrguez l. f. , curiel s. , 2007 , apj , 134 , 1870 trinidad m. a. , rodrguez t. , rodrguez l. f. , 2009 , apj , 706 , 244 | we have performed vlbi and single - dish monitoring of 22 ghz h@xmath0o maser emission from the high mass young stellar object g353.273 + 0.641 with vera ( vlbi exploration of radio astrometry ) and tomakamai 11-m radio telescope .
two maser flares have been detected , separated almost two years .
frequent vlbi monitoring has revealed that these flare activities have been accompanied by structural change of the prominent shock front traced by h@xmath0o maser alignments .
we have detected only blue - shifted emissions and all maser features have been distributed within very small area of 200 @xmath1 200 au@xmath2 in spite of wide velocity range ( @xmath3 100 km s@xmath4 ) .
the light curve shows notably intermittent variation and suggests that the h@xmath0o masers in g353.273 + 0.641 are excited by episodic radio jet .
the time - scale of @xmath5 2 yr and characteristic velocity of @xmath5 500 km s@xmath4 also support this interpretation .
two isolated velocity components of c50 ( -53 @xmath6 7 km s@xmath4 ) and c70 ( -73 @xmath6 7 km s@xmath4 ) have shown synchronised linear acceleration of the flux weighted @xmath7 values ( @xmath5 -5
km s@xmath4 yr@xmath4 ) during the flare phase .
this can be converted to the lower - limit momentum rate of 1.1 @xmath1 10@xmath8 m@xmath9 km s@xmath4 yr@xmath4 .
maser properties are quite similar to that of @xmath10 20126 + 4104 especially .
this corroborates the previous suggestion that g353.273 + 0.641 is a candidate of high mass protostellar object .
the possible pole - on geometry of disc - jet system can be suitable for direct imaging of the accretion disc in this case . [ firstpage ] ism : jets and outflows masers stars : early - type stars : formation . |
recently , a new particle of mass about 125 gev has been discovered at the cern large hadron collider ( lhc ) with a total production rate consistent with that of the standard model ( sm ) higgs boson @xcite . confirming this higgs - like particle as the one responsible for the electroweak symmetry breaking is of paramount importance in particle physics because , for one thing , it explains the origin of mass for elementary particles . while further detailed examinations are required
, the current lhc data show some deviations in the pattern of its decay branching ratios from the sm expectation .
this leads to the speculation that the higgs sector may not as simple as the one in sm . in certain new physics models such as supersymmetry
, the higgs sector has to be extended with additional nontrivial isospin @xmath8 scalar multiplets for consistency or to explain new phenomena .
such an extension also holds the capacity to provide additional cp - violating sources for low - energy phenomena as well as baryon asymmetry of the universe . for example , the two - higgs doublet model ( 2hdm ) @xcite is an extensively studied prototype in which an additional scalar doublet is introduced .
@xmath8 triplet higgs fields also occur in some new physics models , such as the left - right symmetric model @xcite and little higgs models @xcite . by introducing a complex triplet higgs field
, it is possible to have an effective dimension-5 operator for generating tiny majorana mass for neutrinos .
therefore , it is important to determine the true higgs sector in order to exactly know what kind of new physics models exist at the tev or higher energy scales . in this paper , we want to focus on the phenomenology of the extended higgs sector in the model proposed by georgi and machacek ( gm ) @xcite in mid-80s .
we investigate how one can distinguish it from the other higgs - extended models at the lhc .
the gm model contains a higgs doublet field @xmath9 and a triplet field @xmath10 , with the latter containing a hypercharge @xmath11 component and a @xmath12 component .
the model is of great interest because it can provide tiny mass to neutrinos la the seesaw mechanism , dubbed the type - ii seesaw @xcite .
moreover , it has been shown that the higgs potential in this model can be constructed to maintain a custodial @xmath4 symmetry at the tree level @xcite , keeping the electroweak @xmath0 parameter at unity to be consistent with the experimental constraint . in the model , there are 5-plet higgs bosons @xmath1 ( = @xmath13 ) , 3-plet higgs bosons @xmath2 ( = @xmath14 ) and singlet higgs boson @xmath15 under the classification of the @xmath4 symmetry .
the masses of the higgs bosons belonging to the same @xmath4 multiplet are the same at the tree level as the consequence of custodial symmetry .
the doubly - charged higgs boson @xmath16 , for example , is an important but not unique feature of the model .
finding particles in one higgs multiplet and checking their ( near ) mass degeneracy would better verify the model .
strategies of discovering such higgs bosons , however , depend largely on the vacuum expectation value ( vev ) of the higgs triplet field , @xmath17 . in the minimal higgs triplet model ( htm ) where only one additional complex higgs triplet is introduced , the doubly - charged higgs bosons couples dominantly to a pair of like - sign leptons when @xmath18 gev .
the collider phenomenology of this scenario has been extensively studied recently @xcite .
the doubly - charged higgs boson has been searched for at the tevatron @xcite and the lhc @xcite by looking for like - sign lepton pairs with the same or different flavors .
a lower mass bound of about @xmath19 gev has been obtained for most scenarios . on the other hand , the doubly - charged higgs bosons couples dominantly to a pair of like - sign @xmath20 bosons when @xmath21 gev .
this possibility is less explored experimentally . besides , the triplet vev in the htm is constrained by the @xmath0 parameter to be less than a few gev , limiting significantly the discovery reach at the lhc . in the gm model , a larger triplet vev is allowed due to the custodial symmetry .
it is therefore interesting to consider signatures of the like - sign gauge boson decays . in ref .
@xcite , collider phenomenology of the gm model has been discussed in the case of light triplet - like higgs bosons , _
e.g. _ , less than 100 gev .
a recent study by one of the authors and collaborators @xcite finds that with @xmath22 gev and appropriate cuts , the current lhc can reach up to @xmath23 gev for the doubly - charged higgs mass . in this work ,
we further explore consequences of the custodial symmetry in the higgs sector of the gm model and study the phenomenology of its entire higgs sector at the lhc .
we find that the single production of @xmath1 via the weak vector boson fusion process is useful to test the mass degeneracy among the @xmath1 bosons .
we also find that the drell - yan process , where @xmath1 and @xmath2 are simultaneously produced can be used to check the mass degeneracy among @xmath2 .
the structure of this paper is organized as follows .
we review the gm model in section [ sec : model ] .
the higgs bosons are first classified according to their group representations under the custodial symmetry .
we then consider possible mixings between the two triplets and between the two singlets , and work out their masses . a mass relation among the higgs bosons of different representations
is obtained in the decoupling limit when the triplet vev vanishes .
finally , we show the yukawa couplings between sm fermions and the physical higgs bosons . in section [ sec : constraints ] , we consider both theoretical constraints of perturbative unitarity and vacuum stability and the experimental constraint from the @xmath24-pole data of @xmath25 decay at one - loop level . in particular , they impose bounds on the triplet vev and the higgs triplet mass
. in section [ sec : decays ] , we discuss in detail how the higgs bosons decay in scenarios with or without hierarchy in the masses of the physical higgs singlet , 3-plet , and 5-plet .
the collider phenomenology of the model can be drastically different in different regions of the @xmath17-@xmath26 ( @xmath26 is the mass difference between @xmath1 and @xmath2 ) space .
section [ sec : pheno ] discusses how the higgs bosons can be searched for at the lhc . finally , we compute the decay rates of @xmath27 and @xmath28 in the model in section [ sec : diphoton ] .
our findings are summarized in section [ sec : summary ] .
in the gm model , the higgs sector is composed of the sm isospin doublet higgs field @xmath29 with hypercharge @xmath30 and two isospin triplet higgs fields @xmath31 with @xmath11 and @xmath32 with @xmath12 .
these fields can be expressed in the form : @xmath33 where @xmath9 and @xmath10 are transformed under @xmath34 as @xmath35 and @xmath36 with @xmath37 and @xmath38 being the @xmath39 generators . the neutral components in eq .
( [ eq : higgs_matrices ] ) can be parametrized as @xmath40 where @xmath41 , @xmath42 and @xmath43 are the vev s for @xmath44 , @xmath45 and @xmath46 , respectively . when the two triplet vev s @xmath42 and @xmath43 are taken to be the same , _
i.e. _ , @xmath47 , the @xmath34 symmetry is reduced to the custodial @xmath4 symmetry . the phase convention for the component scalar fields
are chosen to be @xmath48 ,
@xmath49 , @xmath50 , @xmath51 and @xmath52 . the relevant lagrangian involving the higgs fields can be written as @xmath53 where @xmath54 , @xmath55 , @xmath56 and @xmath57 are the kinetic term , the yukawa interaction between @xmath29 and the fermions , the neutrino yukawa interaction between @xmath31 and the lepton doublets , and the higgs potential , respectively .
the most general higgs potential invariant under the @xmath58 symmetry in terms of the fields defined in eq .
( [ eq : higgs_matrices ] ) is @xmath59 ^ 2 + \lambda_3\text{tr}[(\delta^\dagger\delta)^2]\notag\\ & + \lambda_4\text{tr}(\phi^\dagger\phi)\text{tr}(\delta^\dagger\delta ) + \lambda_5\text{tr}\left(\phi^\dagger\frac{\tau^a}{2}\phi\frac{\tau^b}{2}\right ) \text{tr}(\delta^\dagger t^a\delta
t^b)\notag\\ & + \mu_1\text{tr}\left(\phi^\dagger \frac{\tau^a}{2}\phi\frac{\tau^b}{2}\right)(p^\dagger \delta p)^{ab } + \mu_2\text{tr}\left(\delta^\dagger t^a\delta t^b\right)(p^\dagger \delta p)^{ab } , \label{eq : pot}\end{aligned}\ ] ] where @xmath60 are the pauli matrices , @xmath61 are the @xmath62 matrix representation of the @xmath39 generators given by @xmath63 and the matrix @xmath64 is defined as @xmath65 as in the htm , the sm electroweak symmetry breaking can induce the triplet field to develop a vev @xmath17 through the @xmath66 term in the higgs potential . to our knowledge ,
most of the previous analyses ignore both @xmath66 and @xmath67 interactions in their phenomenology studies .
we will keep these terms in this work . using the tadpole conditions ,
@xmath68 the parameters @xmath69 and @xmath70 can be eliminated as [ eq : m1m2 ] @xmath71 where @xmath72 and @xmath73 with @xmath74 and @xmath75 . in eq .
( [ eq : m1m2 ] ) , we introduce @xmath76 and @xmath77 as @xmath78 the second and third conditions in eq .
( [ eq : vc ] ) give the same constraint in eq .
( [ eq : m2 ] ) as long as @xmath79 .
before we discuss the mass matrices and the mass eigenstates for the higgs bosons , it is convenient to classify the higgs boson states according to the custodial @xmath4 symmetry .
the triplet field @xmath10 , which can be understood as a @xmath80 representation of the @xmath4 multiplet , can be decomposed into the irreducible representations @xmath81 . likewise , the doublet field @xmath9 being the @xmath82 representation of the @xmath4 multiplet , can be decomposed into @xmath83 .
the @xmath84 representation of @xmath9 can be identified as the nambu - goldstone ( ng ) bosons of the sm as long as there is no mixing between the @xmath84 representations of @xmath10 and @xmath9 .
the 5-plet ( @xmath16 , @xmath85 and @xmath86 ) , the 3-plet ( @xmath87 and @xmath88 ) and the singlet ( @xmath89 ) originating from @xmath10 can be related to the original component fields as @xmath90 it is seen that @xmath86 and @xmath91 are cp - even states , whereas @xmath88 is a cp - odd state . in eq .
( [ eq : cust ] ) , the scalar fields with a tilde are not mass eigenstates in general .
they can in principle mix with the corresponding scalar fields from the higgs doublet field .
the mass of the doubly - charged higgs boson @xmath16 is @xmath92 the mass matrix for the cp - odd higgs states in the basis of ( @xmath93 , @xmath88 ) and that for the singly - charged states in the basis of ( @xmath94 , @xmath95 , @xmath96 ) are given by @xmath97 the mass matrix for the cp - even higgs states in the basis of ( @xmath98 , @xmath89 , @xmath99 ) is @xmath100 where the elements of the @xmath101 submatrix are @xmath102.\end{aligned}\ ] ] [ eq : meven_ele ] the mass eigenstates are related to the above - mentioned states via the following unitary transformations @xmath103 where @xmath104 and @xmath105 are the ng bosons for the longitudinal components of the @xmath106 and @xmath24 bosons .
the explicit forms of the unitary matrices are @xmath107 where @xmath108 , @xmath109 and the mixing angle @xmath110 is defined by @xmath111 the masses of the singly - charged higgs bosons ( @xmath85 and @xmath112 ) , the cp - odd higgs boson ( @xmath113 ) and the cp - even higgs bosons ( @xmath86 , @xmath15 and @xmath5 ) are then @xmath114 it is observed that @xmath16 , @xmath85 and @xmath86 are degenerate in mass and so are @xmath112 and @xmath113 because of the custodial invariance in the higgs potential . therefore , the higgs boson masses can be conveniently written as @xmath115 the five dimensionless couplings in the potential , @xmath116 , can be substituted by the five physical parameters @xmath117 , @xmath118 , @xmath119 , @xmath120 and @xmath110 as follows : @xmath121,\notag\\ \lambda_3 & = \frac{1}{v^2s_h^2}\left[c_h^2(2m_1 ^ 2 - 3m_{h_3}^2)+m_{h_5}^2-m_2 ^ 2\right],\notag\\ \lambda_4&= \frac{1}{6v^2s_hc_h}\left[\frac{\sqrt{6}}{2}s_{2\alpha}(m_h^2-m_{h_1}^2)+3s_hc_h(2m_{h_3}^2-m_1 ^ 2)\right],\notag\\ \lambda_5&= \frac{2}{v^2}(m_1 ^ 2-m_{h_3}^2 ) .
\label{lambdas}\end{aligned}\ ] ] the decoupling limit of this model can be obtained when we take the @xmath122 limit ( or equivalently @xmath123 ) . in this limit , the mass formulae of the higgs bosons reduce to @xmath124 notice that @xmath77 is proportional to @xmath125 , and thus it becomes zero in this limit for a fixed value of @xmath67 .
if one wants to fix @xmath77 at a finite value , @xmath67 has to be taken to infinity to compensate @xmath123 and eventually violates perturbativity in this model .
therefore , @xmath126 is the natural choice in this limit . on the other hand , @xmath76 is proportional to @xmath127 . even in the @xmath123 limit
, we can take a finite value for @xmath76 as long as @xmath128 at the same rate as @xmath129 .
consequently , the triplet - like higgs bosons decouple when @xmath130 , and only @xmath5 remains at the electroweak scale and acts like the sm higgs boson .
in addition , in the decoupling region @xmath131 , we find a simple mass relation for the triplet - like higgs bosons : @xmath132 for the convenience in discussing interactions between leptons and the higgs triplet field , we reorganize the higgs fields as follows : @xmath133 the relationship between the two representations in eqs .
( [ eq : higgs_matrices ] ) and ( [ 22mat ] ) are given in appendix a. with the introduction of the @xmath31 field above , the yukawa interactions between the lepton doublets and the higgs triplet are @xmath134 if we assign two units of lepton number to @xmath31 , then the @xmath135 and @xmath66 terms in the higgs potential violate the lepton number . if we then take @xmath136 , @xmath113 becomes massless and corresponds to the ng boson for the spontaneous breakdown of the global @xmath137 lepton number symmetry . in fact , @xmath112 are also massless in that case because of the custodial symmetry . the majorana mass of neutrinos is derived as @xmath138 this mass matrix can be diagonalized by the pontecorvo - maki - nakagawa - sakata matrix @xmath139 , and the yukawa matrix @xmath140 can be rewritten as @xmath141 the left - handed neutrino fields are then transformed as @xmath142 for simplicity , we hereafter assume that @xmath139 is the unit matrix and the mass eigenvalues of @xmath143 are degenerate : @xmath144 . in terms of the scalar mass eigenstates , the interaction terms are @xmath145\notag\\ & + \frac{2m_\nu}{s_hv}\left[\frac{1}{\sqrt{3}}(h_5 ^ 0+\sqrt{2}s_\alpha h+c_\alpha h_1 ^ 0)+i(g^0 s_h + h_3 ^ 0c_h)\right]\overline{\nu_i^c}p_l\nu_i + \text{h.c.}\label{ny}\end{aligned}\ ] ] the yukawa interaction between the fermions of one generation and the higgs doublet @xmath29 is given by @xmath146 with @xmath147 . in terms of the fermion masses @xmath148 and the physical higgs states , the interaction terms are expressed as @xmath149 -\sum_{f = u , d , e}\frac{m_f}{v } \left[\frac{c_\alpha}{c_h}\bar{f}fh-\frac{s_\alpha}{c_h}\bar{f}fh_1 ^ 0+i\text{sign}(f ) \tan\theta_h\bar{f}\gamma_5fh_3 ^ 0\right ]
\notag\\ & -\frac{\sqrt{2}v_{ud}}{v}\left[\tan\theta_h\bar{u}(m_u p_l - m_d p_r)dh_3^+\right]+\frac{\sqrt{2}m_e}{v}\tan\theta_h\bar{\nu}p_reh_3^++\text{h.c . } ,
\label{yukawa1}\end{aligned}\ ] ] where @xmath150 is one element of the cabibbo - kobayashi - maskawa ( ckm ) matrix , sign@xmath151 and sign@xmath152 .
finally , we discuss the kinetic terms for the higgs fields @xmath153 where the covariant derivatives are @xmath154 the masses of the gauge bosons are obtained under the condition of @xmath47 as @xmath155 thus , the electroweak rho parameter @xmath156 is unity at the tree level .
one - loop corrections to @xmath0 have been calculated in ref .
@xcite for the gm model .
the deviation of @xmath0 from unity depends on the logarithm of the triplet - like higgs boson masses and , therefore , the one - loop effect is not important in this model .
the gauge - gauge - scalar ( gauge - scalar - scalar ) vertices are listed in table [ ggs ] ( table [ gss ] ) in appendix b. we note that there is the @xmath157 vertex at the tree level in the gm model ( see table [ ggs ] ) . in the higgs - extended models with @xmath158 at the tree level and having singly - charged higgs bosons ( _ e.g. _ , the 2hdm ) ,
the @xmath159 vertex is absent at the tree level @xcite and can only be induced at loop levels . therefore , the magnitude of this vertex in such models is much smaller than that in the gm model .
thus , this vertex can be used to discriminate models with singly - charged higgs bosons .
the possibility of measuring the @xmath159 vertex has been discussed in refs .
@xcite for the lhc and in ref .
@xcite for future linear colliders .
in this section , we discuss constraints on the parameter space of the gm model .
first , we consider the theoretical constraints from perturbative unitarity and vacuum stability . secondly , as experimental constraints
, we consider the @xmath160 data and other @xmath161 physics data .
-@xmath117 plane . in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively .
, title="fig:",width=188 ] -@xmath117 plane . in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively .
, title="fig:",width=188 ] -@xmath117 plane .
in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively . , title="fig:",width=188 ] -@xmath117 plane .
in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively .
, title="fig:",width=188 ] -@xmath117 plane . in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively .
, title="fig:",width=188 ] -@xmath117 plane . in all the plot , the uncolored regions are allowed , and the 3-plet higgs mass is taken to be 300 gev , @xmath162 mev and @xmath163 .
blue and pink shaded regions are respectively excluded by the vacuum stability bound and the unitarity bound .
the left , center and right plot show the case of @xmath164 , 200 gev and 230 gev , respectively .
, title="fig:",width=188 ] the perturbative unitarity bound for the gm model has been studied in ref .
@xcite and can be directly applied to our analysis . before doing so
, we will make a change in the parameterization .
this is because eqs .
( [ lambdas ] ) suggest apparent divergences in @xmath165 in the limit @xmath166 .
however , this is only an artefact that can be avoided by reparameterization
. we therefore select the following parameterization @xmath167 in terms of which all the dimensionless couplings can be rewritten for @xmath168 as @xmath169 it is seen that the @xmath17 dependence drops out in @xmath165 and no divergent @xmath170 s appear even when @xmath171 . for the vacuum stability condition , we require that the potential is bounded from below in any direction with large scalar fields
. this condition imposes constraints on the dimensionless coupling constants @xmath116 . in the gm model
, we then derive the following inequalities @xmath172 they have taken into account the positivity of all combinations of two non - zero scalar fields , as have been discussed in ref .
@xcite for the htm . fig .
[ theory_const1 ] shows the regions excluded by the unitarity and the vacuum stability constraints for the case of @xmath173 gev and @xmath174 mev .
the left , center and right plots show the cases for @xmath175=0 , 300 , 350 gev , respectively . for the unitarity bound
, we consider the @xmath176-wave amplitudes for elastic scatterings of two scalar boson states and require their absolute values of the eigenvalues to be less than 1 .
it is observed that the allowed regions by the unitarity bound for larger @xmath175 is smaller than those for smaller @xmath175 .
this is because the @xmath177 coupling increases as @xmath175 becomes larger .
in fact , the excluded regions are determined by the following unitarity condition @xcite ; @xmath178 on the other hand , the vacuum stability bound becomes milder as @xmath175 is taken to be a larger value because of the increasing @xmath177 coupling . for a fixed value of @xmath117 and @xmath175 ,
a larger @xmath179 value is allowed ( excluded ) by the unitarity ( vacuum stability ) bound .
[ theory_const2 ] also shows the regions excluded by the unitarity and the vacuum stability conditions for the case of @xmath180 gev and @xmath174 mev .
the allowed regions are much smaller than those in the case of @xmath173 gev .
the excluded regions from the vacuum stability for smaller ( larger ) values of @xmath179 are determined by the third ( fourth ) inequality in eq .
( [ vs ] ) .
in the case of larger @xmath17 values ( _ e.g. _ , @xmath181 gev ) , the regions excluded by the unitarity ( vacuum stability ) condition are larger ( smaller ) compared to the small @xmath17 case .
this is because the @xmath182 coupling becomes larger .
in addition , the singlet higgs boson mass gets a larger value , so that the regions excluded due to @xmath183 are smaller in the larger @xmath17 case .
the renormalized @xmath160 vertex is defined by @xcite @xmath184 where the renormalized coupling @xmath185 can be expressed as @xmath186 where @xmath187 ( @xmath188 ) denote the one - loop corrections to the @xmath160 vertices from the sm ( gm ) contributions , where the @xmath20 boson and the ng boson ( @xmath112 ) are running in the loop , @xmath189 ( @xmath190 ) is the third component of the isospin ( the electric charge ) for the field @xmath191 , and @xmath192 and @xmath193 .
the analytic formulas for @xmath187 is given in ref .
@xcite , and their numerical values are calculated as @xcite @xmath194 the one - loop correction @xmath195 is given in terms of the passarino - veltman function @xcite by @xmath196 .
\end{aligned}\ ] ] on the other hand , @xmath197 can be neglected because the corrections are proportional to the bottom quark mass [ see eq .
( [ yukawa1 ] ) ] . we can also neglect the contributions from @xmath113 loop diagrams for the same reason .
the renormalized couplings @xmath185 can be compared to the experimental value of @xmath198 @xcite @xmath199 data given in eq .
( [ rb_exp ] ) on @xmath17 as a function of @xmath118 .
the region above the black ( red ) line is excluded at 1@xmath200 ( 2@xmath200 ) level .
, width=302 ] in fig .
[ bound_zbb ] , we show the excluded parameter space in the @xmath118-@xmath17 plane using the @xmath201 data in eq .
( [ rb_exp ] ) . basically , the upper bound on @xmath17 increases monotonically with @xmath118 .
the 2@xmath200 bound is about 25 gev more relaxed than the 1@xmath200 bound over the considered range .
we note in passing that the constraint of the @xmath202 data for the gm model is similar to that in the type - i 2hdm @xcite and is milder than the @xmath201 constraint .
( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 . we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots .
, title="fig:",width=188 ] ( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 .
we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots .
, title="fig:",width=188 ] ( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 .
we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots . , title="fig:",width=188 ] + ( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 .
we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots .
, title="fig:",width=188 ] ( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 .
we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots .
, title="fig:",width=188 ] ( left ) , @xmath96 ( center ) and @xmath86 ( right ) as a function of @xmath17 .
we take @xmath173 gev , @xmath203=125 gev and @xmath204 in all the plots .
the mass difference @xmath26 is taken to be 0 for the upper three plots , and to be @xmath205 gev for the lower three plots .
, title="fig:",width=188 ] in this section , we discuss the decay of the triplet - like higgs bosons , namely the 5-plet higgs bosons @xmath1
( @xmath206 or @xmath86 ) , 3-plet higgs bosons @xmath2 ( @xmath207 or @xmath113 ) and the singlet higgs boson @xmath15 . decay branching ratios of the higgs bosons depend on the mass parameters @xmath117 , @xmath118 and @xmath119 , the vev of the triplet field @xmath17 , and the mixing angle @xmath110 .
the mass of the sm - like higgs boson @xmath5 is fixed at 125 gev . using the mass relation given in eq .
( [ m1_pred ] ) , we can treat @xmath119 as an dependent parameter determined by @xmath118 and @xmath117 .
hereafter , we take @xmath208 , @xmath118 and @xmath17 as the input parameters , and assume @xmath163 for simplicity .
once we apply the mass relation , there are three different patterns of masses for the triplet - like higgs bosons . in the case of @xmath209 ,
all the masses of the triplet - like higgs bosons are degenerate : @xmath210 , whereas in the case of @xmath211 ( @xmath212 ) , the mass spectrum is then @xmath213 ( @xmath214 ) .
first , we consider the decays of the 5-plet higgs bosons . in the case of @xmath215
, the 5-plet higgs bosons can decay into weak gauge boson pairs or lepton pairs depending on the magnitude of @xmath17 .
when @xmath216 , the 5-plet higgs bosons can decay into a 3-plet higgs boson and a gauge boson , such as @xmath217 and @xmath218 , in addition to the two decay modes allowed in the case of @xmath215 . in fig .
[ decay_150_vd_2 ] , the decay branching ratios of @xmath219 , @xmath96 and @xmath86 are shown as a function of @xmath17 in the case of @xmath173 gev . when @xmath220 ( upper row ) , the main decay modes of @xmath219 , @xmath96 and @xmath86 change from @xmath221 , @xmath222 , and @xmath223 to @xmath224 , @xmath225 , and @xmath226 or @xmath227 at around @xmath228 gev , respectively .
here @xmath86 decays more dominantly into @xmath226 than @xmath227 because of the mass threshold effect .
when @xmath229 gev ( lower row ) and for the wide range of @xmath230 gev , the main decay mode of @xmath219 is @xmath231 , those of @xmath96 are @xmath232 and @xmath233 , and those of @xmath86 are @xmath234 and @xmath235 . , but for @xmath180 gev.,title="fig:",width=188 ] , but for @xmath180 gev.,title="fig:",width=188 ] , but for @xmath180 gev.,title="fig:",width=188 ] + , but for @xmath180 gev.,title="fig:",width=188 ] , but for @xmath180 gev.,title="fig:",width=188 ] , but for @xmath180 gev.,title="fig:",width=188 ] fig .
[ decay_300_vd_2 ] shows the decay branching ratios of the 5-plet higgs bosons for @xmath180 gev .
the general behavior here is roughly the same as the case with @xmath173 gev .
the crossing point for the main decay modes in each of the upper plots ( @xmath209 ) slightly shifts to a smaller @xmath17 ( @xmath236 gev ) .
( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow .
, title="fig:",width=188 ] ( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow .
, title="fig:",width=188 ] ( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow .
, title="fig:",width=188 ] + ( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow .
, title="fig:",width=188 ] ( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow . ,
title="fig:",width=188 ] ( left column ) , @xmath96 ( center column ) and @xmath86 ( right column ) on the @xmath17-@xmath237 plane ( with @xmath216 ) .
we take @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) three plots show the case for @xmath173 gev ( 300 gev ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow .
, title="fig:",width=188 ] fig .
[ contour3 ] shows the contour plots of the decay branching ratios of @xmath219 , @xmath96 and @xmath86 on the @xmath17-@xmath237 plane ( with @xmath216 ) for the cases with @xmath173 gev ( upper plots ) and @xmath180 gev ( lower plots ) .
there are always three distinct regions in this plane . in the region of small @xmath237 and small ( large )
@xmath17 , the main decay modes of the 5-plet higgs bosons are the a pair of leptons ( weak bosons ) . in the region of large @xmath237 , they are a 3-plet higgs boson and a gauge boson , denoted by @xmath238 in the plots ( @xmath239 or @xmath24 ) , where it is understood that all the possible channels of @xmath238 should be summed over . secondly , we consider the decays of the 3-plet higgs bosons .
the 3-plet higgs bosons can decay into a pair of fermions through the yukawa interactions given in eq .
( [ yukawa1 ] ) and a pair of leptons through the neutrino yukawa interaction given in eq .
( [ ny ] ) , depending on the value of @xmath17 in the case of @xmath240 . in the region dominated by fermionic decays ,
the main decay mode strongly depends on @xmath118 .
when @xmath118 is smaller than the top quark mass , @xmath241 ( @xmath113 ) mainly decays into @xmath242 or @xmath243 ( @xmath244 ) , whereas in the case of @xmath245 , @xmath241 ( @xmath113 ) decays into @xmath246 ( @xmath244 ) .
furthermore , when @xmath118 is larger than @xmath247 , @xmath113 decays dominantly into @xmath248 , and @xmath241 still mainly into @xmath246 .
in addition , the 3-plet higgs bosons can decay into the sm - like higgs boson @xmath5 and a gauge boson , _
e.g. _ , @xmath249 and @xmath250 if @xmath118 is larger than @xmath203 .
when @xmath211 ( @xmath212 ) , the 3-plet higgs bosons can decay into a gauge boson and a 5-plet ( singlet ) higgs boson .
( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively . , title="fig:",width=226 ] ( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively . , title="fig:",width=226 ] + ( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively . , title="fig:",width=226 ] ( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively .
, title="fig:",width=226 ] + ( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively . , title="fig:",width=226 ] ( left column ) and @xmath113 ( right column ) as a function of @xmath17 .
we take @xmath173 gev , @xmath251 gev and @xmath163 in all the plots .
the mass difference @xmath26 is fixed to 0 ( top plots ) , 50 gev ( middle plots ) and @xmath205 gev ( bottom plots ) , respectively . , title="fig:",width=226 ] , but for @xmath180 gev .
, title="fig:",width=226 ] , but for @xmath180 gev .
, title="fig:",width=226 ] + , but for @xmath180 gev . ,
title="fig:",width=226 ] , but for @xmath180 gev .
, title="fig:",width=226 ] + , but for @xmath180 gev . ,
title="fig:",width=226 ] , but for @xmath180 gev .
, title="fig:",width=226 ] in fig .
[ decay_150_vd ] , the decay branching ratios of @xmath241 and @xmath113 are shown as a function of @xmath17 for @xmath173 gev .
the mass difference @xmath26 is taken to be 0 , 50 gev and @xmath205 gev in the top , middle and bottom plots , respectively . from the top two figures
, it is seen that the dominant decay modes of @xmath241 ( @xmath113 ) change from @xmath222 ( @xmath223 ) to @xmath242 or @xmath243 ( @xmath244 ) at around @xmath252 gev . in the case of @xmath253 gev ( middle plots ) and for a wide range @xmath254 gev , the 3-plet higgs bosons mainly decay into a 5-plet higgs boson and a weak gauge boson , _
i.e. _ , @xmath255 , @xmath256 and @xmath257 for @xmath241 decays and @xmath258 and @xmath259 for @xmath113 decays . on the other hand , in the case of @xmath260 gev ( bottom plots ) ,
the main decay modes of @xmath241 ( @xmath113 ) are @xmath261 ( @xmath262 ) in the range of @xmath263 gev .
[ decay_300_vd ] shows the decay branching ratios of @xmath241 and @xmath113 as a function of @xmath17 for @xmath180 gev .
the mass difference @xmath26 is taken to be 0 , 50 gev and @xmath205 gev in the top , middle and bottom plots , respectively .
when @xmath220 ( top plots ) , the main decay modes of @xmath241 ( @xmath113 ) change from @xmath222 ( @xmath223 ) to @xmath246 and @xmath264 ( @xmath265 ) at @xmath266 gev . when @xmath267 gev ( middle plots ) and @xmath268 gev ( bottom plots ) , the main decay modes are the same as in the case of @xmath173 gev in the range of @xmath269 gev .
( left column ) and @xmath113 ( right column ) on the @xmath17-@xmath26 plane .
we take @xmath173 gev , @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) two plots show the case with @xmath211 ( @xmath216 ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow.,title="fig:",width=283 ] ( left column ) and @xmath113 ( right column ) on the @xmath17-@xmath26 plane .
we take @xmath173 gev , @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) two plots show the case with @xmath211 ( @xmath216 ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow.,title="fig:",width=283 ] + ( left column ) and @xmath113 ( right column ) on the @xmath17-@xmath26 plane .
we take @xmath173 gev , @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) two plots show the case with @xmath211 ( @xmath216 ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow.,title="fig:",width=283 ] ( left column ) and @xmath113 ( right column ) on the @xmath17-@xmath26 plane .
we take @xmath173 gev , @xmath203=125 gev and @xmath163 in all the plots .
the upper ( lower ) two plots show the case with @xmath211 ( @xmath216 ) .
each solid ( dashed ) curve represents the branching ratio of 50% ( 90% ) for the corresponding decay mode indicated by the arrow.,title="fig:",width=283 ] , but for @xmath180 gev.,title="fig:",width=283 ] , but for @xmath180 gev.,title="fig:",width=283 ] + , but for @xmath180 gev.,title="fig:",width=283 ] , but for @xmath180 gev.,title="fig:",width=283 ] in fig .
[ contour1 ] , we give the contour plots of the decay branching ratios of @xmath241 and @xmath113 for @xmath173 gev . the mass difference @xmath26 is taken to be positive ( negative ) in the upper ( lower ) two figures . in this figure , br@xmath270 and br@xmath271 denote the sums of the decay branching ratios of the modes with a 5-plet higgs boson and a gauge boson .
br@xmath272 and br@xmath273 denote the sum of the decay branching ratios of @xmath274 and @xmath275 and that of @xmath276 and @xmath277 , respectively .
similar to fig .
[ contour3 ] , it is seen that there are three distinct regions in this plane . in the small @xmath26 and small ( large ) @xmath17 region ,
the main decay modes are @xmath222 ( @xmath242 ) for @xmath241 and @xmath223 ( @xmath244 ) for @xmath113 . on the other hand , in the large @xmath26 region , the decay modes associated with a 5-plet ( singlet ) higgs boson dominate in the case of @xmath278 ( @xmath216 ) .
we notice that the regions where the @xmath279 decay is dominant are wider than the corresponding one where the @xmath280 decay is dominant .
this is because of a larger number of decay modes in @xmath279 .
[ contour2 ] shows the contour plots of the branching ratios of the 3-plet higgs bosons for @xmath180 gev . in the plots of the left column
, there is no dashed curve corresponding to the branching ratio of 90% for the @xmath281 decay mode .
this is because the @xmath282 decay mode is also kinematically allowed at the same time when the @xmath281 is open , and the former amounts to around 30% .
we here comment on the decays of the singlet higgs boson @xmath15 .
when we take @xmath163 , the decay property of @xmath15 is similar to that of @xmath86 . in the case of @xmath283
, @xmath15 can decay into @xmath223 ( @xmath226 or @xmath227 ) for smaller ( larger ) values of @xmath17 .
when @xmath284 , @xmath15 can decay into a 3-plet higgs boson and a weak gauge boson .
if @xmath285 , @xmath15 can mix with @xmath5 and can thus decay into fermion pairs via the mixing in addition to the above - mentioned modes .
-@xmath237 plane .
, width=377 ] throughout this section , the decay properties of the 5-plet higgs bosons and the 3-plet higgs bosons can be separately considered for four different regions in the @xmath17-@xmath26 plane , as schematically shown in fig .
[ schematic ] . in region
i , all the triplet - like higgs bosons mainly decay leptonically : @xmath286 in this region , the mass of the 5-plet higgs bosons is constrained to be @xmath287 gev by the search at the lhc for doubly - charged higgs bosons decaying into same - sign dileptons @xcite . in region ii , the 5-plet higgs bosons mainly decay into the weak gauge boson pairs , while the 3-plet higgs bosons decay into the fermion pairs .
when the mass of the 3-plet higgs bosons is less than the top quark mass , the main decay modes are @xmath288 for region iii and region iv , one has to separately consider the cases whether the sign of @xmath26 is positive or negative . in the case of @xmath284 , the 5-plet higgs bosons mainly decay into the lepton pairs ( weak gauge boson pairs ) in region iii ( region iv ) .
the 3-plet higgs bosons mainly decay into a 5-plet higgs boson and a weak gauge boson : @xmath289 in the case of @xmath212 , the main decay modes in both region iii and region iv are @xmath290
in this section , we discuss how the custodial symmetry of the gm model can be tested at the lhc .
there are characteristic features of the triplet - like higgs bosons that mainly originate from the triplet field @xmath10 of the model .
( 1 ) the masses of the higgs bosons belonging to the same @xmath4 multiplet are the same .
( 2 ) the 5-plet and the singlet higgs bosons have the gauge - gauge - scalar type of couplings as listed in table [ ggs ] , but not the yukawa couplings given in eq . ( [ yukawa1 ] ) , while the 3-plet higgs bosons have the yukawa couplings , but not the gauge - gauge - scalar type of couplings .
these features can be used to test the custodial symmetry of the gm model . as is discussed in the previous section ,
it is important to study the decay pattern of the triplet - like higgs bosons . in particular , feature ( 2 ) mentioned above can be most clearly tested in region ii because the 5-plet ( 3-plet ) higgs bosons mainly decay into weak gauge boson pairs ( fermion pairs ) in this region . in the following discussion ,
we focus on region ii and the detectability of the 5-plet and 3-plet higgs bosons . .
the upper - left , upper - right , and bottom plots show the production cross sections for @xmath16 , @xmath291 , and @xmath86 , respectively . in all the plots , we take @xmath292 gev and @xmath220 . the lhc collision energy is assumed to be 8 tev ( dashed curves ) and 14 tev ( solid curves ) .
, title="fig:",width=302 ] .
the upper - left , upper - right , and bottom plots show the production cross sections for @xmath16 , @xmath291 , and @xmath86 , respectively . in all the plots , we take @xmath292 gev and @xmath220 . the lhc collision energy is assumed to be 8 tev ( dashed curves ) and 14 tev ( solid curves ) .
, title="fig:",width=302 ] + .
the upper - left , upper - right , and bottom plots show the production cross sections for @xmath16 , @xmath291 , and @xmath86 , respectively . in all the plots , we take @xmath292 gev and @xmath220 . the lhc collision energy is assumed to be 8 tev ( dashed curves ) and 14 tev ( solid curves ) .
, title="fig:",width=302 ] there are several production modes for the 5-plet higgs bosons @xmath1 and the 3-plet higgs bosons @xmath2 , as listed below . throughout this section , @xmath293 and
those with bars denote light quarks and anti - quarks . 1 .
: : * the drell - yan process * + @xmath1 and @xmath2 can be produced in pairs via @xmath294 and @xmath24 , _
e.g. _ , @xmath295 and @xmath296 .
the cross section is determined by the gauge coupling as well as the higgs masses @xmath117 and @xmath118 , independent of the value of @xmath17 .
2 . : : * the mixed drell - yan ( mdy ) process * + @xmath1 and @xmath2 can be produced at the same time , _ e.g. _ , @xmath297 , which we call the mixed drell - yan ( mdy ) process to be separated from the usual drell - yan process mentioned above .
the cross section is proportional to @xmath298 , and is thus relatively suppressed in comparison with the drell - yan process , especially in the large @xmath17 case .
: : * the weak vector boson fusion ( vbf ) process * + the single production of @xmath1 occurs via the @xmath299 process .
the cross section is proportional to @xmath300 , so that this mode can be important in the large @xmath17 case .
4 . : : * the weak vector boson associated process * + in addition to the vbf process , @xmath1 can also be produced in association with a weak gauge boson , _
e.g. _ , @xmath301 .
the cross sections of such modes are proportional to @xmath300 as for the vbf production mode .
thus , this mode can also become important when the vbf process is important .
5 . : : * the yukawa process * + @xmath2 can be produced through the yukawa interactions given in eq .
( [ yukawa1 ] ) as the gluon fusion process for the sm higgs boson : @xmath302 .
there are t - channel @xmath112 and @xmath113 production modes : @xmath303 and @xmath304 .
these production cross sections are proportional to @xmath305 .
: : * the top quark decay * + when @xmath118 is smaller than the top quark mass , @xmath112 can be produced from the top quark decay .
the decay rate of the @xmath306 depends on @xmath305 . among these production processes , channels
3 and 4 can be useful to discriminate the gm model from the others with doubly - charged higgs bosons and to test the mass degeneracy of @xmath1 . in the htm , for example , the doubly - charged higgs boson can in principle be produced via the vbf and the vector boson associated processes .
however , these cross sections are much suppressed due to the tiny triplet vev required by the electroweak rho parameter . in the gm model ,
@xmath17 can be of order 10 gev , so that these production processes become useful .
the mdy process is also a unique feature of the gm model because the higgs bosons @xmath1 and @xmath2 having different decay properties are produced at the same time . in particular ,
when region ii is realized , the main decay modes of these two higgs bosons are distinctly different .
thus , this process can be useful not only to test the mass degeneracy of @xmath2 but also to distinguish the model from the others also having doubly - charged and/or singly - charged higgs bosons . in fig .
[ cross ] , production cross sections of the 5-plet higgs bosons from channels 2 , 3 , and 4 are shown as a function of @xmath117 for the lhc running at 8 and 14 tev .
we take @xmath307 gev and @xmath308 as an example in all the plots .
it is noted that the dominant production mechanism is the vbf process for a sufficiently large @xmath117 .
we will first discuss the vbf process and the vector boson associated process to study the mass degeneracy among @xmath16 , @xmath85 and @xmath86 .
then we turn to the mdy process .
let us consider the case with @xmath173 gev , @xmath309 gev ( _ i.e. _ , @xmath310 gev ) and @xmath311 gev as an example in region ii . in this case , the 5-plet higgs bosons decay into gauge boson pairs almost 100% ( the branching fractions of @xmath312 and @xmath313 being 67% and 33% , respectively ) .
on the other hand , @xmath112 decays to @xmath314 at 66% and @xmath315 at 29% , and @xmath113 decays to @xmath244 at 89% .
we note that the branching fraction of @xmath316 here is around 0.4% .
the upper limit of the top quark decay into a charged higgs boson and the bottom quark is 2 - 3% in the case where the charged higgs boson mass is between 80 and 160 gev , under the assumption that the charged higgs boson decays to @xmath317 at 100% @xcite .
thus , the selected parameter set is allowed by the constraint from the top quark decays .
the signal events from the vbf production processes for the 5-plet higgs bosons are given by @xmath318 from the vector boson associated processes , we have the following events @xmath319 where the associated weak gauge bosons are assumed to decay hadronically so that they have the same final states as the vbf process .
moreover , we consider the case where the weak gauge bosons produced from the decay of @xmath1 decay leptonically .
then the final states of the signal events have same - sign ( ss ) dileptons plus dijets and missing transverse energy ( @xmath320 ) for the @xmath16 production mode , where @xmath321 denotes collectively the light leptons @xmath322 and @xmath323 hereafter .
the final state of the @xmath291 production mode includes trileptons plus dijets and missing transverse energy ( @xmath324 ) , while that for the @xmath86 production mode has opposite - sign ( os ) dileptons plus dijets and missing transverse energy ( @xmath325 ) .
the corresponding background events for these signal events are from the @xmath326 for the @xmath16 production , @xmath327 for the @xmath291 production , and @xmath248 , @xmath328 and @xmath329 for the @xmath330 production .
we simulate the signal and the background event rates by using madgraph 5 @xcite at the parton level for the cases where the lhc operates at the center - of - mass ( cm ) energy @xmath331 of 8 tev and 14 tev .
we impose the following basic kinematic cuts @xmath332 where @xmath333 and @xmath334 are the transverse momenta of the jet and the lepton , respectively , @xmath335 and @xmath336 are the pseudorapidities of the jet and the lepton , respectively , and @xmath337 is the distance between the two jets .
the cross sections for the signal and background events are listed in table [ sig_and_bg ] , where the signal cross section includes contributions from the vbf production and the vector boson associated production .
an integrated luminosity of 100 fb@xmath338 is assumed in the simulations . in this table ,
the signal significance is defined by @xmath339 where @xmath176 and @xmath161 are the numbers of the signal and background events , respectively .
the significance of the @xmath320 event from the @xmath16 production process exceeds 5 even using simply the basic cuts .
however , the significances for the remaining two events from the @xmath85 and @xmath86 production processes are less than 1 . for the @xmath325 event , in particular , the background is larger than the signal by 3 to 4 orders of the magnitude because of the huge @xmath248 background .
.signal and background cross sections in units of fb after each kinematic cut , along with the significance @xmath340 defined by eq .
( [ significance ] ) based on an integrated luminosity of 100 fb@xmath338 .
the numbers without ( with ) parentheses correspond to the case with a cm energy of @xmath341 tev ( 14 tev ) .
the signal cross section includes contributions from both the vbf production and the vactor boson associated production processes . for the @xmath325 events
, we further impose the requirement of the @xmath342-jet veto for each jet to reduce the background , where the @xmath342-tagging efficiency is take to be 0.6 @xcite . [ cols="^,^,^,^,^,^,^,^,^,^ " , ]
the loop functions appearing in the calculations of the sm - like higgs boson decay to diphotons are given in terms of the passarino - veltman functions @xcite as @xmath343,\\ i_{1/2}&= -4m_f^2 \left[\frac{2}{m_h^2}-\left(1-\frac{4m_f^2}{m_h^2}\right)c_0(0,0,m_h^2,m_f , m_f , m_f)\right],\\ i_1&= 2m_w^2\left[\frac{6}{m_h^2}+\frac{1}{m_w^2}+6\left(\frac{2m_w^2}{m_h^2}-1\right)c_0(0,0,m_h^2,m_w , m_w , m_w)\right ] .
\label{i_func}\end{aligned}\ ] ] those for the @xmath7 process are given by @xmath344,\\ j_{1/2}= & -\frac{4m_f^2(\frac{1}{2}i_f - s_w^2q_f)}{s_wc_w(m_h^2-m_z^2)}\big[2+(4m_f^2-m_h^2+m_z^2)c_0(0,m_z^2,m_h^2,m_f , m_f , m_f)\notag\\ & + \frac{2m_z^2}{m_h^2-m_z^2}(b_0(m_h^2,m_f , m_f)-b_0(m_z^2,m_f , m_f))\big],\\ j_1= & \frac{2m_w^2}{s_wc_w(m_h^2-m_z^2)}\bigg\{\left[c_w^2\left(5+\frac{m_h^2}{2m_w^2}\right)-s_w^2\left(1+\frac{m_h^2}{2m_w^2}\right)\right]\notag\\ & \times\left[1 + 2m_w^2c_0(0,m_z^2,m_h^2,m_w , m_w , m_w)+\frac{m_z^2}{m_h^2-m_w^2}(b_0(m_h^2,m_w , m_w)-b_0(m_z^2,m_w , m_w))\right]\notag\\ & -6c_w^2(m_h^2-m_z^2)c_0(0,m_z^2,m_h^2,m_w , m_w , m_w)+2s_w^2(m_h^2-m_z^2)c_0(0,m_z^2,m_h^2,m_w , m_w , m_w)\bigg\}. \label{j_func}\end{aligned}\ ] ] t. d. lee , phys .
d * 8 * , 1226 ( 1973 ) . j. c. pati and a. salam , phys .
d * 10 * , 275 ( 1974 ) r. n. mohapatra and j. c. pati , phys .
d * 11 * , 2558 ( 1975 ) ; g. senjanovic and r. n. mohapatra , phys .
d * 12 * , 1502 ( 1975 ) .
h. georgi and m. machacek , nucl .
b * 262 * , 463 ( 1985 ) .
t. p. cheng and l. f. li , phys .
d * 22 * , 2860 ( 1980 ) ; j. schechter and j. w. f. valle , phys . rev .
d * 22 * , 2227 ( 1980 ) ; g. lazarides , q. shafi and c. wetterich , nucl .
b * 181 * , 287 ( 1981 ) ; r. n. mohapatra and g. senjanovic , phys . rev .
d * 23 * , 165 ( 1981 ) ; m. magg and c. wetterich , phys .
b * 94 * , 61 ( 1980 ) .
m. s. chanowitz and m. golden , phys .
b * 165 * , 105 ( 1985 ) .
t. han , b. mukhopadhyaya , z. si and k. wang , phys .
d * 76 * , 075013 ( 2007 ) ; a. g. akeroyd , m. aoki and h. sugiyama , phys .
d * 77 * , 075010 ( 2008 ) ; p.
fileviez perez , t. han , g. -y .
huang , t. li and k. wang , phys .
d * 78 * , 015018 ( 2008 ) ; a. g. akeroyd and c. -w .
chiang , phys .
d * 80 * , 113010 ( 2009 ) ; a. g. akeroyd and c. -w .
chiang , phys .
d * 81 * , 115007 ( 2010 ) ; a. g. akeroyd , c. -w .
chiang and n. gaur , jhep * 1011 * , 005 ( 2010 ) .
d. acosta _ et al .
_ [ cdf collaboration ] , phys .
lett . * 93 * , 221802 ( 2004 ) ; v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 93 * , 141801 ( 2004 ) ; v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 101 * , 071803 ( 2008 ) ; t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
* 101 * , 121801 ( 2008 ) .
g. aad _ et al . _ [ atlas collaboration ] , phys .
d * 85 * , 032004 ( 2012 ) ; s. chatrchyan _ et al . _
[ cms collaboration ] .
j. f. gunion , r. vega and j. wudka , phys .
d * 42 * , 1673 ( 1990 ) . c. -w .
chiang , t. nomura and k. tsumura , phys .
d * 85 * , 095023 ( 2012 ) .
s. chakrabarti , d. choudhury , r. m. godbole and b. mukhopadhyaya , phys .
b * 434 * , 347 ( 1998 ) ; a. melfo , m. nemevsek , f. nesti , g. senjanovic and y. zhang , phys .
d * 85 * , 055018 ( 2012 ) ; m. aoki , s. kanemura and k. yagyu , phys .
d * 85 * , 055007 ( 2012 ) .
j. f. gunion , r. vega and j. wudka , phys .
d * 43 * , 2322 ( 1991 ) . j. a. grifols and a. mendez , phys .
d * 22 * , 1725 ( 1980 ) .
e. asakawa , s. kanemura , phys .
* b626 * , 111 - 119 ( 2005 ) ; e. asakawa , s. kanemura , j. kanzaki , phys .
* d75 * , 075022 ( 2007 ) ; m. battaglia , a. ferrari , a. kiiskinen , t. maki , [ hep - ex/0112015 ] ; s. godfrey , k. moats , phys
. rev .
* d81 * , 075026 ( 2010 ) .
a. arhrib , r. benbrik , m. chabab , g. moultaka , m. c. peyranere , l. rahili and j. ramadan , phys .
d * 84 * , 095005 ( 2011 ) .
h. e. haber and h. e. logan , phys .
d * 62 * , 015011 ( 2000 ) .
b. w. lynn and r. g. stuart , phys .
b * 252 * , 676 ( 1990 ) .
g. passarino and m. j. g. veltman , nucl .
b * 160 * , 151 ( 1979 ) .
v. d. barger , j. l. hewett and r. j. n. phillips , phys .
d * 41 * , 3421 ( 1990 ) .
m. aoki , s. kanemura , k. tsumura and k. yagyu , phys .
d * 80 * , 015017 ( 2009 ) .
s. chatrchyan _ et al . _
[ cms collaboration ] , jhep * 1207 * , 143 ( 2012 ) .
j. alwall , m. herquet , f. maltoni , o. mattelaer and t. stelzer , jhep * 1106 * , 128 ( 2011 ) . v. d. barger , a. d. martin and r. j. n. phillips , phys
. lett .
b * 125 * , 339 ( 1983 ) ; v. d. barger , t. han and j. ohnemus , phys .
d * 37 * , 1174 ( 1988 ) ; j. bagger , v. d. barger , k. -m .
cheung , j. f. gunion , t. han , g. a. ladinsky , r. rosenfeld and c. -p .
yuan , phys .
d * 52 * , 3878 ( 1995 ) .
the atlas collaboration , calibrating the b - tag efficiency and mistag rate in 35 pb1 of data with the atlas detector , atlas - conf-2011 - 089 , june , 2011 .
j. s. gainer , w. -y .
keung , i. low and p. schwaller , phys .
d * 86 * , 033010 ( 2012 ) .
m. carena , i. low and c. e. m. wagner , jhep * 1208 * , 060 ( 2012 ) . c. -w .
chiang , and k. yagyu , arxiv:1207.1065 [ hep - ph ] . | we study how the custodial symmetry in the higgs sector of the georgi - machacek ( gm ) model can be tested at the lhc . as the minimal extension of the higgs triplet model , in which tiny neutrino masses are generated via the type - ii seesaw mechanism , the gm model keeps the electroweak @xmath0 parameter at unity at tree level . in the gm model , there are 5-plet ( @xmath1 ) , 3-plet ( @xmath2 ) and singlet ( @xmath3 ) higgs bosons under the classification of the custodial @xmath4 symmetry , in addition to the standard model - like higgs boson ( @xmath5 ) .
these new higgs bosons have the following characteristic features at the tree level : ( 1 ) the masses of the higgs bosons belonging to the same @xmath4 multiplet are degenerate ; and ( 2 ) @xmath1 and @xmath3 couple to the electroweak gauge bosons but not sm quarks , whereas @xmath2 couples to the quarks but not the gauge bosons .
we find that the @xmath1 production from the weak vector boson fusion process and the drell - yan process associated with @xmath2 are useful in testing the custodial symmetry of the higgs sector at the lhc .
in addition , these processes can also be used to discriminate from other models that contain singly - charged higgs bosons and extra neutral higgs bosons .
we also investigate a possible enhancement in the @xmath6 as well as @xmath7 decays . |
since the demonstration of the isolation of a single atomic plane of graphite ( graphene ) on a standard @xmath3 substrate @xcite , it has suddenly become possible to experimentally verify many theoretical predictions about the peculiar physical behavior of graphene - based systems .
indeed , the linear @xmath4 dispersion curves , and the consequent relativistic - like behavior of charge carriers in monolayer graphene , have been broadly confirmed by several experimental observations @xcite .
interestingly , stacking a number @xmath0 of graphene layers on top of each other can lead to new physical systems exhibiting completely different properties .
for instance , some peculiar gate - tunable electronic and optical properties have been recently reported in the bilayer case @xcite .
in fact , the interlayer coupling induces a gradual departure from the electronic bands of monolayer graphene @xcite , until the bulk limit ( graphite ) is reached for @xmath0 large enough .
therefore , there is a range of @xmath0 where the physical properties of graphene stacks are sensitive even to a variation @xmath5 .
for instance , a qualitative difference between the transport properties of bilayer and trilayer graphene was evidenced in recent experiments @xcite . moreover ,
the optical absorption spectra are predicted to systematically vary with the layers number within the effective mass approximation for @xmath6 @xcite , reflecting the gradual modification of the band structure .
here we show that a clear splitting of the raman @xmath2 band is observed for @xmath7 and @xmath8 , when the multilayer graphene symmetry is broken by heavy doping of the top layer , whereas the splitting is not observed for @xmath9 and @xmath10 .
the presence or absence of splitting highlights different electron - phonon interactions , which are influenced by doping and symmetry breaking in distinct ways in the various @xmath0-layer graphenes .
the reported results confirm two experimental reports in literature about the @xmath2 band splitting in bilayer graphene ( obtained by gate field effect ) @xcite , adding new important information such as a systematic study of the splitting as a function of the layers number and of the doping .
raman spectroscopy is a very powerful tool for studying graphene , yielding information on the electronic structure and on the electron - phonon coupling ( epc ) in the material @xcite .
in fact , this technique allows to clearly distinguish a monolayer and a bilayer from a few - layer graphene , through the analysis of the @xmath11 band @xcite , and to estimate the charge carrier density and type in monolayer graphene from the spectral positions and relative intensities of the @xmath2 and @xmath11 bands @xcite .
the @xmath2 band ( at @xmath12 @xmath13 ) is due to a first - order raman scattering process involving zone - center in - plane optical phonons . in stacked graphene layers
, the vibrations in different atomic planes can combine with each other in various ways , depending on the number of layers and on the stacking order .
moreover , the epc is affected by the number and symmetry of the stacked layers , so that the @xmath2 band may be used to study the effects of symmetry breaking and doping on the electronic and phononic properties of multilayer graphenes .
the knowledge of these effects is of fundamental importance for the development of graphene - based field - effect electronic devices .
graphene layers studied in this work were deposited on a 285 nm thick @xmath14 on @xmath15 substrates , by adhesive tape exfoliation of natural graphite .
then , the samples were analyzed by optical microscopy , in order to estimate the number of graphene layers composing the deposited thin flakes .
it can be seen in fig .
[ f1 ] that the contrast ( defined as @xmath16 , where @xmath17 and @xmath18 are the reflected light intensities from the @xmath19 substrate with and without graphene , respectively ) measured on many semi - transparent flakes increases in a stepwise manner .
the analysis of the @xmath11 raman band confirmed that the lowest two steps correspond to monolayer and bilayer graphenes , indicating that few graphene layers can be counted by contrast analysis .
such a behavior is related to the optical absorption of graphene , which was found to be directly proportional to the number of layers for @xmath0 small enough @xcite . using appropriate filters in order to select the wavelengths at which the contrast variation is high
, we were able to distinguish up to 6 layers .
moreover , the experimental contrast values were checked by theoretical calculations within the fresnel coefficients approach @xcite .
= 550 nm , are compared to theoretical values . ] after a preliminary annealing in vacuum ( 1x10@xmath20 mbar for two days ) in order to remove possible adsorbed impurities from the graphene surface , the first run of raman measurements was carried out .
raman spectra were acquired by means of a jobin - yvon u1000 raman spectrometer equipped with a microscope ( 100x objective ) and with an ar - kr laser , using the excitation wavelength @xmath21 = 514.5 nm .
the incident laser power focused on the sample was adjusted to be less than 5 mw , to avoid any local heating effect .
various @xmath0-layer graphene flakes were analyzed , displaying the standard @xmath2 and @xmath11 bands reported in literature.the @xmath22 band at @xmath23 1350 @xmath13 , related to lattice defects , was never observed in the experiments , confirming the good quality of our graphene samples .
moreover , the analysis of the monolayer spectra ( @xmath2 peak at @xmath23 1582 @xmath13 with fwhm @xmath23 13 @xmath13 ) indicates that the unintentional doping level in the pristine samples was relatively low ( about 1x10@xmath24 @xmath25 ) @xcite .
then , the samples were subjected to a radio frequency ( rf ) plasma treatment in chf@xmath26 gas , and immediately ( within a few minutes ) re - characterized by raman spectroscopy .
previous studies reported in literature @xcite have shown that the following radical species can be found in a chf@xmath26 plasma : @xmath27 atoms and @xmath28 ( @xmath29 ) radicals . in dry etching processes ,
@xmath27 atoms normally act as the reactive species ( responsible for the etching ) , whereas @xmath28 radicals are passivation precursors giving rise to polymer deposition .
it has been observed that , in the case of chf@xmath26 plasma processes , the etch rate decreases with increasing flow ( in contrast with what happens in the case of cf@xmath30 plasma ) , due to the low concentration ratio @xmath31/[cf_{x}]$ ] @xcite .
this means that the action of @xmath27 atoms becomes more effective with decreasing the flow rate , because the passivating action of @xmath28 radicals is reduced . in our experiments ,
we have observed an increasing modification of the graphene raman spectra , with decreasing the flow rate in the plasma treatment .
in particular , we performed a preliminary study on monolayer graphene processed at various chf@xmath26 flow rates .
[ f2 ] shows that the plasma treatment induces a blue shift of both @xmath2 and @xmath11 peaks , which increases with decreasing the gas flow .
moreover , below a flow rate threshold ( about 6 sccm ) , two new peaks arise at about 1350 @xmath13 ( @xmath22 peak ) and 1620 @xmath13 ( @xmath32 peak ) , which indicate the presence of defects in the @xmath33 @xmath34 lattice .
also the effect on the raman spectra of @xmath0-layer graphenes , which is going to be discussed in detail in the following section , was remarkably reduced with increasing the gas flow rate .
flow rates ] in analogy with the mechanism involved in the etching processes , the interaction of @xmath27 atoms with the graphene surface is likely reduced at high flow rate , because of the passivating action of the @xmath28 radicals .
therefore , the effect of chf@xmath26 plasma on graphene raman spectra can be ascribed to the adsorption of @xmath27 atoms on the surface . with increasing the @xmath27 coverage
, the modification of the graphene properties becomes more important , passing from a p - type doping effect ( blue - shifted @xmath2 and @xmath11 peaks ) @xcite to a structural modification ( @xmath22 and @xmath32 peaks ) at a very low flow rate ( less than 6 sccm ) .
the doping effect can be ascribed to an electron transfer from graphene to adsorbed @xmath27 atoms , i.e. a mechanism analogous to the observed charge transfer between graphene and adsorbed @xmath35 atoms @xcite . at very low gas flow ,
when the passivating action of the @xmath28 radicals is minimized , chemical modification ( fluorination ) of graphene may eventually occur , with a transition from @xmath33 to @xmath36 @xmath34 hybridization similar to that observed in graphane formation by plasma hydrogenation @xcite
. this evolution may be analogous to the transition from semi - ionic to covalent @xmath37 bonding observed in carbon nanotubes treated in @xmath38 plasma @xcite . here
, we discuss the results of processes carried out at a gas flow of 6 sccm ( pressure of 100 mtorr ) for 5 minutes , at rf power = 15 w. such experimental conditions lead to a very high doping , without structural modification of the graphenes .
moreover , the symmetry of stacked graphenes is broken by the dipole moment generated by the charge transfer from graphene to the adsorbed @xmath27 atoms .
therefore , the situation under study is very similar to that found in field - effect experiments , and the results here reported may be useful for the study of gated graphene - based devices .
we focus now on the effect of top doping on the raman @xmath2 band , which displays very distinct features depending on the number @xmath0 of stacked graphene layers .
the change of the @xmath2 band induced by plasma treatment in the various cases is visible in fig .
[ f3 ] , where the spectra at @xmath39 and @xmath40 were acquired on the same substrate before and immediately after the treatment , respectively .
the @xmath2 peak of monolayer graphene is largely blue - shifted ( to @xmath23 1590 @xmath13 ) and narrowed ( fwhm @xmath23 6 @xmath13 ) .
both the observations are consistent with an increase of the doping level , which induces a hardening of the mode , due to the non - adiabatic removal of a kohn anomaly for zone center optical phonons @xcite , and a reduction of the linewidth , due to pauli exclusion principle which inhibits phonon decay into electron - hole pairs when the fermi level surpasses half the phonon energy @xcite . on the other hand , the bilayer and 4-layer spectra display a very evident splitting of the @xmath2-mode , whereas the behavior of trilayer spectrum is similar to that of monolayer .
these results have been confirmed on different flakes on the same sample and on different samples .
furthermore , we observed that the modification induced by the plasma treatment was not stable under ambient conditions , as the raman spectra changed with the passing of time , slowly tending to their pristine form ( fig .
indeed , the initial condition can be restored by a vacuum annealing , so that the plasma treatment can be repeated for several times in a reproducible way .
such a reversible behavior is consistent with the absence of the @xmath22 peak , which indicates the lack of structural modification of the material ( in contrast with the case of the chemical modification obtained by plasma treatment at very low gas flow ) .
as previously discussed , the monolayer spectra , showing the hardening of both the @xmath2-mode and the @xmath11 mode , tell us that a strong @xmath41-type doping is achieved @xcite upon plasma processing .
therefore , we were able to gradually vary the doping level on top of each @xmath0-layer graphene , studying the effect for different values of @xmath0 .
the effect of charged adsorbates has already been studied on epitaxial bilayer graphene on sic by arpes measurements @xcite .
it was shown that the electronic bands of bilayer graphene are strongly affected by a potassium atoms coverage on the top surface , due to the @xmath42-type doping induced by the adsorbates . indeed ,
because of the short screening length along the @xmath43-axis of stacked graphenes @xcite , the majority of the doping charge is localized in the top layer and a dipole moment is formed , giving rise to a symmetry breaking and to the consequent band gap opening @xcite . in our case ,
an analogous effect is likely to occur : the doping charge density rapidly decreases going from the top layer to the bottom , thus altering the symmetry of the multilayers .
the symmetry breaking can affect very much the physical properties of the system , as it eliminates the symmetry constraints which rule both phonon raman activity and epc .
in these conditions , all phonon modes included in the @xmath2 band can become raman active and strongly mixed with each other @xcite , and the epc is affected by the modification of the electronic bands . in fig .
[ f3 ] , the @xmath2 band dependence on the doping is visible in all cases , but it looks different for different @xmath0 values .
an estimation of the doping level may be obtained from monolayer graphene spectra , basing on the @xmath2 peak position and linewidth , and on the intensity ratio of the @xmath2 and @xmath11 peaks @xcite , but we have verified that different monolayers can display slightly different @xmath2 peak positions on the same sample , due to the difficulty of controlling the native doping of graphene in ambient atmosphere @xcite . furthermore , monolayer graphene is likely to have a different reactivity with respect to that of few - layer graphene , as experimentally observed in the case of hydrogenation @xcite . also in our experiments , we have seen that , after plasma treatment at very low flow rate , the @xmath22 peak is hardly observed in the few - layers spectra , whereas the monolayer spectrum displays a very evident @xmath22 peak .
therefore , it seems not correct to extrapolate the doping values from the monolayer analysis to interpret the few - layer spectra .
however , although a precise quantitative estimate of carrier concentration in each case is not possible in our experiments , we can monitor the different behavior of @xmath0-layer graphene raman spectra with decreasing the top doping , starting from high doping levels ( more than @xmath44 ) as suggested by the analysis of monolayers spectra . the effect in the bilayer
can be interpreted on the basis of some recent literature .
indeed , the splitting of the raman @xmath2 band was recently observed in gated bilayer graphene @xcite , and ascribed to the inversion symmetry breaking induced by the gate field effect and to two distinct epc involved in the @xmath2 band .
this is due to the fact that the @xmath2 band of bilayer graphene includes two doubly degenerate modes , @xmath45 and @xmath46 , which are symmetric and antisymmetric with respect to the inversion symmetry , respectively .
consequently , only the @xmath45 mode is normally raman active , unless inversion symmetry breaking switches the @xmath46 mode on , too .
it has been shown by ando and koshino @xcite that , in the presence of asymmetry in the potential of the two stacked layers , symmetric and antisymmetric modes are strongly mixed with each other and two peaks appear in the raman spectrum .
moreover , phonons can be considerably modified by resonant electronic interband transitions , when the asymmetry opens up a gap comparable to the phonon energy .
ab initio calculations have been performed to compute the @xmath2 band of bilayer graphene in asymmetric conditions as a function of the carrier concentration in the top and bottom layer ( @xmath47 and @xmath48 ) @xcite , predicting the behavior of the two modes which become raman visible at certain values of @xmath47 and @xmath48 . in fig .
[ f4 ] , we report the analysis of the two lorentzian peaks which can fit the bilayer @xmath2 band at various stages of the plasma modification , for two different samples .
the experimental evolution of the peaks position has been fitted by the theoretical curves obtained from ref .
@xcite , assuming that about the @xmath49 of the total charge carriers is confined in the top layer , according to ref .
@xcite , and using the total carrier concentration @xmath42 as the free parameter in the fitting .
then , the other spectral features ( intensity and linewidth ) have been compared to the behavior predicted in ref .
@xcite , considering the carrier concentration values obtained from the best fit of the peaks position curve .
it can be seen that a good qualitative agreement is obtained for all the analyzed parameters .
in particular , the threshold at about @xmath50 , after which a steep variation of both the intensity and linewidth ratios is predicted by theory , is well reproduced by the experimental data , thus confirming the consistence of the carrier concentration results obtained by fitting the peaks position curves .
therefore , the behavior of the bilayer @xmath2 band is well interpreted in the framework of an asymmetric carrier distribution model .
band as a function of doping .
circles and triangles are experimental data from two different samples , and lines are the theoretical findings of ref . @xcite . in ( a ) ,
the peaks positions have been fitted by the theoretical curves , assuming that about the @xmath49 of the total charge carriers is confined in the top layer . in ( b ) and ( c ) , the width and intensity ratios of the low ( l ) frequency and high ( h ) frequency peak are reported . ] in order to interpret the spectra for @xmath51 , it is worth to consider , first of all , the evolution of the @xmath2 band with @xmath0 as predicted by group theory . indeed , applying the group theory to bernal stacked graphene layers , the irreducible representations of the infrared and raman active modes at the @xmath52 point for @xmath0-layer graphene can be obtained @xcite , as listed in table [ t1 ] .
the modes related to the @xmath2 band are the @xmath53 , @xmath46 , @xmath54 and @xmath55 in - plane modes , while the @xmath56 modes are related to out - of - plane phonons .
.irreducible representations of the ir and raman active modes at the @xmath52 point for @xmath0-layer graphene [ cols="^,^,^",options="header " , ] @xmath53 and @xmath46 modes are found in case of inversion symmetry of the system ( @xmath0 even ) , whereas @xmath54 and @xmath55 modes appear for mirror symmetry ( @xmath0 odd ) .
importantly , only the inversion symmetry inhibits the raman activity of antisymmetric modes
. it can be obtained by ab initio calculations @xcite that in trilayer graphene two @xmath54 modes and one @xmath55 mode can be found at the g - band frequencies , whereas in 4-layer graphene two @xmath53 modes and two @xmath46 modes vibrate at the @xmath2 band frequencies . the inversion symmetry breaking , produced by the top doping , switches on the previously raman silent antisymmetric modes @xmath46 in @xmath0 even - layer graphene , so that all the phonon modes become raman active and mixed in @xmath0-layer graphene for every @xmath0 value .
moreover , like in the bilayer case previously discussed , the epc can be strongly affected by the lack of symmetry constraints and by the change of the electronic band structures .
the absence of splitting for trilayer graphene can be qualitatively interpreted considering the allowed electronic interband transitions which give rise to the phonon energy renormalization ( kohn anomaly ) when @xmath57 ( low doping ) . in fig .
[ f5 ] it is shown that both @xmath54 and @xmath55 phonons can couple with electronic transitions when the fermi level is at the dirac point , whereas in bilayer graphene only the symmetric @xmath53 mode can efficiently create electron - hole pairs , due to energy conservation and symmetry selection rules . indeed
, first principles calculations @xcite have shown that in trilayer graphene the phonon linewidths of symmetric and antisymmetric modes are all of the same order of magnitude , whereas in the bilayer the antisymmetric linewidth is two orders of magnitude smaller than that of the symmetric mode . therefore , almost the same epc strength is expected for the three phonon modes in trilayer graphene , giving rise to the phonon energy renormalization at @xmath58 and to the consequent hardening of all phonons with moving the fermi level , like in monolayer graphene .
this is a qualitative different case with respect to that of bilayer , where only the symmetric phonon energy is renormalized at @xmath59 .
the presence or absence of the @xmath2 band splitting in the two cases reflects the presence or absence of distinct epc for phonons of distinct symmetry .
.the electronic bands are taken from ref .
the transitions indicated by dotted lines are not involved in the phonon energy renormalization , because their energy is much higher than the @xmath2 band phonon energy ( @xmath60 ) , which is reported as a scale bar on the left . ]
the observation of splitting in the 4-layer spectrum suggests the presence of distinct epc for the different phonon modes in analogy with the bi - layer behavior . in fig .
[ f6 ] we show the experimental position of the two lorentzian peaks fitting the 4-layer @xmath2 band , as a function of doping as obtained from the bilayer analysis .
it is worth noting that a minimum can be clearly identified in the curve of the low frequency peak , indicating that a maximum is likely to occur in the epc for some phonon modes at a given doping value ( @xmath61 ) . .
the electronic bands are taken from ref .
@xcite , and the @xmath2 band phonon energy ( @xmath60 ) is reported as a scale bar .
( right ) positions of the two peaks fitting the splitted @xmath2 band of 4-layer graphene as a function of doping .
the observed dip may be related to a strong epc due to the intraband transitions shown on the left . ]
such a feature can be interpreted by taking into account the band structure of the 4-layer , which can be approximated by two bilayer type band structures @xcite . in particular , intraband transitions occurring between two sub - bands separated by @xmath62 , which corresponds to an experimentally observed absorption by ir spectroscopy @xcite , may give rise to an almost resonant coupling with phonons ( @xmath63 ) , and to a consequently strong renormalization of phonon energy .
this epc is expected to have a maximum when the fermi level reaches the high energy sub - band ( see fig . [ f6 ] ) , when the number of possible intraband transitions is maximized . therefore , the doping value @xmath64 , at which a minimum for the raman peak position is observed , is likely to correspond to the fermi energy touching the high energy sub - band at @xmath65 .
a theoretical analysis of the 4-layer band structure and density of states in the presence of asymmetric doping may confirm this hypothesis .
in summary , we have experimentally investigated the raman @xmath2 band for @xmath0-layer graphene ( @xmath66 ) in the presence of high asymmetric doping , finding two different types of behavior .
for @xmath0 odd , the @xmath2 band is always fitted by a single lorentzian peak , which is blue - shifted with increasing the doping level .
this is due to a strong epc for all phonon modes when @xmath67 , which decreases with increasing the fermi energy . on the other hand , for @xmath0 even , an evident splitting of the @xmath2 band is observed , related to the presence of distinct epc for phonons of distinct symmetry . in particular , in the 4-layer case a signature of the van hove singularity at @xmath68 is likely to be observed as a minimum of the low energy peak position .
insights into the electron - phonon interactions in @xmath0- layer graphenes in the presence of top doping can be useful for the study of field effect graphene - based devices .
moreover , the @xmath69 plasma treatment may be a powerful technique for the study of graphene in the presence of a coverage of highly electronegative atoms such as fluorine .
finally , the variety of the results reported in the literature about the raman @xmath2 band in heavily doped bilayer graphene @xcite suggests that the repartition of the additional charge carriers is not well understood in most experiments .
therefore , suspended graphene samples may be a good testbed to further investigate the distribution of the doping , especially in the bilayer case .
this work was carried out within the euramet joint research project `` ulqhe '' .
the research within this euramet jrp receives funding from the ec fp7 , era - net plus , under grant agreement no . | an experimental study of raman scattering in @xmath0-layer graphene as a function of the top layer doping is reported . at high doping level , achieved by a chf@xmath1 plasma treatment ,
we observe a splitting of the @xmath2 band in the spectra of bilayer and 4-layer graphene ( @xmath0 even ) , whereas the splitting is not visible in case of monolayer and trilayer graphene ( @xmath0 odd ) .
the different behaviors are related to distinct electron - phonon interactions , which are affected by symmetry breaking and fermi level position in different ways in the various @xmath0-layer graphenes . in trilayer graphene , a weakening of the electron - phonon coupling as a function of the fermi energy induces a hardening of all zone - center in - plane optical phonon modes , like in monolayer graphene . on the other hand , in 4-layer graphene two distinct trends
are observed in the @xmath2 band as a function of doping , suggesting the presence of two different groups of electron - phonon interactions , like in bilayer graphene . |
one of the goals of atom optics is the creation of efficient analogues of optical elements for atoms .
considerable progress has been made demonstrating mirrors , lenses , and beamsplitters @xcite .
these elements might be combined into devices such as atomic interferometers @xcite .
evanescent wave atomic mirrors have been demonstrated in a number of experiments @xcite . a reflective diffraction grating for atoms
was reported by christ _
they observed diffraction of a slowed metastable neon beam from an evanescent optical grating formed by counterpropagating laser beams . up to 3% of incident atoms
were diffracted by up to 50 mrad from the main reflected beam .
this constitutes a large angle reflective beamsplitter .
diffraction of fast metastable neon atoms has recently been observed by brouri _
_ @xcite .
diffraction is possible because grazing incidence of the atomic beam produces an atomic de broglie wavelength perpendicular to the grating which is comparable to the grating periodicity .
the diffraction angles are determined by energy and momentum conservation @xcite .
however to determine the fraction of atoms which are diffracted the interaction of the atoms with the evanescent field must be analyzed in detail .
two - level atom models have dominated the theoretical work on evanescent wave devices @xcite .
although two - level models predict diffraction @xcite we have previously reported @xcite that they can not explain the _ particular _ diffraction observed by christ _ et al .
_ @xcite . in this paper
we show that multi - level atoms can explain the observed diffraction .
it can be understood using the quasipotential theory of deutschmann _ et al .
. the richer structure of the multi - level theory allows avoided crossings between ground state quasipotentials that are impossible in the two - level theory .
these produce diffraction by allowing atoms to exit the evanescent field in different quasipotential eigenstates than they entered the field @xcite .
the multi - level model not only explains the atomic diffraction observed by christ _ et al .
_ but also suggests further work .
in particular we find a sensitive dependence on the polarization of the laser beams @xcite , and on the incoming atomic zeeman level . for pure s - polarisation , the system can be represented as a series of non - interacting two - level systems and hence we do not see any diffraction .
our numerical model incorporates the ten magnetic sublevels of the 3s[3/2]2 @xmath0 3p@xmath1[3/2]2 transition of metastable neon . by numerically solving the time dependent schrdinger equation
we find diffracted beam fractions consistent with the observations of christ _
_ @xcite .
the geometry of the diffraction experiment is shown in fig .
[ schematic ] .
the direction perpendicular to the quartz surface is @xmath2 and the direction parallel to the surface @xmath3 .
the origin of the @xmath2 coordinate is at the interface and the positive direction is into the vacuum .
we assume the atom is initially propagating towards the interface in the @xmath4 plane with a positive @xmath3-component of velocity .
[ schematic ] the modulus of the evanescent field wavevector @xmath5 and its inverse decay length @xmath6 are @xcite @xmath7 where @xmath8 is the free field wavevector modulus , assumed to be the same for each laser , @xmath9 is the refractive index of the quartz , and @xmath10 is the angle of incidence of the lasers with the quartz surface .
the values of these parameters for the experiment we shall model are given in table 1 .
.parameters used in experiment of christ
_ et al_. @xcite .
[ cols="^,^",options="header " , ] we assume that the atoms are initially in an eigenstate @xmath11 of the @xmath3-component of their momentum with eigenvalue @xmath12 .
then because the photon momentum of the copropagating ( counterpropagating ) field is @xmath13 ( @xmath14 ) the @xmath3-component of the atomic momentum is restricted to the eigenvalues @xmath15 , with @xmath16 any integer .
we denote the corresponding set of allowed centre - of - mass @xmath3-momentum eigenstates by @xmath17 .
a ground ( excited ) state atom has @xmath16 even ( odd ) .
the 3s[3/2]2 @xmath0 3p@xmath1[3/2]2 transition of metastable neon is between states with total angular momentum j=2 .
hence there are five ground @xmath18 , and five excited @xmath19 , sublevels labeled by the magnetic quantum numbers @xmath20 , fig .
[ levels ] . for the transitions are shown . ]
[ levels ] the _ discrete _ basis states for the @xmath3-component of momentum and the internal state of the atom are therefore @xmath21 given our initial condition of ground state atoms the magnetic quantum number @xmath22 is @xmath23 ( @xmath24 ) if @xmath16 is even ( odd ) . the atomic motion in the @xmath2-direction , perpendicular to the quartz surface ,
is represented by one coordinate basis wavefunction @xmath25 for each discrete basis state @xmath26 .
we use the following ansatz for the complete state of the atom @xmath27 where @xmath28 is the atomic mass .
the exponential prefactor accounts explicitly for the initial atomic kinetic energy in the @xmath3-direction .
the hamiltonian @xmath29 for the system can be divided into two parts : the @xmath2-component of the kinetic energy @xmath30 , with continuous eigenvalues , and the rest @xmath31 , which has discrete eigenvalues , @xmath32 @xmath33 ( @xmath34 ) is the @xmath2-component ( @xmath3-component ) of the atomic momentum perpendicular ( parallel ) to the quartz surface .
@xmath35 is the sum of the atom s internal energy and the electric dipole interaction energy between the evanescent field and the atomic transition @xmath36 , @xmath37 this hamiltonian is in an interaction picture with the atomic dipoles rotating at the average of the laser frequencies @xmath38 , where @xmath39 is the frequency of the laser copropagating ( counterpropagating ) with the atoms .
these frequencies can be different , as in the experiment of stenlake _ et al .
_ @xcite . in that case
they must be sufficiently similar that the approximation of equal photon momentum magnitudes holds .
the atomic detuning @xmath40 is the difference between the ( degenerate ) atomic transition frequencies and @xmath41 .
the dipole interaction energy is @xmath42 where h.c .
means hermitean conjugate and @xmath43 is the positive frequency part of the transition electric dipole moment operator @xmath44 @xmath45 @xmath46 is the positive frequency part of the total electric field @xmath47 @xmath48 @xmath49 ( @xmath50 ) is the amplitude of the copropagating ( counterpropagating ) field at the quartz surface , and is assumed real for convenience .
the @xmath51 are the corresponding field polarization vectors .
@xmath52 is half the frequency difference between the two lasers . in the examples we consider in the next section @xmath53 .
we take the @xmath54 direction as the quantization axis and work with a spherical basis of polarization vectors @xmath55 defined in terms of the cartesian unit vectors by @xmath56 the wigner - eckart theorem @xcite allows us to express the dipole matrix elements in terms of clebsch - gordon coefficients @xmath57 and a reduced matrix element @xmath58 @xmath59 where @xmath60 is @xmath61 for @xmath62 transitions .
the values of the non - zero clebsch - gordon coefficients are indicated in fig .
[ levels ] .
the dipole interaction hamiltonian eq .
( [ hamed ] ) then becomes @xmath63 the @xmath64 are the spherical polarization components of the two lasers . substituting the ansatz eq .
( [ ansatz ] ) into the schrdinger equation corresponding to the hamiltonian eq .
( [ hamiltonian ] ) gives the set of coupled schrdinger equations @xmath65\psi_{n
, m_{e / g } } \nonumber \\ & & + \sum_{n',m_{e / g } ' } \langle n , m _
{ e / g } | v |n',m_{e / g}'\rangle \psi_{n',m_{e / g } ' } \quad , \label{coupledse}\end{aligned}\ ] ] @xmath66 we have solved this set of partial differential equations numerically by the well known split operator method @xcite .
our numerical solutions were checked for accuracy by demanding that changing either the time - step or the spatial grid size did not significantly alter the final result .
before presenting numerical solutions of the schrdinger equations we consider the quasipotentials of the hamiltonian eq .
( [ hamiltonian ] ) @xcite .
these are the eigenvalues of that part of the hamiltonian having discrete eigenvalues , namely @xmath67 eq . ( [ discreteham ] ) .
since @xmath35 includes the interaction with the evanescent field the quasipotentials are a function of @xmath2 , the distance from the quartz surface .
the eigenstates corresponding to the quasipotentials may be thought of as atomic states doubly dressed by the two evanescent fields .
the eigenstates of @xmath35 corresponding to the quasipotentials are adiabatically followed by atoms moving sufficiently slowly towards the surface , except near avoided crossings of the quasipotentials .
hence they behave like actual potentials , for example slowing atoms down as they climb them , which is the origin of atomic reflection @xcite . at the avoided crossings non - adiabatic transitions between the quasipotentials
may occur .
these make diffraction possible , since the atoms may leave the evanescent field in a different superposition of quasipotential eigenstates than that in which they entered , and different eigenstates may have different momenta .
a two - level atom model allows diffraction for certain ranges of parameters .
in particular the diffraction considered by deutschmann _
et al_. @xcite occurred for much lower atomic detunings than were used in the experiment of christ _
et al_. @xcite .
the lower detunings would have produced unacceptable levels of atomic excitation and hence of spontaneous emission .
since excited states spontaneously emit , diffraction into ground states , with even @xmath16 , is of most interest .
the optimum probability for the @xmath68 diffraction orders was estimated to be about 6% by deutschmann _
et al_. @xcite .
this is because at least four avoided crossings are required , with an optimal transfer probability of 50% at each , and @xmath69 .
the quasipotentials for a two - level model of the experiment are shown in fig .
[ qp](a ) .
note that the incoming @xmath70 quasipotential only has avoided crossings with high order quasipotentials , the first being with the @xmath71 quasipotential @xcite .
the transition between the corresponding eigenstates would involve a 21 photon process and hence be very weak , giving negligible @xmath72 order diffraction .
this is inconsistent with the observations of christ _
et al_. @xcite . away from p - polarization .
] [ qp ] the quasipotentials for the multi - level atom model of the experiment are shown in fig .
[ qp](b ) .
the multi - level model makes possible raman transitions between the ground zeeman levels .
hence there is now an avoided crossing between @xmath70 and @xmath72 quasipotentials , so diffraction into an @xmath72 eigenstate is expected . the experiment of christ _
_ @xcite that we have modelled reported up to 3% diffraction .
the ratio of the intensities of the copropagating to counterpropagating waves was fixed at 1.64 . to calculate the diffracted fraction we numerically solved the schrdinger equations ( [ coupledse ] ) , as described in the previous section .
our results are quite sensitive to the polarizations of the two laser beams @xcite .
[ diffvpol ] shows the percentage of diffraction into the @xmath72 order as a function of the polarization angle of the lasers .
the results in this figure were calculated for an equal mixture of magnetic sublevels in the incoming atomic beam .
we did not calculate the diffraction from all initial @xmath23 states for all polarization angles .
this was because each point on fig .
4 took approximately 30 minutes to compute on a vpp300 supercomputer . typically the spatial grid in the @xmath2 direction had 2048 elements and 13 @xmath16 states , @xmath73 to @xmath74 were used .
the integration over 1.6 @xmath75s was performed with 0.2 ns timesteps .
these numerical parameters were varied to ensure the insensitivity of the solutions .
however we did calculate for all @xmath23 states for several representative polarisation angles , and in all cases it was found that they produced amounts of diffraction bearing a fixed ratio to each other .
this is because only one of the @xmath76 quasipotentials is involved in the diffraction process , and the various @xmath23 states enter this quasipotential in fixed ratios .
this observation allowed us to deduce the overall diffraction for a mixture of magnetic sublevels .
order diffraction versus polarization angle for an incoming @xmath77 atom .
the percentage of diffraction is plotted versus the linear polarization angle , in degrees , of the two laser beams from pure p - polarization , @xmath78 .
the solid curve is a fit to the calculated points .
other parameters are as in table 1 . ]
[ diffvpol ] we have also modelled the conditions which gave maximum diffraction according to unpublished experimental data @xcite : the polarization of the copropagating beam rotated by 5 degrees from perfect p - polarization ( electric field in the plane of incidence ) and the counterpropagating beam rotated 15 degrees from perfect p - polarization @xcite . despite uncertainty as to the accuracy of these figures , we found that these parameters gave a large degree of diffraction ( about 14% ) .
furthermore , a ratio of 1.64 between the co - propagating and counterpropagating beams was cited as an important condition for diffraction to occur @xcite .
computationally , we found that this condition produced a local maximum in the amount of diffraction .
these results demonstrate that a multi - level model is able to account for the diffraction observed by christ _
the difference between the diffraction we find ( 14% for experimental parameters ) and that observed , 3% , is reasonable given the non - ideal aspects of the experiment and the lack of delicate control over the parameters .
state atom .
the probability density latexmath:[$p_{n } = \sum_{m } @xmath16 and either the distance from the glass in microns ( a ) and ( b ) , or the velocity perpendicular to the glass ( c ) and ( d ) . in each case
the top figure is the initial condition and the bottom figure is 1.6 @xmath75s later , after reflection .
there is no probability in the orders not shown .
parameters are as in table 1 with laser polarizations of @xmath80 away from p - polarization . ]
[ wf ] one of the advantages of numerically solving the time - dependent schrdinger equation is that `` movies '' of the wavefunction evolution are available .
the wavefunction can be visualized in either coordinate or momentum space .
[ wf ] shows an initial @xmath81 probability density evolving to produce @xmath82 order diffraction .
we have quantitatively modeled , from first principles , the reflection grating atomic diffraction experiment of christ _
our results are consistent with the levels of diffraction observed in that experiment .
the model predicts strong dependence on the polarization of the copropagating and counterpropagating laser beams .
this is physically reasonable since the atomic zeeman structure makes the shifts of the various atomic transitions polarization dependent .
there is some evidence for this effect in the experiment @xcite .
computational modelling of experiments , such as we have reported , is particularly useful if it can suggest new experiments .
our results show that control of the polarization of both laser beams is crucial in reflection grating atomic diffraction experiments .
furthermore we found that under the conditions of the experiment of christ _
_ @xcite the zeeman @xmath83 state produced most of the @xmath72 order diffraction .
hence optical pumping into this state could potentially increase the diffraction .
our model is easily adapted to other atoms and we plan to use it to model a caesium evanescent wave diffraction experiment currently underway in our group .
we are particularly indebted to r. deutschmann and m. schiffer for correspondence and unpublished data concerning their experiment .
we also acknowledge discussions with the anu atom optics group , especially , i. littler and j. eschner .
the computations were performed at the australian national university supercomputer facility . | diffraction of multi - level atoms by an evanescent wave reflective diffraction grating is modeled by numerically solving the time - dependent schrdinger equation .
we are able to explain the diffraction observed in experiments with metastable neon .
this is not possible using a two - level atom model .
the multi - level model predicts sensitive dependence of diffraction on the laser polarization and on the intensity ratio of incoming and reflected laser beams . |
driven dissipative systems made of colloidal particles suspended in liquid afford a way to study many - body non - equilibrium dynamics at the single particle level @xcite .
for example , using a 1d crystal of oil droplets driven along a microfluidic channel , non - equilibrium hydrodynamically - induces phonons were recently studied @xcite . surprisingly , the crystal vibrations , in that system , did not decay due to thermal fluctuations , as could be expected at such low reynolds numbers ( re@xmath0 ) , but continued to propagate acoustically at a sound velocity of @xmath1 m/s @xcite .
the dispersion relations of this 1d crystal were different from those of a harmonic crystal , they exhibited symmetry breaking due to flow directionality .
here we study the collective excitations of a different driven dissipative system . in our system particles
are driven along a ring - like optical trap , which means that they are confined to a quasi one dimensional track , but interact hydrodynamically in 3d rather than in 2d as in the previously studied system @xcite .
moreover , particles in our system tend to pair due to a hydrodynamically induced pseudo - attraction @xcite .
we show here , that this pairing interaction greatly affects the particles collective excitations .
m polystyrene beads trapped in the optical vortex rotate at approximately constant angular velocity . ]
we use holographic optical tweezers ( hots ) @xcite to confine suspended colloidal particles and drive them along a 1d circular optical trap .
thus particles never leave the field of view , and can be followed for long periods of time ( i.e. an hour ) .
one type of such a circular trap is created with hots by projecting a helical beam of light ( optical vortex ) , carrying angular momentum @xcite , into the sample plain ( fig [ fig : system ] ) .
when particles are trapped in an optical vortex beam , this angular momentum is transfered to them , causing them to rotate around the ring - shaped optical trap @xcite . as a result ,
the trapped colloidal particles constitute a driven dissipative system in which the particles are driven at constant angular velocity and are coupled to a thermal bath ( the suspending media ) . many properties of this model system have bean calculated @xcite and measured @xcite in the past two decades . on a broader context , a single particle driven along a ring trap constitutes a model for the motion of cilia @xcite , and has been used to study synchronization among cilia hairs in arrays of ring traps ( e.g. @xcite ) .
the same mechanism that is responsible for particle pairing of two particles rotating in a single ring @xcite , causes synchronization between single particles rotating in two adjacent rings @xcite . in both cases
the effect depends on the ratio between tangential driving and radial confinement along the ring .
the equations of motion for @xmath2 suspended colloidal particles subjected each to a force , @xmath3 , and interacting hydrodynamically , are given by @xcite : * v*_i=_j=1^n*h*_ij*f*_j .
[ eq : vel ] where @xmath4 is the velocity of particle @xmath5 .
it is common practice to assume pairwise additivity of the hydrodynamic interactions @xcite in the stokeslet approximation to derive @xmath6 ; * h*_ij_ij+ ( * i*+*r*_ij*r*_ij ) ( 1-_ij ) [ eq : stokes ] where @xmath7 is the stokes drag on the sphere , and @xmath8 is the dynamic viscosity of the fluid . in an optical vortex the optical forces acting on a typical particle are a constant tangential driving force @xmath9 , and a radial confining force @xmath10 , where @xmath11 is the trap stiffness , @xmath12 is the radial position of particle @xmath5 , and @xmath13 is the radius of the optical vortex .
these equations of motion were linearized , at zero temperature , around a configuration of equidistant arrays of trapped particles in @xcite .
the angular velocity of the particles , @xmath14 , was separated into a constant value , @xmath15 and a time varying component , @xmath16 . assuming the radial confinement to be infinitely strong , @xmath15 and @xmath17 are given by : & = & ( 1+_ij ) + & = & - _ ij_j [ eq : motion ] where @xmath18 , @xmath8 the fluid viscosity , and @xmath19 the radius of the colloidal particles .
the solution of these equations reveals several oscillatory modes of motion which the system can adopt @xcite . in experiments , trapped
colloidal particles seem to alternate between the different modes @xcite .
note that these modes are inherently different from the hydrodynamic modes that exist in equilibrated arrays of trapped colloids @xcite as they have real eigenvalues @xcite .
it was recently shown @xcite that when only two particles are trapped in an optical vortex , they pair .
this pairing attraction is a result of symmetry breaking induced by the curved path the particles are forced to take .
consequently , the average angular velocity of particles in an @xmath2 particle vortex can be predicted from the zero mode motion of the aforementioned equations of motion @xcite , by treating each particle pair as an effective particle with an effective mobility @xcite .
one important aspect of the pairing interaction is that it is a pure non - equilibrium effect , it arises only through driving .
this means that the effects of driving are observable even in the co - rotating frame , i.e. the formation of particle pairs .
particle pairing can be overcome in two ways : by increasing temperature , or by increasing the ratio between the radial confinement force and the tangential driving force , @xmath20 @xcite .
in this paper we study the collective excitations of particles trapped in an optical vortex as a function of the pairing interaction strength .
we start by demonstrating , for the first time , that @xmath21 can be controlled by tuning laser powers .
we then correlated between the trapping laser power and @xmath21 , in a one - particle vortex , and find a range in which @xmath21 changes linearly with laser power .
we proceed to verify the assumption that @xmath21 controls pairing by measuring the average angular separation between two particles trapped in an optical vortex as a function of @xmath21 measured from single particle experiments .
finally , we turn our attention to the collective vibrational modes of many particles trapped in an optical vortex .
we start by calculating theoretically and numerically the dispersion relations of particles in a vortex as a function of the driving force and particle number .
we then compare our calculations to the experimentally measured dispersion relations and show that they are non - decaying , similar to what was found for a 1d crystal of oil droplets @xcite , but are better described as fluctuations of particle pairs rather than as fluctuations of single , separated , beads .
* experiments*- colloidal suspensions of polystyrene beads , 1.1 @xmath22 in diameter ( invitrogen lots # 742530 ) , were place in between a glass slide and a cover - slip to form an approximately 40 @xmath22 thick sample . during experiments the suspended colloidal particles
were subjected to an optical vortex trap generated by holographic optical tweezers @xcite . in our setup a gaussian laser beam ( coherent , verdi @xmath23 nm ) is transformed into a helical beam carrying a topological charge of @xmath24 using a spatial light modulator ( hamamtsu , x10468 - 04 ) .
the beam is then inserted into an inverted microscope ( olympus , ix71 ) and focused through the microscope objective ( 100x oil immersion na=1.4 ) forming a ring - like optical trap with a diameter of @xmath25 @xmath22 and diffraction limited thickness .
we performed three sets of experiments trapping either 1,2 or 21 particles , driving them at laser powers of 0.5 - 4 w ( at the output of the laser ) , and imaging them at either 10 fps ( hamamtsu , orca flash 4 ) , or at 70.4 fps ( gazzele , point gray ) .
particle tracking was done using conventional video microscopy @xcite . during each experiment particles
were allowed to reach a steady state before measurement commenced . * simulations*- we use the stokesian dynamics protocol @xcite to simulate particles driven in an optical vortex due to its tested ability to correctly account for many particle hydrodynamic interactions , and its incorporation of thermal forces .
extending eq .
[ eq : vel ] to include thermal forces and replacing the stokeslet approximation ( eq . [ eq : stokes ] ) with the rotne
prager operator @xcite , we have : @xmath26 where @xmath27 the diffusion tensor , @xmath28 boltzmann s constant , @xmath29 the absolute temperature , and @xmath30 represents the stochastic thermal force : @xmath31 we obtain @xmath32 by employing the cholesky decomposition on the diffusion matrix : @xmath33 , with @xmath34 the lower triangular matrix , and @xmath35 the basis set of random gaussian variables @xmath36 , @xmath37 the discretized equations of motion then read , @xmath38 with the rotne prager tensor @xcite given by : @xmath39 with @xmath40 . in the simulations particles
are allowed to move in two dimensions confined to a ring of radius @xmath13 with a harmonic restoring force @xmath41 , and a driving force @xmath42 .
a weeks - chandler - anderson repulsive pair potential @xcite is used to represents the hard core repulsion between colloidal particles .
simulation parameters were chosen to match experiments in terms of particle size , vortex radius , driving and confining forces and temperature .
the time interval was equivalent to @xmath43 .
there are two main differences between the experimental system and the simulated system .
experiments are conducted near a glass wall whereas simulations assume no near boundary . as a result
the hydrodynamic interactions in experiments should be weaker .
in addition , in simulations the optical potential landscape is perfectly smooth , unlike the situation in experiments .
a colloidal particle driven along an optical vortex experiences , ideally , two types of optical forces : an intensity gradient force confining it radially @xmath44 , and a phase gradient force , due to radiation pressure , driving it tangentially @xmath45 @xcite .
we measure these forces by tracking a single particle rotating in an optical vortex ( fig .
[ fig : system2]a ) . the tangential driving force can be extracted from the average angular velocity of the bead @xmath15 , since @xmath46 .
the radial confining force @xmath47 , can be extracted from the radial position distribution of the bead , using @xmath48 , where @xmath49 is the stiffness of the trap in the radial direction , @xmath50 m is its width , and @xmath51 is the variance in the bead s radial position .
, as a function of laser power .
inset shows the radial @xmath47 and tangential @xmath52 forces . ]
the optical potential landscape of an optical vortex trap is not completely uniform @xcite , due to imperfections in the optical train . as a result ,
the particles move over a rugged landscape , similar to the motion over a periodic potential with a small amount of randomness ( for example see theoretical treatment in @xcite ) .
the roughness of the optical potential is characterized indirectly , by studying the tangential diffusion of a single particle ( fig .
[ fig : system2]b ) at the co - rotating frame ( the frame of reference rotating at the mean angular velocity of the particle ) .
we observe that at low laser powers the particle diffusion becomes sub - diffusive ( @xmath53 with @xmath54 ) , while at high laser powers it is super - diffusive ( @xmath53 with @xmath55 ) .
we attribute these changes to the effect of the rough optical potential . at intermediate laser powers
the particle diffuses normally , which implies that the landscape effect is minimal .
we therefore perform our experiments in this regime . in fig .
[ fig : kappa ] the forces @xmath47 , @xmath52 , and their ratio , @xmath21 , are shown as a function of laser power . here
the same dynamical regimes are observed , at low driving force @xmath21 is large and widely distributed . at high driving force
@xmath21 decreases with laser intensity and has a narrow distribution . at intermediate driving forces ( @xmath56 ) , where the particle diffusion is normal ( see fig . [
fig : system2]b ) , and we perform our experiments , @xmath21 increases linearly with laser intensity . , the ratio between the radial confinement force and the tangential driving force for simulations ( blue circles ) , and experiments ( green squares ) . in experiment
laser powers vary between @xmath57 .
the inset shows a typical time series from experiment of the angular separation between two particles ( i=2 w , @xmath58 ) , the pairing angle was averaged over the paired state , e.g. in the inset , of angles lower than @xmath59 . ]
when a second particle is added to the optical vortex trap , the system relaxes to a new steady state in which the two particles come close and their angular separation @xmath60 fluctuates around some mean value .
the inset of fig .
[ fig : dtheta ] shows a typical time series of @xmath60 exhibiting significant fluctuations during experiments . from simulations and previous work @xcite
we deduce that our experiments are performed at an intermediate temperature regime in which the system translates between paired and unpaired states continuously .
the angular separation between the particles has a main peak with secondary lobes . we define a paired state as a state in which the angular separation lies in the main peak , and the pairing angle as the average angular separation between the particles at such a paired state . at a low laser power of @xmath61w
no distinct particle pairing is observed , i.e. the inter - particle distance varies significantly with time .
increasing laser power to @xmath56 , corresponding to @xmath62 , particles become paired ( see fig . [
fig : dtheta ] ) .
the average separation angle , @xmath63 , depends linearly on @xmath21 , regardless of laser power ( compare to fig . [
fig : kappa ] ) , and ranges from @xmath64 to @xmath65 .
this corresponds to distances of 1.4 to 2.6 particle diameters , meaning that although particles come close to touching during experiments , occasionally , they are further apart at their preferred angular separation . in simulations ( fig .
[ fig : dtheta ] ) pairing angle decreases with increasing @xmath21 as well , however , unlike the experimental findings , the pairing angle show particles almost touching .
in addition , above a certain @xmath21 , the particles no longer pair .
this crossover is not experimentally accessible .
we now turn our attention to the collective excitation of particles on an optical vortex . in our system , a vortex of topological charge @xmath24 can hold as many as 21 particles for a long period of time ( several hours ) .
this means that the 1d volume fraction of particles is @xmath66
. a larger number of particles can be trapped in the ring , but with time particles start leaving the vortex , rotating alongside other particles , and reentering the trap occasionally @xcite .
we note here that odd numbers of particles in a vortex beam , at high density , seem more stable than even numbers at similar densities , in experiments .
taking into account our results from the one- and two - particle experiments , we choose to work at intermediate laser powers ( @xmath57 ) . in this range
an increase in laser power corresponds to a decrease in pairing attraction .
the mean velocity of the particles increases with particle number @xcite and with driving power ( see fig . [
fig : system2]a ) , ranging from @xmath67 m/s to @xmath68 m/s for our 21 particle vortex . in fig .
[ fig : particles ] a snapshot of 21 particles rotating in an optical vortex at @xmath69 is compared with snapshots from simulation at room temperature and zero temperature .
the colloidal configuration at low temperature shows a periodic structure , that is part of a limit cycle motion ( see for example supplementary movie @xcite ) including many particle pairs . at room temperature the particles configuration ( fig .
[ fig : particles]b ) and dynamics is not as ordered ( see supplementary movie @xcite ) . from these images
it is not clear if , and to what extent , the hydrodynamic induced pairing interaction affects the collective particle excitations , especially at room temperature . in diameter ,
trapped in an optical vortex of topological charge @xmath24 , at @xmath69 .
b ) simulation results for similar conditions and room temperature , and c ) simulation results at zero temperature . ] in order to address this question we analyze the fluctuations in the angular position of the particles in the co - rotating frame in terms of their dispersion relations , @xmath70 , following the analysis of @xcite .
we start by defining new coordinates , the angular separation between adjacent particles , @xmath71 we apply a discrete fourier transform both on angular separation and time and obtain the power spectrum of particle fluctuations .
the dispersion relations are extracted from the peaks of the power spectrum . in fig .
[ fig : dispersion ] the experimentally measured dispersion relation of the angular fluctuations of 21 particles driven at @xmath72 is presented .
two branches can be observed ; a linear relation due to particles responding to intensity variation along the optical vortex ( arising from imperfections in the optical train ) , and a tilted sine - like curve arising from inter - particle hydrodynamic interactions .
similar to the observations for oil droplets in a microfluidic channel @xcite , we observe symmetry breaking of the dispersion relations due to driving , acoustic wave behavior near @xmath73 propagating at sound velocities of @xmath74 m/s to @xmath75 m/s , and group velocity sign change within the brillouin zone . , for longitudinal waves at @xmath72 .
two branches are seen , a sine - like curve due to hydrodynamic interactions between colloidal particles ( the peak position of which is present by black dots ) , and a straight line due to the response of particles to imperfections in the optical vortex ( dashed line ) . ]
figure [ fig : dispersion_compare]a shows the measured dispersion relation , normalized by the average angular velocity @xmath15 , of the 21 particles driven in the ring at different @xmath21 .
the dispersion relation of a simulated system are presented as well , multiplied by a factor of 2.7 to fit the experimental data .
we find that in experiments the dispersion relation shape changes significantly as @xmath21 increases .
a much weaker dependence on @xmath21 was observed in simulation .
there are two possible effects that can explain the discrepancy between experiment and simulation : the weaker hydrodynamic interaction in experiments due to the presence of a nearby glass wall , and the experimental rugged optical landscape over which the particles are driven . in order to obtain more intuitive understanding of the collective excitations in our system , we calculate theoretically the dispersion relation of particles in an optical vortex .
we assume that the equilibrium configuration of particles is a uniform distribution around the ring , and use eq .
( 1 ) to describe their motion . substituting _
m = e^i(mk - t ) into eq .
( 1 ) and solving for @xmath76 , we have = 2_m=1^(m)(mk ) [ eq : dispersion ] where @xmath77 , and n is the number of particles trapped on the ring . at low temperatures and low @xmath21
we expect particles to form stable uniformly distributed pairs @xcite .
assuming that particle pairs are point objects with double the force acting on them we can use the same equation ( eq . ( [ eq : dispersion ] ) ) to model their dispersion relations provided we multiple @xmath52 by 2 and divide the number of particles by 2 .
clearly , neither configuration describes our experiments well , as our particles ( and particle pairs ) are not point objects , radial confinement is not infinitely strong , and particles positions fluctuate significantly ( see fig .
[ fig : particles]a and movies @xcite ) .
surprisingly , this crude estimation affords insight into the collective excitations observed .
a theoretical curve for 21 particles arranged uniformly around the vortex ( dashed black line ) and a curve for 10 particle pairs arranged uniformly around the vortex driven at twice the force ( solid blue line ) are compared in fig .
[ fig : dispersion_compare]b to experiment and simulation .
the theoretical calculations are normalized by the calculated average angular velocity of particles in the vortex .
we observe , from eq .
( [ eq : dispersion ] ) , that the dispersion relation amplitude depends on the force acting on each bead , and its shape on the configuration of the beads . comparing between experiment and theory
, we find that particle pairing affects the collective excitations of the beads .
the fact that the normalized experimental curves do not collapse to a single curve ( fig .
[ fig : dispersion_compare]a ) reflects the change in pairing strength with @xmath21 , as was observed in the two - particle vortex ( fig .
[ fig : dtheta ] ) .
the sound velocity @xmath78 approaches that of an equally distributed array of single particles at low pairing strength , and that of an equally distributed array of particle pairs at high pairing strength .
it is interesting to note that the sound velocity of the particle waves ( @xmath74m / s@xmath79 m/s ) is higher than the average velocity of the particles ( @xmath80m / s@xmath81 m/s ) .
the shape of the experimental curve changes gradually with pairing strength approaching the single particle curve at weak pairing .
for all cases , at low frequencies , or at large distances the dispersion relations are closer to those calculated for single particles , but at high frequencies , or short distances coupling prevails .
increases and pairing decreases the experimental curves depart from the simulated curve and approach the theoretical curve for single particles .
b ) comparison of experiment , theory and simulation . while the shape of the dispersion relation on the simulated system is identical to that obtained from experimental data , the amplitude is 2.7 times higher in experiment ( with @xmath69 ) than in simulation . ] comparing our results with a driven 1d crystal oil droplet suspension studied recently @xcite , we find many similarities . both systems exhibit symmetry breaking in the dispersion relations due to the directionality of driving . in both systems
hydrodynamic interactions maintain vibrational modes in spite of the overdamped conditions , and the sound velocity of the collective vibrational modes is of the order of several hundreds of micrometers per second .
unlike the oil droplet crystal , instabilities in the optical vortex are subdued due to the strong radial confinement . however , when the 1d volume fraction of particles , approaches @xmath82 trapping becomes unstable . in such cases particles
are continuously kicked out of the ring trap and are later on absorbed back in @xcite .
another unique trait of our system , is that hydrodynamic interactions are responsible both for creating the underlying configuration of equidistant particle pairs and for maintaining its vibrational excitations .
this is different from the microfluidic 1d suspension whose crystalline structure stems from initial conditions of formation . even in room temperature , where particle pairing is relatively weak ( see fig . [ fig : particles]a ) , we find that the pairing effect is dominant in the collective vibrations of the driven particles .
the dynamics of the driven particles and their collective excitations are reproduced reasonably well by numerical simulation , as discussed above .
however , the effect of thermal fluctuations on the dynamics should be examined in the future . while in simulation we expect thermal fluctuation to be decoupled from the periodic motion of the particles and hence result in a trivial cloaking of the motion , in experiments this need not be the case .
the issue of whether thermal fluctuations or the vortex inhomogeneities are the dominant source of noise remains unclear and requires a more detailed study .
the beating of large assemblies of cilia hairs can be random , synchronized ( with identical phase ) , or metachronal ( traveling wave ) @xcite .
arrays of particles , each rotating around a ring , were used to model these collective modes both theoretically @xcite and experimentally @xcite . since particle pairing in a single vortex
is related to the synchronized motion of two cilia hairs , it seems that the collective excitations observed in our system should correspond to collective beating of assemblies of cilia hairs . in this paper ,
the collective excitation found for many particles in a single ring is of the traveling wave mode , full synchronization would correspond to all particles moving in union in a single group .
it would be interesting , to find the transition between the modes of motion in the optical vortex system and compare their onset to beating cilia , especially due to the larger number of particles accessible experimentally compared to multiple ring systems .
the authors wish to thank roy bar - ziv , haim diamant , and yair shokef for fruitful discussions .
in particular , we have benefited from extensive discussion with tsevi beatus .
this research was supported by the james franck german - israeli binational program in laser - matter interactions , by the israel science foundation grant 1271/08 , and by the us - israel binational science foundation grant 2008483 . | two colloidal particles , driven around an optical vortex trap , have been recently shown to pair due to an interplay between hydrodynamic interactions and the curved path they are forced to follow .
we demonstrate here , that this pairing interaction can be tuned experimentally , and study its effect on the collective excitations of many particles driven around such an optical trap .
we find that even though the system is overdamped , hydrodynamic interactions due to driving give rise to non - decaying excitations with characteristic dispersion relations .
the collective excitations of the colloidal ring reflect fluctuations of particle pairs rather than those of single particles . |
in its simplest form , the idea of extended unitarity ( eu ) states that the @xmath6 pair in a single partial wave should have the same phase variation with @xmath5 in _ all _ reactions as in elastic scattering .
this idea originates from aitchison @xcite and has been adopted in various guises by many authors .
his arguments will be presented in detail in section 2 , so as to expose the assumptions and consequences . at the time the idea was introduced , it was a reasonable conjecture ; now modern data allow it to be checked accurately , but disagree with it .
many experimental groups have made extensive fits to production data using a @xmath7-matrix approach based on eu .
these fits are excellent ; no criticism is intended of their quality .
experimentalists have found empirically the necessary freedom to get good fits to data .
however , on close inspection , this freedom is inconsistent with strict eu .
this whole topic has been the subject of extensive discussion with many authors .
there is a bewildering jungle of claims and counter - claims .
my objective is to cut a path through this tangle and expose where problems lie ; this makes the presentation pedantic in places .
aitchison s essential point is that all processes should be described by a universal denominator @xmath8 $ ] , where @xmath7 is the same as for elastic scattering ; @xmath9 is lorentz invariant phase space .
the assumption which is being made is that watson s theorem @xcite applies to the coherent sum of all components in the @xmath10 partial wave .
this is a step beyond watson s derivation , which referred only to a single eigenstate ; watson did not consider overlapping resonances .
-12 mm -7 mm in @xmath6 elastic scattering , the @xmath3 is superimposed on a slowly rising amplitude associated with the @xmath2 pole .
cern - munich data @xcite show that the phases of these two components add . below the @xmath11 threshold ,
both @xmath2 and @xmath3 @xmath12-matrices @xmath13 are confined to the unitarity circle if we neglect the tiny inelasticity due to @xmath14 .
unitarity may be satisfied by multiplying @xmath15-matrices @xmath16 of @xmath2 and @xmath3 , as suggested by dalitz and tuan @xcite .
this fits the data within errors of @xmath17 .
1 shows the argand diagram for the @xmath18 @xmath6 s - wave from my recent re - analysis of these ( and other ) data @xcite . from bes data on @xmath19
, the @xmath3 has a full - width at half - maximum of @xmath20 mev @xcite .
the combined phase shift rises rapidly from @xmath21 at 0.88 gev to @xmath22 near 1.1 gev .
there is a deep dip in the cross section where the combined phase goes rapidly through @xmath23 .
the crucial point of eu is that this feature should be common to production processes .
-15 mm -8 mm bes data on @xmath24 @xcite immediately require a modification of the rudimentary form of eu . the @xmath6 mass spectrum in these data is reproduced in fig .
there is a dominant @xmath3 contribution and a small interfering @xmath2 contribution ; this is very different to elastic scattering .
l " ahde and meissner @xcite modify the conjecture of eu to apply separately to strange and non - strange components , i.e. to the scalar form factors for @xmath6 and @xmath11 .
the dip in the elastic cross section at 989 mev is a very delicate feature .
if , for any reason , relative magnitudes of @xmath2 and @xmath25 change , the zero at 989 mev can disappear quickly ; here and elsewhere , @xmath25 will denote @xmath3 unless there is confusion with other @xmath25 s . if the phase of @xmath25 changes with respect to the @xmath2 , the mass at which the dip appears will likewise change .
the interference region between @xmath2 and @xmath3 is an ideal place to check extended unitarity .
two considerations will play a critical role : unitarity and analyticity .
unitarity is often quoted and plays an essential role in setting up the current @xmath7-matrix formalism which treats both elastic scattering and production on the same basis . for a production reaction , aitchison conjectures a unitarity relation for the production amplitude @xmath26 : @xmath27 he defines his @xmath26 to be proportional to @xmath28 , where @xmath29 refers to production : @xmath30 dividing both sides of ( 1 ) by @xmath31 , @xmath32 it is odd that @xmath29 on the right - hand side is multiplied by @xmath33 , unless @xmath34 .
however , experiment will require different contributions to @xmath29 from @xmath2 and @xmath25 .
consider next analyticity .
dispersion relations connect magnitudes and phases .
if the relative magnitudes of @xmath25 and @xmath2 change from those of elastic scattering because of matrix elements , their relative phases _ must _ also change .
conversely , analyticity predicts that if the phase variation with @xmath5 of the amplitude is universal , as eu demands , so is the variation with @xmath5 of the magnitude ( up to a constant scaling factor ) ; for the simplest situation where only resonances are present , the relative magnitudes of @xmath2 , @xmath3 and any further @xmath25 must be _ almost _ the same in production as elastic scattering .
this is a point which has almost always been ignored .
the word ` almost ' represents a caveat : there may in addition be a polynomial is @xmath5 which can be different between elastic scattering and production .
it turns out that one can plausibly limit deviations of relative magnitudes within @xmath35 .
this question is discussed in subsection 2.1 .
experiment requires larger deviations than this in the four sets of data discussed here .
this implies phases must change from those predicted by eu .
experimentalists have correctly allowed for this by using complex coupling constants in the isobar model .
section 3 compares the prediction of eu with 3 sets of data .
the first concerns bes data for @xmath36 @xcite .
the non - strange components of @xmath2 and @xmath25 dominate both elastic scattering and @xmath0 . from aitchison
s algebra and that of lhde and meissner , it follows that the @xmath25 amplitude should have almost the same magnitude as the @xmath2 amplitude , as well as the same phase as elastic scattering .
this prediction is contradicted by the data , where no @xmath3 is visible and a fit to the data places a low limit on it
. the next two sets are crystal barrel data for @xmath37 , where @xmath2 and @xmath3 are clearly visible @xcite .
one set is for annihilation in liquid hydrogen and the other for gaseous hydrogen .
annihilation from the @xmath38 initial state is 13% in liquid and 48% in gas , allowing a clear separation of amplitudes for production of @xmath2 and @xmath25 from @xmath39 and @xmath38 .
results for both are inconsistent with the deep dip of elastic scattering predicted by eu .
section 4 concerns data from the afs experiment on central production : @xmath40 @xcite .
here one expects the protons in the final state to act as spectators .
however , eu still fails conspicuously to fit the data .
this important result leads to a revised form of the unitarity relation , as follows .
-87 mm -52 mm fig .
3 sketches the usual diagrammatic approach to the unitarity relation .
it may be derived by cutting the diagram down the middle , along the dashed line . for a 2-body system of @xmath6 , @xmath11 , @xmath41 , etc .
the resulting relation is well known : @xmath42 the application of 2-body unitarity assumes that the pions interact only with one another , not with any spectator .
in most sets of data there are large signals where pions do interact with the spectator . for @xmath0 , as an example , the @xmath43 channel accounts for 40% of events @xcite .
some of it may be generated by pions from decays of @xmath2 or @xmath25 rescattering from the spectator ; this is a so - called triangle graph .
aitchison himself remarks that this can distort the unitarity relation .
this provides one reason why eu may fail for the first three sets of data ; it does not explain the fourth , where some further effect is required .
there are fundamental differences between elastic scattering and production . in @xmath0 , for example , matrix elements @xmath44 and @xmath45 dictate the magnitudes of these amplitudes ; any values are possible .
this differs from elastic scattering , where @xmath2 and @xmath25 magnitudes are fixed purely by their coupling constants @xmath46 to @xmath6 .
equation ( 3 ) is an asymmetric relation , allowing @xmath2 and @xmath25 to be produced with different magnitudes , but requiring that they rescatter an in elastic scattering .
a more logical alternative is the symmetric relation @xmath47 hence @xmath48 .
this relation fits afs data for central production naturally , whereas eu does not .
if the @xmath25 is absent from production data , ( 5 ) reduces to the obvious relation latexmath:[${\rm im}\ , t^\sigma = does not allow the @xmath25 to be absent , as we shall see in section 2 .
section 5 suggests a new way of fitting 2-body data .
section 6 then summarises conclusions .
in a two - body process , the scattering of a pair of pions to final states @xmath6 , @xmath11 , @xmath41 , @xmath50 and @xmath51 must satisfy unitarity .
the @xmath12-matrix for these coupled channels may be written in terms of a real @xmath7-matrix as @xmath52 it is normalised here so that @xmath53 . below the inelastic threshold @xmath54
the @xmath12-matrices used here will include couplings to all channels .
however , it simplifies the presentation of essential points to reduce the formalism initially to a single @xmath6 channel .
this simplication is sufficient to expose the basic issues , and can be generalised later to include inelasticity .
the approach of aitchison @xcite will now be outlined .
i am grateful to him for clarifying the algebra in more detail than is to be found in the original publication .
suppose @xmath15-matrices multiply , i.e. phases add .
let @xmath55 and @xmath56 be @xmath7-matrices for @xmath2 and @xmath25 respectively .
the elementary expression for @xmath57 then gives a @xmath7-matrix for elastic scattering @xmath58 from which it follows that the @xmath12-matrix for elastic scattering is @xmath59 aitchison now conjectures that an amplitude @xmath26 for producing a two - body channel present in @xmath60 may be written in terms of a vector @xmath61 , with @xmath62 where @xmath63 and @xmath64 are constants for production couplings . with this ansatz ,
the relation @xmath65 known as extended unitarity , is automatically satisfied .
it is a consequence of ( 10 ) that @xmath26 has the same phase as @xmath66 . substituting ( 8) and ( 11 ) in ( 10 ) gives , in this one - channel case @xmath67/\rho , \end{aligned}\ ] ]
where @xmath68 . from ( 13 ) , the phase of @xmath26 is indeed @xmath69 , as imposed by ( 10 ) .
equation ( 14 ) will play the decisive role in comparisons with experiment . in ( 14 ) , @xmath70 and @xmath71 . at 989 mev , @xmath72 and @xmath73 .
so both terms are very close to zero , regardless of the values of @xmath63 and @xmath64 .
this predicts that production data should have the same deep dip at this energy as elastic scattering .
there is a further point . in the second term ,
@xmath74 over a sizable mass range . in the first term
, @xmath75 should be conspicuous , since it has a rapid phase variation and the same peak magnitude as @xmath76 , which is itself clearly visible in all sets of data considered here .
however , the data all require the magnitude of the @xmath3 to be smaller than predicted .
the key point is that the factor @xmath77 of ( 13 ) leads directly to the factor @xmath78 in the first term of ( 14 ) .
the first and third sets of data will require @xmath79 of ( 14 ) to be small . in the elastic region ,
the first term becomes @xmath80 the bizarre conclusion of eu is that the phase of the @xmath3 is present even though @xmath81 .
this is inconsistent with analyticity .
it will be shown in section 3 that experiment disagrees with eu even without the constraint of analyticity .
however this additional constraint makes conclusions more definitive . for purely elastic scattering ,
the omn ` es relation @xcite reads [ including a factor @xmath82 in @xmath83 : @xmath84,\end{aligned}\ ] ] where p denotes the principal value integral .
we shall not actually need to evaluate ( 16 ) .
it plays only a conceptual role and this needs considerable explanation .
the basic point is that @xmath85 contains both real and imaginary parts , so @xmath86 determines both . for elastic scattering @xmath87
it arises from the left - hand cut , i.e. meson exchanges between the two pions . with inelasticity
, corresponding relations may be written in a 2-channel form .
then @xmath88 is replaced by @xmath89 , the angle @xmath12 makes to the real axis when measured from the origin of the argand diagram , see fig .
12 of subsection 4.1 .
if eu is valid , the production amplitude may be written @xmath90 . in principle
@xmath91 could be anything , depending on production dynamics .
however , we have quite precise experimental information about it .
an extreme view is that @xmath63 and @xmath64 of ( 13 ) can be arbitrary and complex .
however if they are complex this leads directly to a conflict with eu .
equation ( 14 ) contains two parts @xmath92 and @xmath93 .
substituting breit - wigner formulae for @xmath76 and @xmath75 , the first term becomes @xmath94 .
this is real but has a specific @xmath5-dependence in the numerator as well as in the denominator .
if @xmath63 or @xmath64 becomes complex , the numerator becomes complex .
this then introduces a phase variation separate from @xmath85 .
the `` prediction '' of eu is distorted by this extra phase .
only if @xmath91 is real does eu survive in its strict form .
many experimental groups have used the p - vector approach using complex coupling coefficients , without realising that this destroys the universality of the phase variation with @xmath5 .
this is what experiment demands , so they have done the right thing .
but the use of the universal denominator @xmath95 $ ] is no longer logically correct .
one might as well fit directly in terms of complex coupling constants and individual @xmath12-matrices for each resonance .
there is a further fundamental point . for elastic scattering
, @xmath96 is uniquely related to @xmath97 by both unitarity and analyticity . at first sight
it appears that analyticity relates @xmath91 in the same way to @xmath85 in production reactions , with the result @xmath98 , where @xmath31 is a constant .
this requires @xmath99 : if the phase of the @xmath6 amplitude is universal , relative magnitudes of @xmath2 and @xmath25 must also be universal .
there is however a caveat .
a more fundamental form of the unitarity relation ( 12 ) is that the discontinuity of @xmath26 across the elastic branch cut is @xmath100 .
then @xmath26 may be multiplied by a polynomial @xmath91 , providing it does not have a discontinuity along the real @xmath5-axis .
a few examples will hopefully clarify ideas .
firstly , a form factor in @xmath5 is one such example , arising from the sizes of particles , i.e. from matrix elements .
secondly , in @xmath101 , the e1 transition has an intensity proportional to the cube of the photon momentum ; this inflates the lower side of the @xmath3 .
thirdly , in @xmath38 @xmath102 , there is an @xmath103 centrifugal barrier for the production process .
fourthly , in some special cases , matrix elements may go through zero as a function of @xmath5 .
taking @xmath91 to be real , let us write in general @xmath104 a feature of all production data considered here is a strong low - mass @xmath6 peak due to the @xmath2 pole .
this peak is not present in elastic scattering because of an adler zero in the elastic scattering amplitude at @xmath105 , just below the @xmath6 threshold .
the elastic amplitude rises approximately linearly with @xmath5 and there is no low mass peak . the origin of the difference has been known to theorists for at least 20 years .
au , morgan and pennington @xcite pointed out that the difference between elastic scattering and central production data can be accomodated by using the same breit - wigner denominator for both , but replacing the numerator @xmath96 by something close to a constant .
this polynomial is allowed because @xmath106 is outside the physical region .
data require @xmath107 , hence @xmath108 .
more exactly , @xmath109\ ] ] for the parametrisation of the @xmath2 amplitude in @xcite . in practice , quadatic and
cubic terms in @xmath5 are very small and under tight control from fits to data up to 1.8 gev . for @xmath0 ,
the @xmath2 pole is visible by eye in fig .
4(a ) below .
the phase of the @xmath2 amplitude in this reaction is experimentally the same as in elastic scattering within @xmath17 @xcite .
values of @xmath110 can be determined directly from the data .
the same is true of the @xmath111 @xcite , which likewise has an adler zero in the numerator for elastic scattering , but not for production . in both cases
, @xmath112 is consistent within errors with a constant ; the adler zero in the numerator of elastic scattering has disappeared .
one can try fitting the @xmath2 and @xmath111 poles in production data with the conventional form factor @xmath113 , where @xmath114 is momentum in the production channel . for both
, @xmath115 optimises at slightly negative values , which are unphysical . for the @xmath2 , @xmath116 @xmath117 with @xmath118 confidence and for the @xmath111 , @xmath119 @xmath117 at the same level
. a crucial piece of information in testing eu will be relative magnitudes of @xmath2 and @xmath25 .
the magnitude of the @xmath2 amplitude is easily separated from the tail of the @xmath3 at 920 mev , three half - widths from 989 mev . if @xmath91 is determined at this energy and at 400 mev , the @xmath2 amplitude changes by @xmath120 at 989 mev for the @xmath118 confidence level quoted above .
however , this exaggerates the error , since there are compensating changes in the line - shape fitted to the amplitude .
realistically , changes are half this . adding in quadrature uncertainties in the @xmath2 line - shape due to the opening of the @xmath11 threshold , the uncertainty in the amplitude extrapolated from 920 to 989 mev is @xmath121 with @xmath118 confidence .
this provides a tight constraint on the relative magnitudes of @xmath2 and @xmath3 amplitudes if eu is correct .
this disposes of the paradox that the phase of @xmath3 can be present with @xmath122 . from this point
onwards , it will be assumed that eu should be supplemented with the condition imposed by analyticity within @xmath35 .
formulae and numerical parameters for the @xmath2 amplitude are to be found in refs .
@xcite and @xcite . the @xmath4 coupling has been fitted to four sources : ( i ) phase shifts deduced from cern - munich data by ochs @xcite , ( ii ) @xmath123 data of pislak et al .
@xcite , ( iii ) predictions of @xmath6 phase shifts by caprini et al .
@xcite using the roy equations , and ( iv ) bes data on @xmath124 @xcite .
the coupling to @xmath11 and @xmath41 has been fitted to available data on those channels @xcite , and in @xcite the @xmath50 coupling has been derived from cern - munich data . there is close consistency between all these sets of data .
because refs . @xcite and @xcite fit the same data with different formulae , the amplitudes agree within errors up to 1.2 gev . those of ref
@xcite are more cumbersome to use , since they allow for the dispersive effect of the @xmath50 threshold
. therefore the first and fourth sets of data discussed below are fitted with the formulae from ref .
@xcite .
the general procedure adopted here is to allow the parameters of the @xmath2 to have the freedom allowed in earlier determinations of its parameters , but no extra freedom in the vicinity of @xmath3 . in testing eu , the magnitude of @xmath25
is restricted to the @xmath35 discussed above ; its phase is allowed the freedom with which its parameters are known .
the @xmath3 is so narrow that it is readily separated from the @xmath2 . the bes data on @xmath125 shown in fig .
2 exhibit a very clear @xmath126 signal .
an important point is the excellent mass resolution and mass calibration of the bes detector , @xmath127 mev .
both are easily checked for the @xmath11 channel against the very precisely known parameters of @xmath128 .
data from the same publication @xcite on @xmath129 contain a clear @xmath130 peak , and the two sets of data determine accurately the ratio @xmath131 of couplings to @xmath11 and @xmath6 .
an important detail is an error in units in @xcite : @xmath132 is reported as 165 mev ; this should read 0.165 gev@xmath133 . the breit - wigner denominator for the @xmath3 amplitude is @xmath134 $ ] and @xmath135 has to be continued analytically below the @xmath11 threshold as @xmath136 . without direct information on the @xmath11 channel , this term gets confused with @xmath137 @xcite .
any form factor applied to @xmath138 adds to the confusion .
one only has to glance at the particle data tables @xcite to see the large spread in parameters fitted to the @xmath3 ( and @xmath139 ) in experiments having no direct information on the branching ratio between @xmath11 and @xmath6 .
unfortunately , the pdg does not report the bes determination of @xmath140 , which is the best in the published literature because of the availability of a clear signal in @xmath141 .
it was shown in @xcite that the bes parameters for @xmath3 are closely consistent with kloe data @xcite on @xmath142 when one allows for interference between @xmath2 and @xmath25 .
this paper determines @xmath143 .
bes data on @xmath144 confirm a large value @xmath145 for this ratio @xcite .
the @xmath146 full width at half - maximum ( 34 mev ) agrees well with cern - munich data @xmath147 mev @xcite .
it also agrees closely with the full - width of the @xmath3 signal in crystal barrel data @xmath148 mev ) .
the bes parametrisation will therefore be adopted in fitting the afs data .
further checks from belle , babar and cleo c will be very welcome .
considerable detail needs to be given of fits to experimental data , in order to pin down the disagreements with eu . the prime conclusions which will emerge are that ( i ) the @xmath6 amplitude does not follow that of elastic scattering
, ( ii ) the magnitude of the @xmath3 amplitude , relative to @xmath2 , is much smaller than predicted by eq .
( 14 ) , regardless of analyticity which also requires that their relative magnitudes should be the same within @xmath35 . -12 mm -5 mm fig .
4 displays features of bes data for @xmath0 @xcite .
there is a clear peak in fig .
4(a ) at @xmath150 gev due to the @xmath2 pole .
its shape is cleanly separated from the @xmath151 contribution up to 1.05 gev , where the @xmath152 rapidly overtakes it .
the data include a slowly varying @xmath153 background which is included in the fit .
there is also a well defined slowly varying component due to the reflection of @xmath154 .
both are shown in the experimental publication .
there are two amplitudes for production of @xmath2 and @xmath3 , with orbital angular momenta @xmath155 and @xmath156 in the production reaction .
the @xmath157 amplitude includes a centrifugal barrier for production which optimises at a radius of 0.8 fm ( roughly as expected for convolution of wave functions of @xmath2 , @xmath158 and @xmath25 ) . for the isobar model fit
, different magnitudes are allowed for @xmath155 and 2 , with complex coupling constants , though it turns out that the @xmath157 contribution from @xmath3 is negligible
. in the eu fit the relative phases of @xmath2 and @xmath25 components are constrained to be the same , and relative magnitudes are constrained to be the same within @xmath35 . in figs .
4(a ) and ( c ) , the lower histogram shows the @xmath159 intensity : from the isobar model fit in ( a ) and from eu in ( c ) .
the upper histogram , close to data points , shows the result of the full fit .
unfortunately , the mass projection alone is not definitive , for reasons explained shortly .
one low point in the mass projection just above 1 gev hints at a dip following the eu prediction .
however , there is a great deal more information contained in the dalitz plot of fig .
4(b ) and also in the correlation between the @xmath158 decay plane and the @xmath160 production plane . in the dalitz plot , there are strong horizontal and vertical bands due to @xmath154 .
these bands interfere with @xmath2 and @xmath151 ( and @xmath3 if present ) .
note that the @xmath154 decays mostly through @xmath15-waves , so its intensity would be almost constant across the dalitz plot in the absence of interferences .
also note from the lower histogram of fig .
4(a ) that the @xmath2 amplitude at 950 mev is sizable .
cross - hairs on fig .
4(b ) show where a @xmath6 pair of mass 1 gev intersects the @xmath161 band .
if @xmath3 were present with the same magnitude as @xmath2 and with the phase predicted by eu , a large but narrow interference with @xmath161 can not be avoided due to @xmath3 , whatever the relative phase of @xmath154 .
it should have a full width of @xmath162 gev@xmath133 along the @xmath161 band : the line - width of @xmath126 .
there should also be a dip somewhere along the diagonal at @xmath163 gev .
4(d ) shows an enlarged view of this region .
there is no sign of these features .
the @xmath158 decay plays an important role .
the spin of the @xmath158 lies along the normal to its decay plane ; information on this decay plane is a key ingredient in determining helicity amplitudes .
these are essential to determine interferences between the amplitudes for @xmath164 , @xmath165 and @xmath43 .
there are in principle five @xmath151 amplitudes corresponding to production with @xmath155 ( one amplitude ) , @xmath166 ( three ) and @xmath167 ( one ) .
it turns out that the @xmath167 amplitude and one of the three @xmath166 amplitudes are negligible .
there are large interferences between @xmath2 and the remaining three @xmath151 amplitudes and even larger interferences with @xmath154
. with the eu hypothesis , the best fit to the dalitz plot ( upper histogram ) and the @xmath158 decay plane fills in the predicted dip at @xmath168 gev@xmath133 with other interferences , see fig .
4(c ) ; however , the price is a considerable deterioration of log likelihood compared with the fit of fig .
the isobar model fit is better than eu by 259 in log likelihood .
there are two reasons .
firstly , the narrow @xmath25 does not appear in interferences with either @xmath161 or @xmath152 .
the mass resolution of the bes data is 4 mev in @xmath6 at 950 mev ; searching for the @xmath3 with accurately known line - shape is limited purely by statistics and there are @xmath169k events .
secondly , in the fit based on eu , there are strong conflicts between interference terms amongst @xmath43 , @xmath165 and @xmath164 .
the data want slowly varying interferences between @xmath161 and the broad @xmath152 , instead of rapidly varying interferences of the narrow @xmath25 with @xmath161 and/or @xmath152
. there are two extra fitting parameters for the isobar model fit : i.e. two complex coupling constant for @xmath164 instead of one .
[ the @xmath170 @xmath157 amplitude is negligible ] .
the definition of log likelihood is such that a change of 0.5 corresponds to a one standard deviation change in one variable . for large statistics ,
@xmath171 is approximately twice the change in log likelihood .
so the fit of fig .
4(a ) is better than eu by 19.7 standard deviations statistically .
it is important to remark that the bes publication gives a second fit done independently .
this shows a @xmath6 mass projection almost identical with the scalar form factor .
this requires a very strong @xmath3 , in conflict with what one can see by eye in fig .
however , this analysis made no use of the @xmath158 decay plane . without that information ,
it is impossible to disentangle the magnitudes of the five @xmath151 amplitudes ; the determination of the magnitude and phase of the @xmath3 signal becomes very poor .
i have rechecked the analysis omitting information from the @xmath158 decay plane .
no stable fit emerges with the @xmath6 mass projection of the second fit in the bes publication , and there is no evidence for the presence of @xmath3 at all in this fit .
the only explanation i can find of this second bes fit is that the relative magnitudes of @xmath3 and @xmath2 have been constrained to agree with the scalar form factor .
this is , of course , exactly what needs to be checked in the present work .
-12 mm -0.5 cm .
5 shows the argand diagram for @xmath172 from my fit to @xmath2 and @xmath25 ; the factor @xmath9 is included so that the diagram goes to zero at the @xmath6 threshold .
the ratio of @xmath25 and @xmath2 amplitudes is @xmath173 at 0.99 gev and the @xmath25 lags the @xmath2 by @xmath174 . at the top of a large loop due to the @xmath2 ,
there is a small structure due to @xmath3 . for comparison purposes ,
the dashed curve shows the @xmath2 amplitude alone and the dotted curve shows the eu prediction normalised to the full curve at 0.47 gev , the peak of the @xmath2 .
the essential point is that the deep dip in this latter curve near the @xmath25 mass is missing from the data .
there is one qualitative feature of fig . 5 which will be important later .
the amplitudes in a production process are free to add vectorially in any way required by the data .
this situation is quite different from elastic scattering , where the amplitude is constrained to the unitary circle .
it is now necessary to consider a number of systematic questions which might blur the argument against eu .
corresponding remarks will apply to fits to the other three sets of data .
firstly , could the disappearance of the predicted @xmath3 arise from effects of the @xmath11 threshold ?
the answer is no , because the @xmath25 appears clearly in elastic scattering and the prediction of eu is that it should be almost identical in production . a precise cancellation of the magnitude and phase of the predicted @xmath25
would be needed .
this is ruled out by the known couplings of both @xmath2 and @xmath25 to @xmath11 .
the analytic continuation for the @xmath3 below the @xmath11 threshold is accurately determined over the small mass range concerned by bes data on @xmath175 and @xmath141 . in the @xmath2 amplitude ,
the @xmath11 inelasticity rises over a mass scale of 200 mev as shown in fig .
11 of ref .
any flexibility in the analytic continuation of the @xmath11 term then has a scale of @xmath176 mev .
furthermore , it is closely constrained by the fit to kloe data on @xmath177 .
so this explanation is highly implausible .
secondly , could the bes data be fitted assuming @xmath178 , followed by @xmath179 ? this has been tested by adding @xmath180 fitted freely .
its magnitude optimises at zero within errors .
it improves log likelihood only by 2 .
thirdly , aitchison argues @xcite that a further term @xmath181 might be added to the @xmath61-vector due to dynamics of the production process .
a similar approach was used in fits by bowler et al . and basdevant and berger to the @xmath182 to allow for the deck effect @xcite @xcite @xcite .
an additional term @xmath183 appears in @xmath184 .
if @xmath185 , the second term can cancel the term @xmath186 of eu , leaving @xmath76 and the first of the additional terms , which is close to 0 since @xmath187 .
this removes almost all the structure due to @xmath3 .
however , this cancellation also leaves a term @xmath188 and when @xmath189 is small this becomes @xmath190 , which is much larger than the term @xmath76 itself if @xmath181 is a constant .
this is ruled out by the data , so it becomes necessary to tailor the @xmath5-dependence of @xmath181 to reproduce the magnitude of the @xmath6 amplitude . however , this is not the end of the story .
phase information is also available . in @xcite
it is shown that interference between @xmath2 and @xmath154 measures the phase of the @xmath2 amplitude ( plus any background term @xmath181 ) and requires it to be the same as for elastic scattering down to 450 mev with an error of @xmath191 in the worst scenario , see fig . 2 of that paper .
it is worth mentioning that there is a similar result for the @xmath111 in @xcite .
4(a ) of that paper shows that the @xmath111 phase from bes data on @xmath192 agrees with the lass effective range formula for elastic scattering down to 750 mev within @xmath193 .
4(f ) of the same paper shows a corresponding agreement for e791 data on @xmath194 within a similar error .
these results rule out any background different from @xmath2 and @xmath111 with a magnitude larger than @xmath35 and a phase difference above @xmath195 .
they of course agree with watson s original statement that @xmath85 of a single resonance should be universal .
there is no obvious source of the background proposed by aitchison , as there was in @xmath196 , where a deck background was visible in the data and made a @xmath197 contribution to cancelling the @xmath182 amplitude . a similar fortuitous cancellation with unidentified backgrounds
will be required also for all three of the following sets of data . in all cases ,
the additional background gives rise to a phase variation different to strict eu .
if the background removes the @xmath25 , only the @xmath2 is left , i.e. @xmath198 , which contributes only part of the phase of elastic scattering ; so the amplitude does not satisfy eu .
-13 mm -0.7 cm a fresh analysis of crystal barrel data on @xmath200 annihilation at rest to @xmath201 has been completed recently @xcite .
data are available with statistics of @xmath202k events ( and @xmath203 experimental backgrounds ) in both liquid and gaseous hydrogen ; these allow a good separation of @xmath39 and @xmath38 initial states .
although the @xmath3 appears clearly in @xmath39 annihilation , the magnitude of the @xmath6 s - wave amplitude does not go to zero on the argand diagram near 1 gev , as eu predicts .
this argand diagram is reproduced in fig .
its magnitude is smallest at 0.98 gev , but is still very distinct from zero .
for @xmath38 annihilation , the fitted @xmath3 amplitude , relative to @xmath2 , is much smaller , see fig .
the magnitude of the @xmath2 amplitude is quite large near 1 gev .
[ htb ] -16 mm -0.7 cm the question arises how robust these solutions are .
could they ` bend ' to accomodate eu ? the published analysis requires some @xmath5-dependence of the numerator fitted to the @xmath2 amplitudes , of the form @xmath204 , with complex @xmath205 .
can extra flexibility reach agreement with eu ?
a second point is that the analysis does not include the repulsive @xmath206 @xmath6 s - wave .
could this bring conclusions into line with eu ?
the brief answer is definitely no , and will be presented graphically .
7(a ) shows the @xmath6 mass projection for data in liquid from the current analysis ; it fits the data points accurately . in fig .
7(b ) , the @xmath207 combination of the isobar model is replaced by the eu combination ( with the constraint from analyticity that magnitudes of @xmath2 and @xmath3 should be equal within @xmath35 ) .
initially , only the coupling constant of this combination is refitted , leaving other amplitudes untouched ; this is for the purpose of illustrating the change required by eu .
the @xmath2 component is large and can not change much ; a deep dip appears at 1 gev because of the corresponding dip in the elastic amplitude .
it is of course necessary to re - optimise all components .
the resulting mass projection is shown in fig .
7(c ) and is still in severe disagreement with the data .
a measure of the disagreement may be obtained from @xmath208 . here , it is necessary to point out that even the fit of fig .
7(a ) has a @xmath208 larger than 1 .
this is probably because of the enormous statistics and small , slowly varying systematic errors in acceptance .
the procedure adopted here is to renormalise @xmath208 to 1 per point for this fit and apply the same scaling factor to all other fits .
the fit of fig .
7(c ) then has a renormalised @xmath208 of 40599 for 3500 bins ; this is a 170 standard deviation discrepancy , allowing for the reduction in the number of fitted parameters by 2 .
a much better fit is possible if relative magnitudes of @xmath2 and @xmath3 are allowed to vary .
the phase of the s - wave amplitude in fig .
6(a ) is close to eu , and only a 10 mev shift is required for a perfect fit .
this is consistent with the energy calibration and resolution of the crystal barrel detector .
however , the fitted combination of amplitudes no longer agrees with the crucial equation ( 14 ) .
7(f ) shows the effect of including the @xmath209 s - wave amplitude , using the formulae of section 4.1 of ref .
there is only a rather small improvement , because the slow @xmath5-dependence of this amplitude can not fill the narrow dip at 1 gev .
the renormalised @xmath208 falls to 34345 , a discrepancy of @xmath210 .
-12 mm -0.7
cm it is not just the @xmath6 mass projection which governs @xmath208 .
one should inspect discrepancies in @xmath208 all over the dalitz plot .
8 makes such a comparison .
panel ( a ) shows discrepancies in @xmath208 where the fit is above the liquid data and ( b ) the discrepancies where the fit is low .
panels ( c ) and ( d ) show results for data in gas .
one sees striking systematic discrepancies all over the dalitz plot , arising from interferences .
such discrepancies are almost completely absent from the fit of fig .
7(a ) using the isobar model .
there are two further points for discussion . in the work shown on figs .
7(e ) and ( f ) , the @xmath209 s - wave was fitted without factoring out the adler zero which occurs at @xmath211 . if this is done ( as for the @xmath2 amplitude )
, the broad @xmath209 amplitude gets confused with the @xmath2 amplitude and leads to minor improvements all over the dalitz plot .
however , none of these is distinctive enough to require the @xmath209 amplitude definitively .
secondly , could a more complicated polynomial than @xmath204 multiply the @xmath2 amplitude and give a successful fit ?
extensive tests were made in @xcite with the objective of improving the fits reported there .
if one chooses too free a polynomial , the fitted @xmath38 component in gaseous hydrogen can fluctuate wildly from the 50% predicted from stark mixing by reifenrofer and klempt @xcite . to avoid this ,
the fitted @xmath38 component is constrained within the range @xmath212 . unless the numerator of the eu amplitude is designed to include a narrow dip at 1 gev , no large improvement is observed .
one further observation from the crystal barrel data is worth reporting .
relative intensities of @xmath213 and @xmath151 are quite different in @xmath200 annihilation to those in elastic scattering . in @xmath200 data ,
the integrated intensities of these two resonances are equal within @xmath35 after allowing for the ( modest ) effects of centrifugal barriers for production ( @xmath166 for @xmath39 annihilation and @xmath103 for @xmath38 ) . however , in elastic scattering the fitted @xmath214 width is a factor 4 smaller than that of @xmath215 .
this again disagrees with eu .
central production of a @xmath6 pair in @xmath216 was fitted using eu by au , morgan and pennington ( amp ) @xcite . in data from the afs experiment at the isr @xcite
, the two final - state protons are produced with very small 4-momentum transfers @xmath217 gev@xmath133 .
it is routinely assumed that the @xmath6 pair is unaffected by final state interactions with these protons , which are separated from the central region by a gap in rapidity .
-15 mm -6 mm amp draw attention to structure in the @xmath6 mass spectum similar to the dip in the @xmath6 s - wave elastic cross section .
9 shows the afs data as open squares . there is a peak at low @xmath6 mass , close to that of the @xmath2 pole in bes data , but not quite identical , for reasons discussed shortly .
triangles on fig .
9 show @xmath6 elastic cross sections derived from cern - munich phase shifts ; this is done by dividing the @xmath6 amplitude of elastic scattering by @xmath87 , leaving only the term @xmath218 .
the curve shows my fit to elastic data after dividing the @xmath6 amplitude by @xmath96 .
the agreement between the curve and triangles demonstrates that the parametrisation of the @xmath2 reproduces cern - munich phase shifts .
results are similar to the afs data , but there is a distinct difference in the vicinity of @xmath3 .
what is clearly evident is constructive interference between @xmath25 and @xmath2 immediately below 989 mev , where eu predicts a zero . even without fitting
, one can see that eu will fail .
a detail is that , up to the @xmath11 threshold , one piece of information from each cern - munich moment is sufficient to determine phase shifts for both s- and p - waves . above the @xmath11 threshold , however , the separation of inelasticity parameters @xmath219 and phase shifts @xmath220 can not be made without further assumptions . just above the @xmath11 threshold , the solution of ochs @xcite has @xmath2210.6 - 0.7 , whereas the bes line - shape for @xmath3 demands @xmath219 parameters dropping to @xmath222 at 1.01 gev . in view of this large discrepancy
, predictions from cern - munich phase shifts are not shown above the @xmath11 threshold .
-12 mm -6 mm since the work of morgan and pennington @xcite , information on both @xmath2 and @xmath3 has improved enormously .
the @xmath2 amplitude is taken ( within errors ) from ref .
@xcite , using fit ( iii ) given there .
the resulting cross section is shown by the chain curve of fig .
although this fit follows the general features of the data , it is not accurate for low @xmath6 masses .
varying parameters of @xmath3 within errors quoted by bes has negligible effect on the quality of the fit .
some additional flexibility is clearly needed in the numerator @xmath91 of the @xmath2 amplitude .
there are two obvious sources .
firstly , amp point out that central production goes via two intermediate pomerons : @xmath223 , where @xmath224 exchange is allowed .
this alone does not achieve a good fit .
the second explanation arises from regge factors in the production process .
amp argue for an @xmath5-dependent factor @xmath225 in the numerator arising from the leading regge trajectory .
there may be contributions also from a daughter trajectory . the full curve of fig .
10(a ) shows an isobar model fit using a complex constant for @xmath3 , but without any @xmath226 .
it uses a scaling factor for the cross section shown by the dashed curve .
this takes the form of a gaussian dip : @xmath227 .
this scaling factor gives rise to a slow modulation of the @xmath2 amplitude over hundreds of mev .
it has only a small effect on what is fitted to @xmath3 .
a detail is that a gaussian mass resolution of 10 mev , quoted by the afs collaboration @xcite , is folded into each point of the fit ; it is significant only near the @xmath11 threshold .
the fit to @xmath6 data has a @xmath208 of 35.2 for 34 degrees of freedom .
consider now the @xmath3 mass range .
the isobar model fit of fig .
10(a ) requires an @xmath25 magnitude only @xmath228 of the eu prediction .
this fit requires constructive interference between @xmath2 and @xmath25 below the @xmath25 mass and destructive interference above .
it implies the @xmath25 phase is @xmath229 below the eu prediction .
this immediately throws doubt on eu .
a fit using eu is conspicuously bad . as an example , if the first 29 points up to 988 mev are fitted alone , @xmath230 if fitted by @xmath2 and @xmath3 only .
eu predicts an amplitude which is almost zero at 989 mev , whereas the data at 988 mev are far above this .
if the eu fit is extended to 1300 mev including the @xmath226 with the elasticity fitted in @xcite , @xmath208 is @xmath231 , because the @xmath226 contribution is far too large . even if the @xmath226 is fitted freely in magnitude , @xmath232 for 37 points .
the fit , shown by the chain curve of fig .
10(b ) is particularly bad close to the @xmath11 threshold , where the dip of elastic scattering is predicted by eu .
the fit to @xmath11 data is also poor , with a @xmath208 of 18 for 5 points . the dotted curve of fig .
10(b ) shows the effect of fitting freely an additional contribution from the @xmath206 s - waves : @xmath233 = 59.6 . this fails to cure the poor fit .
it makes almost no difference whether the @xmath206 amplitude is divided by a factor @xmath234 like the @xmath2 amplitude .
the basic difficulty is that the slowly varying @xmath206 amplitude can not cure the rapid structure due to @xmath3 .
furthermore , the relative magnitudes of the fitted @xmath206 amplitude and the eu amplitude is 0.46 , whereas it is only 0.18 for elastic scattering at 1 gev .
such a large @xmath206 amplitude is implausible .
obviously the problem with eu is that the magnitude of the @xmath25 amplitude needs to be smaller than for elastic scattering . from analyticity
, this also requires a difference in phase .
that is also clear from the absence of the predicted dip at 989 mev in the data . if the phase of the @xmath3 amplitude is constrained to the eu value , but relative magnitudes of @xmath2 and @xmath25 are set free , @xmath235 , which is bad .
most of @xmath171 comes from points immediately around the @xmath11 threshold , showing that the data reject also the phase variation of eu . at this point
, one could argue that the mechanism of the production reaction is unknown and might generate a phase for @xmath3 different to that of the @xmath2 .
this argument is not specific , though the isobar model can fit the data well .
however , the usual argument for a different phase for @xmath25 and @xmath2 is multiple scattering of the pions with spectator particles . in afs data
, there is an empty rapidity gap isolating the central region .
remember also that cern - munich phase shifts are derived in the first instance from data on @xmath236 at high momentum and small momentum transfer , a similar configuration .
the conjecture of eu will now be replaced with an alternative ansatz .
the treatment of production data needs to be able to cope with the case where one resonance amplitude is zero .
eu does not , since a universal phase equal to elastic scattering requires a production amplitude @xmath237
. the correct production amplitude should reduce to @xmath238 when the @xmath25 is absent .
a small @xmath25 amplitude should produce a small perturbation to @xmath198 .
suppose the 2-body @xmath6 amplitude is written as @xmath239,\ ] ] where @xmath240 and @xmath79 are real .
this allows freedom in @xmath79 and includes a phase @xmath241 which becomes the same as for elastic scattering when @xmath242 .
it is necessary to choose as a the state with the larger amplitude on resonance , so that @xmath243 .
if one could create this @xmath6 system in ` free space ' , the appropriate 2-body unitarity relation below the @xmath11 threshold would be @xmath244 an alternative way of formulating the basic physics ( with the same outcome ) is in terms of the schwinger - dyson equation . instead of the conventional relation @xmath245 my conjecture is to replace this with @xmath246 here @xmath247 is the ` potential ' generating the final state and @xmath248 is the propagator . for the case of purely elastic scattering , a closed form for @xmath241 of ( 19 ) may be derived by substituting @xmath76 and @xmath75 in the form @xmath249 into ( 20 ) .
after simple cancellations between left- and right - hand sides , @xmath250 or @xmath251 if @xmath252 , this is a different relation from purely elastic scattering .
the improvement in the fit is dramatic .
immediately an excellent fit to points below the @xmath11 threshold is obtained with @xmath253 for 29 points and 24 degrees of freedom .
the term @xmath254 in ( 24 ) differs from @xmath189 by @xmath255 .
this is just what is needed to produce the interference between @xmath2 and @xmath25 observed in the isobar model fit . above the @xmath11 threshold ,
it is tempting to satisfy unitarity by introducing a @xmath7-matrix .
however , the @xmath7-matrix depends on the assumption that the 2-body system is confined to the unitary circle , but that is no longer the case in a 3-body situation .
the fit may be extended above the inelastic threshold by writing @xmath6 and @xmath11 amplitudes as @xmath256/\rho _ 1 \\ & = & \alpha \left [ t^\sigma_{11 } \left(1 + \frac { \gamma g_2^\sigma r}{g_1^\sigma } \right ) + t^f_{11 } \left(\beta + \frac { \epsilon g_2^f r}{g_1^f } \right ) e^{2i\psi } \right]\big/ \rho_1 \\\nonumber f_{kk } & = & \alpha [ t^\sigma_{12 } + \gamma t^\sigma_{22 } + ( \beta t^f_{12 } + \epsilon t^f_{22 } ) e^{2i\psi ' } ] /\rho _ 2 \\ & = & \alpha \left [ t^\sigma_{11}\frac { g^\sigma_2 } { g^\sigma_1 } \left ( \frac { 1}{r } + \frac { \gamma g_2^\sigma } { g_1^\sigma } \right ) + t^f_{11 } \frac { g^f_2}{g^f_1 } \left(\beta + \frac { \epsilon g_2^f } { g_1^f } \right ) e^{2i\psi ' } \right]\big/ \rho _ 1.\end{aligned}\ ] ] eqs .
( 25 ) and ( 26 ) expose the explicit dependence of @xmath180 on the ratio @xmath257 ; the experimental group divides out the phase space @xmath258 and @xmath259 in the @xmath6 and @xmath11 channels . as explained in section 2.1 , all @xmath12 have the numerator of elastic scattering replaced by a constant .
my @xmath180 is defined so as to contain a factor @xmath260 and @xmath261 is defined to contain a factor @xmath258 . with these definitions ,
the unitarity relations become @xmath262 eqs . ( 25 ) and ( 26 ) satisfy these relations by construction , except that @xmath263 and @xmath264 need to be constrained to obey ( 27)-(29 ) . above the @xmath11 threshold , this is done using freely fitted @xmath263 and @xmath264 for every data point and introducing into @xmath208 a penalty function which applies ( 27)-(29 ) with @xmath265 errors ; in practise these constraints are easily satisfied and discrepancies at the end of the fit are below the @xmath266 level ; this is well below experimental errors .
in fact the @xmath11 data are not very precise , leaving large flexibility in @xmath264 above the @xmath11 threshold .
in other words , the data are easy to fit above the @xmath11 threshold , but highly definitive below it .
a detail is that @xmath267 needs to include form factors both below and above the @xmath11 threshold , such that it falls quite rapidly on both sides of the threshold ; formulae are given in @xcite .
-16 mm -6 mm the best fit with @xmath2 and @xmath3 alone has @xmath268 for 37 points and 32 degrees of freedom .
the fit is good up to 1.1 gev , but is inadequate for @xmath4 data near 1.3 gev .
this may be cured straightforwardly by adding a small @xmath226 component .
a technical detail is that the @xmath226 amplitude is multiplied by a factor @xmath269 and ( 24 ) is iterated ; the contribution of @xmath226 to @xmath180 and @xmath261 is negligible .
the resulting fit , shown on fig .
11(a ) , has @xmath270 and the @xmath226 amplitude is 0.18 times that of the @xmath2 amplitude at 1.3 gev .
the @xmath208 for @xmath6 data is 28.7 for 29 degrees of freedom .
the omega collaboration reports a significant contribution from @xmath226 to their data on central production of @xmath271 @xcite .
their fitted mass and width agree closely with the line - shape fitted to @xmath226 in ref .
@xcite .
11(b ) shows the fit to afs @xmath272 data .
a detail here is that the @xmath272 data of afs are scaled up by a factor @xmath273 to allow for isospin clebsch - gordan coefficients in @xmath274 and @xmath272 systems .
the @xmath11 data constrain the coefficients of @xmath180 and @xmath261 amplitudes .
the lowest afs @xmath11 point has small acceptance which may have significant systematic uncertainty @xcite .
-8 mm -4 mm fig .
12 shows the argand diagram for the fitted amplitude .
this illustrates the form of eq .
there is a geometrical relation between the imaginary part of the amplitude and its modulus squared , but it is a different relation to eu . the @xmath3 amplitude is smaller than that of the @xmath2 and their relative phases are different to eu .
the same is true of fig .
5 , the argand diagram fitting bes data ; there the @xmath3 contribution is very small .
let us now return to equations ( 20 ) and ( 24 ) and review their general features .
the last term of ( 24 ) , @xmath275 as @xmath242 and @xmath276 .
furthermore it approaches this limit non - linearly as @xmath242 .
the interesting point is that two - body elastic scattering emerges as a limiting case of a more general relation .
furthermore it has pathological properties as the relative magnitude of a and b crosses 1 ( or -1 ) . if b becomes larger than a , it is necessary to interchange the roles of a and b. as @xmath79 approaches 1 from below , @xmath277 and one recovers the standard result of elastic scattering .
further pathological cases arise if @xmath278 ( impossible in the 2-body elastic case ) .
elastic scattering is in fact a very special case .
so eu can fail very badly as @xmath79 departs from 1 . as @xmath79 drops from 1 , the phase @xmath263 measured from the origin of the argand diagram soon changes by only a modest amount over the @xmath25 . in this case
, the isobar model becomes an excellent approximation : the @xmath25 has its usual dependence on @xmath5 through its phase shift @xmath220 and the @xmath25 amplitude adds vectorially to the @xmath2 ; the isobar model can then fit a constant phase to both quite successfully if the term @xmath279 is small . in summary
, the modified form of eu suggested in ( 20 ) and ( 24 ) gives a much better fit than eu .
the fit of morgan and pennington @xcite using eu requires an additional third - sheet pole at @xmath280 mev .
this additional pole can not be accomodated by bes @xmath175 data , which require only a second - sheet pole at @xmath281 mev and a broad third - sheet pole at @xmath282 mev .
these two poles have a natural explanation .
if the coupling of @xmath3 to @xmath11 is allowed to decrease gradually to zero , leaving other parameters unchanged , the two poles coalesce towards the same pole position @xmath283 mev ; it is the effect of @xmath284 which moves the second - sheet pole to the @xmath11 threshold and moves the third - sheet pole away .
there is no pole in this mass range from the @xmath2 amplitude .
a second remark is that all of @xmath2 , @xmath111 , @xmath139 and @xmath3 may be reproduced as a nonet by the model of rupp and van beveren , where mesons couple to a quark loop @xcite . in this model ,
no additional pole like that of morgan and pennington appears in the mass range close to the @xmath11 threshold .
these two results suggest that the additional pole is a consequence of the constraint of eu .
the relation ( 20 ) is a new conjecture .
are there forseeable snags ?
the @xmath2 and @xmath25 may mix , and this mixing could be different in elastic scattering and production .
this mixing would alter the apparent width of @xmath3 and could induce an additional phase change relative to @xmath2 .
presently there is no indication of any need for mixing .
such mixing will be absent if @xmath2 and @xmath3 have strictly orthogonal wave functions , as is plausible for members of the same nonet . on fig .
2 , the @xmath2 is definitely visible in @xmath175 data
. it would not be surprising if @xmath175 and @xmath160 channels filter out orthogonal combinations of @xmath2 and @xmath25 . from fig .
2 , one can estimate the relative intensities of @xmath2 and @xmath25 .
it is necessary to allow for the mass resolution , since the @xmath25 amplitude falls extremely rapidly from its peak at 989 mev , particularly above the @xmath11 threshold . doing this ,
the intensity of @xmath2 is @xmath285 of @xmath25 at the peak , i.e. @xmath120 in amplitde .
this is marginally higher than the @xmath25 signal fitted to @xmath160 but within the error , supporting the idea of orthogonal amplitudes .
many authors use the @xmath7-matrix to satisfy unitarity for 2-body scattering , e.g. the coupled channels @xmath6 , @xmath11 , @xmath41 , @xmath286 and @xmath50 .
the popular approach is to add @xmath7-matrices of all resonances appearing in one partial wave .
however , if resonances overlap , as @xmath2 and @xmath3 do , the @xmath7-matrix poles occur at masses where combined phase shifts happen to go through @xmath21 , @xmath22 , etc . ,
i.e. at @xmath287 and 1200 mev .
firstly , an expansion in terms of these poles is problematical for @xmath3 unless other factors or high powers of @xmath5 are included .
secondly , the relation between @xmath7-matrix poles and @xmath12-matrix poles is obscure .
any one @xmath12-matrix pole is built up from all @xmath7-matrix poles ; the converse is obvious .
the prescription that @xmath15-matrices multiply below the inelastic threshold does not appear naturally , but has to be enforced by fitting data .
an attractive alternative can be constructed following the spirit of aitchison s approach ( for 2-body scattering ) .
suppose one combines two resonances according to the prescription @xmath288 below the @xmath11 threshold , this automatically gives the result that phase shifts add .
[ further resonances may be combined by iterating this prescription ] . a nice feature of ( 30 )
is that one can write @xmath289 using the same mass @xmath290 as the usual breit - wigner denominator .
a second attractive feature of ( 31 ) is that the amplitude continues naturally through the @xmath11 threshold , because of the factor @xmath291 in the denominator . my own approach in several papers ,
@xcite has been close to this .
all diagonal elements of @xmath15-matrices are multiplied , as proposed here .
magnitudes of off - diagonal elements of the @xmath15-matrix need to be calculated from unitarity relations .
for example , for a 3-channel system : @xmath292 the phase of these off - diagonal elements has been fitted empirically , whereas ( 30 ) would predict these phases .
this approach successfully fits elastic data , @xmath293 and @xmath41 with one proviso : a good fit requires inclusion of mixing between @xmath2 , @xmath226 and @xmath294 @xcite , using the formulae of anisovich , anisovich and sarantsev @xcite ; these formulae are the modern equivalent of the breit - rabi equation of molecular spectroscopy , generalised to include resonance widths .
crystal barrel data have @xmath3 and @xmath2 components with relative magnitudes seriously different to those predicted by eu plus analyticity .
furthermore , they are different in @xmath295 and @xmath38 annihilation . the bes data for @xmath124 do not reproduce the deep dip of elastic scattering at 989 mev .
afs data likewise do not contain the same dip at this mass .
all these results are in conflict with equation ( 14 ) , which is a direct consequence of eu .
this shows unambiguously that there must be a major flaw in the hypothesis of eu .
experimentalists have dealt with this problem by using complex coupling constants for each resonance . however , as emphasised in subsection 2.1 , the imaginary part of the coupling constant introduces into the numerator of the amplitude an @xmath5-dependent phase variation which alters the universal phase coming from the denominator @xmath296 $ ] .
this destroys the original idea of a universal phase .
one might as well fit directly in terms of the @xmath12-matrix of each individual resonance , along the lines outlined in section 5 .
the form of the @xmath7-matrix suggested there would eliminate differences between @xmath7-matrix and @xmath12-matrix poles , making interpretation of results more direct .
not all experimentalists adopt the p - vector approach .
some fit directly in terms of individual @xmath12-matrices for each resonance , including sequential decays from one resonance to a daughter with different complex coupling constants for each decay mode .
ascoli and wyld @xcite and schult and wyld @xcite consider a multiple scattering series of the type @xmath297 , etc , where @xmath298 is a 3-body resonance and brackets indicate resonances in two - body sub - systems ; this is a unitarity effect of a different form to that considered here . in view of the failure of eu in the 4 cases considered here ,
each new set of data should be inspected on its merits .
let us examine ways of trying to save eu .
firstly , it is possible that unspecified backgrounds can be added to the p - vector so as to side - step the conflict .
however , analyticity independently limits relative magnitudes of @xmath3 and @xmath2 within @xmath35 .
experimental determinations of @xmath2 and @xmath111 phases in @xcite and @xcite constrain phases within @xmath195 .
the probability that unspecified backgrounds can evade eq .
( 14 ) to this accuracy in four sets of data is vanishingly small .
furthermore , if such backgrounds are introduced , eu loses any predictive power .
secondly , crystal barrel data and afs data can not be fitted with eu whether or not the @xmath209 s - wave amplitude is included .
so this is not a satisfactory escape route .
a third likely possibility , applicable to the first three sets of data , is that pions from sigma and/or @xmath3 rescatter from the spectator particle , leading to a breakdown of eu .
aitchison himself pointed this out .
today , we known that such graphs have magnitudes typically @xmath299 of those of the parent processes before the rescattering .
this is sufficient to introduce large phase changes in some cases , but not all . for all of these three sets of data
, the isobar model provides an excellent fit .
the production amplitude is then written @xmath300 with complex @xmath63 and @xmath64 . in this form , no vestige remains of the constraint that @xmath15-matrices must multiply as in elastic scattering .
relative magnitudes of @xmath63 and @xmath64 can arise from matrix elements coupling the initial state to each resonance . for @xmath0 and @xmath301 ,
the @xmath25 component is so small that one can not tell whether the phase alone follows eu or not .
however , for @xmath302 , the @xmath25 signal is large enough to rule out this possibility .
the fourth point is that one would still expect eu to work for afs data , but it does not .
an excellent fit may be obtained by replacing the unitarity relation @xmath303 by the new relation @xmath304 this corresponds to the relation @xmath305 rather than the commonly used form @xmath306 .
equations ( 20 ) and ( 24 ) have pathological behaviour in the vicinity of the 2-body elastic limit @xmath99 .
one can now see the basic problem of eu .
it attempts to impose on the 3-body system a very special behaviour which is narrowly restricted to 2-body scattering . away from the special case @xmath99 ,
the isobar model works successfully .
there are two points about the new unitarity relations .
firstly , it was argued in section 2.1 that a universal phase in the denominator of the amplitude also requires , via analyticity , almost universal magnitudes ; the word ` almost ' covers the possibility that there may be slowly varying form factors or centrifugal barrier factors in production reactions without corresponding changes to @xmath85 .
if relative magnitudes of resonances differ by large amounts between 2-body scattering and production , their relative phases _ must _ also change .
secondly , the new unitarity relation ( 33 ) succeeds quantitatively in accounting for the observed relative phase between @xmath2 and @xmath3 in central production . that is a non - trivial result .
the fit to afs data then requires only two poles in the vicinity of @xmath3 , in agreement with the bes line - shape ( as does the isobar model ) .
eu requires an extra pole for which there is no obvious explanation .
the form of this new unitarity relation is illustrated by the argand diagrams of figs .
5 and 12 . in both ,
the @xmath3 amplitude is small or fairly small and so is @xmath254 of eq .
( 19 ) . as a result ,
the phase @xmath263 measured from the origin of the argand diagram changes rather little over the @xmath25 .
this is an extra source of phases appearing in the isobar model , and has not been appreciated before .
however , one then needs to ask whether this new relation can be used universally in the isobar model .
does it , for example , correctly describe the relative phases of @xmath2 and @xmath25 in @xmath307 data and in @xmath24 ?
the answer is no .
for these two reactions , the agreement between data and equations ( 20 ) and ( 24 ) improves substantially over eu .
however , there are still discrepancies with the new unitarity relation of @xmath308 , which is still significant .
it seems likely that rescattering of pions from the spectator introduces some additional phases .
the new unitarity relation needs to be tested elsewhere .
a possible testing ground is in kloe data on @xmath309 , where both @xmath2 and @xmath25 may contribute , but the decay is electromagnetic ; the small amplitude from @xmath310 introduces a perturbation of only @xmath285 in amplitude and only in well defined parts of the dalitz plot .
the remedy which succeeds well in fitting nearly all data is the isobar model , where both magnitudes and phases of resonances are both fitted freely .
it needs to be emphasised that experimental groups have adopted the flexibility needed to fit existing data , so their results are essentially sound and are not in question .
the hypothesis of eu has mostly been adopted by theorists for making predictions .
those predictions now need to be viewed with suspicion .
although the scalar form factor is well determined for elastic scattering , it is dangerous to assume that this form factor is universal and can predict production processes .
i am greatly indebted to i.j.r .
aitchison for extensive and illuminating comments .
i am also grateful to prof .
g. rupp and prof .
e. van beveren for extensive discussions .
m. ablikim _
[ bes collaboration ] , phys .
b * 607 * 243 ( 2005 ) t.a .
lhde and u.g .
meissner , phys .
d * 74 * 034021 ( 2006 ) m. ablikim _
[ bes collaboration ] , phys .
b * 598 * 149 ( 2004 ) a. abele _ et al .
a * 609 * 562 ( 1996 ) | in production processes , e.g. @xmath0 or @xmath1 , the @xmath2 and @xmath3 overlap in the same partial wave .
the conjecture of extended unitarity ( eu ) states that the @xmath4 pair should have the same phase variation as @xmath4 elastic scattering .
this is an extension of watson s theorem beyond its original derivation , which stated only that the @xmath5-dependence of a single resonance should be universal .
the prediction of eu is that the deep dip observed in @xmath6 elastic scattering close to 1 gev should also appear in production data .
four sets of data disagree with this prediction .
all require different relative magnitudes of @xmath2 and @xmath3 . that being so , a fresh conjecture is to rewrite the 2-body unitarity relation for production in terms of observed magnitudes .
this leads to a prediction different to eu .
central production data from the afs experiment fit naturally to this hypothesis .
pacs numbers : 13.25.-k , 13.25.gv , 13.75.lb * experimental disagreements with extended unitarity * |
detailed observations of individual giant elliptical galaxies have shown that they can be subdivided into two groups with respect to their structural properties @xcite [ bdm , and references therein ] .
faint giant ellipticals are isotropic rotators with small minor axis rotation and disky deviations of their isophotal contours from perfect ellipses
. they might contain faint disks which contribute up to 30% to the total light in the galaxy
. therefore their disk - to - bulge ratios overlap with those of s0-galaxies @xcite .
disky ellipticals have power - law inner density profiles @xcite and show little or no radio and x - ray emission @xcite .
boxy ellipticals , on the other hand , are in general more massive than disky ellipticals .
they have box - shaped isophotes , and show flat cores .
the kinematics of boxy ellipticals is generally more complex than that of disky ellipticals .
they rotate slowly , are supposed to be supported by velocity anisotropy and have a large amount of minor axis rotation .
a number of rotationally supported ellipticals also show boxy isophotes .
these systems are either purely boxy ( this is supposed to originate from tidal interactions with nearby massive companions ) or show disky isophotes in the inner part and boxy isophpotes in the outer part @xcite .
occasionally , boxy ellipticals have kinematically distinct cores @xcite .
these cores inhabit flattened rapidly rotating disk- or torus - like components dominating the light in the central few hundred parsecs @xcite , but they contribute only a few percent to the total light of the galaxy .
the fact that these cores are metal - enhanced shows that gas ( at least in the inner regions ) must have played an important role during their formation @xcite .
boxy ellipticals show stronger radio emission than the average and have high x - ray luminosities consistent with emission from hot gaseous halos @xcite .
the distinct physical properties of disky and boxy elliptical galaxies point to the fact that both types of ellipticals could have different formation histories .
it has been argued by @xcite and @xcite that the observed stellar disks and the high density power law centers in disky ellipticals are signatures of dissipation during their formation . in this sense
they seem to continue the hubble sequence from s0s to higher bulge - to - disk ratios @xcite .
boxy ellipticals with an even higher bulge - to - disk ratio show a stronger kinematical decoupling at their centers and no signature of a disk at all .
therefore the early type hubble sequence from s0s to massive ellipticals might represent galaxies with a disk component embedded in more prominent spheroidal body .
+ the traditional view on the formation and evolution of giant elliptical galaxies is that they are very old stellar systems and formed very early at a redshift of more than two @xcite .
after an intensive initial star - formation phase they experienced very little mass evolution @xcite .
it has been argued by many authors that the stellar evolution of ellipticals is compatible with pure passive evolution models @xcite or models with exponentially decaying star formation @xcite .
alternatively , hierarchical theories of galaxy formation predict that massive galaxies were assembled relatively late in many generations of mergers of disk - type galaxies or smaller subunits and mass accretion .
it has been argued by @xcite and @xcite that this merger scenario is consistent with observations of galaxies at different redshifts .
the idea that elliptical galaxies can form from mergers of disk - galaxies has been originally proposed by @xcite .
thereafter the merger hypothesis has been investigated in great details by many authors .
for a recent review of current models of spheroid formation see @xcite .
@xcite , @xcite and @xcite performed the first fully self - consistent merger models of two equal - mass stellar disks embedded in dark matter halos .
the remnants were slowly rotating , pressure supported and anisotropic systems and generally followed an @xmath4 surface density profile in the outer parts .
however , due to phase space limitations @xcite it was necessary to start with progenitors with massive central bulge components @xcite to fit the observed de vaucouleurs profile also in the inner parts .
these simulations showed that global properties of equal - mass merger remnants resemble those of ordinary slowly rotating massive elliptical galaxies .
more detailed investigations of isophotal shapes of the merger remnants have shown that the same remnant can appear either disky or boxy when viewed from different directions @xcite with a trend pointing to boxy isophotes @xcite . @xcite and @xcite investigated a sample of disk - disk mergers with a mass ratio of 3:1 and found that the remnants are flattened and fast rotating in contrast to equal - mass mergers . @xcite investigated the photometric and kinematical properties of a prototypical 3:1 merger remnant in details and compared the results to observational data of disky elliptical galaxies .
they found an excellent agreement and proposed that fast rotating disky elliptical galaxies can originate from pure collisionless 3:1 mergers while slowly rotating , pressure supported ellipticals form from equal - mass mergers of disk galaxies . in this paper
we extend the analysis of @xcite .
a large number of 112 merger remnants resulting from a statistically unbiased sample of simulations of mergers between disk - galaxies with mass ratios of 1:1 , 2:1 , 3:1 , and 4:1 is investigated .
this large sample allows a much more thorough investigation of the statistical properties of merger remnants as all previous studies . after a short description of the simulation methods in section [ mmod ] we investigate the photometric and kinematic properties of the simulated remnants in section [ pkp ] .
we discuss the drawback of our results on the theory of the formation of elliptical galaxies in section [ con ] .
the spiral galaxies were constructed in dynamical equilibrium using the method described by @xcite .
we used the following system of units : gravitational constant g=1 , exponential scale length of the larger disk @xmath5 and mass of the larger disk @xmath6 .
each galaxy consists of an exponential disk , a spherical , non - rotating bulge with mass @xmath7 , a hernquist density profile @xcite with a scale length @xmath8 and a spherical pseudo - isothermal halo with a mass @xmath9 , cut - off radius @xmath10 and core radius @xmath11 .
we followed a sequence of mass ratios of the progenitor disks from @xmath12 to @xmath13 where @xmath14 is the mass of the more massive galaxy divided by the mass of the merger partner .
the equal - mass mergers were calculated adopting in total 400000 particles with each galaxy consisting of 20000 bulge particles , 60000 disk particles , and 120000 halo particles .
we decided to use twice as many halo particles than disk particles to reduce heating and instability effects in the disk components @xcite .
for the mergers with @xmath15 the parameters of the more massive galaxy were as described above .
the low - mass companion contained a fraction of @xmath16 the mass and the number of particles in each component with a disk scale length of @xmath17 , as expected from the tully - fisher relation @xcite .
the n - body simulations for the equal - mass mergers were performed by direct summation of the forces using the special purpose hardware grape6 @xcite . with this highly efficient hardware one force calculation for 400000 particles
takes approx .
@xmath18 seconds .
the mergers with mass ratios @xmath15 were followed using the newly developed treecode vine ( wetzstein et al . , 2003 ) in combination with the grape5 @xcite hardware for which the code was optimized .
vine uses a binary tree in combination with the refined multipole acceptance criterion proposed by @xcite .
this criterion takes the mass distribution of every node into account .
it enables the user to control the absolute force error which is introduced by the tree construction .
we chose a value of 0.001 which guarantees that the error resulting from the tree is smaller than the intrinsic force error of the grape5 hardware which is of the order of @xmath19 . to compensate for the limited mass resolution of grape5 we limit the maximum mass of a tree node to @xmath20 times the minimum particle mass .
to further increase the speed of the calculation we only use grape5 if forces for more than 50 particles are needed at the same time .
otherwise we compute the forces on the host computer as for small particle numbers the communication time between the board and the host computer exceeds the time for the force calculation on the host . with the parameters given above one force calculation with vine and grape5 for 400000 particles
takes approx .
@xmath21 seconds .
we used a gravitational plummer - softening of @xmath22 and a fixed leap - frog integration time step of @xmath23 . for the equal - mass mergers simulated with direct summation on grape6
the total energy is conserved , vine in combination with grape5 conserves the total energy up to @xmath24 .
the galaxies approached each other on nearly parabolic orbits with an initial separation of @xmath25 length units and a pericenter distance of @xmath26 length units ( same parameters as e.g. @xcite ) .
a study of orbits of merging dark matter halos in cosmological large scale simulations by @xcite has shown that most of the merging halos are indeed on parabolic orbits .
the inclinations of the two disks relative to the orbit plane were @xmath27 and @xmath28 with arguments of pericenter @xmath29 and @xmath30 . in selecting unbiased initial parameters for the disk inclinations
we followed the procedure described by @xcite .
for the spin vector of each disk we defined four different orientations pointing to every vertex of a regular tetrahedron .
the initial orientations we used translate to the following set of angles : for the first galaxy @xmath31 .
the second galaxy has @xmath32 .
these parameters result in 16 initial configurations for equal - mass mergers and 16 more for every mass ratio @xmath15 where the initial orientations are interchanged .
following the simple hypothesis that the orientations of the merging disks are independent of each other and independent of their mutual orbital plane , every merger geometry has an equal probability to be realized @xcite .
the orbital parameters are listed in table [ tbl-1 ] . in total
we simulated 112 mergers .
the total computing time for the entire sample was about 1600 hours wall clock time . for all simulations
the merger remnants were allowed to settle into equilibrium approximately 8 to 10 dynamical times after the merger was complete .
then their equilibrium state was analyzed .
to compare our simulated merger remnants with observations we analyzed the remnants with respect to observed global photometric and kinematical properties of giant elliptical galaxies , e.g. surface density profiles , isophotal deviation from perfect ellipses , velocity dispersion , and major- and minor - axis rotation . defining characteristic values for each projected remnant we followed as closely as possible the analysis described by bdm .
an artificial image of the remnant was created by binning the central 10 length units into @xmath33 pixels .
this picture was smoothed with a gaussian filter of standard deviation 1.5 pixels . the isophotes and their deviations from perfect ellipses
were then determined using a data reduction package kindly provided by ralf bender . to be confident that a once determined isophotal shape is characteristic for the remnant and does not change with time we investigated the time evolution of the ellipticity and the @xmath34 profile starting @xmath35 time units after the merger of the bulge components was complete and followed the evolution for the next @xmath36 time units . in intervals of four time units the luminous part of the remnant
was transformed to the principal axes of its moment of inertia tensor .
the tensor was evaluated using @xmath37 of the most tightly bound particles .
the isophotal properties were then analyzed as seen along the minor axis the characteristic ellipticity @xmath38 for each projection was defined as the isophotal ellipticity at @xmath39 .
figure [ fig1 ] shows the ellipticity profiles for the 3:1 merger remnant 10 in intervals of four time units starting at @xmath40 .
the ellipticity profile shows little evolution with time and so does @xmath38 ( small box ) .
the behaviour of this simulation is characteristic for the whole set of simulations .
following the definition of bdm for the global properties of observed giant elliptical galaxies , we determined for every projection the effective @xmath34-coefficient , @xmath41 , as the mean value of @xmath34 between @xmath42 and @xmath43 , with @xmath44 being the projected spherical half - mass radius . in case of a strong peak in the @xmath34-distribution with an absolute value that is larger than the absolute mean value , we chose the peak value .
being characteristic for all merger remnants figure [ fig2 ] shows the @xmath34-profiles at different times for the 3:1 merger remnant with geometry 10 .
figure [ fig3 ] shows the corresponding @xmath34-profiles for the equal - mass remnant with geometry 2 .
although the individual @xmath34-profiles differ in details at different time steps the global characteristic shape is conserved .
this is reflected in the time evolution of @xmath41 which was assigned to each remnant at the different time steps as shown in the small diagrams of figure [ fig2 ] and figure [ fig3 ] .
the change in @xmath41 due to evolutionary effects ( a real change in the isophotal shape and/or a change in three - dimensional shape leading to a different projection angle ) is of the same order as the error due to the limited particle number .
the bootstrap error bar is shown in the small diagrams of figure [ fig1 ] and figure [ fig2 ] .
these measurements convinced us that the photometric properties of the inner regions ( inside one effective radius ) of the merger remnants evolve relatively fast into an equilibrium configuration .
as the photometric properties do not significantly change with time they can be used for a further statistical analysis .
+ to investigate projection effects we determined for each simulation @xmath41 and @xmath38 for 500 random projections of every remnant .
this resulted in 8000 projections for @xmath45 and 16000 projections for @xmath46 , respectively .
the normalized histograms of projected ellipticities are shown in figure [ fig4 ] .
for equal - mass mergers the distribution rises from a small number of projections with zero ellipticity to a peak around @xmath47 and then falls to zero at @xmath48 .
2:1 remnants show one peak around @xmath49 and a strong peak at @xmath50 falling to zero at @xmath48 .
3:1 and 4:1 remnants show a flat distribution for small ellipticities rising to a strong peak around @xmath51 and @xmath52 , respectively .
no projection leads to ellipticities larger than @xmath53 .
the distribution function for the isophotal shapes of equal - mass remnants peaks at @xmath54 and declines rapidly for more negative values ( figure [ fig5 ] ) . in total 47% of the projected 1:1 remnants
show disky isophotes distributed in a broader wing for positive @xmath41 . for higher mass ratios the distribution functions have a similar shape but are shifted to more positive values of @xmath41 .
2:1 remnants peak at around @xmath55 and @xmath56 of the projected remnants show disky isophotes .
the distributions for 3:1 and 4:1 remnants both peak around @xmath57 . for these systems
the fraction of disky projections is about @xmath58 and @xmath59 , respectively .
in addition , the 500 projected values for @xmath41 and @xmath38 where used to calculate the two - dimensional probability density function for a given simulated remnant to be `` observed '' in the @xmath41-@xmath38 plane .
we added up the probability densities assuming equal weights for every simulation geometry at a given mass ratio .
figure [ fig6 ] shows the results for mergers with @xmath60 , and @xmath61 .
the contours indicate the areas of 50% ( dashed line ) , 70% ( thin line ) and 90% ( thick line ) probability to detect a merger remnant with the given properties . observed data points from bdm
are over - plotted .
filled boxes are boxy ellipticals with @xmath62 while open diamonds indicate disky ellipticals with @xmath63 .
the errors were estimated applying the statistical bootstrapping method @xcite .
the area covered by 1:1 remnants with negative @xmath41 is in good agreement with the observed data for boxy elliptical galaxies .
in particular the observed trend for more boxy galaxies to have higher ellipticities is reproduced .
the disky equal - mass remnants also follow the trend for observed disky ellipticals .
remnants with an @xmath41 around zero can have slightly higher ellipticities than observed .
remnants with @xmath64 , and @xmath61 predominantly populate the region of disky ellipticals and more disky ellipticals also tend to be more flattened .
there is a trend for projections with @xmath57 to have slightly higher ellipticities than observed ( or at a given ellipticity the remnants are not disky enough ) .
observed very disky ellipticals with relatively small ellipticities can not be reproduced . in summary
there is a clear trend for unequal - mass mergers to produce more disky remnants .
responsible for the disky appearance of the 3:1 and 4:1 remnants is the distribution of the particles of the massive disk .
figure [ fig7 ] shows the different contributions from the small and the large progenitor galaxy and the resulting isophotal map of a characteristic 3:1 merger remnant .
the particles originating from the small progenitor accumulate in a torus - like structure that has peanut - shaped or boxy isophotes .
in contrast , the dominant luminous material from the larger progenitor still keeps its disk - like appearance . in combination , the contribution from the larger progenitor since it is three times more massive dominates the overall properties of the remnant .
this result holds for all 3:1 and 4:1 merger remnants .
the more massive disk component is not completely destroyed during the merger event and determines the overall structure of the remnant . for equal - mass mergers
both disks are effected during the merger and they can loose the memory of their initial state .
the isophotal analysis also provides information about the radial change of the relative orientation of the major axes of the isophotes . in general , the amount of isophotal twist depends on the projection angle .
this is demonstrated in figure [ fig8 ] for a more elongated and a nearly round projection of a 3:1 merger remnant . to get a quantitative measure for the isophotal twist we determined the relative position angle @xmath65 between the isophote at 0.5 @xmath66 and 1.5 @xmath66 for every projection of the remnant .
figure [ fig9 ] shows a comparison of isophotal twists for characteristic remnants with @xmath67 , and @xmath61 .
the isophotal twist is in general larger for projections that appear nearly round . for ellipticities larger than @xmath68
the isophotal twist is @xmath69 .
the distribution of @xmath65 versus @xmath38 for random projections of every remnant is consistent with observations of elliptical galaxies @xcite .
however , the simulated remnants can show larger isophotal twists at high ellipticities than observed galaxies .
in addition there seems to exist no obvious correlation between the mass ratio of the galaxies and the amount of isophotal twist .
the central velocity dispersion @xmath70 of every remnant was determined as the average projected velocity dispersion of the luminous particles inside a projected galactocentric distance of @xmath71 .
we defined the characteristic rotational velocity along the major and the minor axis as the projected rotational velocity at @xmath39 and @xmath72 , respectively .
figure [ fig10 ] shows the time evolution for these velocity measurements for a 3:1 merger simulation which is characteristic for all simulations .
the derived kinematical properties of the remnants stay nearly constant over a long time period and are therefore a good measurement of the intrinsic kinematics of the simulated remnants .
again , the bootstrap error is of the same order as the change of the measured values with time .
like for the isophotal shape we computed the statistical kinematical properties of the simulated remnants and compared them with observational data of elliptical galaxies .
the normalized histograms for @xmath73 are shown in figure [ fig11 ] .
there is a clear trend for 1:1 mergers to produce slowly rotating ellipticals with @xmath74 . for 2:1 mergers
the peak value is around @xmath75 .
3:1 and 4:1 merger remnants have peaks around @xmath76 and @xmath77 , respectively .
there are indications for a weak secondary peak around @xmath78 for 3:1 and 4:1 remnants .
figure [ fig12 ] shows the distribution function in the @xmath73-@xmath38 plane .
the area of slowly rotating boxy ellipticals ( filled boxes ) is almost completely covered by remnants of 1:1 mergers while 2:1 mergers have rotational properties resembling faster rotating boxy and slowly rotating disky ellipticals . 3:1 and 4:1 remnants are clearly fast rotating ,
show high ellipticities and can be associated with fast rotating disky ellipticals .
mergers with an increasing mass ratio produce faster rotating ellipticals with higher ellipticities .
all simulated data are in good agreement with observations . however , there is a trend for the simulated remnants to have slightly larger maximum ellipticities than observed .
projections with high ellipticities ( @xmath79 ) show systematically smaller values for @xmath80 than observed .
the effect is strongest for equal - mass merger remnants and is discussed in detail below .
this indicates that the simulated systems are more strongly supported by anisotropic velocity dispersions than the observed systems , despite significant rotation for 3:1 and 4:1 remnants .
observed ellipticals with @xmath81 can not be reproduced as was already reported by @xcite .
the minor - axis kinematics of the simulated remnants was measured as the rotation velocity along the minor axis at @xmath82 .
the amount of minor axis rotation was parametrized as @xmath83 @xcite .
minor axis rotation in elliptical galaxies , in addition to isophotal twist , has been suggested as a sign for a triaxial shape of the main body of elliptical galaxies @xcite .
indeed , 1:1 mergers show a significant amount of minor - axis rotation ( figure [ fig13 ] ) .
this is consistent with their mostly triaxial shape and the measured isophotal twist . 3:1 and 4:1 remnants are more oblate and show weak minor axis rotation .
however , the amount of isophotal twist is comparable to equal - mass merger remnants .
a detailed analysis of the connection between intrinsic shape , isophotal shape , and internal kinematics is beyond the scope of this paper and needs further investigation .
the anisotropy parameter @xmath84 was defined as the ratio of the observed value of @xmath73 and the theoretical value for an isotropic oblate rotator @xmath85 with the observed ellipticity @xmath86 @xcite .
this parameter has been used by observers as a test whether a given galaxy with observed @xmath87 , @xmath70 and @xmath86 is flattened by rotation ( @xmath88 ) or by velocity anisotropy ( @xmath89 ) @xcite .
figure [ fig14 ] shows the normalized histograms for the @xmath84 values of the simulated remnants .
the 1:1 remnants peak around @xmath90 with a more prominent tail towards lower values .
only about 10% of the projections have @xmath91 .
in contrast 82% have @xmath92 and would clearly be interpreted as being supported by anisotropic velocity dispersion .
2:1 remnants peak at @xmath93 with 41% showing @xmath91 . 3:1 and 4:1 mergers with peaks at @xmath94 and @xmath95 are consistent with model predictions of oblate isotropic or slightly prolate rotators .
the percentage of projections with @xmath91 is 64% and 83% , respectively . 3:1 and 4:1 remnants also have predominantly disky isophotes and cover the area populated by observed disky ellipticals in the @xmath96 - @xmath41 diagram ( figure [ fig15 ] ) .
1:1 merger remnants show the most complex behaviour .
a significant number of the projected remnants lie in a region of disky anisotropic systems where no elliptical galaxy is observed . assuming that a projected remnant clearly fails to resemble an observed elliptical
if it has @xmath97 and @xmath98 , 28% of the 1:1 remnants would fall into the forbidden regime . + to understand the behaviour of equal - mass remnants in more detail we illustrate the influence of the initial orientation of the progenitor disks on the properties of the merger remnants .
we defined mean values of the investigated quantities of all 500 projections for every merger geometry .
thereafter we divided the equal - mass remnants into four groups of geometries that produce remnants with almost distinct properties : slowly rotating boxy remnants with @xmath99 and @xmath100 ( group a ; geometries 2,4,5 , and 14 ) , slowly rotating remnants with predominantly disky isophotes having @xmath101 and @xmath102 ( group b ; geometries 1,8,9,11 , and 13 ) , modestly rotating remnants ( @xmath103 ) that appear to be isotropic with @xmath104 ( group c ; geometries 7 , 12 , and 15 ) and remnants with modest deviations in isophotal shape @xmath105 ( group d ; geometries 3,6,12 , and 16 ) .
the results are shown in figure [ fig16 ] .
remnants of group a are slowly rotating anisotropic systems with purely boxy isophotes .
more than 90% of the projected remnants are consistent with observations of boxy anisotropic elliptical galaxies .
this class of equal - mass merger remnants was sampled by @xcite .
most projections of group b show disky isophotes and appear to be anisotropic .
around 60% of the projected remnants of this group fail to fall on the observed correlation .
furthermore the remnants of this group show too high ellipticities at low rotation velocities , a behaviour already described by @xcite .
the shaded regions in figure [ fig16 ] indicate the location of the highly elliptical projections ( @xmath106 ) .
they dominate the disky anisotropic branch of `` not observed '' merger remnants .
it is possible that the very elongated projections of group b would be classified as s0 galaxies .
some of them , as ngc4550 , show weak net rotation and are still very elongated .
geometry 1 produces a very axisymmetric remnant with two counterrotating populations of stars in as it is observed in ngc4550 ( see e.g. @xcite ) .
however , @xcite argued that a merger origin for this particular galaxy is unlikely as it would have heated both disks too much .
group c combines all rotating 1:1 remnants .
they appear slightly disky or boxy depending on the orientation and are isotropic .
it has to be noted that even the disky remnants of this group are consistent with observations and only 10% of the projections are failures .
this group shows a clear connection to the initial conditions as the spins of the progenitor disks were almost aligned .
projections of group d can either have disky or boxy isophotes and are mostly anisotropic .
35% of the projections fall in the forbidden regime .
we used a large set of n - body simulations of collisionless mergers of disk galaxies with mass ratios of @xmath107 and @xmath61 to investigate for the first time statistical properties of disk - merger remnants .
in contrast to previous studies where only projections along the principal axes were investigated we analyzed a large number of randomly projected merger remnants and compared the results in a consistent way to observations .
galaxies on the sky are also viewed along random lines of sight therefore this is the appropriate way of comparison .
we showed that the detailed isophotal and kinematical properties of the simulated merger remnants reach their equilibrium values relatively fast after the merger of the central parts of the galaxies is complete and thereafter do not significantly evolve with time .
the high resolution of the simulations made it possible to keep the errors , especially for @xmath34 , at a reasonably low value and enabled us to perform a statistical analysis .
the basic result is that the projected kinematical and photometric properties of remnants of major mergers of disk galaxies are in surprisingly good agreement with the observational data for elliptical galaxies .
the mass ratio of the progenitor disks determines the global properties of the remnants .
the influence of the initial orientation on the remnants is strongest for equal - mass mergers .
purely boxy , anisotropic and slowly rotating remnants with a large amount of minor axis rotation can only be produced by equal - mass mergers with certain initial orientations . however , if the initial spins of the disks were almost aligned the remnants appear to be isotropic and disky or boxy , depending on the viewing angle . in total 28% of all projected 1:1 remnants show properties that are not observed at all .
as some of these remnants have high ellipticities and very small @xmath84 they appear to be flattened by strongly anisotropic velocity dispersions .
in addition they mostly show very disky isophotes . as we can not exclude that mergers with these geometries have occurred they
should be observed , which is not the case .
if this controversy is not solved in the future it might constitute a serious problem for the merger hypothesis .
however , if observed , these elongated disky objects might have been classified as s0s and have therefore been excluded from the observed sample of bona fide ellipticals . based on the diverse properties of the 1:1 remnants
they more likely resemble the class of intermediate mass giant ellipticals in the transition region from rotating disky to non - rotating boxy ellipticals . in this regime
even boxy rotating galaxies have been observed @xcite . in contrast ,
3:1 and 4:1 mergers form a more homogeneous group of remnants .
they have preferentially disky isophotes , are fast rotating and show small minor axis rotation independent of the assumed projection . in general
the properties are in very good agreement with observations of fast rotating disky ellipticals .
2:1 mergers have intermediate properties with boxy or disky isophotes depending on the projection and the orbital geometry of the merger . globally
, different projection angles and orbits do not change the fundamental properties of 1:1 mergers on the one side and 3:1 and 4:1 mergers on the other side .
+ in summary , many photometric and kinematical properties of low and intermediate mass giant elliptical galaxies can be understood as a sequence of major mergers of disk galaxies with varying mass ratios .
the homogeneous group of massive boxy ellipticals populates the boxy anisotropic area of the @xmath108 - @xmath41 plane .
their distribution is inconsistent with the distribution of simulated 1:1 remnants .
therefore they are most likely not remnants of equal - mass mergers of disk galaxies .
boxy and mildly anisotropic remnants are only reproduced by a small subsample of initial conditions .
there is no reason why other geometries should have been avoided .
it is more likely that massive boxy ellipticals have formed by other processes like mergers of early type galaxies @xcite or multiple mergers in a group environment @xcite .
+ the simulations presented here were purely dissipationless , taking into account only the stellar and dark matter component of a galaxy .
the importance of gas in galaxy - galaxy mergers and the detailed influence on the structure of elliptical galaxies is not fully understood up to now .
numerical simulations of galaxy mergers including gas by @xcite and @xcite have shown that the presence of gas can change the orbital structure and the shape of the merger remnants .
as soon as the gas is driven to the center during the merger the mass concentration seems to be responsible for the destruction of stellar box orbits .
this process makes it even more difficult to explain the formation of giant boxy ellipticals by binary disk mergers .
+ the present study will serve as the basis of a further detailed investigation on the influence of an additional dissipative component .
@xcite proposed a revised hubble sequence with disky ellipticals representing the missing link between the im - spiral - s0 sequence and boxy ellipticals .
they noted that gas infall into the equatorial plane with subsequent star formation could lead to a second disk - like subcomponent .
ellipticals with disks could appear disky when seen edge - on and boxy otherwise @xcite .
our simulations indeed indicate that a disk - like substructure is responsible for producing disky isophotes ( see fig .
[ fig6 ] ) . however , in the present case , the disk is the remnant of the more massive spiral which was not completely destroyed during the minor merger ( see also @xcite ) .
@xcite investigated line - of - sight velocity distributions of dissipationless merger remnants and found a velocity profile asymmetry that is opposite to the observed one .
they concluded that this disagreement can be solved if ellipticals would contain a second disk - like substructure that most likely formed through gas accretion .
the situation is however not completely clear , as another study by @xcite found a good agreement of the observed asymmetries for some cases . @xcite
have shown that extended gas disks can form as a result of a gas rich unequal - mass merger .
@xcite recently presented a first detailed set of equal - and unequal - mass merger simulations of gas - rich galaxies resulting in the formation of extended gas disks .
his simulations demonstrate a very complex dynamical evolution .
it has to be investigated how the presence of gas changes the detailed properties of the remnants in detail .
in addition , star formation during the merger will influence the stellar kinematics ( see e.g. @xcite ) and the stellar populations by adding young and probably more metal rich stars . as metal enrichment provides a futher strong observational constraint on the formation history of early type galaxies ( see e.g. @xcite ) more simulations , including dissipation , star formation and chemical evolution will now be required in order to understand in detail the role of gas for the formation of elliptical galaxies by mergers .
we thank ralf bender and hans - walter rix for helpful comments on the script and the referee , joshua barnes , for valuable suggestions that significantly improved the paper .
all the simulations presented in this work have been performed on the grape cluster at the max - planck - insitute for astronomy in heidelberg ( mpia ) .
thorsten naab is grateful to the mpia for its support .
ccccccc 1/17 & 0 & 0 & 180 & 0 & 2 & 30 + 2/18 & 0 & 0 & 71 & 30 & 2 & 30 + 3/19 & 0 & 0 & 71 & -30 & 2 & 30 + 4/20 & 0 & 0 & 71 & 90 & 2 & 30 + 5/21 & -109 & -60 & 180 & 0 & 2 & 30 + 6/22 & -109 & -60 & 71 & 30 & 2 & 30 + 7/23 & -109 & -60 & 71 & -30 & 2 & 30 + 8/24 & -109 & -60 & 71 & 90 & 2 & 30 + 9/25 & -109 & 0 & 180 & 0 & 2 & 30 + 10/26 & -109 & 0 & 71 & 30 & 2 & 30 + 11/27 & -109 & 0 & 71 & -30 & 2 & 30 + 12/28 & -109 & 0 & 71 & 90 & 2 & 30 + 13/29 & -109 & 60 & 180 & 0 & 2 & 30 + 14/30 & -109 & 60 & 71 & 30 & 2 & 30 + 15/31 & -109 & 60 & 71 & -30 & 2 & 30 + 16/32 & -109 & 60 & 71 & 90 & 2 & 30 + | we perform a large parameter survey of collisionless n - body simulations of binary mergers of disk - galaxies with mass ratios of 1:1 , 2:1 , 3:1 , and 4:1 using the special purpose hardware grape .
a set of 112 merger simulations is used to investigate the fundamental statistical properties of merger remnants as a function of the initial orientation and the mass ratio of the progenitor disks .
the photometric and kinematical properties of the simulated merger remnants are analyzed .
the methods used to determine the characteristic properties are equivalent to the methods used for observations of giant elliptical galaxies . we take projection effects into account and
analyze the remnant properties in a statistical way for comparison with observations .
the basic properties of the remnants correlate with the mass ratio of the progenitor disks .
we find that about 80% of the equal - mass merger simulations lead to slowly rotating merger remnants having @xmath0 .
observers would interpret those objects as being supported by anisotropic velocity dispersions .
all 1:1 remnants show significant minor - axis rotation .
one half of all projected 1:1 remnants shows boxy ( @xmath1 ) and the other half shows disky ( @xmath2 ) isophotes .
a distinct subclass of 4 out of 12 initial orientations leads to purely boxy remnants independent of orientation .
1:1 mergers with other initial orientations show disky or boxy isophotes depending on the viewing angle .
remnants with mass ratios of 3:1 and 4:1 have more homogeneous properties .
they all rotate fast ( maximum value of @xmath3 ) and show a small amount of minor - axis rotation , consistent with models of isotropic or slightly anisotropic oblate rotators .
if observed in projection they would be interpreted as being supported by rotation . about 90% of the projected 3:1 and
4:1 remnants show disky isophotes .
2:1 remnants show intermediate properties .
projection effects lead to a large spread in the data in good agreement with observations .
they do not change the fundamental kinematical differences between equal - mass and unequal - mass merger remnants .
the correlation between isophotal twist and apparent ellipticity of every single merger remnant is in good agreement with observations .
the amount of twisting strongly depends on the orientation of the remnant but is only weakly dependent on the mass ratio of the merger .
the results of this study weaken the disk merger scenario as the possible formation mechanism of massive boxy giant ellipticals as only equal - mass mergers with special initial orientations can produce purely boxy anisotropic merger remnants .
some orientations of 1:1 mergers can even lead to disky and anisotropic remnants which are either not observed or would be classified as s0s based on their morphology .
in general , the properties of equal - mass ( and 2:1 ) merger remnants are consistent with the observed population of giant ellipticals in the intermediate mass regime between low mass fast rotating disky and bright massive boxy giant ellipticals . 3:1 and 4:1 merger remnants , however , are in very good agreement with the class of low luminosity , fast rotating giant elliptical galaxies .
binary mergers of disk galaxies are therefore still very good candidates for being the main formation mechanism for intermediate and low mass giant ellipticals . the homogeneous class of massive boxy ellipticals most likely formed by a different process . |
iron - bearing mgsio@xmath0 perovskite ( si - pv ) is one of the major constituent minerals of earth s lower mantle along with ( mg , fe)o ferropericlase ( fp ) .
unraveling the composition and thermal structure of the lower mantle requires a detailed understanding of the influence of iron ( fe ) on these host minerals .
it is well known that iron in ( mg , fe)o undergoes a pressure induced spin crossover from high - spin ( s=2 ) to low - spin ( s=0 ) state and affects elastic properties @xcite . in the case of si - pv , deciphering the iron spin state has been quite challenging due to complex perovskite structure and possibility of different valence states of iron ( ferric and ferrous ) .
si - pv is also known to undergo a perovskite ( pv ) to post - perovskite ( ppv ) phase transition at @xmath4 125 gpa and this will have geophysical consequences in deep lower mantle region @xcite . understanding the effect of iron , especially , on pv to ppv transition and on elastic properties of pv and ppv phases is crucial to constrain lower mantle composition . in ( mg , fe@xmath1)sio@xmath0-pv
, the pressure induced increase in m@xmath5ssbauer quadrupole splitting ( qs ) above @xmath430 gpa ( low - qs to high - qs transition ) @xcite has been explained by lateral displacement of fe@xmath1 ion which remains in the high - spin ( hs ) state in the entire lower - mantle pressure range @xcite . in the case of ( mg , fe@xmath2)(si , fe@xmath2)o@xmath0-pv , however , first - principles calculations by @xcite reported that fe@xmath2 in b - site undergoes a high - spin ( s=5/2 ) to low - spin ( s=1/2 ) state change at approximately 41 gpa and 70 gpa using lda+u@xmath3 and gga+u@xmath3 methods , respectively , while in the a - site fe@xmath2 remains in the high - spin state .
these findings were in good agreement with experimental observations of @xcite .
there have also been several experimental @xcite and first - principles @xcite studies about the spin crossover in mgsio@xmath0-ppv ( si - ppv ) .
the consensus regarding the spin crossover in iron - bearing si - ppv is that fe@xmath1 in the a - site remains in the high - spin and high - qs state throughout the lower mantle pressure range , while fe@xmath2 undergoes a high to low spin transition in the b site and remains in the high spin state in the a - site @xcite . in spite of these efforts to pin down the state of iron at high pressures , it has been very difficult to ascertain the exact role of iron with its varying spin and valence state on pv to ppv transition , especially because of the extreme pressure and temperature conditions in the deep lower mantle region .
mggeo@xmath0-perovskite ( ge - pv ) is a low pressure analog of mgsio@xmath0 perovskite @xcite and would be a promising candidate to investigate iron spin crossover and its structural influence on mgsio@xmath0-pv and ppv phases . in this study , we have used first - principles dft+u ( lda+u@xmath3 and gga+u@xmath3 ) calculations to investigate the iron spin states and their crossover in the pv and ppv phases of fe@xmath1- and fe@xmath2-bearing mggeo@xmath0 . comparing enthalpy of the different spin states , we have determined the stability of the different states and transition pressure . using these results ,
we have also investigated the effect of fe@xmath1 and fe@xmath2 substitution on pv to ppv transition .
we have used density functional theory ( dft ) within the local density approximation ( lda ) @xcite and generalized gradient approximation ( gga ) @xcite augmented by the hubbard u ( lda / gga+u@xmath3 ) , calculated self - consistently and structurally consistently @xcite .
ultrasoft pseudo - potentials , generated by vanderbilt s method @xcite , have been used for fe , si , ge , and o. for mg , norm - conserving pseudo - potential generated by von barth - car s method , has been used .
the plane - wave kinetic energy and charge density cutoff are 50 ry and 300 ry , respectively .
self - consistent u@xmath3 , calculated by @xcite and @xcite using the linear response approach @xcite ( table [ table : u_sc ] ) , have been used .
all calculations were performed in a 40-atoms super - cell .
electronic states were sampled on a shifted @xmath6 and @xmath7 k - point grid for pv and ppv structure , respectively @xcite .
structural optimization at any arbitrary volume has been performed using variable cell - shape damped molecular dynamics approach @xcite .
structures are optimized until the inter - atomic forces are smaller than 10@xmath8 @xmath9 . for every possible
spin and valence state , structures are optimized at twelve volumes in the relevant pressure - range and the 3@xmath10-order birch - murngharn equation of state has been used to fit these compression data .
iron spin states have been investigated in fe@xmath1 and fe@xmath2-bearing mggeo@xmath0 and mgsio@xmath0 in a 40-atom super - cell .
one fe@xmath1 ion is substituted for a mg@xmath1 ion in the a - site in ( mg@xmath11fe@xmath12)geo@xmath0 and ( mg@xmath11fe@xmath12)sio@xmath0 while two fe@xmath2 s are substituted for one mg@xmath1 and one first neighbor ge@xmath13 or si@xmath13 as in ( mg@xmath11fe@xmath12)(ge@xmath11fe@xmath12)o@xmath0 and ( mg@xmath11fe@xmath12)(si@xmath11fe@xmath12)o@xmath0 .
the nearest - neighbor fe@xmath2-fe@xmath2 substitution has been chosen as it has been found to be the lowest energy configuration in previous first - principles studies @xcite .
relative enthalpies have been used to calculate transition pressures ( p@xmath14 ) and the stability of a state or phase in a given pressure range .
effects of disorder have not been addressed at this point , but it is expected to produce a two - phase loop . here
we wish to analyze ( a ) the potentially different effects of fe@xmath1 and fe@xmath2 in the pv to ppv transition and ( b ) whether the germanate phase could serve as a good low pressure analog to investigate spin transitions in iron . as in the case of fe@xmath1 in si - pv @xcite , fe@xmath1 in ge - pv
remains in the high - spin ( hs , s=2 ) state in the entire lower - mantle pressure range ( 0 - 135 gpa ) .
however , it undergoes a pressure induced lateral displacement producing a transition from low to high m@xmath5ssbauer quadrupole splitting ( qs ) .
the atomic structures for the low- and high - qs states of iron in si - pv are shown in fig .
[ fig1](a ) indicating the lateral displacement of iron in the perovskite a - site .
the pressure dependence of enthalpy differences between high - qs and low - qs states ( @xmath15h = h@xmath16-h@xmath17 ) gives the low- to high - qs transition pressure .
it is well - known that lda usually underestimates transition pressures while gga usually overestimates it . in this study
, we present both lda+u@xmath3 and gga+u@xmath3 results to provide lower and upper bounds for this transition , as shown in fig . [
fig1](b ) .
the low- to high - qs state change in ge - pv occurs at @xmath428.5 gpa with lda+u@xmath3 and at @xmath440.5 gpa with gga+u@xmath3 .
these transition pressures are quite large compared to those in si - pv , i.e. , @xmath49.5 gpa and @xmath422.5 gpa respectively as presented in table [ table : p_t ] . owing to the larger ionic radius of ge@xmath13 ( 0.53 @xmath18 )
compared to that of si@xmath13 ( 0.40 @xmath18 ) , the unit - cell volume and inter - atomic distances in ge - pv are larger than in si - pv .
zero pressure fe - o bond - lengths in si - pv and ge - pv obtained with lda+u@xmath3 are shown fig . [ fig2](a ) .
the average of fe - o bond - lengths , @xmath19fe - o@xmath20 , is also shown in fig . [ fig2](b ) .
the larger @xmath19fe - o@xmath20 value in ge - pv allows fe@xmath1 to sustain higher pressures than in si - pv before changing to the high - qs state .
it is also interesting to note that this transition occurs when the @xmath19fe - o@xmath20 reaches a particular value ( @xmath42.22 @xmath18 ) whether in ge - pv or in si - pv , whether using lda+u@xmath3 or gga+u@xmath3 methods ( fig . [ fig2].(b ) ) .
it should be noted that the pv to ppv transition is expected to occur at approximately @xmath463 gpa in mggeo@xmath0 @xcite and at @xmath4125 gpa in mgsio@xmath0 @xcite .
these pressures are much higher than low- to high - qs transition pressures in both fe@xmath1-bearing compounds .
consequently , this transition has not been investigated in si - ppv or ge - ppv phases .
next , we investigate spin states of fe@xmath2 in the b - site of ge - pv ( low - pressure ) and ppv ( high - pressure ) phases . the crystal structures of si - pv and si - ppv phases are shown in fig . [ fig3](a ) .
si - ppv has a layered structure characterized by corner - sharing and edge - sharing sio@xmath21 octahedra extending along the a - c plane . the pressure dependence of calculated enthalpy differences between high - spin ( hs ) and low - spin ( ls ) states of fe@xmath2 ( @xmath15h = h@xmath22-h@xmath23 ) in pv and ppv phases are shown in fig . [ fig3 ] ( b ) and ( c ) , respectively .
the calculated hs to ls transition pressures in ge - pv are 56.5 gpa with lda+u@xmath3 and 85.0 gpa with gga+u@xmath3 . similarly , these transition pressures in ge - ppv are 49.5 gpa and 82.5 gpa with these methods .
our calculated transition pressures in si - pv are 41.0 gpa with lda+u@xmath3 and 69.5 gpa with gga+u@xmath3 , which are in very good agreement with previous first - principles study @xcite . in the case of si - ppv ,
our calculated transition pressures are 28.0 gpa and 59.5 gpa , respectively , which are quite larger than previously reported values of -30 gpa and 36 gpa @xcite .
these transition pressures in si - pv and si - ppv are also very similar to those reported by @xcite in ( mg , al)(si , fe)o@xmath0-pv and ppv phases where it was observed that the presence of aluminum does not affect the spin transition pressure significantly . considering the overall agreement of the stability of the iron spin states and pressure induced crossover in si - pv and si - ppv phases with previous studies @xcite , the present calculations about hs to ls transition in fe@xmath2 at b site seems to be consistent and reliable . in order to explain the increase of hs to ls transition pressures in mggeo@xmath0 phases , we compare atomic configurations around fe@xmath2 in the b - site of si - pv and ge - pv phases . as shown in fig .
[ fig4](a ) , the coordination number of fe@xmath2 is six and fe - o bond lengths in this site are smaller than those of fe@xmath1 in the a - site .
the pressure dependence of the average bond - length , @xmath19fe - o@xmath20 , for fe@xmath2 in the b - site in si - pv and ge - pv phases are shown in fig . [ fig4](b ) . as in the case of low- to high - qs transition in fe@xmath1 ( fig .
[ fig1 ] ) , the hs to ls transition in fe@xmath2 in ge - pv and ge - ppv phases are also large compared to those in si - pv and si - ppv phases ( fig .
[ fig3 ] ) due to relatively larger fe - o bond - lengths and unit - cell volumes in germanate phases at a given pressure ( fig .
[ fig4 ] ) .
similar to low- to high - qs transition , the hs to ls transition also occurs only when the @xmath19fe - o@xmath20 reaches a particular value ( i.e. , @xmath41.86 @xmath18 in the present case ) irrespective of ge - pv or si - pv , or of the exchange - correlation functional used ( fig . [ fig4](b ) ) .
this is a useful insight for predicting pressure induced spin state changes in different materials .
having investigated the stability of different iron states in the lower mantle pressure range in the previous section , we now focus on the potential effect of fe@xmath1 and fe@xmath2 substitution on the pv to ppv transition .
this is a plain static calculation to access the overall influence of iron substitution on the pv to ppv transition in these phases .
we have not included finite temperature entropic or vibrational effects , therefore , we are not able to produce a full phase diagram with a two phase loop region .
this is simply the first step towards the calculation of such phase diagrams .
it is well known that the lda functional underestimates the equilibrium volume while gga overestimates it .
after inclusion of vibrational effects lda equations of state and elastic properties tend to agree better with experimental data than gga results @xcite .
therefore , we have chosen lda+u@xmath3 results to investigate the effect of fe@xmath1 and fe@xmath2 in the pv to ppv transition in both compounds .
previous studies based on _ ab initio _ calculations @xcite and experimental measurements @xcite have reported the pv to ppv transition in mgsio@xmath0 and mggeo@xmath0 to occur around @xmath4125 gpa and @xmath463 gpa , respectively , at @xmath42500 k. as indicated in fig . [
fig1](b ) and in table [ table : p_t ] , the low- to high - qs transition in si - pv and ge - pv occurs at much lower pressures than the pv to ppv transition in these minerals @xcite .
considering this fact , only the high pressure high - qs state of fe@xmath1 was considered in the investigation of the pv to ppv transition .
similarly , near the pv to ppv transition pressure , the stable state of fe@xmath2 substituted in both si - pv and si - ppv phases is hs state in the a - site ( hs - a ) and ls state in the b - site ( ls - b ) .
however , in germanate phases , the hs - b to ls - b transition pressure ( fig .
[ fig3 ] and table [ table : p_t ] ) is near the pv to ppv transition pressure . in order to find the appropriate spin states of fe@xmath2 across the transition we have considered all possibilities in the calculation of enthalpy differences ( @xmath15h = h@xmath24-h@xmath25 ) ( see fig .
[ fig5 ] ) . as indicated in fig .
[ fig5](b ) the transition pressure 55.5 gpa involves a transition between hs - b in ge - pv to ls - b in ge - ppv . obviously iron in the a site remains in the hs state . at high temperatures , contributions by other states might be relevant as well , but this issue is well beyond the scope of this paper .
this transition and transition pressure , i.e. , hs - b in pv and ls - b in ppv , is the only combination consistent with b(hs ) to b(ls ) transition pressures of 56.5 gpa and 49.5 gpa in ge - pv and ge - ppv , respectively .
the effect of fe@xmath1 and fe@xmath2 on the pv to ppv transition in si - pv and ge - pv is summarized in fig .
[ fig6 ] showing the pressure dependence of enthalpy differences ( @xmath15h = h@xmath24-h@xmath25 ) for the configurations indicated in the previous paragraph .
static ( zero temperature ) pv to ppv transition pressures in iron - free ( x=0 ) mggeo@xmath0 and mgsio@xmath0 are 43.2 gpa and 94 gpa , respectively .
the clapeyron slope for the pv to ppv transitions have been previously estimated to be @xmath4 7.5 mpa / k for mgsio@xmath0 @xcite and 7.8 mpa / k for mggeo@xmath0 @xcite . taking these clapeyron slopes approximately into account , transition pressures at @xmath42500
k should be @xmath4 62.7 gpa and @xmath4 112.8 gpa for mggeo@xmath0 and mgsio@xmath0 , respectively .
these results are in good agreement with previous experimental measurements and _ ab initio _ lda calculations @xcite .
a summary of transition pressure shifts caused by iron in silicate and germanate phases is indicated in ( fig .
[ fig6 ] , table [ table : pv - ppv - p_c ] ) .
as indicated , fe@xmath1 decreases the pv to ppv transition pressure in both mggeo@xmath0 and mgsio@xmath0 .
these observations are consistent with previous experimental @xcite and _ ab initio _ studies @xcite . however , in the case fe@xmath2 , the pv to ppv transition pressure increases noticeably . to the best of our knowledge , this fact has not been verified experimentally yet . to understand these opposite effects of fe@xmath1 and fe@xmath2 substitution on the pv to ppv transition , the volume difference between si - pv and si - ppv phase ( @xmath15v = v@xmath24-v@xmath25 ) is shown in fig .
[ fig7](a ) .
with respect to iron - free case , the volume difference is smaller and larger in magnitude with fe@xmath1 and fe@xmath2 substitution , respectively .
these smaller and larger volume difference will contribute to enthalpy difference between pv and ppv phase in such a way that will decrease the trasniton pressure with fe@xmath1 substitution while it will increase when fe@xmath2 is substituted .
this can be seen explicitly in fig .
[ fig7](b and c ) , where pressure dependence of energy ( @xmath15e ) and p@xmath26v term [ @xmath15(pv ) ] of relative enthalpy ( @xmath15h ) are shown separately . transition pressure points , where @xmath15e and @xmath15(pv ) terms cancel each other , are shown by vertical dashed lines .
using _ ab initio _ dft+u calculations , we have undertaken a comparative investigation of state changes of ferrous ( fe@xmath1 ) and ferric ( fe@xmath2 ) iron in mgsio@xmath0 and mggeo@xmath0 perovskite ( pv ) and post - perovskite ( ppv ) phases .
the goal was to verify the possibility of using mggeo@xmath0 as a low pressure analog of mgsio@xmath0 for better understanding of spin state changes in iron .
unfortunately , owing to the larger ionic radius of germanium , fe - o inter - atomic distances in the germanate phases are larger than in the silicate phases , allowing the germanates to sustain higher pressures than analog silicates before state changes in iron take place .
it has been clearly shown that such state changes , whether the displacement transition ( low- to high - qs ) of fe@xmath1 in the a - site or the hs to ls state change in fe@xmath2 in the b - site , take place at approximately the same average fe - o , @xmath19fe - o@xmath20 , distance irrespective of the chemistry or the exchange and correlation functional used in the calculations .
this encompassing investigation also allowed us to address the effect of fe@xmath1 and fe@xmath2 substitutions on the pv to ppv transition . despite the simplicity of these static calculations , previously measured and calculated effect of fe@xmath1 on this transition in mgsio@xmath0
is well reproduced : fe@xmath1 substitution decreases the transition pressure , whether in germanate or silicate phases .
however , we report the opposite effect for fe@xmath2 substitution in the b - site . this transition should involve the ls - b state in both pv and ppv silicate phases but hs - b and ls - b states in pv and ppv germanates , respectively .
for both types of transitions the effect is the same : coupled substitution of fe@xmath2 increases the pv to ppv transition in both materials .
this observation has not been validated by experiments yet and should be relevant for better understanding the d discontinuity and the nature of the deep lower mantle .
this work was supported primarily by grants nsf / ear 1319368 and nsf / career 1151738 .
computations were performed at the minnesota supercomputing institute ( msi ) and at the blue waters system at ncsa .
we thank han hsu for helpful discussions .
antonangeli , d. , j. siebert , c. m. aracne , d. l. farber , a. bosak , m. hoesch , m. krisch , f. j. ryerson , g. fiquet , and j. badro ( 2011 ) , spin crossover in ferropericlase at high pressure : a seismologically transparent transition ?
, _ science _ , _ 331 _ , 64 - 67 .
caracas , r. , and cohen , r. e. , ( 2005 ) , .
effect of chemistry on the stability and elasticity of the perovskite and post - perovskite phases in the mgsio@xmath0fesio@xmath0al@xmath27o@xmath0 system and implications for the lowermost mantle .
, _ geophys .
_ , _ 32 _ , l16310 .
catalli , k. , s .- h .
shim , v. b. prakapenka , j. zhao , w. sturhahn , p chow , y. xiao , h. liu , h. cynn , w.j .
evans ( 2010a ) , spin state of ferric iron in mgsio@xmath0 perovskite and its effect on elastic properties , _ earth planet .
_ , _ 289 _ , 6875 .
catalli , k. , s .- h .
shim , v. b. prakapenka , j. zhao , w. sturhahn ( 2010b ) , x - ray diffraction and moessbauer spectroscopy of fe@xmath2-bearing mg - silicate post - perovskite at 128138 gpa , _ american mineralogist _ , _ 95 _ , 418421 .
ito , e. , yamazaki , d. , yoshino , t. , fukui , h. , zhai , s. , shatzkiy , a. , katsura , t. , tange , y. , funakoshi , k. , ( 2010 ) , pressure generation and investigation of the post - perovskite transformation in mggeo@xmath0 by squeezing the kawai - cell equipped with sintered diamond anvils , _ earth planet .
_ , _ 293 _ , 8489 .
jackson , j.m .
, w. sturhahn , g. shen , j. zhao , m. y. hu , d. errandonea , j. d. bass , and y. fei ( 2005 ) , a synchrotron mssbauer spectroscopy study of ( mg , fe)sio@xmath0 perovskite up to 120 gpa , _ am . mineral .
_ , _ 90 _ , 199205 .
karki , b. b. , r. m. wentzcovitch . s. de gironcoli , and s. baroni ( 1999 ) , first principles determination elastic anisotropy and wave velocities of mgo at lower mantle conditions , _ science _ , _ 286 _ , 1705 .
li , j. , v. v. struzhkin , h. k. mao , j. shu , r. j. hemley , y. fei , b. mysen , p. dera , v. prakapenka , and g. shen ( 2004 ) , electronic spin state of iron in lower mantle perovskite , _ proc .
_ , _ 101 _ , 1402714030 .
li , j. , w. sturhahn , j. m. jackson , v. v. struzhkin , j. f. lin , j. zhao , h. k. mao , and g. shen ( 2006 ) , pressure effect on the electronic structure of iron in ( mg , fe)(si , al)o@xmath0 perovskite : a combined synchrotron mssbauer and x - ray emission spectroscopy study up to 100 gpa , _ phys . chem .
minerals _ , _ 33 _ , 575585 .
lin , j. , h. watson , g. vanko , e. e. alp , v. b. prakapenka , p. dera , v. v. struzhkin , a. kubo , j. zhao , c. mccammon , and w. j. evans ( 2008 ) , intermediate - spin ferrous iron in lowermost mantle post - perovskite and perovskite , _ natl .
geosci _ , _ 1 _ , 688691 .
lin , j. f. , e. e. alp , z. mao , t. inoue , c. mccammon , y. xiao , p. chow , and j. zhao ( 2012 ) , electronic spin and valence states of iron in the lower - mantle silicate perovskite by synchrotron mssbauer spectroscopy , _ american mineralogist _ , _ 97 _ , 592597 .
lin , j. f. , s. speziale , z. mao , and h. marquardt ( 2013 ) , effects of the electronic spin transitions of iron in lower - mantle minerals : implications to deep - mantle geophysics and geochemistry , _ rev .
_ , _ 51 _ , 244275 .
mao , w.l .
, g. shen , v. b. prakapenka , y. meng , a. j. campbell , d. l. heinz , j. shu , r. j. hemley , and h .- k . mao ( 2004 ) , ferromagnesian postperovskite silicates in the d layer of the earth .
101(45 ) _ , 1586715869 .
mao , z. , j.f .
lin , c. jacobs , h. c. watson , y. xiao , p. chow , e. e. alp , and v. b. prakapenka ( 2008 ) , electronic spin and valence states of fe in cairo@xmath0-type silicate post - perovskite in the earth s lowermost mantle , _ geophys .
_ , _ 37 _ , l22304 .
mccammon , c. , i. kantor , o. narygina , j. rouquette , u. ponkratz , i. sergueev , m. mezouar , v. prakapenka , and l. dubrovinsky ( 2008 ) , stable intermediate - spin ferrous iron in lower - mantle perovskite , _ nat .
_ , _ 1 _ , 684687
. mccammon , c. , l. dubrovinsky , o. narygina , i. kantor , x. wu , k. glazyrin , i. sergueev , and a. i chumakov ( 2008 ) , low - spin fe@xmath1 in silicate perovskite and a possible layer at the base of the lower mantle , _ phys .
earth planet .
_ , _ 180 _ , 215221 . nunez - valdez , m. , z. wu , y. g. yu , j. revenaugh , and r. m. wentzcovitch ( 2012a ) , thermoelastic properties of ringwoodite ( fe@xmath28mg@xmath29)@xmath27sio@xmath30 : its relationship to the 520 km seismic discontinuity , _ earth planet .
_ , _ 351 - 352 _ , 115122 . ross , n. l. , and a. navrotsky ( 1988 ) , study of the mggeo@xmath0 polymorphs ( ortho - pyroxene , clinopyroxene , and ilmenite structures ) by calorimetry , spectroscopy , and phase - equilibria , _ american mineralogist _ , _ 73 _ , 13551365 . runge , c. , a. kubo , b. kiefer , y. meng , v. prakapenka , g. shen , r. cava , and t. duffy ( 2006 ) , equation of state of mggeo@xmath0 perovskite to 65 gpa : comparison with the post - perovskite phase , _ phys . and chem . minerals _ , _ 33 _ , 699709 .
shieh , s.r .
, t. s. duffy , a. kubo , g. shen , v. b. prakapenka , n. sata , k. hirose , and y. ohishi ( 2006 ) , equation of state of the postperovskite phase synthesized from a natural ( mg , fe)sio@xmath0 orthopyroxene , _ proc .
103(9 ) _ , 3039 - 3043 . wentzcovitch , r. m. , b. b. karki , m. cococcioni , and s. de gironcoli ( 2004 ) , thermoelastic properties of of mgsio@xmath0-perovskite : insights on the nature of the earth s lower mantle , _ phys .
_ , _ 92 _ , 018501 .
wentzcovitch , r. m. , j. f. justo , z. wu , c. r. s da silva , a. yuen , and d. kohlstedtd ( 2009 ) , anomalous compressibility of ferropericlase throughout the iron spin cross - over , _ proc .
_ , _ 106 _ , 21 .
wu , z. , j. f. justo , c. r. s. da silva , s. de gironcoli , and r. m. wentzcovitch ( 2009 ) , anomalous thermodynamic properties in ferropericlase throughout its spin crossover , _ phys .
b _ , _ 80 _ , 014409 .
yu , y. g. , h. hsu , m. cococcioni , and r. m. wentzcovitch ( 2012 ) , spin states and hyperfine interactions of iron incorporated in mgsio@xmath0 post - perovskite , _ earth planet .
_ , _ 331 - 332 _ , 1 - 7 . -bearing mgsio@xmath0 in pv phase at ambient pressure .
fe , mg , o , and si are represented as purple , orange , red , and blue sphere , respectively .
mggeo@xmath0-pv structures are similar and not shown here .
( b ) enthalpy difference ( @xmath15h = h@xmath16-h@xmath17 ) for fe@xmath1-bearing si - pv ( blue ) and ge - pv ( red ) .
lower and upper bounds correspond to lda+u@xmath3 and gga+u@xmath3 results , respectively . ,
width=302 ] -bearing si - pv and ge - pv phases .
( a ) fe - o distances ( in @xmath18 ) in si - pv , and ge - pv structures at ambient pressure are shown next to the oxygen atoms .
( b ) pressure dependence of average fe - o distance .
vertical lines represent the low - qs to high - qs transition .
lower and upper bounds correspond to lda+u@xmath3 and gga+u@xmath3 results , respectively.,width=302 ] -bearing mgsio@xmath0 pv and ppv phase .
structures of pv and ppv phases of fe@xmath2-bearing mggeo@xmath0 are similar and not shown here .
fe , mg , o , and si are represented as purple , orange , red , and blue sphere , respectively .
enthalpy difference ( @xmath15h = h@xmath22-h@xmath23 ) for fe@xmath2-bearing mgsio@xmath0 ( blue ) and mggeo@xmath0 ( red ) in ( b ) pv and ( c ) ppv phases .
lower and upper bounds correspond to lda+u@xmath3 and gga+u@xmath3 results , respectively.,width=302 ] in si - pv and ge - pv phases .
( a ) zero pressure fe - o bond lengths ( @xmath18 ) in si - pv and in ge - pv obtained with lda+u@xmath3 .
( b ) pressure dependence of average fe - o bond lengths for mgsio@xmath0 ( blue ) and mggeo@xmath0 ( red ) phases .
vertical lines represent the high- ( hs ) to low - spin ( ls ) transitions in the b - site .
lower and upper bounds correspond to lda+u@xmath3 and gga+u@xmath3 results , respectively.,width=302 ] h = h@xmath24-h@xmath25 ) between fe@xmath2-bearing ge - pv and ge - ppv phase using lda+u@xmath3 method with different possible spin - state combinations of fe@xmath2 at b - site . ( a ) pv(hs ) and ppv(hs ) , ( b ) pv(hs ) and ppv(ls ) , ( c ) pv(ls ) and ppv(hs ) , ( d ) pv(ls ) and ppv(ls).,width=302 ] v = v@xmath24-v@xmath25 ) , ( b ) internal energy ( @xmath15e = e@xmath24-e@xmath25 ) , and ( c ) pv term ( @xmath15(pv)=(pv)@xmath24-(pv)@xmath25 ) of the enthalpy for iron - free ( red ) , fe@xmath1-bearing ( blue ) , and fe@xmath2-bearing ( green ) mgsio@xmath0 .
results represent lda+u@xmath3 calculation .
vertical dashed lines represent pv to ppv transition point.,width=302 ] | mggeo@xmath0-perovskite is known to be a low - pressure analog of mgsio@xmath0-perovskite in many respects , but especially in regard to the post - perovskite transition . as such , investigation of spin state changes in fe - bearing mggeo@xmath0
might help to clarify some aspects of this type of state change in fe - bearing mgsio@xmath0 .
using dft+u calculations , we have investigated pressure induced spin state changes in fe@xmath1 and fe@xmath2 in mggeo@xmath0 perovskite and post - perovskite .
owing to the relatively larger atomic size of germanium compared to silicon , germanate phases have larger unit cell volume and inter - atomic distances than equivalent silicate phases at same pressures . as a result ,
all pressure induced state changes in iron occur at higher pressures in germanate phases than in the silicate ones , be it a spin state change or position change of ( ferrous ) iron in the perovskite cage .
we showed that iron state transitions occur at particular average fe - o bond - length irrespective of mineral composition ( silicate or germanate ) or functionals ( lda+u@xmath3 or gga+u@xmath3 ) .
ferrous iron substitution decreases the perovskite to post - perovskite ( ppv ) transition pressure while coupled ferric iron substitution increases it noticeably .
iron - spin crossover , ferrous and ferric iron , pv to ppv transition |
in combinatorics , a typical problem is to establish algorithms for generating and counting of all objects of a given type .
we are interested in the situation when : such an algorithm can be described in terms of generating operations , which make possible to build the objects under scrutiny recursively from the objects with lesser number of elements ; each object can be associated with a number or a monomial in formal variables , in such a way that the generating operations correspond to certain algebraic or differential operations . as a result
, the set of all objects is mapped into a sequence of numbers or polynomials governed by certain recurrence relations .
a remarkable fact is that in some cases the generating functions for these sequences appear also in the theory of integrable equations .
we mention the relation between the number partitions and statistical mechanics @xcite , the combinatorics of painlev transcendents @xcite , of kdv solitons @xcite and of kp tau - function @xcite , as just few examples of interconnection of both theories .
a well known example , which is more close to the subject of this paper , is given by the bell polynomials which describe the statistics of set partitions and also define the potential burgers hierarchy @xcite .
our goal is to generalize this observation for several special types of set partitions with restrictions on the structure of blocks . in sections [ s : sp ] , [ s : b ]
we consider the generic set partitions and @xmath0 type partitions related , respectively , with the burgers and ibragimov shabat linearizable hierarchies .
section [ s : n - seq ] is the only one , where set partitions are replaced by other combinatorial objects ; the corresponding equations are related with the burgers hierarchy by the hodograph transformation . in all these cases
the definitions of monomials rely upon counting of elements in blocks . in section [ s :
nop ] , we study the non - overlapping partitions associated with the kdv hierarchy ; section [ s : ap ] is devoted to the atomic partitions and the kaup
broer hierarchy ( a gauge equivalent version of the nonlinear schrdinger hierarchy ) . here
, the combinatorics becomes more complicated , as well as its statistics : the definition of the monomial corresponding to a partition is based on the sequence of operations which bring to this partition , rather than on the size of its blocks .
moreover , the combinatorics becomes more disguised : in the burgers case , the generating function is related intermediately with the higher flows , while in the kdv and nls cases it becomes the asymptotic series in powers of the spectral parameter for the logarithmic derivative of @xmath1-function , governed by riccati equation .
the structure of all sections is uniform : first , we define the generating operations for the combinatorial objects under consideration , next , we introduce the polynomials which describe their statistics and derive the recurrence relations , finally , a comparison with an integrable hierarchy is given .
recall that a _ set partition _ is a set of nonempty , pairwise disjoint subsets ( _ blocks _ ) of the set , such that their union gives the whole set .
the notation @xmath2 $ ] means that @xmath3 is a partition of the set @xmath4=\{1,\dots , n\}$ ] .
the set of all partitions of @xmath4 $ ] is denoted @xmath5 and its subset consisting of all partitions with @xmath6 blocks is denoted @xmath7 , @xmath8 . as a rule , we denote the blocks by the same letter as the partition , with a subscript : @xmath9 . for @xmath10 ,
it is convenient to accept that the set @xmath11=\varnothing$ ] admits the unique partition @xmath12 which contains no blocks , that is , @xmath13 .
let @xmath14 denote the number of elements of the set @xmath15 .
then @xmath16 is the number of blocks in the set partition @xmath3 , and @xmath17 is the size of the block @xmath18 .
a block is called a singleton if @xmath19 , and a multiplet if @xmath20 .
the total number of elements in the partitioned set will be denoted @xmath21 . by definition ,
a partition is an unsorted set , that is , the order of blocks and order of elements in blocks do not matter .
nonetheless , the ordering defined by the enumeration of elements can be used for a definition of unique representation of partitions . in the _ canonical form _
, a block of a partition is represented by a list of elements sorted by increase , and the partition itself is represented by a list of blocks , sorted by increase of their first ( minimal ) elements . for short ,
it is common to write , for instance , @xmath22 , with the blocks separated by vertical bars .
alternatively , a set partition in the canonical form is represented by the sequence @xmath23 , where @xmath24 is the index of the block containing the element @xmath25 ; in the above example , it is @xmath26 .
this gives rise to the integer sequences characterized by the _ restricted growth _ condition : @xmath27 the set of such sequences of length @xmath28 will be denoted @xmath29 . throughout the paper ,
we use the graphical representation of set partitions as illustrated by fig .
[ fig : p - djm ] .
the elements are enumerated from left to right , the rows represent the blocks sorted from top to bottom by increase of minimal elements .
the line between the outermost elements of a block is called its _ support _ : @xmath30_{\mathbb r}$ ] .
this notion is used in sections [ s : nop ] , [ s : ap ] , devoted to the so - called _ non - overlapping _ and _ atomic _ partitions .
the problem of construction of all set partitions @xmath5 admits the following solution .
let us consider the operations @xmath31 defined as adding of a new element either to one of existing blocks of the partition @xmath9 , or as a new singleton : @xmath32 these operations generate partitions in the canonical form automatically , if the element @xmath33 ( or the set @xmath34 ) is appended to the corresponding list , see fig .
[ fig : p - djm ] , where the column on the right contains all possible vacancies for the new element .
, @xmath35 append new element at one of the vacancies marked by empty circles .
a block of size @xmath36 is associated with the variable @xmath37 . ]
[ st : sp.dm ] any partition of @xmath4 $ ] is generated , in a unique way , by operations @xmath38 applied to the seed partition @xmath12 .
the sequence of operations is recovered uniquely by deleting the elements in the reverse order , from @xmath28 to @xmath39 . in terms of the restricted growth sequences
@xmath40 , the generating operations amount to appending of an integer @xmath41 $ ] , where @xmath42 , and the operation @xmath35 corresponds to the value @xmath43 .
in fact , @xmath44 encodes the sequence of operations which generate a given partition .
alternatively , it can be written as follows .
let @xmath45 denote the set of words consisting of @xmath28 characters @xmath46 and @xmath35 , such that the subscript for any character @xmath47 may take values in the range from 1 to the number of occurrences of @xmath35 to the right from this character .
statement [ st : sp.dm ] establishes the bijection @xmath48 between @xmath45 and @xmath5 .
although we do not use this representation of set partitions in this section , its generalizations will be important in the study of special partition in sections [ s : nop ] , [ s : ap ] .
the generating operations provide a plain and obvious construction algorithm for the set partitions , which is implemented , in particular , in the _ combinatorica _ package @xcite .
it should be mentioned , that there exist also more effective algorithms based on the gray codes ( * ? ? ?
* ch.7.2.1.5 ) , @xcite .
let us turn to the problem of counting of the set partitions .
let the variable @xmath37 be assigned to any block @xmath18 with @xmath36 elements . in the corresponding restricted growth sequence
, @xmath36 is the multiplicity of the entry @xmath49 .
a partition is associated with the monomial equal to the product of variables assigned to all blocks : @xmath50 summation over all partitions of @xmath4 $ ] defines the ( _ complete exponential _ ) _ bell polynomials _ ( see e.g. @xcite ) @xmath51 in the polynomial @xmath52 , the degrees in any monomial @xmath53 are related by equality @xmath54 ( that is , the polynomial is homogeneous with respect to the weight @xmath55 ) , while the coefficient of the monomial is equal to the number of partitions of @xmath4 $ ] into @xmath56 blocks with 1 element , , @xmath57 blocks with @xmath58 elements , for instance : @xmath59{rc } y_1= & v_1 \\ n= & 1 \\[5pt ] & 1 \end{array}\qquad \begin{array}[t]{rccc } y_2= & v_2 & + & v^2_1 \\ n= & 2 & = & 1 + 1 \\[5pt ] & 12 & & 1|2 \end{array}\qquad \begin{array}[t]{rccccc } y_3= & v_3 & + & 3v_1v_2 & + & v^3_1 \\ n= & 3 & = & 1 + 2 & = & 1 + 1 + 1 \\[5pt ] & 123 & & 1|23 & & 1|2|3 \\ & & & 12|3 & & \\ & & & 13|2 & & \end{array}\]]@xmath59{rccccccccc } y_4=&v_4 & + & 4v_1v_3 & + & 3v^2_2 & + & 6v^2_1v_2 & + & v^4_1 \\
n=&4 & = & 1 + 3 & = & 2 + 2 & = & 1 + 1 + 2 & = & 1 + 1 + 1 + 1 \\[5pt ] & 1234 & & 1|234 & & 12|34 & & 1|2|34 & & 1|2|3|4 \\ & & & 123|4 & & 13|24 & & 1|23|4 & & \\ & & & 124|3 & & 14|23 & & 1|24|3 & & \\ & & & 134|2 & & & & 12|3|4 & & \\ & & & & & & & 13|2|4 & & \\ & & & & & & & 14|2|3 & & \end{array}\ ] ] at @xmath10 , we have @xmath60 , in accordance with the common convention that a product over empty set is equal to @xmath39 .
the number of set partitions with blocks of prescribed size is computed explicitly and we arrive to the formula @xmath61 where the sum is taken over all _ partitions of the number _ @xmath28 .
the following statement provides a more effective computation method . in this statement and further on
, @xmath62 denotes the differentiation with respect to the variable @xmath63 , and variables @xmath37 are interpreted as derivatives of a function @xmath64 .
[ st : yn ] polynomials @xmath52 in variables @xmath65 are governed by the recurrence relation @xmath66 the exponential generating function of the sequence @xmath52 is @xmath67 let @xmath68 .
the action of the differentiation @xmath69 on the monomial @xmath70 amounts , according to the leibniz rule , to replacement of @xmath37 by @xmath71 for each factor in turn , taking the multiplicity into account . in the partitions language
, this corresponds to adding of a new element to one of the blocks in turn .
as the result , we obtain the sum of monomials @xmath72 with all admissible values of @xmath49 .
the multiplication of monomial @xmath70 by @xmath73 yields the monomial @xmath74 .
thus , @xmath75 , where the sum is taken over the partitions @xmath76 obtained from @xmath3 by adding the element @xmath33 in all possible ways .
summation over @xmath3 brings to equation ( [ yn - rec ] ) , taking statement [ st : sp.dm ] into account .
this equation can be cast to the form @xmath77 which implies @xmath78 . for the exponential generating
function , we obtain @xmath79 and expansion of @xmath80 into the taylor series with respect to @xmath81 gives ( [ yn - egf ] ) .
@xmath82 @xmath83 the few first polynomials @xmath52 are written down in table [ t : yn ] .
the terms of degree @xmath6 called partial bell polynomials @xmath84 appear in the fa di bruno formula for the @xmath28-th derivative of a composite function : @xmath85 ( see discussion in @xcite ) . by summing up the coefficients of @xmath84 , which is equivalent to identifying all variables @xmath37 , we forget about the sizes of blocks and consider just their number in a given partition .
this gives rise to the bell touchard polynomials in one variable @xmath86 where the coefficient of @xmath87 , the _ stirling number of the second kind _ , is equal to the number of partitions of @xmath4
$ ] into @xmath6 blocks ( * ? ? ?
* http://oeis.org/a048993[a048993 ] ) .
by definition , @xmath88 at @xmath89 and @xmath90 , in accordance with the convention @xmath91 .
the substitution @xmath92 turns the recurrence relation ( [ yn - rec ] ) into @xmath93 and the generating function ( [ yn - egf ] ) takes the form @xmath94 the total number of partitions of a set with @xmath28 elements , the _ bell _ or _ exponential number _
* http://oeis.org/a000110[a000110 ] ) , is defined by equations @xmath95 let us recall the notion of generalized symmetry from the theory of integrable equations .
let @xmath96 and @xmath97 denote , as before , the derivatives with respect to the distinguished variable @xmath63 .
the evolution equation @xmath98 admits the symmetry @xmath99 ( classical at @xmath100 and generalized , or higher , at @xmath101 ) , if the corresponding flows commute , that is , the equality @xmath102 holds identically with respect to @xmath103 , where @xmath104 is the differential operator @xmath105 the equation is considered integrable if it admits a sequence ( hierarchy ) of symmetries of arbitrarily large order .
moreover , in a typical situation , the higher symmetries commute with each other .
the potential burgers hierarchy @xmath106 satisfies this commutativity property .
this follows easily from its equivalence to the linear hierarchy @xmath107 namely , the change @xmath108 results in @xmath109 the mutual commutativity of the flows ( [ heat ] ) is obvious ( @xmath110 ) and it is not difficult to prove that this property is preserved under the point transformations , see e.g. @xcite . more generally , in the theory of integrability it is common to consider equivalent the equations related by point or contact transformations and also some non - invertible differential substitutions , because these changes do not affect such properties as the existence of higher symmetries , conservation laws , lax pairs and so on .
the combinatorial interpretation is , however , not invariant .
in contrast , we have seen that meaningful combinatorics may appear just from nothing , as a result of simple change of variables between equations ( [ heat ] ) and ( [ pot - burgers ] ) .
further examples of this kind are given in sections [ s : u2u2 ] , [ s : is ] .
thus , a combinatorics is associated with a concrete form of a hierarchy .
the commutativity follows also from the identity ( [ y*y ] ) below , which coincides with the recurrence relation ( [ yn - rec ] ) at @xmath111 , @xmath112 , @xmath113 .
the general case can be easily proven by induction with respect to @xmath25 , but instead we will generalize the combinatorial proof of equation ( [ yn - rec ] ) . first , let us introduce the following notation .
the _ concatenation _ of partitions @xmath2 $ ] , @xmath114 $ ] is the partition @xmath115,\ ] ] where @xmath116 denotes the partition of @xmath117 $ ] obtained from @xmath118 by adding @xmath28 to all elements .
( compare with the _ mathematica _ language convention that adding of a scalar to a list acts on each entry : @xmath119 . ) ) . ]
[ st : yy ] the bell polynomials satisfy the identity @xmath120 given @xmath2 $ ] , @xmath114 $ ] , let us construct the partition @xmath121 ( fig . [
fig : concatenation ] on the left ) .
the corresponding monomials ( [ ppi ] ) satisfy the relation @xmath122 .
next , let us consider the partitions obtained from @xmath123 by deleting of one of the size @xmath36 blocks of @xmath116 and adding its elements to blocks of @xmath3 , in all possible ways ( one such partition is shown on the right of fig .
[ fig : concatenation ] ) .
the sum of the corresponding monomials is equal to @xmath124 .
vise versa , given @xmath125 $ ] , let us consider the blocks with the minimal element not greater than @xmath28 .
their intersections with @xmath4 $ ] uniquely define the partition @xmath3 , while the block formed from the rests of these blocks and the rest blocks of @xmath123 constitute the partition @xmath116 . therefore , the above operations give all partitions @xmath126 , without repetitions , when @xmath3 and @xmath118 run , respectively , over @xmath5 and @xmath127 .
the summation of monomials gives @xmath128 as required .
the right hand sides of equations ( [ pot - burgers ] ) do not contain @xmath129 , which makes possible the substitution @xmath130 .
this brings to the burgers hierarchy @xmath131 homogeneous with respect to the weight @xmath132 .
the few first equations are @xmath133 what is the combinatorial interpretation of the coefficients in this case ?
the differentiation in equation ( [ burgers ] ) corresponds to adding , turn by turn , of an element to blocks of partitions @xmath5 .
since we do not add a new block in this case , hence the partitions under consideration are constructed in the same manner as in section [ s : gf ] , but the operation @xmath35 is not used on the last step . as a result , only those of partitions @xmath134 are constructed , which do not contain element @xmath33 as a singleton ; for instance , at @xmath135 : @xmath136 123&&1|23&&1|2|3&&1234 & & 1|234 & & 12|34 & & 1|2|34 & & { \settowidth{\slen}{$1|2|3|4$}\hbox to 0pt{\rule[0.5ex]{\slen}{0.5pt}\hss}1|2|3|4}\\ & & 12|3 & & & & & & { \settowidth{\slen}{$123|4$}\hbox to 0pt{\rule[0.5ex]{\slen}{0.5pt}\hss}123|4}&&13|24&&{\settowidth{\slen}{$1|23|4$}\hbox to 0pt{\rule[0.5ex]{\slen}{0.5pt}\hss}1|23|4 } & & \\ & & 13|2 & & & & & & 124|3 & & 14|23 & & 1|24|3 & & \\ & & & & & & & & 134|2 & & & & { \settowidth{\slen}{$12|3|4$}\hbox to 0pt{\rule[0.5ex]{\slen}{0.5pt}\hss}12|3|4 } & & \\ & & & & & & & & & & & & { \settowidth{\slen}{$13|2|4$}\hbox to 0pt{\rule[0.5ex]{\slen}{0.5pt}\hss}13|2|4 } & & \\ & & & & & & & & & & & & 14|2|3 & & \end{array}\ ] ] renumbering makes possible to replace the element @xmath33 with any other one , and we arrive to the following statement .
[ st : burgers ] in the burgers hierarchy , the coefficient of the monomial @xmath137 is equal to the number of partitions of the set @xmath138 $ ] into @xmath139 blocks with 1 element , , @xmath57 blocks with @xmath140 elements , under the additional condition that a distinguished element of the set is not a singleton .
the total number of partitions of @xmath141 $ ] such that a distinguished element is not a singleton is found by setting all variables equal to @xmath39 : @xmath142 these integers form the sequence @xmath143 according to ( * ? ? ? * http://oeis.org/a005493[a005493 ] ) , it can be characterized also as the total number of blocks in all partitions of @xmath4 $ ] .
indeed , the partitions of @xmath141 $ ] under consideration are obtained from partitions of @xmath4 $ ] by enlarging of one of the blocks , and this operation is applied exactly as many times as many blocks there are in all partitions .
let @xmath144 denote the set of integer sequences @xmath145 satisfying the _ natural growth _ condition @xmath146 . the sequence @xmath147 can be matched to an upper triangular @xmath148 matrix , with one unit in each column and rest elements equal to zero , see fig .
[ fig : n - seq ] .
it is clear that @xmath149\times[2]\times\cdots\times[n]$ ] , that is , the structure of this set is simpler comparing to the set @xmath29 of the restricted growth sequences ( [ rgs ] ) .
in particular , @xmath150 , instead of the bell numbers for @xmath151 .
nevertheless , these sequences also support a quite meaningful generating function . for any sequence @xmath152 ,
let us count the occurrences of integers from @xmath39 to @xmath28 and form the monomial @xmath153 summation over all sequences defines the polynomials @xmath154 analogous to the definition of the bell polynomials ( [ yn ] ) for the restricted growth sequences . in both cases , the definition of monomials is related with counting of elements in the level sets , which are the rows of the upper triangular matrix or the partition blocks .
notice , that one can count the units in the diagonals of the matrix instead of the rows , resulting in the monomial @xmath155 . in general ,
@xmath156 , but , nevertheless , the summation over all @xmath147 gives the same result .
indeed , if we reverse a part of each column , from the first element to the diagonal one , then a matrix of the same type appears , and this transform is bijective and takes the rows into the diagonals .
for the sequences , it is given by the map @xmath157 . in order to derive the recurrence relation for @xmath158 we assume , as before , that @xmath159 , @xmath62 . .
distribution by rows gives @xmath160 , distribution by diagonals gives @xmath161 . ]
[ st : hn ] the polynomials @xmath158 in variables @xmath162 satisfy the recurrence relation @xmath163 that is , @xmath164 .
let us interpret the sequence @xmath147 as a distribution of enumerated balls over enumerated boxes , possibly empty , such that @xmath165-th ball is allowed to be only in first @xmath165 boxes .
the variable @xmath166 is assigned to each box with @xmath36 balls , and the monomial ( [ pa ] ) is equal to their product .
generating operations for @xmath144 are extremely simple : the passage to @xmath167 amounts to appending of an arbitrary integer @xmath168 $ ] .
let @xmath169 , @xmath170 .
these operations are interpreted as follows : first , we add the empty box number @xmath33 , this multiplies @xmath171 by @xmath172 ; next , we put the ball number @xmath33 into the box number @xmath49 , this replaces the variable @xmath166 assigned to this box with @xmath173 . as a result , @xmath171 is mapped into the sum of monomials over all sequences @xmath174 with the same first @xmath28 elements as in @xmath147 : @xmath175 summation over @xmath152 completes the proof .
in contrast to the bell polynomials , @xmath158 are homogeneous not only with respect to the weight @xmath176 , but also with respect to the degree .
the few first polynomials are written down in table [ t : hn ] , with the terms collected by powers of the distinguished variable @xmath172 .
let the rest variables be equal to @xmath129 and denote @xmath177 , this gives the homogeneous polynomials in two variables @xmath178 where the _ euler number _
@xmath179 is equal to the number of sequences from @xmath144 which take @xmath6 different values ( * ? ? ?
* http://oeis.org/a008292[a008292 ] ) .
the total sum of the coefficients is , obviously , @xmath180 .
@xmath181 @xmath182 the above substitution replaces the operator @xmath69 with @xmath183 and the recurrence relation ( [ hn - rec ] ) takes the form @xmath184 which implies the symmetry property and the recurrence relation for the euler numbers @xmath185 it should be mentioned , that the euler numbers admit also many other interpretations , related with the permutations of @xmath28 elements , in particular , @xmath179 is equal to the number of permutations @xmath186 with @xmath187 descents ( that is , positions @xmath165 , such that @xmath188 ) , and also to the number of permutations with @xmath187 exceedances ( positions @xmath165 , such that @xmath189 ) .
in fact , the permutations admit a wide variety of combinatorial interpretations and representations , resulting in diversity of associated statistics , see e.g. @xcite .
the natural growth sequences also appears in the theory of permutations , as the inversion vectors and lehmer codes .
recall , that the inversion vector is a sequence @xmath190 , such that @xmath191 is equal to the number of elements @xmath123 on the left of @xmath165 and greater than @xmath165 , that is @xmath192 .
it is clear that @xmath193 $ ] , therefore , adding 1 to the entries of an inversion vector and taking its reverse bring to a sequence from @xmath144 .
this mapping @xmath194 is a bijection , see e.g. @xcite .
the polynomials @xmath158 appear , up to some denotation , in the computation of the inverse function derivatives , like the bell polynomials appear in the formula for the derivatives of a composite function ( [ faa ] ) .
indeed , let @xmath196 , @xmath197 , then @xmath198 ( cf . with the definition of @xmath158 in ( * ? ? ?
* http://oeis.org/a145271[a145271 ] ) ) .
essentially , this formula is used in the hodograph type transformation of the linear hierarchy ( [ heat ] ) which we rewrite now in the form @xmath199 under the change of variables @xmath200 , @xmath201 , @xmath202 , the derivatives are computed according to the rule @xmath203 and the equation takes the form @xmath204 .
further differential substitution @xmath205 brings to equations @xmath206 in particular , first two flows are @xmath207 ( a discussion of this example can be found in ( * ? ? ?
* sect.20.1 ) ) .
the commutativity of equations ( [ unun ] ) follows , like in section [ s : pot - burgers ] , either from their relation to equations ( [ heat ] ) or intermediately from the identity @xmath208 which can be easily proved by induction with respect to @xmath25 .
it would be interesting to find its combinatorial meaning , like in statement [ st : yy ] .
set partitions of @xmath0 type @xcite , see also @xcite , appear if we employ the reflection symmetry of the set under consideration .
a partition @xmath3 of the set @xmath209 is called @xmath0 type partition if : \1 ) @xmath210 , that is , for any block @xmath211 also @xmath212 ; \2 ) @xmath3 contains just one block @xmath213 ( called @xmath214-block ) , such that @xmath215 .
all such partitions are denoted @xmath216 , and partitions with @xmath6 block pairs are denoted @xmath217 , @xmath218 .
the canonical form of @xmath0 type partition is defined as follows : in @xmath214-block , all negative elements are removed ; in all blocks , the elements are sorted by increase of the absolute value ; from any block pair we keep only the block with positive first element ; the blocks are sorted by increase of their first elements .
moreover , it is convenient to write @xmath219 instead of @xmath220 .
for instance , the partition @xmath221 is encoded as @xmath222 .
the pictorial representation is clear from fig .
[ fig : bp - djm ] .
notice , that deleting elements @xmath223 from any partition @xmath216 yields a partition from @xmath224 .
therefore , @xmath216 is obtained from @xmath224 by adding @xmath223 in all possible ways , namely : @xmath225 , adding both elements @xmath223 to 0-block ; @xmath226 , adding @xmath223 to the blocks @xmath227 , @xmath228 ; @xmath229 , adding @xmath223 to the blocks @xmath230 , @xmath228 ; @xmath231 , adding a new block pair @xmath232 , @xmath233 .
these operations generate , in a unique way , any @xmath0 type partition starting from the trivial partition of the set @xmath234 .
type partition ; the blocks are sorted downwards by increase of elements with minimal absolute value . on the right : representation of the canonical form ; negative elements are marked @xmath235 , positive @xmath236 . generating operations
add element to one of the vacant positions marked by empty triangles .
the corresponding monomials ( [ bppi ] ) are written near to the blocks . ]
let us define the mapping @xmath237 from @xmath216 into the set of monomials of variables @xmath238 .
given a block @xmath239 , let @xmath240 denote the number of its positive elements : @xmath241 it is clear that the number of negative elements in the block is equal to @xmath242 . let a partition @xmath243 consists of 0-block @xmath213 and block pairs @xmath244
, @xmath245 , such that the elements with minimal absolute value are positive in blocks @xmath18 .
we assign the monomial of degree @xmath246 @xmath247 to such a partition , so that all partitions @xmath216 are associated with the polynomial @xmath248 which serves as @xmath249-analog of the bell polynomial @xmath52 . as an example , let us write down all partitions @xmath250 , collecting together the partitions which correspond to one and the same monomial ( as usual , @xmath172 ) : @xmath251{rccccccccc } a_3= & u_3 & + & 5u^2u_2 & + & 8uu^2_1 & + & 9u^4u_1 & + & u^7 \\[5pt ] & 0123 & & 0|123 & & 0|12\bar3 & & 0|12|3 & & 0|1|2|3 \\ & & & 0|1\bar2\bar3 & & 0|1\bar23 & & 0|1\bar2|3 & & \\ & & & 012|3 & & 01|23 & & 0|13|2 & & \\ & & & 013|2 & & 01|2\bar3 & & 0|1\bar3|2 & & \\ & & & 023|1 & & 02|13 & & 0|23|1 & & \\ & & & & & 02|1\bar3 & & 0|2\bar3|1 & & \\ & & & & & 03|12 & & 01|2|3 & & \\ & & & & & 03|1\bar2 & & 02|1|3 & & \\ & & & & & & & 03|1|2 & & \end{array}\ ] ] [ st : an ] polynomials ( [ an ] ) in variables @xmath252 , @xmath62 , satisfy the recurrence relation @xmath253 let @xmath254 and let us track the monomial @xmath70 under the action of generating operations : @xmath255 : the factor @xmath256 is replaced with @xmath257 ; @xmath46 : the factor @xmath258 is replaced with @xmath259 ; @xmath260 : the factor @xmath261 is replaced with @xmath262 ; @xmath35 : two factors @xmath263 are added .
therefore , the application of all possible operations @xmath264 replaces the monomial @xmath70 with @xmath265 ; the summation over @xmath3 completes the proof .
@xmath266 @xmath267 the few first polynomials @xmath268 are given in table [ t : an ] .
let us simplify the statistics by merging the terms of the same degree ; this brings to the polynomials @xmath269 .
the coefficients of @xmath270 are the numbers of partitions @xmath217 , that is , the analogs of the stirling numbers of the 2nd kind for @xmath0 type partitions ( * ? ? ?
* http://oeis.org/a039755[a039755 ] ) .
the total coefficient sums @xmath271 give the total numbers of partitions @xmath216 which constitute the sequence of @xmath0-analogs of the bell numbers , or the _ dowling numbers _ ( * ? ? ?
* http://oeis.org/a007405[a007405 ] ) . the diagram @xmath272 \qquad\updownarrow \psi^2=s & & \qquad\updownarrow u^2=v \\[7pt ] s_{,t_3}=d\bigl(s_2-\dfrac{3s^2_1}{4s}\bigr ) & & v_{,t_3}=d\bigl(v_2-\dfrac{3v^2_1}{4v}+3vv_1+v^3\bigr)\\[7pt ] \qquad\uparrow s = q_1 & & \qquad\uparrow v = w_1 \\[7pt ] q_{,t_3}=q_3-\dfrac{3q^2_2}{4q_1 } & \overset{q = e^{2w}}{\leftarrow\!\!\!-\!\!-\!\!\!\rightarrow } & w_{,t_3}=w_3-\dfrac{3w^2_2}{4w_1}+3w_1w_2+w^3_1 \end{array}\ ] ] describes the chain of point transformations and substitutions of introducing a potential type which linearizes the ibragimov
shabat equation @xcite .
the whole hierarchy consists of equations related by the same transformations with equations @xmath273 .
the absence of even - order flows is explained by the fact that , although the changes look quite harmless , they partially destroy the symmetry algebra . for @xmath274 , the change @xmath275 brings to an equation with the total derivative in the right hand side : @xmath276 in contrast , the analogous equation for @xmath277 contains the term @xmath278 outside the parentheses , that is @xmath279
; so that the further substitution @xmath280 leads out of the class of evolution equations .
the structure of odd flows is described by the following statement .
denote @xmath281 , @xmath282 , @xmath283 , then the ibragimov
shabat hierarchy is equivalent to equations @xmath284 and the coefficients @xmath268 coincide with the polynomials ( [ an - rec ] ) .
let us consider the generating function @xmath285 and define @xmath15 by equation @xmath286 .
the change of variables @xmath287 maps the identity @xmath288 into equation ( [ a ] ) , which is exactly equivalent to recurrence relations ( [ an - rec ] ) .
next , let @xmath289 , then ( cf . with ( [ stn ] ) ) @xmath290 by applying @xmath291 , we obtain @xmath292 which implies @xmath293 the second equality in ( [ ut ] ) is obtained by eliminating of derivatives in virtue of ( [ a ] ) .
the definition of non - overlapping partitions @xcite makes use of the ordering of the set @xmath4 $ ] .
[ def : nop ] blocks @xmath294 and @xmath239 of a partition @xmath2 $ ] overlap , if @xmath295 a partition is called _ non - overlapping _ if no pair of its blocks overlap .
equivalently , the supports of any two blocks either do not intersect or one of them contains another one . for instance , the partition in the fig .
[ fig : p - djm ] does not satisfy this requirement , because the blocks @xmath296 overlap ( as well as @xmath297 ; the other pairs do not ) .
all non - overlapping partitions of @xmath4 $ ] are denoted @xmath298 , and those which contain @xmath6 multiplets ( and any number of singletons ) will be denoted @xmath299 , @xmath300 .
it is natural to consider @xmath12 non - overlapping partition , that is , @xmath301 .
also , let @xmath302 denote the set of non - overlapping partitions of @xmath4 $ ] without singletons . in papers
@xcite , @xmath303 denotes the number of non - overlapping partitions of @xmath4 $ ] with @xmath6 blocks of any size .
counting multiplets only turns out to be more convenient in our construction method for partitions . in this approach ,
one may collect , in the mind s eye , all singletons of a partition into one distinguished block ( possibly , empty ) .
notice , that this establishes a bijection between @xmath298 and a subset of @xmath304 which consists of partitions with elements @xmath39 and @xmath305 in one block .
the generating operations for non - overlapping partitions are more complicated than in the previous sections .
these include the operations @xmath306 which we now describe . for non - overlapping partitions .
] _ operation @xmath255 _ , adding of the singleton @xmath34 to a partition .
this coincides with the operation @xmath35 for generic set partitions from section [ s : gf ] , but we use another notation and interpret @xmath255 as adding the element into the team of singletons .
notice , that @xmath307 . _
operation @xmath46 , @xmath308 _ , adding the element @xmath33 to a multiplet .
let @xmath309 be all multiplets in @xmath310 , sorted by increase of their minimal elements .
if we just add the element @xmath33 to @xmath311 then the multiplets with the supports embracing @xmath311 will overlap with this block ( see fig .
[ fig : nop - dj ] above ) .
let these multiplets are those with the indices @xmath312 .
let us divide them into the left and right parts with respect to @xmath313 : @xmath314 and form new multiplets by cyclic permutation of the right parts ( fig .
[ fig : nop - dj ] below ) : @xmath315 notice , that the block @xmath316 , which acquires the element @xmath33 , contains 3 elements or more .
it is easy to see that the described procedure brings to the blocks which do not overlap with each other and with the rest multiplets . _
operation @xmath317 .
_ let @xmath310 , @xmath318 , then @xmath319 this is a kind of concatenation ( [ | ] ) , with the wrapping of the second partition by a doublet ( see fig .
[ fig : nop - p ] ) .
in particular , @xmath320 . for non - overlapping partitions .
] [ st : nop.dp ] applying all admissible operations @xmath321 to the seed partition @xmath12 generates all non - overlapping partitions : @xmath322 moreover , any partition appears only once .
the last of operations leading to a given partition @xmath323 is defined by the block @xmath239 with the element @xmath33 : if @xmath324 then @xmath325 ; if @xmath326 then @xmath327 ; if @xmath328 then @xmath329 , where @xmath49 is the maximal index of the multiplet with the support containing the element @xmath330 which precedes @xmath33 . in all cases , the partitions @xmath118 or @xmath331
are determined uniquely and consist of lesser number of elements , so that the induction can be applied .
the generating of a given partition is encoded by expression @xmath332 , builded from a formal variable @xmath263 by the binary operation @xmath333 and operations @xmath334 , @xmath335 , where @xmath336 is equal to the number of occurrences of @xmath263 in @xmath147 .
let the set @xmath337 consist of all such expressions containing @xmath33 characters @xmath338 , and the set @xmath339 include those of them which contain @xmath43 characters @xmath263 .
then @xmath340 moreover , @xmath300 , because the number of occurrences of @xmath317 in @xmath341 is always one less than the number of occurrences of @xmath263 ( that is , it is equal to @xmath342 ) .
a comparison with the action of the operations on the partitions leads to the following bijection , taking the statement [ st : nop.dp ] into account . [
st : nop.phi ] the substitution @xmath343 is a one - to - one mapping @xmath344 , for all @xmath345
. table [ t : phi * ] contains a complete list of expressions @xmath337 and respective partitions @xmath298 for @xmath346 .
notice , that @xmath347 at @xmath348 , and in @xmath349 only the partition @xmath350 is eliminated .
the non - overlapping partitions without singletons @xmath302 are builded by the same scheme , but with @xmath255 excluded from the set of generating operations .
the corresponding set @xmath351 consists of all expressions @xmath337 with no occurrence of @xmath255 .
@xmath352 in sections [ s : sp ] , [ s : b ] we assigned a monomial to any partition by counting the elements ( or positive elements ) in the blocks and by multiplying the corresponding variables @xmath166 .
for the non - overlapping partitions this does not bring to a meaningful result , because the operations @xmath46 change the sizes of blocks unpredictably . in this case , a suitable definition comes from the encoding of partitions by expressions @xmath353 .
we define the monomial @xmath354 as the value of expression @xmath341 , computed according to the following rules : \(i ) operation @xmath334 adds 1 to the subscript of @xmath263 for @xmath355-th occurrence of @xmath263 in @xmath147 , counting from the left ( as usual , we assume that @xmath172 ) ; \(ii ) all operations @xmath333 are applied after @xmath46 , and their result is equal to the product @xmath356 .
let us define the polynomials @xmath357 by summation over all expressions @xmath337 , and consider the generating function @xmath358 [ st : fn ] polynomials @xmath359 in variables @xmath252 , @xmath62 , are governed by recurrence relations @xmath360 equivalent to the riccati equation @xmath361 equation ( [ fn ] ) follows from ( [ pi * ] ) and identities @xmath362 where @xmath353 , @xmath363 .
the second identity follows from the rule ( ii ) . in order to prove the first one ,
let us preserve the same order of variables @xmath364 in a monomial , as they appear after applying of operations @xmath46 according to the rule ( i ) . then ,
if a monomial corresponding to expression @xmath365 is equal to @xmath366 , then @xmath367 , @xmath368 , and we only have to sum over @xmath49 . in this proof , the operation @xmath46 is identified with the action of differentiation @xmath69 onto one of the factors according to the leibniz rule .
this leads to the following simple description : _ in polynomials @xmath359 , the coefficient of any monomial is equal to the number of ways to obtain this monomial by multiplication and differentiation .
_ for instance , there are 4 ways for the monomial @xmath369 ( see table [ t : phi * ] ) : first multiply @xmath263 by @xmath263 and next apply @xmath69 to the first or second factor , or first apply @xmath69 to @xmath263 and then multiply by @xmath263 from left or from right .
the few first polynomials @xmath359 are written down in table [ t : fn ] .
now , let us consider the set @xmath302 of non - overlapping partitions without singletons .
since the corresponding expressions @xmath341 do not contain @xmath255 , hence the first occurrence of @xmath263 is not differentiated when computing the monomial @xmath354 , and therefore @xmath354 is divisible by @xmath263 .
so , let us assign to these partitions the polynomial @xmath370 in particular , one finds easily from table [ t : phi * ] @xmath371 it turns out that these polynomials are very simply related with the polynomials for all non - overlapping partitions .
@xmath372 @xmath373 [ st : tfn ] the generating function for @xmath374 is equal to @xmath375 by repeating the arguments in the previous proof , we obtain that @xmath376 satisfies the equation @xmath377 . from here ,
all @xmath374 are determined uniquely through recurrence relations which can be easy written down .
the change @xmath378 brings to equation ( [ df ] ) , and since the leading term of @xmath379 is equal to @xmath380 , hence @xmath379 coincides with the series ( [ f ] ) .
let us pass to the simplified statistics , as in the previous sections , by merging the terms of the same degree in @xmath359 under the substitution @xmath381 . since @xmath382 is one more than the number of multiplets in the corresponding partition @xmath383 , hence the coefficient of @xmath384 is equal to the cardinality of the set @xmath339 , or , what is the same , of the set @xmath299 .
[ st:|pi*nk| ] the number @xmath385 of non - overlapping partitions of @xmath4 $ ] with @xmath6 multiplets is equal to the coefficient of @xmath384 in the polynomial in one variable @xmath386 , defined by recurrence relations @xmath387 the generating function @xmath388 satisfies the equation @xmath389 [ st:|tpi*nk| ] the number @xmath390 of non - overlapping partitions @xmath4 $ ] with @xmath6 multiplets and without singletons is equal to the coefficient of @xmath391 in the polynomial @xmath392 , where @xmath393 the substitution @xmath394 , @xmath395 reduces ( [ f ] ) to the bessel equation @xmath396 with the linearly independent solutions @xmath397 , where @xmath398 is the bessel function @xmath399 this function satisfies the recurrence relations @xmath400 generating functions @xmath401 satisfy the corresponding riccati equations , as formal asymptotic series with respect to the parameter @xmath81 . taking into account that @xmath402 , we obtain the asymptotic expansion @xmath403 _ bessel numbers _ @xmath404 counting the non - overlapping partitions of @xmath4 $ ] are equal to the total sums of the coefficients of @xmath359 .
analogously , _ 2-associated bessel numbers _
@xmath405 count the non - overlapping partitions without singletons .
for the first time , both sequences were defined in @xcite , where an asymptotic expansion was obtained which coincides with ( [ fj ] ) under the substitution @xmath406 .
however , its derivation was based on a different combinatorial technique : instead of the above generating operations which bring to the riccati equation , the so - called path diagrams were used @xcite which lead to a representation of @xmath407 as a continued fraction , which is equivalent to the three - term recurrence relation for @xmath398 . from here , a functional equation for the generating function @xmath408 can be derived , which implies a recurrence relation intermediately for the numbers @xmath409 ( * ? ? ?
* eq.(22 ) ) , see also @xcite , ( * ? ? ?
* http://oeis.org/a006789[a006789 ] ) and
* http://oeis.org/a099950[a099950 ] ) ( sequence @xmath410 ) .
this is equivalent to the following statement after elimiantion of @xmath411 .
[ st : bb * ] sequences @xmath412 satisfy the recurrence relations @xmath413 the equation for @xmath414 follows from the corollary [ st:|tpi*nk| ] , the equation for @xmath409 appears if we enlarge the partitions without singletons until the necessary size , by inserting singletons in all possible ways .
the diagonal entries of the number triangle in table [ t : fn ] , that is , the coefficients of `` dispersionless terms '' @xmath384 in @xmath415 , form the sequence of catalan numbers ( * ? ? ?
* http://oeis.org/a000108[a000108 ] ) .
these monomials correspond to partitions builded only by operation @xmath317 , that is , consisting only from non - overlapping doublets , which can be easily identified with the strings of balanced parentheses .
correspondingly , the replacement of the initial condition in equation ( [ fn ] ) by 1 leads to the recurrence for the catalan numbers interpolated with zeroes ( * ? ? ?
* http://oeis.org/a126120[a126120 ] ) : @xmath416 recall that the riccati equation ( [ df ] ) plays the key role in the theory of kdv equation @xmath417 as a tool for computing both conservation laws and higher symmetries @xcite .
the kdv hierarchy is defined by the compatibility conditions for the schrdinger equation @xmath418 and equations of the form @xmath419 , where @xmath420 .
a straightforward computation brings to equations @xmath421 and the recurrence relations @xmath422 this implies the equation for the generating function @xmath423 @xmath424 which uniquely defines all @xmath425 as polynomials , homogeneous with respect to the weight @xmath426 ( the right hand side of ( [ kdv.g ] ) can be replaced with arbitrary constant series in powers of @xmath427 , then the homogeneous flows are replaced by their linear combinations ) .
it is easy to check that solution of equation ( [ kdv.g ] ) is expressed through solutions of equation ( [ df ] ) : @xmath428 which is equivalent to equations @xmath429 complementary to recurrence relations ( [ fn ] ) .
equation ( [ df ] ) is brought to ( [ kdv.psi ] ) by substitution @xmath430 and , therefore , equation ( [ gf ] ) is equivalent to relation @xmath431 .
the equalities @xmath432 demonstrate that @xmath379 is a generating function for the densities of conservation laws , and the densities @xmath433 are trivial .
although the relation between the generating functions @xmath379 and @xmath434 is very simple and looks like equation ( [ tf ] ) , no combinatorial interpretation is known for the polynomials @xmath425 or recurrence relations ( [ grec ] ) , in terms of non - overlapping partitions or any other combinatorial objects .
nevertheless , there exist explicit , although rather complicated , expressions for the coefficients of these polynomials .
one of them was obtained already in @xcite , it defines the coefficient of a given monomial as a certain multiple integral .
the structure of expressions from @xcite is more combinatorial .
atomic partitions were introduced in the study of the algebra of symmetric polynomials in noncommutative variables ( these set partitions enumerate a certain basis of generators for this algebra ) @xcite . like in the case of non - overlapping partitions ,
the definition makes use of the ordering of the underlying set .
equivalently , no non - empty partitions @xmath331 exist such that @xmath438 , where @xmath439 is the concatenation ( [ | ] ) .
yet another equivalent definition reads : the set partition is atomic if the supports of its blocks cover the whole interval of the partition , @xmath440_{\mathbb r}.\ ] ] let @xmath441 and @xmath442 denote the sets of all atomic partitions of @xmath4 $ ] and those which consist of @xmath6 blocks , @xmath443 at @xmath444 and @xmath445 at @xmath101 .
construction of atomic partitions employs the operations @xmath446 two first operations are similar to ( [ sp.djm ] ) , but the difference is that the new element is added at the @xmath28-th position , displacing the old element from this position to @xmath33-st position , cf .
[ fig : p - djm ] and [ fig : ap - djm ] ; moreover , operation @xmath35 is applied only for @xmath101 . in more details , we first replace the element @xmath28 by @xmath33 in the partition @xmath435 $ ] : @xmath447 next , operation @xmath46 adds the element @xmath28 to the block @xmath448 ( in particular , if @xmath18 contains @xmath28 then this operation just adds @xmath33 to this block ) ; operation @xmath35 adds @xmath28 as a new singleton : @xmath449 operation @xmath317 is a modification of concatenation , with displacing of the last elements of both partitions ( cf . with ( [ nop.p ] ) ) .
let @xmath450 , @xmath451 , @xmath452 , then @xmath453 in other words , we first concatenate the partitions with the maximal elements removed , then we place these elements back to their blocks , in the last but one position for the block from the first partition and in the last position for the block from the second one , see fig . [
fig : ap - p ] .
let us prove that our operations preserve the atomicity property .
notice , that in partition @xmath456 , the block @xmath18 containing @xmath28 is not a singleton if @xmath101 , since otherwise the partition is of the form @xmath457 .
therefore , the passage to the partition @xmath458 enlarges the support of this block : @xmath459_{\mathbb r}$ ] .
the further adding of the element @xmath28 does not reduce the blocks supports and it follows by ( [ ap.supp ] ) that partitions @xmath460
, @xmath461 are atomic as well .
the exceptional case is @xmath444 , for which @xmath462 , this is why the definition of this operation includes the restriction @xmath463 . in ( [ ap.p ] ) , analogously , the blocks of partitions @xmath464 and @xmath465 cover , respectively , the intervals @xmath466_{\mathbb r}$ ] and @xmath467_{\mathbb r}$ ] ( again , take into account that if @xmath452 then the blocks with maximal elements are not singletons ) , and their union covers @xmath468_{\mathbb r}$ ] .
now let us prove that any partition @xmath469 can be represented , in a unique way , as @xmath470 or @xmath471 or @xmath327 , where @xmath472 are atomic partitions .
let @xmath28 ( last to the end element of @xmath3 ) belongs to the block @xmath18 .
construct the partition @xmath473 $ ] by deleting this element and moving @xmath33 at the vacant position .
the following cases are possible .
\1 ) @xmath18 is a singleton .
then @xmath474 and @xmath470 ; moreover , @xmath3 can not be obtained in other way , because the operations @xmath46 , @xmath317 add the element @xmath28 to a block which already contains some elements .
\2 ) @xmath18 is not a singleton and @xmath474 . then @xmath471
; the uniqueness follows from the fact that if @xmath327 then @xmath475 and @xmath49 is uniquely determined as the index of a block containing the element @xmath28 .
\3 ) @xmath18 is not a singleton and @xmath475 .
then @xmath476 is uniquely represented as @xmath477 , where @xmath478 , @xmath479 .
the partition @xmath123 is characterized by the property that the union of the supports of @xmath480 covers the interval to the right of the rightmost gap in the union of the supports of @xmath476 .
moreover , the block @xmath481 belongs to the partition @xmath482 and appending of @xmath483 to this block gives us an atomic partition @xmath118 such that @xmath327 .
we will consider two statistics for the atomic partitions .
first , let us copy the definition of the bell polynomials @xmath52 ( [ yn ] ) : @xmath484 it turns out that polynomials @xmath485 and @xmath52 are closely related ( cf . with statement [ st : tfn ] ) .
pay attention that equation ( [ tyy ] ) involves the ordinary generating functions , in contrast to the exponential one in ( [ yn - egf ] ) .
the differential equations for this functions can be solved in quadratures , but these formulae are of little value for us , since @xmath486 and @xmath487 serve just as asymptotic expansions for solutions , like the series in section [ s : nop - stat ] .
let @xmath97 , @xmath62 .
it is easy to prove that the generating operations ( [ ap.djm ] ) , ( [ ap.p ] ) satisfy the property @xmath491 statement [ st : ap.djmp ] implies the recurrence relations @xmath492 which are equivalent to the riccati equation @xmath493 the change ( [ tyy ] ) brings to equation @xmath494 which is equivalent to the recurrence relations ( [ yn - rec ] ) for the bell polynomials .
let @xmath495 denote the number of atomic partitions @xmath4 $ ] with @xmath6 blocks ( * ? ?
* http://oeis.org/a087903[a087903 ] ) and @xmath496 be the total number of atomic partitions ( * ? ?
* http://oeis.org/a074664[a074664 ] ) , see table [ t : rn ] .
the substitution of @xmath497 and @xmath498 into ( [ tyy ] ) establishes the connection of these numbers with the stirling numbers of the 2nd kind and the bell numbers : @xmath499 for the first time , these relations were found in ( * ? ? ?
* eq.(1),(2 ) ) .
the so - called unsplittable partitions with a close statistics were studied in @xcite .
the total number of such partitions of size @xmath28 coincides with the number of atomic partitions , but another number triangle appears instead of @xmath502 .
a rather complicated bijection between these two classes of partitions was established in @xcite .
it is not clear at the moment , whether this correspondence is described by some transformation of generating functions or variables @xmath103 .
let us consider expressions @xmath332 , builded from the variable @xmath263 by use of operations @xmath333 , @xmath503 and @xmath334 , @xmath504 , where @xmath336 is equal to the common number of occurrences of @xmath263 and @xmath35 in @xmath147 .
let the set @xmath505 , @xmath463 , consists of all such expressions containing @xmath506 characters @xmath507 , and the set @xmath508 includes those with the degree equal to @xmath6 . according to statement [ st : ap.djmp ] ,
the map @xmath509 defines a bijection between @xmath508 and @xmath442 , for @xmath463 ( notice , that the partition @xmath454 which makes up @xmath510 , is actually not involved into ( [ pi^a ] ) ) .
in addition to the variable @xmath172 , we introduce the variable @xmath511 and define the monomial @xmath512 , as the value of expression @xmath341 computed according to the following rules , applied in the given order : summation over all expressions defines the polynomials @xmath514 see table [ t : phi^a ] , where the list of expressions @xmath505 is given for @xmath515 , along with the respective partitions and monomials , and table [ t : rn ] containing few first polynomials @xmath516 . @xmath517 3 & d_1u & 123 & u_1 \\ & mu & 13|2 & uv \\[3pt ] 4 & d_1d_1u & 1234 & u_2 \\ & d_1mu & 134|2 & u_1v \\ & d_2mu & 14|23 & uv_1 \\ & md_1u & 124|3 & u_1v \\ & mmu & 14|2|3 & uv^2 \\ & p(u , u ) & 13|24 & u^2\\ \hline \end{array}\ ] ] a comparison of tables [ t : phi * ] and [ t : phi^a ] demonstrates that non - overlapping partitions @xmath298 are in one - to - one correspondence with the atomic partitions @xmath518 constructed without the use of operation @xmath35 .
notice , that in partition @xmath519 the blocks of @xmath118 and @xmath123 are enumerated sequentially , and operation @xmath35 adds singleton as the last block . on the step ( i ) , the variable @xmath129 is introduced as the second argument , hence after this step the enumeration of variables @xmath520 from left to right is consistent with the enumeration of blocks in respective partition . therefore , after the step ( ii ) , a block with @xmath36 elements corresponds to variable @xmath521 or @xmath522 , depending on whether this block was generated by operation @xmath317 or @xmath35 .
this implies the relation between the monomials @xmath237 and @xmath523 , and therefore , between polynomials @xmath485 and @xmath516 : @xmath524 ( except for @xmath525 , because the polynomial @xmath526 is not defined ) .
equation ( [ rn ] ) follows from ( [ pi^a ] ) and identities @xmath530 the two last identities are obvious . in order to prove the first ,
let us write the variables @xmath364 , @xmath103 in the monomial in the same order as they stay after applying the operations @xmath35 and @xmath46 according to the rules ( i ) , ( ii ) .
then the difference between the monomials @xmath531 and @xmath512 is that the subscript of @xmath49-th factor is increased by 1 , and we only have to take the sum over @xmath49 .
linearization of the riccati equation ( [ dr ] ) brings to equation @xmath532 the associated hierarchy of nonlinear equations appears from the compatibility conditions of ( [ kb.psi ] ) with equation @xmath533 , where @xmath534 and @xmath535 are polynomials in @xmath81 .
straightforward computations brings to equations @xmath536 where the subscript @xmath537 denotes the deleting of negative powers of @xmath81 , and to equation for the generating function @xmath538 @xmath539 which uniquely determines all @xmath540 .
moreover , the compatibility conditions are equivalent to the system @xmath541 the choice of the right hand side in ( [ kb.g ] ) provides the homogeneity with respect to the weight @xmath426 , @xmath542 .
the basic equations of the hierarchy are of the form @xmath543 this is the kaup broer system @xcite , one of gauge equivalent forms of the nonlinear schrdinger equation ( the chain of substitution can be found in ( * ? ? ?
* eq . ( a ) ) ) .
let us write down also the 3-rd order symmetry : @xmath544 a comparison of the riccati equations ( [ dr ] ) and ( [ dty ] ) proves that all flows admit the reductions @xmath545 and @xmath546 , both leading to the burgers hierarchy ; a comparison with the riccati equation ( [ df ] ) proves that all odd flows admit the reduction @xmath547 to the kdv hierarchy .
the generating functions @xmath434 and @xmath548 from statement [ st : rn ] are related by the formula @xmath549 , where the series @xmath550 is another solution of the riccati equation ( [ dr ] ) . like in the case of equation ( [ kdv.g ] ) ,
an intermediate combinatorial interpretation of polynomials @xmath540 is unknown .
i. avramidi , r. schimming . a new explicit expression for the korteweg de vries hierarchy .
219:1 * _ ( 2000 ) http://dx.doi.org/10.1002/1522-2616(200011)219:1<45::aid-mana45>3.0.co;2-s[4564 ] n. bergeron , c. reutenauer , m. rosas , m. zabrocki . invariants and coinvariants of the symmetric group in noncommuting variables .
_ canad . j. math . *
60 * _ ( 2008 ) http://dx.doi.org/10.4153/cjm-2008-013-4[266296 ] .
n. bergeron , m. zabrocki .
the hopf algebras of symmetric functions and quasi - symmetric functions in non - commutative variables are free and co - free
. _ j. algebra appl . * 08 * _ ( 2009 ) http://dx.doi.org/10.1142/s0219498809003485[581600 ] .
chen , t.x.s .
li , d.g.l .
a bijection between atomic partitions and unsplitable partitions .
_ electron .
j. combin .
* 18 * _ ( 2011 ) http://www.emis.ams.org/journals/ejc/volume_18/pdf/v18i1p7.pdf[#p7 ] .
p. flajolet , r. schott .
non - overlapping partitions , continued fractions , bessel functions and a divergent series .
j. combin .
* 11:5 * _ ( 1990 ) http://dx.doi.org/10.1016/s0195-6698(13)80025-x[421432 ] .
gelfand , l.a .
asymptotic properties of the resolvent of sturm liouville equations , and the algebra of korteweg de vries equations .
_ russian math .
surveys * 30:5 * _ ( 1975 ) http://dx.doi.org/10.1070%2frm1975v030n05abeh001522[77113 ] .
f. lambert , i. loris , j. springael , r. willox . on a direct bilinearization method : kaup s higher - order water wave equation as a modified nonlocal boussinesq equation .
_ j. phys .
a : math . gen . * 27 * _ ( 1994 ) http://dx.doi.org/10.1088/0305-4470/27/15/028[53255334 ] .
mikhailov , a.b .
shabat , r.i .
extension of the module of invertible transformations .
classification of integrable systems .
115:1 * _ ( 1988 ) http://dx.doi.org/10.1007/bf01238850[119 ] .
the on - line encyclopedia of integer sequences , published electronically at http://oeis.org , 2010 : http://oeis.org/a000108[a000108 ] , http://oeis.org/a126120[a126120 ] catalan n. ; http://oeis.org/a000110[a000110 ] bell n. ; http://oeis.org/a005493[a005493 ] 2-bell n. ; http://oeis.org/a006789[a006789 ] bessel n. ; http://oeis.org/a007405[a007405 ] dowling n. ; http://oeis.org/a008277[a008277 ] , http://oeis.org/a048993[a048993 ] stirling-2 n. ; http://oeis.org/a008292[a008292 ] euler n. ; http://oeis.org/a039755[a039755 ] @xmath0-analogs of stirling-2 n. ; http://oeis.org/a074664[a074664 ] n. of atomic partitions ; http://oeis.org/a087903[a087903 ] analogs of stirling-2 n. for atomic partitions ; http://oeis.org/a099950[a099950 ] n. of 2-sparse nonoverlapping partitions ; http://oeis.org/a145271[a145271 ] expansion of @xmath551^ng(x)$ ] . | we demonstrate that statistics for several types of set partitions are described by generating functions which appear in the theory of integrable equations . _
key words : _ set partition , @xmath0-type partition , non - overlapping partition , atomic partition , bell polynomial , dowling number , bessel number , generating function , integrable hierarchy _ msc : _ 05a18 , 37k10 02.10.ox , 02.30.ik |
the study of the binder cumulants for isolated systems of various sizes has been an important tool to determine the critical temperature and universal quantities of spin models @xcite and lattice gauge theory models at finite temperature @xcite .
the basic idea is that the intersections among binder cumulants curves for systems of different sizes obey scaling relations often referred to as finite size scaling ( fss ) .
linearization in the scaling variables is often used to derive approximate fss relations for the binder cumulants and the critical temperature that allow , in principle , simple linear extrapolations to infinite volume @xcite . however , in practice , the errors bars on the intersections are often of the same order as the finite size variations and the linearity of the data is not always obvious .
examples of such graphs are figure 7 in @xcite or figures 3 , 4 and 6 of @xcite .
for this reason , authors will sometimes prefer to rely on a single intersection for the two largest volumes available that has potentially smaller errors @xcite .
a possible source of systematic error in these infinite volume extrapolations are the nonlinear terms that have been neglected in the derivation of the linear fss formulas . if these effects are small enough , it is difficult to disentangle them from the statistical errors .
typically , we are talking about a 1 - 2 percent effect .
this may sound small , however , for the @xmath0 hierarchical sigma models where very accurate calculations can be performed , the differences between the binder cumulants at successive values of @xmath1 are less than 2 percent for low values of @xmath1 @xcite . consequently , it is reasonable to set accuracy goals of less than 1 percent for the binder cumulants , especially if the estimate is used to establish that a particular model belongs to a known class of universality . in this article , we discuss quantitatively the effects of the nonlinear terms in fss with an example and we propose a new procedure to obtain infinite volume estimates that are minimally distorted by the nonlinear effects .
the procedure can be used for spin models or lattice gauge models and @xmath2 is a notation that can be used interchangeably for either the inverse temperature in ising - like models or @xmath3 in gauge theory .
one purpose of the article is to provide a specific improvement to the common practice of using data at the same , equally spaced , values of @xmath2 for all value of the linear size @xmath1 ( we call this the fixed interval procedure " ) .
clear examples of the fixed interval procedure can be seen in figure 6 of @xcite , figure 11 of @xcite , figure 7 of @xcite , figure 1 of @xcite and figure 1 of @xcite where the fourth order cumulant is plotted versus @xmath2 or the temperature . in the large volume limit , the cumulant becomes a function ( that we will call @xmath4 ) of the reduced variable @xmath5 , with @xmath6 the usual correlation length exponent .
this function @xmath4 can be approached at sufficiently large volume by plotting the cumulant versus this reduced variable .
when this is done , the data `` collapses '' up to some finite size corrections .
it seems clear that in order to reduce the nonlinear effects , we should shrink the range of @xmath2 as @xmath1 increases .
this idea seems simple , however its practical implementation is non - trivial .
first , we need to estimate the size of the nonlinear effects .
second , we only know @xmath7 with a finite accuracy and the interval can not be shrunk to a size much smaller than the uncertainties on @xmath7 .
we need to find a compromise between two conflicting requirements . on one hand
, the interval needs to be small enough to reduce the nonlinear effect . on the other hand ,
if the interval is too small , we may not cover properly the region where the linear approximation is valid which could result in errors for the intersections .
the main technical question that we address here is : can we find an optimal width for the interval were both requirements can be reasonably satisfied ?
it is difficult to answer this question for the gauge or spin models discussed above because it is hard to extract the nonlinear effects from the statistical errors .
in addition , we have no independent estimates of the infinite volume cumulants and calculations at large volume are cpu demanding
. we will address this question in a model where the practical issues mentioned above can be avoided , namely dyson hierarchical model @xcite with an ising measure . in this model , it possible to block spin very accurately with methods @xcite that have been recently reviewed in @xcite .
this allows us to make very accurate calculations at very large volume ( linear size up to @xmath8 sites ) that can not be reached with ordinary monte carlo methods .
the free parameter of this model has been fixed in such a way that free gaussian fields scale like in 3 dimensions .
the technical details are given in [ sec : appa ] .
these large volume numerical calculations will be used later to check the accuracy of estimates made using data at much smaller volumes ( linear size up to 256 sites as in existing monte carlo ( mc ) calculations ) .
our presentation is focused on the estimation of the infinite volume critical temperature and the ( universal ) fourth order binder cumulant .
this quantity is defined precisely in section [ sec : notations ] where we introduce other notations and state the problem .
we provide a parametrization of the nonlinear effects for the fourth order cumulant in terms of two new parameters ( see equation ( [ eq : param ] ) ) .
accurate numerical values of these parameters are given in [ sec : appa ] in the case of the hierarchical model .
we then discuss the linearization and graphs testing its validity .
the new method that we propose here ( the shrinking interval procedure " ) is presented in section [ sec : shrink ] .
it proceeds in two steps .
first , we need to estimate the size of the nonlinear effects using data at relatively small volume .
we explain that the important quantity is the ratio of the linear to quadratic leading amplitudes ( called @xmath9 ) .
we show in [ sec : pencil ] that this quantity can be estimated with reasonable accuracy using a pencil and a ruler on a graph showing the binder cumulant versus @xmath2 at relatively small volume .
we then propose to use a shrinking range @xmath10 where @xmath11 is an adjustable small parameter which we shall see , can picked in an optimal way by localizing crossings in infinite volume extrapolations . in section [ sec :
compare ] we compare the shrinking interval procedure with the fixed interval procedure .
the calculations are done using a number of values of @xmath2 and linear sizes that are typical in existing mc calculations .
we emphasize that in doing these calculations , we did not use our prior knowledge of the critical temperature , binder cumulant or amplitudes given [ sec : appa ] , however we used our knowledge of the critical exponents @xmath6 and @xmath12 .
we first used a fixed value of @xmath11 ( 0.05 ) and found much better result than with the fixed interval method .
we then compared infinite volume extrapolations for @xmath7 and the cumulant , for different values of @xmath11 .
the results are displayed in figure [ fig : extraps ] where we see remarkable crossings ( among extrapolation curves for different maximal volume ) at approximately the same value of @xmath11 for the two quantities .
we compared the results for this optimal value with the accurate ones and found relative errors of less than @xmath13 for @xmath7 and of less than @xmath14 for the cumulant .
figure [ fig : extraps ] is the most important result of the paper .
it shows that an optimal value of @xmath11 can selected using the numerical results .
it is crucial to realize that the crossing means that for the optimal value of @xmath11 , results at not too large volume provide very accurate results .
one should also appreciate the number of large volume calculations involved in producing the two graphs that would take a prohibitively long time for 4 dimensional lattice gauge theory . in section [ sec : app ] , we discuss the effects of the nonlinear terms for the calculations done in the literature and we conclude with possible applications of the new method .
the new method proposed in this article can be used for two types of models .
first , the spin models where @xmath2 denotes the inverse temperature in appropriate units . in the literature on the 3 dimensional ising model @xcite @xmath2
is often denoted @xmath15 .
second , the lattice gauge models with gauge group @xmath16 at finite temperature @xmath17 where @xmath18 is the number of sites in the euclidean time direction and @xmath19 the lattice spacing . for these models @xmath20 is not the inverse temperature .
however , if we use the one - loop scaling @xcite , @xmath21 this formula illustrates that for the finite temperature gauge model , @xmath22 implies @xmath23 because the ordered phase of these models corresponds to high @xmath24 and high @xmath2 . in summary
, the notation @xmath2 can be used for both types of models with a consistent meaning : the ordered phase corresponds to @xmath25 .
for this reason , we will define a reduced @xmath2 ( or @xmath15 ) variable : @xmath26 other choices such as @xmath27 lead to similar linear behavior near @xmath7 but have different nonlinear behavior .
these considerations are important if we want to connect with other expansions @xcite .
we now consider finite size scaling for isolated blocks in 3 dimensions .
the spin models are defined on symmetric cubic lattice @xmath28 sites .
the gauge models are defined @xmath29 lattices .
we use the convention @xmath30 for the variable number of sites ( while @xmath18 is kept fixed ) in order to have unified notations . in both cases ,
the system is not in contact with a larger system but isolated and defined with suitable boundary conditions . under a rg transformation where the lattice spacing @xmath31 , we have @xmath32 and @xmath33 , where the @xmath34 are the nonlinear scaling variables which transform multiplicatively .
we denote @xmath35 the only relevant scaling variable ( we will not deal with external fields ) and we will only consider the effect of the first irrelevant variable denoted @xmath36 .
we assume that the exponents @xmath37 , @xmath38 are known with good precision as it is the case when we are testing the hypothesis that a particular system belongs to a well studied class of universality .
in addition , we assume the expansions @xmath39 we define the fourth binder cumulant as @xmath40 with @xmath41 the average of the order variable ( spin for ising , polyakov s loop in the time direction for gauge theory ) in 3 dimensions .
directly related quantities appear in the literature such as @xmath42 or @xmath43 etc . due to the lack of consensus , we have picked a form that has limits that are easy to remember .
both numerator and denominator are unsubtracted averages and so this quantity can be calculated at large volume without running into loss of accuracy problems occurring for subtracted averages . in the symmetric phase , @xmath44 is dominated by @xmath45 and the the correction is suppressed by one power of the volume ( the coefficient is a renormalized coupling constant ) .
consequently , if @xmath46 ( @xmath47 ) , @xmath48 tends to 3 in the infinite volume limit . on the other hand in the ordered phase ,
@xmath44 is dominated by @xmath49 and the limit is 1 . at fixed @xmath1 , for sufficiently small @xmath2 , @xmath48 is close to 3 and for sufficiently large @xmath2 , @xmath48 is close to 1 . as @xmath1 increases
, the transition sharpens as illustrated in figure [ fig : b4 ] . versus @xmath2 ( left ) and versus @xmath50 ( right ) , for @xmath51 4 , 8 , 16 , 32 , 64 and 128 , for the ising hierarchical model.,title="fig:",width=230 ]
versus @xmath2 ( left ) and versus @xmath50 ( right ) , for @xmath51 4 , 8 , 16 , 32 , 64 and 128 , for the ising hierarchical model.,title="fig:",width=230 ] finite size scaling @xcite implies that @xmath52 since @xmath53 , when the volume is large enough , @xmath48 can be approximated by a function of @xmath54 only . if we plot @xmath48 versus @xmath54 , the curves at various @xmath1 approximately collapse .
this is illustrated in figure [ fig : b4 ] where one can see that the larger violations of this approximation are observed at smaller volume .
the most common procedure to study the intersections of the various curves at fixed @xmath1 is to linearize equation ( [ eq : exactfss ] ) and try to extrapolate the results at infinite @xmath1 . in the rest of this article
, we will consider the effects of nonlinear terms on this procedure .
our main assumption will be that @xmath55 note that we have not included terms of the form @xmath56 which should appear as a consequence of the presence of @xmath57 terms in the expansion of @xmath35 because they are suppressed by a factor @xmath58 compared to the term @xmath59 .
we have also not included terms of order @xmath60 and @xmath61 . in the linear approximation ( @xmath62 )
, we recover the standard linear fss formula for the point of intersection denoted @xmath63 between the two curves @xmath64 and @xmath65 , namely @xmath66 with @xmath67 these formulas have been written in a way that makes the symmetry under the interchange of @xmath1 and @xmath68 obvious
. they may look more familiar if we replace @xmath68 by @xmath69 and factor out the powers of @xmath1 . since we assume that @xmath6 and @xmath12 are known in good approximation , linear fits can used to determine the remaining unknown parameters in equation ( [ eq : linearfss ] ) .
graphs based on these formulas can be found in @xcite ( for @xmath70 = 2 ) , @xcite and in figures [ fig : fixedbet ] and [ fig : fixedbin ] below . for reasons explained in the introduction
, we will make model calculations for dyson s hierarchical model .
for this model , the local potential approximation is exact and the change in the local measure under block spinning can be calculated very accurately at very large volume and very close to @xmath7 @xcite ( see @xcite for a recent review ) .
the free parameter of this model , usually denoted @xmath71 has been fixed in such a way that free gaussian fields scale like in 3 dimensions . with this choice , as the blockspinning reduces the number of sites by a factor 2 , the linear sizes are of the form @xmath72 .
the details of the calculations and accurate estimates of the parameters entering in equation ( [ eq : param ] ) are given in [ sec : appa ] .
in this section , we present the shrinking interval procedure advocated in the introduction
. it may be useful to state all the steps that need to be followed .
first , we perform linear fits of @xmath48 as a function of @xmath2 at fixed volume in a sufficiently small @xmath2 interval .
we then use the linear fits at different volumes to determine the intersections called @xmath73 and @xmath74 in equation ( [ eq : linearfss ] ) .
finally , we select a set of pairs @xmath75 and perform linear fits to determine the unknown coefficients in the linear fss formula equation ( [ eq : linearfss ] ) .
the final result is an infinite volume extrapolation for @xmath7 and @xmath48 .
we now discuss the first step .
we need to specify the @xmath2 interval ( in other words , its center and width ) for calculations of @xmath48 at a given volume . given a linear size @xmath1 , we propose to use our best estimate @xmath76 of @xmath7 obtained from smaller sizes .
this is an iterative procedure .
we need to start with some reasonable estimate ( for instance obtained with the finite interval procedure at small size ) and then keep improving .
we propose to restrict the calculation of @xmath48 to the interval parametrized in the following way : @xmath77 we do not need a very accurate value for @xmath9 . we show in [ sec : pencil ] that it is easy to estimate the order of magnitude of @xmath9 from a graph such as figure [ fig : b4 ] , using a pencil and a ruler .
the value of @xmath11 needs to be chosen carefully . on one hand
, we need @xmath11 small enough in order to control the nonlinear effects . on the other hand ,
if @xmath11 is very small , we need a correspondingly good estimate of @xmath7 . in addition ,
when @xmath11 is too small , the intersections may be far away from the regions where we have values of @xmath48 .
these two effects could in principle compensate .
unfortunately , in the example where we have done accurate calculations , they go in opposite directions .
namely , the values of @xmath76 obtained from small volume data are below the true @xmath7 while the small volume intersections are above @xmath7 as can easily be seen from figure [ fig : b4 ] .
we want to figure out if the two conflicting requirements can be partially satisfied for some optimally chosen value of @xmath11 .
the errors on @xmath73 and @xmath74 depend not only on @xmath11 but also on @xmath76 .
the choice of @xmath11 is clearly a difficult optimization problem and it is useful to make numerical experiments . in the following section
we show that values of @xmath78 lead to results that compare very well with the results of [ sec : appa ] and that a reasonable compromise between the two requirements discussed above is possible .
we should also mention that the use of linear fits to determine the intersections could be replaced by more sophisticated search methods .
however , in calculations that are cpu demanding , and where we have only data for a few values of @xmath2 , linear fits is the most common practice .
in this section , we compare fss results using the shrinking interval procedure discussed above and the fixed interval procedure where we use fixed values of @xmath2 for all values of @xmath1 .
the calculations were made for the ising hierarchical model .
for the fixed interval calculation , we took the same 9 values of @xmath2 : 1.176 , 1.177 , ... , 1.184 for all possible @xmath1 between 8 and 256 , or in other words , @xmath79 with @xmath80 the number of blockspinnings . for the shrinking interval procedure
, we started with the fixed interval data with @xmath81 and an estimate of @xmath82 corresponding to this data .
we then followed the procedure described in section [ sec : shrink ] with @xmath83 .
this value of @xmath11 was selected from the compromise of having linear fits that were not blatantly off the data and at the same time keep most of the intersections within the fitted range .
we also picked 9 values of @xmath2 within the specified range . at each @xmath1 , we used all the possible intersections involving the 5 closest sizes to determine @xmath76 using equation ( [ eq : linearfss ] ) .
this means that we made linear fits with 15 points and the results are quite robust under small changes in the procedure .
linear fits were performed for each @xmath1 in order to express @xmath48 as a linear function of @xmath2 . for each pair
@xmath75 , it is possible to determine the intersection @xmath63 of the corresponding lines .
these empirical values are plotted versus the calculable values @xmath84 and @xmath85 , defined in ( [ eq : lm ] ) , in figures [ fig : fixedbet ] and [ fig : fixedbin ] . in order to make the graphs readable and the dependence on @xmath1 and @xmath68 clear ,
we have displayed 4 sets of 6 intersections denoted @xmath75 in the graphs .
the meaning of @xmath75 is that we take the intersection between @xmath68 and all the possible lower values starting at @xmath1 .
for instance , @xmath86 is a short notation for the 6 intersections : 8 and 32 , @xmath87 and 32 , .... , and , @xmath88 and 32 .
obtained with the fixed interval procedure ( left ) and with the shrinking interval procedure ( right ) versus @xmath84 for 4 sets of 6 pairs of values defined in the text .
the solid line is the linear fit for the set @xmath86 ( left ) and the set @xmath89 ( right ) .
the dash line is the behavior expected from equation ( [ eq : linearfss ] ) and the accurate values of [ sec : appa].,title="fig:",width=230 ] obtained with the fixed interval procedure ( left ) and with the shrinking interval procedure ( right ) versus @xmath84 for 4 sets of 6 pairs of values defined in the text .
the solid line is the linear fit for the set @xmath86 ( left ) and the set @xmath89 ( right ) .
the dash line is the behavior expected from equation ( [ eq : linearfss ] ) and the accurate values of [ sec : appa].,title="fig:",width=230 ] versus @xmath85 obtained with the fixed interval procedure ( left ) and with the shrinking interval procedure ( right ) for 4 sets of 6 pairs of values defined in the text .
the solid line is the linear fit for the set @xmath86 ( left ) and the set @xmath89 ( right ) .
the dash line is the behavior expected from equation ( [ eq : linearfss ] ) and the accurate values of section [ sec : appa].,title="fig:",width=230 ] versus @xmath85 obtained with the fixed interval procedure ( left ) and with the shrinking interval procedure ( right ) for 4 sets of 6 pairs of values defined in the text .
the solid line is the linear fit for the set @xmath86 ( left ) and the set @xmath89 ( right ) .
the dash line is the behavior expected from equation ( [ eq : linearfss ] ) and the accurate values of section [ sec : appa].,title="fig:",width=230 ] for the fixed interval procedure , one can see that values corresponding to the smallest volumes , namely the @xmath86 set , the behavior is approximately linear .
the extrapolations at infinite volume for the fit with the six points of this set are 1.17848 for @xmath7 and 2.53435 for @xmath48 , which is not too far from the accurate values but not very accurate either . however ,
if we increase the volume the linearity and the accuracy of the extrapolations degrade rapidly .
on the other hand , with the shrinking interval method , the accuracy seems to improves with the size . in both graphs ,
the fit made with the @xmath89 set is hardly distinguishable from the accurate result .
in order to give a more detailed idea of the effects on the extrapolated values , the numerical values of the parameters entering in equation ( [ eq : linearfss ] ) are shown in table [ tab : shrin ] for various sets of data .
@xmath90&@xmath48&@xmath9115&1.1788071 & 3.74 & 2.5245 & - 0.76116 & 1.1789292 & 3.54 & 2.51486 & - 0.71517 & 1.1789933 & 3.40 & 2.50724 & - 0.67318 & 1.1790389 & 3.28 & 2.49907 & - 0.62119 & 1.1790832 & 2.99&2.48794 & - 0.53420 & 1.1790042 & 3.51 & 2.50940 & - 0.69721 & 1.1790198 & 3.35 & 2.50336 & - 0.64722 & 1.1790267 & 3.20 & 2.49945 & - 0.60623 & 1.1790301 & 3.04 & 2.49636 & - 0.56424 & 1.1790316 & 2.85 & 2.49387 & - 0.518expected&1.1790302&2.91&2.49642&-0.529 we have also performed fits of the intersections for each value of @xmath1 using all the possible intersections involving the 5 values of the linear size immediately below .
this amounts to 15 intersections , just as for the determination of @xmath76 .
this procedure maximizes the number of data points for a given size interval and should be suitable when larger numerical errors are present .
we have repeated the calculation with smaller and larger values of @xmath11 .
the results for the @xmath7 and @xmath48 are shown in figure [ fig : extraps ] .
one sees that the results become erratic for @xmath92 .
there is a line crossing that appears in both graphs for @xmath93 .
we believe that this crossing reflects the compromise discussed above , even though it occurs at a value of @xmath11 about two times larger than initially guessed .
the estimates at the crossing are @xmath94 and @xmath95 . in general , we do not know precisely the crossing value and so larger error bars should be set by using the variation with @xmath11 .
from the figures , this should be at most @xmath96 for @xmath7 and 0.005 for @xmath48 for @xmath97 .
the accurate values are clearly within the errors bar for the estimates obtained with the 6 and 15 points fits .
it is also instructive to compare the 15 points fits in figure [ fig : extraps ] at @xmath98 with the values of the 6 point fits given in table [ tab : shrin ] for the same values of @xmath80 .
this gives an idea of the variability of the extrapolations as we change the data used to make the linear fits of equation ( [ eq : linearfss ] ) . and
@xmath48 based on 15 point linear fits from the intersections among the @xmath48 curves at @xmath72 and the 5 values of @xmath1 immediately below , for @xmath80 between 19 and 24.,title="fig:",width=326 ] and @xmath48 based on 15 point linear fits from the intersections among the @xmath48 curves at @xmath72 and the 5 values of @xmath1 immediately below , for @xmath80 between 19 and 24.,title="fig:",width=326 ] it is important to realize that near the value of @xmath11 where the various curves of figure [ fig : extraps ] cross , the extrapolations are approximately the same for all the values of @xmath80 .
this means that picking @xmath11 properly allows to get optimally accurate estimates with volumes not too large .
one should also appreciate that there are 35 values of @xmath11 in each graph of figure [ fig : extraps ] and that for each values of @xmath11 , we need more than 100 calculations of @xmath48 to determine the intersections . doing the same number of calculations using the mc method for spin models or lattice gauge theory would take a very long time .
in this section , we comment on existing numerical results found in the literature .
we first discuss existing estimates of @xmath48 for the ising universality class .
the methods used can be divided into those that rely on intersections @xcite , and those that attempt to work as closely as possible to the nontrivial fixed point @xcite .
note that @xcite takes into account the nonlinear effect of @xmath99 and also the effect of the second irrelevant direction .
the results are summarized in table [ tab : ising ] .
the error bars in @xcite are based on the visual estimate @xmath100 from figure 7 .
the agreement between the two results based on improved action that reduce the effect of the irrelevant variables @xcite as well as the agreement with @xcite indicates that 1.603 should be accurate with about one part in a 1000 .
the discrepancy with @xcite seems to be of statistical origin : if we use the fit given in equation ( 14 ) of @xcite together with @xmath97 and @xmath101 we obtain an absolute error less than @xmath102 for @xmath48 if we neglect @xmath99 . intersectionsising 3 & 1.590 ( 15 ? ) & @xcite & mc + reweighting & extrapolationising 3 & 1.591 ( 9 ) & @xcite & mcrg + reweighting&(128,256)ising 3 & 1.604(1 ) & @xcite & mc at best @xmath7 & -ising 3 & 1.603(1 ) & @xcite & improved action & -ising 3 & 1.603(1 ) & @xcite & improved action & -@xmath103 @xmath104 & 1.620(30 ) & @xcite & mc at various @xmath2 & extrapolation @xmath103 @xmath105 & 1.570(80 ) & `` & & ' ' @xmath103 @xmath106 & 1.520(100 ) & `` & & ' ' @xmath103 @xmath107 & 1.622(3 ) & @xcite & mc at various @xmath2 & ( 24,32)@xmath103 @xmath104 & 1.602(27 ) & `` & & ' ' @xmath103 @xmath106 & 1.584(39 ) & `` & & ''
we now discuss the nonlinear effects for specific calculations made with the fixed interval procedure .
for the 3 dimensional ising model , @xmath7 ( usually denoted @xmath108 ) is known with great accuracy and consequently , most of the graphs showing @xmath48 use a very narrow @xmath109 interval where the differences between linear fits and numerical data are difficult to see .
for instance , if we use the @xmath110 data in figure 6 of @xcite , it is possible to draw a line that stays within the error bars of all the data points .
consequently , we can only set a lower bound on @xmath9 from this figure by assuming that @xmath111 , defined in [ sec : pencil ] , is less than the error bars that we estimated to be less than 0.02 .
this bound is consistent with the numerical estimate of @xcite .
a similar situation is encountered with figure 2 of @xcite for @xmath103 lgt with @xmath112 .
we focus our analysis on the @xmath113 data points .
if we draw a line from the point with the lowest horizontal coordinate to the point with the largest one , the three points inside appear to be above the line . however , the line is within the three error bars . using the data points kindly provided by the author
, we obtained @xmath114 which is slightly less than the errors bars .
the lower bound quoted in the table is based on @xmath115 . on the other hand
, a clearly non - zero value for @xmath111 can be seen in figure 11 of @xcite and figure 9 of @xcite and relatively stable values can be obtained for @xmath9 .
if we now denote @xmath116 the maximum value of @xmath117 used in the linear fits made to determine the intersections ( assuming that such a fits were performed , otherwise we rely on the width of the figure ) in each reference , we can estimate the relative size of the nonlinear effects by calculating @xmath118 the numerical values are shown in table [ tab : emax ] .
there are no large values of @xmath119 .
this is because when @xmath120 , the failure of linear fits is obvious .
on the other hand , systematic errors due to unaccounted nonlinear effects may be reduced by using a smaller @xmath119 with the procedure discussed in section [ sec : shrink ] .
@xmath121 & @xmath122 & @xcite&@xcite&@xcite&@xcite data used & fig .
6 , @xmath110&fit on fig .
11&fig . 9 , @xmath123 & fig .
2 , @xmath113 @xmath9&@xmath124&3&2&@xmath125 @xmath116 & 5.5 & 0.4 & 0.47 & 0.4 @xmath119 & @xmath126&0.13&0.24 & @xmath127
we have proposed a new method designed to reduce possible nonlinear effects in the estimates of @xmath7 and @xmath48 . for the ising hierarchical model with a volume corresponding to a linear size of 256 in 3 dimensions
, the method gives results which agree with independent accurate estimates with a relative accuracy better than one part in 100,000 for @xmath7 and 5 parts in 1000 for @xmath48 .
the intrinsic nonlinear effects ( measured in terms of @xmath128 ) in this model are roughly of the same size as those found the other models that we discussed in section [ sec : app ] . figures [ fig : fixedbet ] and [ fig : fixedbin ] shows the potentially disastrous effect of ignoring nonlinear effects . in graphs with large numerical errors
, this effect may be overlooked .
the small values of @xmath129 found in the literature indicate that nothing drastic should appear in the cases considered .
however , it would be interesting to repeat and extend these calculations for larger volumes using the method proposed here .
the optimal value of @xmath11 can in principle be obtained by looking at crossings of extrapolated values .
this suggests that the procedure should be followed for a few different values of @xmath11 .
we are planning to consider the effects of statistical errors on the new method by applying it to the 3 dimensional ising model and finite temperature gluodynamics .
we thank a. velytsky for valuable discussions and comments and for providing his numerical data .
this research was supported in part by the department of energy under contract no .
fg02 - 91er40664 .
for completeness , we explain how we calculated @xmath48 for the hierarchical model and we give the details of the numerical calculations of the parameters .
the interest of the hierarchical model is that we can blockspin exactly .
the basic formula for the fourier transform of the local measure is @xmath130 with @xmath131 note that we do not rescale the field as in a renormalization group transformation .
@xmath132 is a constant that we adjust so that @xmath133 .
when it is the case , we have @xmath134 with @xmath135 the sum of all the spins .
if we write @xmath136 then @xmath137 the critical values are obtained by plugging the values of @xmath138 and @xmath139 from the nontrivial fixed point .
see @xcite for details and references .
the value of @xmath140 is obtained by increasing @xmath1 but keeping @xmath50 constant .
the asymptotic values for different @xmath50 can be fitted very well with a line which gives @xmath91 and @xmath141 .
we then subtracted the effect of @xmath140 and obtained @xmath142 and @xmath99 using discrete derivatives near @xmath143 .
we used calculations at very large volume ( @xmath144 ) . in the text
, we show that these accurate results can be reproduced with a new procedure using data at much smaller volume . from refs .
@xcite , @xmath145 using these values , and numerical results a fixed values of @xmath50 and large @xmath1 , we first determine the @xmath146 terms and find @xmath147 subtracting these effect and taking discrete approximation of the derivative with respect to @xmath109 near 0 , we find @xmath148 one should appreciate that we have been able to determine numerically the 8 parameters of equation ( [ eq : param ] ) with good numerical stability .
in this appendix , we explain how to estimate the order of magnitude of @xmath9 from a graph illustrating approximate data collapse such as figure [ fig : b4 ] , using a pencil and a ruler .
a subset of the data of figure [ fig : b4 ] is shown in figure [ fig : penrule ] together with lines that can be drawn on the original graph .
the line joins two points with approximately opposite @xmath149 coordinates .
we denote these two points @xmath150 and @xmath151 .
we call @xmath111 the difference between the line and the data ( assumed to be a quadratic function ) at 0 .
it is then easy to show that @xmath152 in figure [ fig : penrule ] , we have approximately @xmath153 , @xmath154 , @xmath155 and @xmath156 , which gives the estimates @xmath157 and @xmath158 which are in reasonably good agreement with the previous estimates .
note that we are mostly interested in the ratio @xmath159 and that this ratio is invariant under a multiplicative rescaling .
this is useful if we are considering graphs with other quantities displayed ( for instance @xmath160 ) . for the model considered here , @xmath161 .
the above estimates are based on the data with @xmath162 . in figure
[ fig : penrule ] , we also see part of the data for @xmath163 and we see that it is not much above the @xmath162 data . from equation ( [ eq : param ] ) , it is clear that if we neglect @xmath141 , the entire curve is translated uniformly in the vertical direction .
but the estimates of @xmath142 and @xmath99 are based on differences and do not depend on this translation provided that we only use data for one value of @xmath1 .
this is confirmed by using the @xmath110 data as shown in figure [ fig : penrule ] where we obtain an estimate of @xmath9 very close to the one quoted above for @xmath162 .
this shows that up to a subtraction and rescaling of the horizontal coordinate , we can rely on a graph of @xmath48 versus @xmath2 at relatively small volume .
it is clear that the subtraction of @xmath7 requires to have an estimate of this quantity .
however , this estimate does not need to be very precise to get the order of magnitude of @xmath9 . versus @xmath50 , for @xmath51 16 ( below ) , 32 and 64 ( above ) for the ising hierarchical model.,title="fig:",width=230 ] versus @xmath50 , for @xmath51 16 ( below ) , 32 and 64 ( above ) for the ising hierarchical model.,title="fig:",width=230 ] hasenbusch m , pinn k and vinti s 1998 _ physical review b _ * 59 * 11471 campostrini m , pelissetto a , rossi p and vicari e 1999 _ phys . rev . _ * e60 * 35263563 binder k and luijten e 2001 _ phys .
rept . _ * 344 * 179253 | we point out that ignoring nonlinear effects in finite size scaling may lead to errors in estimates of the critical temperature and binder cumulants .
we show that the order of magnitude of these effects can be estimated from data at relatively small volume . using this estimate
, we propose to use linear fits in increasingly small temperature regions as the volume is increased ( rather than using a fixed temperature interval ) .
the choice of the exact coefficient of proportionality can be optimized and reveals interesting crossing patterns among estimates .
we show that the new procedure works very well for dyson hierarchical model .
we discuss applications of the method for 3 dimensional spin models and finite temperature lattice gauge theories and comment on the nonlinear effects for existing calculations . |
in the magellanic clouds ( mcs ) giant loops of h@xmath0ii regions can be recognized in deep pictures taken in h@xmath14 light .
these h@xmath0ii structures were divided by goudis & meaburn ( 1978 ) into two distinct groups : shell structures called giant shells ( gss ) with diameters of 20@xmath15pc and the huge supergiant shells ( sgss ) with diameters of 600@xmath16pc . employing unsharp masking techniques and high contrast copying of the long exposed h@xmath14 images of davies et al .
( 1976 , hereafter dem ) , meaburn and collaborators ( see meaburn 1980 ) subsequently identified 85 gss and 9 sgss in the large magellanic cloud ( lmc ) and 1 sgs in the small magellanic cloud ( smc ) . unlike gss
whose stucture can be explained by the combined activity of supernova explosions , stellar winds and radiation pressure of the central star association(s ) , sgss ( i.e. structures about @xmath17kpc in diameter ) need very effective , gigantic mechanisms for their creation .
such large scale features , which rival in size only with spiral structures , might be created , according to the appraisal of mechanisms by tenorio - tagle & bodenheimer ( 1988 ) , by the collision of high velocity clouds ( hvcs ) with the disk of the galaxy or by stochastic self - propagating star formation ( sspsf ) .
the infall of an hvc would cause a well defined velocity deviating from the velocity of the original disk gas .
also , because of the short time scale of such a collision , an almost identical age is expected for all stars whose formation was triggered by that event . in the case of sspsf ( see feitzinger et al .
1981 ) star formation propagates from one point ( e.g. the centre of a sgs ) in all directions ( e.g. towards the rim of a sgs ) .
thus one should see in projection two velocity components in the h@xmath0i layer ( a receeding and an approaching component ) as well as a clear age gradient of the star populations inside a sgs
. however , the observation of a central area of a sgs with little gas ( i.e. a hole in the h@xmath0i layer ) and of an approximately @xmath18pc thick shell of neutral material , which is ionized at the inner edge ( visible as h@xmath14 filaments ) by the radiation from the associations containing numerous early type stars ( see lucke & hodge 1970 ; lucke 1974 ) in the centre of the sgs , would be consistent with both scenarios .
the importance of understanding the formation and the structure of sgss is evident if one realizes how big they are .
they may dominate large portions of a galaxy , in particular of the smaller irregular galaxies , such as the magellanic clouds are .
knowledge about the creation of sgss leads to a better knowledge of the more recent development of the mcs and of their youngest star populations .
the shape and size of lmc@xmath04 is such , that sspsf has been considered as the most likely explanation for the formation of this sgs ( see dopita et al .
however , observations of star groups along the edge of lmc@xmath04 showed that the ages of these groups were of the same order as the time needed to create the sgs in the case of sspsf ( see vallenari et al .
1993 ; petr et al .
1994 ) .
reid et al . ( 1987 ) found from @xmath2 , @xmath19 photometry of shapley constellation iii ( the southern half of the sgs lmc@xmath04 , mckibben nail & shapley 1953 ) no clear age gradient , also inconsistent with a global sspsf model .
furthermore , domgrgen et al . (
1995 ) presented an investigation of lmc@xmath04 based on h@xmath0i data and iue spectra .
one result is that there is only one distinct velocity component towards us with @xmath20 and just a diffuse rear component .
this is not consistent with an undisturbed expanding shell , but it indicates a break - out of the sgs at the back side of the lmc .
so lmc@xmath04 resembles a cylinder rather than a sphere which should be expected in a galaxy with an h@xmath0i scale height well below @xmath21pc .
we note , however , that it is notoriously difficult to derive depth structure in a reliable way .
= 8.95 cm all these points show that we still do not well understand the history of lmc@xmath04 .
to improve this situation , photometry of star groups _ inside _
lmc@xmath04 was taken in 1993 with the goal to derive ages .
this paper presents that data and the results .
additionally the star formation history is further investigated in sect . 4 by looking for overlapping age groups in the derived mass functions .
our datajbraun / diploma.html in the www for further information ) . ] were taken in april , 1993 ( during 13 nights in dutch time ) with the @xmath22 m dutch telescope ( 1st dataset ) . a second ,
very limited dataset was taken on the 6th february , 1996 with the @xmath23 m danish telescope .
the dutch telescope was equipped with a @xmath24pix@xmath25 tektronix ccd ( eso # 33 ) with a scale factor of @xmath26 pix@xmath27 ( @xmath17pix = @xmath28 m ) and a total field of view of @xmath29 , the danish telescope with a @xmath30pix@xmath25 loral / lesser ccd ( w11 - 4 chip ) with a scale factor of @xmath31 pix@xmath27 ( @xmath17pix = @xmath32 m ) .
cccccccrccr field sequence & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath2 & @xmath37 & & @xmath38 & @xmath39 & + number & [ pix ] & [ pix ] & [ pix ] & [ pix ] & [ mag ] & [ mag ] & & [ mag ] & [ mag ] & + @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & & @xmath40 & @xmath40 & + 2.0030 & 678.1 & 437.7 & 340.9 & 266.8 & 16.145 & 0.074 & @xmath410.153 & 0.116 & @xmath412.696 & @xmath410.263@xmath42 + 2.0031 & 853.4 & 441.3 & 165.7 & 270.4 & 16.130 & 0.075 & 1.697 & 0.119 & @xmath412.711 & 1.587@xmath42 + 2.0032 & 757.4 & 450.6 & 261.6 & 279.7 & 11.866 & 0.073 & 0.000 & 0.116 & @xmath416.975 & @xmath410.110@xmath42 + 2.0033 & 732.9 & 453.7 & 286.2 & 282.8 & 17.990 & 0.079 & 0.918 & 0.130 & @xmath410.852 & 0.808@xmath42 + 2.0035 & 799.3 & 458.3 & 219.7 & 287.4 & 16.559 & 0.074 & @xmath410.143 & 0.116 & @xmath412.282 & @xmath410.253@xmath42 + @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & @xmath40 & & @xmath40 & @xmath40 & + the 1st dataset contains 25 ccd fields ( 024 , see fig .
3 ) with an overlap of about 80@xmath13pix .
they form two coherent strips , one @xmath3pc strip through the ob superassociation lh@xmath077 from east to west with 10 fields ( i.e. ccd positions which cover an area of @xmath43 each ) and an @xmath5pc strip from the south to the north with 15 fields , reaching the rim of the supergiant shell lmc@xmath04 ( see figs . 1 and 3 ) . the whole j-shaped area is about 298@xmath44 in size ( without the 16% of overlapping regions ) . for each position we have long exposed frames of 10@xmath13min in the @xmath1 and 5@xmath13min in the @xmath2 passband ( eso # 419 and # 420 ) reaching down to approximately @xmath45mag ( fields 04 corresponding to region a , see sect .
4 ) or to @xmath46mag ( fields 524 corresponding to regions b e ) and short exposed frames of @xmath17min in @xmath1 and @xmath47min in @xmath2 , all with a seeing of [email protected]@xmath48 .
the 2nd dataset contains 3 ccd fields ( @xmath49 , @xmath50 and @xmath51 , see fig .
3 ) at the beginning and the end of the area covered by the 1st dataset . this set was obtained to get a better calibration of the main dataset , so we took short exposed frames in the @xmath1 and @xmath2 passbands ( eso # 450 and # 451 ) reaching down to @xmath52mag with a seeing of [email protected]@xmath48 .
the data reduction was carried out with midas and the built - in photometry package daophot ii of stetson ( 1987 ) .
additionally some special steps ( e.g. correction of bad double columns and reading the time information of the header ) was done with iraf . to calibrate the frames we used the following standard fields of landolt ( 1992 ) : rubin@xmath0149 , pg@xmath00918@xmath53029 and pg@xmath01633@xmath53099 . because of wrong time information in the file headers we were only able to use the standard fields of night 8 and 9 of the 1st dataset ( i.e. fields 1116 ) and to calibrate the beginning ( i.e. fields 03 ) and the end ( i.e. fields 2324 ) of the j-shaped region by the 2nd dataset . to get a homogeneous dataset ( see table 1 )
we adjusted the magnitude levels of the fields 410 and 1722 by their overlapping areas .
cc@ c@ c@ c@ ccll & + & + & & + & & & @xmath2 & @xmath54 & 18 & & @xmath55 & @xmath56 + & & & @xmath57 & @xmath58 + & 18 & @xmath59 & @xmath2 & @xmath54 & 20 & & @xmath60 & @xmath61 + & & & @xmath62 & @xmath63 + & 20 & @xmath59 & @xmath2 & & & & @xmath64 & @xmath65 + & & & @xmath66 & @xmath67 + the calibration of the 1st dataset was done with the following equations : @xmath68{l } v_{\rm n } - 3.210\,(13)\;\mbox{mag } + 0.029\,(7 ) \cdot \\[0.2 cm ] \left [ ( b - v)_{\rm n } - 0.450\,(18)\;\mbox{mag } \right ] \end{array}\ ] ] @xmath69{l } \left [ ( b - v)_{\rm n } - 0.450\,(18)\;\mbox{mag } \right ] / 0.907\,(9 ) \end{array}\ ] ] with index n indicating the normalization to exposure time @xmath17s and airmass @xmath70 and of the shift from point spread function ( psf ) to aperture magnitudes . the airmass correction was made by means of the atmospheric extinction coefficients measured on la silla by the geneva group ( burki et al .
1995a , b ) .
the mean errors of our photometry are given in table 2 .
the total error contains all statistical and systematical errors ( including psf fit , calibration and psf to aperture shift as their main part ) and is rather an overestimation ( see fig .
2 ) , while the statistical error from daophot is a clear overestimation of the reached accuracy . cr@ r@ r@lr@ r@ r@lccc macs & & & field sequence & @xmath35 & @xmath36 + name & [ @xmath71 & @xmath72 & @xmath73 & ] & [ @xmath74 & @xmath75 & @xmath76 & ] & number & [ pix ] & [ pix ] + @xmath40 & & & @xmath40 & @xmath40 & @xmath40 + j0533@xmath41669#053 & 5 & 33 & 59&.941 & @xmath4166 & 59 & 56&.69 & 2.0202 & 216.1 & 412 .6 + j0534@xmath41670#002 & 5 & 34 & 02&.625 & @xmath4167 & 00 & 32&.02 & 2.0175 & 254.6 & 338 .9 + j0534@xmath41670#003 & 5 & 34 & 03&.094 & @xmath4167 & 00 & 59&.24 & 2.0032 & 261.6 & 279 .7 + j0534@xmath41670#004 & 5 & 34 & 03&.522 & @xmath4167 & 01 & 49&.54 & 2.0117 & 273.7 & 170 .4 + j0534@xmath41670#005 & 5 & 34 & 04&.026 & @xmath4167 & 00 & 10&.46 & 2.0047 & 269.8 & 386 .5 + @xmath40 & & & @xmath40 & @xmath40 & @xmath40 + = 18.0 cm
there is no ccd photometry inside lmc@xmath04 in the literature overlapping with our data , so we compared the brightest stars of spectral type b0 to a9 and luminosity class i / ia marked by sanduleak ( 1969 ) with the @xmath1 , @xmath2 magnitudes of rousseau et al .
( 1978 ) .
= 8.70 cm this leads to an unexpected high deviation of the 1st dataset , which concerns only up to five of the brightest ( evolved ) stars of a ccd field , so it wo nt effect the results of the isochrone fit .
for all other stars the values of the 1st and 2nd dataset agree within the scope of the errors . to calculate absolute magnitudes we used the distance modulus of @xmath77mag ( see westerlund 1990 , and refs . therein ) corresponding to a distance of @xmath78kpc . for identification purpose we looked for stars having macs entries ( magellanic catalogue of stars ) to get good positions ;
rrrrr region & field & & & + & & & & + e & 24 & @xmath79 & n & 0.08 + & 23 & @xmath80 & n & 0.08 + & 22 & @xmath81 & n & 0.09 + & 21 & @xmath82 & n & 0.10 + & 20 & @xmath83 & n & 0.11 + d & 19 & @xmath84 & 11 & 0.11 + & 18 & @xmath85 & 11 & 0.11 + & 17 & @xmath86 & 10 & 0.11 + & 16 & @xmath87 & 10 & 0.09 + & 15 & @xmath88 & 10 & 0.09 + c & 14 & @xmath89 & 11 & 0.04 + & 13 & @xmath90 & 11 & 0.00 + & 12 & @xmath91 & 11 & 0.09 + & 11 & @xmath92 & 11 & 0.11 + & 10 & @xmath93 & 11 & : 0.11 + b & 9 & @xmath94 & 14 & : 0.11 + & 8 & @xmath95 & 13 & : 0.11 + & 7 & @xmath96 & 10 & : 0.11 + & 6 & @xmath97 & 9 & : 0.11 + & 5 & @xmath98 & 9 & : 0.11 + a & 4 & @xmath99 & 10 & : 0.11 + & 3 & @xmath100 & 13 & 0.08 + & 2 & @xmath101 & 13 & 0.11 + & 1 & @xmath102 & 11 & 0.11 + & 0 & @xmath82 & 16 & 0.11 + : indicating rough values ( see sect .
3 ) + n number of ms stars too small for accurate age determination , but consistent with @xmath103myr + e.g. sk@xmath10467@xmath0198 ( sanduleak 1969 ) equals macs j0534@xmath41670#003 ( de boer et al . 1995 ; tucholke et al .
1996 ) with the coordinates ( 2000 ) : ra : @xmath105 and dec : @xmath106 ( see fig . 3 and table 3 ) .
examples of the resulting colour - magnitude diagrams ( cmds ) are shown in fig .
we now proceed and will fit isochrones by eye to the colour - magnitude diagram of each field separately , with @xmath107)$ ] and as fit parameters .
we used the isochrones of the geneva group ( schaerer et al .
1993 ) for lmc metallicity ( @xmath108 or [ fe / h]@xmath109dex ) .
these isochrones include the effect of convective core overshooting and are based on the new opacities . the use of the isochrones of the padova group ( alongi et al .
1993 ) would lead to slightly younger ages ( about @xmath110 in logarithmic age ) ; but we stuck to the geneva data to stay consistent with the analyses of other regions in and around lmc@xmath04 by olsen et al .
( 1997 ) , petr ( 1994 ) and wilcots et al .
( 1996 ) , as discussed in sect .
5 ( see table 6 ) . in all
, we estimate that the accuracy of the determined ages is @xmath111 in logarithmic units . according to ratnatunga & bahcall ( 1985 )
the number of foreground stars is negligible in respect of the fit .
= 18.0 cm thus one can expect about 3 stars of our galaxy towards the lmc in the colour range of @xmath112mag and in the apparent visual magnitude range of @xmath113mag @xmath114mag ( corresponding to the absolute visual magnitude range for lmc objects of approximately @xmath115mag @xmath116mag ) . as examples of this fit to the 1st dataset we show four cmds in fig .
2 : the ccd field 0 with the largest age of the analysed area ( a ) , field 6 with one of the smallest ages ( b ) , and field 22 to demonstrate the impossibility of fitting an isochrone to the last five fields ( i.e. fields 2024 corresponding to region e , see sect .
4 ) because of a poorly populated upper main sequence ( c ) .
we also show the cmd for the whole j-shaped region ( i.e. all ccd fields , 024 , without overlap ) with the appropriate isochrones and the main sequence cut used to set up the mass function ( see braun 1996 for all cmds ) .
we find that all our fields indicate an age between 9 and 16@xmath13myr .
the results of all 25 fields are listed in table 4 .
the top of the main sequence region may contain data points for evolved stars . given
the age derived we expect their number to be small so that they can not effect the age determination in a significant way .
the values for the reddening are in accordance with the foreground reddening of @xmath117\;$]mag ( oestreicher et al .
in performing the calibration of the fields in sequence from 3 to 11 we found small systematic inconsistencies .
since field 3 and 11 could be calibrated in an absolute manner , we have interpolated the calibration for the fields in between , leading to less certain values of the stellar colour and thus of @xmath118 , as given in table 4 .
we judge the effect of this uncertainty on the ages as negligible . for all our fields one has to realize that we implicitly assume that all stars seen in a field belong to the same population . however , projection effects can mask age gradients , because in fitting isochrones one is only sensitive to the youngest star population . considering that star formation has been going on in all regions of the lmc
some influence has to be expected . to handle this problem
we investigate the luminosity and mass function for our region in the next section .
to get a first hint about the possibility of a superposition of stellar populations with different ages - the isochrone fitting would only be sensitive for the youngest population if the turn - off points ( tops ) of the main sequence ( ms ) are not too different - we combine 5 ccd fields to one region ( with an area of @xmath119 on the sky , see fig .
1 ) . for each region
we construct the luminosity function ( with slope @xmath120 ) and the mass function ( with slope @xmath121 ) , as well as for the whole analysed area .
this was made for equidistant intervals @xmath19 with the usual approach ( see scalo 1986 ) : @xmath122 + \mbox{const.}\ ] ] @xmath123 c@r@rl@rl region & & & & & + e & @xmath124 & @xmath125 & @xmath126 & @xmath125 & @xmath127 + d & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 + c & @xmath133 & @xmath134 & @xmath135 & @xmath136 & @xmath137 + b & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + a & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 + total@xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 + as an example we present in fig . 4 the luminosity and the mass function of region b. the translation of brightness to mass for stars on the ms was done with the best fitting geneva isochrone of 10@xmath13myr , which can be described with the numerical relation : @xmath154{l } 4.184 - 1.609 \cdot m_v/[{\rm mag } ] \\[0.2 cm ] + 0.303 \cdot \left ( m_v/[{\rm mag } ] \right)^2 \\[0.2 cm ] - 0.036 \cdot \left ( m_v/[{\rm mag } ] \right)^3 \\[0.2 cm ] - 0.000\,94 \cdot \left ( m_v/[{\rm mag } ] \right)^4 \\[0.2 cm ] + 0.001\,73 \cdot \left ( m_v/[{\rm mag } ] \right)^5 \end{array}\ ] ] with a rms error of 0.031 , beeing valid up to the top at @xmath155mag ( corresponding to @xmath156 ) .
the resulting slopes are , as shown in table 5 , very close to the standard values of @xmath157 and @xmath158 ( will 1996 ) .
a higher negative slope would indicate the superposition of different star populations ( note that incompleteness in the faint bins causes a flatter distribution than it really would be ) .
since the slopes we find are the same as those found in general , we conclude that there is no significant contribution of older age groups in these fields up to 300@xmath13myr ( equivalent to stars of 3@xmath13 still being on the main sequence ) .
even older star formation events may be present but they would contribute at fainter levels than the useable limit of our photometry .
the ages of the various star groups of lmc@xmath04 have been collected in table 6 .
note that the location of the various groups can be found in fig .
1 . for ngc 1948
we list only the newer age determination of will et al .
( 1996 ) superseding the one by vallenari et al .
( 1993 ) .
lrrl object & & & + & & & + ngc@xmath01978 & 2@xmath0200 & 0.08 & bomans et al .
1995 + ngc@xmath01948@xmath159 & 510 & 0.20 & will et al .
1996 + lh@xmath072@xmath160 , north & 815 & 0.000.04 & olsen et al .
1997 + lh@xmath072@xmath160 , south & 5 & 0.060.17 & olsen et al .
1997 + region e@xmath161 & n & 0.080.11 & this paper , table 4 + region d@xmath161 & 1011 & 0.090.11 & this paper , table 4 + region c@xmath161 & 11 & 0.00:0.11 & this paper , table 4 + region b@xmath162 & 914 & : 0.11 & this paper , table 4 + region a@xmath162 & 1016 & 0.08:0.11 & this paper , table 4 + ngc@xmath02004 & 16 & 0.09 & sagar & richtler 1991 + lh@xmath063@xmath163 & 14 & 0.07 & petr 1994 + lh@xmath060@xmath164 & 9 & 0.04 & petr 1994 + lh@xmath054@xmath165 & 6 & 0.10 & petr 1994 + lh@xmath076@xmath166 & 25 & 0.09 & wilcots et al . 1996
+ + + + + + + + + + + the ages we derived for the j-shaped region are in the range of 9 to 16@xmath13myr .
also ngc@xmath01948 , ngc@xmath02004 , lh@xmath072 north , lh@xmath063 , and lh@xmath060 are 9 to 16@xmath13myr old .
this seems to suggest that most of lmc@xmath04 was formed as an entity some 9 to 16@xmath167myr ago .
much younger are lh@xmath054 and lh@xmath076 , with an age of 5@xmath13myr , but these associations are located on the very rim of lmc@xmath04 .
n@xmath051 , with lh@xmath063 , 60 and 54 , shows an age gradient ( petr et al .
1994 ; petr 1994 ) , a clear hint at sspsf on smaller scales .
much more recent starformation was triggered at the edge of lmc@xmath04 , after the entire interior was in existence for some 5@xmath13myr ( the time between formation of the lmc@xmath04 interior and that of lh@xmath054 ) .
it is interesting to note that 5@xmath0myr is about the time the more massive stars need to become supernova .
also lh@xmath072 south is young , but it may have formed indeed more recently , in the aftermath of events in lh@xmath072 north .
note that the globular cluster ngc@xmath01978 , the rather conspicuous object near the northern rim of lmc@xmath04 and having an age of approximately @xmath168gyr ( bomans et al .
1995 ) , has nothing to do with the creation and/or evolution of sgs lmc@xmath04 .
this statement is valid for all objects older than @xmath169myr .
both ngc@xmath01948 and ngc@xmath01978 are part of the @xmath170 key region e centered at ra : @xmath171 and dec : @xmath172 ( de boer et al . 1989 , 1991 ; will et al .
lmc@xmath04 contains a substantial population of young stars of age 10@xmath13myr , while a quite older ( @xmath173myr ) background population may exist .
a huge gas cloud must have been present in which the conditions were favourable for the burst of star formation @xmath174myr ago .
the consequences of that burst are observed today .
these can be summarized as follows .
a volume of which we see the projected surface area of about @xmath17kpc@xmath25 is filled with essentially coeval stars .
the volume contains little neutral gas in numerous low column density filaments ( domgrgen et al .
1995 ) , and at the same time clearly contains ionzed gas ( see bomans et al .
1996 ) .
inside lmc@xmath04 we have a large number of young main sequence stars . from the present day mass function ,
extended with the same slope to the mass rich end to also include the original mass rich stars , we can calculate the number of stars in each mass range @xmath175 $ ] initially present in our j-shaped region : @xmath176{l } \int\limits_{m_{\rm l}}^{m_{\rm u } } 10^{4.972\,(22 ) } \cdot ( { \cal m}/{\mbox{${\cal m}_{\odot}$}})^{-2.716\,(44)}\;d{\cal m } \end{array}\ ] ] a fit to the geneva evolutionary tracks ( schaerer et al .
1993 ) for an initial mass @xmath177 yields a stellar lifetime relation of : @xmath178{l } ( 86 \pm 10)\;{\rm myr } \cdot ( { \cal m}/{\mbox{${\cal m}_{\odot}$}})^{-0.722\,(62 ) } \end{array}\ ] ] stars with an initial mass above @xmath179 will become supernovae of type ii . combining eqs . ( 6 ) and ( 7 ) with @xmath180 we get the number of supernovae after a given time : @xmath181{l } 1.4 \cdot ( t/[{\rm myr}])^{2.38 } \end{array}\ ] ] for star populations one would expect all stars with @xmath182\;{\mbox{${\cal m}_{\odot}$}}$ ] to have exploded into sne in the first 10@xmath13myr .
thus we get @xmath183 , for roughly 5% of the lmc@xmath04 area . extrapolating from the j-shaped region to the entire interior of lmc@xmath04
means that in this sgs of 10@xmath13myr age about 57@xmath184 supernovae have exploded . with an average supernova energy output of 10@xmath185erg
the total number of past sne will have dumped at least 10@xmath186erg into lmc@xmath04 .
on the other hand , the supernova rate right after starformation is very small and eq .
( 8) indicates that after 5@xmath13myr only @xmath187 200 sne have exploded inside lmc@xmath04 , or less than 10% of the total in 10@xmath13myr . at present
there is little neutral gas inside lmc@xmath04 , whereas the structure must have had lots of relatively dense neutral gas for the star formation .
we can estimate the minimum energy input required to dissolve the birth cloud . assuming a thickness of originally 500@xmath13pc with a gas density in the volume @xmath2 , the energy needed for total ionization is @xmath188ev = @xmath189erg @xmath190 .
a further calculation shows that roughly @xmath191erg @xmath190 can accelerate all particles of the entire birth cloud to 100@xmath13 , enough to disrupt the cloud .
this energy is easily provided by the supernovae .
in fact , a substantial fraction of the energy needed is already released between 5 and 8@xmath13myr . summarizing , based on the recognition that all young stars of lmc@xmath04 are nearly coeval at 10@xmath13myr , and on the sequel that supernovae will go off everywhere inside lmc@xmath04 at a fair rate dumping energy rather evenly in the birth cloud
, we can explain the structure of lmc@xmath04 as we see it today . in consequence of that , once the first supernovae occur , their individual ( but soon the collective )
blast waves may trigger star formation at the edges .
this fits with the age derived for lh@xmath072 south .
starformation at the edges of lmc@xmath04 will have taken place at very recent times as secondary process .
we expect that an age determination for the stars in any of the h@xmath0ii regions to the east of lmc@xmath04 will show very young ( @xmath54 5@xmath13myr ) ages , like the ones at the nw ( will et al .
1996 ) and the sw ( petr et al . 1994 ) .
the present supernova rate inside lmc@xmath04 as calculated from our equations is about 1 per 670@xmath13yr which is comparable to the sn rate of the whole lmc derived from counting supernova remnants ( chu & kennicutt 1988 ) .
our mass function and main sequence lifetime indicate an increase of the sn rate to 1 per 120@xmath13yr in 25@xmath13myr from now , at a time when the stars of originally 8@xmath192 will explode .
the ages found for the stars inside lmc@xmath04 lie in the range of @xmath6myr to @xmath7myr without a visible correlation with the distance to the lmc@xmath04 centre ( see table 4 and fig .
this means that there must have been one triggering event for star formation inside the greatest sgs of the lmc with a diameter of @xmath8kpc , whereas at the rim of lmc@xmath04 there is evidence for sspsf ( petr et al .
1994 ; petr 1994 ) .
thus sspsf ca nt be the creation mechanism of lmc@xmath04 as claimed by the first models ( see e.g. dopita et al . 1985 ) , even if one is able to find examples of sspsf in this region on scales below @xmath193pc .
jmb , jmw and djb acknowledge support from the deutsche forschungsgemeinschaft ( dfg ) , in the frame of the graduiertenkolleg `` the magellanic system and other dwarf galaxies '' ( grk 118/2 - 96 ) .
djb is grateful for a feodor - lynen - fellowship of the alexander von humboldt - foundation .
alongi m. , bertelli g. , bressan a. , et al .
, 1993 , a&as 97 , 851 bomans d.j .
, vallenari a. , de boer k.s . , 1995 , a&a 298 , 427 bomans d.j . ,
de boer k.s . ,
koornneef j. , grebel e.k .
, 1996 , a&a 313 , 101 braun j.m . ,
1996 , diploma thesis , university of bonn burki g. , rufener f. , burnet m. , et al .
, 1995a , a&as 112 , 383 burki g. , rufener f. , burnet m. , et al .
, 1995b , eso messenger 80 , 34 chu y .- h .
, kennicutt r.c . , 1988 , aj 96 , 1874 davies r.d .
, elliot k.h . , meaburn j. , 1976 , mem .
r. astron .
81 , 89 ( dem ) de boer k.s . , azzopardi m. , baschek b. , et al . , 1989 , eso messenger 57 , 27 de boer k.s . ,
spite f. , franois p. , et al .
, 1991 , eso messenger 66 , 12 de boer k.s .
, tucholke h .- j .
, seitter w.c . , 1995 , eso messenger 81 , 20 domgrgen h. , bomans d.j .
, de boer k.s . , 1995 , a&a 296 , 523 dopita m.a . , mathewson d.s .
, ford v.l . , 1985 ,
apj 297 , 599 feitzinger j.v . ,
glassgold a.e . , gerola h. , seiden p.e . , 1981 , a&a 98 , 371 goudis c. , meaburn j. , 1978 , a&a 68 , 189 henize k.g .
, 1956 , apjs 2 , 315 hodge p.w . ,
wright f.w . , 1967 ,
`` the large magellanic cloud '' , smithsonian press kennicutt r.c . ,
hodge p.w . , 1986 ,
apj 306 , 130 landolt a.u . , 1992 , aj 104 , 340 lucke p.b . , 1974 ,
apjs 28 , 73 lucke p.b . ,
hodge p.w . , 1970 , aj 75 , 171 mckibben nail v. , shapley h. , 1953 , proc .
39 , 358 meaburn j. , 1980 , mnras 192 , 365 oestreicher m.o . , gochermann j. , schmidt - kaler t. , 1995 , a&as 112 , 495 olsen k.a.g . ,
hodge p.w . , wilcots , e.m . , pastwick l. , 1997 , apj 475 , 545 petr m. , 1994 , diploma thesis , university of bonn petr m. , bomans d.j .
, grebel e.k . , 1994 , in the `` 3rd ctio/ eso workshop on the local group '' , layden a. , smith r.c . ,
storm j.(eds . ) , eso conf .
ratnatunga k.u .
, bahcall j.n . , 1985 , apjs 59 , 63 reid n. , mould j. , thompson i. , 1987 , apj 323 , 433 rousseau j. , martin n. , prvot l. , et al . , 1978 ,
a&as 31 , 243 sagar r. , richtler t. , 1991 , a&a 250 , 324 sanduleak n. , 1969 , ctio , contribution no .
89 scalo j.m . , 1986 , fund .
11 , 1 schaerer d. , meynet g. , maeder a. , schaller g. , 1993 , a&as 98 , 523 stetson p.b .
, 1987 , pasp 99 , 191 tenorio - tagle g. , bodenheimer p. , 1988
, ara&a 26 , 145 tucholke h .- j . ,
de boer k.s . ,
seitter w.c .
, 1996 , a&as 119 , 91 vallenari a. , bomans d.j .
, de boer k.s . , 1993 , a&a 268 , 137 westerlund b.e . , 1990 , a&ar 2 , 29 wilcots e.m . , hodge p.w .
, king n. , 1996 , apj 458 , 580 will j .- m . , 1996 , ph.d .
thesis , university of bonn will j .- m . ,
bomans d.j .
, tucholke h .- j . , et al .
, 1995 , a&as 112 , 367 will j .- m . ,
bomans d.j .
, vallenari a. , schmidt j.h.k .
, de boer k.s . , 1996 , a&a 315 , 125 | the youngest stellar populations of a j-shaped region inside the supergiant shell ( sgs ) lmc@xmath04 have been analysed with ccd photometry in @xmath1 , @xmath2 passbands .
this region consists of 2 coherent strips , one from the east to west reaching about @xmath3pc across the ob superassociation lh@xmath4 and another extending about @xmath5pc from south to north .
the standard photometric methods yield 25 colour - magnitude diagrams ( cmds ) which were used for age determination of the youngest star population by isochrone fitting .
the resultant ages lie in the range from @xmath6 m to @xmath7 m without correlation with the distance to the lmc@xmath04 centre .
we therefore conclude that there must have been one triggering event for star formation inside this great lmc sgs with a diameter of @xmath8kpc .
we construct the luminosity function and the mass function of five regions consisting of 5 fields to ensure that projection effects do nt mask the results .
the slopes lie in the expected range ( @xmath9 $ ] and @xmath10 $ ] respectively ) .
the greatest values of the slope occur in the north , which is caused by the absence of a young , number - dominating star population .
we have calculated the rate with which supernovae ( sne ) have exploded in lmc@xmath04 , based on the finding that all stars are essentially coeval .
a total of 57@xmath11 supernovae has dumped the energy of @xmath12erg over the past 10@xmath13myr into lmc@xmath04 , in fact enough to tear the original star - forming cloud apart in the time span between 5 and 8@xmath13myr after the starformation burst .
we conclude that lmc@xmath04 can have been formed without a contribution from stochastic self - propagating star formation ( sspsf ) , although the ring of young associations and h@xmath0ii regions around the edge have been triggered by the events inside lmc@xmath04 . |
increased interest in the use of organic materials@xcite for a variety of optical and electronic applications , including the fabrication of electroluminescent devices , field - effect transistors , lasers etc .
, has followed the successful demonstration of devices with useful lifetimes and performance .
the new advances depend on a better choice of organic materials , and on new technologies borrowed from recent developments in semiconductor technology , thin films , surface science and materials preparation .
the luminescence of organic light emitting diodes ( oled ) proceeds by the injection of electrons and holes which form excitons and then decay radiatively to the ground state.@xcite although better design of the diode structure can improve the quantum efficiency of the device considerably , selection rules for optical transitions in these organic materials limit the radiative channel to singlet excitons .
hence the three - times more abundant triplet excitons decay by _ nonradiative _ channels and are wasted .
this raises the possibility of an enormous gain in efficiency if the triplet excitons could also be harvested for light emission .
baldo et al .
attempted to tap the triplet excitons by energy transfer from the host organic to a porphine - platinum flourescent dye.@xcite the objective of this work is to design the _ host material itself _ so that the singlet - triplet selection rules become irrelevant .
this is done using heavy atoms where spin - orbit interactions ( @xmath1 coupling ) become important , so that @xmath2 is no longer a good quantum number .
we note in passing that the spin selection rules can not be circumvented by applying magnetic fields , or incorporating local magnetic impurities since the spin remains a good quantum number even in the presence of such fields .
it is necessary to bring the electrons into atomic sites where the spin - orbit coupling intervenes to break the spin selection rules . since metal chelates ( e.g. ,
al , in , sc ) of 8-hydroxyquinoline have provided some of the most successful oled materials , we will consider molecules based on modifications of these standard materials .
since alq@xmath0 , i.e. , al(8-hydorxyquinoline)@xmath0 , has been a very successful host material ( fig .
1 ) in oleds , we consider two structures derived from alq@xmath0 where the quinoline part is modified .
these are shown in figs . 2 and 3 , and
are obtained by first replacing the n atom by an as atom ( an arsenoquinoline , denoted by al(asq)@xmath0 ) , and then replacing the carbon atom _ para _ to the as atom by a boron atom , cf . , fig . 3 , to give al(asbq)@xmath0 .
we proceed to investigate if such molecules are stable , and what luminescence properties could be expected from them , using sophisticated state of the art quantum chemical calculations .
these molecules were chosen for this investigation following detailed studies of alq@xmath0 itself .
for example , on examining the electron densities of the ground state and the excitonic states of alq@xmath0 , it is found that the al ion itself does not play a strong role in determining the exciton density .
that is , replacing the al by a heavy chelating atom where the @xmath1 coupling is strong would not help the luminescence .
this is also experimentally known to be true , e.g. , in the scandium complex . on the other hand ,
the n atom in the quinoline ring participates decisively in the relevant excitonic states . thus replacing the n atom by
as is appealing .
of course , the spin - orbit interaction is even stronger in sb , but replacing n by sb would be too strong a chemical modification and the synthesis of the corresponding molecule may be even more difficult . hence here we examine the case of replacing n by as .
preliminary calculations showed that the n@xmath3as substitution had the effect of shifting the luminescence to the red , while the shorter wavelength of the alq@xmath0 is more attractive for device applications .
it was envisaged that replacing the _ p-_c atom , i.e , the c atom in the _
para_-position to the n - atom , by boron would shift the spectrum back to the shorter wavelength regime . in this communication
we examine these questions using sophisticated quantum calculations .
we find that the n@xmath3as leads to a strong improvement in the spectrum , without too much of an adverse red shift , while the further substitution of the _ p-_c atom by b is not useful .
the ground states of the metal chelates were calculated using the gausssian-98 code at the b3lyp/3 - 21 g * level.@xcite .
both al(asq)@xmath0 and al(asbq)@xmath0 were found to be stable .
hence it should be possible to achieve a synthesis of these materials .
the vertical excitation spectra were calculated at the tdb3lyp/3 - 21 g * level.@xcite . at this point
we have _ not _ included spin - orbit coupling as yet .
this is because the chelate structures , containing three 8-hydroxyquinoline groups , are too large for a direct implementation of relativistic effects .
this would be undertaken as the next stage of this investigation . in anticipation of those results
we have simply replaced the triplet oscillator strengths which are zero by an analogous singlet oscillator strength and plotted the excitation spectra ( see also table i ) . some justification for the assumption that the singlet @xmath3 groundstate oscillator strength and that for the triplet @xmath3 ground become very similar in the presence of the spin - orbit interaction in as is given in the appendix .
the resulting enhanced triplet+singlet spectra as well as the standard ( singlet only ) spectra are shown in fig . 4 the excitations marked with an asterisk in table i , for alq@xmath0 and al(asq)@xmath0 are transitions which involve the h ( i.e. , homo ) and l ( i.e , lumo ) orbitals in a major way . thus in alq@xmath0 , the first singlet ,
i.e. , s1 , is approximately 0.98h @xmath3 l + other configurations . while the t4 is approximately 0.96h @xmath3 l + other configurations . in the as substituted form al(asq)@xmath0 we have t4 being 0.9h
@xmath3 l + other configurations , while the s1 is higher in energy and is an admixture with 0.94h @xmath3 l + other configurations .
hence these transitions may be identified with the homo - lumo gaps used in simplified theoretical schemes .
such simplified schemes are of course modified when ci is included , where in various excited states ( based on the ground state determinant ) get included .
the deficiencies of the simpler schemes become quite evident when we go to the boron substituted form al(asbq)@xmath0 . here
it is not possible to identify a transition where the initial state is mostly homo ( i.e. , @xmath4 90% ) .
in fact , t5 , at 2.105 ev has 0.9h but the final orbital is l+4 admixed with other configurations .
the h @xmath3 l transitions are at t8 with 0.76h in the initial orbital , and s4 with 0.79h in the initial orbital .
this shows that the b substitution effectively introduces transitions in the nominal h @xmath3 l gap of simplified theories .
four additional transitions of al(asbq)@xmath0 , which might be labeled t10 , s7 , s8 , and s9 in the notation of table i are at 2.908 , 2.913 , 3.057 , and 3.103 ev and complete the same energy window as in the other two compounds .
comparison of the calculate al(q)@xmath0 spectra with experiment is not shown here , since such a comparison was carried out in a previous publication.@xcite such comparison require taking due account of solvent effects , or aggregation effects etc , and as well as considerations of non - radiative pathways , since the experimental spectral intensities are affected by many such factors .
the objective of this study is to examine the al(q)@xmath0 , al(asq)@xmath0 , and al(asbq)@xmath0 spectra calculated under the _ same _ conditions , and in the same energy window .
these results show the al(asq)@xmath0 spectrum in the same energy window as al(q)@xmath0 ; however , its spectral features are more intense .
the main intensity is red shifted by 0.3 ev , confirming the conclusions based on more elementary considerations .
however , there are now transitions in the blue as well .
the substitution of the _ p-_carbon by boron , i.e , al(asbq)@xmath0 , shows that the stronger absorption lines are somewhat blue shifted .
the center of gravity of the spectral intensity is slightly blue shifted , although many peaks in the middle of the energy range are red shifted .
the substitution of a nominally trivalent boron atom for a nominally @xmath5 c site leads to some distortion of the original ring structure .
the improvement of the spectrum is questionable .
hence , it is clear that we need to look for a better choice than boron .
in fact , fig .
4 suggests that the al(asq)@xmath0 provides an excellent improvement over the more usual alq@xmath0 while its synthesis would not be as difficult as the boron substituted form .
we show that the spectra of metal chelated 8-hydorxyquinolines can be manipulated and improved to obtain much better luminescence properties .
we have shown how the spectrum can be shifted and that the intensity would increase if the triplet excitations could be harvested .
the replacement of the n in alq@xmath0 by an as atom provides a definite improvement , while the further substitution of the _ p-_carbon by b is not seen to provide a useful improvement .
this is a preliminary study which anticipates a more detailed calculation where the @xmath1 coupling would be explicitly included .
the relativistic effects for heavy atoms can be included up to @xmath6 , where @xmath7 is the fine structure constant , in first principles pseudopotentials .
the result splits into two terms of the form : @xmath8<l|\ ] ] where the @xmath9-sum is a sum over angular momentum states .
the first term is the scalar part , and is not significant for the problem of the singlet - triplet bottleneck . the second term in the above equation
, @xmath10 introduces mixing between orbital and spin angular momentum states , leading to the break down of the usual spin selection rules .
however , the full inclusion of the spin - orbit term in first - principles ci calculations of the type discussed in this paper is numerically quite prohibitive , given the large number of electrons which need to be handled in typical led materials .
leave aside ci calculations , the numerical evaluation even with a pseudopotential scheme is extremely demanding.@xcite the spin - orbit operator is highly local in the sense that it is mostly sensitive to changes in electron density close to the nucleus , where relativistic effects dominate .
thus we may look at calculations for systems containing as as a guide . in order to get at least a simple `` grosso modo '' estimate , we may compare the homo - lumo gap ( bandgap ) of as obtained from a non - relativistic ( i.e , standard ) calculation with a calculation where some effort is put into an evaluation of the relativistic effects.@xcite thus gonze et al .
have carried out calculations which can be used for a simple estimate of the effect of the spin - orbit effect in as .
the s - levels in as are about 9 ev lower than the p - levels .
the p - levels suffer a splitting of about 0.36 ev in as , and is a measure of the strength of the spin - orbit term in as .
since we are concerned with singlet and triplet excitons , we should compare this spin - orbit splitting with the gap between hole energies and the fermi energy , and electron energies and the fermi energy in as .
numerical values for these , for as have been given by j .-
issi,@xcite , and by priestly et al.,@xcite .
issi gives 0.154 ev and 0.202 ev for electrons and holes , while priestly et al give 0.177 and 0.190 ev .
hence the total effect for an electron - hole pair is of the order of 0.35 ev by one estimate , and 0.36 ev by the second estimate .
these numbers are in fact very similar to the spin - orbit energy ( @xmath11 0 .
these considerations suggest that when estimating the oscillator strengths for the processes @xmath12groundstate@xmath4 and @xmath13groundstate@xmath4 , we may , as a _ grosso modo _ estimate take the oscillator strengths to be of the same magnitude .
r. bauernschmitt and r. ahlrichs , _ chem . phys .
, * 256 * , 454 ( 1996 ) ; m.e .
casida , c. jamorski , k.c.casida and d.r .
salahub , _ j. chem
_ , * 108 * , 4439 ( 1998 ) ; r.e .
stratmann , g.e .
scuseria and m.j .
_ , * 109 * , 8218 ( 1998 ) . cccccccccccc transition & alq@xmath0 & & transition & al(asq)@xmath0 & & transition & al(asbq)@xmath0 & @xmath14 & + type & @xmath15 & @xmath16 & type & @xmath15 & @xmath16 & type & @xmath15 & @xmath16 & + + t1 & 2.152 & & t1 & 1.847 & & t1 & 1.870 & & + t2 & 2.190 & & t2 & 1.854 & & t2 & 1.932 & & + t3 & 2.228 & & t3 & 1.867 & & t3 & 1.945 & & + s1*&2.718 & 0.0047 & t4 * & 2.613 & & t4 & 2.102 & & + t4*&2.741 & & s1 * & 2.631 & 0.0129 & t5 & 2.105 & & + s2 & 2.909 & 0.0590 & t5 & 2.744 & & t6 & 2.181 & & + s3 & 2.924 & 0.0180 & s2 & 2.769 & 0.0060 & s1 & 2.333 & 0.0008 & + t5 & 2.925 & & s3 & 2.897 & 0.1066 & s2 & 2.337 & 0.0015 & + s4 & 3.000 & 0.0423 & s4 & 2.998 & 0.0265 & s3 & 2.409 & 0.0011 & + s5 & 3.029 & 0.0027 & t6 & 3.002 & & t7 & 2.595 & & + t6 & 3.056 & & t7 & 3.054 & & * t8 & 2.609 & & + s6 & 3.187 & 0.0173 & s5 & 3.116 & 0.0242 & * s4 & 2.614 & 0.0169 & + t7 & 3.228 & & s6 & 3.195 & 0.0244 & s5 & 2.643 & 0.0138 & + t8 & 3.255 & & t8 & 3.232 & & t9 & 2.753 & & + s7 & 3.273 & 0.0049 & s7 & 3.324 & 0.0741 & s6 & 2.756 & 0.0004 & + fig.1 structure of the light emitter alq@xmath0 .
see ref.@xcite for bond lengths , angles etc .
+ fig.2 n atoms of the quinolines have been replaced by as in al(asq)@xmath0 + fig.3 the c atom in the _
para_-position to n has been replaced by b in al(asbq)@xmath0 .
the kekule - type structure is only schematic .
+ fig.4 calculated spectra of the three compounds .
solid curves : only the singlet is available for light emission ; dashed curves : the triplet exciton is also harvested ( using a @xmath16 comparable to the nearest singlet ( see appendix ) .
note that the intensity axis is logarithmic .
the calculation is for isolated molecules , and hence a broadening of 0.025 ev has been included . | electroluminescence in organic light emitting diode ( oled ) materials occurs via the recombination of excitonic electrons - hole pairs
. only the singlet excitons of commonly used oled materials , e.g. , aluminum trihydroxyquinoline ( alq@xmath0 ) , decay radiatively , limiting the external quantum efficiency to a maximum 25% .
thus 75% of the energy is lost due to the triplet bottleneck for radiative recombination .
we consider molecules derived from alq@xmath0 which bypass the triplet bottleneck by designing structures which contain strong spin - orbit coupling . as a first stage of this work
, we calculate the groundstate energies and vertical excitation energies of al - arsenoquinolines and al - boroarsenoquinolines .
it is found that the substitution of n by as leads to very favourable results , while the boron substitution leads to no advantage . |
natural numbers which are integer multiples of their digit reversals are called _ palintiples_. the most well - known examples of base-10 palintiples include @xmath0 and @xmath1 . in general , letting @xmath2 represent the natural number @xmath3 where @xmath4 is a natural number and @xmath5 for all @xmath6 , we say @xmath7 is an @xmath8-_palintiple _ provided @xmath9 for some natural number @xmath10 .
base-@xmath11 palindromes and examples with leading zeros are not considered so that @xmath12 and @xmath13 . previous work on palintiple numbers mostly restricts its attention to a limited number of digits @xcite or on characterizing base-10 palintiples @xcite .
other work includes constructive methods for finding palintiples when they exist @xcite .
both of these papers use a graph - theoretical framework to gain insight into the problem .
the graphs essentially visualize an efficient search a for palintiples using the potential carries as nodes . as sloane @xcite and young @xcite
have noted , such graphs have interesting properties in and of themselves .
the work of @xcite establishes some general properties of palintiple numbers of any base having an arbitrary number of digits .
the methods therein pay particular attention to the structure of the carries which naturally separates all palintiples into three mutually exclusive and exhaustive classes .
letting @xmath2 be an @xmath8-palintiple with carries @xmath14 , @xmath15,@xmath16 , @xmath17 , these classes are defined as follows : we say that @xmath7 is _ symmetric _ if @xmath18 for all @xmath6 and that @xmath7 is _ shifted - symmetric _ if @xmath19 for all @xmath20 .
a palintiple that is neither symmetric nor shifted - symmetric is called _
asymmetric_. the examples given above are symmetric . among other things , @xcite finds all shifted - symmetric palintiples .
this paper continues these efforts by further conjecturing the nature of all symmetric palintiples while further characterizing them .
a characterization of shifted - symmetric palintiples is also presented in order to gain insight into the asymmetric case .
the result is a characterization of asymmetric palintiples which involves a only a simple condition on the base and multiplier .
lastly , we address the problem of bases for which asymmetric palintiples do not exist . for more examples of palintiples as well as a more thorough introduction to the topic
, the reader is directed to @xcite and @xcite .
for convenience , we state two results found in @xcite .
let @xmath21 be an @xmath8-palintiple and let @xmath22 be the @xmath23th carry .
then @xmath24 for @xmath6 . from these equations the digits
may be stated in terms of the carries : @xmath25 for @xmath6 .
we now consider a conjecture which , if proved , would answer an open question posed by @xcite and further characterize symmetric palintiples .
suppose @xmath26 is an @xmath8-palintiple with carries @xmath27 .
then , if @xmath7 is symmetric , then @xmath28 for @xmath20 . [ carries_mod_n_minus_one ] if the conjecture [ carries_mod_n_minus_one ] is true , then an @xmath8-palintiple is symmetric if and only if @xmath29 divides @xmath11 .
[ n_plus_one_divides_b ] suppose @xmath26 is an @xmath8-palintiple with @xmath27 as its carries .
the reverse implication is established in @xcite .
suppose then @xmath7 is symmetric .
then @xmath30 .
supposing conjecture [ carries_mod_n_minus_one ] holds , @xmath31 . if @xmath32 then @xmath33 by equation [ digits ] .
therefore since @xmath34 for all @xmath20 as shown in @xcite , @xmath35 .
another application of equation [ digits ] finishes the proof .
provided conjecture [ carries_mod_n_minus_one ] , then the following characterization of symmetric palintiples also holds .
@xmath26 is a symmetric @xmath8-palintiple if and only if the following conditions hold : 1 .
@xmath36 for all @xmath20 where @xmath37 is an integer such that @xmath38 and @xmath39 is a palindromic binary sequence .
2 . @xmath40 and @xmath41 .
3 . @xmath42 implies @xmath43 and @xmath44 implies @xmath45 for all @xmath46 .
[ reg1 ] suppose @xmath7 is a symmetric @xmath8-palintiple with carries @xmath27 .
corollary [ n_plus_one_divides_b ] implies that @xmath38 for some integer @xmath37 . since @xmath34 for all @xmath20 , an application of equation [ digits ] and the * conjecture *
show that the first condition is satisfied . by definition @xmath47 so that @xmath40 , and if @xmath48 , then @xmath49 which is a violation of the condition that there are no leading zeros .
if @xmath42 but @xmath45 , then @xmath50 which is not a base-@xmath11 digit . similarly , if @xmath44 but @xmath43 , then @xmath51
now suppose conditions 1 , 2 , and 3 hold .
conditions 2 and 3 guarantee that @xmath52 and that there are no leading zeros .
from here it is a routine matter to show that @xmath7 is an @xmath8-palintiple .
condition 1 guarantees that @xmath29 divides @xmath11 which , by corollary [ n_plus_one_divides_b ] , proves that @xmath7 is symmetric .
casting the above into the light of recent work by kendrick @xcite who showed that the @xmath8-young graph , denoted here as @xmath53 , is isomorphic to @xmath54 ( otherwise known as the `` 1089 graph '' since @xmath55 is a @xmath56-palintiple ) if and only if @xmath29 divides @xmath11 ( as conjectured by sloane @xcite ) , we ask if the following are equivalent for an @xmath8-palintiple @xmath7 with carries @xmath57 : 1 .
@xmath7 is symmetric 2 .
@xmath58 3 .
@xmath59 4 .
@xmath29 divides @xmath11 . if @xmath58 , the work of kendrick @xcite shows that any node of has the form @xmath60 $ ] , @xmath61 $ ] , @xmath62 $ ] , or @xmath63 $ ] which establishes @xmath64 .
@xmath65 is easily established since by equation [ digits ] , @xmath66 and @xmath67 .
@xmath68 is demonstrated in @xcite .
proving the above equivalence ( which as demonstrated above is tantamount to proving ( 1 ) @xmath69 ( 2 ) , and we leave it as an open question ) would prove conjecture [ carries_mod_n_minus_one ] .
in contrast to their symmetric and shifted - symmetric counterparts , it is not well understood under what conditions asymmetric palintiples exist . whereas
, conditions on @xmath10 and @xmath11 which characterize symmetric and shifted - symmetric palintiples given by corollary [ reg1 ] and theorem [ reg2condition ] below also assure their existence ( see @xcite and the proof of theorem [ regbases ] below ) , the same can not be said for asymmetric palintiples .
moreover , the behavior of asymmetric palintiples is considerably more haphazard . on one hand ,
@xmath70-digit symmetric palintiples exist in every base for all @xmath71 .
similarly , the existence of a shifted - symmetric @xmath8-palintiple ensures the existence of a @xmath70-digit example for all @xmath72 . on the other hand , minimal examples of asymmetric palintiples
might exist for only a certain number of digits .
( a @xmath70-digit @xmath8-palintiple is considered minimal if no @xmath8-palintiples exist for less than @xmath70 digits . )
this minimal number of digits can be quite large : the smallest @xmath73-palintiple is 105 digits long .
furthermore , unlike the symmetric and shifted - symmetric cases , the existence of a @xmath70-digit asymmetric @xmath8-palintiple does not guarantee the existence of a @xmath74-digit example .
( we note , however , that symmetrically concatenating digit strings of @xmath8-palintiples yields other representations of larger @xmath8-palintiples ( @xcite ) .
thus , the existence of a @xmath70-digit minimal example always guarantees the existence of a @xmath75-digit example for any natural number @xmath76 . )
what is more , asymmetric palintiples seem to completely lack the kinds of patterns which are found in the carries and digits ( figure 1 ) of symmetric and shifted - symmetric palintiples
. * figure 1 * : a plot of @xmath77 versus @xmath78 for all minimal palintiples for @xmath79 .
the next theorem states a simple condition between @xmath10 and @xmath11 which characterizes shifted - symmetric palintiples . an @xmath8-palintiple is shifted - symmetric if and only if @xmath80 .
[ reg2condition ] suppose @xmath2 is an @xmath8-palintiple with carries @xmath14 , @xmath15,@xmath16 , @xmath17 . if @xmath7 is shifted - symmetric , then , by results in @xcite , @xmath81 must have non - trivial solutions less than or equal to @xmath82 .
that is , @xmath83 which establishes the forward implication .
suppose then that @xmath80 .
equation [ digits ] implies that @xmath84 .
consequently , @xmath85 .
hence , @xmath86 divides @xmath87 . without loss of generality
suppose @xmath88 .
then by our hypothesis , it follows that @xmath89 for some integer @xmath90 .
but since @xmath34 for all @xmath20 , we see that @xmath91 so that @xmath92 .
equation [ digits ] implies that @xmath93 .
suppose now that @xmath19 and that @xmath81 . equation [ digits ] implies @xmath94 .
this with our inductive hypotheses proves that @xmath95 , that is , @xmath96 . the same argument given above for @xmath97 shows that @xmath98 and the proof is complete .
if conjecture [ carries_mod_n_minus_one ] holds , then an @xmath8-palintiple is asymmetric if and only if @xmath99 and @xmath100 .
[ irreg ] if conjecture [ carries_mod_n_minus_one ] holds , then two @xmath8-palintiples are both either symmetric , shifted - symmetric , or asymmetric .
[ pal_type ] corollary [ irreg ] sheds some light upon the case of asymmetric palintiples by narrowing down possible values of @xmath10 and @xmath11 for which they may exist .
it should be noted , however , that this result does not specify conditions which guarantee their existence .
finding such conditions is key to finding all palintiples .
the work of kendrick @xcite reveals the asymmetric class to be more diverse than previously thought .
given the large variety of young graph isomorphism classes that arise for even relatively low bases ( @xmath101 ) and the apparent rate at which the number of isomorphism classes grows with @xmath11 , the difficulties presented by the asymmetric class are not surprising .
such difficulties naturally lead to the question of bases which do not allow for their existence ( that is , bases for which all palintiples are known ) . considering bases up to 20 ,
asymmetric palintiples are known to exist in bases 8 , 11 , 14 , 15 , 17 , 18 , 19 , and 20 . on the other hand ,
methods developed by young @xcite along with corollary [ irreg ] show that if conjecture [ carries_mod_n_minus_one ] holds , that no asymmetric palintiples exist in bases 5 , 7 , 9 , and 13 . in similar fashion
, it is conjectured that only symmetric palintiples exist in bases 10 , 12 , and 16 .
arguments in @xcite show that no asymmetric palintiples exist in bases 3 , 4 , and 6 .
suppose @xmath103 is composite .
choose @xmath12 such that @xmath29 divides @xmath103 .
then @xmath104 .
therefore @xmath80 so that there is at least one non - trivial solution @xmath105 to the congruence @xmath106 that is less than or equal to @xmath82 .
methods in @xcite then allow us to construct a shifted - symmetric @xmath8-palintiple .
hence , @xmath11 can not be a strongly symmetric base .
it is still unknown if infinitely many symmetric or strongly symmetric bases exist .
this number may very well be finite : electronically implementing the methods of young @xcite we see that , with the exception of those mentioned above , there are no symmetric bases less than 544321 .
thus , the converse of theorem [ regbases ] fails to hold in most cases .
we leave these as open questions . | a palintiple is a natural number which is an integer multiple of its digit reversal . a previous paper partitions all palintiples into three distinct classes according to patterns in the carries and then determines all palintiples belonging to the shifted - symmetric class .
in this paper we consider a strategy for finding all symmetric palintiples based upon recent work .
we also discuss the last case of asymmetric palintiples and consider bases for which asymmetric palintiples do not exist . |
observations of fluctuations in cosmic microwave background ( cmb ) can provide important constraints on cosmological models .
large angular scale ( @xmath0 ) observations probe the initial conditions , in particular the amplitude and the slope of primordial power spectrum ( @xcite ; @xcite ; @xcite ) .
these scales could also provide information on the geometry and the matter content of the universe ( @xcite ; @xcite ; @xcite ) .
however , theoretical interpretation of measurements on these scales is complicated by cosmic variance and this intrinsically limits the accuracy with which these parameters can be estimated using large angular scale measurements alone .
while small scale measurements suffer less from cosmic variance , their interpretation is complicated by the microphysics during recombination and/or reionization .
theoretical models often give wildly different predictions for the anisotropy power spectra when the parameter values are only slightly changed , while some combinations of parameters seem to provide nearly identical spectra ( e.g. @xcite ) .
the purpose of this letter is to clarify which combinations of physical parameters affect the cmb fluctuations and what are the physical processes that lead to these fluctuations .
our two - fluid model for adiabatic fluctuations , presented in 2 , generalizes previous theoretical approximations that investigated cmb fluctuations in the limiting cases of large and small angles ( @xcite ; @xcite ) .
the model is accurate enough that it can be used for a quantitative analysis of various models , yet it is also simple enough that it can clearly separate between different physical processes that affect the cmb fluctuations . in
we use this model to identify the physical parameters that can be determined using cmb measurements over a large angular range , extending previous studies that were limited to a smaller range of angles and/or parameters ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . in
4 we present the conclusions .
in this section we present a method for computing the cmb anisotropies that is a generalization of analytic methods first introduced by sachs & wolfe ( 1966 ) and is based on the line - of - sight integration along the photon past light cone .
we will assume that the photons and baryons are tightly coupled prior to recombination , which will allow a simple two - fluid description of perturbations ( @xcite ) .
our analysis will be restricted to the linear perturbation theory of adiabatic perturbations and we will neglect any possible tensor contributions .
different theoretical models will be compared using angular power spectrum of cmb anisotropies , @xmath1 , where @xmath2 is the multipole of temperature anisotropy @xmath3 and @xmath4 is the spherical harmonic .
expected cmb anisotropy for a given experiment can be calculated from @xmath5 , where @xmath6 is the experiment window function .
the temperature fluctuation @xmath7 in the direction @xmath8 can be expressed as a line - of - sight integral , @xmath9e^{-\mu } d\tau .
\label{bolt - sol}\ ] ] here @xmath10 is the conformal time with the value @xmath11 today , @xmath12 is the gravitational potential , @xmath13 is the photon density perturbation and @xmath14 is the electron velocity .
we introduced the thomson opacity along the past light cone @xmath15 with @xmath16 , where @xmath17 is the expansion factor , @xmath18 the ionization fraction , @xmath19 the electron number density and @xmath20 the thomson cross section .
the above expression is written in the gauge invariant formalism ( @xcite ) .
we neglected the anisotropic stress contribution and the unobservable monopole contribution arising from the local gravitational potential . in the limit of infinitely thin lss
the function @xmath21 approaches to a dirac delta - function @xmath22 , where @xmath23 denotes the conformal time at recombination . equation [ bolt - sol ] then reduces to ( kodama & sasaki 1984 ) @xmath24 the velocity term in equation [ bolt - sol - rec ] can be rewritten using the linear theory approximation @xmath25 into @xmath26 , where @xmath27 is the radial coordinate .
we may decompose the sources in equation [ bolt - sol - rec ] into an orthonormal spherical basis set with fourier amplitudes @xmath28 , @xmath29 and @xmath30 , where @xmath31 .
after the angular and ensemble averaging we obtain the following expression for the multipole moments ( @xcite ) , @xmath32 where @xmath33 is the spherical bessel functions and @xmath34 its derivative .
all the perturbed quantities are evaluated in @xmath35-space at @xmath23 .
@xmath36 denotes the primordial power spectrum of potential @xmath12 , usually expressed as a power law @xmath37 . for later purpose we introduced the function @xmath38 , which incorporates the damping effects . in the case of a non - flat universe
the functions @xmath33 need to be substituted by their appropriate generalizations .
this is unimportant for @xmath39 and the results presented here will be valid for non - flat universes , provided that the relation between angles and physical sizes is expressed using angular size distances .
last term in equation [ cl ] gives the so - called integrated sachs - wolfe ( isw ) contribution .
the time dependence of the potential is denoted with @xmath40 , where @xmath41 .
this term vanishes in a flat , matter dominated @xmath42 universe , but is present in the case of a vacuum energy dominated universe ( @xcite ) , curvature dominated universe ( @xcite ) or when the universe is in transition epoch from being radiation to being matter dominated ( @xcite ) . to calculate the anisotropy power spectrum we need to evaluate the source contributions in equation [ cl ] at the epoch of recombination .
the photon evolution equations in @xmath35-space are given by ( @xcite ; @xcite ; @xcite ) @xmath43 we also need the evolution equations for baryon and cdm perturbations , @xmath44 where @xmath45 and @xmath46 are the baryon and photon mean densities , respectively .
the energy and momentum constraint equations give the equations for @xmath12 and @xmath47 , @xmath48 where @xmath49 and @xmath50 . here
@xmath51 and @xmath52 are the neutrino and cdm mean densities , respectively .
we replaced neutrino density and velocity perturbations with the corresponding photon perturbations .
this becomes invalid on small scales due to the free - streaming of neutrinos , but does not affect significantly the final results .
we also neglected the anisotropic shear and the possible curvature terms .
the above equations are supplemented by the friedmann equation , which at early times ( when a possible cosmological constant or curvature term can be neglected ) is given by @xmath53 the solution to this equation is @xmath54 where @xmath55 ( assuming three flavors of massless neutrinos ) , @xmath56 for the standard recombination , @xmath57 , @xmath58 is the value of matter density today in units of critical density and @xmath59 is the value of hubble constant today in units of 100km / s / mpc
. we will now assume the tight coupling limit @xmath60 , which is a good approximation on scales larger than the silk damping scale ( @xcite , @xcite ) .
in this case the photons and baryons are coupled into a single fluid with @xmath61 and @xmath62 .
the above equations rewritten in terms of dimensionless time @xmath63 and dimensionelss wavevector @xmath64 become @xmath65 \nonumber \\ \phi=&-{3\over 2 } ( \eta/\kappa)^2(\delta+3\eta v /\kappa ) \,,\ \ \ & \dot{\phi}=-\eta \phi+ { 3 \eta^2 v \over 2 \kappa } \nonumber \\ \delta = & { \delta_\gamma[1+{3\over 4}(y - y_c)]+y_c\delta_c \over 1+y}\ , , \ \ \ \ & v = { v_\gamma({4 \over 3 } + y - y_c)+y_cv_c \over 1+y } , \label{eqs}\end{aligned}\ ] ] where the derivatives are taken with respect to @xmath63 , @xmath66{\omega_{b } \over \omega_{m}}y= 1.68 { \omega_{b } \over \omega_{m } } y$ ] , @xmath67 and @xmath68 .
equations [ eqs ] are a coupled system of 4 first order differential equations . the appropriate initial conditions at @xmath69 ( when the universe is radiation dominated ) and @xmath70 ( when the mode is larger than the hubble sphere radius ) are @xmath71 .
\label{init}\end{aligned}\ ] ] the above equations need to be evolved until @xmath72/\alpha$ ] . the temperature anisotropy expressed with the dimensionless variables is given by @xmath73 ^ 2 , \label{clfinal}\ ] ] where @xmath74 is the angular distance to the lss in units of @xmath75 and we assumed @xmath76 .
the term @xmath77/10y(x_{rec})$ ] arises from the isw effect due to the potential varying with time during the transition period from the radiation dominated to the matter dominated universe ( @xcite ) . for simplicity
we dropped the isw contribution from possible curvature or cosmological constant , which is only important at the lowest values of @xmath78 ( @xmath79 ) .
the damping transfer function @xmath80 is approximately unity for low values of @xmath78 ( @xmath81 , but gradually decreases afterwards .
its main contributions come from the silk damping and from the finite width of lss .
the first effect can be calculated analytically by expanding equations [ photon ] and [ cdmb ] to second order in @xmath82 and neglecting the effects of gravity and expansion . in the matter - dominated era one
obtains @xmath83 ( @xcite ) , where @xmath84 is the silk damping scale in units of @xmath75 , @xmath85 .
second effect can be analytically estimated by performing the line - of - sight integral in equation [ bolt - sol ] in the limit where the sources are slowly changing over the timescale on which the visibility function @xmath21 is non - negligible ( @xcite ) .
visibility function can be approximated as a gaussian @xmath86 $ ] , where for standard recombination @xmath87 . in the limit @xmath88 ( where @xmath89 is the wavevector which gives the dominant contribution to @xmath90 )
, we obtain the damping factor @xmath91 .
therefore , the damping effects can be written as @xmath92 this works reasonably well for the standard ionization history .
note however that if the limit @xmath88 is not satisfied , then the damping due to the finite thickness of lss is not exponential , but is proportional to @xmath93
. this will be the case , for example , in reionized models .
equations [ eqs]-[transfer ] are all is needed to evaluate the temperature fluctuations . although equations [ eqs ] can not be solved analytically in general , they have analytic solutions in the limits of small and large @xmath94 . in the first limit where the modes are larger than the hubble sphere radius
the amplitude of perturbations at a given time is a constant ( figure [ fig1 ] ) .
this gives the standard sachs - wolfe expression for cmb fluctuations , as can be verified by evolving the initial conditions in equations [ init ] into the matter - dominated era and neglecting the velocity term in equation [ clfinal ] . in the second limit ( large @xmath94 )
the equations can be solved using the wkb approximation and the solution is given by the acoustic oscillations of photon - baryon plasma ( @xcite , @xcite ) . in the intermediate regime , which is of main interest for us , the equations need to be solved numerically , but the physics can be well understood by the two limits above . as shown in figure [ fig1 ] , equations [ eqs ] give an excellent approximation to the exact results over a large range of wavevector @xmath94 .
although the above system of equations is particularly useful for the standard recombination scenarios , one can also use it to calculate anisotropy power spectrum in reionized models . as one can see from equation [ bolt - sol ] , the primary fluctuations will be suppressed by a factor @xmath95 $ ] , where @xmath96 is thomson opacity at recombination .
in addition , there will be secondary fluctuations generated at the new last - scattering surface , which can be calculated using the same method as above , except that one needs to replace @xmath97 with @xmath98 in the regime where the compton drag is negligible .
since the modes larger than the hubble sphere radius at the new lss do not evolve in the matter - dominated regime this simply regenerates the sachs - wolfe expression for low values of l. on smaller scales the thick new lss damps the secondary fluctuations and in many scenarios these become negligible ( although on arcminute scales the second - order terms may become important , @xcite ) .
provided that one is interested in degree angular scales , then the effect of reionization is to suppress the fluctuation power spectrum relative to the large scales by a factor of @xmath99 $ ] .
the comparison between our approximation and the exact solutions of perturbed boltzmann equation ( seljak & bertschinger 1994 ; seljak 1994 ) is presented in figure [ fig2 ] for several cosmological models . all of the multipole moments are normalized relative to @xmath100 , which is approximately fixed by the cobe experiment and where the curvature effects and isw effects due to @xmath101 can be neglected .
one can see that the agreement is excellent over a large range of @xmath78 .
the deviations at large @xmath78 arise because of improper treatment of damping effects , while the deviations at small @xmath78 can be attributed to the neglection of neutrino anisotropic shear .
another effect that introduces small deviations is the dependence of @xmath102 on cosmological parameters , which slightly offsets the position of the peaks .
nevertheless , our approximation correctly predicts the positions and amplitudes of doppler peaks ( also called acoustic or sakharov oscillations ) with a 20 % accuracy over most of parameter range .
given the high accuracy of our model we may now investigate how the anisotropy power spectrum depends on the cosmological model .
our goal is to identify the parameters that can be determined using the cmb measurements and to give a physical understanding of how they affect the anisotropies . in equations [ eqs ] the free parameters are @xmath103 and @xmath104 .
we can replace them with physically more relevant parameters @xmath105 and @xmath106 .
in addition to these two we have the parameters @xmath96 , @xmath107 , @xmath74 and @xmath84 .
these six parameters will uniquely determine all the cmb power spectra within our approximation .
of the parameters above , @xmath107 , @xmath96 and @xmath84 all suppress the power on small scales relative to large scales ( for @xmath108 ) .
the suppression is different in the three cases , being proportional to @xmath109 ( neglecting the possible tensor contribution ) , @xmath95 $ ] and @xmath110 , respectively .
this , in principle , allows to separate the different suppression effects and to determine the three parameters separately ( but see @xcite ) .
note that these parameters do not change the positions of the doppler peaks , only their amplitude .
since the effects of these parameters are physically transparent , we will restrict in the following to the case of @xmath111 and @xmath112 . moreover , silk damping is important only for large values of @xmath78 and it can be neglected if one concentrates on the first few doppler peaks .
we are thus left with @xmath74 , @xmath113 and @xmath114 , which uniquely determine positions of the doppler peaks .
position of the first doppler peak is determined by the angular size of the hubble sphere radius at decoupling , which , expressed in terms of our variables is given by @xmath115 . since @xmath116 for typical values of @xmath113 ,
position of the first doppler peak mainly depends on @xmath74 .
assuming @xmath117 we have @xmath118 for the model with negligible cosmological constant and @xmath119 for the model with negligible curvature .
the latter result shows that the first doppler peak only weakly depends on @xmath120 .
this is because the angular size distance at large redshifts scales with @xmath121 in approximately the same way as does the hubble sphere radius at decoupling ( @xcite ; @xcite ) .
this is , however , not true in general and the position of the first doppler peak depends on @xmath121 and @xmath122 when both curvature and cosmological constant are important .
the value of @xmath78 at which the maximum of the first doppler peak lies is given approximately by @xmath123 .
figure [ fig3]a compares @xmath90 s of curvature and cosmological constant dominated models with those of @xmath42 model at fixed values of @xmath113 and @xmath114 .
one can see that the position of the first doppler peak can accurately determine @xmath121 in curvature dominated universe ( @xcite ) , but can not precisely determine @xmath121 in cosmological constant dominated universe ( @xcite ; @xcite ) .
however , even in this model the positions of secondary doppler peaks are already significantly displaced relative to each other when @xmath121 changes from 0.25 to 1 .
this would thus allow independent determination of @xmath121 even in cosmological constant dominated universe , once @xmath113 and @xmath114 are known ( see below ) .
the dependence of the doppler peak positions and amplitudes on @xmath114 and @xmath113 is more complicated , since both parameters appear in the evolution equations [ eqs ] and change the properties of acoustic oscillations . moreover , the two parameters enter into the equations differently and have different physical effects : @xmath114 is related to the properties of photon - baryon plasma and determines its effective sound velocity at recombination , whereas @xmath113 is related to the time evolution of the expansion factor , since it determines the epoch of matter - radiation equality .
this means that one can not expect the anisotropy spectra to remain invariant under a certain combination of the two parameters and both @xmath114 and @xmath113 are required for a complete description of the doppler peaks .
figures [ fig3]b and [ fig3]c show how the doppler peaks change when one of the two parameters is changing while the other is held fixed .
if one concentrates on the first doppler peak then it is not possible to determine the two parameters simultaneously , since both increasing @xmath114 and decreasing @xmath113 lead to an increase in the first doppler peak .
the physical mechanisms that lead to this are different : while an increase in @xmath114 increases the amplitude of the first wave in @xmath97 and @xmath124 ( fig .
[ fig1 ] ) , a decrease in @xmath113 also leads to an increased isw contribution .
once the secondary peaks are observed as well , then different effects of the two parameters become significant and allow one to determine the two parameters simultaneously ( figs .
[ fig3 ] b , c , d ) . figure [ fig3]d shows how changing @xmath113 at a fixed value of @xmath125 affects the multipole moments .
again , since the epoch of matter - radiation equality is changing with @xmath113 , one does not expect the multipole moments to remain unchanged and both our approximation and exact calculations confirm this .
therefore , changing @xmath59 at a fixed @xmath126 and @xmath121 changes the anisotropy power spectrum , contrary to some recent claims ( @xcite ; @xcite ) .
the approximation for calculating anisotropy power spectrum presented here is a generalization of the sachs - wolfe approximation , which itself is only valid on scales larger than the hubble sphere radius at recombination . by modelling the cosmological perturbations as a two - component fluid plasma we extended this approach to all scales .
the approximation is useful both for developing the physical understanding of processes that affect cmb fluctuations , as well as for a quantitative prediction of multipole moments for various cosmological models .
the main approximations used in our model are a two - fluid approximation , neglection of anisotropic shear , a simplified treatment of thomson scattering effects , neglection of curvature effects and neglection of vector and tensor contributions .
none of these assumptions is essential for the method and one can generalize the approach presented here to obtain exact results ( seljak 1994 ) .
this will lead to a computationally more demanding system of equations , but the main physical effects that lead to the creation of doppler peaks will still be determined by the equations presented in this letter . by rewriting the equations in their dimensionless form we identified all the dimensionless parameters that affect the anisotropy power spectra .
measurements of cmb fluctuations can only determine these parameters .
for example , neutrinos enter into our equations indirectly through the friedmann equation [ friedmann ] and through the energy - momentum constraint equations [ constr ] .
the presence of a massive neutrino only weakly changes these equations and the resultant multipole moments are almost indistinguishable from the ones with the massless neutrino .
therefore , the question of whether neutrino has a mass has little hope to be answered using the cmb measurements .
the most interesting aspect of the cmb power spectra is the peculiar pattern of the doppler peaks , which allows a simultaneous determination of @xmath114 and @xmath113 .
this would provide an independent test of nuclesynthesis prediction of @xmath114 ( e.g. @xcite ) and would also constrain the parameter space on @xmath121 and @xmath59 . in addition , the position of the first doppler peak determines @xmath121 in curvature dominated model . in cosmological constant dominated model the position of the first doppler peak does not allow one to determine @xmath121 accurately , but positions of secondary doppler peaks could be used to constrain @xmath121 ( although for accurate determination exact calculations should be used in this case ) .
another way to break the degeneracy between @xmath121 , @xmath126 and @xmath59 is to determine the silk damping scale @xmath84 , which depends only on these three parameters and can not be expressed as a combination of @xmath114 and @xmath113 .
this would require a separation of silk damping from the damping due to the finite thickness of lss in reionized models and from the @xmath108 suppression of small scales relative to large scales ( including the possible tensor contribution ) .
this is possible , because the three effects suppress the small scale power differently .
thus , a combination of cmb measurements over a large range of angles could be used to separately determine the baryon mass density , matter mass density and the hubble constant value .
bond , j. r. , & efstathiou , g. 1984 , , 285 , l45 bond , j. r. , crittenden , r. , davis , r. l. , efstathiou , g. , & steinhard , p. j. 1994 , , to be published fugukita , m. , sugiyama , n. , & unemura , m. 1990 , , 358 , 28 g ' orski , k. m. et al .
1994 , , to be published g ' orski , k. m. , & stompor , k. m. 1994 , , 422 , l41 jrgensen , h.e .
, kotok , e. , naselsky , p. , & novikov , i. 1994 , , to be published kamionkowski , m. & spergel , d. n. 1994 , , to be published kamionkowski , m. , spergel , d. n. , & sugiyama , n. 1994 , , 426 , l57 kodama , h. , & sasaki , m. 1984 , prog .
78 , 1 kodama , h. , & sasaki , m. 1986 , int . j. mod
a1 , 265 kofman , l. , & starobinsky , a. 1985 , sov .
lett . , 11 , 271 ma , c. p. , & bertschinger , e. 1994 , , to be published muciaccia , p. f. , mei , s. , de gasperis , g. , & vittorio , n. 1993 , , 410 , l61 padmanabhan , t. 1993 , structure formation in the universe , cambridge : cambridge university press peebles , p. j. e. , & yu , i. t. 1970 , , 162 , 815 sachs , r. k. & wolfe , a. m. 1966 , , 147 , 73 seljak , u. 1994 , in preparation seljak , u. , & bertschinger , e. 1994 , to appear in `` present and future of the cosmic microwave background '' , eds
. j. l. sanz , e. martinez - gonzalez , & l. cayon ( in press ) silk , j. 1968 , , 151 , 459 smoot , g. f. et al .
1993 , , 396 , l1 sugiyama , n. & silk , j. 1994 , , in press vishniac , e. t. 1987 , , 322 , 597 vittorio , n. , & silk , j. 1985 , , 297 , l1 walker , t. p. , steigman , g. , schramm , d. n. , olive , k. a. , & kang , h. 1991 , , 376 , 51 wilson , m. l. , & silk , j. 1981 , , 243 , 14 wright , e. l. et al .
1994 , , to be published | we present a simple , yet accurate approximation for calculating the cosmic microwave background anisotropy power spectrum in adiabatic models .
it consists of solving for the evolution of a two - fluid model until the epoch of recombination and then integrating over the sources to obtain the cmb anisotropy power spectrum .
the approximation is useful both for a physical understanding of cmb anisotropies , as well as for a quantitative analysis of cosmological models .
comparison with exact calculations shows that the accuracy is typically better than 20 percent over a large range of angles and cosmological models , including those with curvature and cosmological constant . using this approximation
we investigate the dependence of the cmb anisotropies on the cosmological parameters .
we identify six dimensionless parameters that uniquely determine the anisotropy power spectrum within our approximation .
cmb experiments on different angular scales could in principle provide information on all these parameters .
in particular , mapping of the doppler peaks would allow an independent determination of baryon mass density , matter mass density and hubble constant .
# you can latex this if you have the aastex latex macro , three compressed # uuencoded tar figures are added at the end of this file . to unpack them #
just strip off the main text and call resulting file , e.g. , figs.uu # then say csh figs.uu .
if you are not on a unix machine , you should explicitly # execute the commands : uudecode figs.uu ; uncompress figs.tar.z ; # tar -xvf figs.tar . or ftp the compressed postscript # version using anonymous ftp on arcturus.mit.edu in /preprints / cmb_approx.ps.z # 1 |
during their evolution , the different stellar populations of galaxies produce and inject into the interstellar medium metals that congregate to form dust particles of different size and composition .
this dust , heated by the general stellar radiation field , re - emit the absorbed energy in the far - ir . in normal galaxies , those objects dominating in number the nearby universe , the emitting dust has a modified black body spectrum with a peak at @xmath6 200 @xmath5 m rapidly decreasing at longer wavelengths ( boselli et al .
2003 ) . because of the quiescent star formation activity of these objects , however , most of the dust has a relatively cold temperature ( @xmath6 20 k ) and is emitting in the submillimetric domain .
the total dust amount of normal galaxies can thus be measured only by means of observations in the 200 - 1000 @xmath5 m spectral range . + all physical ( star formation activity , gas content , metallicity ; boselli et al . 2001 ; 2002 ; zaritsky et al .
1994 ; gavazzi et al .
2004 ) , structural ( concentration index , light profile ; boselli et al . 1997 ; gavazzi et al .
2000 ; scodeggio et al .
2002 ) , kinematical ( shape of the rotation curve ; catinella et al . 2006 ) and stellar population ( stellar spectral energy distribution ; gavazzi et al . 2002 ; boselli et al .
2003 ) properties of galaxies are strongly related to their total stellar mass ( downsizing effect ) and marginally to their morphological type .
tracing the statistical dust properties of normal galaxies can thus be done by observing a large and complete sample of objects spanning the largest possible range in luminosity and hubble type .
+ although accessible to ground based facilities such as scuba , the observation of such large samples of normal , quiescent galaxies would be too time consuming ( more than 20 hours per galaxy ) ( dunne et al .
2000 ) . given its large field of view ( 4x8 ) and
its sensitivity ( @xmath6 7 mjy for a 5 @xmath7 detection for a point source in 1 hour integration ) , spire on herschel is the ideal instrument for such a survey .
for such a purpose , the spire extragalactic group selected a volume limited , complete sample of @xmath6 300 galaxies to be observed in 100 hours of guaranteed time in the three spire bands at 250 , 360 and 520 @xmath5 m .
to span the largest possible range in mass , the sample has been selected in the @xmath8 band ( from 2mass ) since near - ir luminosities are linearly related to the total dynamical mass of galaxies ( gavazzi et al .
1996 ) . a distance constraint of 15 @xmath0 @xmath1 @xmath0 25 mpc has been introduced to construct a volume limited sample .
this distance range limits distance uncertainties due to local peculiar motions and secure the presence of low - luminosity , dwarf galaxies , non accessible at higher redshift . to minimize galactic cirrus contamination ,
galaxies have been selected at high galactic latitude ( @xmath9 @xmath10 @xmath11 54@xmath12 ) and in low galactic extinction regions ( @xmath13 @xmath0 0.15 ; schlegel et al .
our primary sample contains all galaxies ( from ellipticals to spirals ) with @xmath8 @xmath0 9 ( @xmath6 126 objects ) . a secondary sample of @xmath6 187 late - type galaxies with 9 @xmath0 k @xmath0
12 has been added to cover the low luminosity range .
this secondary sample does not include the dust poor ellipticals whose emission would be hardly detected within reasonable integration times .
+ figure 1 shows the histogram of the morphology distribution of the selected sample .
as selected , the sample includes both relatively isolated and cluster object ( virgo and fornax ) and is thus ideal for a statistical study on the effects of the environment on the dust properties of galaxies ( figure 2 ) .
dwarf irregulars and blue compact galaxies , here undersampled in spite of the selected large range in luminosity , will be the targets of another spire key program ( madden et al .
, this conference )
. pointed observations will be done with integration times of 30 minutes for ellipticals and lenticulars , and 12 minutes for late - types . given a slewing overhead of 3 minutes per observation
, this will ensure an efficient use of the telescope . for early - type galaxies
the integration time has been determined by assuming a lower limit to their total dust mass of @xmath6 10@xmath14 m@xmath15 as determined from iras observations ( bregman et al .
1998 ) . with 30 minutes of integration a detection limit of 11 mjy at @xmath6 4 @xmath7 will be achived . for spirals
the integration time is dictated by the need of detecting the cold dust emission outside the optical radius . by combining isophot ( alton et al .
1998 ) and scuba ( valahakis et al . 2005 ) observations of extended sources with spectral energy distribution of normal galaxies of different type ( boselli et al .
2003 ) , and assuming a standard dust to gas ratio we estimate that 12 minutes of integration are sufficient to detect the cold dust associated to the extended hi disc .
with 12 minutes of integration time we will get to a detection limit of 22 mjy at @xmath6 5 @xmath7 .
the selected sample will allow us to trace for the first time the variation of the cold dust properties ( dust mass and temperature , dust to gas ratio , ... ) of normal galaxies along the hubble sequence and as a function of luminosity .
combined with multifrequency datasets , the spire data will be used to study the role of dust in the physics of the interstellar medium . through the shielding of the interstellar radiation field , in particular of the uv light , dust
participate in the cooling of the gas and thus plays a major role in the process of star formation .
dust is also an important catalyzer in the formation of the molecular hydrogen , and thus is critical entity for the study of the feedback .
modeling the energetic balance between emitted and absorbed light , uv to sub - millimetric spectral energy distributions of these 300 galaxies will be used to quantify the dust obscuration in different objects .
this analysis will allow us to define standard recipes for correcting uv and optical data , a useful tool for the interpretation of all modern surveys .
+ the presence of dust shells and discs in ellipticals will be used to study their hierarchical formation history . by studying the relationship between dust mass and other global properties of ellipticals
we will determine how much dust is produced by the old stellar population and how much is the result of mergers .
+ the comparison of cluster and isolated galaxies will allow us to make a detailed study on the effects of the environment on the dust properties of galaxies , and thus understand whether the hot and dense cluster intergalactic medium can be polluted through the gas stripping process of late - type galaxies ( boselli & gavazzi 2006 ) .
this dataset will be used to measure the local luminosity and dust - mass functions and distributions , important benchmarks for the deep herschel surveys planned by the spire team , providing at the same time a unique reference sample for any statistical study .
the proposed analysis can be done only once corollary data covering the whole electromagnetic spectrum will be available . given its definition , the herschel reference sample is easily accessible for ground based and space facilities : the selected galaxies are in fact relatively bright ( @xmath16 @xmath0 15 mag ) and extended ( @xmath6 2 - 3 arcmin ) .
uv data at 1500 and 2300 will be taken from the galex all sky survey or the uv atlas of nearby galaxies ( gil de paz et al .
2006 ) , optical and near - ir images will be secured by the sdss ( abazajian et al .
2005 ) and 2mass ( jarrett et al . 2003 ) surveys .
these data will be used to trace the underlying stellar population and reconstruct the star formation history of the target galaxies .
+ far - ir data in the 10 - 200 @xmath5 m spectral range , needed for characterizing the whole dust emission and thus constraining dust masses and temperatures , will be taken from the akari / astro - f survey at a similar spatial resolution ( @xmath6 20 - 30 arcsec , depending on the wavelength ) .
radio continuum data are already available from the nvss survey ( condon et al .
1998 ) , while hi data will be obtained thanks to the alfalfa survey ( giovanelli et al . 2005 ) .
+ dedicated spectroscopic observations in drift scan mode , needed to measure the metal content of the target galaxies as well as their underlying stellar population at higher spectral resolution than with imaging , is under way with carelec at the ohp 1.93 m telescope .
we also started h@xmath17 imaging observations ( needed to measure the present day star formation activity ) at the san pedro martir observatory .
+ follow up x - ray ( chandra , xmm ) , co ( fcrao , jcmt ) and radio ( westerbork , vla ) observations of the target galaxies are also in program .
abazajian , k. , et al@xmath18 , 2005 , 129 , 1755 alton , p. , trewhella , m. , davies , j. , et al . , 1998 , a&a , 335 , 807 boselli , a. , tuffs , r. , gavazzi , g. , hippelein , h. & pierini , d. , 1997 , a&as , 121 , 507 boselli , a. , gavazzi , g. , donas , j. scodeggio , m. , 2001 , aj , 121 , 753 boselli , a. , lequeux , j. , gavazzi , g. , 2002 , a&a , 384 , 33 boselli , a. , gavazzi , g. , sanvito , g. , 2003 , a&a , 402 , 37 boselli , a. , gavazzi , g. , 2006 , pasp , 118 , 517 bregman , j. , snider , b. , grego , l. , cox , c. , 1998 , apj , 499 , 670 catinella , b. , giovanelli , r. , haynes , m. , 2006 , apj , 640 , 751 condon , j. , et al .
, 1998 , aj , 115 , 1693 dunne , l. , eales , s. , edmunds , m. , ivison , r. , alexander , p. , clements , d. , 2000 , mnras , 315 , 115 gavazzi , g. , pierini , d. , boselli , a. , 1996 , a&a , 312 , 397 gavazzi , g. , franzetti , p. , scodeggio , m. , boselli , a. , pierini , d. , 2000 , a&a , 361 , 863 gavazzi , g. , bonfanti , c. , sanvito , g. , boselli , a. , scodeggio , m. , 2002 , apj , 576 , 135 gavazzi , g. , zaccardo , a. , sanvito , g. , boselli , a. , bonfanti , c. , 2004 , a&a , 417 , 499 gil de paz , a. , boissier , s. , madore , b. , et al .
, 2006 , apjs , in press ( astroph/0606440 ) giovanelli , r. , haynes , m. , brian , k. , et al . , 2005 ,
aj , 130 , 2598 jarrett , t. , chester , t. , cutri , r. , schneider , s. & huchra , j. , 2003 , aj , 125 , 525 schlegel , f. , finkbeiner , d. , davis m. , 1998 , apj , 500 , 525 scodeggio , m. , gavazzi , g. , franzetti , p. , boselli , a. , zibetti , s. , pierini , d. , 2002 , a&a , 384 , 812 valahakis , c. , dunne , l. , eales , s. , 2005 , mnras , 364 , 1253 zaritsky , d. , kennicutt , r. , huchra , j. , 1994 , apj , 420 , 87 | in order to study the dust properties of different galaxies in the nearby universe , the spire extragalactic group selected a volume limited ( 15@xmath0 @xmath1 @xmath025 mpc ) , complete sample of 313 galaxies spanning the whole range in morphological type ( from ellipticals to late - type spirals ) and luminosity ( 8.5 @xmath0 log @xmath2 @xmath0 12 l@xmath3 , -22 @xmath0 @xmath4 @xmath0
-16 ) extracted from 2mass , to be observed in guaranteed time with herschel . the 250 - 360 - 520 @xmath5 m spire data , combined with those collected at other frequencies , will be used to trace the dust properties of normal galaxies and provide a reference sample for studies at high redshift . |
density functional theory ( dft ) within the kohn - sham ( ks ) single - particle formalism has now been established as a standard tool for the study of inhomogeneous quantum many - electron systems .
@xcite the local density approximation ( lda ) and generalized gradient approximation ( gga ) are successful in describing the ground state properties of various systems . on the other hand ,
their eigenvalue spectra are known to be far from the reality , and it has been assumed that ks orbital eigenvalues are auxiliary quantities without any physical meaning except the highest occupied one.@xcite however , it is clear that dft is also a legitimate route to the study of excited state properties via perturbation theory@xcite or time - dependent dft . @xcite
especially , grling has recently formulated a ks formalism for excited states based on dft perturbation theory and demonstrated that the ks eigenvalue difference is a well - defined zeroth order approximation to excitation energies.@xcite this has been also explicitly confirmed for several atoms by accurate quantum monte - carlo ( qmc ) calculations.@xcite it appears that - at least for localized systems - the unrealistic description of eigenvalue spectra is mainly due to the deficiencies of the conventional approximations for the exchange - correlation energy functional such as the lda or gga rather than the ks scheme itself .
the origin of these deficiencies is now more or less understood from the investigations of the exact properties of the ks exchange - correlation potential @xmath2 which the lda or gga does not satisfy : ( i ) @xmath3 as @xmath4 for finite systems,@xcite ( ii ) the eigenvalue of the highest occupied single particle states is the negative of the ionization energy,@xcite ( iii ) @xmath5 is self - interaction free , and ( iv ) @xmath5 exhibits an integer discontinuity upon the addition of an electron .
@xcite however , it has been recognized that approximations to the _ local _ ks potential obtained from the _ orbital _ dependent exact exchange energy functional through the so - called optimized effective potential ( oep ) method@xcite closely observe these conditions.@xcite in spite of these good features , the implementation of the oep method is very complicated and its computational cost is excessively high .
only spherical atoms,@xcite jellium surfaces,@xcite jellium spheres,@xcite and periodic solids with a small unit cell@xcite have been extensively studied , and there exists no investigation of molecular systems with the oep method . however , about a decade ago , krieger , li , and iafrate ( kli ) proposed an approximation scheme to the oep , and have shown that the kli potential with the exact exchange - only energy functional ( kli - x ) satisfies the above mentioned properties ( i)-(iv ) of @xmath2 , and accurately reproduces the oep - x results for spherical atoms.@xcite within the kli - x approximation , the workload in constructing the potential is significantly reduced , but this method is still relatively complicated compared with the lda or gga and studies are therefore limited .
atomic systems have been originally investigated by kli,@xcite and more recently by grabo _
et al._@xcite bylander and kleinman studied semiconductors with the kli potential.@xcite although molecules are a very good testing ground of any functional approximation , only some diatomic molecules have been investigated by grabo _
et al._@xcite in this article , we report a study of several closed shell diatomic and polyatomic molecules within the kli scheme .
the questions we address are the performance of the kli approximation to the oep with the exact exchange - only functional ( kli - x ) and supplemented by lda and perdew - burke - ernzerhof gga@xcite correlation energy functionals ( kli - x@xmath6lda - c and kli - x@xmath6gga - c ) for both structural properties and eigenvalue spectra . the comparison with and qualitative differences from the lda / gga and the hartree - fock ( hf ) method will be emphasized .
in addition , since we incorporate the exact exchange , we can examine the accuracy of the lda and gga correlation energy functionals .
the underlying computational algorithm is the three - dimensional ( 3d ) finite - difference pseudopotential method@xcite and , to our knowledge , this is the first 3d kli investigation of molecular systems .
the organization of the paper is as follows . in sec .
[ sec : formalism ] , the formulation of kli is reviewed and the differences of the kli - x from the hf method will be emphasized . the implementation of kli - x into our molecular code and its test results are described . in sec . [
sec : results ] , results and discussions for several molecules will be presented for both structural properties and eigenvalue spectra .
especially , comparisons between the kli - x and the lda / gga or hf method are made , and physical origins of these differences will be discussed . in sec .
[ sec : conclusions ] , we conclude the paper by summarizing our work and future prospects .
we start from the spin - dependent ks equations for ks orbitals @xmath7 for spin channel @xmath8 and ks eigenvalues @xmath9 , with the possibility of fractional occupation numbers @xmath10 ( hartree atomic units are used throughout the paper ) @xmath11 { \psi_{\sigma i}}({{\mathbf r } } ) = \epsilon_{{\sigma i } } \psi_{{\sigma i}}({{\mathbf r}}),\ ] ] with the spin density @xmath12 where @xmath13 is the number of occupied @xmath14 spin orbitals .
the effective ks potential @xmath15 is composed of external , hartree , and exchange - correlation contributions , @xmath16 where the exchange - correlation potential @xmath17 is given by @xmath18}{\delta { n_\sigma}({{\mathbf r}})}.\ ] ] if the exchange - correlation functional is explicitly orbital - dependent @xmath19 \equiv { e_{xc}}[\ { { \psi_{\sigma i}}\}]$ ] , the corresponding @xmath20 can be obtained by the oep integral equation@xcite @xmath21 { \psi_{\sigma i}}^\ast({{\mathbf r } } ' ) g_{{\sigma i}}({{\mathbf r}}',{{\mathbf r } } ) { \psi_{\sigma i}}({{\mathbf r } } ) \ + \ c.c .
\ = \ 0,\ ] ] where the quantity @xmath22 is given by @xmath23}{\delta { \psi_{\sigma i}}^\ast({{\mathbf r}})}.\ ] ] the kli approximation to the oep is obtained by ignoring certain terms whose average over the ground - state spin density vanishes.@xcite it can be also heuristically derived by approximating @xmath24 , appearing in the green s function in eq .
( [ eq : oep ] ) by a constant @xmath25@xcite @xmath26 in which case eq .
( [ eq : green ] ) reduces to integrals over the wave functions independent of the spectrum .
approximation@xcite can be obtained by simplifying the numerator of eq .
( [ eq : green ] ) further to @xmath27 , thus ignoring @xmath28 term.@xcite although the kli approximation is generally applicable to any orbital dependent exchange - correlation energy functional , we will concentrate on the case of the exact exchange energy functional defined as @xmath29 = -\frac{1}{2 } \sum_{\sigma=\uparrow,\downarrow } \sum_{i , j}^{{n_{\sigma } } } { f_{\sigma i}}{f_{\sigma j}}\int d{{\mathbf r}}\int d{{\mathbf r } } ' \frac{{\psi_{\sigma i}}^{*}({{\mathbf r } } ) { \psi_{\sigma j}}^{*}({{\mathbf r } } ' ) { \psi_{\sigma i}}({{\mathbf r } } ' ) { \psi_{\sigma j}}({{\mathbf
r}})}{|{{\mathbf r}}- { { \mathbf r}}'|},\ ] ] which has been actually used in this work . here note that @xmath30 are ks orbitals rather than hf orbitals . in this case
, @xmath31 is just the hf local exchange potential expression evaluated with ks ( instead of hf ) orbitals @xmath30 @xmath32 where @xmath33 the approximation of eq .
( [ eq : green ] ) leads to the kli - x potential @xmath34 where @xmath35 is the slater exchange potential,@xcite with @xmath36 .
the quantities @xmath37 and @xmath38 are the expectation values of @xmath39 and @xmath31 with respect to the kli - x orbital @xmath40 , @xmath41 the condition @xmath42 is automatically satisfied with the exact exchange energy functional,@xcite so only the remaining @xmath43 constants @xmath44 in eq .
( [ eq : kli ] ) need to be determined by the solution of the following linear equation : @xmath45 where @xmath46 the slater potential @xmath47 in eq .
( [ eq : kli ] ) ensures the correct asymptotic behavior of @xmath48 for large @xmath49 when the exact exchange energy expression is employed.@xcite it is known that @xmath47 alone is too deep,@xcite and is a factor of @xmath50 larger than @xmath51 in the homogeneous - electron - gas limit . the second term in eq .
( [ eq : kli ] ) , which was originally missing in the slater s proposition , enables @xmath52 to correctly reduce to @xmath53 in the homogeneous - electron - gas limit , and to preserve the property of the integer discontinuity .
@xcite also , due to eq .
( [ eq : vx_homo ] ) , the kli - x highest occupied eigenvalues satisfy the following condition @xmath54 where @xmath55 is the hf orbital energy expression evaluated with the ks ( rather than hf ) orbital @xmath56 . at this point
, we would like to emphasize that the kli - x ks scheme yields a local potential , while the hf method gives a nonlocal exchange potential .
it is well known that the hf method gives eigenfunctions and eigenvalues for the unoccupied orbitals which are inappropriate for describing actual excited states , because hf unoccupied orbitals do not have self - interaction corrections .
@xcite this point will be further emphasized and discussed in sec .
[ subsec : eigenvalues ] with specific examples .
all the calculations in this work are based on the finite - difference pseudopotential scheme@xcite which we have previously applied to the study of neutral and charged water clusters.@xcite in this formulation , quantities such as ks orbitals , densities , and potentials are expressed on grids in a rectangular simulation box , and the laplacians or gradients are evaluated by higher - order finite - difference expressions .
the most computationally demanding step in the construction of the kli - x potential is the computation of the @xmath57 pairs @xmath58 in eq.([eq : uxsi ] ) .
since these terms have the same structure as that of the hartree potential with density @xmath59 replaced by @xmath60 , we use the techniques we have developed for the solution of the poisson equation , and solve the discretized poisson - like equations @xmath61 to do so , we first generated boundary values for @xmath62 at six sides of the simulation box using a multipole expansion .
multipole expansions have been performed until the boundary value error @xmath63 was less than @xmath64 .
next , a coarse solution for @xmath62 on the entire simulation grid has been generated by the fast fourier transform method using a lower - order ( 3 points along each direction ) finite - difference expression .
this solution has been subsequently relaxed by the iterative preconditioned conjugate gradient method using a higher - order ( typically 13 points along each direction ) finite - difference method .
conjugate gradient relaxation steps have been performed until the error @xmath65 is decreased to below @xmath64 . for the preconditioning , a smoothing type preconditioner has been used .
we observed that potential mixing rather than density mixing is more appropriate , which should be understandable considering that the kli - x potential is orbital - dependent .
in addition , for the gga calculations , we have constructed gga potentials with the scheme of white and bird.@xcite for other computational details including the implementation of the gga potential , we refer to our other publications .
@xcite to assess the accuracy of our 3d kli - x implementation , we considered atoms and compared lda and kli - x results with those from an accurate 1d radial atomic code .
one test was for an appelbaum - hamann type local pseudopotential@xcite modified to bind 5 electrons @xmath66 , where @xmath67 , @xmath68 , and @xmath69 .
this choice was made to reconcile the limitations of the 3d code that it can deal with only smooth pseudopotentials and the 1d code that it can only produce accurate results for spherical atoms because of the central field approximation . using a grid spacing @xmath70 = 0.4 a.u .
, simulation cell length @xmath71 = 32 a.u . , and finite - difference order 13 , we obtained agreement with the radial code of @xmath72 a.u./atom for both lda and kli - x total energies . for the hydrogen atom
, we have used a local potential derived by giannozzi@xcite and again obtained an agreement of @xmath72 a.u./atom using @xmath70 = 0.4 a.u . for several other atoms considered in the next section ( c , n , and o ) , we generated lda pseudopotentials with the method of troullier and martins @xcite and compared lda total energies and eigenvalues from the two codes . with @xmath70 = 0.25 a.u .
, the 3d code reproduced the 1d atomic total energy with an accuracy of @xmath73 a.u./atom which is sufficient for our purpose . before closing this section
, we would like to comment on the issues involved with the use of pseudopotentials . although the rigorous and consistent procedure would be to employ pseudopotentials generated within the same functional approximations , we used lda generated pseudopotentials to carry out all other functional calculations for two reasons .
first , we believe this procedure will not change the qualitative picture , since structural and electronic properties are rather insensitive on the nature of the exchange - correlation energy functional that is used for the small 1s cores of the first - row atoms considered here.@xcite second , we note ambiguities associated with pseudopotential generation procedures . for example , we observed a small but sharp peak at the atomic center in the pbe gga pseudopoential , which caused a serious convergence problem.@xcite moreover , in the kli - x pseudopotential generation , there is an intrinsic problem of slow decaying tails due to the nonlocal nature of the exact exchange functional that somewhat influences the results.@xcite these anomalous behaviors can be only alleviated by some post - processings which introduce additional arbitrariness in pseudopotenials .
we first examine the structural properties of @xmath74 , @xmath75 , @xmath76 , @xmath77 , and @xmath78 given in table [ tab : diatom_ground ] and table [ tab : polyatom_ground ] .
the binding energies @xmath79 , equilibrium bond lengths @xmath80 , and vibrational frequencies @xmath81 of the diatomic molecules were determined from the total energy versus bond - length curves by fitting to the five parameter ( @xmath82 , @xmath80 , @xmath81 , a , and b ) hulburt - hirschfelder function , @xcite @xmath83,\ ] ] where @xmath84 and @xmath80 and @xmath82 are the equilibrium distance and bonding energy , using the simulated annealing method .
@xcite the geometries of the polyatomic molecules have been optimized by employing hellmann - feynman quantum forces .
a grid spacing @xmath85 a.u . has been used for all the molecular calculations . for the calculation of binding energies ,
the energies of each pseudoatom and molecules have been calculated by the same method and with the same @xmath70 .
we expect this procedure will result in a systematic cancellation of errors , and in addition we may treat the atom without central field approximation .
our lda and gga results are essentially a reproduction of previous studies .
@xcite the lda consistently overestimates the stability of the molecules and the gga typically cures this tendency of overestimation to a large extent .
the hf approximation on the other hand substantially underestimates the binding energy .
@xcite bond lengths calculated by the lda are usually overestimated and one obtains corresponding underestimations of vibrational frequencies for the diatomic molecules .
the hf method gives the opposite behavior .
the kli - x approximation typically gives similar results as hf for the structural properties which are mostly determined by the total energy functional.@xcite it is well known that the hf method gives the wrong sign of the dipole moment of @xmath75 , and the lda corrects this .
again , the kli - x calculation yields the wrong sign of the dipole moment of @xmath75 in agreement with the hf method .
the difference between magnitudes of the kli - x and hf dipole moment of @xmath75 results from the fact that the hf value has been evaluated at the experimental bond length .
our kli - x calculation at the experimental bond length gave @xmath86 debye .
the addition of an lda or gga correlation energy functional to the exact exchange energy gives mixed results : binding energies are increased , hence improved the kli - x values , but bond lengths are decreased and become worse .
so , we can conclude that the two correlation functional approximations do not give systematic improvements over the exact exchange energy functional , which indicates that these approximations are not accurate descriptions of correlation effects in molecules . in fig .
[ fig : h2_vxc ] , we compare the lda - xc and kli - x potentials of @xmath74 to reemphasize the different nature of the exchange(-correlation ) potentials from the two methods .
while the lda - xc potential decays exponentially , the kli - x potential reproduces the correct @xmath48 asymptotic behavior , and this causes marked differences in the eigenvalue spectra .
this @xmath48 decay property lets the high - lying ks unoccupied eigenvalues correctly approach a rydberg series , and is also important for other properties that are sensitive to the outer part of the charge density , such as polarizabilities .
eigenvalues of the highest occupied orbitals of the molecules considered in the previous section at their equilibrium geometries ( tables [ tab : diatom_ground ] and [ tab : polyatom_ground ] ) are listed in table [ tab : ioniz_potential ] . compared with the lda or gga
, the kli - x values agree very well with experimental ionization potentials , and they are very close to hf values .
it is well known that hf gives a good approximation to the ionization potential via koopman s theorem due to the fact that omission of correlation tends to be cancelled by the neglect of relaxation in the `` frozen orbital '' approximation.@xcite on the other hand , in the exact dft , the highest occupied orbital eigenvalue equals the ionization potential without any relaxation correction@xcite as we typically see in our kli - x calculations , hence , in principle dft should yield highest occupied orbital eigenvalues in better agreement with experimental ionization potentials than hf . however , in the usual lda or gga , the highest occupied eigenvalues are in serious errors due to the wrong decay property of exchange - correlation potential exemplified in fig .
[ fig : h2_vxc ] .
the errors are about 0.2 a.u . for the molecules we have studied as shown in table [ tab : ioniz_potential ] . on the other hand , due to eq .
( [ eq : ev_homo ] ) , kli - x should give highest occupied orbital eigenvalues in good agreement with experimental ionization potentials when hf values also agree with experimental ionization potential .
the addition of lda or gga correlation potentials to the kli - x potential usually lowers kli - x highest occupied orbital eigenvalues , but this leads to too large ionization potentials . among the molecules examined in this work ,
@xmath76 is a particularly interesting case which shows the advantage and disadvantage of the kli - x scheme at the same time .
first , note that the hf results are in qualitative disagreement with experiment : it puts the @xmath87 state higher than @xmath88 state as the highest occupied orbital in contrast to the experimental data .
this incorrect ordering of the first two ionization potentials is a well - known example of the breakdown of the hf picture and indicates a large effect of correlation upon the orbitals in @xmath76 .
@xcite on the other hand note that kli - x gives the correct ordering of the two orbitals , which implies that the local nature of the kli - x potential can make a difference even for the occupied - orbital eigenvalue spectrum .
the fact that the lda also gives the correct ordering ( although the value is worse than the kli - x one due to the above - mentioned reasons ) shows that there are cases where the local potential in the ks scheme is superior to the nonlocal hf potential .
the highest occupied state in the kli - x scheme @xmath89 is equal to @xmath90 [ according to eq .
( [ eq : ev_homo ] ) ] which results in the biggest kli - x error among the molecules studied since this also has the largest hf error .
better quantitative agreement with experiment can be obtained by incorporating the correlations , e.g. through many - body perturbation type approach.@xcite in addition to the eigenvalues of the highest occupied orbitals , we have also examined eigenvalue differences between the highest occupied and the lowest unoccupied orbitals , which are compared with experimental lowest triplet and singlet vertical excitation energies in table [ tab : exctn_energy ] .
first , we point out that the hf calculations give positive orbital energies for all the virtual orbitals in the molecules considered here.@xcite this is because unoccupied orbitals in hf do not really correspond to excited states of the system , in which the excited state electron would have been removed from one of the lower states and acted on by @xmath91 remaining electrons,@xcite but rather the bound states ( if any ) of the singly - charged negative ion , in which the extra electron sees the field due to other @xmath92 electrons.@xcite to make this argument more explicit , we follow slater@xcite and rewrite eqs .
( [ eq : uxsi ] ) and ( [ eq : ksji ] ) ( now using the hf orbitals @xmath93 instead of ks orbitals @xmath94 , so @xmath95 is the true hf local exchange potential ) as @xmath96 / \bigr [ { \psi_{\sigma i}}^{hf \ast}({{\mathbf r } } ) { \psi_{\sigma i}}^{hf}({{\mathbf r } } ) \bigr ] } { |{{\mathbf r}}-{{\mathbf r}}'|},\ ] ] and identify @xmath97 as an exchange charge density . because the @xmath93 s are orthonormal
, the exchange charge density integrated over @xmath98 will be minus one for the occupied orbital ( @xmath99 term exists in the summation @xmath100 and @xmath101 ) , but zero for the unoccupied orbital ( @xmath93 is not included in @xmath100 ) .
hence , the self - consistent field in which an unoccupied orbital moves is that of the nucleus and all @xmath92 electrons instead of @xmath91 electrons . on the other hand , in dft , as mentioned in sec .
[ sec : intro ] , differences of ks eigenvalues are well - defined approximations for excitation energies .
@xcite umrigar _
have confirmed that this is the case for several atomic systems by showing that the exact ks eigenvalue differences obtained from ks potentials derived from accurate qmc densities almost always lie between the triplet and singlet experimental excitation energies.@xcite they further claimed that this is because of the fact that , in addition to the well known exact properties of @xmath2 ( sec .
[ sec : intro ] ) , @xmath2 agrees with the quasiparticle amplitude@xcite for @xmath4 not only up to @xmath102 but up to order @xmath103 inclusive . however , in the lda or gga , no bound unoccupied state exists in many cases , due to the shallow nature of the corresponding exchange - correlation potentials .
we can expect that the kli - x is a much better approximation than the lda / gga , since it satisfies eq .
( [ eq : ev_homo ] ) and has the correct asymptotic @xmath48 behavior .
our kli - x results indeed always give negative virtual orbital eigenvalues for the studied molecules and also show good agreement with experimental values .
the single exception is @xmath77 in which the difference between the highest occupied and lowest unoccupied orbital eigenvalues is bigger than the experimental triplet and singlet excitation energies by @xmath104 a.u .
further detailed agreements with different multiplet states can be obtained by employing a more involved theory such as dft perturbation theory@xcite or time - dependent dft.@xcite we close this section by reconsidering the meaning of eigenvalues in the oep - x ( or kli - x ) and hf schemes to make the differences between the two be clear . in the fractional occupation number formalism , kli have shown that @xmath105 where @xmath106 is the hf orbital energy expression for the orbital @xmath40 evaluated using the oep - x orbitals @xmath94 .
@xcite we note that this can be generalized to the correlated oep case .
@xcite equation ( [ eq : janak ] ) can be rearranged to the form in which the relationship between the two methods is transparent : @xmath107 note that from the condition @xmath108 , satisfied by both oep - x and kli - x [ eq . ( [ eq : vx_homo ] ) ] , eq .
( [ eq : ev_homo ] ) is reproduced from eq .
( [ eq : oep_hf ] ) for the @xmath109 state .
hence , the eigenvalue difference between the highest occupied orbital state @xmath56 and the unoccupied orbital state @xmath110 is given by @xmath111 assuming that the differences between the hf and oep - x orbitals are negligible , we can see that hf orbital eigenvalue differences are much bigger than oep - x values , because the @xmath112 term is small due to the property of the exchange charge density [ eq . ( [ eq : rhox ] ) ] for the unoccupied orbital as discussed above , while the @xmath113 term is clearly quite large .
in this work , we have studied structural and excitation properties of small molecules using the kli approximation to the oep with the exact exchange energy functional only ( kli - x ) and augmented by lda and gga correlation functionals . for structural properties ,
the kli - x gave comparable results to those of the hf method .
for excitation properties , the kli - x results in good eigenvalue spectra for _ both the highest occupied and unoccupied _ orbitals , since it has the correct asymptotic large @xmath49 behavior of @xmath2 unlike the lda or gga .
especially , we find that unoccupied orbital eigenvalues are better described by kli - x than by hf , which illustrates the advantage of the ks scheme in general
. this work has been supported by the national science foundation under grant no .
dmr 98 - 0273 ( y .- h .
k. and r. m. m. ) and by the office of naval research under grant no .
n0014 - 98 - 1 - 0594 ( m. s. ) .
we acknowledge profs .
j. b. krieger and j. p. perdew for critical readings , prof . e. k. u. gross for helpful discussions , prof .
j. soler for providing us with his gga exchange - correlation routine , and dr .
d. sanchez - portal for performing gga calculations for comparisons .
t. grabo , t. kreibich , s. kurth , and e. k. u. gross , in _ strong coulomb correlations in electronic structure : beyond the local density approximation _ , edited by v. i. anisimov ( gordon & breach , tokyo , 1998 ) , and references therein .
ldddddddddd & & & + & @xmath79 & @xmath80 & @xmath81 & @xmath79 & @xmath80 & @xmath81 & @xmath114 & @xmath79 & @xmath80 & @xmath81 + lda - xc & 0.184 & 1.44 & 4228 & 0.466 & 2.13 & 2193 & 0.248 & 0.396 & 2.09 & 2385 + gga - xc & 0.168 & 1.41 & 4228 & 0.424 & 2.12 & 2166 & 0.265 & 0.368 & 2.09 & 2383 + kli - x & 0.136 & 1.39 & 4647 & 0.287 & 2.06 & 2445 & -0.163 & 0.176 & 2.01 & 2634 + + lda - c & 0.185 & 1.36 & 4765 & 0.342 & 2.05 & 2595 & -0.184 & 0.271 & 2.00 & 2710 + + gga - c & 0.166 & 1.38 & 4734 & 0.361 & 2.06 & 2641 & -0.170 & 0.273 & 2.00 & 2805 + hf - x & 0.134 & 1.39 & 4582 & 0.293 & 2.08 & 2431 & -0.279 & 0.195 & 2.01 & 2730 + exp . &
0.174 & 1.40 & 4400 & 0.414 & 2.13 & 2170 & 0.123 & 0.364 & 2.07 & 2358 + lddddd & & + & @xmath115 & @xmath116 & @xmath115 & @xmath117 & @xmath118 + lda - xc & 0.743 & 2.07 & 0.420 & 1.82 & 104.2 + gga - xc & 0.681 & 2.06 & 0.370 & 1.81 & 104.5 + kli - x & 0.536 & 2.05 & 0.257 & 1.76 & 106.0 + +
lda - c & 0.678 & 2.04 & 0.344 & 1.76 & 106.0 + + gga - c & 0.672 & 2.03 & 0.339 & 1.77 & 105.3 + hf - x & 0.522 & 2.05 & 0.258 & 1.78 & 106.1 + exp .
& 0.668 & 2.05 & 0.374 & 1.81 & 104.5 + lddddd & @xmath119 & @xmath120 & @xmath121 & @xmath122 & @xmath123 + lda - xc & 0.37 & 0.34 & 0.38/0.44 & 0.26 & 0.34 + gga - xc & 0.38 & 0.34 & 0.38/0.43 & 0.27 & 0.34 + kli - x & 0.60 & 0.55 & 0.64/0.69 & 0.50 & 0.54 + kli - x + lda - c & 0.60 & 0.61 & 0.69/0.75 & 0.56 & 0.60 + kli - x + gga - c & 0.58 & 0.59 & 0.67/0.74 & 0.55 & 0.58 + hf - x & 0.60 & 0.55 & 0.64/0.62 & 0.51 & 0.55 + exp .
& 0.58 & 0.58 & 0.57/0.62 & 0.46 & 0.53 + .differences between highest occupied and lowest unoccupied orbital states eigenvalues of @xmath74 , @xmath75 , @xmath76 , @xmath77 , and @xmath78 at their respective equilibrium geometries given in table i and ii .
a @xmath124 mark indicates that the lowest unoccupied orbital state is in the continuum and hence unstable .
experimental values are vertical excitation energies to final triplet and singlet states shown in parentheses .
energies are in a.u . [ cols="<,^,^,^,^,^ " , ] | we report density - functional studies of several small molecules ( @xmath0 , and @xmath1 ) within the krieger - li - iafrate ( kli ) approximation to the exact kohn - sham local exchange potential , using a three - dimensional real - space finite - difference pseudopotential method . it is found that exchange - only kli leads to markedly improved eigenvalue spectra compared to those obtained within the standard local - density approximation ( lda ) , the generalized gradient approximation ( gga ) , and the hartree - fock ( hf ) method . for structural properties , exchange - only kli results are close to the corresponding hf values .
we find that the addition of lda or gga correlation energy functionals to the kli exact exchange energy functional does not lead to systematic improvements . |
it is broadly accepted now that active galactic nuclei are powered by accretion onto a supermassive black hole and that the flow is a multi - phase ( at least two - phase ) medium .
a hot optically thin phase is responsible for the x - ray and @xmath6-ray emission while a cool phase of considerable optical depth is responsible for the thermal optical / uv emission as well as for the preprocessing of a significant fraction of the radiation produced by the hot phase ( for a review , see e.g. mushotzky et al .
there is mounting evidence based on the shape of @xmath7 line profile as well as on the variability in soft and hard x - rays , that the release of the major part of the x - ray energy , of the thermal euv / soft x - ray emission and the reprocessing of x - ray by cold material takes place in a very compact region , of order of @xmath8 schwarzschild radii and the cold matter is approximately in keplerian motion ( see e.g. nandra et al .
1997a ) .
however , the geometrical arrangement of the two phases is still under debate .
the three most viable models are : a disk with a hot corona , very optically thick clouds embedded in a hot medium , and moderately optically thick clouds also coexisting with a hot medium . in order to differentiate between these models detailed studies of all of them are necessary .
the disk model is the best studied so far ( e.g. ross & fabian 1993 , stern et al .
1995 , sincell & krolik 1997 ) .
optically thick clouds were studied for example by sivron & tsuruta ( 1993 ; see also the references therein ) .
clouds of moderate optical depth were broadly advertised as a free - free emission model ( e.g. antonucci & barvainis 1988 ) .
however , clouds which are optically thin for electron scattering are not a viable model for an agn ( barvainis 1993 , collin - souffrin et al .
1996 , kuncic et al . 1997
; see also lightman & white 1988 ) . on the other hand clouds , which are optically thick for electron scattering but optically thin for absorption in the uv and moderately optically thick in the optical band ( hereafter otcm model ) ,
can reproduce the required fraction of the x - ray emission and roughly produce the correct multi - wavelength spectra , as well as provide enough material to support the nuclear activity ( collin - souffrin et al .
1996 ; hereafter paper i ) . as the model may have an advantage over clouds optically thick for absorption it deserves a detailed study of its observational consequences .
in this paper we concentrate on the comparison of the predictions of the otcm with the broad band mean spectrum for radio quiet quasars determined by laor et al .
( 1997 ) . as direct analysis of the x - ray emission
does not give any reliable answer to the question of the nature of the x - ray and @xmath6-ray emission due to the still poor determination of the spectra of radio quiet agn in the @xmath9mev range ( see e.g. gondek et al .
1997 for combined seyfert 1 spectra from gro ) we analyze three representative cases of the sources of emission external to the clouds .
as is customary and convenient , we call this incident radiation for the clouds the `` primary '' radiation although this is unjustified when the x - ray emission is itself the emission of the clouds comptonized by a hot medium .
the overall outline of the cloud model and of the radiation transfer within a cloud was presented in paper i so we only summarize here the basic assumptions .
we consider a distribution of clouds surrounding a hot plasma which is a source of incident ( primary ) radiation for the clouds .
sides of the clouds exposed to the central source are bright while the unexposed sides are relatively dark and cold because of the considerable optical depth of the clouds .
the schematic geometry is shown in figs . 1 and 2 .
all clouds are assumed to have the same density and to be exposed to the same incident flux since any assumption about the radial distribution of the cloud properties would be completely arbitrary without an underlying dynamical model .
however , we carefully calculate the radiative transfer in the clouds . the resulting broad band spectra from the ir to x - rays are combinations of the radiation transmitted , emitted and scattered by clouds as well as a fraction of the primary radiation .
the relative weight of these components depends on the covering factor and the size of the source of primary emission . the radiation transfer code used in this paper (
dumont & collin - souffrin , in preparation ) has been improved in comparison with the one used in paper i. the ionization of all hydrogenic ions ( not only hydrogen itself ) proceeds from all 5 levels and interlocking ( subordinate ) lines are included .
these modifications influence to some extent the line emission and the lyman edge , particularly due to its coincidence with the second level of he@xmath10 . since the radiative transfer code of dumont & collin - souffrin is constrained in the present version to photons with energy below 24 kev (
although compton heating by harder x - rays is included ) we calculate the shape of the reflected spectrum above 24 kev using the method given by lightman & white ( 1988 ) .
we use the opacities for the neutral gas since the ionization level is unimportant for these high energy photons .
the continuity between the two reflected spectra was achieved by adjusting the opacity parameter c of lightman & white ( 1988 ) in their eq .
the value of this parameter was in all cases about @xmath11 , by a factor of two lower than the value suggested by george ( private communication ) for a neutral medium and elemental abundances used in george & fabian ( 1991 ) .
= 80 mm = 80 mm since , at this stage , we do not introduce a full dynamical model describing the cloud formation and disruption and their motion the radial distribution of the clouds themselves and of their properties is arbitrary .
therefore , in order to keep the parameterization as simple as possible we assume that all clouds have the same ( constant throughout a cloud ) density @xmath12 and column density @xmath13 . since clouds optically thin for electron scattering were ruled out as a model for optical / uv / soft x - ray continuum we study clouds with @xmath14 , usually concentrating on the @xmath15 case .
higher @xmath13 strongly suppress the emission from the unilluminated side of the cloud so only reflection is seen .
the clouds are opaque for the x - ray radiation so even if some clouds are in our line of sight towards the primary x - ray source this would not lead to the presence of the transmitted , strongly absorbed spectral component .
the primary emission intercepted by the clouds is partially reflected and partially reemitted by the dark side of the cloud ( i.e. the side opposite to the central source ) in the form of thermal emission . the distribution of the clouds is described by the covering factor @xmath16 with respect to the x - ray source
. if @xmath16 is close to 1 , the distribution necessarily is almost spherical .
if @xmath16 is smaller ( e.g. 0.5 ) then the distribution may either be still spherical , or rather constrained to the symmetry plane .
these two cases are different if doppler effects connected with the motion of the clouds are taken into account since plane confinement would result in broader features for the same velocity field .
the nature of the primary emission is still not known . in the past , non - thermal models were favored . in such models
a ( usually power law ) distribution of relativistic electrons produces the radiation either by direct synchrotron emission or by comptonization of soft photons .
these soft photons were either their own synchrotron photons ( ssc ) or came from external source like an accretion disk . in these models the pair creation process was frequently important , particularly for the shape of the spectrum above @xmath17 .
these models are still very popular in the case of radio loud agn . however , in the studies of radio quiet agn the attention recently shifted towards thermal models . in these models
the hot plasma is thermal , at a temperature of order of a few hundreds of kev , and it produces the radiation through comptonization of soft external photons . in these models
the effect of pair creation is usually negligible .
these two basic families of models differ also in the extension of the primary component into the low energy range . in non - thermal models the primary emission may ( but does not necessarily have to ) extend down to the far - ir / mm band .
it was an attractive possibility in the late 70s and 80s when ir emission was generally thought to be dominated by non - thermal processes .
evidence of a significant contribution of dust to the ir emission as well as some arguments based on variability ( e.g. done et al .
1990 for ngc 4051 ) diminished the popularity of that view .
however , a single underlying ir / x - ray continuum is still sometimes advocated on the basis of observational arguments ( e.g. walter & fink 1993 , loska & czerny 1990 , fiore et al .
thermal models , on the other hand , predict a contribution of the hard component only above the energy of the seed photons for comptonization , i.e. mostly starting from the uv band . as a result , in their computations of the overall continua , various authors used different specific assumptions about the shapes of the incident spectra for reprocessing .
lightman & white ( 1988 ) consider a non - thermal model with an energy index 0.7 and a spectrum extending from 1 ev to 3 mev , guilbert & rees ( 1988 ) have the primary non - thermal spectrum with energy index 0.5 ( they also include the effect of pair creation on the shape of the primary ) .
sivron & tsuruta ( 1993 ) assume that the initial synchrotron radiation with an energy index 1 extending from 0.1 ev to 1 mev is filtered by cfr cloudlets ( celotti et al .
this special , very compact , population of cloudlets , with densities @xmath18@xmath19 , hydrogen column @xmath20@xmath21 and covering factor 1 , absorbs the ir emission ( through the free - free process ) but it is transparent at the optical band and above .
the absorbed energy is reemitted in the form of free - free emission with the adopted value of the temperature @xmath22 k. such a preprocessed spectrum serves as incident spectrum for the main clouds occupying the space between @xmath23 and @xmath24 .
in order to study the influence of the shape of the incident flux on the resulting spectrum predicted by the model we choose three representative shapes of the primary continuum justified by general theoretical arguments .
they are named models ( a ) , ( b ) and ( c ) , and they are shown in fig .
this incident radiation resembles closely the incident radiation adopted in our initial study ( paper i ) as well as by kuncic et al .
we assumed a low frequency cut - off at 0.1ev as in paper i. in order to reproduce well the observed spectrum ( according the present knowledge ) we assume an energy index either 0.9 , after the classical paper on seyfert 1 galaxies ( pounds et al .
1990 ) or steeper since in the case of quasars the situation is not clear : laor et al .
( 1997 ) give a slope of 1 for radio quiet objects while williams et al .
( 1992 ) give 0.92 when flat spectrum objects are excluded .
we assume the values of the cutoff energy 100 kev and 280 kev since this last value seems to be suggested by gro data for seyfert 1 galaxies ( grandi et al .
1998 for mcg8 - 11 - 11 ; madejski et al .
1995 for ic 44329a , gondek et al .
1997 for a composite spectrum ) .
no direct data constraints for cut - off energy in radio quiet quasars are available .
such a model corresponds to the presence of a relatively strong magnetic field within the hot plasma , such that the energy density of the magnetic field is larger than the energy density of the soft photons available due to the presence of cool clouds .
models of that type were studied e.g. by maraschi et al .
( 1982 ) , with lower frequency cut - off @xmath25 hz determined by self absorption .
the hypothesis of the existence of such an ir / x - ray power law was observationally tested in the case of the source ngc 4051 ( done et al .
1990 ) and its existence was not confirmed since strong x - ray variability in this source was not accompanied by any optical variability . on the other hand ngc 4051
is not a typical example of an agn and its optical emission may be strongly dominated by starlight . a number of other sources like ngc 5548 ( korista et al .
1995 ) and ngc 4151 ( edelson et al . 1997 ) show coherent day to day variations in optical , uv and x - ray band , with unmeasurable delays smaller than several hours .
we do not analyze the conditions for the production of such a primary emission in the present paper .
we simply assume the parametrization by a single power law and we fix the low energy cut - off at 0.1 ev , as in paper i. the input model parameters are : the ionization parameter , @xmath26 ( where @xmath27 is the bolometric luminosity ) , the cloud density , @xmath12 , and its hydrogen density column , @xmath13 , ( or cloud size ) .
the observational appearance of the system is further parameterized by the relative contribution of the primary ( or incident ) radiation and the emission from dark sides of clouds with respect to the reflected component . for purely random cloud distribution both parameters
are uniquely determined by the covering factor , @xmath28 ( see eq .
2 ) . our second model is a power law characterized by the same energy index and high energy cut - off as before but the adopted low frequency cut - off is set at 30 ev i.e. in the uv band ( @xmath29 hz ) .
this simple model is supposed to represent a situation when the gas is not simply in equipartition with the magnetic field but dominates the behaviour of the matter ( rees 1987 ) . as the expected magnetic field is in this case two orders of magnitude or more higher than in the previous case ( @xmath30 gauss ,
see celotti et al .
1992 ) the self - absorption frequency is proportionally higher .
additionally , in such a strong magnetic field a new family of cloudlets , or filaments , may form in the innermost part of the flow .
such cloudlets are opaque to photons below , again , @xmath31 hz due to their high density and free - free absorption so they filter the primary emission before this emission may reach the main clouds outside ( sivron & tsuruta 1993 ) .
therefore the incident radiation in the case of magnetic field dominating innermost flow would not extend beyond optical / uv . absorption by cloudlets leads additionally to reemission of the absorbed radiation in the form of a black body radiation .
however , as we can not predict the temperature of such radiation and the covering factor by the inner cloudlets we simply neglect this component in the present consideration .
the input model parameters are the same as in the previous section .
this model is based on the assumption that amplification of the magnetic field within the flow is not efficient .
therefore , clouds are the only source of the soft photons .
we also assume this time that the hot medium is thermal , i.e. basically characterized by the optical depth and the temperature .
hard x - ray emission in this model forms by compton upscatter of a fraction of the soft photons from the clouds while soft photons result from interception of a fraction of hard x - ray emission by clouds . in order to obtain the appropriate shape of the hard x - ray continuum
as described above we have to adopt a temperature of the hot medium equal to @xmath32 k and an optical depth @xmath33 . adopting higher ( lower ) value for the temperature would require a lower ( higher ) value of the optical depth of the hot medium .
the compton parameter @xmath34 for such a plasma is 1.1 .
the fraction of the soft photons intercepted by the hot medium results from the model ( see paper i ) and is approximately given by @xmath35 where @xmath36 is the fraction of photons reflected or reemitted by the illuminated side of a cloud , determined by the physical conditions in the gas .
our model is therefore more flexible than the original version of the corona model ( haardt & maraschi 1991 ) which requires @xmath37 equal to 0.5 and @xmath38 , thus practically fixing the value of the compton parameter @xmath34 .
recently developed clumpy corona models are based on an additional covering factor for the hot medium which is equivalent to our @xmath39 ratio from the point of view of the efficiency of the comptonization but the two models nevertheless differ with respect to overall geometry and its ( angular - dependent ) appearance .
the input parameters of the model are , as usual , the ionization parameter , the cloud density , its hydrogen density column ( or size ) , and the weights of the primary and dark side components .
= 80 mm however , the computations of the model require an iterative procedure to achieve a self - consistent solution for the soft and hard emission . in order to avoid an
arbitrarily adopted spectral shape in the optical band we start with a mechanically heated cloud with temperature @xmath40 k. next we compute the comptonized spectrum which results from the interception of soft photons by hot plasma with the parameters given above . for that purpose
we use the method of czerny & zbyszewska ( 1991 ) , appropriate for a thin plasma .
this radiation now serves as incident radiation .
next we calculate the * reflected * ( not emitted ) component as the cloud faces the hot medium clearly with its hot hemisphere .
the comptonization of this reflected component gives a second approximation to the hard x - ray emission spectrum and the model is iterated until it converges .
the normalization is fixed by the adopted value of the ionization parameter and the ratio of the comptonized reflected radiation to renormalized radiation gives us accurately the fraction of soft photons intercepted by the hot gas , or the relative extension of the two regions , roughly estimated by eq .
the observed spectrum of an agn within the frame of the otcm is in general a sum of the primary , a reflected component and an emission from the dark sides of the clouds , with the relative weight of the last two given by the covering factor ( see paper i ) .
we make first a general qualitative discussion without adopting any specific value of this parameter ( sect .
. more detailed discussion , however , clearly involves the covering factor as additional parameter of the model ( sect .
3.2 and 3.3 ) .
= 80 mm = 80 mm the clouds are parameterized by the density , @xmath12 , and the hydrogen density column , @xmath13 .
the reflected radiation , i.e. the radiation of the bright illuminated sides of the clouds is not very sensitive to the column density as long as the total optical depth of a cloud is high and in this paper we deal with such a clouds . in fig .
4 we show two examples of the reflected spectra for two values of the column density ( for model a ) .
we see that the spectrum is softer in the optical band and slightly harder in the soft x - rays .
the column density is important , however , for determination of the emission of the dark sides of the clouds ( see fig .
larger column density leads to almost a black body type of emission although atomic features are clearly visible while lower column density is characterized by very strong atomic features superimposed on the continuum .
the comparison with optically thin clouds was already discussed in paper i. examples of the spectra calculated for three values of the ionization parameter @xmath0 ( 300 , 1000 and 3000 ) were shown in paper i. the dependence on the density of the clouds is weak since we parameterize the incident radiation not by flux but by the ionization parameter @xmath0 .
clouds with lower density ( @xmath41 @xmath19 ) are characterized by slightly bigger hydrogen and helium edges .
clouds with larger densities ( @xmath42 @xmath19 ) are slightly optically thicker in the optical range thus leading to lower emission in those wavelengths .
we tested two values of the high energy extension of the incident spectrum : 100 kev and 280 kev .
the effective change of the bolometric luminosity required to preserve the value of the ionization parameter is practically negligible .
however , high energy photons penetrate more easily deeper layers of the clouds so the radiation is thermalized more effectively and the uv part of the reflected spectrum as well as the dark sides of clouds are slightly brighter .
more important role for the shape of the reflected spectrum is played by the slope of the incident radiation . in fig .
6 we show two examples of model a ( reflected spectrum ) calculated for the energy index @xmath43 equal to 0.9 and 1.1
. steeper spectrum of the incident radiation results in steeper ( i.e. softer ) spectrum both in the optical band and in the soft x - rays .
= 80 mm in order to show the dependence of the otcm model on the assumption about the primary emission we present examples of solutions in which identical distributions of clouds are exposed to primary emission with the same ionization parameter but different spectral properties .
the shapes of the incident spectra were shown in fig .
3 . in the ir / optical band model ( a ) is different from the other two .
if the covering factor of the source is not close to 1 the primary emission directly contributes to this spectral band and may strongly modify the optical slope of the resulting spectrum ( see also section 3.2 ) making it softer .
the visibility of the primary emission in that energy band also affects the variability since in this case a fraction of the optical emission is not expected to be delayed with respect to the x - ray band . in the euv / x - ray band model ( c ) is significantly different from the other two models .
power law models are featureless while the comptonization of the radiation emitted by the clouds preserves traces of atomic features if the optical depth of the hot medium is lower than @xmath44 .
therefore in synchrotron models all line emission ( in particular , fe @xmath2 line ) has to be delayed with respect to the power law while in the last case the clear division of the x - ray emission into primary and reflected component is difficult and a fraction of the line flux does not have to be delayed with respect to power law .
= 80 mm the extension of the primary emission into gamma band simply reflects the adopted assumptions .
the spectra reflected by the clouds are shown in fig.7 .
the emission lines are included ; they are plotted adopting a spectral resolution @xmath45 equal to 30 .
the properties of the three spectra are influenced by the choice of the primary emission mechanism .
the model ( a ) presented here differs with respect to the model given in fig .
9a of paper i mostly because of the change in the shape of the incident radiation ( slope of 0.9 instead of 1.0 , and the high energy cutoff ( 280 kev instead of 100 kev ) .
the new spectrum is generally harder because the incident radiation is harder and elastically scattered photons constitute a significant fraction of the reflected spectrum .
the improvement of the physical input into the code resulted in slightly stronger emission lines and a more shallow lyman edge .
the optical / uv slope of the reflected component is significantly flatter in the case of model ( a ) than in the other two .
however , the traces of a balmer edge and of a significant lyman edge can be seen in all these spectra .
the soft x - ray part of the reflected spectrum looks similar in all three models as it is mostly determined by the adopted value of the ionization parameter .
there is considerable emission below 2 kev with an approximate photon index 2.5 between 0.2 and 2 kev .
the strong fe @xmath2 line is also characteristic for all three spectra , but the @xmath2 edge is the weakest for model ( a ) and the most profound for model ( c ) .
the broad band spectral index @xmath46 , measured customary between 2500 and 2 kev , for the reflected component itself is equal to 1.09 for model ( a ) , 0.94 for model ( b ) and 1.14 for model ( c ) .
the high frequency part of the spectrum is again similar in all three cases which is due to the fact that the shape of the reflection component at high frequencies is simply determined by the klein - nishina cross - section for scattering .
= 80 mm an example of radiation spectra emitted by the dark sides of the clouds is shown in fig .
the figure shows case ( c ) but the other two spectra are also characterized by an extremely large lyman discontinuity in emission unless an extra heating of the dark sides of the clouds is allowed ( dotted line ) . for these spectra
the @xmath46 index is high in all cases since the contribution of the dark side of the clouds to the spectrum at 2 kev is negligible .
therefore any contribution of the dark sides of the clouds to the resulting spectra would give steeper @xmath46 than pure reflection spectra .
any contribution from primary emission would work in the opposite direction , and additionally it would modify the soft x - ray slope .
= 80 mm we show an expanded fragment of the spectrum calculated from model ( a ) in fig .
we consider two extreme cases interesting from the observational point of view .
the lower curve shows a pure reflection component , i.e. corresponds to the case when the primary emission is not seen directly because it comes from a compact source and is shadowed by one of the clouds .
the upper curve was obtained assuming that half of the primary source emission reaches the observer . in both cases
we see the characteristic atomic features : noticeable balmer edge , lyman edge and weak @xmath47 line .
equally profound spectral features are seen in models ( b ) and ( c ) since the spectral shape of the incident radiation has little effect .
the comptonization within the clouds is not expected to remove these features for the clouds of the adopted properties and @xmath48 ( paper i ) .
the clear difference between model ( a ) and models ( b ) and ( c ) is in the slope of this part of the spectrum .
we compare the models with the mean spectrum for radio quiet quasars derived recently by laor et al .
( 1997 ) . our case ( a ) nicely coincides with the data in the optical part ( if primary and the dark sides of the clouds are not seen ) but is far too bright in the uv which means that the temperature of the clouds is too high . in the case of model ( b ) and
( c ) the reflection spectra are definitely too hard in the optical band and any contribution from the primary emission does not change that . the comparison with the mean quasar spectrum obtained by francis et al .
( 1991 ) is more favorable for model ( a ) and in that case a minor contribution from the primary emission would even be allowed .
however , models ( b ) and ( c ) are still too hard to match observations for the ionization parameter @xmath48 .
any contribution from the emitted spectra would give still harder spectra , in contradiction with the data .
the choice of lower value of the ionization parameter @xmath0 results in a lower value for the cloud temperature which in principle helps to reconcile the models ( b ) and ( c ) with the data .
we computed the spectra assuming @xmath49 .
however , in that case the spectral features ( balmer edge in absorption and lyman edge in emission ) are strong , giving a change in the continuum by 50 % and 100 % , respectively , in the case of a reflection component and even more ( 100 % and 500 % respectively ) in the case of emission from the dark sides of the clouds .
= 80 mm since the spectrum emitted by the dark sides does not contribute to the soft x - rays we can restrict our study of that band to the reflected component and the contribution from the primary emission .
we show an expanded version of model ( a ) in fig .
10 , in two versions : pure reflection spectrum and reflection plus half of the primary emission .
we assumed @xmath50 .
the overall shape of the spectrum looks like a power law .
when pure reflection is considered the slope of this power law is similar to the mean spectrum of radio quiet quasars ( laor et al .
if a significant contribution from the primary is allowed the slope is too flat .
models ( b ) and ( c ) are generally similar in this spectral band so they are also an adequate description of the data .
precise values of the slope between 0.2 kev and 2 kev for the reflected component in the three models are equal to 1.48 , 1.35 and 1.57 , respectively .
the slope in the laor et al .
( 1997 ) sample is 1.6 so model ( c ) is the best representation of the data . even in the last model
there is no place for a significant contribution from the direct primary emission since the slope would be too flat .
we see a number of quite strong emission lines , with oviii at 0.653 kev being the most prominent .
complex spectral features are observed in a number of sources both in absorption and in emission , and they are usually explained by warm absorbers , i.e. optically thin clouds with a column density @xmath51 and located at a distance of several light weeks ( e.g. otani et al .
the same features are seen in models ( b ) and ( c ) . neither the mean quasar spectrum of laor et al .
( 1997 ) nor observations of sigle quasars ( e.g. leach et al .
1995 for 3c 273 ) indicate the presence of such features .
a higher value of the ionization parameter would give too flat slopes for all three models .
lower values of the ionization parameter give systematically steeper slopes . model ( a ) calculated for @xmath52 requires a contribution of the primary emission with a weight of 0.2 in order to reproduce the observed slope . for model ( c ) the allowed contribution of the primary is always slightly higher for a given value of the ionization parameter , and for @xmath53 the required weight of the primary is @xmath54 .
the line emission is much stronger in that case . in any case
, the otcm model for quasars explains the soft x - ray emission as a reflection from a partially ionized gas .
the same mechanism was suggested to produce weak soft x - ray excesses in a number of seyfert 1 galaxies ( czerny & .
zycki 1994 ) .
however , in their case there were some problems to achieve the required ionization state of the gas within the frame of the adopted model .
no such problems are encountered here .
= 80 mm = 80 mm = 80 mm so far we discussed the comparison of the models with the mean quasar spectrum in the optical / uv band and in the soft x - ray band separately , i.e. with an arbitrary normalization in both bands .
the comparison of the data with the model in the entire optical / uv / soft x - ray band is shown in fig .
11 ( model c , pure reflection ) .
if the normalization is adjusted to soft x - rays we notice that the observed flattening towards high x - rays ( at @xmath55 kev ) is reproduced by the models .
the @xmath46 index of the mean quasar spectrum of laor et al .
( 1997 ) is equal to 1.46 while in the reflected spectra shown in fig .
7 it is equal to 1.09 , 0.94 and 1.14 , respectively .
however , a contribution of the emission of the dark sides of the clouds helps to fill the gap .
since the temperature of the dark side of a cloud depends significantly on the value of the ionization parameter @xmath0 we can adjust its value to model the uv / euv spectrum ( apart from the spectral features present in the model and absent in the data ) . in the case of model ( a ) the best value of the ionization parameter is @xmath56 and an example of the resulting spectrum
is shown in fig .
even the slight bend below @xmath570.2 kev is reproduced by the models .
the contribution of the dark sides of the clouds required to provide the flux in the optical band is rather moderate and the entire spectrum is mostly dominated by reflection . in the case of model ( c ) the best value of the ionization parameter is slightly higher , @xmath58 ( see fig .
however , there is a large deficit of emission in the optical band which shows that a number of additional , cooler clouds should be included in the model , i.e. a single cloud population located at a given distance ( i.e. parametrized by a single value of ionization parameter ) is not an adequate description of the data in the case model ( c ) .
the spectral features in the optical / uv band are quite large in that case .
the kinematic broadening of these features due to the motion of the clouds may slightly reduce and broaden those features ( see fig . 15 for this effect in the soft x - ray band ) but will not really remove them and we intend to devote a special paper to that problem since it is one of the major issues in all realistic models , including accretion disks .
the high energy cut - off of the spectrum is not well constrained observationally in the case of radio quiet quasars .
therefore , it is difficult to formulate any preferences for any of the discussed models on the basis of this spectral band .
= 80 mm the intensity of the iron @xmath2 line given by the model is clearly large , in contradiction with observations of quasars , since a significant contribution from the primary emission is not allowed . for @xmath59
the ew is @xmath60 ev if the direct primary emission is not seen ( lower curve in fig .
14 ) but it is reduced to @xmath61 ev if the contribution with a weight 0.5 would be allowed ( upper curve in fig .
14 ) . for lower @xmath0
it is still larger , and equals @xmath62 kev and @xmath63 ev , respectively .
this is not surprising if the reflected spectrum dominates since the line intensity expected in such case is large ( e.g. .
zycki & czerny 1994 ) .
observations of moderately bright quasars ( nandra et al .
1997b ) give the equivalent width of the iron @xmath2 line not higher than 300 ev , with a clear trend for a decrease of this value with an increase of a quasar luminosity .
we used our otcm model from paper i in order to analyze the requirements which result from the comparison of the three variants of the model to laor et al .
( 1997 ) mean spectra of radio quiet quasars .
the advantage of concentrating on quasars was in a reliable description of their broad band spectra from optical band to hard x - rays .
model ( a ) reproduces well the entire broad band spectrum of quasars , including the optical slope of the data . on the other hand , present computations carried for that model allow only a minor contribution from the primary emission which results in a too high equivalent width of the iron @xmath2 line and a need of some special geometry for cloud distribution in order to explain this level of primary emission ( see section 4.3 )
. model ( b ) is the least attractive since it displays the same problems as model ( a ) with respect to the level of primary emission and it does not explain the optical slope of quasars .
model ( c ) is also too steep ( i.e. too hard ) in the optical band in comparison with the mean quasar spectrum although it models well the uv / x - ray band .
this may mean that model ( c ) well represents well the innermost part of the accretion flow but it has to be supplemented by the presence of more distant clouds and/or an outer accretion disk .
however , this model also display the problem of too high equivalent width of the iron @xmath2 line since it again allows only a marginal contribution of the direct primary emission .
the modelling of the overall shape of the spectrum was very sensitive to the choice of the ionization parameter @xmath0 since this parameter determines the cloud temperature and the emission from their dark sides which dominate the optical / uv part of the spectrum . our analysis based on the overall spectral shape clearly favored an @xmath0 of order of 300 - 500 to fit the mean quasar spectrum of laor et al .
the value @xmath48 gave a cloud temperature too high to fill the gap in the optical band between the model predictions and the observed spectra if the soft x - ray band was modelled correctly .
it also left no room for any presence of the primary emission since the soft x - ray slope of the reflected component was just marginally steep enough for model ( c ) to match the data .
we can compare this result with more detailed spectroscopic constraints .
the most direct estimate of the ionization level of the cool gas in quasars comes from the analysis of the position of the @xmath64 line .
asca results show that for moderately bright quasars the line is at 6.4 - 6.5 kev , while for bright quasars it is at 6.57 kev ( nandra et al .
this suggests that the ionization parameter for bright quasars is of the order of a few hundreds ( e.g. . zycki & czerny 1994 ) .
in fact , the flux averaged position of the iron line in our computations is 6.82 kev for @xmath50 , model ( a ) , 6.43 kev for @xmath49 , and 6.61 kev for @xmath65 , model ( c ) .
therefore , also from this point of view , all intermediate value of @xmath0 better corresponds to the observations . in the case of purely random distribution of clouds with a covering factor @xmath66 and averaged observing angle the relative contribution of the primary emission should also be determined by the covering factor , thus giving a formula appropriate for modelling a mean spectrum : @xmath67 where @xmath68 is the amplification factor taking into account the multiple scattering @xmath69 where a is the effective albedo approximately equal 0.85 ( see paper i ) .
this formula gives the maximum efficiency since it assumes that the source of primary emission does not intercept repeatedly scattered radiation .
other quantities have the same meaning as in paper i , i.e. @xmath70 is the primary ( incident ) radiation , @xmath71 is the emission of the dark sides of the clouds and @xmath72 is the radiation reflected / reemitted by the bright sides of the clouds . in the case of model(c )
the amplification factor is decreased by the presence of hot plasma intercepting a fraction of the photons and the appropriate formula is given by @xmath73 such an approach reduces the number of the original free parameters of the model as all weights are now expressed by the covering factor @xmath74 which allows us to test the consistency of the derived model parameters with the random distribution of clouds . in the case of model ( a ) and @xmath52 the laor et al .
( 1997 ) data were reproduced by the model with the relative weights of the components in eq .
( 2 ) equal to 0.15 , 1.2 and 1.0 .
if we rely on the relative weight of the second and third term , this result can be translated into a covering factor @xmath75 .
the role of the amplification is taken into account ( @xmath76 ) .
such a cloud distribution , while reproducing precisely the normalization of the last two components , predicts a significantly larger value of the primary contribution ( 0.90 instead of 0.15 ) .
lower amplification still widens this gap .
this means that either the cloud distribution is not random or our theoretical spectra are too hard ( i.e. flat ) in the soft x - ray band .
since the equivalent width of the fe @xmath2 line produced by the model is also too large if the primary level is as low as 0.15 it strongly suggests that the geometry is correct but the reflection spectra should be improved .
the size of the primary source could in principle be estimated from the statistical properties of the observational sample of objects .
we can only infer that the source is probably not too compact in comparison with the clouds distance from the center since a point like source would give two distinct classes of objects : a fraction of objects dominated by primary emission and the rest of the sources with the primary completely hidden from an observer , which seems not to be the case . in the case of model ( c )
a slightly higher value of @xmath0 seems to be favored , of order of 500 .
the relative weights of the components are equal to 0.15 , 5.0 and 1.0 .
therefore , dark sides of the clouds are contributing more to the total spectrum . however , now the amplification is weaker by a factor @xmath77 due to the size of the hot cloud . combining the requirements for the relative contribution of the reflected component , dark side component and equation ( 1 ) we obtain @xmath78 and @xmath79 .
such a model predicts the contribution from the primary ( hot plasma in that case ) , of order of 0.65 , a factor of 4 higher than allowed by the data .
it is interesting to note that the condition for the size of the hot plasma cloud ( eq .
1 ) leads to a reasonable value of the @xmath39 ratio for the value of the compton parameter @xmath34 which resulted from the choice of the plasma parameters giving the hard x - ray slope of index @xmath80 .
= 80 mm the model specifies the @xmath39 ratio , but not @xmath81 itself .
however , if we assume that the mean quasar spectrum actually corresponds to a mean value of the accretion rate @xmath82yr@xmath83 ( i.e. total luminosity @xmath84 erg s@xmath83 ) and a mean value of the black hole @xmath85 ( after zheng et al .
1997 ) the adopted value of the cloud density and the obtained value of the ionization parameter @xmath53 suggest immediately that the representative value of @xmath81 is of order of @xmath86 .
we can not fully rely on the value of the accretion rate and the mass of the black hole since they depend on the choice of a specific model .
however , any reasonable assumption about the luminosity to the eddington luminosity ratio ( of order of 0.1 in this case ) would give a similar order of magnitude .
therefore , the kinematic effects are important . in fig .
15 we show how the soft x - ray emission features are affected by the motion of the clouds .
we see that the result is quite sensitive to the actual distances of the clouds from the center .
larger distances , of about @xmath87 leave most of the spectral features almost unaffected while considerably smaller values , @xmath88 , smear out all the features apart from a strong oviii line which is very broad . therefore , if soft x - ray emission is really produced as a reflection spectrum the presence and the strength of the spectral features should help to test the model .
unfortunately , the available data for quasars are not yet of the appropriate quality and there is an additional problem of confusion with the spectral features due to warm absorber . the comparison between the theoretical models and the data shown in sect
. 4 did not include spectral fitting in terms of determination of the @xmath89 statistics since the cloud model is not ready yet for that kind of quantitative analysis .
clearly , very careful , more advanced computations of both the reflected spectrum and the emission of the dark side of a cloud are necessary in order to decide whether the soft x - ray part of the quasar spectrum can be identified with the radiation reflected by the irradiated sides of the clouds , with some contribution from the primary radiation .
actually , preliminary computations taking into account multiple scattering ( see paper i ) indicate that the amplification is wavelength - dependent as the spectra are steeper in soft x - rays due to this effect .
more careful computations based on radiative transfer of soft photons and monte carlo computations of x - ray photons show that the slope of the spectrum for @xmath52 has the slope as steep as 2.2 for the reflection component in the 0.2 - 2 kev band due to the multiple reflection ( abrassart et al .
, in preparation ) .
such a steep reflection spectrum would allow a primary contribution as high as 0.5 which is only lower by a factor two than expected from a random distribution of the clouds .
the geometrical arrangement of the origin of the primary radiation does not seem equally important as the physical input of the radiative transfer code since the geometrical parameters will adjust themselves to the shapes of the basic spectral components .
considerable help may be provided by the variability constraints . in the otcm
there are three kinds of variations expected .
the first one is connected with variations of the primary source , or the comptonizing hot medium .
clouds respond to those variations both in optical / uv and in soft x - rays with an average delay of order of the travel time through the region they occupy , @xmath90 , of order of some ten thousand of seconds for a @xmath91 black hole . the second one is connected with a single cloud motion , i.e. the visibility of the primary .
these variations are mostly limited to the primary variations and some changes in optical / uv or soft x - rays resulting from the temporary change in geometry are not expected to be delayed in any systematic way .
the time - scale of these variations should be of order of the cloud period diminished by the factor describing the relative size of the x - ray source , @xmath39 , thus not considerably longer than the previous time - scale . variations in quasars on those time - scales are unmeasurably weak in the optical band ( below 1.2 % in time - scales of hours in 3c 273 , von montigny et al .
1997 ) .
finally , any systematic changes in the accretion rate in the innermost part of the flow which would both include a change of the level of primary emission as well as of the covering factor should happen on considerably longer time - scales .
however , at the present stage the otcm does not give quantitative predictions of the relative changes of these two factors , although we generally expect that the covering factor should grow with accretion rate , i.e. with bolometric luminosity , thus resulting in harder uv spectra and larger @xmath46 for larger luminosity .
the variability in hard x - ray band for radio quiet quasars also seems to be the best and most direct probe of the nature of the primary source .
high quality data for galactic sources allow us to compute the time delays as functions of the fourier phase as well as the coherence function ( e.g. cui et al .
1997 ) which strongly support the comptonization mechanism for the production of primary emission and provide a potentially powerful method to constrain the distribution of the hot gas ( e.g. hua , kazanas & titarchuk 1997 ) .
the time - scales involved are in the range of @xmath92s to 1s for cyg x-1 so simple minded scaling may suggest time - scales from days to years .
our results suggest that the quasar spectra can be explained within the frame of the optically thick cloud model ( otcm ) , i.e. as originating in a distribution of irradiated clouds optically thick for electron scattering .
one of the important points is that the spectrum in the soft x - ray band is attributed mostly to the reflection from the illuminated partially ionized sides of the clouds .
we are not able to definitively distinguish at present if the primary emission comes from the unspecified central source in the form of synchrotron emission or if it originates in the central hot plasma cloud by upscattering of soft photons coming from the clouds
. however , we favor the second possibility as the presented results already lead to more self - consistent description for that case . an alternative picture based on the accretion disk and a corona requires very hot plasma to form the hard x - ray emission and moderately hot but optically thicker plasma to explain the soft x - ray quasar spectra .
these two pictures differ with respect to the presence of spectral features since the reflection is accompanied by emission lines and absorption edges which may be used for the purpose of diagnostics .
these features are expected to be smeared to some extent by the cloud motion which may help to distinguish them from features arising in a distant warm absorber . within the frame of otcm ,
the overall shape of the spectrum is very sensitive to the value of the ionization parameter @xmath0 .
interestingly , the value favored by the comparison of the model to the mean quasar spectrum of laor et al .
( 1997 ) is @xmath58 , in agreement with the energy of the fe @xmath2 line in moderately bright quasars ( nandra et al .
present results allow too low a contribution ( by a factor of a few ) from the primary emission to account for the observed equivalent width of @xmath2 but we expect that more advanced computations may remove this problem ( abrassart et al .
, in preparation ) .
the problem which remains is the prediction of the strong spectral features in the optical / uv band .
observations seem to indicate that no such features are observed in quasar spectra while both otcm and advanced accretion disk models predict such atomic features although kinematic effects connected with the cloud motion decrease them to some extent .
we will address this problem in the future .
we are grateful to suzy collin for many helpful discussions and detailed comments on the manuscript .
we thank piotr .
zycki for his help with the high energy reflection code .
this work was partially supported by grant 2p03d00410 of the polish state committee for scientific research and by jumelage / cnrs no 16 `` astronomie france / pologne '' . | we assume a distribution of clouds optically thick for electron scattering ( otcm ) which are moderately optically thin for absorption and we consider them as a model of the mean quasar spectrum of laor et al .
( 1997 ) .
we show that the model is particularly sensitive to the value of the ionization parameter @xmath0 and that for @xmath1 the model well reproduces the optical / uv / x - ray mean quasar spectrum , in agreement with the estimates of the ionization parameter based on the energy of the iron @xmath2 line .
we can not definitively reject synchrotron emission as a source of primary radiation but we favor the model in which the hard x - ray emission is produced by compton scattering of soft photons in a central hot medium surrounded by cool clouds . in such a model clouds are located typically at the distance of @xmath3 , with the covering factor about 0.88 and the radius of hot plasma is @xmath4 .
the model explains optical / uv emission as predominantly due to the dark sides of the clouds and the soft x - ray emission as due to the reflection by the irradiated sides of the clouds .
therefore , atomic features are expected in these bands although they are hardly present in the observational data .
the kinematical effects connected with the cloud motion affect those features but do not remove them .
the level of the primary emission required to model the mean quasar spectrum is too low to reproduce the equivalent width of the iron k@xmath5 line correctly but more detailed computations may resolve this problem . |
one of the strongest motivations for studying low - mass x - ray binaries ( lmxbs ) has been the aim to use these systems as probes of fundamental physics .
it has been argued that the study of high frequency ( 40450 hz ) quasi - periodic oscillations ( hfqpos ) found in lmxbs containing a stellar mass black hole ( bh ) is one of the most promising tools to do so .
the hfqpos are the fastest phenomenon observed in these systems and their high frequency suggests that they are produced in the innermost region of the accretion flow near the bh event horizon . once the processes involved in their formation are firmly understood , they will give a wealth of information on bh mass and spin as well as the structure of strongly curved spacetime ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the frequency of the hfqpos usually occur at specific values , different in each source . in a few cases , pairs of qpos have been detected .
the highest frequency appears to scale inversely proportional to the to bh mass ( e.g. , * ? ? ?
* ; * ? ? ?
hfqpos are weak , transient , and energy dependent , and are mostly detected only at high count rates . in some cases
they are found just above detection levels , so there is considerable uncertainty about their exact properties .
however , typically the quality factor q ranges between less than 2 and a few tens , and amplitudes are between 0.5 and 5% rms ( depending on the source and the energy range used ) .
twin qpos at a stable 300 hz and 450 hz @xcite have been observed in the bh gro j165540 ; at 240 and 160 hz in h1743322 @xcite and at 188 and 268 hz in xte j155056 ( note however , that in this case it is not clear how stable the frequency of these qpos is , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
single qpos or peaked noise ( i.e. when @xmath42 ) features at a rather stable frequency have been detected in xte j1650500 ( @xmath1250 hz , see , e.g , * ? ? ? * ) , 4u 163047 ( variable frequency , see * ? ? ? * ) and xte j1859 + 226 ( @xmath1180 hz , see , e.g. , * ? ? ?
the remaining bh which shows hfqpos is the peculiar system grs 1915 + 105 , which is a 33.5 days orbital period binary system harboring a @xmath5 bh @xcite . at a distance of @xmath6 kpc @xcite ,
it is very often at eddington or super - eddington luminosity ( e.g. , * ? ? ?
until recently , grs 1915 + 105 has been unique in that its x - ray light curves exhibit more than a dozen different patterns of variability usually called `` classes '' ( which are referred to with greek letters ) , most of which are high amplitude and highly - structured ( e.g. , * ? ? ?
most of this structured variability occurs on timescales of seconds or longer , and it is thought to be due to limit cycles of accretion and ejection in an unstable disk ( e.g. , * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
grs 1915 + 105 has hfqpos at the rather low frequencies of @xmath140 hz @xcite and @xmath167 hz @xcite compared with the rest of the bh systems ; however , as in the case of gro j165540 and h1743322 , the qpo frequencies remain almost constant as the x - ray flux changes ( any change in the frequency of the @xmath140 hz oscillation during the sequence of observations was @xmath71% ; the frequency of the 67 hz oscillation varies by only a few percent as the x - ray flux varies by a factor of several ) , suggesting a common origin between systems .
@xcite reported on the detection of qpos at @xmath127 hz ; given the weakness of the signal @xcite could not determine whether the @xmath127 hz was related to the @xmath167 hz qpo ( at the time of @xcite , the @xmath140 hz qpo was not yet reported ) ; however , they were able to conclude that the presence / absence of hfqpos is intimately linked to the slower oscillations and variations that happen on timescales longer than seconds .
an additional peak at @xmath1170 hz was also found after applying a very specific class - count rate - hardness selection @xcite ; however , this barely significant peak is very broad , with a q value or @xmath12 .
if the frequency of the hfqpos depends on the mass and/or spin of the bh , one would expect ( as observations appear to suggest ) , that different systems with different parameters should show qpos at different frequencies . the recent discovery of a bh ( igr j170913624 ) which shows very similar high amplitude and highly - structured variability to that seen in grs 1915 + 105 ( see * ? ? ?
* and references therein ) opened a new window of opportunities to understand the physical mechanism that produces the highly structured x - ray variability and it relation with much faster ( sub - second ) variability .
igr j170913624 has been seen active in 1994 , 1996 , 2001 , 2003 , 2007 and 2011 @xcite . only in the recent 2011 outburst igr
j170913624 started to show the high amplitude and highly - structured variability @xcite typical of grs 1915 + 105 . in the context of the variability classes defined by @xcite for grs 1915 + 105
, @xcite found that igr j170913624 shows the @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 and @xmath13 classes as well as quiet periods which resemble the @xmath14 class , all occurring at 2 - 60 kev count rate levels which can be 10 - 50 times lower than observed in grs 1915 + 105 .
@xcite also found that the difference in flux combined with the circumstance that currently neither the distance to igr j17091 - 3624 nor the mass of its compact object are known lead to the conclusion that either all models requiring near eddington luminosities for grs 1915 + 105-like variability fail , or igr j170913624 lies at a distance well in excess of 20 kpc or , it harbors one of the least massive bhs known ( @xmath15 ) .
the similarities and differences between igr j170913624 and grs 1915 + 105 raises the question of at what frequency should we observe the hfqpos in igr j170913624 . assuming 1/m scaling , and the possibility that igr j170913624 has a mass which is a factor of @xmath15 smaller than grs 1915 + 105 , then the hfqpos in igr j170913624 should be at a frequency significantly higher than the @xmath140 hz and @xmath167 hz seen in grs 1915 + 105
. finding similar hfqpo frequency between sources would imply that the frequency does not depend on the mass or that igr j170913624 lies at a distance well in excess of 20 kpc .
triggered by this question , we searched all the _ rossi x - ray timing explorer _
( rxte ) available data of igr j170913624 for the presence of qpos at frequencies higher than the @xmath17 - 10 hz reported by @xcite . in this letter
we report the discovery of a highly significant qpo at @xmath165 hz , and a marginally detection of a qpo at @xmath1165 hz and briefly discuss the implication of our detections .
igr j170913624 was observed with the proportional counter array ( pca ; * ? ? ?
* ) on - board rxte almost daily since the outburst began in february 2011 until october 15th , 2011 , when rxte could not point to igr j170913624 any more due to visibility constraints .
a total of 228 observations , covering @xmath16 days , sample the 2011 part of the current outburst ( at the moment of submission it is not known whether the source is still active or not ) . in this work we do not use data taken between mjd 55600 and 55614 ( 14 observations ) , given that these observations were affected by a bright source in the field of view ( e.g. , * ? ? ? * ) .
we use the 16-s time - resolution standard 2 mode data to calculate the crab - normalized average per observation 220 kev intensity ( see , e.g. , * ? ? ?
* and references therein for details on the method ) .
power spectra and light curves were produced from the pca using standard techniques ( e.g. , * ? ? ?
* ; * ? ? ?
for the fourier timing analysis we used data in the 2 - 25 kev range ( absolute channels 560 ) from the good xenon mode . to fit the high - frequency ( @xmath17 hz ) part of power spectra , we used one or two lorentzian functions plus a power - law when needed .
we give the frequency of the lorentzians in terms of characteristic frequency @xmath18 @xcite .
for the quality factor @xmath19 we use the standard definition @xmath20 .
_ fwhm _ is the full width at half maximum and @xmath21 the centroid frequency of the lorentzian .
the strength of the qpos are given in terms of fractional rms amplitude .
in the upper and lower panel of fig . [ fig : lc ] we show the average 220 kev intensity for each observation and the 0.01 - 10 hz rms amplitude , respectively . during the 265 days period , igr
j170913624 has been at an average intensity between 20 and 65 mcrab .
( note that during single flares or heartbeats the source was much brighter ) .
the rms amplitude also varied significantly , where igr j170913624 underwent periods of very low ( @xmath110% ) and very high ( @xmath2260% ) amplitude .
figure [ fig : lc ] shows that there is no clear correlation between intensity and rms amplitude .
a similar study of grs 1915 + 105 shows the same ( this will be reported elsewhere ) . at
first we searched each observation power spectrum for indications of qpos at higher frequency than 10 hz and at a significance larger than 3.5@xmath2 .
we found a clear detection ( 4.7@xmath2 , single trial ) of a qpo at @xmath23 hz , @xmath24 and rms amplitude of @xmath25% ( 2 - 25 kev , obsid : 96420 - 01 - 36 - 03 , mjd 55865.5 ) .
we also found that in the period mjd 55830 - 55880 many power spectra revealed power excesses in the 60 - 70 hz range . to further investigate this , we added all data in this period , which corresponds to the low - variability period marked with a grey area in fig .
[ fig : lc ] . in the upper panel of fig .
[ fig : pds ] we show that the resulting power spectrum is clearly dominated by a strong feature at @xmath10.18 hz , but also shows an 8.5@xmath2 qpo at @xmath26 hz ( @xmath27 ; @xmath28% rms amplitude ) .
different sub - selections on time and intensity of these data lead to significant changes in q , amplitude and significance , but not in frequency .
this suggests that that the 66 hz qpo is not always present , similar to what occurs in grs 1915 + 105 . in several of our sub - selections we saw an excess of power in the 150 - 200 hz range . in the left panel of fig .
[ fig : pdsselect ] we show the result of averaging all observations after mjd j55841 .
the qpo at @xmath29 hz ( @xmath30 significant , @xmath31 , @xmath32% rms amplitude ) is accompanied with a a qpo at @xmath33 hz ( @xmath14.5@xmath2 , @xmath34 , @xmath35% rms amplitude ) ; in the inset we show the rms amplitude vs. energy of the 66 hz qpo ; its amplitude increases with energy from @xmath15% at 3 kev , to @xmath114% at 13 kev .
the 67 hz qpo in grs 1915 + 105 also has a hard spectrum , although less extreme than this one @xcite .
the low s / n did not allow a similar analysis for the 164 hz qpo .
we also explored whether the 65 hz qpo is also present before mjd 55830 which is not detected due to the low - statistics . averaging all data in this period led to a broad feature at @xmath160 - 80 hz whose parameters were unconstrained .
when we selected only high count rates ( @xmath36 cts / sec / pcu ) , we detected a 4.5@xmath2 broad ( @xmath37 ) feature at @xmath38 hz ( @xmath39% rms amplitude ) .
we show our best fit n the right panel of fig .
[ fig : pdsselect ] .
we did not detect any qpo at 40 hz nor 27 hz as seen in grs 1915 + 105 .
this is not surprising , as those qpos are can be weaker and complex data selection might be needed to find them ( e.g. , * ? ? ? * ; * ? ? ?
* ) the period mjd 55830 - 55880 ( where the 66 hz qpo is strongest , see section [ sec : khz ] ) corresponds to an interval of relative low ( 10 - 30% ) rms amplitude .
the 1-sec light curves during this period are relatively flat , apart from quasi - periodic oscillations with a typical time scale of 10 s and/or the presence of sharp ` dips ' with a typical duration of a few seconds . in the upper panel of fig .
[ fig : lccomp ] we show a representative lightcuve during this period .
when compared with the grs 1915 + 105 variability classes defined by @xcite , the igr j170913624 light curves during mjd 55830 - 55880 are reminiscent of the @xmath3 class . as a comparison , in the lower panel of fig .
[ fig : lccomp ] we plot a representative @xmath3 class data - segment for grs 1915 + 105 ( for the full light curve , see figs .
2f and 15b , d in * ? ? ?
* ) . a comparison between power spectra between the igr j170913624 and grs 1915 + 105 is shown in fig . [ fig : pds ] .
we discovered the presence of a hfqpo in the rxte data of the igr j170913624 , a system whose x - ray properties are very similar to those of grs 1915 + 105 .
the centroid frequency of the strongest peak is @xmath166 hz , its quality factor above 5 and its rms is between 4 and 10% .
we found a possible additional peak at 164 hz when selecting a subset of data ; however , we consider this detection marginal .
the hfqpos are detected during a period where igr j170913624 is in the @xmath3 variability class @xcite .
@xcite reported observations until mjd 55800 where so far igr j170913624 was seen in at least the @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 , @xmath13 and @xmath14 classes .
the fact that now we also identified the @xmath3 class probably means that igr j170913624 is still evolving between classes .
the frequency of the 66 hz qpo in igr j170913624 is surprisingly similar to that of the main hfqpo in grs 1915 + 105 ( 65 - 69 hz , see * ? ? ?
in addition , the frequency of the 4.5@xmath2 significant broad feature at @xmath33 hz in igr j170913624 also matches that of a similar broad feature found at 170 hz in grs 1915 + 105 @xcite .
the circumstance that currently neither the distance to igr j17091 - 3624 nor the mass of its compact object are known led @xcite to the conclusion that either all models requiring near eddington luminosities for grs 1915 + 105-like variability fail , or igr j170913624 lies at a distance well in excess of 20 kpc or , it harbors one of the least massive bhs known ( @xmath15 ) .
the surprisingly match between the @xmath170 hz and @xmath1170 hz frequency therefore raises more questions than answers .
the frequencies of dynamical motions near compact objects are thought to scale with mass ( e.g. , * ? ? ?
the slower highly - structured low - frequency variability seen in grs 1915 + 105 as compared with igr j17091 - 3624 supported the idea that the later was less massive @xcite .
however , the fact that the hfqpos occur at the same frequency could be at odds , meaning that either the frequency of the hfqpos or the highest - frequency of the ` heartbeats ' in these two sources scale with mass , or that none of them do , or that they both do thanks to a difference in bh spin . if the 67 hz hfqpos frequency scales with mass ( assuming same mass , same spin ) , then either igr j170913624 is further than 20 kpc or it is close , in which case all models requiring near eddington luminosities for grs 1915 + 105-like variability fail @xcite . under the assumption that the centroid frequencies represent that of a keplerian circular motion at the inner - most stable orbit ( isco ) , then at first approximation ( e.g. , * ? ? ? * ) no solution is possible for the 67 hz qpos independently of the spin , unless either the hfqpos are not produced at the isco or the mass of both bhs is significantly higher than the @xmath114@xmath40 estimated for grs 1915 + 105 .
the 160 - 170 hz broad qpos found in both in igr j170913624 ( this letter ) or grs 1915 + 105 would be consistent with a 14 - 20@xmath40 , but only if the spin parameter ( @xmath41 ) in these systems is significantly lower than 1 ( i.e. , at odds with the reported @xmath42 for grs 1915 + 105 , e.g. , * ? ? ? * ; * ? ? ? * ) . in the framework of the relativistic precession model ( rpm , * ? ? ?
* ) , and relaxing the assumption on equal spin , we can identify the higher frequency ( 160 - 170 hz ) as representing a keplerian motion at a certain radius and , the lower frequency ( 66 hz ) as the relativistic periastron precession at the same radius ( notice that this radius must be larger than the isco , where the two frequencies are the same ) . to test this identification , we searched for the combinations between spin parameter and bh mass which allow a solution .
we find solutions for masses ranging between 7 m@xmath43 ( for @xmath44 ) and 19 m@xmath43 ( for @xmath45 ) , the latter consistent with the @xmath5 and @xmath46 reported for grs 1915 + 105 @xcite .
this identification implies that igr j170913624 could be considerably less massive than grs 1915 + 105 ( and coincidentally have a much lower spin than grs 1915 + 105 ) , explaining in this way why the highest frequency of the heartbeats is significantly higher in igr j170913624 than in grs 1915 + 105 @xcite .
we note that in the framework of the relativistic precession model the 41 hz frequency would be unexplained .
solutions to the 41,67 hz pair for grs 1915 + 105 are possible , but only for masses larger than 25m@xmath43 and considering no spin
( i.e @xmath44 ) .
if the hfqpos frequency _ does not _ scale with mass , then it is intriguing why their hfqpos frequencies are the same in these two sources .
one possibility is that the 67 hz hfqpos are different from the hfqpos seen in other sources ( e.g. , * ? ? ?
* ) , and that their frequency is instead related with the physical process that produces the highly - structured low - frequency variability ( only seen ) in these two sources . if true , this could have major implications on works that use the @xmath47 hz hfqpo in grs 1915 + 105 as a standard to estimate the mass of galactic and/or super massive bhs ( e.g. * ? ? ?
our results add another hfqpo to the small number already known .
its properties are not different from the others , in particular its elusiveness .
there are thousands of observations of more than 20 black - hole transients in the rxte archive , but only very few significant peaks have been found . as rxte will cease operations in 2012 january , this sample is not going to be increased until the indian astronomical mission astrosat is launched , while a major step forward will only come with the next generation of x - ray instruments ( e.g. loft , see * ? ? ? * ) .
* acknowledgments : * tb acknowledge support from grant inaf - asi i/009/10/0 .
the research leading to these results has received funding from the european community s seventh framework programme ( fp7/2007 - 2013 ) under grant agreement number itn 215212
`` black hole universe '' . | we report the discovery of @xmath0 high - frequency quasi - periodic oscillations ( hfqpos ) at 66 hz in the rxte data of the black hole candidate igr j170913624 , a system whose x - ray properties are very similar to those of microquasar grs 1915 + 105 .
the centroid frequency of the strongest peak is @xmath166 hz , its quality factor above 5 and its rms is between 4 and 10% .
we found a possible additional peak at 164 hz when selecting a subset of data ; however , at 4.5@xmath2 level we consider this detection marginal .
these qpos have hard spectrum and are stronger in observations performed between september and october 2011 , during which igr j170913624 displayed for the first time light curves which resemble those of the @xmath3 variability class in grs 1915 + 105 .
we find that the 66 hz qpo is also present in previous observations ( 4.5@xmath2 ) , but only when averaging @xmath1235 ksec of relatively high count rate data .
the fact that the hfqpos frequency in igr j170913624 matches surprisingly well that seen in grs 1915 + 105 raises questions on the mass scaling of qpos frequency in these two systems .
we discuss some possible interpretations , however , they all strongly depend on the distance and mass of igr j17091 - 3624 , both completely unconstrained today . |
we are going to study the lyapunov exponents of a rational function @xmath1 acting on the riemann sphere , of degree at least @xmath2 .
in particular , continuing the investigations in @xcite , we are interested in the case that the map @xmath3 is exceptional . slightly modifying ( * ? ? ?
* section 1.3 ) , we call @xmath3 _ exceptional _ if there exists a finite , nonempty , and forward invariant set @xmath4 such that @xmath5 here @xmath6 is the julia set of @xmath3 and @xmath7 is the set of critical points of @xmath3 .
every such set @xmath8 has at most @xmath9 points ( see lemma [ lem : at most four ] ) , hence there is a maximal set with this property , which we denote by @xmath10 .
if @xmath3 is non - exceptional we put @xmath11 . when @xmath3 is clear from the context we denote @xmath10 simply by @xmath12 . given @xmath13 , denote by @xmath14 and @xmath15 the _ lower _ and _ upper lyapunov exponent _ at @xmath16 , respectively .
if both values coincide then we call the common value the _ lyapunov exponent _ at @xmath16 and denote it by @xmath17 .
similarly , for a @xmath3-invariant probability measure @xmath18 we denote by @xmath19 its _ lyapunov exponent_. let @xmath20 be the set of all @xmath3-invariant borel probability measures supported on @xmath21 and @xmath22 be the one of all non - atomic ones .
let @xmath23 and @xmath24 be the sets of ergodic measures contained in @xmath25 and @xmath20 , respectively .
let @xmath26 ( see corollary [ def ] for equivalent definitions of @xmath27 ) . for given numbers
@xmath28 we consider the level sets @xmath29 we denote by @xmath30 the set of _ lyapunov regular points _ with exponent @xmath31 .
we will describe the complexity of such level sets in terms of their hausdorff dimension @xmath32 .
to do so , given a parameter @xmath33 let us consider the potential @xmath34 and the pressure function @xmath35 notice that if @xmath36 is nonempty the potential @xmath37 is unbounded and @xmath38 does not coincide with the the classical topological pressure for @xmath39 ( see @xcite ) .
we define the _ hidden variational pressure _
@xmath40 ( following the terminology in @xcite ) .
after makarov and smirnov ( * ? ? ?
* theorem b ) , the pressure function @xmath41 fails to be real analytic on the interval @xmath42 if and only if @xmath3 is exceptional and @xmath43 moreover , by ( * ? ? ?
* theorem a ) the function @xmath44 is real analytic on the interval @xmath42 and @xmath45 for any @xmath46 let @xmath47 our main result is the following theorem .
[ main1 ] let @xmath3 be a rational function of degree at least @xmath2 .
for any numbers @xmath31 , @xmath48 with @xmath49 we have @xmath50 in particular , for any @xmath51 \setminus \{0\}$ ] we have @xmath52 for @xmath53 we have @xmath54 moreover , @xmath55 and @xmath56 the result of the above theorem has been shown in @xcite in the particular case that @xmath3 is non - exceptional . to prove our main result , in this paper we will create new technical tools in order to deal with exceptional rational maps and then show how
these tools can be applied to adapt the original proofs in @xcite .
the paper is organized as follows . in section [ sec:2 ]
we collect some known results about exceptional maps that will be used in the rest of the paper . in section [ sec : hidden ] we will introduce the concept of hidden pressure using backward branches of @xmath3 , analogously to the tree pressure from @xcite . in the case of exceptional rational maps
we not always have at hand a finite conformal measure with dense support , see proposition [ p : known conformal ] .
for that reason , in section [ sec : bal ] we introduce @xmath0-finite conformal measures that are associated to the hidden pressure . finally , in section [ sec : final ] we apply these tools to prove theorem [ main1 ] . in section [ sec : final-1 ] we provide a lower bound for dimension using the fact that for any rational map we can find an increasing family of uniformly expanding cantor repellers contained in @xmath21 using a construction of bridges that has been established in @xcite and applies to the setting of this paper without changes . in section [ sec : final-2 ] we provide an upper bound for dimension applying frostman s lemma to an appropriate @xmath0-conformal measure at a conical point .
finally , in section [ sec : final-3 ] , we show the existence of periodic orbits in @xmath57 with exponent as large as possible .
we give an alternative proof of this result in appendix [ s : alternative ] via a variant of bowen s periodic specification property , @xcite .
for a critical point @xmath58 we will denote by @xmath59 the local degree of @xmath3 at @xmath60 .
the following result has been proved by the same computation first in ( * ? ? ?
* lemma 2 ) .
[ lem : at most four ] if @xmath8 is a finite subset of @xmath61 such that @xmath62 , then @xmath63 . if @xmath3 is a polynomial then @xmath64 . using that @xmath3 has @xmath65 critical points counted with multiplicity , by we have @xmath66 so @xmath67 . if @xmath3 is a polynomial , then it has at most @xmath68 finite critical points counted with multiplicity , so in this case @xmath69 .
the following is an example of a one parameter family of exceptional rational maps such that for some parameters the exceptional set contains a critical point : for @xmath70 put @xmath71 the point @xmath72 is critical of multiplicity @xmath73 , the point @xmath74 is fixed of multiplier @xmath75 , and the point @xmath76 is critical of maximal multiplicity and the only preimage of @xmath72 .
thus , when @xmath77 belongs to the julia set we have @xmath78 . by choosing suitable @xmath79
, the fixed point @xmath77 could be repelling , cremer , etc .
if @xmath3 is exceptional , then the set @xmath12 contains at least one periodic point . observe that it hence must consist of a finite number of periodic points plus possibly some of their preimages .
we write @xmath80 , where @xmath81 denotes the subset of all neutral periodic points in @xmath12 plus its pre - images and where @xmath82 denotes the subset of all expanding periodic points in @xmath12 plus its pre - images .
we refer to @xcite for further details on exceptional maps and numerous examples .
we will say that @xmath3 has a _ phase transition in the negative spectrum _ if the function @xmath83 fails to be real analytic on @xmath84 .
in this case we put @xmath85 we have @xmath86 and , since the function @xmath83 is convex , for each @xmath87 we have @xmath88 . in the following proposition
we gather several results in @xcite .
a measurable subset @xmath89 of @xmath61 is said to be _ special _ if @xmath90 is injective . given a function @xmath91 , a borel probability measure @xmath92 on @xmath21
is said to be _
@xmath93-conformal outside _
@xmath94 if for every special set @xmath95 we have @xmath96 if @xmath97
we simply say that @xmath92 is _
[ p : known conformal ] let @xmath3 be a rational map of degree at least @xmath2 and let @xmath99 .
then we have the following properties . 1 .
suppose that @xmath3 does not have a phase transition in the negative spectrum , or that @xmath3 has a phase transition in the negative spectrum and @xmath100 .
then @xmath101 and there is a finite @xmath102-conformal measure whose support is equal to @xmath21 .
2 . suppose that @xmath3 has a phase transition in the negative spectrum and that @xmath103 .
then @xmath3 is exceptional , there is an expanding periodic point @xmath104 such that @xmath105 and for every neighborhood @xmath106 of @xmath107 and every measure @xmath92 that is @xmath108-conformal measure outside @xmath36 we have @xmath109 the equality @xmath110 in part @xmath111 follows from the definition of @xmath112 .
the existence of the conformal measure in part @xmath111 follows from ( * ? ? ?
* lemma @xmath113 ) if @xmath114 and from ( * ? ? ? * theorem a ) if @xmath115 .
the fact that @xmath3 is exceptional and that there is an expanding periodic point @xmath104 such that @xmath116 in part @xmath2 is given by ( * ? ? ?
* theorem b ) . to complete the proof of part @xmath2 ,
let @xmath92 be a @xmath108-conformal measure @xmath36 .
since @xmath3 is topologically exact on @xmath21 , it follows that the support of @xmath92 is equal to @xmath21
. let @xmath117 be the period of @xmath107 and let @xmath118 be sufficiently small so that @xmath119 and so that the inverse branch @xmath120 of @xmath121 fixing @xmath107 is defined on @xmath122 and satisfies @xmath123 .
then there is a distortion constant @xmath124 such that , if we put @xmath125 , then for each integer @xmath126 we have by the conformality of @xmath92 @xmath127 thus @xmath128 proving the proposition .
the goal of this section is to prove equivalence of three pressure functions : the hidden variational pressure defined in as well as the hidden hyperbolic pressure and the hidden tree pressure defined in and below . given @xmath33 , the _ hidden hyperbolic pressure _ is defined as @xmath129 where the supremum is taken over all compact @xmath3-invariant ( i.e. @xmath130 ) isolated expanding subsets of @xmath131 .
we call such a set _ uniformly expanding repeller_. here _ isolated _ means that there exists a neighborhood @xmath132 of @xmath133 such that @xmath134 for all @xmath135 implies @xmath136 . [ p.primeiro ] @xmath137 for every @xmath33 .
the inequality @xmath138 follows from the variational principle . on the other hand ( *
* theorem 11.6.1 ) implies that for any @xmath139 we have @xmath140 and hence @xmath141 . before defining the hidden tree pressure ,
let us recall some concepts from @xcite , @xcite , and ( * ? ? ?
* chapter 12.5 ) . given @xmath142 and @xmath33 ,
we consider the _ tree pressure _ of @xmath37 at @xmath143 defined by @xmath144 a point @xmath142 is said to be _ safe _ if @xmath145 where @xmath146 denotes the spherical distance .
a point @xmath142 is said to be _ expanding _ if there exist numbers @xmath147 and @xmath148 such that for all sufficiently large @xmath149 the map @xmath121 is univalent on @xmath150 and satisfies @xmath151 . here , for a subset @xmath132 of @xmath61 and @xmath152 we denote by @xmath153 the connected component of @xmath154 containing @xmath143 .
we point out that every point in @xmath61 outside a set of hausdorff dimension zero is safe and that for each safe point @xmath155 we have @xmath156 , see @xcite and compare with ( * ? ? ?
* theorem 3.4 ) .
[ lem : exisafe ] there exists an expanding safe point in @xmath57 .
notice that @xmath157 since @xmath36 is finite and @xmath158 for any @xmath159 and @xmath160 , this inclusion implies that the set of points that fail to be safe has zero hausdorff dimension .
thus , the existence of an expanding safe point outside @xmath161 follows from the existence of uniformly expanding cantor repellers outside @xmath161 , for example as derived in ( * ? ? ?
* lemma 4 ) .
note that such repellers always have a positive hausdorff dimension by bowen s formula ( see , for example , ( * ? ? ?
* chapter 9.1 ) ) .
let us now define the hidden tree pressure that is an analogue of the tree pressure , obtained by considering a restricted tree of preimages .
given a subset @xmath106 of @xmath21 and @xmath162 which is not in the forward orbit of a critical point , we define @xmath163 and we consider the _ hidden tree pressure _ of @xmath37 at @xmath143 defined by @xmath164 usually the point @xmath143 will be expanding safe and the set @xmath106 will be a neighborhood of @xmath12 .
note that after lemma [ lem : exisafe ] there are such @xmath143 and @xmath106 .
[ l.correct ] if @xmath165 , @xmath106 is a sufficiently small neighborhood of @xmath12 , and @xmath162 is expanding safe , then the pressure @xmath166 does not depend on @xmath106 . to prove the above lemma we need the following technical lemma . [
lem : caf1 ] for an arbitrary neighborhood @xmath106 of @xmath12 and an arbitrary number @xmath167 there exists a number @xmath168 and positive integers @xmath169 such that for every point @xmath170 there exist numbers @xmath171 , @xmath172 and a point @xmath173 such that the set @xmath174 is @xmath175-dense in @xmath21 and satisfies @xmath176 by the locally eventually onto property of @xmath3 on @xmath21 there is an integer @xmath177 such that for each @xmath178 the set @xmath179 is @xmath175-dense in @xmath21 .
we put @xmath180 for each integer @xmath181 let @xmath182 be defined by @xmath183 we will show that for each @xmath184 there is an integer @xmath185 such @xmath186 . since for each @xmath187 the function @xmath188 is continuous and for each @xmath189
the sequence @xmath190 is non - decreasing , it follows that there is a number @xmath191 so that @xmath188 is strictly positive on @xmath192 .
this will imply the desired assertion with @xmath193 we distinguish three cases : + * 1 ) * if @xmath194 is not in the forward orbit of a critical point then @xmath195 .
+ * 2 ) * if @xmath194 is in the forward orbit of a citical point that is not pre - periodic then there exists a number @xmath196 such that @xmath197 is disjoint from @xmath198 . hence , we obtain that @xmath199 . + *
3 ) * if @xmath194 is in the forward orbit of a pre - periodic critical point then , there is @xmath200 and an infinite backward trajectory starting at @xmath197 that is disjoint from @xmath36 and in particular this backward trajectory is longer than @xmath201 .
hence , we can choose numbers @xmath202 , @xmath203 and a point @xmath204 such that @xmath205 is not in the forward orbit of a critical point and such that for each @xmath206 we have @xmath207 .
in particular , we have @xmath208 .
thus , if we put @xmath209 then @xmath210 .
let @xmath211 , @xmath212 be two neighborhoods of @xmath12 . without loss of generality we can assume that @xmath213 . by lemma [ lem : caf1 ] , every backward branch of @xmath214 starting at @xmath143 and ending at some point @xmath215 can be modified to end at some @xmath216 .
the modification involves only removing at most @xmath217 last steps , that decreases @xmath218 at most by a constant factor because @xmath165 , and replacing them by at most @xmath219 steps , which stay in a uniformly bounded from below distance from critical points .
hence we conclude that @xmath220 and @xmath221 differ at most by @xmath222 .
this proves the lemma .
we denote by @xmath223 the maximal distortion of a map @xmath224 on a set @xmath225 .
we establish one preliminary approximation result . [ p.schuh ] given @xmath165 , a sufficiently small neighborhood @xmath106 of @xmath12 and an expanding safe point @xmath226 , for every @xmath167 there exists a uniformly expanding repeller @xmath227 such that @xmath228 we start by recalling the construction used in ( * ? ? ?
* proposition 2.1 ) to prove an analogous statement for @xmath115 and then we modify it using lemma [ lem : caf1 ] to prove the proposition .
as @xmath143 is expanding safe , there exist @xmath147 , @xmath229 and @xmath230 so that for all @xmath231 the map @xmath232 is univalent on @xmath233 and @xmath234 . hence
, in particular , the distortion @xmath235 is bounded from above uniformly in @xmath236 by some number @xmath237 .
given @xmath238 , let @xmath239 be the smallest integer satisfying @xmath240 .
hence , with the above , we have @xmath241 and @xmath242 , where @xmath243 and @xmath244 .
let @xmath245 be such that @xmath246 for any @xmath247 .
let us choose positive constants @xmath31 , @xmath248 and @xmath249 large enough so that @xmath250 and that for every @xmath251 , @xmath252 , @xmath253 for every point @xmath254 on the component @xmath255 the map @xmath256 is univalent and satisfies @xmath257 note that with this choice we have for large @xmath149 , @xmath258 as @xmath259 and @xmath260 covers @xmath21 , we can conclude that for every preimage @xmath261 there exists a component @xmath262 of @xmath263 contained in @xmath264 .
the map @xmath265 , and hence @xmath266 , is univalent on @xmath262 . thus , the map @xmath267 p_proof.eps ( 6,79)@xmath268 ( 40,71)@xmath269 ( 15.5,18.5)@xmath143 ( 102,25)@xmath270 ( -20,34)@xmath271 ( 65,25)@xmath272 ( 10,55)@xmath232 ( 40,5)@xmath121 ( 68,63)@xmath273 has no critical points , and @xmath274 is a uniformly expanding repeller with respect to @xmath275 . let us now slightly modify the construction of @xmath225 by and ignore all those backward branches @xmath276 that correspond to a point @xmath277 . given @xmath278 , let us consider the positive integers @xmath169 and the number @xmath168 provided by lemma [ lem : caf1 ] . then , by lemma [ lem : caf1 ] , for each point @xmath279 there exist numbers @xmath280 , @xmath281 , a point @xmath282 in @xmath283 .
any such branch stays @xmath284-far from @xmath36 .
note that the distortion of @xmath285 on @xmath286 is bounded by a constant @xmath287 independent of @xmath272 and @xmath149 .
given an integer @xmath288 , put @xmath289 note that for distinct @xmath272 and @xmath290 in @xmath291 the sets @xmath292 and @xmath293 are disjoint .
setting @xmath294 the sets @xmath295 are uniformly expanding repellers for @xmath296 and @xmath3 , respectively .
both of these sets are disjoint from @xmath12 by construction . on the other hand there , letting @xmath297 we have @xmath298 since @xmath299 , there is @xmath300 such that @xmath301 hence , if we put @xmath302 , then @xmath303 since @xmath304 , @xmath217 , @xmath305 are independent of @xmath149 and @xmath306 , we obtain the desired assertion by taking a sufficiently large @xmath149 .
we are now ready to prove one further equivalence .
[ sngk ] given a sufficiently small neighborhood @xmath106 of @xmath12 and an expanding safe point @xmath307 , for every @xmath308 we have @xmath309 by proposition [ p.schuh ] , we have @xmath310 . in view of lemma [ l.correct ] , to prove the inequality
@xmath311 it is enough to show that for each expanding repeller @xmath133 that does not intersect @xmath12 , there is a neighborhood @xmath312 of @xmath12 such that @xmath313 .
notice that for every @xmath136 and every neighborhood @xmath312 of @xmath12 disjoint from @xmath133 we have , @xmath314 this follows easily considering the contribution of the backward branches of @xmath315 contained in @xmath133 in the sum in .
let @xmath316 be a neighborhood of @xmath133 on which @xmath317 is uniformly expanding .
thus , there is a constant @xmath318 and for every @xmath319 there is @xmath320 such that for every integer @xmath117 and every @xmath321 there is @xmath322 shadowing @xmath323 and so that @xmath324 it follows that for each neighborhood @xmath312 disjoint from @xmath316 we have @xmath325 . by the eventually onto property of @xmath3 on @xmath21 , we have @xmath326 for some @xmath126 .
fix @xmath327 and let @xmath312 neighborhood of @xmath12 disjoint from @xmath316 .
then we have @xmath328 this shows @xmath329 and completes the proof of the inequality @xmath330 .
recall that @xmath3 is a rational map @xmath3 of degree at least @xmath2 .
if @xmath3 is exceptional , then @xmath12 is the maximal finite and forward invariant subset of @xmath21 satisfying @xmath331 .
otherwise @xmath332 . in the following proposition
we adapt the classical method by patterson and sullivan to construct a @xmath333-conformal measure on @xmath21 for each @xmath114 . for a map without a phase transition in the negative spectrum or for a map with a phase transition in the negative spectrum at some parameter @xmath334
, we obtain a finite conformal measure supported on @xmath21 , as in part @xmath111 of proposition [ p : known conformal ] . for a map having a phase transition in the negative spectrum at some parameter @xmath335
this construction gives us a conformal measure outside @xmath36 , which is finite outside each neighborhood of @xmath12 .
recall that by part @xmath2 of proposition [ p : known conformal ] , existence of phase transition implies that there does not exist a _ finite _
@xmath333-conformal measure for @xmath336 .
[ fsfjl ] let @xmath3 be a rational function of degree at least @xmath2 .
for each @xmath114 there exists a borel measure on @xmath21 that is @xmath337-conformal outside @xmath36 , finite outside any neighborhood of @xmath12 , gives zero measure to @xmath338 and whose support is equal to @xmath21 . as a first step we will apply the patterson - sullivan method while considering only those inverse branches outside a given neighborhood @xmath106 of @xmath12 .
we obtain in this way a measure that is @xmath339-conformal outside the set @xmath340 .
we will obtain a measure @xmath333-conformal outside @xmath36 by taking the limit of the measures obtained by repeating this construction with @xmath106 replaced by smaller and smaller neighborhoods .
we start with the following lemma .
recall that @xmath81 denotes the set of neutral periodic points in @xmath12 plus its preimages .
[ l : zero pressure ] given @xmath341 , for every @xmath342 there exist positive numbers @xmath343 and @xmath89 such that for every @xmath189 and every integer @xmath231 we have @xmath344 where the sum is taken over all @xmath345 satisfying @xmath346 for every @xmath347 .
let @xmath118 be sufficiently small so that for each periodic point @xmath348 of minimal period @xmath117 we have @xmath349 hence , there is some constant @xmath350 such for every integer @xmath231 and every point @xmath205 satisfying @xmath346 for every @xmath351 we have @xmath352 reducing @xmath343 if necessary , we may assume that for every @xmath348 the map @xmath3 is injective on @xmath122 and the set @xmath353 is disjoint from @xmath354 .
so for each @xmath348 and @xmath355 there is at most one point @xmath356 such that @xmath357 . by induction
we can conclude that for each @xmath231 , @xmath189 , and @xmath107 , @xmath358 there is at most one point @xmath345 such that @xmath359 thus the assertion follows with @xmath360 .
we now continue in proving the proposition .
let @xmath361 be an expanding safe point as provided by lemma [ lem : exisafe ] . given @xmath362 ,
let @xmath343 and @xmath89 be the positive numbers provided by lemma [ l : zero pressure ] .
reducing @xmath118 if necessary , we can assume that @xmath363 and by proposition [ sngk ] we can assume that @xmath364 satisfies @xmath365 .
there exists a sequence @xmath366 of positive reals such that @xmath367 and @xmath368 ( see , for example , ( * ? ? ?
* lemma 3.1 ) ) . given @xmath341 and @xmath369 ,
let us define @xmath370 and for each neighborhood @xmath371 of @xmath12 define the measure @xmath372 where @xmath373 denotes the dirac measure supported at @xmath16 .
observe that the measure @xmath374 is probabilistic for any @xmath369 .
if @xmath375 is a neighborhood of @xmath12 the measure @xmath376 is not probabilistic in general , however it is finite as shown in the following lemma .
let us denote by @xmath377 the total mass of a measure @xmath18 , that is , let @xmath378 .
[ lsls ] for every neighborhood @xmath371 of @xmath12 contained in @xmath106 there is a positive constant @xmath379 such that for every @xmath380 we have @xmath381 since by assumption @xmath382 , we have @xmath383 .
it only remains to prove the upper bound .
put @xmath384 note that @xmath385 is the subset of @xmath106 consisting of all points that under forward iteration do not leave @xmath106 for at least @xmath200 steps .
in particular , @xmath386 for all @xmath387 .
we will first consider a simple case and assume that we have @xmath388 for some integer @xmath389 .
we establish an upper bound for @xmath390 and hence for @xmath391 .
observe first that for every @xmath392 we have @xmath393 by for any point @xmath394 we have @xmath395 .
hence , we can estimate @xmath396 since @xmath397 , we have @xmath398 .
so , if we put @xmath399 and @xmath400 , we obtain @xmath401 now let @xmath402 and @xmath403 . since @xmath404 , we obtain by induction in @xmath200 that @xmath405 finally , recalling that @xmath406 , this proves the lemma in this first simple case that we considered . naturally ,
if @xmath407 then we can choose @xmath106 in such a way that @xmath408 and then it would be enough to consider the above case in which @xmath409 is eventually contained in @xmath371 .
however , if @xmath410 , that is , if @xmath12 contains a neutral periodic point in @xmath21 then this is not possible by the existence of siegel compacta ( * ? ?
* theorem 1 ) .
let us now consider the general case .
recall that @xmath411 .
let @xmath375 be an arbitrary neighborhood of @xmath12 .
certainly we can take @xmath412 sufficiently large such that @xmath413 . increasing @xmath200 if necessary
, we can assume that for every integer @xmath414 we have @xmath415 .
for each @xmath416 one of the following two cases can occur : either a ) @xmath417 or b ) @xmath418 , and hence @xmath419 in evaluating the latter term observe that to each point @xmath420 we may find some branch of preimages determined by a point @xmath421 satisfying @xmath422 and hence @xmath423 for every @xmath424 .
however , by lemma [ l : zero pressure ] , given any @xmath425 the contribution of all such branches can be estimated by @xmath426 where each sum is taken over all @xmath205 such that @xmath423 for every @xmath424 . thus , summing over all such branches that could occur , by our previous choice of @xmath200 and @xmath79 we can estimate @xmath427 note that @xmath428 . together with this completes the proof of the lemma .
given a neighborhood @xmath371 of @xmath12 contained in @xmath106 , as @xmath429 there exists a non - zero finite measure that is a weak * accumulation point of the family of measures @xmath430 .
furthermore , each such measure is @xmath333-conformal outside the set @xmath431 .
first observe that , by lemma [ lsls ] the total mass of any of the measures in @xmath432 is uniformly bounded from above and below by some positive constant
. hence this family of measures is relatively compact in the weak * topology and thus possesses a non - zero and finite accumulation point proving the first claim . following the construction in ( * ?
? ? * section 3 ) , we have for every special set @xmath89 disjoint from @xmath433 @xmath434 thus , @xmath435\\ & \phantom{=\frac{1}{m_{t , p}}\big\lvert } - b_1\sum_{x\in a\cap f^{-1}(z)}e^{\widetilde{p}(\varphi_t)-p}\big\rvert \\ & \le\frac{1}{m_{t ,
p}}\sum_{n=1}^\infty\sum_{x\in a\cap f^{-(n+1)}(z ) } b_{n+1}\big\lvert \frac{b_n}{b_{n+1}}-e^{\widetilde{p}(\varphi_t)-p}\big\rvert e^{p-\varphi_t(x ) } e^{s_{n+1}\varphi_t(x)-(n+1)p}\\ & \phantom{=\frac{1}{m_{t , p}}\big\lvert } + \frac{1}{m_{t , p}}b_1 \deg f \cdot e^{\widetilde{p}(\varphi_t ) - p}. \end{split}\ ] ] recall that , by the choice of @xmath436 in , we have @xmath368 and @xmath437 .
hence , we obtain @xmath438 uniformly in @xmath89 .
the assertion now follows like in ( * ? ? ?
* section 3 ) ( see also section 12.1 or lemma 12.5.5 and remark 12.5.6 in @xcite ) .
this proves the lemma .
we are now prepared to finish the proof of the proposition . note that in we use the same normalization factor @xmath439 for all measures @xmath376 for any neighborhood @xmath371 .
hence given @xmath369 for any pair of neighborhoods @xmath371 and @xmath440 of @xmath12 such that @xmath441 we have @xmath442 using a diagonal argument we can conclude that , as @xmath443 and @xmath444 , there exists a weak * accumulation measure @xmath445 of the family @xmath446 and that @xmath445 is @xmath447-conformal outside @xmath448 . replacing @xmath445 by the restricted measure @xmath449 ,
if necessary we can assume that @xmath445 does not give weight to @xmath338 and hence that @xmath445 is conformal outside @xmath36 .
lemma [ lsls ] and together imply that then @xmath445 is finite outside each neighborhood of @xmath12 .
finally , the fact that the support of @xmath445 is equal to @xmath21 follows from the property that @xmath3 is locally eventually onto on @xmath21 .
this finishes the proof of proposition [ fsfjl ] .
in this section we prove theorem [ main1 ] . in section [ sec : final-1 ] we make use of the bridges construction in @xcite to prove the lower bound , which follows along the same lines as in ( * ? ? ?
* sections 2.2 , 2.3 , 5 ) .
we point out that , after a careful observation , in fact it applies without any changes to our present more general setting . in section [ sec : final-1 ]
we also give another application of the bridges construction ( lemma [ help ] and its corollary [ def ] ) , which is used in section [ sec : final-3 ] .
the upper bound is shown in section [ sec : final-2 ] , where in the case @xmath114 we use the @xmath0-conformal measure given by proposition [ fsfjl ] .
we complete the proof of theorem [ main1 ] in section [ sec : final-3 ] by showing that the upper lyapunov exponent of each point in @xmath57 is at most @xmath27 .
we refer the reader to @xcite for all the notation in this subsection .
we call a point @xmath13 _ non - immediately post - critical _ if there exists some preimage branch @xmath450 , @xmath451 , @xmath252 that is dense in @xmath21 and disjoint from @xmath36
. there are at most finitely many non - immediately post - critical points . on the other hand , it is easy to see that each periodic point not in @xmath12 is non - immediately post - critical .
it follows that every uniformly expanding set disjoint from @xmath12 contains at least one non - immediately post - critical point . a set @xmath452 is called @xmath3-_uniformly expanding cantor repeller _ ( ecr ) if it is a uniformly expanding repeller and limit set of a finite graph directed system satisfying the strong separation condition with respect to @xmath3 .
the following proposition generalizes ( * ? ? ?
* proposition 1 ) . given an integer @xmath453 and
a function @xmath224 defined on @xmath21 put @xmath454 [ p.asympt ] there exists a sequence @xmath455 of positive integers and a sequence @xmath456 of subsets of @xmath57 , such that for each @xmath457 the set @xmath458 is @xmath459-invariant and uniformly expanding topologically transitive set , in such a way that for every @xmath460 we have @xmath461 recall the definition of the hidden hyperbolic pressure @xmath462 in . by propositions [ p.primeiro ] and [ sngk ] this pressure coincides with @xmath463 and
is obtained by taking a supremum over uniformly expanding repellers .
note that , given @xmath33 and @xmath167 and a uniformly expanding repeller @xmath452 , by ( * ? ? ?
* lemma 3 ) there exists a positive integer @xmath149 and an @xmath121-ecr @xmath464 such that @xmath465 note that in this case @xmath466 since @xmath467 .
hence @xmath468 contains non - immediately post - critical points .
note further that , given any two @xmath3-ecr s @xmath469 and @xmath470 that both contain non - immediately post - critical points , by ( * ? ? ?
* lemma 2 ) there exists an @xmath3-ecr @xmath471 containing @xmath472 and thus with pressure at least equal to the maximum of pressures of @xmath469 and @xmath470 .
based on these arguments , we can conclude that for any @xmath473 and @xmath167 we can find an integer @xmath117 and a topologically transitive @xmath121-ecr @xmath474 so that @xmath475 for all @xmath476 .
this proves the proposition .
the existence such an approximating sequence of repellers and ( * ? ? ? * theorem 3 ) together imply the following estimate , which is part of theorem [ main1 ] .
for @xmath477 we have @xmath478 consider now a family @xmath479 of @xmath459-ecrs as provided by proposition [ p.asympt ] and assume that the spectrum of lyapunov exponents of @xmath458 eventually contains any exponent in @xmath480 .
given @xmath31 , @xmath481 $ ] we can choose a sequence @xmath482 so that @xmath483 and @xmath484 and that each @xmath485 is a lyapunov exponent of @xmath486 . recall that hence for each @xmath457 there exist a unique number @xmath487 so that @xmath488 . moreover , there exists an equilibrium state @xmath489 for the potential @xmath490 with respect to @xmath486 with lyapunov exponent ( with respect to @xmath459 ) equal to @xmath491 and satisfying @xmath492 by proposition [ p.asympt ] and , we can conclude that @xmath493 if @xmath494 and @xmath495 if @xmath496
. together with ( * ? ? ?
* theorem 3 ) , this proves the proposition .
there is one more useful application of the bridges construction .
[ help ] given an expanding periodic point @xmath497 and @xmath167 , there exist a uniformly expanding repeller @xmath452 disjoint with @xmath12 of positive hausdorff dimension , containing @xmath107 , and an ergodic non - atomic measure @xmath18 supported on @xmath452 such that @xmath498 we start from the orbit @xmath499 of the periodic point @xmath107 .
as @xmath497 , @xmath107 is non - immediately post - critical .
hence , we can find a bridge from @xmath499 going back to @xmath499 and construct @xmath452 as in ( * ? ? ?
* lemma 2 ) .
we can then distribute a gibbs measure on @xmath452 choosing potential in such a way that probability of the backward branch going through the bridge is very small .
[ def ] in the definition of @xmath27 , instead of nonatomic ergodic measures one can use ergodic measures with support outside @xmath12 or ergodic measures giving measure zero to @xmath12 .
recall that a point @xmath16 is called _ conical _ if there exists a number @xmath500 , a sequence of numbers @xmath501 , and a sequence @xmath502 of neighborhoods of @xmath16 such that @xmath503 , the map @xmath504 is univalent on @xmath505 and that distortion @xmath506 is bounded uniformly in @xmath236 and @xmath16 by a constant @xmath507 ( the latter condition follows from the former one from koebe s distortion lemma by replacing @xmath343 by say @xmath508 ) . the following proposition will allow us to restrict our considerations concerning dimension to conical points with positive exponents .
[ conic1 ] the set of points @xmath13 that are not conical and satisfy @xmath509 has hausdorff dimension zero .
we are now ready to prove an upper bound for the dimension .
[ prop : upper ] let @xmath510 .
we have @xmath511 the proof will follow the same ideas as the proof of ( * ? ? ?
* proposition 2 ) .
the only difference is that in the case @xmath3 has a phase transition in the negative spectrum and @xmath512 , we will use a @xmath0-finite conformal measure constructed in section [ sec : bal ] instead of a conformal probability measure . by proposition [ conic1 ]
it is sufficient to study the subset @xmath513 of points that are conical .
recall that , by the frostman lemma , if there exist a finite borel measure @xmath18 and a number @xmath514 such that for every @xmath515 we have @xmath516 then @xmath517 ( see also ( * ? ? ? * theorem 7.2 ) ) . given a conical point @xmath194 with @xmath518 and @xmath519 , there exist numbers
@xmath520\setminus \{0\}$ ] , @xmath500 , and @xmath521 , and a sequence of numbers @xmath522 such that @xmath523 and that @xmath524 for all @xmath236 and all @xmath525 ( compare , for example , ( * ? ? ? * lemma 7 ) ) . by omitting finitely many @xmath526
we can assume that the right hand side of is not greater than @xmath343 .
replacing @xmath526 by @xmath527 and @xmath528 by @xmath529 if necessary , we can also freely assume that @xmath530 .
given @xmath515 , let us fix @xmath118 satisfying @xmath531 denote @xmath532 observe that @xmath533 does not intersect @xmath534 .
indeed , by our assumption it does not intersect @xmath36 .
further , if it intersected @xmath12 then either @xmath535 would intersect @xmath36 for some @xmath536 or @xmath537 would intersect @xmath12 .
the former is impossible because the map would not be univalent there ( we remind that @xmath538 ) , the latter is impossible because @xmath539 .
given @xmath33 let @xmath540 be the @xmath0-finite measure that is @xmath541-conformal outside @xmath36 as provided by proposition [ fsfjl ] and if @xmath542 then let @xmath540 be the finite @xmath541-conformal measure provided by proposition [ p : known conformal ] .
since in all cases @xmath540 is finite outside every neighborhood of @xmath12 , if we put @xmath543 then @xmath248 is finite and for every @xmath236 we have , @xmath544 as @xmath16 is conical , we have @xmath545 together with and this yields @xmath546 recall that this is true for every @xmath515 and @xmath99 .
now concluding as in the proof of ( * ? ? ?
* proposition 2 ) , using the frostman lemma , we obtain that @xmath547 , as wanted .
we establish the following gap in the spectrum of upper exponents .
[ prop : monic ] if @xmath3 is exceptional then @xmath548 for every @xmath549 .
we give two proofs of this proposition , one in this section and the other one in appendix [ s : alternative ] .
we will denote the spherical distance by @xmath146 .
recall that for a rational map @xmath224 and a critical point @xmath550 of @xmath224 we denote by @xmath551 the local degree of @xmath224 at @xmath60 . by corollary [ def ]
we have @xmath552 if and only if there is a periodic point in @xmath12 whose exponent is strictly larger than @xmath553 .
hence , to prove proposition [ prop : monic ] we need to control the exponent of any point @xmath184 whose orbit stays most of the time close to @xmath12 .
any orbit piece that shadows some ( periodic ) orbit in @xmath12 for a long time inherits its exponent , however right before it must have passed close to some critical point which results in a drop of the exponent .
let us make this more precise .
for @xmath554 let @xmath555 be the minimal integer such that @xmath556 is a periodic point , put @xmath557 and if @xmath3 is exceptional then we put @xmath558 we remark that there are examples where there is a point @xmath559 so that @xmath560 for some minimal number @xmath561 . if @xmath562 then @xmath563 . if @xmath564 then @xmath565 is a critical point in @xmath566 and @xmath567
thus , in none of these cases the `` essential exponent '' of @xmath568 and hence of an orbit piece that would shadow some periodic orbit @xmath569 , @xmath570 in @xmath12 would have large exponent . hence , in what follows we can restrict ourselves to the case that @xmath571 .
we have the following result . [ l.neu ]
suppose @xmath3 is exceptional .
let @xmath572 and let @xmath555 be the minimal integer such that @xmath564 is periodic . then there exist constants @xmath573 and @xmath124 such that for every @xmath189 near @xmath568 , but different from @xmath568 , and every integer @xmath574 such that for every @xmath575 we have @xmath576 , the following estimate holds @xmath577 put @xmath578 . to prove the lemma
, it suffices to notice that if @xmath284 is sufficiently small then for any orbit piece @xmath205 , @xmath579 , @xmath252 , @xmath580 that stays @xmath284-close to the periodic orbit of @xmath581 we have @xmath582 .
thus @xmath583 on the other hand , @xmath584 , so @xmath585 and @xmath586 .
given a subset @xmath106 of @xmath61 , let @xmath587 [ l.heuleni ] if @xmath3 is non - exceptional then there is an ergodic measure @xmath18 supported on @xmath21 and such that @xmath588 .
if @xmath3 is exceptional and @xmath106 is a neighborhood of @xmath12 , then one of the following cases holds : 1 .
either there exists an ergodic measure @xmath18 such that @xmath589 2 . or @xmath590 let @xmath106 be empty if @xmath3 is not exceptional and let @xmath106 be a neighborhood of @xmath12 otherwise .
without loss of generality in the latter case we can assume that @xmath106 is open .
let @xmath573 be given by lemma [ l.neu ] .
for each @xmath591 let @xmath592 be a point satisfying @xmath593 and consider the probability measure @xmath594 consider a measure @xmath18 that is accumulated by the sequence of measures @xmath595 in the weak * topology .
notice that @xmath18 is @xmath3-invariant and satisfies @xmath596 .
it follows that there is a @xmath3-invariant an ergodic measure @xmath597 such that @xmath598 . if @xmath3 is not exceptional or if @xmath3 is exceptional and @xmath599 , then we are done . to prove the remaining case ,
assume that @xmath3 is exceptional and @xmath600 .
fix @xmath601 and let @xmath573 and @xmath124 be given by lemma [ l.neu ] . augmenting @xmath602 and reducing @xmath284 if necessary we can assume that @xmath603 and that for each @xmath189 and each integer @xmath231 so that @xmath604 for every @xmath605 , we have @xmath606 let @xmath607 be a neighborhood of @xmath82 that is contained in @xmath608 and such that all preimages of a point in @xmath607 are either in @xmath607 or close to @xmath609 .
for an integer @xmath117 put @xmath610 and @xmath611 we have @xmath612 .
fix @xmath149 and let @xmath555 and let @xmath613 , @xmath252 , @xmath614 be all the integers @xmath615 such that @xmath616 and @xmath617 .
similarly , let @xmath618 be the number of blocks of trajectory of @xmath16 contained in @xmath619 .
for each @xmath620 let @xmath621 be the largest integer @xmath622 such that for each @xmath623 we have @xmath624 . then @xmath625 furthermore , for each @xmath626 so that @xmath627 the point @xmath628 is close to @xmath566 and we thus have @xmath629 hence , for @xmath630 we have @xmath631 since @xmath601 is arbitrary , this implies @xmath632 finishing the proof of the lemma .
[ l : periodicexcep ] suppose @xmath3 is exceptional . then for each @xmath572 and @xmath601 there is a periodic point @xmath633 close to @xmath568 such that @xmath634 in particular , @xmath635 .
let @xmath555 be the least integer such that @xmath636 is periodic , put @xmath578 and let @xmath231 be the period of @xmath107 .
let @xmath573 be sufficiently small so that there is a local inverse @xmath120 of @xmath232 fixing @xmath637 defined on @xmath638 , in such a way that @xmath639 is compactly contained in @xmath638 and for some constant @xmath640 and every @xmath117 and @xmath641 we have @xmath642 .
since @xmath643 there is a point @xmath644 and an integer @xmath126 such that @xmath645 and such that @xmath646 . let @xmath647 be sufficiently small so that the connected component @xmath371 of @xmath648 containing @xmath16 is such that @xmath649 and @xmath650 .
then for every sufficiently large integer @xmath117 there is a connected component @xmath651 of @xmath652 compactly contained in @xmath653 .
it follows that @xmath651 contains a periodic point @xmath654 of @xmath3 of period @xmath655 .
we will now estimate its lyapunov exponent .
since @xmath656 we have @xmath657 , so there is a constant @xmath658 such that for every @xmath659 we have @xmath660 therefore @xmath661 and @xmath662 . in view of lemma [ l.heuleni ] and corollary [ def ] , the proposition is a direct consequence of the lemma [ l : periodicexcep ] .
we finally state the following immediate corollary .
let @xmath663 where the maximum is taken over all critical points @xmath664 and where @xmath665 denotes the minimal integer such that @xmath666 is a periodic point .
we have @xmath667
the purpose of this section is to give an alternative proof of proposition [ prop : monic ] .
we will obtain this proposition as an easy consequence of lemma [ lem : caf2 ] below .
we shall conclude the appendix with a more precise version of this lemma , corresponding to bowen s _ periodic specification property _
, @xcite . [
lem : caf2 ] given a neighborhood @xmath668 , for every @xmath167 there exists a periodic point @xmath669 , so that @xmath670 .
we first collect some preliminary definitions and results .
recall that @xmath591 is said to be a
_ pliss hyperbolic time _ for @xmath16 with exponent @xmath671 if @xmath672 given @xmath167 , by the telescope lemma , see ( * ? ? ?
* lemma 9 ) or @xcite , there exist positive constants @xmath673 , @xmath674 so that for every @xmath675 , every pliss hyperbolic time @xmath149 for a point @xmath16 with exponent @xmath676 , and every @xmath677 , @xmath252 , @xmath678 we have @xmath679 where @xmath680 now let @xmath681 . let us briefly write @xmath682 . by definition of @xmath683
there exists @xmath684 so that for every @xmath685 we have @xmath686 on the other hand , for every @xmath149 large enough we can choose a point @xmath687 so that @xmath688 . notice that we can assume that @xmath149 is a pliss hyperbolic time for @xmath689 with exponent @xmath690 and satisfies @xmath685 .
more precisely for the original @xmath149 let @xmath691 be an integer such that at @xmath692 the expression @xmath693 attains its maximum .
clearly @xmath694 is a pliss hyperbolic time for @xmath16 .
moreover , since @xmath695 , @xmath696 , and the function @xmath697 is upper bounded , we obtain @xmath698 as @xmath699 .
so we can replace @xmath149 by @xmath694 , thus assuming we have a sequence of pairs @xmath700 so that @xmath701 , @xmath702 is a pliss hyperbolic time for @xmath703 with exponent @xmath690 , @xmath704 , and @xmath705 as @xmath706 . in the sequel
we shall omit the index @xmath707 .
we can assume that , possibly after slightly increasing @xmath175 , additionally we have @xmath708 .
indeed , let @xmath709 be the largest integer such that @xmath710 .
since we assume that @xmath149 is a pliss hyperbolic time for @xmath16 and since @xmath205 is close to a critical point , @xmath711 is small and hence the number @xmath712 must be large by . in particular , @xmath713 .
recall that @xmath12 contains periodic points together with their non - critical pre - images .
let the forward trajectory of @xmath205 follow a periodic trajectory of a point @xmath714 . denote by @xmath715 the least period of @xmath633
. then , by we have @xmath716 which yields @xmath717 provided @xmath106 is small enough , where the factor 2 takes in account the distortion in a neighborhood of the trajectory of @xmath633 .
so in the case that @xmath149 is a pliss hyperbolic time and @xmath718 we can consider the smallest integer @xmath719 for which @xmath720 . then there exists an integer @xmath721 between @xmath722 and @xmath723 which is a pliss hyperbolic time for @xmath16 with exponent @xmath724 .
note that @xmath722 and @xmath721 exist , provided @xmath725 and @xmath106 is small enough . by for any point @xmath726 with @xmath727 and @xmath728 we have @xmath729 thus , together with with @xmath730 , for any such @xmath205 we conclude @xmath731 and hence @xmath732 therefore , such @xmath205 must be in some distance to critical points and satisfy @xmath733 where @xmath734 is some positive constant . as concluded before ,
@xmath735 and @xmath12 are disjoint .
now we pull back @xmath736 and show that for @xmath343 small enough no pullback @xmath737 defined in contains a critical point . to show this
let us assume that the initially chosen @xmath343 also satisfies @xmath738 and that @xmath739 .
first , if @xmath457 satisfies @xmath740 we consider two cases : + * 1 ) @xmath741 : * then and for @xmath742 imply @xmath743 . + * 2 ) @xmath744 : * then implies that @xmath737 is very small .
so , if there were a critical point @xmath745 then we would have @xmath746 .
since @xmath12 is forward invariant , this would imply @xmath747 which is a contradiction .
hence @xmath748 .
+ second , in the remaining finite number of cases if @xmath457 satisfies @xmath749 we can assure , possibly after decreasing @xmath343 , not depending of @xmath750 ( possible since @xmath149 are pliss hyperbolic times with common @xmath671 ) , that @xmath748 .
thus , we can conclude that none of the pullbacks @xmath737 captures a critical point and therefore @xmath751 is univalent on @xmath737 for every @xmath677 , @xmath252 , @xmath678 . given @xmath343 , by lemma [ lem : caf1 ]
there exists @xmath168 and positive integers @xmath304 , @xmath217 , @xmath200 , and @xmath707 with @xmath169 and @xmath752 , @xmath172 , and a point @xmath753 such that @xmath754 is @xmath508-dense in @xmath21 and satisfies @xmath755 for every @xmath756 , @xmath252 , @xmath757 .
now we can choose @xmath758 independent of @xmath16 so that for this point @xmath753 and every @xmath756 , @xmath252 , @xmath757 we have @xmath759 yielding that @xmath760 is univalent on @xmath761 defined by @xmath762 for every @xmath763 as well as @xmath764 thus , if @xmath149 is large enough so that @xmath765 then with @xmath766 and using with @xmath767 we obtain @xmath768 and hence for some choice of @xmath763 ( that will in general depend on @xmath769 ) we have @xmath770 and @xmath771 is univalent on @xmath772 .
hence the latter set contains a periodic point @xmath107 . using and distortion estimates
, it can be achieved that @xmath773 , if @xmath149 is large enough .
this proves the lemma .
the idea of the proof of lemma [ lem : caf2 ] is taken from an unpublished note @xcite , where the supremum in was taken over all @xmath13 but allowing the periodic point @xmath107 to belong to @xmath12 .
the proof was simpler in that case as it did not rely on the telescope result .
the result in @xcite has been applied and referred to in @xcite .
another proof of the weaker statement , that is , allowing @xmath774 , can be given by constructing a measure @xmath18 that is an accumulation of the sequence of measures @xmath775 as @xmath699 , where @xmath776 denotes the dirac measure supported at @xmath205 . here
@xmath16 and @xmath149 should be chosen to give an approximation of @xmath683 by @xmath777 .
one finds @xmath107 using katok s method , see ( * ? ? ?
* chapter 11.6 ) .
this easy ` ergodic ' proof is in fact a part of the proof in section [ sec : final-3 ] .
notice that the proof in section [ sec : final-3 ] does not yield the periodic specification property below because it bases on a specification of , maybe short , special sub - blocks of a given piece of a trajectory . by lemma [ lem : caf2 ] for every @xmath778 we have @xmath779 by corollary [ def ] this
bound is less than or equal to @xmath553 .
this proves the proposition .
[ prop : specification ] for every @xmath106 and @xmath780 as in lemma [ lem : caf2 ] there exist an integer @xmath473 and @xmath781 such that for every point @xmath170 and @xmath782 with @xmath708 satisfying @xmath783 , there exists an integer @xmath784 and a periodic point @xmath785 of period at most @xmath786 , such that 1 .
@xmath787 + 2 .
@xmath788 for all @xmath789 .
we just look more carefully at the proof of lemma [ lem : caf2 ] .
consider an arbitrary small @xmath160 .
assume that @xmath16 and @xmath149 satisfy @xmath790 . by definition of @xmath682
we can also assume , compare ( [ e.defchi ] ) , that @xmath791 for all @xmath792 , for some constant @xmath793 depending only on @xmath794 .
then we find a number @xmath795 that is a pliss hyperbolic time for @xmath16 with exponent @xmath690 , as in the proof of lemma [ lem : caf2 ] . to estimate @xmath694 , notice that @xmath796 whereas @xmath797 hence , for @xmath798 , we obtain @xmath799 .
notice also that for @xmath800 we have @xmath801 for @xmath149 large enough since @xmath802 . in consequence ,
the positive integer @xmath803 maximizing @xmath804 is bigger or equal to @xmath805 .
finally we choose @xmath806 so that @xmath807 .
next in the proof of lemma [ lem : caf2 ] we have increased @xmath694 to achieve @xmath808 . here
the increase is not beyond @xmath149 since we have already assumed @xmath708 .
the rest of the proof is the same .
the mysterious indices @xmath809 in the assertion comes from lemma [ lem : at most four ] implying that at most fourth iteration of any point has a backward branch omitting critical points , hence @xmath810 .
kg has been supported by the alexander von humboldt foundation germany , cnpq and faperj , brazil .
the research of fp and mr were supported by the eu fp6 marie curie programmes spade2 and cody and by the polish mnisw grant nn201 0222 33 ` chaos , fraktale i dynamika konforemna ' .
jr - l was partially supported by fondecyt n 1100922 f. przytycki , j. rivera - letelier , _ nice inducing schemes and the thermodynamics of rational maps _ , preprint .
f. przytycki , j. rivera - letelier , and s. smirnov , _ equality of pressures for rational functions _ ,
ergodic theory dynam .
systems * 24 * ( 2004 ) , 891914 . | we study the dimension spectrum for lyapunov exponents for rational maps acting on the riemann sphere and characterize it by means of the legendre - fenchel transform of the hidden variational pressure .
this pressure is defined by means of the variational principle with respect to non - atomic invariant probability measures and is associated to certain @xmath0-finite conformal measures .
this allows to extend previous results to exceptional rational maps . |
entanglement entropy and other related information - theoretic quantities such as mutual information , are by now regarded as valuable tools to study different phenomena in quantum field theories and many body systems @xcite .
these quantities provide a new kind of information that can not be obtained from more standard observables such as two point correlation functions .
namely , both the entanglement entropy and the mutual information , are sensitive probes able to detect non - local signatures of the theory such as topological order which can not be detected by any local observable . concretely , the mutual information @xmath0 between two arbitrary regions @xmath1 and @xmath2 has certain advantages over the entanglement entropy .
first , @xmath0 can be viewed as an entropic correlator between @xmath1 and @xmath2 defined by , _
ab = s_a + s_b - s_a b , [ mutinf ] where @xmath9 is the entanglement entropy of the region @xmath10 and @xmath11 is the entanglement entropy of the two regions . by its definition , @xmath0 is finite and , contrarily to entanglement entropy , is non uv - cutoff dependent .
in addition , the subadditivity property of the entanglement entropy states that when @xmath1 and @xmath2 are disconnected , then , [ ssub ] s_a + s_b s_a b , which inmediatly leads to realize that @xmath12 .
subadditivity is the most important inequality which entanglement entropy satisfies .
a standard approach to compute both the entanglement entropy and the mutual information makes use of the replica trick @xcite .
unfortunately , these calculations are notoriously difficult to carry out , even in the case of free field theories . in the context of the ads
/ cft @xcite-@xcite , however , ryu and takayanagi ( rt ) have recently proposed a remarkably simple formula @xcite-@xcite to obtain the entanglement entropy of an arbitrary region @xmath1 of a @xmath13 dimensional cft which admits a classical gravity dual given by an asymptotically ads@xmath14 spacetime . according to the rt formula
, the entanglement entropy is obtained in terms of the area of a certain minimal surface @xmath15 in the dual higher dimensional gravitational geometry ; as a result , the entanglement entropy @xmath16 in a cft@xmath17 is given by the celebrated area law relation , [ hologee ] s_a = , where @xmath18 is the number of space dimensions of the boundary cft and @xmath19 is the @xmath18-dimensional static minimal surface in ads@xmath14 such that @xmath20 .
the @xmath21 is the @xmath22 dimensional newton constant .
the rt formula provides a simple tool to prove the subadditivity of entanglement entropy from the properties of minimal surfaces @xcite .
otherwise it has to be laboriously derived from the positive definiteness of the hilbert space .
here we consider the mutual information between two disconnected regions @xmath1 and @xmath2 in the ground state of an strongly coupled quantum field theory with a gravity dual given by the ads / cft correspondence . using ( [ hologee ] ) in ( [ mutinf ] ) , this quantity reads , [ holomi ] _
ab= , where @xmath23 is the area of the minimal surface related to @xmath24 .
recently , the holographic mutual information ( [ holomi ] ) has been considered in a quite remarkably amount of different settings @xcite-@xcite .
a striking prediction for the holographic mutual information arises when analyzing the behaviour of the minimal surface @xmath25 .
in @xcite it is shown how , for certain distances between the two regions , there are minimal surfaces @xmath26 connecting @xmath1 and @xmath2 . for those regimes ,
the holographic mutual information has a nonzero value proportional to the number of degrees of freedom in the gauge theory lying on the boundary of ads@xmath14 .
however , when the separation between the two regions is large enough compared to their sizes , then a disconnected surface @xmath27 with , ( _ a b^dis ) = ( _ a ) + ( _ b ) , is both topologically allowed and minimal . in this case , ( [ hologee ] ) yields @xmath28 and a sharp vanishing of @xmath0 then occurs .
this result is quite surprising from a quantum information point of view since , when the mutual information vanishes , the reduced density matrix @xmath29 factorizes into @xmath30 , implying that the two regions are completely decoupled from each other and thus , all the correlations ( both classical and quantum ) between @xmath1 and @xmath2 should be rigorously zero .
indeed , it seems , at least counterintuitive , that all the correlations should strictly vanish at a critical distance , in a field theory in its large @xmath31 limit .
this behaviour is a general prediction of the rt formula ( [ hologee ] ) which is valid for any two regions of any holographic theory . as a matter of fact ,
both ( [ hologee ] ) and ( [ holomi ] ) come from classical gravity in the bulk and provide the correct results to leading order in the @xmath32 expansion
. when the boundary field theory is a large @xmath31 gauge theory , these terms are of order @xmath33 .
thus , one might expect some corrections coming from quantum mechanical effects in the bulk theory , with the first correction appearing at order @xmath34 ( @xmath35 ) @xcite .
these @xmath35 order corrections are small enough not to modify the shape of the surfaces and , as have been argued in @xcite , jointly with the leading classical contributions , they obey the strong subadditivity condition . in this note it is shown that , at least in the case that has been considered , the mutual information ( [ holomi ] ) between two disjoint regions @xmath1 and @xmath2 in the large separation regime , while undergoing a phase transition at a critical distance , instead of strictly vanishing , decays with a law whose leading contributions are given by the exchange of pairs of the lightest bulk particles between @xmath1 and @xmath2 .
these bulk particles correspond to operators in the boundary field theory with small scaling dimensions as stated by the standard ads / cft dictionary @xcite-@xcite . in order to achieve this result , first we propose to interpret the mutual information in terms of correlators of surface operators . these can be realized in terms of a probe d3-brane using the ads / cft correspondence @xcite .
an operator product expansion ( ope ) for the long distance mutual information written in terms of these correlators is then provided .
the expansion is in accordance with a recent proposal given in @xcite where authors provide a long distance ope for the mutual information @xmath0 between disjoint regions inspired by an ope for the mutual information in cft previously discussed in @xcite and @xcite .
there , the expected leading contributions come from the exchange of pairs of operators @xmath36 located in @xmath1 and @xmath2 each with an small scaling dimension @xmath37 .
the ope reads as , [ eq0 ] _
ab~c_^_a^_b ^2~c _ ( ) ^4 + , where @xmath38 is the distance between @xmath1 and @xmath2 and @xmath39 comes from squares of ope coefficients .
thus , when considering a cft theory with a gravity dual , one must deal with a quantum field theory in a fixed background geometry and the long distance expansion for the mutual information reduces to an expression similar to ( [ eq0 ] ) , where now one should consider the exchange of the lightest bulk particles .
the direct computation of the one - loop bulk corrections to the holographic entanglement entropy and rnyi entropies of two wide separated disjoint intervals in a 1 + 1 cft has been explicitly addressed in @xcite . here
we ask if a simpler procedure can be used to learn , at least , some basic properties of the long range expansion of the @xmath0 in higher dimensional theories .
our aim is to provide an ope for the holographic mutual information in ads@xmath40 in terms of correlators of surface operators @xmath41 of the dual @xmath3 sym gauge theory . to this end , in this section , we first present some general properties of @xmath0 for subsystems that are weakly coupled to each other .
we show a result that foreshadows the long distance expansion ( [ eq0 ] ) on very general grounds .
then we review the twist operators and their role in computing the entanglement entropy and the mutual information in quantum field theory through the replica trick method @xcite . based on this , an ope for the long distance mutual information is given .
we also discuss on the strong analogies between the twist operators and the surface operators in gauge theories . we assume , following @xcite , that the nearly factorized density matrix of two subsystems @xmath1 and @xmath2 separated by a distance @xmath38 much bigger than their characteristic sizes is given by , [ eqa01 ] _ a b = _ 0 + _ 1 + ^2 _ 2 , where @xmath42 with @xmath43 , @xmath44 and @xmath45 . as a result , at order @xmath46 , the entanglement entropy @xmath11 may be written as , [ eqa01b ] s_a b & = & - tr [ _ a b _ a b ] + & = & - tr [ ( _ 0 + _ 1 + ^2 _ 2 ) ( _ 0 + _ 1 + ^2 _ 2 ) ] + & & - tr ( _ 0_0 ) - tr ( ( _ 1 + _ 2 ) _ 0 ) - ^2 tr ( _ 0 ^ -1 _ 1 ^ 2 ) + & = & s_a + s_b - tr ( ( _ 1 + _ 2 ) _ 0 ) - ^2 tr ( _ 0 ^ -1 _ 1 ^ 2 ) , so
, the mutual information at this order reads as , [ eqa01b2 ] _ ab ~ tr ( ( _ 1 + _ 2 ) _ 0 ) + ^2 tr(_0 ^ -1 _ 1 ^ 2 ) . thus , it is straightforward to realize that at first order in @xmath47 , the mutual information must vanish since @xmath47 could take either sign while @xmath0 is always non - negative . hence , the first non zero contribution to the mutual information is given by , [ eqa02 ] _ ab~^2
tr(_0 ^ -1 _ 1 ^ 2 ) , which does not depend on @xmath48 .
it can be shown that the @xmath46 term in eq.([eqa02 ] ) does not generically vanish . furthermore , since the non vanishing connected correlators between operators located in @xmath1 and @xmath2 are given by @xmath49 , then one might expect that , [ eqa03 ] _
ab~^2 _ a _
b^2 ~c ( ) ^4 , as far as @xmath50 .
this behaviour obeys the general bound given in @xcite , [ eq_wolf ] _ ab , where @xmath51 is the absolute value of the maximum eigenvalue .
we consider now the computation of the entanglement entropy of a region ( interval ) @xmath1 in a ( 1 + 1)-dimensional cft where @xmath16 is computed via the replica trick @xcite as , [ eq01 ] s_a=-_n tr _ a^n_n=1 = -_n tr _ a^n_n=1 , with @xmath52 the reduced density matrix of the region @xmath1 and @xmath53 .
the method relies on the computation of @xmath54 as a path integral over a @xmath55-sheeted riemann surface , each sheet containing a copy of the cft under consideration .
this path integral happens to be equivalent to the path integral of the symmetric product of the @xmath55 copies of the original cft ( whose central charge is given by @xmath56 ) , defined on a single @xmath57 sheet .
remarkably , @xmath54 can be written as the two point function of two vertex - like point operators @xmath58 and @xmath59 called twist operators , inserted at the two boundary points @xmath60 of @xmath1 in the path integral , i.e , [ eq02 ] tr _
n^-(v ) . the twist operators are actually primary operators with scaling dimensions @xmath61 related to the central charge of the cft and the number of replicas @xmath55
. they account for the conical singularities appearing as one joins the @xmath55 copies of the cft in the @xmath55-sheeted surface formulation of the path integral . in @xmath13 dimensions ,
one may also compute @xmath54 as a path integral over an @xmath55-sheeted riemann surface .
this multi - sheeted surface has a conical singularity along the boundary @xmath62 of the region @xmath1 for which one is computing the entropy .
it is expected that this path integral can be written as a path integral on a single - sheeted surface with an inserted twist - like operator @xmath63 $ ] defined along the boundary @xmath62 .
thus , @xmath64 \right\rangle $ ] and , in absence of further operator insertions , @xmath65 . here
, the operator @xmath66 $ ] is no longer point - like , becoming instead an extended operator such as a line operator in 2 + 1 dimensions or a surface operator in 3 + 1 dimensions . as pointed out in @xcite , a key realization about twist fields in a ( 1 + 1 ) dimensional cft is their resemblance with operators builded as the exponential of a massles field , i.e , a vertex operator in a free boson cft . in practice , the construction and properties of these twist fields beyond ( 1 + 1 ) dimensions
is poorly understood .
nevertheless , let us briefly discuss on how these higher dimensional @xmath63 $ ] operators exhibit significant analogies with extended operators in gauge theories .
assuming a vertex - like functional structure for a higher dimensional twist - field amounts to argue that it is the exponential of a certain type of massles spatial @xmath67-form @xmath68 , [ eq03 ] _
n _ a f^(d-1 ) ) , where @xmath69 must be fixed so as to obtain the correct prefactor for the entanglement entropy , which in a strongly coupled field theory is proportional to @xmath70 .
as long as the region @xmath1 is compact , it is easy to show that @xmath68 and thus @xmath63 $ ] has a `` gauge symmetry '' @xcite , [ eq04 ] f^(d-1)f^(d-1 ) + d^(d-2 ) , with @xmath71 an arbitrary spatial @xmath72-form .
let us to further illustrate the _ ansatz _ in eq.([eq03 ] ) by considering a scalar field theory @xmath73 in 3 + 1 dimensions and a set of @xmath55 replica fields @xmath74 .
these fields amount to a representation of the cyclic permutation subgroup of @xmath75 generated by the twist operator @xmath63 $ ] , _ n : _
n 1 n. in other words , the twist operator @xmath63 $ ] is the analog in the original multi - sheeted surface of moving from one sheet to the next ( previous ) one .
now , it is useful to introduce the linear combination of the replica fields , _ k _ j=1^n e^2i j _ j , k=0,1 , , n-1 , which are phase shifted by the factor @xmath76 as they encircle the codimension-2 spacetime region on which the twist operator is defined , i.e , they diagonalize the twist operator , _
n _ k = _ k _ k. namely , the twist operator @xmath77 $ ] can be written as a product of operators @xmath78 $ ] acting only on @xmath79 , _ n= _ k=0^n-1 _ n , k[a ] , with @xmath78 \widetilde{\phi}_{k'}=\widetilde{\phi}_k $ ] if @xmath80 and @xmath78 \widetilde{\phi}_{k}= \lambda_k\ , \widetilde{\phi}_k$ ] .
the way the field @xmath81 picks up the phase shift @xmath82 resembles the aharonov - bohm effect .
namely , since @xmath83 , one might introduce 2-form gauge fields @xmath84 to give account for these phase shifts .
these fields are normal gauge fields with a singular behaviour along the codimension-2 locus where the twist operator is defined . in a ( 3 + 1 ) dimensional theory ,
this locus amounts to a closed 2-dimensional surface .
therefore , the twist operator @xmath63 $ ] would be some 2-dimensional _ surface _ operator introducing a branch cut in the path integral over the @xmath55-fold replicated theory . in case
that the entangling surface @xmath62 is a static @xmath85 sphere , the twist operator residing on it , acts by opening a branch cut over the ball on the interior . noticing that @xmath75 acts on @xmath86 as a global @xmath87 charge symmetry , the twist operator @xmath78 $ ] can be defined by ( see eq.([eq03 ] ) ) , _ n , k~(i _ a f^(k ) ) , where @xmath84 encodes the flux which generates the phase shift @xmath82 .
a similar analysis has been carried out in @xcite in the two dimensional case , when the twist field is point - like and local .
there authors first discussed the interpretation of the twist fields as vortex - like operators . to finalize , we also note that the mutual information between two regions @xmath1 and @xmath2 can be written in terms of the twist operators @xmath63 $ ] and @xmath88 $ ] as , [ eq06 ] _
ab=_n_n=1 , which amounts to compute the connected correlation function between @xmath63 $ ] and @xmath88 $ ] . as an example , in cft@xmath89 , if one considers two disconnected intervals @xmath90 $ ] , @xmath91 $ ] @xmath92 such that @xmath93 and @xmath94 , then eq.([eq06 ] ) may be written as @xcite , [ eq07 ] _
ab=_n_n=1 , where @xmath95 are the point - like twist operators mentioned above .
it has been argued in @xcite that the minimal area prescription in eq.([hologee ] ) and eq.([holomi ] ) , though providing tempting hints about the structure of correlations in holographic theories at order @xmath32 , hides an important part of that structure in situations such as the long distance regime of the mutual information .
here we argue that it might result helpful to rephrase these quantities in terms of correlators of twist operators ( eq.([eq06 ] ) ) since , once taken this approach , it is in principle possible , to have an ope of these correlators from which @xmath96 , @xmath97 corrections to eq.([holomi ] ) might be obtained .
let us settle on this claim .
the twist operator @xmath63 $ ] can be expanded in a series of local operators @xmath98 when probed from a distance @xmath38 much larger than the characteristic size @xmath99 of the region @xmath1 as , [ eqa04 ] _ n= _
n ( 1 + _ i _ i^a(a , _
i,0 ) ^a_i(0 ) ) , where @xmath100 are the conformal dimensions of the operators .
the exact form of the expansion coefficients @xmath101 is unknown but generally , they should depend both on the scale @xmath99 and of the reference point at which the operator @xmath98 is inserted . here
, we have choosen the reference point as the center of the spherical region enclosed by the twist operator , i.e , the origin . for the sake of subsequent arguments in this paper , the operators @xmath98 are conformal primaries inserted at a single copy of the @xmath55-folded replica trick construction , while in general , they consist in products of two or more of such operators inserted at the same point but in different copies of the cft @xcite .
a similar expansion also holds for the twist operator @xmath88 $ ] defined along the boundary of a region @xmath2 with characteristic size @xmath99 located at a distance @xmath38 from the origin , [ eqa05 ] _
n ( 1 + _ j _ j^b(a,_j , l ) ^b_j(l ) ) .
thus , the opes and their coefficients @xmath102 appear as one replaces the regions @xmath1 , @xmath2 by a sum of local cft operators , the conformal dimensions and insertion points of the expansion coefficients @xmath103 , and @xmath104 . ] .
assuming that the vacuum expectation value of a single operator @xmath105 , the connected correlator in eq.([eq06 ] ) can be written as , [ eqa06 ] ~_i , j
j^b ^a_i(0 ) ^b_j(l ) .
however , recalling eqs.([eqa02])-([eqa03 ] ) , one notices that this ope for the mutual information should not be valid , as only involves @xmath106 terms @xmath107 contrarily to the expected @xmath108 ones @xmath109 .
let us fix this point by focusing on the 1 + 1 cft case .
if one performs a sort of ope such as the one given by eq.([eqa06 ] ) on the quantity within the brackets in eq.([eq07 ] ) , then the computation of @xmath0 singles out the term that is linear in @xmath110 .
it turns out that terms @xmath111 in that expansion are proportional to @xmath112 as shown in @xcite , and therefore , their contribution vanishes after doing the derivative and taking the @xmath113 limit . as a result , one might be compelled to consider an alternative ope for @xmath0 which , while using the long distance expansion for correlators of twist fields , takes into account eqs.([eqa02])-([eqa03 ] ) .
we first notice that the long distance expansion for the operator @xmath66 $ ] with a chiral primary operator ( cpo ) @xmath114 inserted at @xmath115 is given by , [ eqa07 ] = _ k^a
^a_k(0 ) ^b_k(l)~_k^a ( ) ^2_k , where @xmath38 is the distance between regions @xmath1 and @xmath2 and @xmath116 is the scaling dimension of the cpo @xmath114 .
similarly , the long distance expansion for the correlator of @xmath88 $ ] with a cpo @xmath117 inserted at @xmath62 is given by , [ eqa08 ] = _ m^b ^a_m(0 ) ^b_m(l)~_m^b ( ) ^2_m , with @xmath118 the scaling dimension of the cpo @xmath117 . as a consequence ,
it results reasonable to propose a long distance ope for the mutual information which jointly takes into account the long distance correlators of each one of the twist fields with all the cpo which one might find inserted on the other region .
this can be written as , [ eqa09 ] _ ab & ~ & _ n _ n=1 + & = & _ k c_k ( ) ^2_k + _ k m _ n_n=1 ( ) ^_k + _ m , with @xmath119_{n=1 } $ ] .
this `` ope '' accomodates to the very general requeriments for the behaviour of @xmath0 between weakly coupled regions showed above , while its coefficients are a byproduct of the ope between the twist fields and the cpo of the cft . at this point
it is worth to note that , while little is known about twist fields in higher dimensional cfts , not to say about the coefficients @xmath120 of the ope . as discussed above , those seem to be line or surface - like operators of a sort with analogous properties to the better known line and surface operators of gauge theories .
therefore , it might result tempting to access the properties of the mutual information in higher dimensional theories through the properties of these higher dimensional gauge operators , especially in situations where the benefits of computing through the ads / cft correspondence are manifest .
this also relates to the question of , up to what extent , some information theoretic quantities such as the mutual information might determine the underlying qft @xcite . in this sense , one may realize following @xcite , that as the entropy @xmath11 for very distant regions @xmath1 and @xmath2 approaches the sum of entropies @xmath121 , the vacuum expectation value ( vev ) of product of operators @xmath122 and @xmath123 defined on @xmath1 and @xmath2 , factorizes into the product of vev , so the exponential ansatz for @xmath0 , e^ _ ab = , where @xmath124 is a number , is exactly what one might expect in order to account for the clustering properties of correlators and entropies .
this ansatz is a mapping that must respect both poincar symmetry and causality .
the causality constraint imposes that @xmath122 , which in principle is a product of operators fully supported on @xmath1 , should be the same for all the spatial surfaces with the same boundary as @xmath62 .
this implies that @xmath122 must be localized on @xmath62 , which in more than one spatial dimensions , once more suggests that it may be some kind of generalized wilson loop operator of the theory under consideration . here
, it is worth to recall that in eqs.([eqa07 ] , [ eqa08 ] , [ eqa09 ] ) , one must deal with the correlators of the twist operators @xmath125 with the primary operators @xmath126 of the theory inserted at a single copy of the cft , for instance , the first of the @xmath55 copies . at this point , we follow @xcite in order to construct ( at least formally ) a surface - like effective twist operator @xmath127 which only acts within the first copy of the cft by reproducing any correlator of the form , _
n = _ n _ , [ effectwist ] where the subscript on the second correlator means that its computation is carried out on the first single copy of the cft . as in the two dimensional case , it is reasonable to assume that some roles of these effective twist operators , such as to impose the correct boundary conditions on the fields of the theory through their vortex - like singularities , are common to codimension-2 surface - like operators of the cft . under this assumption
, our approach here will consist in modifying eq.([eqa09 ] ) by means of the effective twist operator construction in eq.([effectwist ] ) and then to supersede @xmath128 \
, \mo\rangle_{\mathbf{1}}$ ] with the correlation function @xmath129 \ , \mo\rangle$ ] between a surface operator @xmath130 $ ] of the cft and a primary operator @xmath126 , with @xmath6 as the spatial surface on which the operators are defined . as a consequence ,
provided they can be computed , one may probe the long distance behaviour of @xmath0 by means of the ope between the surface operators @xmath131 $ ] , @xmath132 $ ] and the cpo of the gauge theory under consideration , & = & _ k^a
^a_k(0 ) ^b_k(l)~_k^a ( ) ^2_k , + & = & _ m^b ^a_m(0 ) ^b_m(l)~_m^b ( ) ^2_m , where coefficients @xmath133 depend explicitly on the characteristic size of the spatial regions @xmath1 and @xmath2 and both the insertion points and the scaling dimensions of the cpo .
finally , the long distance expansion for @xmath0 written in terms of these correlators reads as , [ su_mi : ope ] _ ab & ~ & _ k , m + & = & , where @xmath134 . the sums arise by considering all the possible local primary operators of the cft which one might expect to find inserted at each one of the surfaces @xmath135 .
this is precisely the scenario that will be considered in the remainder of this paper . as in the two dimensional case @xcite , the leading contributions to @xmath0 in eq.([su_mi : ope ] )
are controlled by the conformal primaries of the theory . nevertheless ,
while in ( 1 + 1 ) cft the expansion coefficients only depend on the correlation function of these operators , in the higher dimensional case , these coefficients non trivially depend on the geometry of the regions @xmath1 and @xmath2 as has been mentioned above .
we analyze the the mutual information between two static spherical 3-dimensional regions @xmath1 and @xmath2 with radius @xmath99 and separated by a distance @xmath138 , in the @xmath139 sym theory dual to type iib superstring theory on ads@xmath4 . to this aim , we first briefly review the holographic realization of surface operators in the gauge theory and then , using the arguments exposed above , a long distance expansion of the mutual information in terms of the correlators between these operators and the chiral primaries of the theory is provided .
there are different kinds of operators in a 4-dimensional gauge theory attending to the spacetime locus on which they are supported .
codimension-4 operators are point - like local operators that have been extensively studied in the ads / cft correspondence .
codimension-3 operators are one dimensional operators such as the wilson and thooft loops .
two dimensional surface operators @xmath7 are defined along a codimension-2 surface @xmath140 , where @xmath141 is the spacetime manifold on which the theory is defined .
the later were studied by gukov and witten in the context of the geometric langlands program , where they classified them in order to understand the action of s - duality @xcite . in a theory with a gauge group @xmath142 , gauge field , but indeed , for @xmath143 , there are different types of surface operators labelled by partitions of @xmath31 . ]
surface operators are disorder operators which , like thooft operators , can be defined by requiring the gauge field to have a prescribed vortex - like singularity along the surface @xmath6 : f = 2 _ + smooth , where @xmath144 is the gauge field curvature 2-form and @xmath145 is 2-form delta function that is poincar dual to @xmath6
. then , the new path integral is over fields with this prescribed singularity along @xmath6 .
this amounts to introduce a phase factor @xmath146 in the path integral by inserting the operator , ( i _ f ) .
thus , one needs to consider the path integral with a special prescribed singularity along a codimension-2 manifold @xmath6 .
the fields of the theory acquire the phase factors @xmath146 as they encircle the codimension-2 surface @xmath6 due to their singular behaviour near it .
as puzzling as they may seem , these singularities are rather ubiquitous in theories with vortex - like disorder operators such as the discontinuities induced on the fields of the theory by twist ( or effective twist ) operators in a higher dimensional cft . as in the case of a two dimensional cft ,
these discontinuities are consistent as far as the correlation functions of physical operators remain well defined .
some remarkable calculations involving disorder - like surface operators in the context of the ads / cft correspondence have been carried out both in a four dimensional gauge theory @xcite and in a three dimensional theory @xcite . in the large @xmath31 and large thooft coupling @xmath147 limit of the four dimensional @xmath3 sym theory ,
the vortex - like surface operators can be holographically described in terms of a d3-brane in ads@xmath4 with a worldvolume @xmath148 , where @xmath149 and @xmath150 ads@xmath40 is a volume minimizing 3-manifold with boundary , q = .
likewise , the holographic m - theory representation of a one dimensional vortex - like operator in the abjm three dimensional @xmath151 supersymmetric chern - simons theory @xcite , amounts to an m2-brane ending along one dimensional curve on the boundary of ads@xmath152 .
both descriptions are a probe brane approximation .
those are valid when the vortex - like operators under consideration have singular values only in the @xmath87 factor of the unbroken gauge group @xmath153 , which is the case that will be considered in this paper .
when the singular behaviour of the gauge fields are not such specifically restricted , then the disorder operators correspond to arrays of branes from which a pure geometric description in terms of regular `` bubbling '' geometries can be obtained @xcite .
we go back to the arguments given at the end of section 2 and thus consider the ope for the mutual information ( [ su_mi : ope ] ) written in terms of the correlators of surface operators @xmath154
with the chiral primary operators @xmath155 , [ eq9 ] , where @xmath156 or @xmath115 , are two static 2-dimensional spherical regions with radius @xmath99 , @xmath157 is the scaling dimension of the primary operator and @xmath38 is distance between them . as stated above , in the supergravity approximation , when @xmath158 and the thooft coupling @xmath159 , the surface operator @xmath154 is related with a d3-brane @xmath160 ads@xmath40 ending on the boundary of the spacetime with a tension given by @xmath161 ( in the units where the ads@xmath40 radius @xmath162 ) .
the correlator ( [ eq9 ] ) is calculated by treating the brane as an external source for a number of propagating bulk fields in ads and then computing the brane effective action @xmath163 for the emission of the supergravity state associated to the operator @xmath164 onto the point on the boundary where it is inserted @xcite . the prescription to compute this correlator is to functionally differentiate @xmath163 with respect to the bulk field @xmath165 .
this yields a correlator which scales with the distance @xmath38 as , [ eq9bis ] = - .
thus , in the following , the quantities that one might be concerned to compute , are the ope coefficients @xmath166 .
we will outline the calculations just below , but full details of it , can be found in @xcite . as a result , our proposal for the long distance expansion of the mutual information given in eq.([su_mi : ope ] ) , may be holographically realized in terms of the mutual exchange of bulk particles between the codimension-2 regions @xmath62 and @xmath115 on which the disorder surface operators @xmath7 are defined .
namely , its leading contributions should be given by the exchange of pairs of the lightest supergravity particles ( smaller scaling dimensions @xmath116 ) , while its coefficients arise as a byproduct of the ope coefficients appearing in the correlators of these surface operators with the chiral primary operators of the theory ( see figure 1 ) .
this proposal thus resembles the picture provided in @xcite . and @xmath2 ( shaded grey ) of radius @xmath99 separated by a long distance @xmath138 whose boundaries @xmath62 and @xmath115 define the surfaces @xmath167 and @xmath168 respectively ( the figure is represented in one lower dimension for convenience ) .
@xmath169 represents the _ radial _ coordinate in ads .
top left : the emission of a supergravity particle ( dotted line ) from the d3-brane realization of the surface operator @xmath170 onto a point ( x @xmath171 ) on the boundary of ads where the cpo @xmath172 is inserted .
bottom left : the emission of a particle from the d3-brane realization of the surface operator @xmath173 onto a point ( x @xmath174 ) on the boundary of ads where the cpo @xmath175 is inserted .
right : a leading contribution to the long distance ope for @xmath176 is given by the exchange of a pair of the lightest supergravity particles between the surfaces @xmath167 and @xmath168.,width=434 ] we outline the procedure to compute the correlation function ( [ eq9 ] ) .
the coupling of the supergravity mode @xmath177 ( dual to @xmath178 ) to the d3-probe brane realizing the operator @xmath7 is given by a vertex operator @xmath179 . this can be determined by expanding the d3-brane action @xmath180 to linear order in the fluctuations @xcite .
when the local operator @xmath181 emits the supergravity field @xmath177 at a point @xmath182 on the boundary , if it contributes to the correlator of @xmath178 with the surface operator @xmath7 , this supergravity mode propagates on the background ads@xmath4 and is then absorbed by the vertex operator , which must be integrated over the d3-brane realizing the operator @xmath7 in ads .
the bulk field @xmath177 has a simple propagator , however , it has a rather complicated set of couplings with the supergravity fields accounting for the brane fluctuations . in order to proceed ,
one may first write the scalar @xmath177 in terms of a source @xmath183 located at point @xmath182 on the boundary , [ eap1 ] s^(,z)=d^4 g _ ( ;,z ) s_0^ ( ) . here , @xmath184 is the bulk to boundary propagator describing the propagation of the supergravity mode from the insertion point @xmath182 of the cpo to the point @xmath185 on the d3 probe brane , [ eap2 ] g _ ( ;,z )
= c ( ) ^ , where the constant @xmath186 is fixed so as to require the normalization of the two - point correlation function @xmath187 . as the surface operator @xmath154 is probed from a distance @xmath38 larger than its radius @xmath99 , it is possible to approximate , [ eap3 ] g _
( ;,z ) c .
then , it is necessary to write the fluctuations of @xmath163 in terms of the field @xmath177 given by eq.([eap1 ] ) .
this inmediatly leads to determine @xmath179 .
furthermore , it also allows to write the linearized fluctuation contribution of the d3-brane action as , [ eap4 ] s_d3 = t_d3 d v _ s^ , with @xmath177 given in ( [ eap1 ] ) and @xmath161 . in the last expression @xmath188 refers to the volume element of the probe d3-brane .
the correlation function is obtained from functionally differentiating the previous expression with respect to the source @xmath183 , [ eap5 ] & = & - t_d3d d^4 v _ g _ ( ;,z ) s_0^ ( ) + & = & -t_d3 d v _
g_(_0;,z ) .
if we let @xmath189 to be parametrized as @xmath190 , then , integrating out this expression and using the approximation ( [ eap3 ] ) one thus obtains @xmath191 explicitly as @xcite , _ , p= c _ , p ( 1 + ( -1)^ ) , [ coeff_explicit ] where @xmath192 is the momentum of the scalar field in @xmath193 , @xmath194 is a parameter of the surface operator related with the geometric embedding of the d3-brane and @xmath195 is a constant related with the spherical harmonics in @xmath193
. for 10-dimensional supegravity compactified on ads@xmath4 , the ten - dimensional fields may be written as , [ sugra : fields ] = _ p , i _ p y_(p ,
i ) , where @xmath196 is a five dimensional field and @xmath197 are the spherical harmonics on @xmath193 with total angular momentum @xmath198 .
the full spectrum of 10d - supergravity compactified on @xmath193 was obtained in @xcite but , in what follows , we will focus only in the lightest scalars @xmath177 , whose exchange will dominate the long distance behaviour of @xmath0 .
these light scalar fluctuations couple to the @xmath3 sym operators @xmath178 of the lowest dimensions @xmath37 which appear in the ope for the surface operators @xmath7 and @xmath0 .
these states solve the klein - gordon equation in ads@xmath40 , [ eq13 ] _ ^s^=(-4)s^2 .
note that the field @xmath177 has a negative mass for @xmath199 .
however , these modes are not tachyonic , since they propagate on a space of negative curvature . in @xcite
it has been shown in full detail how to obtain the correlator between a surface operator and the lightest of these fields , i.e the scalar with @xmath200 . for @xmath201 this yields , [ eq14 ] = _ 2,0 ( ) ^4= ( ) ^4 which is of order @xmath34 . from eq.([eq14 ] ) one may determine the contribution of the lightest scalar ( @xmath202 ) to @xmath0 .
this amounts to the leading contribution to the long distance expansion given in eq.([su_mi : ope ] ) . defining @xmath203
, this expansion reads as , [ eq16 ] _ ab~(_2)^2 ( ) ^8 + = ( ) ^8 + ( l^-4 , 3 ) , which only depends on @xmath147 . as a result , it has been checked that the leading order of the long distance @xmath0 provided by the ope ( [ su_mi : ope ] ) , is @xmath204 .
this @xmath31 dependence is subleading with respect to the expected @xmath70 dependence which holds when a fully connected minimal surface @xmath26 between the regions @xmath1 and @xmath2 is allowed in an holographic computation .
thus , the holographic mutual information @xmath0 experiences a phase transition marked by a change in the @xmath31 dependence of its leading contributions but does not suffer a sharp vanishing due to large @xmath31 effects .
namely , it smoothly decays following a power law given by eq.([eq16 ] ) while parametrically saturates the bound given by eq.([eq_wolf ] ) .
in this note , we have investigated the structure of the quantum corrections to the holographic mutual information @xmath0 between two wide separated regions in the @xmath3 sym gauge theory dual to ads@xmath4 . to this end , first we have recasted the correlators of twist field operators related to the computation of the mutual information , in terms of correlators between _ surface _ operators in gauge theories .
namely , it is reasobable enough to claim that the twist field operators in a @xmath13 theory would be some kind of codimension-2 disorder - like surface operators .
as so little is known about the higher dimensional versions of the twist field operators , here we have only relied on the most basic analogies between them and the disorder - like surface operators .
it is worth to note that , by no means we have tried to establish an exact identification between them .
further investigations in this direction are surely needed in order to obtain some explicit ( holographic or field theoretical ) constructions of the twist operators in higher dimensions . in spite of this
, we feel that the commented analogies are strong enough to obtain valuable information about the @xmath31-dependence of the first non vanishing quantum corrections to the mutual information . under this assumption
, we have used the ads / cft realizations of the surface operators in the probe approximation , to provide a long distance expansion for the @xmath0 .
the coefficients of this expansion arise as a byproduct of the ope for the correlators of the surface operators with the chiral primary operators of the theory .
the results show that in the case under consideration , the mutual information @xmath0 undergoes a phase transition at a critical distance marked by a change in the @xmath31 dependence of its leading contributions .
namely , in the large separation regime @xmath205 , so instead of strictly vanishing , it smoothly decays with a power law shaped by the exchange of pairs of the lightest bulk particles between @xmath1 and @xmath2 .
the author is grateful to g. sierra , a.v .
ramallo and e. tonni for giving very valuable insights at different stages of this project .
jmv has been supported by ministerio de economa y competitividad project no .
fis2012 - 30625 .
i thank a.v .
ramallo and j. mas for their hospitality at universidad de santiago de compostela where this project took its initial steps .
v. balasubramanian , a. bernamonti , n. copland , b. craps , f. galli , `` thermalization of mutual and tripartite information in strongly coupled two dimensional conformal field theories , '' phys.rev .
* d.84 * 105017 , ( 2011 ) . | we investigate the behaviour of the mutual information @xmath0 between two `` small '' and wide separated spherical regions @xmath1 and @xmath2 in the @xmath3 sym gauge theory dual to type iib string theory in ads@xmath4 . to this end
, the mutual information is recasted in terms of correlators of surface operators @xmath5 defined along a surface @xmath6 within the boundary gauge theory .
this construction relies on the strong analogies between the twist field operators appearing in the replica trick method used for the computation of the entanglement entropy , and the disorder - like surface operators in gauge theories . in the ads
/ cft correspondence , a surface operator @xmath5 corresponds to having a d3-brane in ads@xmath4 ending on the boundary along the prescribed surface @xmath6 .
then , a long distance expansion for @xmath0 is provided .
the coefficients of the expansion appear as a byproduct of the operator product expansion for the correlators of the operators @xmath7 with the chiral primaries of the theory .
we find that , while undergoing a phase transition at a critical distance , the holographic mutual information , instead of strictly vanishing , decays with a power law whose leading contributions of order @xmath8 , originate from the exchange of pairs of the lightest bulk particles between @xmath1 and @xmath2 .
these particles correspond to operators in the boundary field theory with the smallest scaling dimensions .
[ ] |
sn1987a was the closest observed supernova in 380 years , and the evolution of the remnant of this supernova has been studied in great detail over the past twenty years . the radio intensity is growing and its rate of increase is increasing , with the spectral index being observed to flatten @xcite .
imaging at radio wavelengths has enabled the expansion of the inner ring to be traced , and is revealing increasing structure in the inner ring .
the x - ray fluxes observed with xmm and chandra also continue to rise almost exponentially @xcite .
around @xmath14000 days after the supernova the x - ray light curve increased dramatically , attributed to the arrival of the supernova blast wave at the equatorial ring of circumstellar material .
there is some evidence that the x - ray flux is mainly thermal , indicating the blast - wave interaction with dense matter @xcite .
@xcite compared radio and x - ray data with their calculations and concluded that there is a high ( amplified ) magnetic field inside the supernova remnant . the predicted tev gamma - ray flux , resulting mostly from the decay of neutral pions produced in interactions of the accelerated cosmic rays , is almost one order below the level obtained in previous searches by imaging atmospheric cherenkov telescopes ( iacts ) such as cangaroo - ii in 2001 @xcite and h.e.s.s . in 2003
the wide field of view searched by h.e.s.s .
did , however , yield marginal excesses in a region north - east of sn 1987a and at lmc x-1 .
cangaroo - iii is one of two stereoscopic iacts located in the southern hemisphere . with a three - fold coincidence
, the sensitivity is significantly improved from the previous single telescope ( cangaroo - ii ) .
we also increased the observation period by a factor of two in the 2006 observations compared to the 2004 and 2001 seasons .
we here report the results of these observations . the previous h.e.s.s .
analysis considered not only sn 1987a but also four other high energy objects within the same field of view @xcite .
we follow that idea and present upper limits on the integral gamma - ray fluxes from the super - bubble 30 dor c , the crab - like pulsar psr b0540@xmath069 , the x - ray binary lmc x-1 , and the supernova remnant n157b .
super - bubbles are expected to be efficient accelerators of cosmic rays @xcite and a shell feature near the star 30 dor c , within 5 arcmin of sn 1987a , shows an indication of non - thermal hard x - ray emission @xcite .
the energy required to produce the bubble is estimated to be 7@xmath210@xmath3erg , which is higher than generally expected for a single supernova remnant @xcite .
the other three sources belong to categories of high - energy astrophysical objects that may accelerate particles to tev energies , but , considering their large distance of 50kpc , they are less likely to be detected by current iacts than their galactic counterparts . however , as they are well within the field of view of the cangaroo - iii telescopes , and given the marginal h.e.s.s .
excess toward lmc x-1 , it is straightforward to search for any evidence of gamma - ray emission from them .
the cangaroo - iii stereoscopic system consists of four imaging atmospheric cherenkov telescopes located near woomera , south australia ( 31@xmath4s , 137@xmath4e ) .
each telescope has a 10 m diameter segmented reflector consisting of 114 spherical mirrors each of 80 cm diameter @xcite , providing a total light collecting area of 57.3m@xmath5 .
the spherical segments are mounted on a parabolic frame with a focal length of 8 m .
the first telescope , t1 , which was the cangaroo - ii telescope @xcite , was not used for these observations due to its smaller field of view and higher energy threshold .
the second , third , and fourth telescopes ( t2 , t3 , and t4 ) were used for the observations described here .
the camera systems for t2 , t3 , and t4 are identical and are described in @xcite .
the telescopes are located at the east ( t1 ) , west ( t2 ) , south ( t3 ) and north ( t4 ) corners of a diamond with sides of @xmath1100 m @xcite .
the observations were carried out in the periods from 2004 nov 11 to 14 ( mjd 5332053323 ) and from 2006 dec 12 to 27 ( mjd 5408154096 ) using wobble mode " in which the pointing position of each telescope was shifted in declination between @xmath60.5 degree every 20 minutes @xcite from the target : ( ra , dec [ j2000 ] ) = ( 83.866139@xmath4 , @xmath069.269577@xmath4 ) .
we took no off source runs .
the most sensitive region of the field of view is within one degree of the average pointing position .
lmc x-1 is located 0.6 degrees south - east from sn 1987a .
considering our angular resolution of 0.24 degree , there is some overlap between sn 1987a , 30 dor c , psr b0540@xmath069 , and n157b . in the 2004 observation ,
data with gps time stamps were recorded for t2 , t3 and t4 individually when more than four photomultiplier ( pmt ) signals exceeded 7.6 photoelectrons ( p.e . ) . in the offline analysis stage
we combined the data when the three telescope s gps times coincided . in the 2006 observation ,
a hardware coincidence was used to select any two triggered telescopes @xcite .
the images in all three telescopes were required to have clusters of at least five adjacent pixels exceeding a 5p.e .
threshold ( offline three - fold coincidence ) .
the event rate was reduced to 5@xmath18hz by this criterion depending on the elevation angle and year . based on time dependence of these rates
we can remove data taken in cloudy conditions .
the effective observation times for 2004 and 2006 were 632 and 1316 min , respectively .
the corresponding mean zenith angles were 42.1@xmath4 and 40.0@xmath4 .
the light collecting efficiencies , including the reflectivity of the segmented mirrors , the light guides , and the quantum efficiencies of the photomultiplier tubes were monitored by a muon - ring analysis @xcite with the individual trigger data in the same periods .
the light yield per unit arc - length is approximately proportional to the light collecting efficiencies .
deterioration is mostly due to dirt and dust settling on the mirrors .
here , we briefly describe the analysis procedures , which are identical with those described in @xcite .
more details can be found in @xcite and @xcite .
first , the image moments @xcite were calculated for the three telescopes .
the incident gamma - ray direction was determined by minimizing the sum of squared widths ( @xmath7 : weighted by the photon yield ) of the three images seen from the assumed position ( fitting parameter ) with a constraint on the distances from the intersection point to each image center . in order to derive the gamma - ray likeliness
, we used the fisher discriminant ( hereafter @xmath8 ) @xcite .
input parameters were @xmath9 where @xmath10 are energy - corrected @xmath11 and @xmath12 for the t2 , t3 , and t4 and assume that a linear combination of @xmath13 provides the best separation between signal and background , then the set of linear coefficients ( @xmath14 ) should be uniquely determined as @xmath15 where @xmath16 is a vector of the mean value of @xmath17 for each sample and @xmath18 is their correlation matrix .
we previously used it in vela pulsar analysis @xcite to separate `` sharp '' ( gamma - ray like ) images from `` smeared '' ( background ) ones .
the values of @xmath19 , @xmath20 , @xmath21 , and @xmath22 can be calculated from the monte - carlo and observational data ( off - source runs ) , respectively .
we rejected events with any hits in the outermost layer of a camera ( edge cut " ) .
these rejected events suffer from finite deformations , especially in the @xmath23 distribution , which results in deformations of the @xmath8 .
we then derived @xmath8 distributions position by position .
comparing those in the signal region and control background region , we can determine the gamma - ray like events .
here , we use monte - carlo simulations to determine the @xmath8 distribution of gamma - ray signals .
note that in the gamma - ray simulations we used a spectrum proportional to @xmath24 .
the fit of the @xmath8 distribution of source position with the above emulated signal and control background functions were carried out , to derive the number of gamma - ray like events .
this is a one - parameter fitting with the constraint that sum of signal and background events corresponds to the total number of events .
these coefficients can be derived exactly analytically .
the threshold of this analysis is considered to be @xmath11tev which is higher than the typical cangaroo - iii threshold due to the larger zenith angle of the observations , of around 40 degrees . in order to derive morphology
, we segmented the field of view into [email protected] degree@xmath5 square bins .
the @xmath8 distributions for corresponding bins are made and fitted .
the control - background region is defined as the second closest layer , which is more than 0.3 degree from the center of target region , i.e. , larger than the 0.24 degree point spread function ( psf ) .
the statistics of the control - background is , therefore , sixteen times larger than that of signal bin .
the results are shown in figs .
[ fig1 ] .
-3 cm -1 cm the left panel is obtained from the 2004 data and the right from 2006 .
the average telescope pointing position is indicated by the cross centered and labeled as
sn 1987a " , with the other four high energy objects within this field of view indicated by crosses .
our sensitivity is considered to be limited up to one degree in radius from the center .
the psf is considered to be a 0.24 degree radius circle .
the significance distributions ( excess divided by the statistical error ) are approximately normal ( gaussian ) distributions .
the best fit gaussians have mean values of @[email protected] ( 2004 ) and @[email protected] ( 2004 ) , and standard deviations of [email protected] ( 2004 ) and [email protected] ( 2006 ) , within systematic uncertainties .
at first we investigate the region of sn 1987a .
note that this region at some energies is contaminated by 30 dor c , a highly energetic source . to indicate the excess distribution from its center , we show so - called @xmath25 distributions in figs .
[ fig2 ] . here
, the control - background sample was selected in the region @xmath25=(0.10.2 ) [ degree@xmath5 ] .
the background - subtracted signals are shown .
both 2004 and 2006 data have statistically insignificant excesses near @xmath27 .
the @xmath28dofs ( degree of freedom ) for null assumptions are 18.6/15 and 13.2/15 for 2004 and 2006 , respectively .
these statistically insignificant excesses appear as the small excess count peaks around the centers of the plots in fig .
note that our psf corresponds to @xmath290.06 degree@xmath5 .
we , therefore , proceed assuming that there is no signal present . in order to show the consistency between 2004 and 2006 ,
the vertical unit is unified to count rate . in order to determine whether or not there is a gamma - ray excess around sn 1987a
, we made the @xmath8 distributions within the psf .
the control - background was made from the region @xmath25=(0.10.2 ) .
the fitting results are shown in fig .
[ fig3 ] .
the background level differs between the two years .
this is due to sky conditions , mirror reflectivity , time - dependence of the blur spot size ( which improves after maintenance periods ) , etc .
these are considered to be the major systematics for estimating the observational gamma - ray flux in this work . for the light correction efficiency
, we may have 10% uncertainty at maximum .
again there is no statistically significant excess .
we carried out a similar analysis , year by year , energy threshold by energy threshold .
the obtained 2@xmath30 upper limits ( uls ) are listed in table [ table1 ] . .the
2@xmath30 upper limits to the integral fluxes from sn1987a at two epochs with three different energy thresholds . [ cols="^,^,^,^ " , ] these are two year averages .
note that , considering our psf of 0.24 degree , sn 1987a and 30 dor c are highly confused , with other pairs being less confused . also some of the control - background regions are confused .
the acceptance is a slowly and smoothly decreasing function and is reduced to 65% at a distance of one degree from the center .
our morphology indicates a negligibly small excess near the center of field of view both in the 2004 and 2006 data .
this is statistically @xmath312@xmath30 each year .
we note that a similar feature can also be seen in the h.e.s.s .
data shown in figs . 2 and 3 of @xcite .
a deep survey , i.e. , longer observation , around this region is awaited .
the region around lmc x-1 is consistent with zero in our data .
the upper limits of sn 1987a in table [ table1 ] are plotted in fig .
[ fig4 ] .
the results presented here ( in blue and black ) are compared with cangaroo - ii ( red ) @xcite and h.e.s.s .
( green ) @xcite results .
the upper limits in this work are improved by factors of up to 10 on the 2001 result .
that is consistent with the longer observation time and better signal - to - noise ratio resulting from the three - fold coincidence .
however , the theoretical prediction even 7300 days after the supernova ( the dashed curve ) @xcite is still a factor of three below the results of the observations . note that the predictions @xcite were made in the frame of spherical symmetric approach .
the existence of strongly asymmetric interstellar matter structure , which is a dense inner ring , makes their prediction of gamma - ray flux uncertain . according to their rough estimation ,
this uncertainty is not very large ( about a factor of 2 ) due to the stronger supernova shock deceleration in the denser medium .
the solid curve is their prediction for 8249 days after the supernova , still lower than the upper limits obtained in this work . in order to constrain this model , we need further improvements on the hardware and/or analysis methods . a deeper observation by h.e.s.s .
, such as t@xmath3230 h , around @xmath12010 might give conclusive results .
also proposed future large - scale iact arrays such as cta @xcite would be very promising .
the limit obtained for 30 dor c in table [ table2 ] is about 15% crab level at @xmath11tev . considering the distance of 50kpc to lmc
, this upper limit corresponds to a huge cosmic - ray density comparing with that of the crab nebula ( distance of 2kpc ) .
even if there is supernova complex there , the expected tev gamma - ray flux is still below our sensitivity limit . increasing the cosmic - ray target density such as either inter - stellar - matter or photon density within plausible ranges
, the expected flux does not exceed this high upper limit . according to the standard estimation on the hadronic gamma - ray production such as @xcite with an assumption of the target density of 1p/@xmath33
, a total kinetic explosion energy of @xmath34ergs gives a 1 milli - crab flux of gamma - rays at the solar system .
again future larger - scale projects are required to probe such flux levels . for the other three high energy objects , we set upper limits far below those of previous observations .
we , however , could not find any physically important constraints on their activity at high energies again due to the distance of 50kpc .
we have observed sn 1987a at two epochs : mjd 5332053323 ( 2004 nov ) and mjd 5408154096 ( 2006 dec ) , approximately 6500 and 7200 days after the supernova , respectively .
the effective observation times are 10.5 and 20h , respectively .
no statistically significant gamma - ray signals were obtained and we set 2@xmath30 upper limits on the integral fluxes of 617% crab at @xmath11tev .
these are slightly lower than the previous cangaroo - ii and h.e.s.s.upper limits .
although theory predicts a factor of @xmath353 lower flux 7300 days after supernova , it also predicts that the future flux level might exceed the sensitivity limit of the existing and future arrays
. continued monitoring of this object at tev energies is still therefore meaningful in constraining theoretical models .
it is also important to improve the sensitivity of the observation methods , as proposed by the cta project @xcite .
in addition , we set 2@xmath30 upper limits for four more high energy objects inside our field of view : 30 dor c ( super bubble ) , psr b0540@xmath069 ( crab - like pulsar ) , lmc x-1 ( x - ray binary ) , and n157b ( supernova remnant ) .
we thank l.t .
ksenofontov for discussion on the estimated gamma - ray flux from sn 1987a .
this work was supported by a grant - in - aid for scientific research by the japan ministry of education , culture , sports , science and technology , the australian research council , jsps research fellowships , and the inter - university research program of the institute for cosmic ray research .
we thank the defense support center woomera and bae systems .
bamba , a. , ueno , m. , nakajima , h. , & koyama , k. , 2004 , apj , 602 , 257 berezhko , e.g. , & ksenofontov , l.t .
, 2006 , apj , 650 , l59 burrows , c. j. , et al .
1995 , , 452 , 680 daum , a. , et al .
1997 , astropart .
, 8 , 1 drury , l.oc . , aharonian , f.a . , &
vlk , h.j . , 1994 ,
a&a , 287 , 959 enomoto , r. , et al .
2002a , nature , 416 , 823 enomoto , r. , et al . 2002b , astropart .
phys . , 16 , 235 enomoto , r. , et al . , 2003 ,
apj , 591 , l25 enomoto , r. , et al .
2006a , apj , 638 , 397 enomoto , r. , et al .
2006b , apj , 652 , 1268 fisher , r. a. 1936 , annals of eugenics , 7 , 179 habert , f. , geppert , u. , aschenbach , b. , & hasinger , g. , 2006 , a&a , 260,811 .
hermann , g. , 2007 , cherenkov telescope array ( cta ) , http://www.mpi-hd.mpg.de/htm/cta/ hillas , a. m. proc .
19th int .
cosmic ray conf .
( la jolla ) 3 , 445 kabuki , s. , et al .
2003 , nucl .
instrum .
meth . , a500 , 318 kabuki , s. , et al .
2007 , apj , v667 , in press ( arxiv:0706.0367 ) kawachi , a. , et al .
2001 , astropart .
, 14 , 261 nasa , 2007 , skyview , http://skyview.gsfc.nasa.gov/ morris , t. , & podsiadlowski , p. , 2007 , science , 315 , 1103 naito , t. & takahara , f. , 1994 , j. phys . g. , 20 , 477 nishijima , k. , et al . 2005 , proc .
29th int .
cosmic ray conf .
( pune ) , og2.7 , 101 parizot , e. , marcowith , a. , van der swaluw , e. , & tatischeff , v. , 2004 , a&a , 424 , 747 park , s. , burrows , d.n . ,
garmire , g.p . ,
mccray , r. , racusin , j.l . , & zhekov , s.a . , 2007 , to appear in the proceedings of `` supernova 1987a : 20 years after supernovae and gamma - ray bursters '' , held in aspen co. usa , feb 19 - 23 , arxiv:0704.0209 rowell , g. , hinton , j. , benbow , w. , & h.e.s.s .
collaboration , 2005 , in aip conf .
745 , high energy gamma - ray astronomy , ed .
aharonian , h.j .
vlk , & d. horns ( melville : aip ) , 299 staveley - smith , l. , ball , l. , gaensler , b. , kestreven , m. , manchester , d. , & tzioumis , t. , 2007 , atnf news , issue no .
61 , issn 1323 - 6326 , http://www.atnf.csiro.au/news/newsletter/feb07/
ueno , m. , bamba , a. , & koyama , k. , 2003 , in `` the universe viewed in gamma - rays '' , proc .
univ . of tokyo
symposium , ed .
r. enomoto , m. mori , & s. yanagita ( universal academy press : tokyo ) , 311 wright , a. e. , griffith , m. r. , burke , b. f. , & ekers , r. d. 1994 , , 91 , 111 | optical images of sn 1987a show a triple ring structure .
the inner ( dust ) ring has recently increased in brightness and in the number of hot spots suggesting that the supernova shock wave has collided with the dense pre - existing circumstellar medium , a scenario supported by radio and x - ray observations .
such a shocked environment is widely expected to result in the acceleration of charged particles , and the accompanying emission of very high energy gamma - rays . here
, we report the results of observations made in 2004 and 2006 which yield upper limits on the tev gamma - ray flux , which are compared with a theoretical prediction . in addition
, we set upper limits on the tev flux for four high energy objects which are located within the same field of view of the observation : the super - bubble 30 dor c , the crab - like pulsar psr b0540@xmath069 , the x - ray binary lmc x-1 , and the supernova remnant n157b . |
supersymmetry is a prime candidate for new physics beyond the standard model ( sm ) . among various supersymmetric models
the most extensively studied is the minimal supersymmetric model ( mssm ) @xcite .
another popular supersymmetric model , which may be equally or more attractive compared with the mssm , is the next - to - minimal supersymmetric model ( nmssm ) @xcite since it can solve the @xmath6-problem and alleviate the little hierarchy .
these models will soon be put to the test at the lhc .
thus , their phenomenological study is important and urgent . although the most convincing evidence of supersymmetry is the detection of sparticle ( sfermions , gauginos or higgsinos ) productions , the indirect probe through detecting the quantum effects of virtual sparticles in some measurable interactions will play a complementary role .
if the sparticles are relatively heavy ( say above a few tev ) and hence can not be directly detected at the lhc , the indirect probe through quantum effects could be important . for this end
, the higgs boson yukawa interactions may play a special role since they can harbor sizable quantum effects of heavy sparticles when the higgs sector is not so heavy ( i.e. @xmath0 is not so large ) .
the dominant quantum effects of sparticles are from the supersymmetric qcd interaction and in the literature the calculations have been performed in the mssm for such supersymmetric qcd effects in the higgs boson yukawa couplings @xcite and the associated higgs production processes at the lhc @xcite . the studies in the decoupling limit with heavy sparticles showed that for a light higgs sector ( @xmath0 is not large ) the supersymmetric qcd can leave over large residual quantum effects in both the decay @xmath7 @xcite and the productions at the lhc @xcite . given the popularity of the nmssm , it is necessary to extend the study to the nmssm .
this is the aim of this work .
in this work we will focus on the supersymmetric qcd effects in the @xmath1 coupling ( @xmath2 is the lightest cp - even neutral higgs boson ) and perform a comparative study for the two popular supersymmetric models : the mssm and nmssm .
such a study is interesting for two points : * the study of the nmssm can accommodate the study of the mssm , and in some limit the nmssm results can reduce to the mssm results .
one can envisage that the supersymmetric residual effects in the mssm can be magnified in the nmssm .
so the future measurements of such effects may be useful in telling the difference of the two models . *
compared with the mssm , the nmssm may predict a different tree - level @xmath1 coupling and different loop contribution . in the nmssm the mass matrix and mixings of the higgs bosons
are enriched and thus the components of the lightest cp - even higgs boson @xmath2 are different from the mssm values .
the residual supersymmetric qcd effects in the nmssm may be larger and more interesting .
note that supersymmetry is a decoupling theory and all low - energy observables will recover their corresponding sm predictions when the mass scale of all supersymmetric particles ( including the masses of sparticles and @xmath0 ) take their heavy limits .
the large residual quantum effects of sparticles in the higgs yukawa couplings happen only in case that sparticles are heavy but @xmath0 is light . if both sparticles and @xmath0 take their heavy limit , the residual effects of supersymmetry do vanish . since so far such a split scenario ( with light higgs bosons and relatively heavy sparticles ) remains possible , we should check its phenomenological consequence .
we start our analysis by recapitulating the basics of the nmssm . in the nmssm
a singlet higgs superfield @xmath8 is introduced and the higgs terms in the superpotential are given by @xmath9 where @xmath10 and @xmath11 are the higgs doublet superfields , and @xmath12 and @xmath13 are the dimensionless constants .
note that there is no explicit @xmath6-term and an effective @xmath6-parameter is generated when the scalar component ( @xmath14 ) of @xmath8 develops a vev @xmath15 : @xmath16 .
the corresponding soft susy breaking terms are given by @xmath17 so the scalar higgs potential is given by @xmath18 where @xmath19 and @xmath20 are the coupling constant of @xmath21 and @xmath22 , respectively .
so we can see that in the limit of vanishing @xmath23 and with the input of the effective @xmath24 , the nmssm can reduce to the mssm . with the vevs @xmath25 , @xmath26 and @xmath15 ,
the scalar fields are expanded as @xmath27 the mass eigenstates can be obtained by unitary rotations of interaction states , e.g. , for the cp - even neutral mass eigenstates ( @xmath28 ) we have @xmath29 so the lightest cp - even neutral higgs boson @xmath2 is composed by @xmath30 in the mssm , @xmath2 can be decomposed in the same way except that without the singlet component , @xmath31 is a @xmath32 matrix and can be parameterized in terms of a mixing angle @xmath33 .
the sbottom squared - mass matrix in the nmssm is same as in the mssm with @xmath6 replaced by @xmath24 . in the basis of ( @xmath34,@xmath35 ) ,
the squared - mass matrix is given by @xmath36 where @xmath37 with @xmath38 , @xmath39 , @xmath40 and @xmath41 being respectively the isospin and electric charge of the @xmath42-quark , @xmath43 and @xmath44 being the soft breaking masses , and @xmath45 being the soft breaking trilinear coupling .
the sbottom mass eigenstates @xmath46 are obtained by the unitary rotation of the interaction eigenstates : @xmath47 where the unitary matrix @xmath48 parameterized by a mixing angle @xmath49 diagonalizes the mass matrix in eq.([eq : sbottommatrix ] ) .
the coupling of sbottoms with the singlet component @xmath50 of the higgs boson @xmath2 comes from the f - term of the superpotential , given by @xmath51 in terms of the mass eigenstates of sbottoms , the vertex @xmath52 takes the form @xmath53 where @xmath54 is only for the nmssm ( or equivalently set @xmath55 for the mssm ) , while @xmath56 and @xmath57 are present for both models .
they are given by @xmath58z_d^{1i}z_d^{1j } \nonumber\\ & & + \left[2m_b-\frac{2m^2_zs_w^2\cos^2\beta}{3m_b}\right]z_d^{2i}z_d^{2j}\nonumber\\ & & + a_b\left(z_d^{1i}z_d^{2j}+z_d^{2i}z_d^{1j}\right ) \
, , \\ c_u^{ij } & = & \frac{m^2_z\sin 2\beta}{2m_b}\left(1 -\frac{2}{3}s_w^2\right)z_d^{1i}z_d^{1j } \nonumber \\ & & + \frac{m^2_zs^2_w\sin 2\beta}{3m_b}z_d^{2i}z_d^{2j } \nonumber \\ & & + \mu_{eff}\left(z_d^{1i}z_d^{2j}+z_d^{2i}z_d^{1j}\right ) \ , , \label{eq : cu } \\
c_s^{ij}&=&\frac{\sqrt{2}\lambda m_w \sin\beta}{g_2 } \left(z_d^{1i}z_d^{2j}+z_d^{2i}z_d^{1j}\right ) .
\label{eq : css}\end{aligned}\ ] ] now we calculate the susy qcd corrections to the vertex @xmath1 . at tree level
it takes the same form in both the mssm and nmssm , given by @xmath59 the one - loop susy qcd corrections come from the feynman diagrams shown in fig.[eff_hbb ] where we do not show the self - energy loop of @xmath60
. in our calculations we use the on - shell renormalization scheme @xcite and take the external @xmath42 and @xmath60 quarks on shell to get the effective vertex . with the corrections
the effective vertex takes the form @xmath61 where @xmath62 denotes the vertex correction from fig.[eff_hbb](a ) and @xmath63 is the counter term from the renormalization of @xmath64 and the wave functions of @xmath42 and @xmath60 .
they are given by @xmath65 ( m^2_b , m^2_h , m^2_b , m^2_{\tilde g},m^2_{\tilde b_{i}},m^2_{\tilde b_{j } } ) , \label{eq24 } \\ \delta^{{\rm ct}}&= & - \frac{\alpha_s}{3\pi } \biggl\ { \frac{m_{\tilde g}}{m_b } \sin(2\theta_{\tilde b } ) \big[b_0 ( m_b^2,m_{\tilde g}^2,m_{\tilde b_1}^2 ) \nonumber \\ & & - b_0(m_b^2,m_{\tilde g}^2,m_{\tilde b_2}^2 ) \big ] - 2 m_b^2 \big[b_1^{\prime}(m_b^2,m_{\tilde g}^2,m_{\tilde b_1}^2 ) \nonumber \\ & & + b_1^{\prime}(m_b^2,m_{\tilde g}^2,m_{\tilde b_2}^2 ) \big ] - 2 m_b m_{\tilde g } \sin(2\theta_{\tilde b})\nonumber \\ & & \times
\big[b_0^{\prime}(m_b^2,m_{\tilde g}^2,m_{\tilde b_1}^2 ) - b_0^{\prime}(m_b^2,m_{\tilde g}^2,m_{\tilde b_2}^2 ) \big ] \biggr\ } , \label{eq : con}\end{aligned}\ ] ] where @xmath66 is the strong coupling constant , @xmath67 is the gluino mass , and @xmath68 , @xmath69 and @xmath70 are the scalar loop functions @xcite which can be calculated by using looptools @xcite .
in our calculation we use the package nmssmtools @xcite for the mass spectrum and the rotation matrix of the higgs fields ( to get the corresponding mssm results we take the limit of very small values for @xmath12 , @xmath13 and @xmath71 , and results are checked by using the package feynhiggs @xcite ) . to study the decoupling limit of susy particles , we assume all the soft breaking mass parameters ( @xmath43,@xmath44,@xmath67,@xmath45 ) and the parameter @xmath24 are degenerate , which are collectively denoted by @xmath72 . then in the mssm the susy parameters are ( @xmath72 , @xmath73 , @xmath74 ) , while in the nmssm there are three additional parameters ( @xmath12 , @xmath13 , @xmath71 ) . in our calculations we scan over these three additional parameters in the ranges @xmath75 note here we did not consider a large @xmath12 or @xmath13 . theoretically , the requirement of perturbativity up to some cut - off scale will set upper bounds on @xmath12 and @xmath13 at weak scale ( if the cut - off scale is chosen to be the gut scale , a stringent bound @xmath76 is obtained @xcite ) .
phenomenologically , a large @xmath12 or @xmath13 will incur stringent constraints from current experiments ( see the last reference in @xcite ) . before displaying the numerical results
, we make some clarifications regarding to our numerical calculations : * in our calculations we used the package nmssmtools @xcite which considered the loop corrections ( especially the stop / sbottom loops ) to the effective potential , the masses and mixing angles of the higgs bosons .
these corrections are important and can not be ignored .
* since we used the package nmssmtools in which the higgs mass matrices are diagonalized numerically , we did not use any approximate unitary transformation for our calculation .
* for the b - quark mass @xmath64 , in the package nmssmtools ( and thus in our calculation ) , it is taken as the running mass @xmath77 ( we take @xmath78 ) , which means that the sizable qcd loop effects are taken into account . actually ,
although both the tree - level coupling @xmath79 and the one - loop susy qcd contributions @xmath80 are proportional to @xmath64 and thus very sensitive to the value of @xmath64 , the relative correction effects @xmath81 ( displayed in our numerical results ) are not so sensitive to the value of @xmath64 .
now we present some numerical results . to see the general feature of the corrections , we first switch off the experimental constraints on the parameter space and in figs.[dcp_as ] and [ d_a ] display the resluts of the relative correction effects @xmath82
we see that for a light @xmath0 the susy qcd with a large @xmath72 can leave over sizable effects in the coupling @xmath1 . but
as @xmath0 gets heavy , such residual effects of susy qcd become small in magnitude , showing the decoupling behavior of supersymmetry .
compared with the mssm results , the effects in the nmssm can vary in a much broader region . while the corrections are always negative in the mssm , in the nmssm the corrections can be both negative and positive .
.some sample points in the parameter space of the nmssm , in which the susy qcd loop effects @xmath83 can be of very different size .
the mass parameters are in unit of gev . here
we fixed @xmath84 gev and @xmath85 . with such fixed parameters and @xmath72 varying from 400 gev to 2 tev
, the value of @xmath86 in the mssm varies from 106 gev to 119 gev while the corresponding values of @xmath83 in the mssm are about @xmath87 . [
cols="^,^,^,^,^,^,^,^ " , ] [ table-1 ] in order to figure out in what areas of the nmssm parameter space the one - loop susy qcd effects on the @xmath88 coupling are sizable , we present a set of sample points in table [ table-1 ] .
we see that for the susy qcd loop effects on the @xmath88 coupling to be sizable in the nmssm , the ratio @xmath89 ( @xmath90 and @xmath91 are respectively the components of @xmath92 and @xmath93 in @xmath2 , as defined in eq.[eq13 ] ) plays the key role for the following reasons . as shown in eq.([eq27 ] ) , @xmath83 is composed of two parts : the counter - term part @xmath63 and the vertex - loop part @xmath62 .
we found that in most of the parameter space allowed by the lep constraints , @xmath63 is negative and dominant in size , which is independent of how @xmath2 is composed of .
whereas , @xmath62 is positive and cancel @xmath94 to some extent .
the size of @xmath62 can be enhanced by the ratio @xmath89 , as shown in eq.([eq24 ] ) . although the ratio @xmath95 can also enhance the size of @xmath62 , its effect is suppressed by the smallness of @xmath96 and @xmath12 in @xmath54 ( in our scan we found that both @xmath12 and @xmath96 are small in order to satisfy the lep constraints encoded in the nmssmtools ) .
therefore , as @xmath89 gets large , @xmath62 becomes more sizable and , due to its cancellation effect , the total correction effects become less sizable .
note that although the value of @xmath86 is below 114 gev for some points , these points can still satisfy the lep constraints encoded in the nmssmtools because the lep bound on the mssm @xmath86 is 92 gev ( see fig.2 in the paper by barger et .
al . in ref.@xcite ) .
by the way , some comprehensive studies on the phenomenology of the nmssm higgs sector ( which may be quite different from the phenomenology of the mssm higgs sector ) have been performed in ref.@xcite , where all the higgs couplings and all the decay modes as well as productions have been intensively studied . from table [ table-1 ] we see that although @xmath12 is scanned in the range of @xmath97 , only small values of @xmath12 ( @xmath98 in magnitude ) survived the lep constraints encoded in the nmssmtools . as @xmath12 gets large , the mixing between singlet and doublet higgs fields becomes large ( although the mixing is not totally determined by @xmath12 and other parameters are also contributing ) and , consequently , is more constrained by lep experiments @xcite .
such a small @xmath12 , together with other nmssm parameters appearing in the mass matrix of the higgs fields , leads to a small mixing between singlet and doublet higgs fields .
as shown in table [ table-1 ] , the singlet component @xmath96 in @xmath2 is small , which means that the lightest higgs boson @xmath2 is not the singlet - like one and instead it is doublet - dominant ( mssm - like ) . for the @xmath88 coupling with such a mssm - like @xmath2 , the nmssm can still allow for quite different susy qcd effects compared with the mssm predictions .
the reason is that , as discussed above , the susy qcd effects are sensitive to the ratio @xmath89 , which is different in these two models ( even for a small @xmath12 , the nmssm can still allow for different - size components @xmath90 and @xmath91 in @xmath2 because other nmssm parameters are also contributing to the mass matrix and mixings of the higgs fields ) .
when we switch on the comprehensive experimental constraints encoded in the nmssmtools @xcite , we obtain the results of @xmath83 displayed in fig.[con_as ] .
these experimental constraints are comprehensive , including the lep ii searches for the higgs boson , various b - decays and the muon anomalous magnetic moment ( muon @xmath99 ) . from fig.[con_as ]
we see that in the special scenario under our consideration ( all soft breaking mass parameters are degenerate ) , @xmath72 is constrained in a certain range .
the range of @xmath100 gev is not allowed by lep experiments , while @xmath101 tev can not explain the muon @xmath99 data ( we require the supersymmetric effects to account for the deviation of @xmath102 at @xmath103 level )
. in the allowed region of the parameter space , the susy qcd corrections to the @xmath1 coupling can still be significant , over @xmath3 in magnitude .
if we switch off the muon @xmath99 constraint ( the hadronic contribution to @xmath104 is not so certain @xcite ) , the allowed parameter space gets much broader , as shown in fig.[con_as ] . in our study
we did not require supersymmetry to explain various plausible evidences of dark matter .
note that the susy contributions to the muon @xmath99 are sensitive to the soft masses in slepton sector and chargino sector , but not dependent on the squark or gluino mass involved in our susy qcd loops .
the stringent constraint from the muon @xmath99 data shown in fig.[con_as ] comes from our simple assumption that all soft masses ( in the squark sector , the slepton sector and the gaugino sector ) are degenerate .
of course , the muon @xmath99 constraint is not necessary ; without such an assumption of degeneracy , the constraint is lifted . since the susy qcd residual effects can be over @xmath3 for the @xmath1 coupling and thus over @xmath4 for
the ratio of the branching fractions ( susy qcd does not contribute to @xmath105 at one - loop level ) @xmath106 future measurements at the lhc or ilc may unravel such supersymmetric effects .
this ratio @xmath107 is proposed in @xcite as a probe to new physics . to measure this ratio at the lhc , one may count the event numbers of the production of @xmath1 followed respectively by the decay @xmath108 and @xmath105 , and the ratio of these two event - numbers can be a measure of @xmath107 ( the difference of efficiency for @xmath42-tagging and @xmath109-tagging should be taken into account ) .
finally , we make some remarks regarding to our results : * we only investigated the susy qcd corrections , which is @xmath110 and thus should be the most important among the susy corrections . among the susy electroweak corrections , the higgs - top yukawa corrections , which are @xmath111 , may also be sizable although they are seemingly not as large as the @xmath110 susy qcd corrections . with the consideration of the susy electroweak corrections , our conclusion will remain qualatively unchanged . *
since our main interest is the residual effects of heavy sparticles , we assumed all soft susy - breaking mass parameters are equal to @xmath72 .
this is a very strong simplifying assumption .
if we lift such an assumption and consider the multiple free soft parameters , we will obtain the results which numerically are different to some extent while qualitatively exhibit the same feature , i.e. , the nmssm allows broader corrections than the mssm , because the nmssm parameters can complicate the mass matrix and mixings of higgs fields .
* we should stress again that supersymmetry is a decoupling theory and the large residual quantum effects of sparticles in the higgs yukawa couplings are present only in case of a light @xmath0 .
if both @xmath72 and @xmath0 take their heavy limit , the susy effects will vanish .
such a decoupling behavior is similar in the nmssm and mssm , as shown in fig.2 .
we focused on the susy qcd effects in the @xmath1 coupling and performed a comparative study for the two popular susy models : the mssm and nmssm .
we found that for both models the susy qcd can leave over large residual quantum effects in @xmath1 coupling if the sparticles are relatively heavy ( a few tev ) while the higgs sector is not so heavy ( @xmath0 is not so large ) . compared with the mssm results ,
the nmssm can allow for a much broader region of such residual effects .
since these residual effects can be over @xmath3 in magnitude , future measurements of the @xmath1 coupling may unravel such supersymmetric effects or even distinguish the two models .
this work was supported in part by the national natural science foundation of china under grant nos .
10821504 , 10725526 and 10635030 , and by the natural science foundation of beijing under grant no . 1072001 .
99 for a review , see , e.g. , h. e. haber , g. l. kane , phys .
rept . * 117 * , 75 ( 1985 ) .
see , e.g. , j. r. ellis , _ et .
al . _ , 39 , 844 ( 1989 ) ; m. drees , int .
j. mod .
phys . a * 4 * , 3635 ( 1989 ) .
s. f. king , p. l. white , 52 , 4183 ( 1995 ) ; b. ananthanarayan , p.n .
pandita , 353 , 70 ( 1995 ) ; 371 , 245 ( 1996 ) ; int . j. mod
a12 , 2321 ( 1997 ) ; b. a. dobrescu , k. t. matchev , 0009 , 031 ( 2000 ) ; v. barger , p. langacker , h .- s .
lee , g. shaughnessy , 73,(2006 ) 115010 ; r. dermisek , j. f. gunion , 95 , 041801 ( 2005 ) ; g. hiller , 70 , 034018 ( 2004 ) ; f. domingo , u. ellwanger , 0712 , 090 ( 2007 ) ; z. heng , _ et al .
_ , 77 , 095012 ( 2008 ) ; r. n. hodgkinson , a. pilaftsis , 76 , 015007 ( 2007 ) ; 78 , 075004 ( 2008 ) ; j. cao , , h. e. logan , j. m. yang , arxiv:0901.1437 [ hep - ph ] ; j. cao , j. m. yang , 0812 , 006 ( 2008 ) ; 78 , 115001 ( 2008 ) ; c. s. li , j. m. yang , 315 , 367 ( 1993 ) ; a. dabelstein , 456 , 25 ( 1995 ) ; h. eberl , _ et al . _ , 62 , 055006 ( 2000 ) ; m. carena , d. garcia , u. nierste , c. e. wagner,577 , 88 ( 2000 ) ; h. e. haber , et al . , 63 , 055004 ( 2001 ) ; j. cao , g. gao , r. j. oakes , j. m. yang , 68 , 075012 ( 2003 ) ; g. gao , _ et al .
_ , 66 , 015007 ( 2002 ) ; 71 , 095005 ( 2005 ) .
d. j. miller , r. nevzorov , p. m. zerwas , 681 , 3 ( 2004 ) .
a. denner , fortschr .
* 41 * ( 1993)4 g. passarino and m. j. g. veltman , 160 , 151 ( 1979 ) ; g. t hooft and m. j. g. veltman , 153 , 365 ( 1979 ) .
t. hahn , m. perez - victoria , comput .
commun . * 118 * , 153 ( 1999 ) ; t. hahn , nucl .
* 135 * , 333 ( 2004 ) .
u. ellwanger , j. f. gunion , c. hugonie , 0502 , 066 ( 2005 ) . m. frank , _ et al .
_ , 0702 , 047 ( 2007 )
. j. p. miller , _ et .
. phys . * 70 * , 795 ( 2007 ) .
j. guasch , w. hollik , s. penaranda , 515 , 367 ( 2001 ) ; t. ibrahim , p. nath , 68 , 015008 ( 2003 ) . | if the sparticles are relatively heavy ( a few tev ) while the higgs sector is not so heavy ( @xmath0 is not so large ) , the higgs boson yukawa couplings can harbor sizable quantum effects of sparticles and these large residual effects may play a special role in probing supersymmetry at foreseeable colliders . in this work ,
focusing on the supersymmetric qcd effects in the @xmath1 coupling ( @xmath2 is the lightest cp - even higgs boson ) , we give a comparative study for the two popular supersymmetric models : the mssm and nmssm . while for both models the supersymmetric qcd can leave over large residual quantum effects in @xmath1 coupling , the nmssm can allow for a much broader region of such effects .
since these residual effects can be over @xmath3 for the @xmath1 coupling ( and thus over @xmath4 for the ratio @xmath5 ) , future measurements may unravel the effects of heavy sparticles or even distinguish the two models . |
fingerprint recognition , orientation field estimation , orientation field models , orientation field compression , orientation field marking , latent fingerprint recognition , fingerprint image enhancement , fingerprint matching
the orientation field ( of ) is a crucial ingredient for most fingerprint recognition systems @xcite .
an of is an image ( or matrix ) which encodes at each pixel of the fingerprint foreground @xcite the orientation @xmath0 degrees ( or @xmath1 in radians ) of a tangent to the ridge and valley flow at location @xmath2 .
modelling and estimating ofs is a fundamental task for automatic processing of fingerprints .
we introduce a locally adaptive global model called the _ extended quadratic differential _ ( xqd ) model .
we show that xqds can model the of of a fingerprint perfectly in the limit .
the major advantage of the xqd model lies in its small number of parameters , each of which has a simple and obvious geometric meaning .
ofs have many important areas of application at various stages of processing fingerprints . in the following part of this section
, we discuss some of the most relevant of these applications .
the rest of this manuscript is organized as follows . in section [ sec : ofliterature ] , we review related work from the literature for estimating , modelling and marking ofs of fingerprints . in section [ sec : xqdm ] , we describe the novel xqd model . in section [ convergence ] , we prove that the xqd model adapts perfectly to the of of a real fingerprint in the limit . in section [ sec : results ] , we present practical results for compressing real fingerprint ofs by xqd models .
we conclude in section [ sec : conclusion ] with a discussion and we point out topics for future work .
most systems for fingerprint verification and identification are based on minutiae templates .
automatic extraction of minutiae from fingerprints can be a very challenging task for images of low and very low quality .
quality loss of fingerprint images acquired on optical scanners can occur if a finger is too dry , or too wet , or contains scars . putting
a finger with too much or too little pressure on sensor can have similar negative effects on the image quality .
poor image quality can cause a minutiae extraction module to miss some true minutiae and to introduce some spurious minutiae .
the goal of fingerprint image enhancement is to avoid these two types of errors by improving the image quality prior to minutiae extraction .
the most effective approach for fingerprint image enhancement is contextual filtering and the most important type of local context is the of .
for example , oriented diffusion filtering @xcite uses only the of to perform anisotropic smoothing along the ridge and valley flow .
curved regions @xcite are computed based on the of and first , they are used for estimating the local ridge frequency and subsequently , of and ridge frequency estimates are joint inputs for curved gabor filtering @xcite .
these two methods estimate the local context and perform filtering in the spatial domain .
alternatively , methods for contextual filtering of fingerprints can also operate in the fourier domain , see e.g. methods proposed by chikkerur _ et al._@xcite and by bartunek _
et al._@xcite .
a hybrid approach with processing steps in the spatial and fourier domain has been suggested by ghafoor
_ et al._@xcite .
software - based fingerprint liveness detection strives to classify an input fingerprint as belonging to one of two classes : an image of an alive , real finger or an image of a fake or spoof finger made from artificial material like wood glue , gelatine or silicone .
developing countermeasures against spoof attacks is a very active research area .
two methods apply the of to compute invariant descriptors : for histograms of invariant gradients ( hig ) @xcite , the gradient direction at each pixel is normalized relative to the local orientation .
convolution comparison patterns ( ccp ) @xcite are obtained from small image patches . to that end ,
rotation - invariant patches are computed by locally rotating each window according to the local orientation at that pixel .
fingerprint alteration is another type of presentation attack @xcite in which the attacker has the goal of avoiding identification ( e.g. attempting to not being found in a watchlist search during border crossing or not being identified in a forensic investigation ) .
altered fingerprints often have a disturbed of .
therefore it is not surprising that in recent comparisons of features for alteration detection @xcite , some of the most effective features are related to the of . in a nutshell ,
dofts @xcite and ofa @xcite are based on the difference between an estimated of and a smoother version of it , coh @xcite relies on the coherence of gradients and spda @xcite makes use of the fact that alterations tend to introduce additional singularities into a fingerprint .
orientation descriptors have been proposed by tico and kuosmanen @xcite for computing the similarity between two minutiae from two templates .
these local similarities are aggregated into a global score which summarizes the similarity between both templates .
improvements of fingerprint recognition performance have been observed by using differences between two aligned ofs for score revaluation @xcite : first , two minutiae templates are matched and the output is a global similarity score and a minutiae pairing .
second , the corresponding ofs are aligned , based on the paired minutiae and , the similarity between the ofs is evaluated . on the one hand , if both ofs fit well together , the aligned ofs confirm the minutiae pairing and the global score is increased . on the other hand ,
if major discrepancies between the aligned ofs are observed , this is considered as an indication of a potential impostor recognition attempt and the score is decreased accordingly .
ofs are used for fingerprint alignment ( also known as registration ) , i.e. finding a global rotation and translation of one of with respect to other which is obtained by optimizing a cost function @xcite .
et al._@xcite considered in their work the alignment of partial fingerprints from fingermarks to full fingerprints , so called rolled fingerprints which are acquired with the help of e.g. a police officer who rolls the finger of subject to capture the full surface from nail to nail .
similar to above described score revaluation @xcite , they found that of alignment improves the recognition performance , in their case , the rank-1 identification rate .
tams @xcite studied the problem of absolute pre - alignment of a single fingerprint in the context of fingerprint - based cryptosystems @xcite and suggested an of based method .
the goal of classification and indexing is to speed up fingerprint identification ( 1 to n comparisons , where n can be in the magnitude of millions for forensic databases , see chapter 5 in @xcite ) .
for example , the class tented arch is observed in about 3% of all fingerprints @xcite . hence , if a query fingerprint belongs to the class tented arch , then the search space can reduced by 97% .
the majority of approaches for classification and indexing relies on ofs , e.g. cappelli _ et al._@xcite proposed a method for fingerprint classification by directional image partitioning . a recent survey by galar _
et al._@xcite lists 128 references and most of them use the of ( or its singular points ) for classification .
the generation of artificial fingerprint images has the advantage that it is possible to create arbitrarily large databases for research purposes e.g. of a million or a billion fingerprints at virtually no cost and without legal constraints .
methods for producing synthetic fingerprints include @xcite .
a detailed discussion of approaches for constructing and reconstructing fingerprint images can be found in @xcite .
all methods have in common that they require an of for the image creation process . the methods by cappelli _
_ @xcite and by araque _
@xcite rely on the global of model by vizcaya and gerhardt @xcite .
in contrast , the realistic fingerprint creator ( rfc ) @xcite uses of estimations by a combination of gradient - based and line sensor methods @xcite from a database of real fingerprints . during a forensic investigation , it can occur that traces at a crime scene are detected where two or more latent fingerprints overlap on a surface .
the task is to separate these fingerprints , so that the separated single fingerprints can individually be utilized for identification .
several research groups have addressed this problem in their work , and the key to the solution are in each work the ofs , see e.g. @xcite . a different forensic problem studied by hildebrandt and dittmann @xcite is latent fingerprint forgery detection . for this application
one may well compute rotationally invariant features such as hig @xcite or ccp @xcite by taking the orientation flow at the latent fingerprint into account .
considering the importance of the of , it is no surprise that a large body of literature is treating the topic of automatic of estimation .
a classic approach is to estimate the of by some form of averaging ( squared ) image gradients ( computed e.g. using the sobel filter ) or symmetry features , see e.g. @xcite .
however , this works only for good quality images .
further approaches include complex 2d energy operators @xcite . for dealing also with medium and low - quality fingerprint images ,
the line sensor method @xcite was developed which recently has been adapted to detect the oriented filaments in microscopy images @xcite .
a dictionary based method @xcite has been proposed for estimating the of in latent fingerprints .
many additional references can be found in @xcite and chapter 3 of @xcite .
the zero - pole model has been introduced by sherlock and monro in 1993 @xcite .
the flow fields generated by the zero - pole model resemble in some generality ofs of fingerprints , however they deviate significantly from ofs of a real finger .
vizcaya and gerhardt improved the simple zero - pole model in 1996 @xcite by suggesting an additional nonlinear bending scheme to better fit the of generated by their model to real ofs .
a global model based on quadratic differentials ( qd ) has been proposed by huckemann _
et al . _ in 2008 @xcite .
the zero - pole model is a special case of this more general model which has five geometrically interpretable parameters .
the qd model better fits real ofs especially for the fingerprint of the type arch .
further global models include the work by ram _ et al._@xcite who apply legendre polynomials for of modelling .
there are two main motivations for manually marking information in fingerprints .
the first is the creation of ground truth information which can be used for evaluating the performance of human experts and algorithms regarding the estimation or extraction of the target information .
and the second use case is the labeling for ( semi)automated retrival of information such as the foreground region , singular points , orientations or minutiae for fingerprint images which are too difficult for automatic processing by current state - of - the - art automatic fingerprint identification systems ( afis ) software .
forensic examiners mark such information in latent fingerprints to identify suspects in criminal investigations .
advancements in latent fingerprint recognition have the goal of minimizing the time and effort required by human experts for successful identifications . in order to compare the performance of different of estimation methods ,
1782 orientations at specific locations in various fingerprints in have been manually marked by gottschlich _
_ in @xcite with a focus on low - quality regions affected by noise .
et al._@xcite addressed the problem of enhancing very low - quality fingerprints and suggested to manually mark the ofs used for contextual filtering .
they proposed to mark local orientations , compute the delaunay triangulation and interpolate the orientation at unmarked pixel locations inside a triangle from the marked orientations at the three vertices of triangle .
a disadvantage of this approach is that a large number of small triangles is required to approximate the true orientation in highly curved regions around singular points .
et al._@xcite and turroni _
et al._@xcite created a ground truth benchmark called foe following the same marking strategy ( 10 good and 50 bad quality prints ) .
they compared the of estimation performance of several algorithms from literature on this benchmark .
the foe benchmark has recently also been used for evaluating the performance of methods which reconstruct ofs from minutiae @xcite . in our work , of compression results using the foe benchmark
are reported in section [ sec : results ] .
we note that xqd models can be used as an alternate interpolation method not suffering from the need of large numbers of support points at high curvature near singularities .
latent fingerprint recognition is still considered to be a difficult problem .
the level of noise for some fingermarks from crime scenes can be high and depending the surface from which fingermarks are lifted ( or directly photographed ) , a complex background can make the recognition task far more difficult in comparison with the processing fingerprints captured by a fingerprint sensor .
typical first steps , among them fingerprint segmentation @xcite and of estimation , are challenging .
recently , a novel image decomposition technique called dg3pd has been introduced which can better cope with these challenging images , see figures 9 and 10 in @xcite .
the goal of fully automatic latent fingerprint identification has not yet been achieved .
even state - of - the - art commercial latent identification software fails for a considerable amount of images and information still has to be manually marked by forensic experts in these cases .
e.g. in a work by yoon _
et al._@xcite , information about the region - of - interest ( roi ) , the location of singular points and the orientation at some sparse locations is still assumed to be manually marked . in the light of these problems ,
a subordinate target is to minimize the time and effort required by a human experts and the xqd model proposed in our work approach can be instrumental in achieving this .
forensic and governmental databases can contain millions of fingerprint images .
storing large volumes of data efficiently is a key issue which can be addressed by image compression @xcite .
et al._@xcite suggested to utilize the of for improving lossless fingerprint image compression .
more specifically , they suggest to increase redundancy by scanning pixels along the orientation , instead of standard procedures like horizontal ( row by row ) or vertical scanning of images .
larkin and fletcher @xcite proposed a method for lossy fingerprint image compression by decomposing an image into four elemental images which are highly compressible .
one of these four images , called the continuous phase , can be converted into an of and vice versa . both approaches by thrn _
et al._@xcite and by larkin and fletcher @xcite can profit from improvements of the of compression by our xqd models .
if in an application e.g. by a law enforcement agency , fingerprint images and their minutiae templates are stored together , an straightforward idea would be to reconstruct the of from the minutiae template . however , a recent evaluation of of reconstruction methods @xcite showed that all existing methods have weaknesses , and especially in proximity to singular points , all methods tend to be very inaccurate . in an analogy
, minutiae templates can viewed as a form of lossy fingerprint image compressions .
a survey of methods for reconstructing fingerprint images from minutiae templates can be found in @xcite .
recently , shao _ et al . _
@xcite studied fingerprint image compression by use of dictionaries of fingerprint image patches .
an additional discussion of texture image compression can be found in section 7.3 in @xcite .
the efficiency of xqd models for of compression will be detailed in sections [ sec : ofcompression ] and [ sec : results ] .
our methods for manually marking and automatically compressing fingerprint ofs are based on the _ quadratic differential _ ( qd ) model of huckemann _
et al._@xcite .
consequently , we shall outline that model first .
the basis of this model is given by a model for the arch type fingerprint .
adding given singular point coordinates , _ i.e. , _ cores and delta , this can be generalized to model the other fingerprint types .
the of of an arch type fingerprint is roughly controlled by two parameters : given a cartesian coordinate system @xmath3 in complex coordinates @xmath4 , the of is linked to the following complex function @xmath5 for @xmath6 and , otherwise , @xmath7 , as follows .
the orientation angle at the coordinate @xmath3 can be obtained by @xmath8 with the main branch of the argument of a complex number taking values in @xmath9 .
the parameter @xmath10 controls the coordinates of two singularities @xmath11 and @xmath12 ( 2nd order zeroes of @xmath13 ) along the abscissa and @xmath14 is a factor controlling vertical stretching . in fig .
[ fig : arch ] a reasonable fit of the qd model to an arch type fingerprint is visualized where @xmath15 , @xmath16 , and the rotation and translation of the coordinate system have been adjusted .
fingerprints of other types ( such as loops , double loops , and whorls ) contain an equal number of deltas and cores where a fingerprint can not contain more than two deltas / cores ; note that a whorl can be considered as a double loop in which the two cores agree or are of small distance .
the following formula extends eq .
( [ eq : arch ] ) to also model an of of a loop type fingerprint of which core and delta coordinates are encoded by the complex @xmath17 and @xmath18 , respectively : @xmath19 for @xmath6 and , otherwise , @xmath20 . here
@xmath21 denotes the complex conjugate of @xmath22 . an of of a loop type fingerprint modeled by the qd model is visualized in fig .
[ fig : loop ] .
similarly , a double loop with complex core coordinates @xmath23 and complex delta coordinates , @xmath24 is modeled by the following : @xmath25 for @xmath6 and , otherwise , @xmath26 . for both models , eq .
( [ eq : loop ] ) and eq .
( [ eq : doubleloop ] ) , orientation angles are computed via eq .
( [ eq : qdmangles ] ) . for a more comprehensive treatment of the qd model
, we refer the reader to @xcite and the literature therein on geometric function theory ; there the inverse @xmath27 of @xmath13 is considered giving the _ quadratic differential _
( qd ) @xmath28 the solution curves of @xmath29 having the orientations from eq .
( [ eq : qdmangles ] ) .
then , in particular the `` zeroes '' of @xmath13 are in fact poles of the qd and `` poles '' of @xmath13 are zeroes of the qd . as can be seen in fig .
[ fig : qdm ] , the qd model can be used to quite well approximate the general ridge flow using few parameters only
. however , the reader quickly recognizes areas in which the model significantly deviates from the evident ground - truth ridge flow which is an unavoidable effect due to the fact that @xmath13 in the qd model has only few degrees of freedom .
consequently , we need to change or extend the model . in this paper
, we propose to attach a variable number of local correction points to which we refer as _ _ thereby obtaining an _ extended quadratic differential _ ( xqd ) model . with these points the local of modeled by a qd can be corrected to better match with the ridge flow of a fingerprint .
[ [ section ] ] + an _ _ is a @xmath30-tuple @xmath31 where @xmath32 is a two - dimensional coordinate , @xmath33 an orientation angle , and @xmath34 and @xmath35 are two postive numbers .
more precisely , @xmath32 denotes a coordinate at which the orientation given by a qd model @xmath36 is to be corrected ; @xmath37 denotes the orientation angle of the true field at @xmath32 which is to become the new orientation angle there ; finally , @xmath34 and @xmath35 control how significantly the orientation correction influences the neighboring orientations around @xmath32 .
even more specifically , given the orientation angles @xmath38 of a qd ( see eq .
( [ eq : qdmangles ] ) ) , a true orientation @xmath33 at an @xmath39 the new orientation angles at any coordinate @xmath3 is computed as @xmath40 where @xmath41 $ ] denotes a correction angle .
the correction angle is defined as @xmath42 where @xmath43 denotes a function that assumes the value @xmath44 at @xmath45 and decays quickly to zero away from it . here
@xmath46 is the coordinate @xmath3 represented w.r.t .
a coordinate system defined by @xmath32 ( origin ) and @xmath33 ( rotation ) ; specifically , @xmath47 for example @xmath43 can be a tent function as in eq .
( [ tent - fcnt : eq ] ) with @xmath48 . to obtain a higher degree of smoothness , in the applications we use the two - dimensional gaussian @xmath49
similarly , given the of @xmath38 of a qd , the correction angle at @xmath3 can be defined recursively from a multiple number of @xmath50 as @xmath51 for @xmath52 and as in eq .
( [ eq : correctionangle0 ] ) for @xmath53 .
this yields our final xqd model @xmath54 in fig .
[ fig : xqdm ] the effect of correcting a qd model s of using an increasing number of is visualized . one important application of our xqd model is to manually mark semi - automatically a fingerprint s of by an expert . from the many choices of orders of tasks , by preliminary experiments
, we found the following strategy useful , the steps of which are visualized in fig .
[ fig : mark ] . 1 .
[ step : marksps ] manually mark the position of all cores and deltas of the fingerprint ( fig .
[ fig : marksps ] ) .
[ step : markinitialof ] manually mark an initial of ( possibly at sparse locations only , see fig .
[ fig : markinitialof ] ) .
[ step : adjustqdm ] adjust the qd model to the initial of by minimizing a suitable objective function , given by eq .
( [ eq : objectivefunction ] ) , say ( fig .
[ fig : adjustqdm ] ) .
[ step : insertaps ] successively insert to the xqd model further minimizing the objective function ( fig .
[ fig : insertaps ] ) .
the final xqd model agrees , within a preselected error bound , say , with the manually marked of .
this and other stopping strategies are discussed and illustrated in section [ sec : results ] .
+ given the of of an xqd model , _
@xmath55 , and an initial of @xmath56 , we can measure the deviation of the xqd model to the initial of by the following objective function , which , depends on all parameters of the xqd model : @xmath57 we note that , if steps [ step : marksps ] and [ step : markinitialof ] have been performed manually by an expert , the remaining steps can be implemented to run ( semi-)automatically by utilizing a steepest descent method applied to the objective function eq .
( [ eq : objectivefunction ] ) .
the key property of the xqd model is its ability to compactly represent , while having the power of arbitrarly well approximating , a fingerprint s of .
more specifically , recall that we count a total of at most @xmath58 real parameters describing a qd model : the parameters @xmath16 and @xmath15 ( see eq .
( [ eq : arch ] ) ) describing size and stretching , a two - fold parameter for translation , and one more parameter for rotation ; further , a fingerprint can contain at most two cores and two deltas . as an xqd model
is influenced by a variable number of each described by @xmath30 real parameters , an xqd model with @xmath59 consumes a total of at most @xmath60 real parameters , where @xmath61 is the number of singular points .
given an uncompressed of , an xqd model can ideally be approximated automatically with a small number of to compress the field . in fig .
[ fig : mark ] a manually marked fingerprint ( from foe , here assuming no ground truth of available ) has been modeled by a xqd with @xmath62 anchor points . at this point
we stress that the xqd model requires a reasonable estimation of the singular points even if they lie outside of the fingerprint s region of interest .
unfortunately , to date there is no method known that robustly estimates all singular points . beyond that
, however , we are able to automatically obtain an xqd model from an of thereby obtaining an effective method for compressing ofs .
it is general consent that fingerprint ofs are smooth except for the singularities at cores and deltas ( e.g. ( * ? ? ?
* section 3.6 ) ) . in consequence , denoting with @xmath63 the complex orientation of the true field at pixel location @xmath64 and denoting by @xmath65 the complex orientation of a qd model at @xmath64 with the same cores and deltas we may assume that there is a lipschitz constant @xmath66 such that @xmath67 for all @xmath68 in the observation window , while of course @xmath69 is unbound near the singularities . for the following
we assume that we have fit a qd model to a fingerprint s of with same singularities , such that we can assume ( [ lipschitz : eq ] ) for @xmath70 according to the algorithm introduced above , given an approximation @xmath71 to @xmath72 and a correction function @xmath73 , for our convergence considerations here we use not the one given by eq .
( [ eq : correctionangle0 ] ) but a tent function @xmath74 for suitable @xmath75 , we first show that the next iterate @xmath76 is closer to @xmath77 than the previous .
building on that we then propose an algorithm , theoretically assuring an asymptotically perfect adaption to the of . for fixed location @xmath78 and radius @xmath75 , with the lipschitz constant @xmath79 from ( [ lipschitz : eq ] )
, the following hold : 1 .
@xmath80 whenever @xmath81 , 2 .
if @xmath82 for some @xmath83 and @xmath22 with @xmath84 then @xmath85 3 .
choosing @xmath86 we have @xmath87 the first assertion follows from construction . for the second set @xmath88 . then @xmath89 taking the maximum of the last expression over @xmath90 yields ( ii ) .
( iii ) : with the choice for @xmath91 , setting @xmath92 , the right hand side of ( [ proof - lem1:eq ] ) attains its maximum @xmath93 at @xmath94 and the value @xmath95 at @xmath96 . with the algorithm of the following proof every fingerprint of can be perfectly adapted in the limit ,
i.e. @xmath97 we detail one iteration step of the algorithm and then show its convergence .
suppose after the @xmath98-st iteration , @xmath99 , we have an approximation @xmath100 with @xmath101 .
set @xmath102 and place a finite number @xmath103 of anchor points @xmath104 such that @xmath105 covers the fingerprint area , here @xmath106 .
setting @xmath107 , define @xmath108 and @xmath109 . in every step @xmath110 , due to ( iii ) ( b ) of the above lemma
, the approximation error within @xmath111 is below @xmath112 , everywhere else , due to ( iii ) ( a ) , the error will still be bound by @xmath113 . according to ( ii ) , the next iteration will change the error within @xmath111 to below @xmath114 .
since the mapping @xmath115 maps the interval @xmath116 to itself , after at most @xmath117 iterations , with some constant @xmath118 independent of @xmath113 , we have @xmath119 now suppose that the sequence @xmath120 would not converge to zero but to to @xmath121 . then due to @xmath122 and pointwise monotonicity in iterates , @xmath123 yielding a contradiction .
this proves that every of can be assymptotically perfectly adapted .
here we report compression results using the ten good quality ofs provided by @xcite as ground truths . as detailed in section [ sec : ofcompression ] , we have first manually marked singular points and afterwards automatically fit xqd models employing the following several optimization strategies .
stopping criteria and specific improvement steps in each iteration depend on the choice of to the main goal which can be : * as fast as possible ( minimal runtime ) in order to achieve a small deviation of the reconstructed of from the ground truth .
* as exact as possible ( minimal deviation from the ground truth of ) where we allow e.g. at most @xmath124 anchor points .
* as compressed as possible ( minimal file size of the stored xqd ) * as sparse as possible ( minimal number of anchor points ) note that in consequence of ( [ eq : numberparameters ] ) the model s sparsity relates directly to the compression rate : minimizing the number of anchor points is equivalent to a aiming for high compression , see table [ table : compress ] and [ table : compressfilesize ] . at every iteration step
several choices are possible .
one may optimize speed by simply adding a few anchor points without optimizing all possible parameters ( e.g. strategy s1 ) .
alternatively , when accuracy is optimized ( e.g. strategy s4 ) , in every iteration step not only all present anchor points are optimized but as well the choice of singular points and the other parameters of the underlying qd model are reconsidered . balancing the three main goals of speed , compression rate and accuracy of the reconstructed of allows for a range of intermediate strategies ( e.g. s2 and s3 )
results for four example strategies using a grid spacing of 12 pixels for ground truth orientation locations are reported in table [ table : compress ] .
@xmath125 @xmath126
in this work we have presented a semi - automatic tool based on a comprehensive xqd model for the orientation field of fingerprints , that achieves arbitrary precision at a very high compression rate in rather short time .
a compression by a factor of @xmath127 , say , at an accuracy of a few degrees in a few seconds ( see table [ table : compress ] and [ table : compressfilesize ] ) this semi - automatic tool can also be used for fast marking of orientation fields of fingerprints , be it for forensic application or in order to generate large orientation field benchmark databases . after labeling ( or accepting the tool s proposals ) of singular points , location , orientation and scaling of a fingerprint image , a very sparse representation of the orientation field with arbitrary precision
is fast and automatically built .
while in order to give a proof of concept , we have used the benchmark foe dataset , in future work orientation field estimation methods ( e.g. @xcite ) can be combined with our xqd model allowing for a next generation comprehensive low - dimensional fingerprint template consisting of minutiae plus segmentation ( e.g. @xcite ) plus anchor points ( xqd ) plus at most 13 parameters ( qd ) .
one may even consider to place the anchor points at minutiae locations , then only their @xmath128s need to be recorded .
additionally , ideal locations of anchor points
these give the deviation from a conformal qd model
deserve to be studied over large databases .
the authors gratefully acknowledge the support of the felix - bernstein - institute for mathematical statistics in the biosciences and the niedersachsen vorab of the volkswagen foundation .
stephan huckemann expresses gratitude to the support by the samsi forensics workshop 2015/16 .
r. cappelli , d. maltoni , and f. turroni . benchmarking local orientation extraction in fingerprint recognition . in _ proc .
. pattern recogn .
( icpr ) _ , pages 11441147 , istanbul , turkey , august 2010 . | fingerprint recognition is widely used for verification and identification in many commercial , governmental and forensic applications . the orientation field ( of )
plays an important role at various processing stages in fingerprint recognition systems .
ofs are used for image enhancement , fingerprint alignment , for fingerprint liveness detection , fingerprint alteration detection and fingerprint matching . in this paper , a novel approach
is presented to globally model an of combined with locally adaptive methods .
we show that this model adapts perfectly to the true of in the limit .
this perfect of is described by a small number of parameters with straightforward geometric interpretation .
applications are manifold : quick expert marking of very poor quality ( for instance latent ) ofs , high fidelity low parameter of compression and a direct road to ground truth ofs markings for large databases , say .
in this contribution we describe an algorithm to perfectly estimate of parameters automatically or semi - automatically , depending on image quality , and we establish the main underlying claim of high fidelity low parameter of compression . |
@xmath0 carinae is one of the most luminous objects in our galaxy ( @xmath4 ) and is well known for dramatic outbursts in which material is ejected outwards from the star .
its most famous mass ejection is thought to have occurred in the early 19th century , where a few solar masses of gas was expelled in a few decades .
this feature is clearly seen in observations as a bipolar nebula , known as the homunculus ( gaviola 1946 ) .
the origin of the shape of the homunculus is unknown , but could occur due to ejection of mass in a non - uniform interstellar medium i.e. if the ambient medium was denser in the equatorial plane , the nebula would preferentially expand along the poles . surrounding
the homunculus is a larger , less dense nebula , known as the ` outer ejecta ' ( e.g. weis 2005 ) which contains numerous filaments and dense condensates not dissimilar to those seen around the young supernova remnant cassiopeia a ( e.g. smith & morse 2004 , fesen et al 2001 ) .
the central star is hidden from view by this structure but is thought to be a luminous blue variable ( lbv ) star with mass @xmath5 ( pittard 1999 ; smith et al .
lbv stars are well known for large luminosities , violent instabilities and periods of large mass loss .
the presence of a binary companion ( e.g. pittard 2003 ) would provide an explanation for the catastrophic mass ejections and the strong variabilities seen in the radio and x - ray ( cox et al .
1995 ; duncan & white 2003 ) .
the cycle of periodicity suggests that any exisiting binary would have a highly eccentric orbit , leading to strong wind - wind collisions such as those seen in wr binaries ( marchenko et al .
2002 ; pittard 1999 ) .
the evolutionary phase of an lbv star can last more than @xmath6yrs and hence in this phase alone , a single star could lose roughly @xmath7 of gas .
episodic eruptions will increase the mass loss further .
it is therefore not unreasonable to assume that the dust mass loss rate could be @xmath8 . for the lifetime of the average lbv phase
, we could expect lbv stars to inject @xmath9 of dust into the interstellar medium and could be responsible for the dust emission seen around the young supernova remnants cas a ( dunne et al .
2003 , wilson & bartla 2005 ) , kepler ( morgan et al .
2003 ) , sn1998j ( pozzo et al . 2004 ) , the crab ( green , tuffs & pospescu 2004 ) , sn2002 ( barlow et al . 2005 ) and more recently sn2003gd ( sugerman et al . 2006 ) .
the structure of @xmath0 car is revealed in the many multiwavelength observations , yet there are issues surrounding the origin of the homunculus .
ir observations at 12 - 17@xmath1 m revealed a small structure within the homunculus which appeared to be concentrated around the ` waist ' of the nebula ( morris et al .
1999 ; hereafter m99 ) .
they found a small bright core a few arcsecs across which they interpreted as a disk / torus of dust with two components at 200k and 110k with @xmath10 of dust .
they suggested ( with a normal gas - to - dust ratio of 100 ) that the gas within this small torus could therefore be @xmath11 .
this argument is flawed since estimating the 110k dust mass relied on observations at the longer infrared wavelengths where resolution is limited to greater than 1 arcminute , so the precise location of the 110k dust component can not be determined from the mid - ir images alone .
davidson & smith ( 2000 ) also argued against the ir emission in a small torus , proposing instead that this was too small to account for the large luminosities radiated at 100@xmath1 m .
the visual image provides further evidence for this argument since there is no evidence of an opaque dusty disk on such a small scale .
hony et al .
( 2001 ) then used high resolution images at 8 - 20@xmath1 m to suggest that the observed ir dust is not warm dust at 110k , but actually hot dust at 200k with a strange overlapping ring structure in the middle of the nebula .
they argue that these rings are similar to those seen in sn1987a and are at different polar axes to each other .
if this scenario is correct , the total dust mass in @xmath0 car decreases to @xmath12 due to the strong dependence of dust mass with temperature ; the gas mass ejected in the homonculus would then be @xmath13 and the evolution of @xmath0 car would not be as violent as currently thought .
this claim has been disputed in smith et al .
( 2000 ; 2003 ) who propose that the strange smooth loop structure seen in the hony et al .
observations are simply due to a disrupted torus .
they point to compact knots and hotspots in the ir with filaments and arcs which could be due to a toroidal distribution of dust which has been disrupted by subsequent mass loss .
they also use temperature maps from iso data to show that cooler ` hidden ' dust may exist in the outer polar lobes of the homunculus and not in a compact torus .
m99 , davidson & smith ( 2000 ) & smith et al .
( 2002 ) have all argued that there is missing mass which has never been accounted for by the observations , and the cooler dust mass which has been found has not been resolved , so its precise location is unknown . if there is missing ejected mass , this will have important implications for the energy budget and estimated parameters of lbv stars as well as its future evolution . to determine if colder dust could exist further out from the centre of the star , wavelength measurements are needed beyond the mid - ir range ( davidson & smith 2000 ) which will dominate the dust mass .
we therefore use archived scuba observations of @xmath0 car to investigate the distribution of the dust and the ejected mass in one of the most studied objects in the nearby universe .
the sub - mm observations of @xmath0 carinae were originally observed by scuba on the jcmt in january 1998 in grade ii weather , simulataneously observing at 450 and 850 @xmath1 m ( with nominal beam size of 8 and 14 arcsec respectively ) .
the data was taken from the jcmt archive ( originally observed by henry matthews , program i d m97bc30 ) and consisted of a single jiggle map observation with a 120 arcsec chop throw in alt - az and position angle @xmath14 .
this was reduced using the standard data reduction pipeline surf ( sandell et al .
2001 ) with sky opacities obtained from cso - fits and skydips to give relatively stable optical depths of @xmath15 and @xmath16 ( see section [ sec : ext ] ) .
the remsky option was used with the outer ring of bolometers flagged as sky bolometers ( as @xmath0 car is almost a central point source in the maps ) and the option for addback was turned on .
the 450@xmath1 m scuba image of @xmath0 car is shown in fig .
[ eta ] with 850@xmath1 m contours overlaid .
white contours start at 5-@xmath17 ( 160 mjy / beam ) incrementing by 2-@xmath17 with black contours starting at 10-@xmath17 .
the central part is not resolved by the scuba field - of - view but we see an interesting feature which appears to be extended in the x direction , along the midplane of the image which is particularly clear at 850@xmath1 m .
it is difficult to determine the significance of this since the source is at such a high airmass .
it is well known that larger chop throws can cause ` smearing ' along the direction of the throw , in this case the x - axis , and the image of @xmath0 car is similar to example images of this problem .
extended emission along the midplane is also seen in the calibrator crl618 at 850@xmath1 m ( figs [ etasubmm ] ( b ) & ( d ) ) , suggesting it may be a combination of the chopping method and the beam pattern at high airmass .
the extension is not as clear in the 450@xmath18 m image of the calibrator .
radial profiles of both the beam ( taken from the calibrator ) and the source show that the extension can not be separated from the beam pattern .
the high airmass of @xmath0 carinae ( @xmath19 ) at the time of observation produces potential difficulties with calibration which we will address carefully in this section .
the calibrator observed on the same night was at a lower airmass of 2.5 .
calibration of the data requires two steps , firstly an extinction correction for the absorption of incoming radiation by the atmosphere , and secondly a comparison with a calibrator of known flux to convert instrumental units ( v ) into jy . for the time the data was taken ,
the jcmt has two methods available to measure the value of the zenith sky opacity ( @xmath20 ) at the wavelengths of interest ; skydips and extrapolation from the cso 1.1 mm @xmath20 .
skydips are taken at the scuba wavelengths ( 450 and 850@xmath1 m ) but are only made infrequently ( every two hours or so ) .
the cso radiometer measures the @xmath20 at 1.1 mm every 10 minutes , and so samples changes in the atmosphere on a much more useful timescale .
the jcmt have created relationships between the cso @xmath20 values and those at 450 and 850@xmath1 m using an extensive data - base of skydips .
they also fit the cso @xmath20 data for each night with a polynomial which allows interpolation between measurements and reduces the effect of any ` spikes ' in the cso measurements .
these polynomial fits are the recommended way to extinction - correct scuba maps . on the night
the data were taken , the sky was extremely stable .
the plot of the cso @xmath20 values , the polynomial fit and the residuals is shown in fig .
[ csotauf ] .
the part of the night considered to be ` stable ' after sunset and before sunrise ( when the @xmath0 car data were taken ) is shown between the dashed lines . the value of @xmath20 which is used to correct the flux of the object for atmospheric absorption , is uncertain - and this uncertainty , @xmath21 , leads to an uncertainty on the flux measured .
due to the high airmass of @xmath0 car , any error in the @xmath20 value is amplified greatly when applied . from dunne & eales ( 2001 ) , we have the fractional error in flux due to the uncertainty in @xmath20 as : @xmath22 where we have taken @xmath21 to be the r.m.s . of the residuals in fig . [ csotauf ] , and
@xmath23 is the airmass .
the value of @xmath21 is 0.00253 and this must be translated to an r.m.s . on the @xmath20 values at 450 and 850@xmath1
m . for this
we use the pre - upgrade revised narrow - band filter relationships between cso tau and 450/850@xmath1 m given on the jcmt webpages , and using propagation of errors we find that @xmath21(850 ) = 0.005 and @xmath21(450 ) = 0.012 . using eqn .
[ fracfluxe ] we calculate the fractional error on the flux due to uncertainties in the extinction corrections as @xmath24 ( 850 ) = 3 per cent and @xmath24 ( 450 ) = 6 per cent .
if we instead take the extreme values of the residuals rather than the r.m.s . we get increased values of 6 ( 850 ) and 15 per cent ( 450 ) , however , these are not statistically correct , and we include them as merely illustrative of a worst case .
, width=302,height=264 ] the second part of the calibration procedure is to compare the signal measured in an aperture on the object with that in the same aperture on a calibrator of known flux .
the calibrator used was crl618 , chosen because it was observed shortly before @xmath0 carinae , and therefore sky conditions and the shape of the dish would be closest to those at the time of the @xmath0 car observation .
a change in the dish shape ( due to thermal relaxation after sunset ) , or a telescope which is poorly focussed will result in changes in beam shape and hence the gain - defined as the ratio of calibrator flux ( jy ) / calibrator flux ( v ) .
thus , if the calibrator is observed under as similar conditions as possible to the objects then this effect is minimised .
a check of all of the calibration maps during the night showed that the beam shape was very poor in the earlier part of the evening ( after sunset ) , despite many attempts at focussing .
however , for the calibrator map taken prior to @xmath0 car , alignments were also made on the secondary mirror in the x - y plane which resulted in a much improved beam shape .
thus we believe that the previous poor quality images were due to a poorly aligned mirror and there is no reason to believe that this will affect the source data which was taken just after the re - alignment . there could be some concern that observing at such low elevations may also affect the shape of the dish and hence the gains , however , the staff at the jcmt have found no evidence for a link between gain and elevation ( iain coulson , private communication ) .
furthermore , the beam shapes look similar between the calibrator and @xmath0 car ( fig .
[ etasubmm ] ) , again suggesting that there is not a serious change in images observed at the lower elevations . any deterioration in dish shape tends to lead to higher gains ( more jy per v ) and therefore any elevation effect on the dish would be leading us to underestimate the flux of @xmath0 car rather than overestimating it ( as the calibrator was observed at a lower airmass ) .
since the earlier calibration maps had been affected by the mis - alignment of the secondary mirror , and therefore do not reflect usual nightly variations in the gain , we will use the information from tables 1 & 2 in dunne & eales ( 2001 ) to estimate the uncertainty on the gain . here the variation of gain values in various sized apertures was calculated using data taken over the same period as these observations ( july 1997 - july 1998 ) .
this gives the uncertainty in the gain in a 45 arcsec aperture as @xmath25 per cent at 850@xmath1 m and @xmath26 per cent at 450@xmath1 m .
there will also be an absolute calibration uncertainty due to errors in the flux of the calibrator and the model used to generate the planet fluxes ( on which the flux of the secondary calibrators is based ) .
the errors on the flux of crl618 are quoted as @xmath27 jy at 850@xmath1 m and @xmath28 jy at 450@xmath1 m .
adding in quadrature an additional 5% error due to the uncertainty in the model , the total error on the calibrator flux is estimated as @xmath29 per cent at 850@xmath1 m and @xmath30 per cent at 450@xmath1 m .
the total calibration error is then the quadratic sum of the three terms for extinction , gain and absolute as follows : @xmath31 giving @xmath32(850 ) = 9 per cent and @xmath32(450 ) = 13 per cent .
while these numbers may look surprisingly low , it must be remembered that this was a night with exceptionally good observing conditions ( during the el nino period ) , and that the best practise has been used in the calibration procedure .
the integrated flux was measured by placing the same aperture over both @xmath0 car and crl618 , measuring the total flux in volts on both .
the fluxes of @xmath0 car at 850 and 450@xmath1 m were found to be @xmath33jy and @xmath34jy respectively as measured in an aperture of diameter @xmath35 .
the radius of the aperture needed to incorporate the entire emission is @xmath36 , which at the distance of @xmath0 car corresponds to a physical size of 0.3pc and is just larger than the visual size of the homunculus .
the dust seen in these images may be probing the ` outer ejecta ' region ( e.g. smith & morse 2004 , weis 2004 ) .
however , we can not resolve the inner @xmath37 of the homunculus and we refer the reader to the recent sub - arcsecond study of @xmath0 car e.g. chesneau et al ( 2005 ) .
due to the high variability in the x - ray , radio , mm and ir and as the sub - mm observations of @xmath0 car were taken during a shell - event in 1998 , we chose literature fluxes taken nearest this epoch ( corresponding to a radio quiet stage ) .
this allows for the best possible direct comparison of the sub - mm with any variability in the radio , as well as determining the evolution of the sed during this time ( e.g. abraham & damineli 1999 ; abraham et al . 2005 ) .
where unavailable , older literature fluxes were used ( cox et al . 1995 ) .
the 1992 millimetre fluxes taken from this work are thought to be at a similar level to those expected in 1998 ( e.g. fig 1 , white et al .
2004 ) .
we fitted the sed of @xmath0 car from 10 - 850@xmath1 m using a modified blackbody ( fig.[etalow450 ] ) requiring a two temperature component at around 200 and 110k .
the dust emissivity exponent @xmath38 , is @xmath39 , lower than the nominal value of 2 for ` normal ' interstellar composition but higher than those estimated in m99 and smith et al ( 2003 ) .
this indicates that the grains could be amorphous , large or non spherical in composition ( e.g. bohren & huffmann 1983 ) .
however , there is likely to be contamination from the radio emission at the sub - mm wavelengths which is dependant on the phase of the radio cycle .
there are two sources of radio emission from the star - free free ionised emission from the stellar wind and emission from the optically thin homunculus region ( cox et al .
both thermal processes depend on frequency , free - free varies as @xmath40 ( wright & barlow 1975 ) and the optically thin region as @xmath41 .
the contamination from the radio to the sub - mm emission can be removed by determining the flux expected at 450 and 850@xmath1 m .
the variation of free - free emission in jy with wavelength from an ionised stellar wind is given in lamers & cassinelli ( 1999 ) .
x - ray modeling of colliding wind binaries fit the observations of @xmath0 car with parameters : mass @xmath42 , @xmath43 , mass loss rate , @xmath44 , terminal velocity @xmath45 and distance 2.6 kpc ( pittard 1999 ) .
the sed of the free - free emission assuming these values and a doubly ionised wind is shown in fig .
[ etalow450 ] ( dashed line ) .
the power law is _ significantly _ below the sub - mm fluxes and also underpredicts the observed 1.2 and 3 mm fluxes .
cox et al . ( 1995 ) suggest that the power law slope from the free - free emission is actually @xmath46 in the mm regime due to hydrogen recombination changing the electron and proton densities in the wind ( see their fig 3 ) .
the emission then turns over at @xmath47 mm to the normal slope of @xmath48 , just below the observed flux at this wavelength , shown on fig [ etalow450 ] .
we use the _ observed _ fluxes rather than the theoretical prescription of a stellar wind to estimate the maximum possible contribution from the ionised stellar wind in the sub - mm regime . thus we would expect a flux of @xmath49 jy at 850 and 18 jy at 450@xmath1 m respectively due to the ionised stellar wind .
it is clear from fig [ etalow450 ] that almost all of the 850@xmath1 m flux could be free - free emission , yet not all of the 450@xmath1 m emission can be explained from the expected power law .
indeed , it is difficult to explain the 450 and 850@xmath1 m flux ratio with a radio emission process .
however , given the uncertainty of how much of the sub - mm flux is due to dust emission or contaminated by free - free , we have two scenarios for the dust mass in @xmath0 car ( 1 ) best case : all of the estimated scuba flux is due to dust and ( 2 ) worst case : 94 and 35 per cent of the 850 and 450@xmath1 m flux respectively is due to the ionised stellar wind .
this effectively gives a lower limit to the range of dust masses expected ( see section [ sec : outer ] ) . as noted in smith et al .
( 2003 ) , the fit to the energy distribution is not unique , a range of values of temperature could achieve a satisfactory fit to the spectrum within the errors on the fluxes . indeed in their work , they reproduce an adequate fit to the morris et al .
( 1999 ) data with dust temperatures of 140k and 200k .
however , none of the previous best - fit seds produce the correct 450/850 ratio as observed here .
the extra fluxes at the long wavelength tail provide another constraint , which forces the fit to have a colder temperature than the previous models at 110k .
it is extremely important to separate the dust emission from free - free at these wavlengths as this affects the dust mass in the nebula .
if the wind is optically thin , for example , most of the ir emission would be due to dust and not contaminated by free - free .
the emission seen in the scuba images and the radial profiles suggests that some of the dust may be located outside the homunculus although due to the beam pattern we can not determine if this is a real extension .
the hst / wfpc2 image is shown in fig .
[ etahst ] with the chandra x - ray observations and the submm emission shown as contours starting at s / n @xmath50 3-@xmath17 ( original image from smith & morse 2004 ) . the 450@xmath1 m emission appears to coincide with the x - ray emission which is thought to occur where previous mass loss from the star is colliding with the surrounding interstellar material , causing shock heating of the gas .
we may therefore be probing dust shells from a previous mass - loss phase . ] . to determine the dust mass in @xmath0 car
, we use @xmath51 where @xmath52 is the dust mass absorption coefficient ( @xmath53 for normal interstellar dust at @xmath54 ) and d is the distance to the star .
the dust mass from the best fit sed to the free - free subtracted fluxes ( scenario 2 ) is @xmath55 depending on the absorption properties of the dust . with an typical gas - to - dust ratio of 100 ,
this suggests that around @xmath56 would have been expelled in the wind .
such a large number may indicate that the gas - to - dust ratio in the atmospheres of massive stars could be higher than predicted ( e.g. ag car - voors et al .
2000 ) . alternatively , the dust could be more efficient at emitting than normal ism dust . previous modelling of the sed of @xmath0 car has shown that the dust does not follow an opacity law of @xmath57 ( robinson et al 1987 ) . if we use the absorption coefficient estimated from dunne et al .
( 2003 ) to represent pristine , newly formed dust in stellar winds , we obtain @xmath58 which would require @xmath59 of gas to be expelled . using the best fit fluxes _ without _ the free - free contamination ( scenario 1 )
, the dust mass would be @xmath60 .
errors quoted are given using 68 per cent confidence intervals from parameters obtained using the bootstrap technique ( whereby 1000 sets of artificial fluxes were created from the original data with each individual set then fitted with our two - temperature model ) .
this is in good agreement with the original hypothesis of m99 and smith et al .
the dust mass calculated here appears to be spread out over a larger area than the torus and quite possibly outside the polar lobes of the homunculus . combining this result with that in m99 and smith et al .
( 2003 ) , we obtain the dust masses for each component as listed in table [ etacomp ] . since we can not resolve the inner homunculus , the dust mass here in the outer ejecta is likely to include the mass postulated to be in the lobes or the torus .
.[etacomp ] [ cols="^,^,^,^,^ " , ] if we assume that the dust around @xmath0 car is similar to dust in the ism , then the homunculus and the surrounding area contains roughly @xmath61 which has been ejected during the the last @xmath6yrs .
the averaged dust mass loss rate over the entire phase is then @xmath62 .
this is an order of magnitude greater than the dust mass loss rate estimated from observations of wolf - rayet binaries ( marchencko et al .
2002 ) and more than suggested by previous ir observations of @xmath0 car .
if the submm dust is located in the polar lobes of the nebula then it must have been ejected during the same singular event which caused the lobes - the great eruption .
this eruption lasted only a few decades which gives a large time - averaged gas mass loss rate of @xmath63 ( for more efficient dust emitters ) .
such a high mass loss rate was first noted in smith et al .
( 2003 ) and suggests that the pulsational mass loss in @xmath0 car is far more extraordinary than imagined .
the sub - mm observations suggest that an upper limit of @xmath64 of dust could exist in the extended dust shells , dust which must have been formed in the stellar wind with effective temperatures @xmath65 .
this indicates that contrary to the study in morgan & edmunds ( 2003 ) , significant amounts of dust could form in the stellar winds of massive stars even when the conditions in the atmospheric envelope are not reached by the available hydrodynamical models .
this previously undetected mass of material in the outer ejecta has important consequences for our understanding of stellar evolution and the formation of dust in massive stars .
massive stars may therefore provide a sustainable dust yield to the ism , in which case their contribution in chemical evolution models could be severly underestimated ( e.g. whittet 2003 ) .
we have used critical long wavelength observations to show that a large mass of dust does exist around @xmath0 carinae and suggest that the dust is distributed over a larger region than previously thought .
this has important implications for stellar evolution and may provide a natural explanation for the origin of the dust seen in young supernova remnants .
however @xmath0 car is extremely massive and highly unstable , so the huge dust mass loss rate is not unexpected .
it is very unlikely that the lower mass stars which were responsible for the supernova events of kepler and cas a would have ejected enough mass to form such a large quantity of dust before the explosion . in order to determine a more accurate dust mass in @xmath0 car , we require simultaneous observations in the sub - mm and mm to separate the contamination of the free - free emission .
we also require better resolution to determine how extended the sub - mm emission is .
future observations with herschel and sma will allow us to overcome this problem .
these observations should also allow us to determine any variability of the star in the submm .
quantifying the dust injection from stars of different masses in different stages of their evolution will have important consequences for estimating the dust budget in both local galaxies and those at high redshifts as well as providing fundamental information about stellar evolution .
hlg is a research fellow of the royal commission for the exhibition of 1851 .
we thank iain coulson and jan wouterloot for help with the data reduction and nathan smith for interesting and informative discussions .
we also thank the referee for constructive comments which have improved the paper .
the james clerk maxwell telescope is operated by the joint astronomy centre on behalf of the particle physics and astronomy research council of the united kingdom , the netherlands organisation for scientific research , and the national research council of canada .
abraham z. , damineli a. , in _ eta carinae at the millennium _ , asp conference series vol . 179 , 1999 eds
. j. a. morse , r. m. humphreys , and a. damineli , p 263 abraham z.,falceta - gonalves d. , dominici t. p. , nyman l .- .
, durouchoux p. , mcauliffe f. , caproni a. , jatenco - pereira v. , 2005 , a & a , 437 , 977 barlow m.j .
, et al . 2005 , apjl , 627 , 113 bohren c.f . , & huffman d.r . , 1983 ,
absorption and scattering of light by small particles , wiley , new york chesneau o. , min m. , herbst t. , waters l.b.f.m .
, hillier d.j . ,
leinert ch .
, de koter a. , pascucci , i. , et al . , 2005 , a & a , 435 , 1043 cox p. , mezger p.g . , sievers a. , najarro f. , bronfman l. , kreysa e. , haslam , g. , 1995 , a & a , 297 , 168 davidson k. & smith n. , 2000 , nature , vol 405 , p.532 duncan r.a . , & white s.m . , 2003 , mnras , 338 , 425 dunne l. , eales s. , 2001 , mnras , 327 , 697 dunne l. , morgan h.l . ,
eales s. , ivison , r. , edmunds , m.g . , 2003 , nature , 424 , 285 dunne l. , eales s. , edmunds m.g . , ivison r. , alexander , p. , clements d.l . , 2000 ,
mnras , 315 , 115 fesen r.a . , morse j.a . , chevalier r.a
. , borkowski k.j . ,
gerardy c.l . ,
lawrence s.s . ,
van den bergh s. , 2001 , aj , 122 , 2644 gaviola e. , 1946 , rev .
astron . , 18 , 252 green d. , tuffs r.m . , popescu c. , 2004 , mnras , 355 , 1315 harvey p.m. , hoffman w.f .
, campbell m. , f. , 1978 , a & a , 70 , 165 hony s. , dominik c. , waters l.b.f.m .
, icke v. , mellema g. , van boekal , r. , de koter a. , morris p.m. , barlow m. , cox p. , kaufl h.u . , 2001 , a & a , 377 , l1 lamers h.j.g.l.m . , & cassinelli j.p . , 1999 ,
_ an introduction to stellar winds _ , 46 , cambridge university press , great britian marchenko s.v . ,
moffat a.f.j . ,
vacca w.d .
, ct , s. , doyon r. , 2002 , apj , 565 , l59 morgan h.l . , edmunds m.g .
, 2003 , mnras , 343 , 427 morgan h.l .
, dunne l. , eales s. , ivison r. , edmunds , m.g .
, 2003 , apj , 597 , l33 morris .
p.w . , et al . , 1999 , nature , 402 , 502 ( m99 ) pittard j. , phd thesis , 1999 , university of birmingham pittard j. , 2003 , astronomy and geophysics , vol 44 , p.1.17 - 1.22 pittard j.m .
, stevens i.r .
, corcoran m.f . ,
ishibashi k. , 1998 , mnras , 299 , l5 polomski e. , telesco c.m .
, pina r.k .
, scott fisher , r. , 1999 , apj , 118 , 2369 pozzo m. , meikle w.p.s .
, fassia a. , geballe t. , lundqvist p. , chugai n.n . , sollerman j. , 2004 , mnras , 352 , 457 robinson g. , mitchell r.m . , aitken d.k . , briggs g.p .
, roche p.f . , 1987 , mnras , 227 , 535 sandell g. , jessop n. , jenness t. , 2001 , scuba map reduction cookbook , starlink cookbook 11.2 smith n. , gehrz r. , 2000 , apj,529 , l99 smith n. , gehrz r.d . , hinz p.m. , hoffman w.f . ,
mamajek , e.e . ,
meyer m.r . , hora j.l . , 2002 ,
apj , 567 , l77 smith n. , gehrz r.d .
, hinz p.m. , hoffmann w.f .
, hora j.l .
, mamajek e.e .
, meyer m.r . , 2003 ,
apj , 125 , 1458 smith n.m . , morse j.a . , 2004 ,
apj , 605 , 854 sugerman b.e.k . , ercolano b. , barlow m.j . ,
tielens a.g.g.m . ,
clayton g.c .
, zijlstra a.a . , meixner m. , speck a. , et al . , 2006 ,
science , 313 , 196 , pre - print , astro - ph/0606132 voors r.h.m .
, et al . , 2000 , a & a. , 356 , 501 weis k. , _ in the fate of the most massive stars _ , 2005 asp conference series , vol .
332 , eds .
r. humphreys and k. stanek , p.275 white s.m .
, duncan r.a .
, lim j. , nelson g.j .
, drake s.a . ,
kundu m.r . , 1994 , apj , 429 , 380 white s.m . , duncan r.a .
, chapman j.m . , koribalski b. , 2005 , _ in the fate of the most massive stars _ , asp conference series , vol .
332 , eds .
r. humphreys and k. stanek , p.129 whittet d.c.b .
, 2003 , dust in the galactic environment second edition , iop , cambridge university press , uk wilson t.l . , batrla w. , 2005 , a&a , 430 , 561 wright a.e . , &
barlow m.j . , 1975 , mnras , 170 , 41 | we present critical , long - wavelength observations of @xmath0 carinae in the submillimetre using scuba on the jcmt at 850 and 450 @xmath1 m to confirm the presence of a large mass of warm dust around the central star .
we fit a two - component blackbody to the ir - submm spectral energy distribution ( sed ) and estimate between @xmath2 of dust exists in the nebula depending on the dust absorption properties and the extent of contamination from free - free emission at the scuba wavelengths .
these results provide further evidence that @xmath0 carinae s circumstellar nebula contains @xmath3 of gas , although this may have been ejected on a longer timescale than previously thought .
circumstellar matter stars : individual : @xmath0 carinae
submillimetre |
when a bulk semiconductor is simultaneously irradiated by an optical field and its phase - coherent second harmonic , quantum interference between one- and two - photon absorption pathways enables excitation of carrier distributions with interesting properties @xcite .
such excitation , even without an external bias , can produce ballistic photocurrents @xcite , spin - polarized currents @xcite , and pure spin currents @xcite . characteristically of quantum interference
, these currents are sensitive to the phases of the two optical fields . in noncentrosymmetric semiconductors ,
the phases can also be used to control the total population of photoexcited carriers @xcite , and the net carrier spin polarization @xcite . which of these effects occur
depends on the polarization states of the fields .
these are examples of `` @xmath3 '' coherent control schemes , in which a two - color light field controls a physical or chemical process by interference of @xmath4- and @xmath5-photon transitions @xcite . in semiconductors ,
`` 1 + 2 '' excitation has been discussed for impurity - band absorption @xcite , free carrier absorption @xcite , quantum wells @xcite , and quantum wires @xcite , but our interest here is `` 1 + 2 '' coherent control of interband transitions in unbiased bulk semiconductors @xcite .
such experiments have been performed with either ( a ) two fields , typically short pulses , one the generated second harmonic of the other @xcite , or ( b ) a single ultrashort pulse having at least an octave bandwidth @xcite .
previous microscopic calculations of `` 1 + 2 '' processes in bulk semiconductors fall into two categories : _ ab initio _ density functional methods have been used for current injection @xcite and population control @xcite , while simple analytical band models perturbative in @xmath2 ( with at most eight spherical , parabolic bands ) have been used for current injection @xcite and spin - current injection @xcite .
the former are best suited for excess energies on the order of evs , while the latter are only valid for excitation close to the band edge and can not be applied to population and spin control , which vanish in such centrosymmetric models . in this article , we calculate `` 1 + 2 '' processes using an intermediate model that diagonalizes the @xmath6 hamiltonian in a basis of 14 @xmath7-point states with remote band effects included perturbatively .
the model contains empirically determined parameters @xcite .
fourteen - band models ( also called five - level models ) have been used to calculate band structures @xcite , linear @xcite and non - linear @xcite optical properties , and spin decoherence properties @xcite of gaas and other semiconductors .
winkler has recently reviewed 14-band models @xcite .
the model is nonperturbative in @xmath2 and includes nonparabolicity , warping , spin - splitting , and interband spin - orbit coupling .
we apply the 14-band model to the zinc - blende semiconductors insb , gasb , inp , gaas , and znse .
we compare these results with analytic expressions derived in the parabolic - band approximation ( pba ) based on an expansion in @xmath2 about the @xmath7 point of @xmath8 , which is the matrix element governing optical transitions .
a one - photon transition is called `` allowed '' if the zeroth - order term in its expansion is nonzero , and called `` forbidden '' otherwise .
two - photon transitions have two velocity matrix elements , and thus have a hyphenated label depending on the lowest - order terms in the expansions for each matrix element . for example , if both matrix elements are independent of @xmath2 to lowest order , the two - photon transition is called `` allowed - allowed '' . for current injection and spin - current injection ,
we use expressions derived previously with an eight band model @xcite . for population control and spin control ,
we derive expressions with the 14-band model .
the comparison between the pba expressions and the numerical calculation establishes an important microscopic difference between current and spin - current control on the one hand , and population and spin control on the other hand .
close to the band - gap , the former result from the interference of allowed one - photon transitions and allowed - forbidden two - photon transitions , whereas the latter result from the interference of allowed one - photon transitions and allowed - allowed two - photon transitions .
this difference was posited previously based on heuristic arguments @xcite .
most of the early theory on semiconductor `` 1 + 2 '' processes processes conceptually separated the optical injection of densities and currents from the relaxation and transport of these quantities .
we follow this approach , and in this article , focus on microscopic calculations of the optical injection .
we note that relaxation and transport have been studied with an effective circuit model @xcite , hydrodynamic equations @xcite , boltzmann transport in the relaxation time approximation @xcite , a non - equilibrium green function formalism @xcite , and the semiconductor bloch equations @xcite .
we model the optical field as a superposition of monochromatic fields of frequency @xmath9 and @xmath10 : @xmath11 and we sometimes write @xmath12 and @xmath13 .
we describe the fourteen - band model in section [ sec : model ] , and use it to study `` 1 + 2 '' current injection in section [ sec : current ] , `` 1 + 2 '' spin - current injection in section [ sec : spincurrent ] , `` 1 + 2 '' population control in section [ sec : population ] , and `` 1 + 2 '' spin control in section [ sec : spin ] .
we calculate the injection of each `` 1 + 2 '' process using microscopic expressions derived using velocity gauge ( @xmath14 ) coupling in the long wavelength approximation , treating the field perturbatively in the fermi s golden rule limit , and using the independent - particle approximation @xcite . for spin - current injection and spin control ,
we use microscopic expressions that include the coherence between spin - split bands . in appendix
[ app : kdepso ] , we justify the neglect of @xmath2-dependent spin - orbit coupling .
the parabolic - band approximation results are derived and discussed in appendix [ app : pba_a - a ] , and compared with the numerical calculations in sections [ sec : current][sec : spin ] .
we summarize and conclude in section [ sec : summary ] .
the fourteen - band model hamiltonian , which includes important remote - band effects to order @xmath15 , and which we denote @xmath16 , is given explicitly by pfeffer and zawadski @xcite .
the fourteen bands ( counting one for each spin ) , are shown in fig .
[ fig:14banddiagram ] .
they comprise six valence bands ( two each for split - off , heavy and light holes ) and eight conduction bands ( the two @xmath17-like ones at the band edge , and the six next lowest ones which are @xmath18-like ) .
we now briefly review the derivation of @xmath16 .
-point states ( right ) , and one- and two - photon transitions .
@xmath19 , @xmath20 , and @xmath21 indicate irreducible representations of the @xmath22 double group , whereas @xmath23 and @xmath24 indicate irreducible representations of the @xmath22 point group .
note that spin - splitting , nonparabolicity , and warping are not shown in this diagram . ]
the one - electron field - free hamiltonian is @xmath25 , where @xmath26 , the potential @xmath27 has the symmetry of the crystal , and the spin - orbit interaction @xmath28 is @xmath29 where @xmath30 is the dimensionless spin operator , @xmath31 .
note that relativistic corrections proportional to @xmath32 have been neglected @xcite .
the eigenstates of @xmath33 are bloch states @xmath34 with energy @xmath35 .
the associated spinor wave function @xmath36 can be written @xmath37 , where the spinor functions @xmath38 have the periodicity of the crystal lattice .
we use the notation @xmath39 to denote the kets for the @xmath40-functions ; i.e. @xmath41 . note that @xmath42 .
the hamiltonian for the @xmath40-function kets , known as the @xmath1 hamiltonian , is @xcite @xmath43 where the velocity operator @xmath44 $ ] is @xmath45 the second term in @xmath46 , the anomalous velocity , which leads to @xmath2-dependent spin - orbit coupling in @xmath47 , can be neglected for the processes we consider as shown in appendix [ app : kdepso ] ; in the rest of this article , we assume that it vanishes .
the states @xmath48 are a complete set of eigenstates for the hamiltonian @xmath33 on the space of cell - periodic functions .
thus cell - periodic eigenstates of @xmath47 can be expanded in the infinite set of states @xmath48 .
the `` bare '' fourteen - band model truncates this expansion to a set of fourteen states , corresponding to the fourteen bands closest in energy to the fundamental band gap at the @xmath7 point @xcite . in a semiconductor of zinc - blende symmetry ,
the states @xmath49 are conveniently expanded in the eigenstates of @xmath50 , @xmath51 , where , under the point group @xmath22 , @xmath52 transforms like @xmath53 , @xmath54 and @xmath55 transform like @xmath56 @xcite .
the @xmath57 comprises the usual spin @xmath58 states : [ eq : pauli ] @xmath59 the non - zero matrix elements of @xmath60 are @xmath61 cyclic permutations of these [ e.g. @xmath62 , and those generated by hermitian conjugation of these .
the above equations define the spin - orbit energies @xmath63 and @xmath64 , and the interband spin - orbit coupling @xmath65 @xcite .
the fourteen basis states @xmath66 for @xmath16 are [ eq : basis ] @xmath67 where @xmath68 and @xmath69 .
the states are labeled with their transformation property under the double group for @xmath22 , and with a pseudo - angular momentum notation . in the basis
, @xmath70 is diagonal except for terms proportional to @xmath71 .
the connection between the eigenvalues of @xmath70 for the @xmath72-point eigenstates and the eigenvalues of @xmath50 is given by pfeffer and zawadski @xcite .
the nonzero matrix elements of momentum , which appear in @xmath47 , are [ eq : p_matrix ] @xmath73 eq
. defines the parameters @xmath74 , @xmath75 , and @xmath76 .
they are sometimes expressed as energies @xmath77 , @xmath78 , and @xmath79 with the connections @xmath80 , etc .
the `` bare '' fourteen - band model has eight empirical parameters @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath65 , @xmath74 , @xmath76 , and @xmath75 .
its quantitative accuracy is improved by adding important remote band effects to order @xmath15 using lwdin perturbation theory @xcite , which adds @xmath2-dependent terms to the truncated @xmath85 hamiltonian so that its solutions better approximate those of the full hamiltonian @xcite .
the remote band effects are governed by the parameters @xmath86 , @xmath87 , @xmath88 , @xmath89 , and @xmath90 .
the parameters @xmath91 , @xmath92 , and @xmath93 are modified luttinger parameters that account for remote band effects on the valence bands .
they are related to the usual luttinger parameters @xmath94 , @xmath95 , and @xmath96 by the couplings with @xmath97 , @xmath98 , and @xmath99 bands , which are already accounted for in the `` bare '' fourteen - band model @xcite : @xmath100 the parameter @xmath89 accounts for remote band effects on the lowest conduction band , essentially fixing its effective mass to the experimentally observed value .
finally , the parameter @xmath90 is the small @xmath2-linear term in the valence bands @xcite .
the remote band effects can be removed by setting @xmath101 and @xmath102 .
the model includes neither remote band effects on the @xmath103 bands , nor remote band effects on the @xmath97-@xmath104 and @xmath97-@xmath105 momentum matrix elements , although such terms exist in principle @xcite . in summary
, @xmath16 is a fourteen - band approximation to @xmath47 that incorporates some remote band effects .
it can be found in eq .
( 5 ) of pfeffer and zawadzki , although with a slightly different notation @xcite . with their notation on the left , and ours on the right : @xmath106 , @xmath107 , @xmath108 , @xmath109 , @xmath110 .
also , our @xmath82 differs from theirs by a minus sign .
other authors have also used different notations @xcite .
the fourteen bands are shown schematically in fig .
[ fig:14banddiagram ] along with the symmetry notation of the @xmath7-point states , and the notation used to label the bands .
numerical values for the thirteen parameters of the model are listed in table [ table : parameters ] for insb , gasb , inp , gaas , and znse .
they are taken from the literature , where they were chosen to fit low - temperature experimental data .
of the two parameter sets discussed by pfeffer and zawadzki for gaas , we use the one corresponding to @xmath111 that they find yields better results @xcite . for inp , gasb , and insb ,
we use parameters from cardona , christensen and fasal @xcite . for cubic znse
, we use the parameters given by mayer and rossler @xcite , we use a calculated value of @xmath90 @xcite , and we use @xmath112 ev to give a @xmath113 conduction band spin - splitting that matches the _ ab initio _ calculation of cardona , christensen and fasal @xcite .
winkler used these same parameters for znse , but took @xmath114 @xcite .
there is more uncertainty in the parameters for znse than in those for the other materials @xcite , but we include it as an example of a semiconductor with a larger band gap .
c|ddddd & & & & & + @xmath81 ( ev ) & 1.519 & 1.424 & 0.813 & 0.235 & 2.820 + @xmath82 ( ev ) & 0.341 & 0.108 & 0.75 & 0.803 & 0.403 + @xmath83 ( ev ) & 4.488 & 4.6 & 3.3 & 3.39 & 7.330 + @xmath84 ( ev ) &
0.171 & 0.50 & 0.33 & 0.39 & 0.090 + @xmath65 ( ev ) & -0.061 & 0.22 & -0.28 & -0.244 & -0.238 + @xmath74 ( ev ) & 10.30 & 8.65 & 9.50 & 9.51 & 10.628 + @xmath76 ( ev ) & 7.70 & 7.24 & 8.12 & 8.22 & 9.845 + @xmath75 ( ev ) & 3.00 & 4.30 & 3.33 & 3.17 & 9.165 + @xmath115 & 7.797 & 5.05 & 13.2 & 40.1 & 4.30 + @xmath116 & 2.458 & 1.6 & 4.4 & 18.1 & 1.14 + @xmath117 & 3.299 & 1.73 & 5.7 & 19.2 & 1.84 + @xmath89 & -1.055 & 0 & 0 & 0 & 0 + @xmath90 ( mev ) & -3.4 & -14 & 0.43 & -9.2 & -14 the parabolic - band approximation calculations use parameters from table [ table : parameters ] , and average effective masses derived from the parameters in table [ table : parameters ] .
the relations between matrix elements of the bloch states and matrrix elements of the @xmath40-function kets are @xmath118 the matrix elements of the velocity operator , @xmath46 , neglecting the anomalous velocity as discussed in appendix [ app : kdepso ] , can be calculated using , , and the right side of .
the matrix elements of the spin operator @xmath119 , can be found from eq . .
the matrix elements of @xmath120 can be similarly found in the basis of eigenstates of @xmath50 .
each of these can then be rotated to the basis in which the states @xmath121 are expanded .
it is well known that , in a crystal , @xmath122 .
more generally , @xmath123 these identities can be proven from the definitions @xmath124 and @xmath125 $ ] , even for a non - local hamiltonian .
but when remote band effects are included in a finite band model , they no longer hold .
that is , @xmath126 calculated using and eigenstates of @xmath16 is not equal to @xmath127 .
we explicitly restore these identities by using @xmath127 to calculate @xmath128 .
this approach can be described as including remote band effects in the velocity operator .
it was used for an eight band calculation of linear absorption by enders et al @xcite .
this step is not critically important for the effects calculated here , since remote band effects are generally small .
the optical calculations in this article have the form @xmath129 , where @xmath130 where @xmath131 depends on matrix elements and energies of eigenstates of @xmath47 , and where @xmath132 .
the integral in is understood to be restricted to the first brillioun zone , but we do not actively enforce the restriction , since the photon energies considered here cause transitions well within the first brillioun zone . writing @xmath133 in spherical coordinates , where @xmath134 is the solution to @xmath135 we have @xmath136 where we have used @xmath137 and the cubic symmetry of the crystal .
it is numerically convenient to do the sum over any degenerate bands before the integral over @xmath138 and @xmath139 .
the calculations of `` 1 + 2 '' effects in the following sections are primarily labeled by the hamiltonian used to approximate @xmath47 .
the complete fourteen - band model is denoted @xmath16 .
the bare fourteen - band model , denoted @xmath140 , is @xmath16 without remote band effects .
the @xmath141 subset of the fourteen band hamiltonian within the basis @xmath142 is denoted @xmath143 .
the spherical eight - band model , denoted @xmath144 , is derived from @xmath143 by setting @xmath145 and replacing @xmath92 and @xmath93 by @xmath146;@xcite it is a spherical approximation to the kane model including remote band effects @xcite .
the aforementioned calculations are non - perturbative in @xmath2 ; that is , in each case , the hamiltonian is solved numerically at each @xmath2 .
the perturbative calculations of appendix [ app : pba_a - a ] are denoted pba ( parabolic - band approximation ) . the microscopic expression for each of the `` 1 + 2 '' effects contains a sum over intermediate bands , which originates from the two - photon amplitude . unless otherwise noted , calculations include all possible intermediate bands ( eg . , @xmath16 includes fourteen intermediate bands , and @xmath144 includes eight intermediate bands ) .
calculations that restrict this sum are secondarily labeled to reflect the restriction .
the label `` @xmath16 , no @xmath103 '' uses @xmath16 , but does not include @xmath103 bands as intermediate states .
the label `` @xmath16 , no @xmath103/@xmath147 '' uses @xmath16 , but includes neither @xmath103 nor @xmath147 bands as intermediate states .
the label `` @xmath16 , 2bt '' uses @xmath16 , but only includes two - band terms ( terms for which the intermediate band is the same as the initial or final band ) .
similar labels are used for @xmath144 , for example , `` @xmath144-pba , no so '' uses the perturbative solution to @xmath144 and does not include @xmath147 intermediate states .
the current injection rate due to the field can be written @xmath148 where @xmath149 is the macroscopic current density , and @xmath150 the third rank tensor @xmath151 describes one - photon current injection ( the circular photogalvanic effect @xcite ) , the fifth rank tensor @xmath152 describes two - photon current injection , and the fourth rank tensor @xmath153 describes `` 1 + 2 '' current injection @xcite .
aversa and sipe showed that @xmath153 is related to a doubly divergent part of the third - order nonlinear susceptibility @xmath154 @xcite . in cubic materials with point group symmetry @xmath22 , @xmath155 or @xmath156 , a general fourth rank tensor has four independent components , but due to the intrinsic symmetry @xmath157 , @xmath158 has only three independent components ; there are 21 non - zero components of @xmath158 in the standard cubic basis : @xmath159 , @xmath160 , and @xmath161 ( the components in parentheses can be exchanged ) , where @xmath162 , @xmath163 , and @xmath164 denote components along the principal cubic axes @xcite . this can be written @xmath165 where @xmath166 is a kronecker delta and the only non - isotropic part is @xmath167 , which we define in the principal cubic basis as @xmath168 when @xmath169 and zero otherwise .
the three independent components are @xmath170 , @xmath171 , and @xmath172 .
thus , in a cubic material , @xmath173 this generalizes the notation we used previously for a calculation in the parabolic - band approximation @xcite , with the connection @xmath174 , and @xmath175 . in that , or any other spherical approximation , @xmath176 . to calculate @xmath158
, we use the microscopic expression first given by atanasov et al .
@xcite , modified to explicitly include the sum over spin states @xcite .
an alternate microscopic expression has been derived in the length gauge @xcite , but it has not yet been used in a calculation . in the independent particle approximation that we employ here , @xmath158 is purely imaginary @xcite and hence @xmath177 , @xmath178 , and @xmath179 are real , although they can be complex if excitonic effects are included @xcite .
( black lines ) , @xmath178 ( red lines ) , and @xmath179 ( blue lines ) for gaas . panel ( a ) shows the contributions from each initial valence band ; dashed , dotted , and dashed - dotted lines include only transitions from the @xmath180 , @xmath181 , and @xmath147 bands respectively , while the solid lines include all three transitions . the thin solid , light brown line in ( a ) is the total @xmath182 .
panel ( b ) separates the total into electron ( dashed ) and hole ( dotted ) contributions.[fig : currentgaas_initial ] ] the spectra of @xmath177 , @xmath178 , and @xmath179 , calculated for gaas , are shown in fig.[fig : currentgaas_initial](a ) along with the contributions to each tensor component from each possible initial valence band . for a given photon energy ,
electrons photoexcited from the @xmath180 band have higher energies and velocities than electrons photoexcited from the @xmath181 band ; hence the dominant component @xmath177 is larger for @xmath180-@xmath164 transitions than @xmath181-@xmath164 transitions .
the smallness of @xmath178 is due to contributions from the @xmath180-@xmath164 transitions having opposite sign to the @xmath181-@xmath164 transitions , as shown previously in the pba @xcite .
figure [ fig : currentgaas_initial](b ) separates each tensor component into an electron contribution and a hole contribution ( denoted @xmath183 and @xmath184 by atanasov et al @xcite ) .
electrons make a larger contribution to @xmath185 than holes , due to the lower effective mass ( and hence higher velocity ) of an electron than of a hole ( much lower , in the case of a heavy hole ) with the same crystal momentum .
holes dominate @xmath186 at lower photon energies , while electrons dominate @xmath186 at higher energies . both electrons and holes
contribute equally to the anisotropic component @xmath179 . to help in understanding the importance of the various intermediate states , in fig .
[ fig : currentgaas_models ] we compare the calculated current injection tensor elements with various degrees of approximation described in sec . [ approximations_defined ] .
the component @xmath177 ( and hence @xmath187 , since @xmath177 is larger than @xmath188 ) is dominated by two - band terms .
three - band terms cause an increase , by as much as 34% , of @xmath177 [ the difference between the dashed and solid black lines in fig . [
fig : currentgaas_models](a ) ] .
although not shown in fig .
[ fig : currentgaas_models ] , most of the increase is due to three - band terms with the @xmath147 band as an intermediate state .
terms with the @xmath103 bands as intermediate states only cause a small increase to @xmath177 ( the difference between the dotted and solid black lines ) .
the warping of the bands is clearly not important for @xmath177 , since the calculation with @xmath144 closely approximates the calculation `` @xmath16 , no @xmath103 '' , which includes the same intermediate states .
surprisingly , the `` @xmath144-pba , 2bt '' result @xcite closely approximates the complete , non - perturbative fourteen - band calculation , even at excess photon energies for which band nonparabolicity is significant .
this is due to a fortuitous compensation between the neglect of nonparabolicity and the neglect of three - band terms .
the compensation is not as complete for all materials .
the component @xmath178 , which determines the current due to orthogonal linearly polarized fields , is less forgiving to approximations than the component @xmath177 .
we have already seen in fig .
[ fig : currentgaas_initial ] that @xmath178 is small due to a near cancellation of @xmath180 and @xmath181 initial states .
reasonable accuracy on @xmath178 thus requires higher accuracy on the contribution from each initial state .
in particular , three - band terms must not be neglected . by comparing the dashed - dotted and solid lines in fig .
[ fig : currentgaas_models](b ) , it can be seen that , whereas the sum of the two - band terms is negative , the sum of the three - band terms is positive and of the same magnitude .
it is useful to divide the three - band terms into three groups : those with intermediate state from the @xmath180 or @xmath181 bands , those with intermediate state from the @xmath147 band , and those with intermediate state from one of the @xmath103 bands .
we find that each group contributes roughly the same positive amount to @xmath178 for excess photon energies less than @xmath63 .
the groups are added successively to the 2bts in the dashed , dotted , and solid lines in fig .
[ fig : currentgaas_models](b ) .
three - band terms with @xmath147 intermediate states are less important at the higher excess photon energies in fig .
[ fig : currentgaas_models](b ) .
the warping of the bands makes a small but non - negligible contribution to @xmath186 , as seen in the difference between the dashed - double - dotted and dotted lines of fig .
[ fig : currentgaas_models](b ) . the solid brown line in fig .
[ fig : currentgaas_models](b ) is the `` @xmath144-pba , no @xmath147 '' result @xcite . at low excess photon energies , it greatly underestimates @xmath178 due to the neglect of @xmath147 and @xmath103 intermediate states , while at excess photon energies greater than 100 mev , this is partly compensated for by the neglect of nonparabolicity .
it appears from the difference between `` @xmath144-pba , no @xmath147 '' and `` @xmath16 , no @xmath103/@xmath147 '' in fig .
[ fig : currentgaas_models](b ) that nonparabolicity becomes important at energies above 70 mev .
the term @xmath179 is purely due to cubic anisotropy by definition ; in any model that is spherically symmetric it is identically zero .
there is no cubic anisotropy in the `` bare '' ( i.e. without remote band effects ) eight - band model on the set @xmath142 .
cubic anisotropy in the fourteen - band model is due to the momentum matrix elements governed by the parameters @xmath79 and @xmath189 , the interband spin - orbit coupling @xmath71 , and remote bands through @xmath190 and @xmath90 . from fig.[fig : currentgaas_models](c ) , it can be seen that three - band terms are important for @xmath179 .
in fact , with only 2bts included , @xmath179 is positive for gaas , whereas it is negative with all terms included . from fig .
[ fig : currentgaas_models](c ) it can also be seen that the @xmath147 band and @xmath103 bands are important as intermediate states for @xmath179 .
our calculation of @xmath158 is of the same order of magnitude as the _ ab initio _ calculation of atanasov et al.@xcite , but its spectral dependence is different .
in particular , @xmath177 agrees more closely with the pba calculation , as seen in fig . [
fig : currentgaas_models](a ) .
atanasov et al .
had attributed the difference between their _ ab initio _ and pba calculations to the assumption of @xmath2-independent velocity matrix elements in the pba @xcite . however , our calculation accounts for the @xmath2-dependence of velocity matrix elements and agrees closely ( for @xmath177 and @xmath191 ) to the pba
. the earlier _ ab initio _
calculation @xcite was , in fact , inaccurate at low photon energies due to various computational issues ; an improved _ ab initio _ calculation agrees with the spectral dependence at low photon energy given here @xcite .
figure [ fig : current_4materials ] shows the spectra of @xmath185 , @xmath178 , and @xmath179 calculated with @xmath16 for insb , gasb , inp , and znse .
the dashed black line in fig.[fig : current_4materials ] is the pba result @xcite .
the pba appears to be a reasonable approximation to @xmath177 for excess energies less than about @xmath192 . in each material , @xmath193 , and in each material except for znse , the sign of @xmath178 varies as a function of frequency .
the component @xmath179 , which arises due to cubic anisotropy , is negative for each material .
the cubic anisotropy of current injection due to colinearly
polarized fields can be significant enough that it should be measurable . for fields
colinearly polarized along @xmath194 , specified by polar angles @xmath195 and @xmath196 relative to the cubic axes , @xmath197 , \ ] ] where @xmath198 . in general , @xmath199 also has a component perpendicular to @xmath194 that is proportional to @xmath179 , but it vanishes for @xmath194 parallel to @xmath200 , @xmath201 , @xmath202 .
the field polarization that maximizes the current injection depends on the relative sign of @xmath179 and @xmath203 .
when they have the opposite sign , current injection is a minimum for @xmath204 ( @xmath205 ) and a maximum for @xmath206 ( @xmath207 ) ; for light normally incident on a @xmath208 surface , the largest current injection occurs when @xmath209 ( @xmath210 ) .
when they have the same sign , current - injection is a maximum for @xmath204 and a minimum for @xmath206 . from the gaas results shown in fig.[fig : currentgaas_initial](a ) , the current injection for the three cases @xmath211 , @xmath212 , and @xmath206 are in the ratio 1 to 1.14 to 1.20 at the band edge , 1 to 1.15 to 1.20 at 200 mev excess photon energy , and 1 to 1.22 to 1.29 at 500 mev excess photon energy .
in contrast , the _ ab initio _ calculation of atanasov et al .
yields larger ratios , for example 1 to 1.32 to 1.43 at 300 mev excess photon energy @xcite .
this disagreement is consistent with the inaccuracy of the _ ab initio _ calculation discussed above .
initial experiments with gaas used @xmath213 $ ] @xcite , whereas roos et al . exploited the larger signal for @xmath214 $ ] @xcite . for each of the materials shown in fig.[fig : current_4materials ] ,
the minimum current injection is for @xmath215 .
it is worth noting that two - photon absorption is also a minimum with @xmath216 for many semiconductors @xcite .
it seems that both `` 1 + 2 '' current injection and two - photon absorption with linearly polarized fields are larger for @xmath194 directed along the bonds .
the cubic anisotropy of `` 1 + 2 '' current injection is pronounced for cross - linearly polarized fields and opposite - circularly polarized fields .
for example , for cross - linearly polarized fields normally incident on @xmath217 with @xmath218 and @xmath219 , @xmath220 .
\end{split}\ ] ] for fields with opposite circular polarizations , the current injection is proportional to @xmath179 and is hence purely anisotropic .
the component @xmath179 causes a type of current injection that has not previously been noted . in all `` 1 + 2 '' experiments considered thus far with light normally incident on a surface
, the direction of current injection lies in the plane of the surface .
however , with co - linearly polarized light fields normally incident on a @xmath221 surface , the current can have a component into ( or out of ) the surface . the current in this case is @xmath222 , % \end{split}\ ] ] where @xmath223 , @xmath224 is the @xmath225 $ ] direction , and @xmath226 is the angle between @xmath194 and the @xmath227 $ ] direction .
thus , @xmath179 governs this `` surfacing '' current .
the carrier injection rate due to the field can be written @xmath228 , where @xmath229 is the density of electron - hole pairs , @xmath230 is one - photon absorption , @xmath231 is two - photon absorption , and @xmath232 is `` 1 + 2 '' population control @xcite .
the third - rank tensor @xmath233 has intrinsic symmetry @xmath234 . in centrosymmetric materials , such as those with the diamond structure ( point group @xmath155 ) ,
@xmath235 is identically zero ; hence , population control requires a noncentrosymmetric material . in a material with zinc - blende symmetry ( point group @xmath22 )
, @xmath233 has only one independent component ; in the standard cubic basis , @xmath236 are the only non - zero components , where @xmath162 , @xmath163 , and @xmath164 denote components along the principal cubic axes .
we calculate @xmath237 with the microscopic expression given by fraser et al . , which was derived in the independent - particle approximation , and is restricted to @xmath238 @xcite . under those conditions , @xmath237 is real and is proportional to the imaginary part of the susceptibility for second harmonic generation ( shg ) @xcite ; specifically , ( in mks ) @xmath239 this connection to shg , which can be derived from considerations of energy transfer and macroscopic electrodynamics @xcite , is important because the imaginary part of @xmath240 has sometimes been presented _ en route _ to a calculation of @xmath241 @xcite . as well , analytic expressions have been derived for the dispersion of shg by using simple band models , with approximations appropriate for @xmath242 near the band gap @xcite .
however , these earlier works did not connect @xmath243 with population control , and in fact typically stated that it was not independently observable .
[ fig : pop_imchi2 ] shows the calculation of @xmath244 for insb , gasb , inp , gaas and znse .
also shown for comparison is the pba expression ( [ eq : popcontrolpba ] ) , derived in appendix [ app : pba_a - a ] .
each spectrum can be divided into roughly three regions .
at very low excess photon energies , visible in the log - log plot fig .
[ fig : pop_imchi2](f ) , the spectrum is roughly independent of @xmath9 .
this flat part of the spectrum disappears if @xmath245 is set to zero ; hence , it is due to the @xmath2-linear term in the @xmath164 band spin - splitting .
next higher in photon energy , up to about 100 mev in gasb , inp , gaas , and znse ( up to about 15 mev in insb ) , is a region where the agreement with the analytic expression ( [ eq : popcontrolpba ] ) is best . in this region , the ratio @xmath246 , defined in appendix [ app : pba_a - a ] , is @xmath247 for insb , @xmath248 for gasb , @xmath249 for inp , @xmath250 for gaas , and @xmath251 for znse . at higher photon energies ,
the dispersion of @xmath252 deviates from the pbaexpression due to band nonparabolicity and warping , @xmath2-dependence of matrix elements , and transitions from the split - off band , which are not included in ( [ eq : popcontrolpba ] ) .
if we remove the two - band transitions @xmath180-@xmath253-@xmath164 , @xmath181-@xmath254-@xmath164 , and @xmath147-@xmath255-@xmath164 , then the calculation of @xmath256 ( or @xmath237 ) is unchanged .
this is expected for materials of zinc - blende symmetry @xcite .
further , many years ago aspnes argued that the so - called `` virtual hole terms '' of the form @xmath181-@xmath257-@xmath164 and @xmath180-@xmath258-@xmath164 make only a small contribution to @xmath259 @xcite .
such terms have been neglected in some previous calculations of @xmath260 dispersion @xcite . by removing the virtual hole terms , leaving only @xmath261-@xmath103-@xmath164 transitions , we find @xmath237
is reduced by only 610% over the range from the band edge to 500 mev above the gap for gaas .
it is thus clear that inclusion of the @xmath103 bands is necessary for a calculation of population control .
for some purposes it is also sufficient , since if remote band effects are removed from the model , leaving the `` bare '' fourteen - band model @xcite , @xmath262 is decreased by only 710% from its full value for gaas . for most materials , the results in fig .
[ fig : pop_imchi2 ] are in reasonable agreement with previous calculations of @xmath263 @xcite , although most previous calculations had poor spectral resolution in this energy range .
however , for znse , the situation is more complicated .
the calculation of huang and chin is about an order of magnitude smaller than ours @xcite , and that of ghahramani et al .
is about 5 times smaller than ours @xcite .
note also that huang and chin calculated @xmath264 for znse to be an order of magnitude smaller than experimental results @xcite .
wagner et al . have measured the dispersion of @xmath265 , which is an upper bound on @xmath266 ; for znse it is about a factor of two smaller than our calculation of @xmath263 @xcite .
note that wagner et al .
give a different set of band parameters than we have used here @xcite .
the magnitude of @xmath267 determines the magnitude of population control , but in an experiment one is more interested in the depth of the phase - dependent modulation of the carrier absorption , i.e. the control ratio @xmath268 @xcite .
it is @xmath269 this ratio is largest for field amplitudes that equalize @xmath270 and @xmath271 @xcite ; in what follows , we assume this condition has been met .
the ratio then depends only on @xmath272 , @xmath273 , @xmath274 , and the polarizations of the two fields .
for light normally incident on a @xmath275 surface , linearly - polarized fields yield @xmath276 , while opposite circularly - polarized fields yield @xmath277 where @xmath278 and @xmath279 are two - photon absorption anisotropy and circular dichroism parameters @xcite .
stevens et al
. found that for light normally incident on a @xmath275 surface of gaas , opposite circularly polarized fields yield the largest ratio @xcite .
for light normally incident on a @xmath280 surface , fields linearly polarized along @xmath281 $ ] yield @xmath282 the polarization configuration that yields a global maximum for the control ratio depends on the material and photon energy ; we have found that ( [ ratio opposite circular ] ) is the maximum except for very close to the band edge , where ( [ e : ratio_111 ] ) is the maximum . to calculate the population control ratio , it is desirable to use values of @xmath237 , @xmath283 , and @xmath284 calculated within the same set of approximations .
we use microscopic expressions for @xmath283 and @xmath284 in the independent - particle approximation @xcite , and calculate them within the fourteen - band model .
note that our calculation of two - photon absorption ( @xmath284 ) is similar to that of hutchings and wherrett @xcite , but that our model includes remote band effects .
[ fig : pop_ratio ] shows the calculated spectra of the population control ratio ( [ ratio opposite circular ] ) for various semiconductors . for each material ,
the ratio is close to unity at the band edge , then drops steeply , but flattens out to some non - zero ratio as photon energy is increased . in general
, the smaller the band gap ( or conduction band effective mass ) of the material , the narrower the range over which the ratio drops , and the lower the ratio at higher excess photon energy .
worth noting is the particularly large ratio for znse .
also plotted in fig.[fig : pop_ratio ] is the ratio appropriate for linearly - polarized fields normally incident on a @xmath275 surface of gaas , which was the configuration in the experiment of fraser et al @xcite .
for all materials , the ratio ( [ e : ratio_111 ] ) reaches exactly unity at the band edge , in agreement with the pba calculation ( [ popcontrol_ratio_bandedge ] ) in appendix [ app : pba_a - a ] .
surface of insb , gasb , inp , gaas , and znse .
the blue , dotted line is the ratio for linearly polarized fields normally incident on a @xmath275 surface of gaas . ]
the only previous theoretical calculation of the population - control ratio , which was for gaas , missed finding the large ratio near the band edge because it was based on _ ab initio _ calculations of @xmath283 , @xmath284 and @xmath237 that had poor spectral resolution near the band edge @xcite . over the rest of the spectrum shown in fig . [
fig : pop_ratio ] , it is about a factor of two smaller than our calculation .
this is consistent with the previous calculation being based on a calculation of the two - photon absorption coefficient @xmath284 that is too large by comparison with other calculations @xcite .
the population - control ratio has been measured only in gaas @xcite .
the measured ratios on @xmath275-gaas , at excess photon energies of 180 mev @xcite and 312 mev @xcite were 4 to 5 times smaller than our calculation .
some of the difference can be attributed to phase mismatch and large sample thickness @xcite .
an experiment on a @xmath285-grown multiple quantum well was complicated by an additional cascaded second harmonic effect @xcite .
spin - current density can be quantified by a second - rank pseudotensor @xmath286 defined as the average value of the product @xmath120 , where @xmath46 is the velocity operator and @xmath119 is the spin operator @xcite .
note that some authors alternately choose the first index to represent spin and the second index to represent velocity @xcite . also , due to the spin - orbit part of the velocity operator
the so - called `` anomalous '' velocity [ the second term in ] @xmath46 and @xmath119 do not commute , and thus @xmath120 is not hermitian .
instead , one should take @xmath287 as the operator for spin - current .
but since we neglect the anomalous velocity ( see appendix [ app : kdepso ] ) , this is not necessary . the spin - current injection rate due to the field can be written @xmath288 where the pseudotensor @xmath289 describes one - photon spin - current injection @xcite , the pseudotensor @xmath290 describes two - photon spin - current injection , and @xmath291 is `` 1 + 2 '' spin - current injection @xcite .
the fifth - rank pseudotensor @xmath292 has intrinsic symmetry @xmath293 . in an isotropic material
, @xmath294 has three independent components , while in a cubic material ( with @xmath22 , @xmath156 , or @xmath155 symmetry ) @xmath294 has six independent components .
the four parameters @xmath295 , @xmath296@xmath297 , that we used previously to describe spin - current injection in an isotropic model @xcite can be reduced to three independent components with identities such as @xmath298 @xcite . for a cubic material
, @xmath294 has 54 non - zero elements in the principal cubic basis , and can be written @xmath299 where the non - isotropic tensor @xmath167 has nonzero components @xmath300 , where @xmath162 , @xmath163 , and @xmath164 denote components along the principal cubic axes .
the six independent components are @xmath301 , @xmath302 , @xmath303 , @xmath304 , @xmath305 , and @xmath306 .
thus in a cubic material , @xmath307 note that the injection of @xmath308 is zero in a cubic material , i.e. , @xmath309 is traceless . in an isotropic model , such as the one we used previously @xcite , @xmath310 .
the connection to our previous notation is @xmath311 , @xmath312 , and @xmath313 @xcite .
the spin - current injection can be divided into a contribution from electrons @xmath314 , and a contribution from holes @xmath315 ; that is , @xmath316 ( similarly , @xmath317 ) .
expressions in the pba for both the electron and hole spin current are given elsewhere @xcite ; here we focus on the electron spin current , since hole spin relaxation is typically very fast @xcite .
a microscopic expression for the spin - current injection was derived previously in the fermi s golden rule ( fgr ) limit of perturbation theory and applied to a model in which all bands are doubly degenerate @xcite .
however , it is unsuitable for a calculation with @xmath16 , which accounts for the small splitting of the spin degeneracy that occurs in materials of zinc - blende symmetry @xcite .
if the spin - split bands were well separated , then the microscopic expression for @xmath314 would be @xmath318 \delta \left ( 2\omega -\omega _ { cv}\left ( \mathbf{k}\right ) \right ) , \end{split}\ ] ] where @xmath319 is a normalization volume ; the one - photon amplitude @xmath320 is @xmath321 where the charge on an electron is @xmath322 ( @xmath323 ) , and the two - photon amplitude @xmath324 is @xmath325 however , for the photon energies and materials studied here , the spin - splitting is small ; it is comparable to the broadening that one would calculate from the scattering time of the states , and also to the laser bandwidth for typical ultrafast experiments .
thus , the spin - split bands should be treated as quasidegenerate in fgr , with the result @xmath326 + c.c .
, \end{split}\ ] ] where the prime on the summation indicates a restriction to pairs @xmath327 for which either @xmath328 , or @xmath164 and @xmath329 are a quasidegenerate pair .
the optical excitation of the coherence between spin - split bands can be justified using the semiconductor optical bloch equation approach , as was done for the one - photon spin properties @xcite .
note that this issue does not arise for `` 1 + 2 '' current injection or `` 1 + 2 '' population control , since @xmath330 vanish between spin - split bands . using the time - reversal properties of the bloch functions , we find that @xmath331 is real , and can be written as @xmath332 \right\ } , \end{split}\ ] ] where @xmath333 .\label{eq : mccpv}\ ] ] that @xmath334 in is purely real is a consequence of the independent - particle approximation @xcite . ,
@xmath181 , and @xmath147 bands respectively , while the solid lines include all three transitions .
panel ( a ) shows @xmath335 ( black lines ) , @xmath336 ( red lines ) , and @xmath337 ( blue lines ) .
panel ( b ) shows @xmath338 ( black lines ) , @xmath339 ( red lines ) , and @xmath340 ( blue lines ) . ]
the spectra of the independent components of @xmath341 , calculated for gaas , are shown in fig.[fig : spincurrentgaas_initial ] and fig.[fig : spincurrentgaas_models ] .
figure [ fig : spincurrentgaas_initial ] also shows contributions from each possible initial valence band .
figure [ fig : spincurrentgaas_models ] shows the spin - current injection calculated with various degrees of approximation described in sec.[approximations_defined ]
. the only other calculation of `` 1 + 2 '' spin - current injection for bulk gaas is our earlier calculation , which used a spherical , parabolic - band approximation to the eight - band model and did not include the @xmath147 band as an intermediate state @xcite ; it is shown in fig.[fig : spincurrentgaas_models ] for @xmath335 , @xmath336 , and @xmath337 .
the term @xmath335 has the largest magnitude of the six independent parameters of @xmath334 .
since it is negative for @xmath180 and @xmath181 transitions but positive for @xmath147 transitions , it peaks in magnitude at @xmath342 just above @xmath343 ( the energy at which @xmath147 transitions become allowed ) .
two band terms make the largest contribution to @xmath335 , followed by three - band terms with @xmath180 or @xmath181 intermediate states .
the @xmath147 and @xmath103 intermediate states make a very small contribution to @xmath335 for excess energies less than 200 mev .
the warping of the bands is not important for @xmath335 , since the calculation with @xmath144 closely approximates the `` @xmath16 , no @xmath103 '' calculation , which includes the same intermediate states .
the `` @xmath144-pba , no @xmath147 '' calculation , which we derived previously @xcite , is a good approximation to @xmath335 at excess energies below 250 mev ; nonparabolicity becomes important at higher energies .
the @xmath180 contribution has a larger magnitude than the @xmath181 contribution in part because three - band terms increase the magnitude of the @xmath180 contribution , but decrease that of the @xmath181 contribution , as expected from the pba expression .
the term @xmath336 is negative for @xmath180 transitions , positive for @xmath181 transitions , and negligible for @xmath147 transitions .
the calculation `` @xmath16 , 2bt '' is a good approximation to the calculation @xmath16 .
however , the three - band terms are not small ; rather , they nearly cancel .
in particular the transition @xmath180-@xmath181-@xmath164 makes a large positive contribution to @xmath336 , while the transition @xmath180-@xmath147-@xmath164 makes a large negative contribution . since our earlier pba calculation included the former but not the latter @xcite
, it is a poor approximation to @xmath336 .
but by including only 2bts , it is a fair approximation for excess energies less than 200 mev .
this agreement is fortuitous , since the calculation @xmath144 underestimates the magnitude of @xmath336 , and the pba leads to an overestimation of the magnitude of @xmath336 .
the term @xmath337 is negligible when only 2bts are included , in agreement with the pba @xcite .
the @xmath180-@xmath181-@xmath164 transitions are positive , while the @xmath181-@xmath180-@xmath164 transitions are negative ; the former is larger , and thus @xmath337 is positive when @xmath147 intermediate states are neglected .
both @xmath181-@xmath147-@xmath164 and @xmath180-@xmath147-@xmath164 are negative and substantial enough to make the total @xmath337 negative .
consequently , our earlier pba result @xcite , which neglects @xmath147 intermediate states , is a poor approximation to @xmath337 .
upper conduction bands make a fairly small contribution to @xmath337 , and warping does not seem to be important for @xmath337 since the calculation with @xmath144 is a good approximation .
as expected , the terms @xmath338 , @xmath339 , and @xmath340 are zero when calculated with @xmath144 .
the term @xmath338 is negligible when only 2bts are included .
transitions with intermediate states in the set @xmath344 comprise roughly two - thirds of @xmath338 .
the anisotropy of these transitions is not simply due to the warping of the @xmath180 and @xmath181 bands , which we have determined by a calculation ( not shown ) using @xmath143 without the remote band contribution to the velocity .
rather , it comes from wave function mixing of the @xmath99 and @xmath98 states into the valence and @xmath164 band states .
the cubic anisotropy of two - photon absorption has been attributed to such wave function mixing @xcite .
the other third of the full @xmath338 is due to transitions with the @xmath103 intermediate state , which would be forbidden close to the @xmath72 point if the material were isotropic .
we also note that each three - band term makes a positive contribution to @xmath338 .
the term @xmath339 is nearly negligible when only 2bts are included .
transitions from the @xmath180 and @xmath181 bands have opposite sign , and those from the @xmath147 band are negligible .
about half of @xmath339 is due to the transitions @xmath180-@xmath181-@xmath164 and @xmath181-@xmath180-@xmath164 , and the other half is due to transitions with the @xmath103 intermediate states .
transitions with @xmath147 intermediate states are negligible . as with @xmath338 ,
the anisotropy of the @xmath180-@xmath181-@xmath164 and @xmath181-@xmath180-@xmath164 transitions is due to the wave function mixing of the @xmath99 and @xmath98 states into the @xmath180 , @xmath181 , and @xmath164 band states .
the term @xmath340 is positive for @xmath180 transitions , negative for @xmath181 transitions , and negligible for @xmath147 transitions .
the transitions @xmath180-@xmath147-@xmath164 and @xmath181-@xmath147-@xmath164 account for most of the value of @xmath340 , but 2bts are not negligible . transitions with @xmath103 intermediate states reduce the value of @xmath340 by as much as 10% .
most of @xmath340 , especially at energies less than 200 mev , is due to the warping of the @xmath180 and @xmath181 bands .
consistent with this , we find that remote band effects are somewhat important for @xmath340 ; when remote band effects are removed , the calculation of @xmath340 is about 25% larger than the full calculation .
note that @xmath340 is far more sensitive to remote band effects than any other optical property calculated in this article . in fig .
[ fig : spincurrent_4materials ] we plot the spectra of the independent components of the spin current density pseudotensor for insb , gasb , inp , and znse .
the spin current tensor is largest for insb in agreement with the pba expressions in appendix [ app : pba_a - a ] .
we also note that @xmath337 is positive for insb and gasb at low excess photon energy , whereas it is negative for inp , gaas , and znse .
co - circularly polarized fields generate a spin - polarized current , which can be characterized by its degree of spin polarization @xmath345 , where @xmath346 is a unit vector normal to the polarization plane of the fields , and @xmath347 is a unit vector in the direction of @xmath348 @xcite .
essentially , @xmath349 .
since this measure aims to characterize the photoexcited distribution of electrons , we neglect holes from both @xmath350 and @xmath351 in this calculation @xcite . for fields normally incident on a @xmath217 surface ( i.e.@xmath352 ) , the spin current is @xmath353 , \end{split}\ ] ] where @xmath354 , the current is @xmath355 , and the degree of spin polarization is @xmath356 for fields normally incident on a @xmath275 surface , @xmath357 , and @xmath358 the degree of spin polarization is plotted for gaas in fig.[fig : sc_measures](a ) .
the cubic anisotropy is small , but clearly seen , especially at low excess photon energies .
the other materials have very similar degrees of spin polarization . a pure spin current , without an electrical current ,
can be generated with cross - linearly polarized fields @xcite .
we consider fields polarized in the @xmath217 plane , with the @xmath359 field polarized at an angle @xmath195 to the @xmath360 axis ( i.e. @xmath361 $ ] ) and the @xmath10 field polarized at an angle @xmath195 to the @xmath362 axis ( @xmath363 and @xmath364 ) .
the spin current is @xmath365 \end{split}\ ] ] this pure spin current is typically measured by the resulting displacement of up and down spins @xcite .
the finite displacement results from transport and scattering of the electrons . using the boltzmann transport equation in the relaxation time approximation with space - charge effects justifiably neglected @xcite , one finds @xmath366 @xcite . here , @xmath367 is the displacement of spins measured with respect to the quantization direction @xmath224 , and @xmath368 is the momentum relaxation time .
we assume the field intensities have been chosen to balance one- and two - photon absorption , a condition that is @xmath195-dependent due to the cubic anisotropy of two - photon absorption .
thus , @xmath369 and @xmath370 where @xmath371 is the two - photon absorption cubic - anisotropy factor given explicitly in the next section @xcite . at @xmath372 and @xmath373
, @xmath374 is parallel to @xmath375 .
the spin separation distance is plotted in fig .
[ fig : sc_measures](b ) , where we have assumed a momentum relaxation time of 100 fs for each material .
this calculation of the spin separation distance is a significant improvement over our initial calculations @xcite , which used the eight - band pba and neglected three - band terms from the two - photon amplitude ( `` @xmath144-pba , 2bt '' ) .
stevens et al .
measured a spin separation distance of 20 nm in a gaas multiple quantum well at an excess photon energy of 200 mev , and estimated a momentum relaxation time of @xmath376 fs @xcite . for @xmath376 fs , we calculate a spin separation distance of 20.0 nm for bulk gaas at 200 mev .
hbner et al . measured a spin separation distance of 24 nm ( the photoluminescence spot separation is half this distance ) in cubic znse at an excess photon energy of 280 mev , and estimated a momentum relaxation time of @xmath377 fs @xcite . the calculation in fig .
[ fig : sc_measures](b ) yields @xmath378 nm for znse at 280 mev . in both cases
, we now find very good agreement with the experiment , whereas the previous model resulted in larger spin separation distances .
of course , this agreement is contingent on the accuracy of the momentum relaxation time estimates .
note that both the degree of spin polarization for co - circularly polarized fields and the spin - separation distance , plotted in fig.[fig : sc_measures ] , have a kink at excess photon energy @xmath82 and decrease at higher excess photon energies .
a similar kink and decrease , due to the onset of transitions from the split - off band , occurs for both one - photon spin injection @xcite and two - photon spin injection @xcite .
the spin injection rate due to the field can be written @xmath379 , where @xmath119 is the macroscopic spin density , @xmath380 is one - photon spin injection @xcite , @xmath381 is two - photon spin injection @xcite , and @xmath382 is `` 1 + 2 '' spin control @xcite . in previous sections and in some of the expressions below
, we use @xmath119 to denote the single - particle spin operator .
it should be obvious by context when @xmath119 refers to the macroscopic spin density and when it refers to that spin operator .
the fourth - rank pseudotensor @xmath383 has intrinsic symmetry on the indices @xmath384 .
such a pseudotensor is zero in the presence of inversion symmetry ; hence , `` 1 + 2 '' spin control requires materials of lower symmetry . for zinc - blende symmetry ( point group @xmath22 ) ,
a general fourth - rank pseudotensor has three independent parameters and 18 non - zero elements in the standard cubic basis ; forcing the @xmath384 symmetry leaves two independent parameters @xmath385 the spin injection has a contribution from electrons @xmath386 , and a contribution from holes @xmath387 ; that is , @xmath388 , and @xmath389 .
we treat the spin - split bands as quasidegenerate when taking the fgr limit of perturbation theory , as discussed for the spin current in section [ sec : spincurrent ] , deriving the microscopic expression @xmath390 + c.c .
, \end{split}\ ] ] where the prime on the summation indicates a restriction to pairs @xmath327 for which either @xmath328 , or @xmath164 and @xmath329 are a quasidegenerate pair . using the time - reversal properties of the bloch functions , we find that @xmath391 is purely imaginary and can be written @xmath392 \right\ } , \label{eq : zeta_micro } \end{split}\ ] ] where @xmath393 is given in eq .
( [ eq : mccpv ] ) .
( black lines ) , @xmath394 ( red lines ) , and @xmath395 ( blue line ) with breakdown into initial states .
dotted lines include transitions from the @xmath181 band , dashed lines include transitions from the @xmath180 band , dashed - dotted lines include transitions from the @xmath147 band , and solid lines include all transitions . ]
( solid black lines ) , `` @xmath16 , no @xmath103 '' ( dotted lines ) , `` @xmath16 , no @xmath103/@xmath147 '' ( dashed - dotted lines ) , `` @xmath16 , 2bt '' ( dashed lines ) , and `` @xmath16-pba '' ( solid grey lines ) . ]
the spectra of @xmath396 and @xmath397 for gaas are shown in figs .
[ fig : spin_gaas_initials ] and [ fig : spin_gaas_models ] .
figure [ fig : spin_gaas_initials ] also shows the contributions from each possible initial valence band . figure [ fig : spin_gaas_models ] shows the spin control calculated with various degrees of approximation described in sec.[approximations_defined ] .
the term @xmath396 decreases from zero at the band edge to a maximum negative value at 40 mev , mostly due to transitions from the @xmath180 band , and is positive at higher excess photon energies , mostly due to transitions from the @xmath181 band .
the low energy behavior is in agreement with the pba result ( [ eq : spin_ia_pba ] ) , in which the ratio of @xmath398 transitions is @xmath399 .
transitions with @xmath147 and @xmath103 intermediate states dominate the decrease in @xmath400 at low excess photon energies , as seen in fig .
[ fig : spin_gaas_models](a ) ; they are the only non - zero transitions in the pba result ( [ eq : spin_ia_pba ] ) .
the contribution from @xmath103 intermediate states is negative and approximately constant over most of the spectrum , whereas the contribution from @xmath147 intermediate states changes from negative to positive as transitions from the @xmath147 band become allowed ( @xmath401 ) .
the contribution from 2bts , which is zero in the pba , is positive over the whole spectrum .
the breakdown of the pba is due to the increase in magnitude of the 2bts .
in fact , the sum of the pba and the 2bts is a good approximation to the full calculation .
we also note that a calculation with @xmath143 for @xmath402 yields a nearly negligible result ; thus , the contribution from intermediate states within the set @xmath403 ( including 2bts ) is due to the mixing of the @xmath98 and @xmath99 wavefunctions with these states .
the term @xmath397 is larger in magnitude than the term @xmath402 over most of the calculated spectrum .
it falls to a maximum negative value at 95 mev , sharply increases when transitions from the @xmath147 band become allowed , and is positive at higher excess photon energy . at lower photon energies , transitions from the @xmath180 band and transitions from the @xmath181 band
both make negative contributions to @xmath397 ; in the pba result ( [ eq : spin_ib_pba ] ) the ratio of @xmath398 transitions is @xmath404 .
fig.[fig : spin_gaas_models](b ) reveals that @xmath397 is essentially due to contributions from @xmath103 intermediate states , and 2bts . over the whole spectrum ,
the former are negative while the latter are positive .
the smallness of the contribution from @xmath147 intermediate states is also seen in the pba result ( [ eq : spin_ib_pba ] ) , since @xmath405 in that expression .
we also note that a calculation with @xmath143 for @xmath406 yields a nearly negligible ; thus , the contribution from intermediate states within the set @xmath407 ( including 2bts ) is due to the mixing of the @xmath98 and @xmath99 wave functions with these states .
we have also calculated the spin - control pseudotensor for the semiconductors insb , gasb , inp , and znse .
the results are shown in fig .
[ fig : spin_4materials ] along with the parabolic - band approximations ( [ eq : spin_ia_pba ] ) and ( [ eq : spin_ib_pba ] ) .
the magnitude of spin control is determined by @xmath408 , but in an experiment one is more interested in the depth of the phase - dependent modulation of the spin polarization signal .
one possible definition for the signal is the ratio of spin injection measured with both @xmath409 and @xmath410 fields to the sum of the spin injections measured with circularly polarized fields of each frequency @xcite .
the amplitude of its modulation is @xmath411 where the argument @xmath412 indicates injection with a @xmath413 polarized field .
this ratio , which is largest for field amplitudes that equalize @xmath414 and @xmath415 , was measured by stevens et al . with excess photon energies of 150 mev and 280 mev @xcite .
the ratio has an undesirable feature : it can exceed unity . close to the band edge in many semiconductors ( at 2 mev in gaas ) , there is a photon energy for which @xmath416 @xcite . at that photon energy , it is impossible to choose field amplitudes to balance one- and two - photon spin injection with circular polarized fields [ @xmath417 , and thus the maximum ratio has a singularity .
even if the condition @xmath418 is relaxed , the ratio can exceed unity .
this is because @xmath419 and @xmath420 have opposite sign close to the band gap @xcite , and thus it is possible , by appropriate choice of field amplitudes , to make the denominator of the ratio arbitrarily small . an alternate ratio to characterize the spin control , which has an upper bound of unity , is @xmath421 it is the amplitude of phase - dependent oscillation of the degree of spin polarization , and it is most useful when there is little or no population control .
we assume the fields are chosen to balance one- and two - photon absorption . for most photon energies and materials
this is nearly the same as balancing one- and two - photon spin injection .
for normal incidence on a @xmath275 sample , opposite circularly polarized fields yield @xmath422 for normal incidence on a @xmath285 sample , opposite circularly polarized fields @xmath423 and orthogonal linearly polarized fields ( @xmath424-polarized ) yield @xmath425 where @xmath426 @xcite , and @xmath427 is the angle between the polarization of the @xmath409 field ( @xmath428 ) and the @xmath429 $ ] axis , which lies in the @xmath285 plane .
the angle that maximizes @xmath430 depends on photon energy through @xmath431 and @xmath371 .
we determine it numerically . ) ] . in ( a ) , for gaas , black lines are @xmath432-incident , opposite circularly polarized fields ; green lines are @xmath433-incident , opposite circularly polarized fields ; and red lines are @xmath434-incident , orthogonal linearly polarized fields . in ( b ) , for insb , gasb , inp , gaas , and znse , solid lines are @xmath275-incident , opposite circularly polarized fields .
the dotted line in ( b ) is @xmath435-incident , orthogonal linearly polarized fields for inp . ]
the ratio @xmath430 for gaas is plotted in fig.[fig : spin_ratio](a ) . for @xmath432-incidence ,
opposite circularly polarized fields yield the highest ratio over the studied range of photon energies .
for @xmath285-incidence , opposite circularly polarized fields yield the highest ratio , except for between 190 mev and 415 mev when @xmath424-polarized fields the highest ratio .
for @xmath424-polarized fields , the angle that yields the largest ratio decreases from 0.99 rad to 0.53 rad from the band edge to 320 mev , and is zero for higher excess energies .
the ratio @xmath430 for the five semiconductors insb , gasb , inp , gaas , and znse are plotted in fig .
[ fig : spin_ratio](b ) . at low photon energy ,
opposite circularly polarized fields normally incident on @xmath275 yield the largest ratio for insb , gasb , gaas , and znse , whereas orthogonal linearly polarized fields normally incident on @xmath436 yield the largest ratio for inp .
we have studied the four `` 1 + 2 '' coherent control effects current injection , spin - current injection , population control , and spin control in bulk semiconductors having zinc - blende symmetry .
we used an empirical , fourteen - band @xmath437 hamiltonian and examined the relative importance to each effect of the possible initial and intermediate states .
we have also studied the crystal orientation and polarization dependencies of each effect .
cubic anisotropy is small in some cases , but large in others .
we have compared the numerical calculation with analytic expressions , derived in the parabolic - band approximation , to show the value and limitations of the latter .
the pba expressions , where they are accurate , are useful to show how the effects scale in different materials .
the comparison between the two approaches establishes that , at low excess photon energies , `` 1 + 2 '' current injection and `` 1 + 2 '' spin - current injection are due to interference of allowed one - photon transitions and allowed - forbidden two - photon transitions , whereas `` 1 + 2 '' population control and `` 1 + 2 '' spin control are due to interference of allowed one - photon transitions and allowed - allowed two photon transitions .
it also explains the large population- and spin - control ratios predicted by the fourteen - band calculation close to the band edge , where allowed - allowed two - photon transitions dominate allowed - forbidden two - photon transitions . neither `` 1 + 2 '' population control , nor `` 1 + 2 '' spin control have yet been experimentally studied in that spectral range .
the anomalous velocity , i.e. @xmath438 , which leads to @xmath2-dependent spin - orbit coupling in @xmath47 from the term @xmath439 , is often neglected in @xmath1 models @xcite .
some authors have treated matrix elements of @xmath440 as additional independent parameters @xcite .
for example , bahder , who gives the matrix for @xmath441 within the eight - band model , defines the model parameter @xcite @xmath442 ostromek used the value @xmath443 ev to fit the eight - band model to experimental results @xcite .
we here relate matrix elements of @xmath440 ( and hence matrix elements of @xmath444 ) to other parameters of the model , thereby demonstrating that they can be neglected for the effects we consider .
bir and pikus showed that the identity @xmath445 = i\hbar \bm{\nabla } v$ ] leads to @xmath446 @xcite .
an application of that identity to the remaining nonzero matrix elements yields @xmath447 and similar results for cyclic permutations and hermitian conjugates of these . the energies @xmath448 , @xmath449 , and @xmath450 are the eigenvalues of @xmath52 , @xmath451 , and @xmath452 with respect to the hamiltonian @xmath50 .
their values are fixed by the requirement that the eigenvalues of @xmath70 yield the parameters @xmath81 , @xmath453 , @xmath63 , and @xmath454 @xcite . neglecting the small contribution from @xmath455 , @xmath456 , @xmath457 , and @xmath458 .
thus , gives matrix elements of @xmath440 in terms of other model parameters .
in particular , with parameters from table [ table : parameters ] for gaas , we find @xmath459 ev . from the point of view of the theory of invariants @xcite ,
@xmath2-dependent spin - orbit coupling amounts to using different values of @xmath74 for @xmath460 and @xmath461 valence bands ( and similar changes for @xmath462 coupling and @xmath76 coupling ) @xcite . in terms of @xmath463 , @xmath464 for couplings with @xmath461 bands and @xmath465 for couplings with @xmath460 bands . from , @xmath466
this is very small , since the numerator is on the order of ev , whereas @xmath467 ev . and
since this relative change in the matrix element depends on the ratio of @xmath463 to @xmath74 , even the overly large coupling value of @xmath443ev has only a small effect on optical properties @xcite . for comparison ,
consider interband spin - orbit coupling parameterized by @xmath65 . in the eight - band model ,
interband spin - orbit coupling is a remote band effect ( since it is a coupling with the @xmath103 bands ) , which effectively causes @xmath468 $ ] and @xmath469 $ ] .
thus , @xmath470 this effect , which is included in our calculation , is small ( it is @xmath471 in gaas ) , but it is orders of magnitude larger than the relative change due to @xmath2-dependent spin - orbit coupling .
the above suggests that @xmath2-dependent spin - orbit coupling can be neglected for the processes we consider in bulk , cubic materials .
in this appendix , we discuss parabolic - band approximation ( pba ) expressions , which are perturbative in the bloch wave vector @xmath2 , for `` 1 + 2 '' coherent control effects . there have been several different calculations of @xmath472 in the pba @xcite . using a two - band model ( one conduction and one valence band ) , atanasov _
et al_. obtained @xmath473 and @xmath474 @xcite . using a three - band model , but only accounting for two - band terms , shiek - bahae studied the approximate scaling of `` 1 + 2 '' current injection spectra with the band gap @xmath81 and concluded that @xmath185 and @xmath178 are proportional to @xmath475 , where @xmath476 @xcite . our earlier pba calculation was based on an 8-band model , included both two- and three - band terms in the two - photon amplitude , but did not include terms with the @xmath147 band as an intermediate state @xcite .
more recently , we included the @xmath147 band as an intermediate state , but only for two - band terms @xcite .
the 2bts in the 8-band model result differ from the 2bts in the three - band model result of sheik - bahae by material independent factors . where @xmath478 , @xmath479 , @xmath480 is given in ref . , and @xmath481
. in and ( @xmath482 and @xmath483 ) , the first term is from the @xmath180-@xmath164 transition , the second term is from the @xmath180-@xmath181-@xmath164 transition , the third term is from the @xmath181-@xmath164 transition , and the fourth term is from the @xmath181-@xmath180-@xmath164 transition . in for @xmath484 ,
the first term is from the @xmath180-@xmath485-@xmath164 transition , and the second term is from the @xmath181-@xmath180-@xmath164 transition .
note that two - band terms make no contribution to @xmath484 . to calculate optical effects due to the interference of allowed one - photon transitions and allowed - allowed two - photon transitions , we approximate the spin and velocity matrix elements and the energy denominator by their values at the @xmath72 point , and approximate the energy bands in the @xmath486-function as spherical and parabolic , neglecting the small @xmath2-linear term @xmath90 and the small @xmath113 spin - splitting .
we used this method previously for two - photon spin injection @xcite .
since bands are degenerate at the @xmath72 point , the lowest - order approximation to the matrix elements still depends on the direction @xmath487 @xcite . however , by averaging the microscopic expression over physical systems rotated by each point group operation [ which is equivalent to averaging over each term in eq .
[ macro_zetaia ] or eq .
[ macro_zetaib ] ] , one can make the calculation using @xmath7-point states with pseudo - angular momentum quantized along @xmath488 .
the integral over @xmath2 becomes a straightforward integral over the density of states in this approximation .
the @xmath7-point basis states are given in .
however , all but the @xmath97 states are not eigenstates at the @xmath72 point due to spin - orbit coupling between upper conduction and valence bands parameterized by @xmath71 . using eigenstates to first order in @xmath71 @xcite , we find where @xmath490 % \end{split } \\
z_{\pm } ^{\prime \prime } & = -\frac{\delta ^{-}e_{p^{\prime } } } { 3 } \frac{1}{e_{0}^{\prime } + \delta _ { 0}^{\prime } } \left ( \frac{1}{e_{0}^{\prime } -\hbar \omega } \pm \frac{1}{e_{0}^{\prime } + \delta _ { 0}^{\prime } -\hbar \omega } \right)\end{aligned}\ ] ] in @xmath491 , the first term is from intermediate @xmath492 states and the second term is from intermediate @xmath493 and @xmath494 states . in @xmath495 ,
the first term is from intermediate @xmath147 states , the second term is from intermediate @xmath492 states , and the third term is from intermediate @xmath493 and @xmath494 states . in @xmath496 ,
the first term is from intermediate @xmath147 states , and the second term is from intermediate @xmath494 and @xmath493 states .
the term @xmath497 can be neglected for typical semiconductors .
note that @xmath498 has contributions only from intermediate @xmath147 and @xmath492 states .
this only includes transitions from initial @xmath180 and @xmath181 states ; transitions from initial @xmath147 states , which contribute when @xmath499 , have been neglected .
we derive an expression for population control using the same method used above for spin control . to first order in @xmath71 , @xmath500 \frac{\sqrt{2\hbar \omega -e_{g}}}{\left ( 2\hbar \omega \right ) ^{3}}\sqrt{e_{q}}\left ( x_{1}+x_{2}+x_{3}\right ) , \label{eq : popcontrolpba}\ ] ]
where @xmath501 , \\ x_{3 } & = -\frac{\delta ^{-}}{3}\frac{e_{p^{\prime } } } { e_{0}^{\prime } + \delta _ { 0}^{\prime } } \left ( \frac{1}{e_{0}^{\prime } -\hbar \omega } + \frac{1}{e_{0}^{\prime } + \delta _ { 0}^{\prime } -\hbar \omega } \right ) .\end{aligned}\ ] ] note that @xmath502 is positive .
for typical semiconductors , @xmath503 can be neglected and @xmath504 in @xmath505 , the most important term is the last , which comes from the interference of @xmath506-@xmath147-@xmath164 two - photon transitions and @xmath506-@xmath164 one - photon transitions .
the expression ( [ eq : popcontrolpba ] ) only includes the allowed - allowed transitions from the @xmath180 and @xmath181 bands . at photon energies for which @xmath507 , one should add the contribution due to the transition @xmath147-@xmath103-@xmath164 . because of , is also an analytical expression for @xmath508 .
jha and wynne have also used @xmath2-independent velocity matrix elements and spherical , parabolic bands to derive an expression for @xmath509 , but they did not include the interband spin - orbit coupling term @xmath65 @xcite . taking the imaginary part of their eq.4.4 for @xmath510 , and correcting a factor of @xmath511 error , reproduces the @xmath512 one would find from but with @xmath513 .
also , they make the approximation @xmath514 in the term @xmath515 . to get a pba expression for the population control ratio requires pba expressions for one- and two - photon absorption .
we take the same approach used to derive , but for simplicity , we take @xmath114 in the following . in the pba , at photon energies @xmath516 ,
one - photon absorption is @xmath517 in a material of cubic symmetry , the two - photon absorption tensor @xmath274 has three independent components @xmath518 , @xmath519 , and @xmath520 , which are alternately parameterized by the set @xmath521 ( see sec.[sec : population ] ) . the allowed - forbidden two - photon absorption in the isotropic kane model , neglecting three- and four - band terms , is @xmath522,\ ] ] where @xmath523 note the additional symmetry , @xmath524 in this isotropic model .
the allowed - allowed two - photon absorption , neglecting @xmath525 , has @xmath526 and @xmath527 which agrees with arifzhanov and ivchenko @xcite .
thus , at photon energies for which allowed - allowed transitions dominate two - photon absorption , @xmath528 whereas when allowed - forbidden transitions dominate two - photon absorption , @xmath529 this work was financially supported by the natural science and engineering research council , photonics research ontario , and the us defense advanced research projects agency .
we gratefully acknowledge many stimulating discussions with ali najmaie , fred nastos , eugene sherman , art smirl , marty stevens , and henry van driel . | quantum interference between one- and two - photon absorption pathways allows coherent control of interband transitions in unbiased bulk semiconductors ; carrier population , carrier spin polarization , photocurrent injection , and spin current injection can all be controlled .
we calculate injection spectra for these effects using a @xmath0 @xmath1 hamiltonian including remote band effects for five bulk semiconductors of zinc - blende symmetry : insb , gasb , inp , gaas , and znse .
microscopic expressions for spin - current injection and spin control accounting for spin split bands are presented .
we also present analytical expressions for the injection spectra derived in the parabolic - band approximation and compare these with the calculation nonperturbative in @xmath2 . |
advances in computational power and increasingly accurate techniques for estimating the current state of the earth s atmosphere have significantly improved numerical weather prediction ( nwp ) @xcite . as state estimation error
is reduced due to improved methods of data assimilation , error in the model tendency plays an increasing role in the uncertainty of predictions at every temporal and physical scale @xcite . in 1978 , leith introduced a statistical technique to correct model tendency error , in which short model forecasts are compared to a time series of reference `` truth '' to estimate both state - independent model bias , and state - dependent error components which are approximated by a least - squares linear function of the model state @xcite .
more recently , empirical correction has been employed with success in atmospheric models with relatively few degrees of freedom ( e.g. @xmath0 in @xcite ) , and low - dimensional modifications of the technique involving , for example , singular value decomposition ( svd ) of the state - dependent correction operator ( the least - squares linear function proposed by leith ) have proven successful in models with as many has @xmath1 degrees of freedom @xcite . in this study , we apply the original technique developed by leith to simple three - dimensional lorenz models @xcite , where in addition to testing the effectiveness of empirical correction , we aim to understand the dynamical ramifications of a statistical approach to the correction of model tendency error . the model tendency @xmath2 is defined as the change in state variables over one timestep of numerical integration , which we denote as a time derivative : @xmath3 where @xmath4 is the atmospheric state - vector , typically with @xmath5 degrees of freedom for nwp . note that @xmath6 represents the model state , whereas @xmath7 will represent the true system state , in terms of model variables .
given the state @xmath4 of a physical system , @xmath8 consists of all the known physics , forcings , and parameterizations of sub grid - scale processes .
to make a one - timestep model forecast , we approximate the true change in state variables over that time by the model tendency , @xmath9 , and the error @xmath10 in that approximation is called the _ tendency error _ : @xmath11 even with perfect estimates of the current state of the atmosphere , the model tendency error would quickly separate forecasts from the truth , due to the atmosphere s chaotic dynamics @xcite . as a result , techniques for reducing the model tendency error represent a current and active research area , and those that are applicable independent of the specific model are of special interest . clearly , one would like to improve the physics represented by @xmath8 from first principles . in what follows
, we assume this improvement has met the limit of diminishing returns , and move towards a statistical approach .
the general strategy of empirical correction is to compare short forecasts generated by a model to observations of the system being modeled over some training period .
if the state - space of the system is well represented by the training period , and the model is a reasonable approximation of the true system , the forecast error statistics can be used to create an empirical correction that pushes the model closer to the truth _ at each timestep _ of numerical integration . adjusting the model
every timestep reduces the nonlinear growth of tendency error , providing more effective error reduction than a posteriori statistical correction @xcite .
this strategy is similar to nudging or newtonian relaxation in a data assimilation ( da ) context , where one is assimilating observations , except that here we are nudging with predicted , rather than observed , forecast error .
the present study is an investigation of a three - step empirical correction procedure inspired by the work of leith @xcite , delsole and hou @xcite , and more recently danforth et .
al . @xcite .
we first test its effectiveness in synchronizing lorenz systems @xcite with varied parameter - values in a perfect model scenario .
we then apply the correction to an alternative model derived by ehrhard and mller @xcite tuned to approximate the evolution of a toroidal thermosyphon , an experimental analogue to the original lorenz system .
the `` true '' climate is represented by a long - time , high - dimensional computational fluid dynamics ( cfd ) simulation of the thermosyphon .
an _ analysis _
, which is an approximation of the true system state in terms of model variables , is then created by three - dimensional variational ( 3dvar ) data assimilation and used for training and verification of the empirical correction .
this process mimics the application of empirical correction in an operational nwp setting .
we also verify the corrected model by direct comparison with observations of the truth .
the results in each experiment suggest that the correction procedure is effective for reducing error .
however , there is an associated cost of this short - term error reduction , which is evidenced by substantial qualitative differences between the dynamics of the corrected model and the true system , differences that were not present in the uncorrected model .
introduction of system - specific knowledge into the correction procedure is shown to mitigate some of that cost , while also improving error statistics further than the entirely general procedure .
the paper is structured as follows . in sec .
[ emcor ] we define the procedure for a general model @xmath8 .
the application of the technique in the perfect model scenario is addressed in sec .
[ pms ] , and in sec .
[ tcm ] we present the thermosyphon model correction .
finally , we discuss the results and conclude the paper in sec .
the correction procedure employed in this experiment consists of three steps : ( a ) training , ( b ) state - independent correction and ( c ) state - dependent correction .
the state - independent correction can be thought of as aligning the time - average of the model state with that of the true state .
likewise , the state - dependent correction can be considered an alignment of the model variance with that of the truth . to determine the correction terms , we compare short model forecasts to observations of the true system over a training period in a process called
_ direct insertion_. in general , comparing model forecasts to a true physical system requires estimates of the true system state in terms of the model state - variables .
consider a vector time - series @xmath12 of such estimates , which we will call the `` reference truth '' .
the amount of time @xmath13 , measured in model timesteps , between estimates is called the _ analysis window _ ; we assume it to be constant .
the process of direct insertion begins with the generation of a time - series @xmath14 of duration-@xmath13 model forecasts , where each forecast in the time - series is initialized from the previous state in the reference truth .
the first vector in the series , for example , will be @xmath15 , which is the model state resulting from an @xmath13-timestep forecast started with initial condition @xmath16 . subtracting each of the model forecast states @xmath14 from the corresponding reference true state @xmath12 produces a third time - series @xmath17 which represents the forecast errors after @xmath13 timesteps .
these errors result from differences between the model rate of change for each variable and the true rate of change , and they are commonly referred to as analysis corrections ( or increments ) .
see fig .
[ fig : dirins ] for a schematic of the procedure . represents a time series of the reference truth , and the analysis window represents the number of timesteps between estimations of the true system state .
@xmath6 represents a time series of forecasts with duration equal to the analysis window , each of which is started from the previous true state .
the time - average of the analysis corrections @xmath18 divided by the number of timesteps in the analysis window @xmath13 approximates the average ( state - independent ) model bias @xmath19 . ] finally , we separate each of the time series into anomalous and time - average components : @xmath20 where the expectation operator @xmath21 denotes averaging over the training period , and the primes denote _ anomalies _ , which are differences from the mean .
the time - average components will be used for state - independent correction as described in sec .
[ sec : si ] and the anomaly time - series will be used for state - dependent correction as detailed in sec .
[ sec : sd ] .
we turn our attention first to a state - independent correction of the form @xmath22 where the constant vector @xmath19 is the average model error ( bias ) to be determined .
recall that our goal here is to empirically align the time - averages of the state - variables in the model to those of the true system .
we call the time - averaged true system state the _ climatology _ , and we approximate it by @xmath23 , the average of the reference true state over the entire training period . the average of the analysis corrections @xmath24 over the training period provides an estimate for the systematic , state - independent error generated by the model during the analysis window , as explained in fig .
[ fig : dirins ] . dividing by the number of timesteps in the analysis window , then , we approximate the model bias by @xmath25 , and the bias - corrected model tendency is thus given by @xmath26 note that at this point we are approximating the model tendency error @xmath10 by the model bias @xmath19 alone .
we also wish to estimate any component of error that may depend on the system state , by approximating @xmath27 , where @xmath28 is a matrix operator to be described in the next section . to generate a linear state - dependent correction operator @xmath28 , we follow leith @xcite and delsole and hou @xcite , by first recomputing the forecast and correction time series in fig . [
fig : dirins ] using the state - independent corrected model , @xmath29 , and then decomposing into mean and anomalous components .
we seek an improved model of the form @xmath30 where @xmath31 includes both stages of correction .
letting @xmath32 $ ] be the tendency error of the improved model , we minimize the expected square tendency error @xmath33 , ( where @xmath34 is the transpose of @xmath35 ) , with respect to * l*. the minimization results in the formula @xmath36 as explained by danforth et .
@xmath37 is the average over the training sample of cross covariance matrices obtained by taking the outer product [ @xmath38\cdot[\mathbf{x}^{t'}(t)]^{tr}$ ] for each time @xmath39 .
@xmath40 is the average true - state covariance matrix , and * l * is known as leith s state - dependent correction operator .
when * l * operates on the current anomalous state @xmath41 , we can think of it as doing two things : ( 1 ) @xmath42 operates on @xmath41 , effectively relating the current state to the reference truth in the dependent sample , i.e. giving the best representation of the current state in terms of past states ; and then ( 2 ) @xmath37 operates on the result , determining what correction should be made .
this allows the model correction to adjust to different regions of state - space , and explains why the state - dependent correction can be thought of as attempting to align the model and true - state variances . in the next section ,
we describe the application of this three - stage procedure to align lorenz systems with different parameter values in a perfect model scenario , and in sec .
[ tcm ] we describe its application to couple a low dimensional model to a high dimensional toy climate simulation .
we also note here that the term _ corrected _ will imply the application of both state - independent and state - dependent correction , unless explicitly stated otherwise .
as a first step in the investigation of the correction technique , we consider its application to a model originally studied by lorenz @xcite .
the system of equations ( [ eq : lorenz ] ) represents fluid flow between two plates , rayleigh - bnard convection , in which convection cells form for certain parameter ranges .
however , with only slight modification ( the details of which appear in sec .
[ tcmed ] ) , they also describe the flow in a natural convection loop @xcite .
lorenz systems are covered exhaustively in publication @xcite , and thus provide a familiar platform on which to perform preliminary tests of strategies for predicting the future state of chaotic systems . in this perfect model scenario ,
the true system and the models share the structure of ( [ eq : lorenz ] ) and only differ in parameter values .
specifically , a true or _ nature _ run was created by integrating the lorenz system @xmath43 with the standard parameter set : @xmath44 .
models with the same @xmath45 and @xmath46 , but with @xmath47-values varying from 25 to 31 in increments of 0.5 ( except for @xmath48 ) were the subjects for correction .
for each of these 12 models , the correction algorithm was performed using the 4 different analysis windows @xmath49 and 8 timesteps , resulting in 48 distinct model - correction pairs in an exponential design .
the training and testing of the corrected models is detailed in the following sections , and a picture showing one particularly positive outcome appears in fig .
[ fig : traj1 ] . ) , a model with @xmath47-perturbation -2 ( @xmath50 ) , and the corrected model using an analysis window of @xmath51 timesteps .
all start from the same initial condition ( circle ) , and represent 2 time - units ( 200 timesteps ) of integration ( concluding with squares ) .
note that the corrected model trajectory is well aligned with the true trajectory for much longer than the uncorrected model trajectory . even after it deviates noticeably ,
the corrected model trajectory changes flow regimes ( switches lobes ) with the true trajectory .
in contrast , the uncorrected model trajectory deviates from the true trajectory almost immediately , and remains in the initial lobe . ]
a 100 time - unit nature run was generated by integrating system ( [ eq : lorenz ] ) from the initial condition @xmath52^{tr}$ ] ( see @xcite ) for 10000 timesteps of @xmath53 time units each , using fourth order runge - kutta .
for each of the 12 models and 4 analysis windows , a time - series of short forecasts was generated by direct insertion ( see fig . [
fig : dirins ] ) . as an illustrative example , consider training with an analysis window of @xmath54 .
the first forecast , @xmath55 , is a 4-timestep model forecast started with the true initial condition @xmath56 .
the second forecast , @xmath57 , is a 4-timestep model forecast started with the true state @xmath58 , and so on , resulting in 10000/4 = 2500 total short forecasts .
the state - independent correction @xmath59 was then computed as described in fig .
[ fig : dirins ] and sec .
[ sec : si ] .
next , the correction time - series @xmath17 was recomputed using a state - independent corrected version of the model , namely eq .
( [ eq : sicortend2 ] ) . correction and
true - state anomalies were calculated as in eq .
( [ eq : tsdecomp ] ) , and the cross covariance and covariance matrices were determined for each time @xmath39 and averaged over the training sample to obtain @xmath37 and @xmath40 as outlined in sec .
[ sec : sd ] . finally , the static leith operator * l * was computed and the training procedure was complete .
note that the training design imparts a statistical disadvantage upon the use of wider analysis windows .
specifically , doubling the analysis window halves the number of samples in the training period .
the design was chosen , despite this prejudice , to more accurately reflect an operational implementation in which the training data is likely to be drawn from a fixed period of time . however , to further support the validity of comparisons between models corrected with different analysis windows , we note that letting the training period be @xmath60 , ensuring that the number of samples is held constant at 10000 , yields results that are qualitatively indistinguishable from those presented here .
( color online ) plots of average rmse over 1000 trials for the uncorrected ( thick solid blue ) and corrected models ( all with @xmath50 ) using analysis windows of 8 , 4 , 2 and 1 , black dot , red dash - dot , green dash , solid magenta respectively . the exact model ( thin solid black )
is the same as the true system , but started from an initial condition perturbed randomly on the order of @xmath61 in each state variable .
time units are on the @xmath62-axis , where one time unit is 100 timesteps in the numerical integration . the technique s performance clearly degrades with widening analysis window . ] a new nature run , 10000 time - units ( one million timesteps ) in length , was generated starting from the last true state in the training period .
the purpose of beginning at the end of the training period was to obtain an _
independent _ truth with which to test the effectiveness of the corrected models . for each of the 48 corrected models ,
1000 randomly selected states from this new nature run were used as the initial state , and both the uncorrected and corrected models were integrated for 20 time units .
in addition , an _ exact _ model , with the same parameters as the truth , was integrated from a random perturbation of that same initial condition , on the order of @xmath61 in each state variable .
the purpose of the exact model is to represent an upper bound for the effectiveness of the empirical model error correction , i.e. demonstrating a case where the impact of the @xmath47-perturbation was corrected exactly , and forecast error comes only from the initial condition discrepancy .
[ fig : traj1 ] depicts the results of a single test with a particularly positive outcome .
trajectories of the truth ( solid black ) , uncorrected model ( blue dot ) , and corrected model ( red dash ) , all start from the same initial state and progress for two time - units .
the corrected model trajectory stays close to the true trajectory for much longer than the uncorrected model trajectory , and even after deviating markedly , the corrected model trajectory still switches lobes with the true trajectory .
two metrics were used to measure forecast accuracy : root mean square error ( rmse ) , and anomaly correlation ( ac ) .
the rmse is given at time @xmath39 by @xmath63 ^ 2 } } \label{eq : test.1}\ ] ] where @xmath6 is the model state and @xmath7 is the true state , and we are summing over the @xmath64 state variables in the lorenz system .
[ fig : allrmse ] plots the average rmse over 1000 trials performed for correction of the @xmath50 ( @xmath47-perturbation of -2 ) model using the 4 different analysis windows .
anomaly correlation is a metric frequently used in weather and climate modeling to determine the length of time for which a model forecast is useful .
the ac is given by @xmath65 where @xmath66 and @xmath67 are the anomalous model state and anomalous true state , respectively , at a particular time .
ac is essentially the dot product of the anomalous model state with the anomalous true state , normalized such that ac @xmath68 for a perfect model .
a forecast is typically considered useful for as long as its ac remains above 0.6 @xcite . as with rmse ,
the ac scores for each model , corrected and uncorrected , were averaged over 1000 trials to provide a good representation of model performance .
see fig .
[ fig : acbyraw ] for ac plots demonstrating the effects of changing analysis window length and parameter perturbation in the original model on the duration of useful forecasts .
+ fig .
[ fig : allrmse ] demonstrates that the tested empirical correction technique succeeds in reducing error in the perfect model scenario .
however , the importance of frequent observations of the truth during the training period is highlighted by the approach of the average corrected model rmse towards the uncorrected model rmse with widening of the analysis window .
it is also noteworthy that the corrected model remains almost as good as the exact model for a full time unit when observations are made every timestep . in this case
the rmse remains below 2 more than 5 times as long the uncorrected model .
[ fig : acbyraw ] demonstrates that along with reduced error , empirical correction has the potential to provide forecasts that are useful for much longer .
training with an analysis window of 1 timestep , the corrected model forecasts are useful for nearly 4 times longer than the uncorrected model forecasts . in light of the sensitivity of ac to analysis window length , the bottom panel of fig .
[ fig : acbyraw ] suggests that the accuracy of parameter values matters less for the effectiveness of the corrected model , as measured by error statistics , than does the frequency of observations in training .
however , it should be noted that error statistics are not the whole story .
the ability of the corrected models to reproduce the qualitative dynamical _ behavior _ of the true system accurately , like switching of lobes in the attractor , is only indirectly indicated by reduced forecast error .
we address this issue in sec .
[ sec : mqd ] . + a summary representation of the data in fig .
[ fig : acbyraw ] is shown in fig .
[ fig : skdurbyraw ] , where the duration of forecast usefulness is plotted vs. analysis window in the top panel , and vs. @xmath47-perturbation in the bottom panel . however , these plots still focus only on two cross sections of the 48 total model - correction pairs that were tested in the perfect model scenario , specifically the 4 corrected @xmath50 models in the top and the 12 corrections across all @xmath47-perturbations using an analysis window of 1 timestep in the bottom . to get a better sense of the relative importance , fig .
[ fig : skillful ] shows a surface plot of duration of useful forecasts for all combinations of analysis window and @xmath47-perturbation tested .
surprisingly , with an analysis window of 2 timesteps , the corrected models with the largest r - perturbation ( parameter error greater than 10% ) produce forecasts that are useful longer than those made by the uncorrected model with the smallest r - perturbation ( parameter error less than 2% ) . for systems with reasonably small model errors , this is an indication that empirical correction may improve forecasts more readily than parameter tuning .
-perturbation tested , top surface corrected , bottom surface uncorrected . on the left face
we can see the cross section represented by fig .
[ fig : skdurbyrpert ] . using an analysis window of 2 timesteps ,
corrected models with largest r - perturbation provide forecasts that are useful longer than those of uncorrected models with smallest r - perturbation.,title="fig : " ] +
we next investigate the effectiveness of the correction procedure in a more realistic situation , where the forecast model is structurally different ( in dynamics , dimension , parameterization , etc ... ) from the true system .
consider a fluid - filled , vertically - oriented natural convection loop , or thermosyphon , with circular geometry , a schematic of which appears in fig .
[ fig : loop ] . the constant temperature imposed on the wall of the lower half of the loop , @xmath69 , is greater than the constant temperature imposed on the wall of the upper half , @xmath70 , resulting in a temperature inversion . for large enough temperature differences
@xmath71 , convection dominates , and the flow undergoes chaotic reversals of direction referred to as _ flow regime changes _ , while remaining laminar @xcite .
these dynamics produce forecasting difficulties similar to those encountered in weather and climate prediction , and thus the thermosyphon provides a useful platform on which to test potential improvements to forecasting methods . , while the bottom half is heated to @xmath72 , creating a temperature inversion .
the three state variables in the low - dimensional model are ( 1 ) @xmath73 , the mass flow rate ( mean fluid velocity ) ; ( 2 ) @xmath74 , the horizontal temperature difference ; and ( 3 ) @xmath75 , the difference between the vertical temperature profile and the value it takes during conduction.,title="fig : " ] + the true system was represented by numerical simulation using the 2-d laminar navier - stokes equations along with the energy equation , and a finite - volume - based flow modeling software package ( fluent 6.3 ) was used to perform the numerical integration , see @xcite for details .
almost 90 days of fluid behavior was generated , in which @xmath76 flow reversals occurred .
we also note that for the rayleigh number used in this experiment , ra @xmath77 , the thermosyphon has two unstable convective equilibrium solutions corresponding to steady clockwise and steady counter - clockwise flow @xcite .
the low - dimensional model used to make forecasts of the cfd simulation is the ehrhard - mller ( em ) system : @xmath78 where @xmath79 is proportional to the mean fluid velocity , @xmath80 is proportional to the horizontal temperature difference in the loop , and @xmath81 is proportional to the deviation of the vertical temperature difference from the value it takes during conduction .
this system was derived from physical principles to model a natural convection loop @xcite , and for this study the parameters @xmath82 were tuned empirically to best match the flow reversal behavior of the cfd simulated thermosyphon .
the primary difference between em and lorenz systems is @xmath83 , a function that determines the velocity dependence of the heat transfer between the fluid and the wall .
this characteristic of the flow is ignored by the lorenz equations ( i.e. @xmath84 ) .
@xmath83 varies as the third root of the magnitude of the mean fluid velocity for @xmath85 , and as a fourth degree polynomial in @xmath86 for @xmath87 to remain differentiable ; the reader may see @xcite for more detail .
we note that when @xmath84 in the em equations ( [ eq : em ] ) , they are identical to the lorenz system ( [ eq : lorenz ] ) with @xmath88 .
physically , the unitary geometric factor ( i.e. @xmath88 ) in em results from the forced single circular convection cell in a thermosyphon , as opposed to the unconstrained flow producing multiple cells between two plates .
using a background forecast created with the em model , and observations of the cfd mean fluid velocity @xmath89 with gaussian noise added to simulate error , 3dvar data assimilation was performed to generate an analysis , or best guess of the true state of the system in terms of the variables of the forecast model @xcite .
one segment of the analysis is used as the reference truth @xmath12 for training , and the remainder is used for testing .
approximately 3 days of 3dvar analysis , corresponding to 432 time - units in the em forecast model , was used as the training period reference truth . as in the perfect model scenario ,
fourth - order runge - kutta was used with a timestep of @xmath53 time - units .
an analysis window of @xmath90 timesteps , corresponding to about 30 seconds of simulated flow , was used to match the frequency of observation in the data assimilation scheme .
thus , 8640 short forecasts were used to compute the model bias and leith operator for the model , as outlined in sec .
[ emcor ] .
three forecast models were compared by verification against both the 3dvar analysis and direct observation of the mass flow - rate @xmath89 in the cfd simulated thermosyphon : ( 1 ) the uncorrected em model with parameters tuned to best represent observations of the cfd simulated mass flow rate ; ( 2 ) the tuned model with correction applied ; and ( 3 ) an em model whose parameters differ from the tuned model by 10% , _ with _ correction applied .
the purpose of the third test - model is to gauge the relative capabilities of empirical correction and parameter tuning for error reduction and prolonging the usefulness of forecasts .
a set of 1000 trial forecasts were performed , starting from randomly chosen points in the analysis _
after _ the training period .
for each forecast , rmse and ac time - series were computed with respect to the analysis and averaged over all trials .
see the top panel of fig .
[ fig : emcor ] for the resulting average ac plot ( rmse not shown ) .
we also verify the model forecasts against observed scalar mass flow - rate @xmath89 , for which time - series of relative error were averaged over the 1000 trials , pictured in the bottom panel of fig .
[ fig : emcor ] .
two details are important in the computation of the relative error .
first , to compare model output to observations , it is necessary to convert the model state - variables to `` observation - space '' variables . in other words ,
an observation operator determined by data assimilation was used to convert the model state - vector @xmath91 to an observation - space value @xmath92 , which is the predicted mass flow - rate of the system .
second , the error is taken relative to the saturation point , which we define as the average absolute difference between the mass flow - rates of the system at randomly chosen points in time .
thus , an average relative error near 1 means that the forecast model is no better than a random guess . + the results presented in fig .
[ fig : emcor ] indicate that corrected models , tuned or not , produce smaller short - term forecast error on average than the uncorrected , optimally tuned model .
corrected model forecasts are thus typically useful for longer .
though this is an important benefit , average short - term error statistics may conceal considerable qualitative differences between model dynamics and those underlying the true system .
stability of equilibrium solutions , and changes of flow regime characterized by aperiodic switching between otherwise confined regions of state - space are examples of qualitative characteristics for which it may be crucial that the model dynamics match those of the truth . in the next section
we address the effect of empirical correction on the dynamical matching capability of the em forecast model . to measure the accuracy of models with regard to matching the flow reversal behavior of the true system , forecasts were generated with both the corrected and uncorrected em models for 5000 initial states throughout the attractor , from the testing portion of the 90-day 3dvar analysis , and the time of the first predicted flow reversal
was recorded for each one .
we investigate the difference between the predicted times and the actual times ( from the analysis ) of first flow reversal , taken @xmath93 , so that positive values indicate late predictions while negative values indicate early predictions .
see fig .
[ fig : rcdiffbyic ] for plots of the results .
+ + three unforeseen costs of empirical correction for this system , in terms of lost dynamical matching capability , can be summarized as follows : * stabilization of convective equilibrium solutions * elimination of flow reversal behavior for states in a neighborhood of either convective equilibrium * spurious dynamical asymmetry between lobes we address these costs in the sections that follow , examining the first two related phenomena in sec .
[ sec : eqstab ] , and then explaining the spurious asymmetry in sec .
[ sec : brokensym ] .
jacobian analysis of the em equations ( [ eq : em ] ) provides analytical confirmation of the instability of the convective equilibria in the uncorrected model .
specifically , the jacobian evaluated at each equilibrium has one negative real eigenvalue , whose eigenvectors are in the local direction of ( tangent to ) the stable manifold of the equilibrium , and a conjugate pair of complex eigenvalues with positive real part , whose 2-d eigenspace is locally tangent to the unstable manifold of the equilibrium .
in fact , for both convective equilibria the positive real parts of these unstable eigenvalues are quite small , on the order of @xmath94 , indicating weakly repelling instability . in the following
we explain analytically the mechanism by which empirical correction overcomes this weak repulsion , producing a forecast model with _ attracting _ , and thus stable , convective equilibria , see fig .
[ fig : falsestab ] .
empirical correction of the em model effectively alters the right - hand side of the differential equations ( [ eq : em ] ) by first adding a constant related to the bias term @xmath19 , and then adding a term that depends linearly on the model state , i.e. , something related to @xmath95 . letting @xmath96
be the vector - valued em differential equation , and @xmath97 be the corrected equation , we write @xmath98 where @xmath99 can be thought of as the computed bias term and leith operator , respectively , for an infinitesimal timestep @xmath100 . because of the nonlinearity of the system , we can not determine the exact relationship between the infinitesimal correctors @xmath101 and @xmath102 , and the bias term @xmath19 and leith operator @xmath28 , respectively , that we compute using a timestep of @xmath53 and analysis window of @xmath103 timesteps .
however , since we discretize @xmath97 in numerical integration , we _ can _ determine the @xmath101 and @xmath102 that we actually apply .
we effectively approximate the correction terms @xmath104 within the fourth order runge - kutta scheme .
now , armed with an analytical representation of the differential equations @xmath97 , we note the following relationship between the corrected model jacobian @xmath105 , and that of the uncorrected em model @xmath106 : @xmath107 since the constant bias term @xmath101 disappears and @xmath102 operates on a translation of the model state . we evaluate the jacobian of the corrected model at each of the convective equilibria , and
determine its eigenvalues .
indeed , the real part of the complex conjugate eigenvalues , for each convective equilibria , is _ negative _ for the corrected model @xmath97 , and thus the convective equilibria have become stable .
the equilibria attract all states inside neighborhoods around them , which thereby separate state - space near the attractor into regions whose trajectories will either change flow regime at least once , or not at all . see fig .
[ fig : falsestab ] for an example of this effect .
furthermore , any trajectories that land in one of these neighborhoods after only one or two flow reversals , which might occur within the expected 17-minute duration of useful forecasts for the corrected model reported in fig .
[ fig : emcor ] , will then approach steady convection .
this behavior is in qualitative _ opposition _ to that of the true system , _ and _ that of the original uncorrected em model , for which steady convection in a single direction is an unstable equilibrium . the size discrepancy between left and right - lobe regions attracted to the convective equlibria of the corrected model revealed in fig .
[ fig : rcdiffbyic ] demonstrates that empirical correction breaks the symmetry of the em system .
as in the conventional lorenz system ( [ eq : lorenz ] ) , the em system ( [ eq : em ] ) is symmetric under the mapping @xmath108 . again letting @xmath96
be the vector - valued em differential equation , this symmetry implies that @xmath96 commutes with a certain matrix @xmath109 , i.e. @xmath110\ ] ] in fact , empirical correction breaks this symmetry in two ways .
first , recall that after bias correction alone we have changed the em system by adding a constant vector to the right - hand side of the differential equation . letting @xmath111
represent the bias - corrected differential equation , @xmath111 _ does not _ commute with @xmath109 unless there is zero bias in @xmath112 and @xmath113 , i.e. @xmath114\ ] ] where @xmath115 can be any constant , and we recall that @xmath116 can be thought of as the computed bias term for an infinitesimal timestep @xmath100 . note that even if no bias in @xmath112 or @xmath113 existed , the probability of statistically computing a bias term @xmath19 that would preserve the symmetry of the em system is zero . in the unlikely case
that a bias term is computed that preserves symmetry , or such a bias term is forced , state - dependent correction will break it . assuming that @xmath111 does commute with @xmath109 , and letting @xmath97 be the fully corrected differential equation , then @xmath117 in other words
, @xmath97 commutes with @xmath109 if and only if @xmath118 commutes with @xmath109 .
this forces the computed @xmath28 to be of the form @xmath119\ ] ] where the @xmath120 can be any constants .
of course the probability of computing such an @xmath28 statistically is also zero .
therefore , both bias and state - dependent correction break the symmetry of the em model .
undesirable effects of empirical correction include the breaking of system symmetry , and altered stability of equilibrium solutions which results in large regions of state - space for which flow - reversals never occur in corrected model forecasts .
we emphasize that the qualitative behavior of the corrected model is substantively different from that of the uncorrected model , which matches the behavior of the cfd simulated truth .
this is true despite the fact that the corrected model shows improved average error statistics .
however , it is possible to adjust the correction procedure to mitigate this effect by directly incorporating dynamical knowledge of the true system , which is the subject of the next section .
note that in doing so , we sacrifice the general applicability of the technique . ) .
the uncorrected em model trajectory ( blue ) behaves as it should , winding away from the equilibrium at the center of the lobe , whereas the trajectory of the corrected model ( red ) collapses toward the equilibrium solution , indicating the false stability produced by the correction .
the source of the false stability is the overcorrective nudging visible as jumps ( every 30 seconds , the length of the analysis window ) on the red curve .
the correction is attempting to lengthen flow regimes , a consequence of the blue regions near the equilibria in the top of fig .
[ fig : rcdiffbyic ] . ] to encode dynamical knowledge of the system in the empirical correction procedure , the state - space is partitioned into regions based on the qualitative behavior of the system , and then a separate bias term @xmath19 and state - dependent correction operator @xmath28 is computed _ for each region_. for example , in the context of weather forecasting , the state - space of the atmosphere could be divided by stage in the el nio oscillation , or by day and night , or local season for regional models .
[ fig : rcdiffbyic ] suggests two ways to partition the state - space in the present context : ( 1 ) by flow regime direction ( lobe ) ; or ( 2 ) by distance from the nearest convective equilibrium solution . in each case , the state - space is decomposed into two regions , left / right lobe , or near / far from equilibrium , respectively .
in addition to testing the correction procedure using each of these strategies individually , a procedure applying them simultaneously , which results in a partition of the state - space into four regions , is also tested . to generate lobe - dependent bias correction terms and state - dependent correction operators , two regions @xmath121 , corresponding to flow regimes of opposite direction , are defined by @xmath122 noting that their union is the entire state - space .
physically , @xmath123 represents all states undergoing clockwise ( counter - clockwise ) convection .
next , the direct insertion procedure illustrated in fig . [
fig : dirins ] is modified to produce two subsequences of the analysis correction time series , @xmath124 and @xmath125 , that correspond to two subsequences of the analysis time series , @xmath126 and @xmath127 , respectively , where @xmath128 and @xmath129 is defined similarly .
these subsequences are separated into mean and anomalous components , as in eqs .
( [ eq : tsdecomp ] ) , using means over each individual subsequence . finally , the separate correction terms and operators are computed by : @xmath130 for @xmath131 . to apply the lobe - dependent correction , at every timestep of numerical integration the current state
is determined to be in either @xmath132 or @xmath133 , and the appropriate bias correction term and state - dependent operator are applied to advance the model .
+ a procedure analogous to that for lobe - dependent correction is applied here .
two regions @xmath134 , corresponding to near and far from equilibrium , respectively , are defined by @xmath135 where @xmath136 and @xmath137 are the convective equilibria ( estimated from the parameters of uncorrected model , @xcite ) , and the critical value @xmath138 is a parameter for the procedure .
we are effectively approximating the neighborhoods attracted to the convective equilibria by spheres of radius @xmath138 .
for all results shown in the paper , the critical value @xmath139 was used , though error statistics and dynamical matching capability were virtually unaffected by changing this parameter by 25% in either direction .
this range of critical values was tested as estimates to the average radius of the dark red region in the left lobe of fig .
[ fig : rcdiffbyic ] ( bottom ) . continuing with the correction scheme ,
direct insertion is modified as in the lobe - dependent correction , and the region - specific bias terms and state - dependent operators are calculated as in eqs .
( [ eq : lkcorterms ] ) , substituting @xmath140 for @xmath141 .
application of the correction to a forecast model also proceeds in the same fashion . defining the lobe regions @xmath121 as in the lobe - dependent section , and the equilibrium regions @xmath134 as in the previous section , we define the four regions for simultaneous lobe and equilibrium - dependent correction by @xmath142 so that we modify direct insertion to produce four subsequences of analysis increments , each paired with the appropriate subsequence of the analysis time series .
note that the critical value @xmath138 used in defining the regions @xmath134 does not depend on the lobe in this scheme . allowing a different @xmath138 for each lobe is a possible modification that was not tested .
we compute the bias terms and leith operators as in eqs .
( [ eq : lkcorterms ] ) , substituting @xmath143 for @xmath141 , where now @xmath144 . again
, application of the correction proceeds as in the individual dynamically informed schemes , where the current state is determined to be in one of the four defined regions , and the appropriate bias term and leith operator are used to advance the model .
first , we discuss the results of applying the dynamically uninformed , and entirely general correction procedure as described in sec .
[ emcor ] , to couple the three - dimensional em forecast model to the high - dimensional cfd simulated thermosyphon
. then we report the results of modifying the correction procedure to directly incorporate dynamical knowledge of the system , as detailed in sec .
[ sec : idk ] .
the ability of the correction procedure to overcome parameter inaccuracies , as suggested by the results in the perfect model scenario , inspired the comparison of three forecast models .
* tuned , uncorrected * untuned , corrected * tuned and corrected fig . [
fig : emcor ] shows that empirical correction produces forecasts that remain useful longer , and demonstrate reduced error in this mock - operational setting , as well as in the perfect model scenario .
verification against both analysis and observations allows more confidence in the success of the procedure in aligning forecasts made by the low - dimensional model with the cfd simulated thermosyphon .
additionally , the results constitute evidence that the correction procedure is more effective than fine - tuning of parameters for improving error statistics in this more realistic setting , as well as in the perfect model scenario .
this is not to say that parameters should not be tuned , but rather that empirical correction is a cheaper and more effective avenue for reducing forecast uncertainty in the present context
. empirical correction of the em forecast model comes at the cost of decoupling qualitative dynamics from those of the true system , however . as shown in fig .
[ fig : rcdiffbyic ] , a large region is created in which initial states have trajectories under the forecast model that behave completely different from how they would in the true system .
an example of such a trajectory appears in fig .
[ fig : falsestab ] . while the general empirical correction procedure reduces forecast error for many initial states , it produces a forecast model that is entirely useless for others . by introducing dynamical knowledge into the correction procedure ,
the results of which appear in the next section , we greatly reduce the number of such initial states , and also further reduce average forecast error .
+ in an effort to produce forecast models that more accurately reflect the qualitative dynamics of the true system , two types of system - specific dynamical knowledge were directly incorporated into the empirical correction procedure : ( 1 ) flow regime direction of the current system - state , and ( 2 ) distance between the state and the nearest convective equilibrium solution .
these dynamical cues were incorporated individually , and also simultaneously , resulting in three dynamically informed , empirically corrected forecast models , as outlined in sec .
[ sec : idk ] .
[ fig : acmfreall ] shows average ac and mass flow - rate relative error over 5000 trials for the three dynamically informed and corrected models , as compared to the original biased model and the dynamically uninformed , corrected model . as described in the caption to the figure , encoding the current flow regime into the correction procedure results in a forecast model that is useful for almost twice as long as the original biased model , and doubles the improvement that was gained by dynamically uninformed emprical correction . encoding the distance to the nearest equilibrium solution ,
on the other hand , does not greatly prolong usefulness , and in fact reduces it slightly when applied simultaneously with lobe - dependent correction . in fig .
[ fig : rcdiffbyicall ] , we see how the three forecast models resulting from dynamically informed empirical correction compare with the dynamically uninformed , corrected model , with regard to matching the qualitative behavior of the true system .
we see a trend of improvement , characterized by smaller regions of states whose dynamics are different in the models than in the cfd simulated thermosyphon ( dark red regions ) , as we apply first lobe - dependent , then equilibrium - dependent , and finally simultaneously lobe and equilibrium - dependent correction .
results shown in figures [ fig : acmfreall ] and [ fig : rcdiffbyicall ] suggest that encoding flow regime direction into the correction procedure primarily enhances forecast statistics , while encoding distance from equilibrium primarily enhances dynamical matching
. simultaneous inclusion of the two types of dynamical cues results in the best dynamical matching , with only a slight cost in average forecast accuracy . for further evidence of this summary conclusion ,
consider table [ tab:1 ] .
.[tab:1 ] median absolute differences between predicted and actual times of first flow reversal ( row 1 ) , along with percentage of trials for which the first flow reversal was predicted within 1 and 2 minutes , ( rows 2 and 3 , respectively ) , for the uncorrected model ( m ) , dynamically uninformed corrected model ( cm ) , lobe - dependent corrected model ( ld ) , equilibrium - dependent corrected model ( ed ) , and simultaneously lobe and equilibrium - dependent corrected model ( ld - ed ) . [ cols="^,^,^,^,^,^ " , ] in the first row , the median absolute differences between predicted and actual times of first flow reversal over the 5000 trials are listed for the uncorrected model ( m ) , dynamically uninformed corrected model ( cm ) , lobe - dependent corrected model ( ld ) , equilibrium - dependent corrected model ( ed ) and simultaneously lobe and equilibrium - dependent model ( ld - ed ) .
the means are highly skewed for the cm , ld and ed models , due primarily to the large basins of attraction for the left - lobe equilibrium solution , see fig . [
fig : rcdiffbyicall ] top left , right and bottom left , respectively , and thus we show only the medians .
the medians show that both the ld and the ld - ed corrected models predict the first flow reversal more accurately than the uncorrected model .
the bottom two rows show the percentage of the 5000 trials for which the first flow reversal was predicted within 1 and 2 minutes , respectively , for each of the models . again
the ld and ld - ed corrected models show the best performance .
note that the ld - ed model boosts both the 1 and 2-minute success rates by approximately 10% over the uncorrected model .
we note , however , that even the ld - ed model exhibits a spuriously stable convective equilibria in each lobe , fig .
[ fig : rcdiffbyicall ] . incorporating dynamical knowledge of the true system in the correction procedure , at least through the partitioning of state - space as we have done here , is not enough to avoid the stabilizing effect of empirical correction on the equilibria of the em model explained in sec .
[ sec : eqstab ] .
it is plausible that models of other systems with weakly repelling ( attracting ) equilibria might be subject to similar stabilizing ( destabilizing ) effects under empirical correction .
such effects might be mitigated by the strategy employed in the present work , but probably not avoided .
( color online ) difference ( in minutes ) between predicted and actual time of first flow reversal , plotted by initial state , for the ( a ) corrected , ( b ) lobe - dependent corrected , ( c ) equilibrium - dependent , and ( d ) lobe equilibrium - dependent models . as in fig .
[ fig : rcdiffbyic ] , the difference was taken @xmath93 , so that positive values ( towards red ) indicate late predictions while negative values ( towards blue ) indicate early predictions .
inset partial histograms show the number of forecasts ( out of 5000 ) predicting the first flow reversal within 3 minutes of the truth ( green ) and predicting that a flow reversal will never happen ( red ) ; the axes have the same scale for comparison , and the bar colors correspond to the dot colors .
( b ) applying lobe - dependent correction ( as opposed to ( a ) dynamically uninformed correction ) increases the number of initial states for which the first flow reversal is predicted accurately , visible as a larger number of green dots ( bigger green bar ) .
also , although the region of dark red ( initial states attracted to convective equilibrium ) in the left lobe has decreased in size , the one in the right lobe has inflated , again as compared to the top left .
( bottom left ) equilibrium - dependent correction shrinks the left - lobe red region without inflating the one in the right lobe .
however , it maintains the region of initial states for which flow reversal predictions are slightly early ( light blue ) , which was reduced by the lobe - dependent correction .
( bottom right ) applying the simultaneously lobe and equilibrium dependent correction , in which there are four different correction regions , the forecast model demonstrates the smallest region of initial states whose qualitative dynamics are different from the cfd simulated thermosyphon . ]
it is apparent from the results of this work that the empirical correction technique tested here is successful in reducing forecast error in the low - dimensional setting . in both the perfect model scenario and the mock - operational experiment , improved
error statistics and prolonged usefulness of forecasts were demonstrated by the corrected models .
furthermore , the empirical correction procedure was shown to provide greater improvement in average forecast accuracy than fine - tuning of parameters . however
, this positive result comes at the cost of altering some important dynamical characteristics of the model .
details and implications of these results are discussed in the next two sections . in the third section
, we make an observation about the relative impact of state - independent and state - dependent correction , in the present context , to conclude the paper .
in the perfect model scenario , empirical correction of models with the greatest parameter error provided forecasts with smaller short - term average error , and that were useful for longer , than those made by the uncorrected models with least parameter error . similarly , empirically correcting the em model with 10% error in _
every _ parameter ( measured from tuned values ) produced a forecast model that outperforms the tuned , but uncorrected model in predicting the cfd simulated thermosyphon . in each experiment ,
superior performance is evident in terms of anomaly correlation and average forecast error , measured against 3dvar analysis , and is verified against observed mass flow - rate of the cfd simulation in the toy climate experiment .
we do not present these results as evidence that empirical correction should replace parameter tuning , nor even that the former is better than the latter in any well - defined sense .
however , the results do suggest that empirical correction could be a viable complement to the tuning of model parameters . particularly as degrees of freedom become large , for example @xmath145 in some currently operational numerical weather models , the computational cost of parameter tuning is very large in comparison to that of empirical correction , when appropriately modified for such models ( see @xcite ) .
an example of such a modification is to compute the first @xmath146 principal components ( pcs ) of @xmath28 by singular value decomposition ( svd ) , where @xmath147 is determined so that a certain percentage of the state covariance is explained by these first @xmath147 pcs .
a combined tuning / correction approach could reduce the number of model integrations necessary in the parameter tuning process without sacrificing model accuracy .
however , the dynamical ramifications of this strategy must be considered , as we explain in the next section . the reduction in average forecast error provided by empirical correction
belies fundamental dynamical disturbances born out of the correction procedure .
stabilization of equilibrium solutions , which results in large regions of initial states for which the corrected model forecasts behavior in opposition to that of the true system , follows from the dynamical modification imposed by empirical correction .
in addition , the symmetry of the model system is broken by empirical correction .
though these costs can be mitigated somewhat by hard - wiring system - specific dynamical cues into the correction procedure , they can not be eradicated without more fundamental alterations of the technique , e.g. , forcing the bias term @xmath19 and leith operator @xmath28 to preserve system symmetry . in operational practice
, empirical correction is known to introduce imbalances , e.g. violating geostrophy , necessitating some mechanism for smoothing the flow into a physically viable region of state space .
in fact , it may be impossible to avoid all dynamical inaccuracies resulting from empirical correction , and even if theoretically possible , it would likely be impractical to do so in any operational setting . in considering the application of the technique in operational settings , then ,
it must be determined if the effects of misrepresented dynamics can be reduced to a tolerable level on a case - by - case basis . in the ideal situation ,
regions of state - space that would be dynamically misrepresented under empirical correction could be reduced , by minor modifications to the correction procedure , to encompass only unrealistic or unlikely physical states . in any case , the technique presented in this study should not be applied without such considerations . state - independent error correction by itself produces almost no improvement in any of the forecast models in this study ( not shown ) .
this is in contrast to what has been observed in operational weather and climate model studies , where state - independent bias correction typically outperforms state - dependent correction in reduction of forecast errors @xcite .
the inaccuracies of ad - hoc forcings included in such models to compensate for external and/or irresolvable phenomena ( e.g. solar and cloud forcings , respectively ) are likely responsible for a large component of the bias . in light of the lack , or at least minimal nature of such external and sub - gridscale influences in the toy models considered here ,
the ineffectiveness of bias correction is logically consistent with this explanation .
the state - dependent leith operator is entirely responsible for the success of the corrected models in this study . in the perfect model scenario
this makes sense because the difference between the forecast models and the `` truth '' model are inherently multiplicative , i.e. the parameters are coefficients weighting the interaction between state - variable values and thus resulting errors _ must _ depend on state . for the em model of the cfd system
, it seems that errors resulting from the low dimensionality of the forecast model may also be multiplicative in nature .
if this is the case , state - dependent correction may reduce error patterns in operational models that result from reduced dimensionality , e.g. coarse resolution .
the correction will not likely compensate for processes that are irresolvable due to coarse resolution , but rather may reduce the propagation of error resulting from the omission of such phenomena .
this hypothesis is consistent with demonstrated improvement of local behavior in state - dependent corrected atmospheric models with @xmath148 degrees of freedom @xcite . in previous studies of state - dependent correction in models
that are much more realistic than those considered here , resulting error reduction has been minuscule in comparison to what is achieved by bias correction .
however , this is not cause to reject the usefulness of parameterizing state - dependent error . though globally averaged error reduction may not be significant , improvement in the local behavior of models
can have a large impact on forecast uncertainty , particularly in an ensemble strategy where state - dependent correction can increase the spread in previously unsampled state - space directions .
simulations and experiments were performed on the vermont advanced computing center ( vacc ) 1400-processor cluster , an ibm e1350 high performance computing system .
the authors would like to thank ross lieb - lappen for assistance in preparing this manuscript .
this study was supported by nsf grant # dms0940271 , the mathematics and climate research network , and a national aeronautics and space administration ( nasa ) epscor grant . | improving the accuracy of forecast models for physical systems such as the atmosphere is a crucial ongoing effort .
errors in state estimation for these often highly nonlinear systems has been the primary focus of recent research , but as that error has been successfully diminished , the role of model error in forecast uncertainty has duly increased .
the present study is an investigation of a particular empirical correction procedure that is of special interest because it considers the model a `` black box '' , and therefore can be applied widely with little modification .
the procedure involves the comparison of short model forecasts with a reference `` truth '' system during a training period in order to calculate systematic ( 1 ) state - independent model bias and ( 2 ) state - dependent error patterns .
an estimate of the likelihood of the latter error component is computed from the current state at every timestep of model integration .
the effectiveness of this technique is explored in two experiments : ( 1 ) a perfect model scenario , in which models have the same structure and dynamics as the true system , differing only in parameter values ; and ( 2 ) a more realistic scenario , in which models are structurally different ( in dynamics , dimension , and parameterization ) from the target system . in each case
, the results suggest that the correction procedure is more effective for reducing error and prolonging forecast usefulness than parameter tuning . however , the cost of this increase in average forecast accuracy is the creation of substantial qualitative differences between the dynamics of the corrected model and the true system .
a method to mitigate the structural damage caused by empirical correction and further increase forecast accuracy is presented . |
over the past few years there has been a growing interest in the investigation of possible high - energy quantum - gravity - induced deviations from lorentz invariance , that would induce modifications of the energy - momentum dispersion ( on - shell ) relation @xcite .
this hypothesis finds support in preliminary results obtained in some popular approaches to the study of the quantum - gravity problem , most notably approaches based on spacetime noncommutativity " @xcite or inspired by loop quantum gravity " @xcite .
it is expected that the scale governing such deviations from lorentz symmetry is the scale where the same frameworks predict a breakdown of the familiar description of spacetime geometry , the quantum - gravity energy scale " , here denoted by @xmath0 , which should be roughly of the order of the planck scale ( @xmath1 ev ) . while the study of particles with energies close to the planck scale is far beyond our reach in particle - physics laboratories and even in astrophysical observatories , it is possible to look for the minute effects that planck - scale deviations from lorentz symmetry produce for particles with energies much lower than the planck scale . from this perspective of the search of small leading - order corrections
, some astrophysical phenomena could provide meaningful insight , by either producing signals of the modified dispersion , or alternatively , providing constraints on the relevant models .
indeed , astronomical observations have already set valuable limits on these leading - order corrections and as such on some of the alternative scenarios for quantum gravity " .
most insightful are the studies based on the high - energy modifications of particle speeds @xcite and on the modifications to high - energy reaction thresholds ( as relevant for the ultra - high - energy cosmic - rays or very - high - energy @xmath2-rays ) @xcite .
the rapid development of astronomical instrumentation over the last years has improved significantly our ability to test these scenarios .
in particular the recent results of the fermi telescope collaboration @xcite , using a time - of - flight analysis of a short gamma - ray burst grb090510 , achieved a sensitivity to planck - scale effects and set for the first time a limit of @xmath3 ( for the case of effects introduced linearly in the quantum - gravity scale , defined below as @xmath4 ) . the conceptual perspective that guides these studies is consistent with the history of other symmetries in physics , which were once thought to be fundamental but eventually turned out to be violated .
there is no essential reason to believe that lorentz symmetry will be spared from this fate .
if the symmetry is broken at a scale @xmath0 , we should expect leading - order corrections to arise even at much lower energies , and hence it is just natural to explore their possible implications .
however , in view of the importance of lorentz symmetry in the logical consistency of our present formulation of the laws of physics , one should ask which modified dispersion relations can be placed into a viable and consistent theory ?
the analysis presented here intends to contribute in this direction .
we investigate this question from a rather general perspective , aware of the fact that it is quantum - gravity research that provides the key motivation for these studies , but also in principle open to the possibility that the conjectured modifications of the dispersion relation might have different origin .
we consider a generic low - energy " , @xmath5 , leading - order modification of the dispersion relation of the form : is here introduced only for the consistency of @xmath0 in the description of speeds with the analogous parameter most commonly used in the relevant literature . ]
@xmath6 the power of this leading correction , @xmath7 , and the parameter @xmath0 are model - dependent and should be determined experimentally .
the conventional speed of light constant , @xmath8 , remains here the low - energy limit of massless particles speed , and we put hereafter @xmath9 ( with the exception of only a few formulas where we reinstate it for clarity ) .
notice that the fact that we work in leading order in the planck - scale corrections allows us to exchange the ( modulus of ) momentum of a photon with its energy in all planck - scale suppressed terms .
the modification of the dispersion relation produces a difference between energy and momentum of a photon , but this is itself a first - order correction and taking it into account in terms that are already suppressed by the smallness of the planck scale would amount to including subleading terms .
the approach based on ( [ lidform ] ) , which is adopted here , has been considered by many authors as the natural entry point to the phenomenology of lorentz invariance violation @xcite .
we are mainly concerned with the conceptual implications of such modifications of the dispersion relation for the way in which the same phenomenon is observed in different reference frames , and for our exploratory purposes it is sufficient to focus on ( [ lidform ] ) .
our results apply to all cases in which ( [ lidform ] ) is satisfied to leading order .
the findings should apply also to scenarios with birefringence , which could be induced by planck - scale effects ( see , _ e.g. _ , ref .
our results may provide a first level of intuition even for the possibility of planck - scale induced fuzziness " , which is the case of scenarios in which there is no systematic modification of the dispersion relation but modifications roughly of the form ( [ lidform ] ) occur randomly , affecting different particles in different ways depending on the quantum fluctuations of spacetime that they experience ( for details see , _
e.g. _ , refs .
@xcite ) . assuming , as commonly done in the related literature @xcite , that the standard relation @xmath10 holds , the dispersion relation ( [ lidform ] ) leads to the following energy dependence of the speed of photons : @xmath11 this is an important phenomenological prediction of the modified - dispersion scenario .
this effect is the basis for the most generic tests of the lorentz - violation theories .
we will consider the case of subluminal motion , corresponding to the ` - ' sign in eq .
( [ lawforspeed ] ) .
a similar analysis can be performed for the case with superluminal speeds .
we here contemplate an experiment where two relatively boosted observers detect two photons , which are emitted simultaneously at the source but arrive separated by a delay due to their different energies according to ( [ lawforspeed ] ) .
the basic idea of our study is to compare , under different theoretical scenarios , the time delay measured in the different reference frames .
this primarily serves as a gedanken experiment that establishes features which modified - dispersion models must include in order to have a consistent scenario .
we show that if in all reference frames photons have a dispersion relation as in ( [ lidform ] ) and a speed law as in ( [ lawforspeed ] ) ( with fixed frame - independent parameters ) , then assuming the transformations between frames are governed by the standard lorentz laws would lead to a contradiction between the measurements of the different observers .
this is expected , but we use our explicit derivation to deduce other requirements the models must fulfill for a solution .
we consider two classes of modified - dispersion models .
the first class consists of theories where the lorentz symmetry is broken ( lsb ) with the existence of a preferred inertial frame ( sometimes attributed to the cosmic microwave background frame ) .
the small - scale structure and hence high - energy behavior is defined in this reference frame .
therefore , the dispersion relation is allowed to take different forms in other frames . in the following
we will consider only the most common lsb theories in which the lorentz transformations are still applicable when going from one frame to another .
the second class of models describes a scenario where the lorentz symmetry is merely deformed , such that the equivalence of all inertial observers is maintained .
this means that the lorentz symmetry makes way for a more complex symmetry , but the theory is still relativistic ( there is no preferred reference frame ) .
these are called doubly special relativity " ( dsr ) models . within this scenario
the laws of transformation are necessarily modified from the standard lorentz transformations .
this is done in such a way that all observers agree on the physical laws , including the energy - momentum dispersion relation .
not all models of dsr predict energy - dependent velocities of photons , but we deal with the analysis of the common dsr framework where the velocities behave as in ( [ lawforspeed ] ) .
this is to say that we will have two complementary scenarios - lsb with standard lorentz transformations but allowing for departures between the dispersion relations in different frames , and dsr where the theory is frame - independent but the transformations between reference frames differ from the lorentz ones .
we observe that the large effort recently devoted to the phenomenology of modifications to lorentz invariance has focused on analyses performed in a single frame .
we believe that deeper insight on the fate of the lorentz symmetry can be gained by also investigating how the same phenomenon is viewed by two different observers .
we argue that our study sheds light on several conceptual issues , which in turn may well provide guidance even for the ongoing effort utilizing the standard
laboratory frame " tools .
we obtain definitive results for our lsb scenario , and for the dsr case we find that some non - classical features of spacetime are required ( while the phenomenological dsr results are severely conditioned by our restrictive assumption of a classical - spacetime implementation ) . in principle
( and perhaps one day in practice , after a lorentz - violating effect is observed in one frame ) our description of the effects of the lsb scenario for observations of the same phenomenon in two reference frames with a relative boost could be even exploited experimentally . in spite of the limitations associated with the assumption of a classical - spacetime implementation for the dsr case , our findings for
the comparison between the lsb scenario and the dsr scenario provide some encouragement for the possibility that our scheme could also be exploited to discriminate between these alternative spacetime - propagation models .
we turn now to the comparison between different observers which is the key point in our analysis .
let us start by stressing that we shall focus for simplicity on flat ( minkowskian ) spacetime .
while in actual studies of astronomical signals the cosmological curvature can play a significant quantitative role in the actual time delay @xcite , we are here mainly concerned with a conceptual analysis and curvature will not change our qualitative results . in a flat spacetime it follows immediately from ( [ lawforspeed ] ) that two photons with energies @xmath12 and @xmath13 , emitted simultaneously by a distant source , will not reach the observer simultaneously . for a source at distance @xmath14 from the observer
, one should find a difference in arrival times at the detector given by : @xmath15 clearly then , if the time of arrival of a low - energy photon is @xmath16 , the time of arrival of a photon with high energy @xmath17 will be @xmath18 .
we are interested in the comparison between the measurements of the same photons , emitted from a distant source , by two observers , @xmath19 and @xmath20 , in relative motion .
we consider a photon of low energy @xmath12 and a photon of high energy @xmath13 .
in addition to the mentioned assumption of flat spacetime , we assume that the two photons are emitted simultaneously from a single spacetime point . for simplicity
we consider a source , @xmath21 , at rest with respect to the observer @xmath19 .
we denote by @xmath22 the spatial distance from the source as measured by @xmath19 ( @xmath23 ) .
we denote by @xmath24 and @xmath25 the position coordinates of the source and the detector respectively , and we denote by @xmath26 the relative speed between observers @xmath19 and @xmath20 .
the two observers are at the same spacetime point , where the faster photon reaches both of them , so they are synchronized there , _
i.e. _ @xmath27 at @xmath28 .
we also implicitly assume that the synchronization of clocks between the two observers is essentially standard : they would perform clock synchronization using low - energy photons , which are practically unaffected by the modification of the dispersion relation . in fig .
[ figure1 ] one can see the space - time diagrams in the two reference frames , the first at rest with respect to @xmath29 and the second at rest with respect to @xmath30 .
only the case in which @xmath20 is moving away from @xmath21 is depicted , but in the formulas a `` @xmath31 '' in front of @xmath26 will clarify , where needed , that the formulas are valid also in the case in which @xmath20 is moving towards @xmath21 .
each observer measures the delay in the time of arrival of the photon @xmath13 with respect to the photon @xmath12 in its own reference system .
this means that @xmath19 measures @xmath32 , while @xmath20 measures @xmath33 .
the measurement of @xmath20 appears dilated in the reference frame of @xmath19 , where it is @xmath34 .
@xmath35 is the time needed by the photon moving at speed @xmath36 to catch up with the observer @xmath20 , that moves at speed @xmath26 and has an advantage of @xmath37 ( this last relation can be immediately verified by looking at fig .
[ figure2 ] ) . , with @xmath38 highlighted.,scaledwidth=50.0% ] the kinematics in the reference frame @xmath19 yields the relation : @xmath39 subtracting both sides by @xmath40 and recalling that @xmath41 and @xmath34 , we get ( the `` @xmath42 '' sign refers to the case of @xmath20 moving away from @xmath21 ) : @xmath43 this relation is central to our analysis , since it was derived in the reference frame @xmath19 ( making no assumptions regarding the behavior of other reference frames ) . the right - hand - side of ( [ conscond ] ) can be expressed using exclusively the law of energy dependence of the speed of photons in the first rest frame ( of @xmath19 and @xmath21 ) .
the formula can be used to relate the results of time - delay measurements performed by the two observers @xmath19 and @xmath20 , respectively @xmath44 and @xmath45 .
this will impose limits on the free parameters , the speed law in the reference frame @xmath20 and the laws of transformation between observers .
regarding the boost transformation that acting on @xmath46 gives @xmath45 , we can write : @xmath47 = \delta t_{i } \left ( \frac{v(e_2)}{v(e_2 ) \mp \beta } \right ) , \ ] ] where we denoted formally by @xmath48 the action of the ( inverse ) boost that takes from @xmath20 to @xmath19 .
@xmath49 reduces to the lorentz transformation in the case of lorentz invariance , and if the photons speed is independent of energy this condition is trivially satisfied .
@xmath49 is also the usual lorentz transformation in our analysis of lsb .
however , it takes a more general form in the case of dsr . the relation ( [ conscondboost ] ) should be viewed as a requirement of logical consistency : if all the assumptions we have made hold and there exist modified dispersion laws and a transformation action @xmath49 , then inevitably they must be connected through ( [ conscondboost ] ) . if the speed of photons depends on their energy , but we keep the classical transformation laws between observers unmodified , it is easy to see that the law giving the quantitative dependence of the speed of photons on energy must be observer - dependent .
we demonstrate explicitly in this subsection that if @xmath49 is the standard lorentz transformation action and both @xmath44 and @xmath45 obeyed the same frame - independent but energy - dependent speed law , the consistency condition would not be met .
this derivation with lorentz transformations will then be used in sec .
[ lsbsec ] as the basis for the lsb analysis . if the transformation laws between the reference frame of @xmath19 and that of @xmath20 are the special - relativistic ones , since the delay @xmath45 is a proper time in the reference frame of @xmath20 , it is related to @xmath46 by the usual time - dilation formula : @xmath50 = \gamma \delta t_{ii}'\ ] ] ( where as usual @xmath51 ) .
we use this to rewrite eq .
( [ conscond ] ) as follows : @xmath52 next we notice that in this equation @xmath44 is of order @xmath53 ( clearly it must be the case that @xmath54 if @xmath55 ) .
therefore , for an analysis in leading order of @xmath53 , where we also assume @xmath56 , it is possible to take @xmath57 on the right - hand - side of ( [ jocz ] ) : @xmath58 where we denoted by @xmath59 the relativistic doppler factor . equation ( [ deltat2primobroken ] ) gives the time delay @xmath45 that @xmath20 must observe for consistency with the time delay @xmath60 seen by @xmath19 .
we should describe @xmath45 in terms of quantities measured by @xmath20 itself , so that we can investigate the form of the time - delay formula according to @xmath20 on the basis of the form postulated for @xmath19 .
first , using the speed law of ( [ lawforspeed ] ) in the reference frame of @xmath19 , as explained , we explicitate @xmath60 in terms of the energies and the distance measured by @xmath19 : @xmath61 putting this in ( [ deltat2primobroken ] ) we have : @xmath62 now we can use the fact that the relativistic doppler factor also connects the energy @xmath17 of a particle as seen by observer @xmath19 and the corresponding energy @xmath63 seen by observer @xmath20 , @xmath64 , so that : @xmath65 \mathcal{d}^{-n } .\ ] ] since @xmath22 is already multiplied by a factor @xmath53 in ( [ jocw ] ) , we can , consistently to leading - order , ignore all effects of @xmath0 on the relationship between @xmath66 and @xmath22 .
accordingly , we can handle @xmath22 as the space component of a light - cone vector ( that is the vector connecting the point with space - time coordinates @xmath67 to the point with coordinates @xmath68 , see fig .
[ figure1 ] ) , and so it also transforms through the doppler factor : @xmath69 the same conclusion is of course drawn if we observe that according to @xmath30 the distance between @xmath21 and @xmath19 is @xmath70 ( standard lorentz contraction of the proper length @xmath22 ) , but in order to obtain @xmath66 ( the distance traveled by the photon according to @xmath30 - the distance between @xmath30 and @xmath21 at emission ) one must consider also the distance that according to @xmath30 the source @xmath21 travels during the time @xmath71 : @xmath72 in agreement with ( [ jocv ] ) .
substituting ( [ jocu ] ) and ( [ jocv ] ) in ( [ jocw ] ) , we can finally describe the time delay seen by observer @xmath30 in terms of other quantities measured by @xmath30 : @xmath73 comparing ( [ delo ] ) with ( [ sec2final ] ) , we see that in such a scenario the speed law is not frame - independent .
the modified dispersion relation is inconsistent with keeping the two familiar ingredients of frame independence and lorentz transformation laws .
we have to abandon at least one of these appealing notions in order to have a consistent solution .
the following sections discuss two different scenarios that can achieve consistency by allowing for these additional departures from special - relativity : ( i ) breaking the symmetry and having a preferred - frame scenario " ( sec .
[ lsbsec ] ) or ( ii ) modifying the lorentz transformations ( sec .
[ dsrsec ] ) .
we have found that if we use the standard lorentz transformations , we necessarily reach a lsb preferred - frame scenario " .
this means that in principle the theory could be formulated in one chosen frame of reference ( typically one where the laws take an appealingly simple form ) and then all other observers see the lorentz - transformed image " of the fundamental laws written in the preferred frame .
this is indeed the conceptual perspective adopted in the most studied lorentz - invariance - violation models , where the scale @xmath0 in ( [ lawforspeed ] ) is expected to take different values for different observers .
it is valuable to establish how exactly the laws change form in going from one reference frame to another , since this is the only way to combine the results of experiments performed by different observers .
the result of the previous derivation turned out to provide a quantitative characterization of the frame dependence of the lsb scenario .
we shall here continue that derivation , aware that it must be viewed in the context of an observer - dependent framework , and obtain further characterizations of the consistent broken - symmetry scenario . according to the analysis of sec .
[ clasanalsec ] , if we assume that the lorentz transformations hold and in a certain ( preferred ) frame the speed of photons behaves as in ( [ lawforspeed ] ) , the photon speed law must be quantitatively different for relatively boosted observers .
the result ( [ sec2final ] ) , derived on the basis of our logical - consistency requirement , implies that starting from the speed law @xmath74 in a frame @xmath19 , one arrives at the following speed law in a frame @xmath30 : @xmath75 we therefore have an explicit characterization of the expected change of the in - vacuo - dispersion between reference frames .
the laws in @xmath19 and in @xmath20 have the same functional form , but they differ quantitatively with the addition of the doppler factor parameter in @xmath20 .
this parameter is determined in each frame by its relative speed with respect to the privileged frame - for any observer it is fixed and the addition is just a numerical constant .
this allows us to treat the difference between observers as a variation in the scale of symmetry - breaking - marking with @xmath0 the scale as measured by @xmath19 , one then finds that the scale as measured by @xmath20 is : @xmath76 this definitive description of the quantum - gravity - scale transformation is in itself an important characterization .
it is especially noteworthy as several authors had expected ( see , _ e.g. _ , ref .
@xcite and references therein ) that @xmath77 should be treated as a length scale , subject to standard lorentz - fitzgerald contraction . equation ( [ jocccc ] ) shows that this is not the case .
another potentially valuable characterization is found upon recalling the assumption that the energy dependence of the speed of photons , ( [ lawforspeed ] ) , is describable as the group velocity of photons governed by a deformed dispersion relation : @xmath78 we now deduce from eq .
( [ lawforspeedprimebroken ] ) with @xmath79 , that in the reference system of @xmath20 the dispersion relation takes the form : @xmath80 here on the right - hand - side we have highlighted a peculiar consequence of the law of transformation of the scale @xmath0 .
its net result is that , for a given photon ( of given energy @xmath17 according to observer @xmath19 ) , the violation of the special - relativistic dispersion relation , @xmath81 , has exactly the same magnitude for all observers .
this magnitude is fixed by the formula for the lorentz - violating term in the preferred frame , which is based on the photon energy @xmath17 , measured in that frame : @xmath82 any observer can compute the lorentz - violating term of the dispersion relation by finding the special " energy @xmath17 , according to its boost relative to @xmath19 , and using the same formula .
we have evidently found , in the context of our frame - dependent scenario , an invariant term , which has the form of a mass term in special - relativity .
we stress that obviously this invariant @xmath83 can not be viewed as an effective mass for photons in the lsb scenario " , as already implied by the fact that the speed law ( [ lawforspeedprimebroken ] ) is not obtained by treating @xmath84 as a mass in the ultrarelativistic limit : is found by observing that @xmath83 is an invariant under passive lorentz transformations but not an invariant under active lorentz transformations , something that is not at all surprising for cases where lorentz symmetry is broken . under passive boosts that establish how the properties of the same particle are measured by different observers
, we have indeed found that @xmath83 is an invariant .
of course the observation reported in ( [ jocjoc ] ) simply reflects the evident fact that @xmath83 is not invariant under active boosts , which for a single observer map a particle of energy @xmath12 into a different particle of energy @xmath13 .
another example of corrections that are invariant under passive lorentz transformations but not invariant under active lorentz transformations is the case of correction terms of the form @xmath85 , with @xmath86 some external " lorentz - symmetry - breaking vector and @xmath87 the four - momentum of the particles : under a passive boost , of course , both @xmath86 and @xmath87 would transform in a way just such @xmath85 is left invariant , but under an active boost only @xmath87 is transformed and the term @xmath85 changes its value . ]
@xmath88 of course , a generic correction @xmath83 to the dispersion relation will affect inertia in the way prescribed for mass only if , unlike our @xmath83 , it is independent of energy - momentum ( so that it would provide inertia in standard fashion in the group - velocity calculation , obtained by @xmath89 ) .
one similarity between our case and the analogy of a massive - particle dispersion is that in transforming to a reference frame where the energy of the particle is smaller , the effect of the correction to @xmath90 becomes larger .
this result for the case of lsb comes in contrast to the intuition , by which the phenomenological effect of lorentz invariance violation should become more significant when a particle has a larger energy which is closer to the quantum - gravity scale . in conclusion , regarding the much - studied quantum - gravity scenario of a high - energy breakdown of lorentz symmetry with the emergence of a preferred frame
, we have formulated explicitly the only allowed form of the modified dispersion relation , which is consistent with the lorentz transformations . to have a quantitative appreciation of the phenomenological effect of the lorentz - violating speed law in different reference frames , we observe again the result , ( [ deltat2primobroken ] ) , obtained for the time delay that would be measured in a general frame . for observations made from different reference frames with a relative speed @xmath26
, the time delays between the photons will differ by a factor of the corresponding @xmath91 . since the difference between the reference frames is not suppressed by another power of @xmath92 ( but only by the smallness of the relative speed ) , it may in principle be an observed phenomenon .
as an example , we may consider a detector moving at a speed of @xmath93 km / s ( the speed of nasa s helios 2 satellite , which was launched in the 1970s ) : @xmath94 ( when the detector is moving away from the photons ) .
this means that if , for instance , a detector at rest " measures a lorentz - violation - induced time delay of @xmath95 s ( which the fermi telescope can measure if indeed a @xmath4 effect exists ) , the moving detector will measure a time delay greater by @xmath96 ms .
this is small but not unrealistic to expect a possible test in the future .
note that our result is that in the lsb context the time delay in the moving frame is larger even though the photons measured energies are lower .
this is in contrast to the intuitive expectation of a smaller time delay in a frame where the photon energy is smaller ( see sec .
[ closingsec ] for further discussion ) .
a different scenario , which has recently generated interest - the dsr scenario , was essentially introduced @xcite to explore the possibility that the quantum - gravity scale may affect the laws of transformation between observers while preserving the relativistic nature of the theory ( no preferred frame ) . within this dsr scenario the departures from ordinary lorentz invariance
would take the shape of a deformation " ( rather than breakdown ) of symmetries .
this can be thought of in a close analogy with the transition from galilean invariance to lorentz invariance .
galilean invariance ensures the equivalence of inertial observers but makes no room for observer - independent relativistically - non - trivial scales .
attempts to accommodate a maximum - speed law first relied on scenarios that would break galilean symmetry ( ether ) , but eventually found proper formulation in the shape of a deformation of galilean invariance .
this formulation is lorentz invariance , which is still without a preferred frame , but the laws of transformation between inertial observers are deformed ( with respect to the galilean case ) in such a way that a velocity scale , setting the speed - of - light maximum value , is observer - independent .
the dsr framework makes another step with a deformation of lorentz invariance that introduces an additional observer - independent scale .
the aim here is exploring the possibility that the existence of a short - distance ( high - energy ) relativistically - invariant scale might eventually become manifest as we probe high - energy regimes more sensitively , just like the presence of a relativistic velocity scale becomes manifest in accurate observations of high - velocity particles .
we are interested in considering the case of dsr models in which the new observer - independent law is a law of in - vacuo dispersion , based on a relativistic energy scale .
this will offer us a framework for investigation which is complementary to our lsb case . in the lsb scenario
one assumes the laws of transformation to remain unchanged , allowing for an observer - dependent scale of in - vacuo dispersion , while in a dsr scenario with in - vacuo dispersion one must insist on observer independence of the scale of in - vacuo dispersion , allowing the laws of transformation between observers to take a new form .
our analysis is somewhat limited by the fact that these dsr scenarios are still work in progress " . among the residual grey areas for a dsr description of in - vacuo dispersion
particularly significant for our purposes are the ones concerning the description of spacetime . in the following we assume a classical - spacetime geometry , within which we adopt , as done by most authors in the field @xcite , concepts such as spacetime points , translational symmetries , energy , momentum and group velocity in the familiar manner .
we find , below , that in - vacuo dispersion of the form ( [ lidform ] ) is inconsistent with dsr in such a classical spacetime .
this provides new and particularly strong elements in favor of a non - classical spacetime picture .
these findings are consistent with several independent arguments giving indirect evidence of the necessity to adopt a novel , quantum " , picture of spacetime ( see , _ e.g. _ , refs .
in such a quantum spacetime the sharp absolute concept of spacetime would be lost , and in particular one might have that two events that are simultaneous at the same spatial point for one observer are not simultaneous for another observer . however , a satisfactory picture of such a quantum spacetime has not yet been found .
the dsr side of our analysis also intriguingly suggests that quantitatively important differences can be found between the lsb and dsr cases when a single phenomenon is observed from two different reference frames .
while clearly conditioned by our restrictive assumptions , and therefore subject to further scrutiny in future dsr studies , the magnitude of the effect is large enough that it would be surprising if more refined analyses removed it completely .
we shall argue that this might open the way to a future test for the discrimination between alternative in - vacuo dispersion scenarios ( provided that in - vacuo dispersion is detected ) .
we proceed , inspecting the possibility of the dsr scenario with in - vacuo dispersion , using the logical - consistency criterion derived earlier .
we here consider the difference in time delays seen by two observers as a result of in - vacuo dispersion in the dsr case .
of course , as long as we consider a single observer there is no distinction between lsb - case and dsr - case in - vacuo dispersion , and also in the dsr scenario we have for observer @xmath19 : @xmath97 the differences between the lsb and dsr scenarios are instead very significant for what concerns the relationship between this formula for observer @xmath19 and the corresponding formula for an observer @xmath20 boosted with respect to @xmath19 .
we now start from the axiom of our dsr scenario , that @xmath0 is an observer - independent scale and the speed law of photons is identical in all reference frames .
this is in contrast to the property of the lsb case that enabled a consistent scenario . here ,
we seek a consistent description through a new form of transformation laws between observers . )
the transformations should converge to the standard lorentz transformations , and in leading order there would be some correction term with @xmath53 .
] exploiting the observer independence of the speed law ( [ lawforspeed ] ) , we directly arrive at the time delay measured by observer @xmath20 : @xmath98 the measurements in the different reference frames , ( [ deltat1joc ] ) and ( [ deltat2primopreservedshort ] ) , have to be related through the criterion ( [ conscondboost ] ) for consistency .
this clearly requires a departure from the standard lorentz transformations . for the purpose of comparing @xmath44 and @xmath45 we should re - express the formula ( [ deltat2primopreservedshort ] ) in terms of energies and distances measured by observer @xmath19 .
we should , generally , write the relations between @xmath22 and @xmath66 and between @xmath17 and @xmath63 with a correction to the lorentz transformations .
however , @xmath45 already contains a factor @xmath53 , so within our leading - order analysis the zeroth order expressions will suffice , _
i.e. _ we can use @xmath99 + \mathcal{o}\left(e_{qg}^{-n}\right ) , \ ] ] which yields : , with the suggestion of ref .
@xcite that was based on a qualitative argument . ]
@xmath100 utilizing now the consistency criterion , @xmath101 we arrive at a formulation of the connection between the time interval @xmath46 and its transformation to the frame of observer @xmath20 : @xmath102 = \mathcal{d}^{n+1 } ( 1 \mp \beta ) \delta t_{ii } + \mathcal{o}\left(e_{qg}^{-2n}\right ) .\ ] ] this last result poses an immediate challenge for the consistency of this scenario .
the corresponding formula for the lorentz transformation of the proper time @xmath45 is : @xmath103 it is obvious that the expression ( [ dsrtranslawdeltat2 ] ) can not be a leading - order correction of ( [ srref ] ) .
the crucial observation is that the difference between the transformation laws of ( [ dsrtranslawdeltat2 ] ) and ( [ srref ] ) is of zeroth order , _
i.e. _ it is not proportional to any powers of @xmath104 . in our analysis
we tried to meet the consistency requirement by allowing for some deformations of the transformation laws , such that at zeroth order the dsr transformations are the same as the special - relativistic ones .
however , higher - order corrections of the transformations were not relevant in our derivation , which consisted only of quantities which are already suppressed by @xmath53 .
since the only essential assumption that was used to derive this paradoxical result concerned the classicality of spacetime geometry , we conclude that there can not be any consistent classical - spacetime formulation of dsr with modified dispersion
. the comparison of ( [ dsrtranslawdeltat2 ] ) and ( [ srref ] ) allowed us to conclude , consistently with indications that emerged in previous dsr studies @xcite , but in our opinion more forcefully than in any previous related investigation , that a classical - spacetime formulation of dsr models with energy - dependent photon velocity can be excluded .
we must either adopt a dsr framework with no energy - dependent photon dispersion or alternatively consider non - classical features of the dsr spacetime , such that there is no classical - algebraic - description of the transformations of spacetime coordinates between reference frames .
instead the transition to dsr should be one that involves a change to the nature of spacetime .
it is , indeed , not very surprising to find a contradiction between the implications of the dsr framework and the properties of classical spacetime , since the quantum - gravity motivation for dsr research is strongly related to the concept of quantum spacetime " .
a quantum description of spacetime entails dramatic modifications to common features of spacetime geometry that are intuitive in our classical description .
the sharp observer - independent identification of an event ( a spacetime point ) is not available in a quantum spacetime .
instead we have some fuzziness " .
the new geometry will have a fundamentally different structure .
for example , a popular quantum - gravity - inspired description of these spacetimes relies on the formalism of spacetime noncommutativity " , where the coordinates of an event are described in terms of a set of noncommuting observables , governed for example by noncommutativity of the type @xmath105 = i { \cal c}^\alpha_{\mu \nu } \frac{x_\alpha}{e_{qg}}$ ] , with @xmath106 a model - dependent dimensionless matrix . among other approaches that resist the temptation of characterizing spacetime nonclassicality in terms of coordinates ,
one of the most popular is the rainbow metric " picture proposed in @xcite , which essentially characterizes the nonclassicality of spacetime in terms of a metric tensor which is perceived differently by particles of different energy .
this formalization attempts to also find support in some results obtained through direct analysis of dsr scenarios with in - vacuo dispersion .
in particular , it is known @xcite that observer - independent modifications of the dispersion relation for massive particles are inconsistent with a classical - spacetime picture .
this is seen considering the case of two particles of masses @xmath107 and @xmath108 with the same velocity according to some observer @xmath19 .
the observer @xmath19 could see two particles with different masses and different energies moving at the same speed and following the same trajectory ( for @xmath19 particles @xmath109 and @xmath110 are near " at all times ) , but , taking into account the dsr - deformed laws of transformation between observers @xcite , the same two particles would have different velocities according to a second observer @xmath20 ( so according to @xmath20 they could be near " only for a limited amount of time ) .
this would clearly be a manifestation of spacetime nonclassicality , since it amounts to stating that a single spacetime point for @xmath19 is mapped into two ( possibly sizeably distant ) points for @xmath20 .
the results so far obtained in this section provide a novel characterization of the virulence " of the modification of spacetime structure required by dsr scenarios with in - vacuo dispersion .
our findings confirm previous indications in support of the idea that these scenarios require a _ soft deformation _ of energy - momentum space but a rather drastic change in the description of spacetime .
with respect to previous studies that provided support for similar characterizations , our analysis contributes a perspective centered on time measurements , whereas earlier works @xcite had focused mainly on spatial - position measurements .
accordingly , our findings , particularly the form of eq .
( [ dsrtranslawdeltat2 ] ) , could provide motivation for further investigation of the definition of time in dsr scenarios .
already in some of the first dsr studies @xcite it was realized that several subtle issues may affect such a definition of time , but these challenges have not been pursued further .
regarding the peculiar properties of quantum spacetime , of particular relevance for our analysis is the concept of simultaneity .
special relativity removes the abstraction of an absolute time " , but still affords us an objective , observer - independent , concept of simultaneity for events occurring at the same spatial point . in a quantum - spacetime setting
, however , events which are simultaneous and at the same spatial point for a certain observer are not necessarily simultaneous for another observer @xcite .
this is easily seen by noticing that most models of spacetime quantization are inspired by the quantization of ( position - velocity ) phase space in ordinary quantum mechanics , where the heisenberg principle obstructs the possibility of two particles objectively having the same position " in phase space ( they can not have sharp values of both position and velocity , so they can not even be attributed sharply to a point of phase space in a classical sense ) .
indeed , the related concept of a generalized uncertainty principle " , introducing an absolute limitation to combined sharp determinations of both position and time of an event , has been introduced in most approaches to the spacetime quantization @xcite .
the fact that simultaneity at the same spatial point might no longer be observer - independent exposes a possibly important limitation of the applicability of our classical - spacetime analysis to a non - classical spacetime : it was in fact crucial for our line of reasoning that we assumed that for both observer @xmath19 and observer @xmath20 the emission of the two photons by the ( point - like ) source was simultaneous .
while this does not weaken our conclusion that only quantum - spacetime formulations of dsr are admissible , it might weaken the insight on properties of these quantum spacetimes that our analysis also provides .
naively one might think that the needed quantum spacetimes should necessarily match our result ( [ dsrtranslawdeltat2 ] ) , but spacetime classicality was assumed in deriving ( [ dsrtranslawdeltat2 ] ) .
therefore it is too restrictive to insist precisely on ( [ dsrtranslawdeltat2 ] ) , if the analysis we here reported is ever generalized to the case of ( some example of ) quantum spacetime .
actually we feel that this is the direction where our analysis provides the most serious challenge ( and perhaps even an opportunity ) for dsr research : while it is true that the paradox we highlighted is not necessarily applicable to quantum - spacetime scenarios , it is clearly at this point necessary to address this issue within some specific quantum - spacetime picture and in explicit physical fashion .
some explicit candidate quantum - spacetime pictures for dsr with energy - dependent speed of massless particles have been proposed @xcite , but their analysis has not gone much further than the merely abstract / mathematical level , whereas we here showed that a serious challenge ( possibly excluding some or even all of these candidates ) will be met only upon a fully physical formulation of these concepts of spacetime quantization .
we have analyzed in this work two scenarios of departures from lorentz invariance .
demanding the consistency of observations in different reference frames , we arrived at a characterization of the requirements from each scenario . both scenarios require a costly departure from the present formulation of the laws of physics . for lsb
we have the emergence of a privileged rest frame " in which @xmath0 is defined and measured ( and in all other frames we find that @xmath0 is given by a specific non - trivial transformation ) . for dsr
we find no classical solution to the consistency problem , as we have shown that the required transformation laws between boosted frames do not yield the lorentz transformations as a limiting case .
a potential resolution of the paradox is the departure from a classical spacetime and the possible emergence of apparently paradoxical spacetime nonclassicality .
the current literature on quantum - gravity finds advocates for each of these new - physics descriptions .
unfortunately , experiments of this nature , that will tell us which if any of these two possibilities should be adopted , can not be imagined at present . however , we recall the recent history of the quantum - gravity research , when , just fifteen years ago , it was not imagined that experiments would at all be able to test lorentz invariance violation at the planck scale
. still , even though an experimental test is unlikely at the immediate future , it is noteworthy what one finds by comparing the results ( [ deltat2primobroken ] ) and ( [ deltat2primopreserved ] ) , describing the behavior of the in - vacuo - dispersion - induced time delay @xmath45 measured by observer @xmath30 w.r.t .
the delay @xmath60 measured by @xmath19 , as the relative speed between the two observers changes : @xmath111 in this comparison we want to stress the relevant fact that in the two cases @xmath45 behaves in the opposite way : when in the lsb case it shrinks w.r.t .
@xmath60 , in the dsr case it grows , and vice - versa .
if we consider again the conservative quantitative example of sec .
[ lsbsec ] , we notice that even with present - day technology a possible measurement by two detectors would give a difference at a level of @xmath112 between the two scenarios in ( [ phenomcomparison ] ) . while the result of the dsr analysis was derived in a classical - spacetime environment , which was shown to be an unfit description , it is probably safe to assume that the correct result , which may be derived when we have the full quantum - spacetime description , will admit our result as at least a rough approximation .
though we do not claim here that our dsr - case formula should be trusted in detail , it appears likely that our preliminary analysis has uncovered a qualitative feature that could be robust - it might well be a valid assumption that for observables such as @xmath45 the dependence on the boost factor @xmath91 of the dsr case would differ significantly from the corresponding feature of the lsb case . and
while we have concentrated here on a conceptual analysis , it is not unreasonable to imagine that if evidence of in - vacuo dispersion is ever found , a two - telescope experiment could be conducted to investigate further the departure from lorentz invariance , following the strategy of analysis we advocated here . since the contrast between the expected results of the different modified - dispersion scenarios appears to be a zeroth order effect , when the sensitivity of time - of - flight tests reaches scales above the quantum - gravity scale , such an experiment can in principle
be performed .
this would help establish in detail the fate of lorentz symmetry .
u. j. thanks the university of rome la sapienza " for hospitality while some of this work was done .
g. a .- c .
is supported in part by grant rfp2 - 08 - 02 from the foundational questions institute ( fqxi.org ) .
t. p. and u. j. are supported in part by an erc advanced research grant and by the center of excellence in high energy astrophysics of the israel science foundation .
u. j. is also supported by the lev - zion fellowship of the israel council for higher education .
we are grateful to lee smolin for encouraging feedback on a draft of this manuscript and for forwarding to us a draft of the manuscript in @xcite .
g. amelino - camelia , j. ellis , n. e. mavromatos , d. v. nanopoulos and s. sarkar , astro - ph/9712103 , _ nature _ * 393 * , 763 ( 1998 ) .
r. gambini and j. pullin , gr - qc/9809038 , _ phys .
d _ * 59 * , 124021 ( 1999 ) .
j. alfaro , h. a. morales - tcotl and l. f. urrutia , gr - qc/9909079 , _ phys . rev .
lett . _ * 84 * , 2318 ( 2000 ) .
r. aloisio , p. blasi , p. l. ghia and a. f. grillo , astro - ph/0001258 , _ phys . rev .
* 62 * , 053010 ( 2000 ) .
j. magueijo and l. smolin , gr - qc/0305055 , _ class .
quantum grav . _
* 21 * , 1725 ( 2004 ) .
r. aloisio , a. galante , a. grillo , e. luzio and f. mndez , gr - qc/0501079 , _ phys .
b _ * 610 * , 101 ( 2005 ) . g. amelino - camelia , gr - qc/0210063 , _ int . j. mod .
d _ * 11 * , 1643 ( 2002 ) . | we investigate the implications of energy - dependence of the speed of photons , one of the candidate effects of quantum - gravity theories that has been most studied recently , from the perspective of observations in different reference frames .
we examine how a simultaneous burst of photons would be measured by two observers with a relative velocity , establishing some associated conditions for the consistency of theories . for scenarios where the lorentz transformations remain valid
these consistency conditions allow us to characterize the violations of lorentz symmetry through an explicit description of the modification of the quantum - gravity scale in boosted frames with respect to its definition in a preferred frame .
when applied to relativistic scenarios with a deformation of lorentz invariance that preserves the equivalence of inertial observers , we find an insightful characterization of the necessity to adopt in such frameworks non - classical features of spacetime geometry , e.g. events that are at the same spacetime point for one observer can not be considered at the same spacetime point for other observers .
our findings also suggest that , at least in principle ( and perhaps one day even in practice ) , measurements of the dispersion of photons in relatively boosted frames can be particularly valuable for the purpose of testing these scenarios . |
the steady states of self gravitating fluid in three dimensions have been studied by a long list of illustrious mathematical physicists .
( for an extensive list of references see [ 1,2,3 ] ) .
the motivation for this research was due to the interest in the shape , stability and evolution of celestial bodies and systems[12 ] .
we now know however that many celestial objects such as galaxies and our solar system exhibit ( effectively ) `` two dimensional structure '' [ 4,5,6,7 ] . furthermore recent discoveries are leading us to believe that systems similar to our solar system are `` abundant '' in the galaxy and their existence might be due to the collapse of a two dimensional interstellar cloud under gravitation ( this is the so called the `` nebular theory '' ) [ 8,9,10,11,17 ] .
this data leads us to believe that there is a fundamental physical process which we do not understand fully as yet that leads to the formation of planetary systems throughout the galaxy ( and beyond ) .
this background motivates us to investigate in this paper the steady states and time - dependent evolution of a self gravitating and rotating fluid in two dimensions .
this problem has been explored by a large number of investigators using elaborate analytic methods and computer simulations which involve , in general , thermodynamic considerations , magnetohydrodynamics modeling and turbulence.(for a complete list of references see [ 8,9,10,13,16 ] ) . while these are important issues we still need , in our opinion , prototype analytic models that are able to capture the evolution of this process and lead to insights about its possible outcomes . in this paper
we attempt to develop such a model using the basic hydrodynamic equations that govern the time - dependent evolution of an isothermal , incompressible , stratified ( i.e non constant density ) and rotating fluid in two dimensions under gravity [ 1,2,3 ] .
( the justification for the reduction from three to two dimensions has been discussed by many authors .
a lucid treatment is given in ref .
[ 12 ] pp.1 - 12).under these assumptions we show that the number of model equations can be reduced from five to a system of two equations for the steady states and three coupled equations for its evolution .
the models contain some `` parameter functions '' which encode information about the asymptotic mass density distribution of the fluid and its momentum .
the steady state model was investigated by us previously.[18,19,20 ] however in this paper we consider a more general model in which the `` gas cloud '' is rotating also with uniform angular velocity @xmath0 and study the impact of this rotation on the matter distribution in the steady state .
to study the predictions of these models we use both analytical and numerical methods to solve their equations under a variety of conditions .
in particular we consider radial solutions to these equations which represent the evolution of an interstellar cloud with isothermal equation of state [ 10 ] .
it might be argued that the hydrodynamic assumptions we are making in this paper are not realistic from astrophysical point of view .
however our main goal is to capture analytically , as far as possible , the nonlinear and time dependent aspects of the processes under consideration .
accordingly our results might be useful to provide some analytic insights and guidelines for more elaborate work on this topic .
the plan of the paper is as follows : in sec 2 we present the basic hydrodynamic equations and show how one can reduce them to a coupled system of three equations .
sec 3 presents further simplifications of these equations .
the first is for the steady states of the model .
the second is for the time dependent evolution of the gas cloud under the assumption of constant vorticity . in sec 4
we present analytical and numerical radial solutions of these equations .
we end up sec 5 with summary and conclusions .
following the standard convention [ 1,14,15 ] we model the time dependent non - relativistic flow of an incompressible fluid in two dimensions @xmath1 by the hydrodynamic equations of inviscid and incompressible stratified fluid @xmath2 @xmath3 @xmath4 @xmath5 @xmath6 where subscripts indicate differentiation with respect to the indicated variable , @xmath7 is the fluid velocity , @xmath8 is its density , @xmath9 is the pressure , @xmath10 is the gravitational field and g is the gravitational constant .
the terms @xmath11 , @xmath12 represent the components of the apparent centrifugal force due to the rotation of the gas cloud with angular velocity @xmath0 .
we can nondimensionalize these equations by introducing the following scalings @xmath13 where @xmath14 are some characteristic length , velocity and mass density respectively that characterize the problem at hand . substituting these scalings in eqs .
( [ 2.1])-([2.5 ] ) and dropping the tildes these equations remain unchanged ( but the quantities that appear in these equations become nondimensional ) while @xmath15 is replaced by @xmath16 .
( once again we drop the tilde ) . in view of eq .
( [ 2.1 ] ) we can introduce a stream function @xmath17 so that @xmath18 using this stream function we can rewrite eq .
( [ 2.2 ] ) as [ 13,15 ] @xmath19 where for any two ( smooth ) functions @xmath20 @xmath21 using @xmath17 the momentum equations ( [ 2.3]),([2.4 ] ) become @xmath22 @xmath23 to eliminate @xmath9 from these equations we differentiate eq .
( [ 2.10 ] ) and eq .
( [ 2.11 ] ) with respect to @xmath24 respectively and subtract .
this leads to @xmath25 where @xmath26 .
the sum of the second and fourth terms in this equation can be rewritten as @xmath27 to reduce the first and third terms in ( [ 2.12 ] ) we use ( [ 2.8 ] ) .
we obtain @xmath28 combining the results of ( [ 2.13 ] ) and ( [ 2.14 ] ) eq .
( [ 2.12 ] ) becomes @xmath29 thus we have reduced the original five equations ( [ 2.1])-([2.5 ] ) to three equations ( [ 2.5 ] ) , ( [ 2.8 ] ) and ( [ 2.15 ] ) .
although ( [ 2.15 ] ) is rather cumbersome in general , it can simplified further under some restrictions which are presented in the following section .
equation ( [ 2.15 ] ) can be simplified further in two cases .
the first is when we consider only steady states of the flow and the second is when the flow vorticity is constant . when we consider only steady states of the flow ( [ 2.8 ] ) implies that @xmath30 and after some algebra [ 18 ] ( [ 2.15 ] ) reduces to @xmath31 where @xmath32 and @xmath33 is some function of @xmath8 .
thus the equations governing the steady state are ( [ 4.1 ] ) , ( [ 2.5 ] ) and @xmath34 and @xmath33 are `` parameter functions '' which determine the nature of the steady state .
the function @xmath34 is a parameter function which is determined by the momentum ( and angular momentum ) distribution in the fluid . from a practical point of view
the choice of this function determines the structure of the steady state density distribution .
the corresponding flow field can be computed then aposteriori ( that is after solving for @xmath8 ) from the following relations.[18 ] @xmath35 the function @xmath33 that appears in eq .
( [ 4.1 ] ) can be determined from the asymptotic values of @xmath8 and @xmath10 on the boundaries of the domain on which eqs ( [ 2.5]),([4.1 ] ) are solved .
when these asymptotic values are imposed or known one can evaluate the left hand side of eq .
( [ 4.1 ] ) on the domain boundaries and re - express it in terms of @xmath8 only to determine @xmath33 on the boundary of the domain . however , the resulting functional relationship of @xmath36 on @xmath8 must then hold also within the domain itself since @xmath36 does not depend on x , y directly .
for example if we assume that on an infinite domain @xmath37 , @xmath38 and the asymptotic behavior of @xmath8 and @xmath10 is given by @xmath39 then ( asymptotically ) ( [ 4.1 ] ) evaluates to @xmath40 to begin with we consider the case where the vorticity is zero and then generalize to the case where the flow vorticity is constant . when the flow vorticity @xmath41 is zero then @xmath42 eq . (
[ 2.15 ] ) becomes @xmath43 however when the vorticity is zero we can introduce the velocity potential @xmath44 which satisfies @xmath45 , @xmath46 .
replacing @xmath17 by @xmath44 in ( [ 2.16 ] ) we obtain @xmath47+\phi,\rho\ } = 0\ ] ] hence @xmath48+\phi = s(\rho)\ ] ] the equations of the flow in this case are @xmath49 ( which replaces ( [ 2.8 ] ) ) , ( [ 2.18 ] ) and ( [ 2.5 ] ) . to generalize this reduction to the case where @xmath50 ( where @xmath51 is any constant ) we define @xmath52 therefore @xmath53 which implies that there exists a function @xmath44 so that @xmath54 hence @xmath55 using these relations to substitute @xmath44 for @xmath17 in ( [ 2.23 ] ) leads to @xmath56 = -j\{\phi,\rho\}.\end{aligned}\ ] ] therefore @xmath57 hence @xmath58+\phi,\rho\}=0.\ ] ] using ( [ 2.8 ] ) we have @xmath59+\phi,\rho\}\ ] ]
it follows then that @xmath60+\phi = s(\rho).\ ] ] if @xmath61 , @xmath17 can be eliminated from this equation if we differentiate with respect to @xmath62 and use ( [ 2.20 ] ) to obtain @xmath63+\phi\right]_y = s(\rho)_y\ ] ]
when we consider the special case where in polar coordinates @xmath64 and @xmath65 the system ( [ 4.1 ] ) and ( [ 2.5 ] ) with @xmath66 reduces to @xmath67 @xmath68 to solve this system of equations we let @xmath69 , solve ( [ 4.2 ] ) for @xmath10 and substitute the result in ( [ 4.3 ] ) .
this leads to the following fourth order equation for @xmath8 @xmath70 the general solution of this equation is @xmath71 where @xmath72 and @xmath73 are bessel functions of the first and second kind of order @xmath74 and @xmath75 assuming no singularity at the origin we set @xmath76 . to assess the impact of the rotation term on the steady state we solved this system for @xmath77 on a circular disk using the boundary conditions @xmath78 and @xmath79 with @xmath80 , @xmath81 .
the results of these computations for different values of @xmath0 are plotted in fig .
@xmath82 . in this figure
we see that the separation between the density peaks become more pronounced as @xmath0 increases .
this might interpreted as leading to the creation of protoplanets around the central core . a strong dependence on @xmath0 is shown in fig .
@xmath83 which has the same parameters as fig .
@xmath82 except that the boundary conditions on @xmath8 are : @xmath84 and @xmath85 .
this figure illustrate clearly the effect that rotation can have on the pattern of density fluctuations within the cloud .
furthermore in this figure the magnitude of the density fluctuations reverses itself as @xmath0 becomes larger viz .
the higher density peaks are placed at larger values of @xmath86 .
( which is reminiscent of the situation in the solar system )
the system ( [ 2.5]),([2.18 ] ) and ( [ 2.19 ] ) can be simplified further if we use polar coordinates and assume that @xmath87 are functions of @xmath86 and @xmath88 only .
we obtain , @xmath89 where @xmath90 . when we consider a steady state solutions of ( [ 2.28 ] )
then @xmath91 and @xmath92 .
if follows from the first equation in ( [ 2.28 ] ) that either @xmath93 or @xmath94 must be zero . in the first case
@xmath8 is constant and we can let @xmath95 without loss of generality .
when @xmath94 is zero we must have @xmath96 and the second equation in ( [ 2.28 ] ) becomes @xmath97+s''(\rho)(\rho_r)^2- c\rho = 0\end{aligned}\ ] ] where primes denote differentiation with respect to @xmath8 .
we consider these two cases separately .
a. steady state with @xmath95 since @xmath95 the function @xmath33 is a constant and the general solution for @xmath10 is @xmath98 where @xmath99 , @xmath100 are arbitrary constants .
the equation for @xmath44 becomes @xmath101 ( @xmath36 can be absorbed in @xmath100 but we leave it in this form as these two constants have different physical meaning ) .
if we let @xmath102 to avoid the singularity at the origin ( [ 2.31 ] ) yields @xmath103\right\ } + c_3\ ] ] b. steady states with @xmath104 in this case the solution of ( [ 2.29 ] ) depends on the nature of the function @xmath33 . in general
this equation has to be solved numerically .
however we present here analytical solutions of this equation for two special cases .
1 . @xmath69 where @xmath105 is a constant .
the solution to ( [ 2.29 ] ) in this case is @xmath106 it follows then that the nature of the steady state is determined by the ratio @xmath107 .
a sample of the resulting @xmath8 profiles is presented in @xmath108 . to obtain this figure we considered a pinched disk with @xmath109 , @xmath110 and @xmath111 .
the resulting steady state has an increase in the material density towards the circumference of the disk .
similar graphs were obtained numerically for @xmath112 , @xmath113 .
2 . @xmath114 + in this case we have @xmath115 where @xmath116 substituting @xmath117 we obtain @xmath118 which might be interpreted as representing a binary system .
we consider in this section a disk of radius @xmath82 with a steady state @xmath119 and @xmath120 .
letting @xmath121 and using ( [ 2.30 ] ) , ( [ 2.32 ] ) ( with @xmath102 ) this yields the following equations for the steady state @xmath122 @xmath123\ ] ] for a perturbation from this state , viz .
@xmath124 we obtain to first order in @xmath125 the following system of equations : @xmath126 the equation for @xmath127 in ( [ 2.36 ] ) can be solved analytically .
its general solution is @xmath128 where @xmath129 is any smooth function of its variable which has to be adjusted to the initial conditions of the perturbation . the second equation in ( [ 2.36 ] )
is a ( reduced ) poisson equation and its general solution can be expressed by quadratures @xmath130 where @xmath131 , @xmath132 have to be determined by the boundary conditions on @xmath133 .
finally one can obtain also an expression for the solution for @xmath44 in terms of quadratures .
for example if the initial perturbation in @xmath8 is @xmath134 where @xmath51 is a constant then @xmath135 and @xmath136 the evaluation of @xmath133 using the second equation in ( [ 2.36 ] ) and ( [ 2.37b ] ) is straightforward .
it should be obvious how one can generalize this example to other expressions for @xmath137 .
a second approach to the solution of the system ( [ 2.36 ] ) is to assume exponential dependence in time , viz .
@xmath138 this ansatz reduces ( [ 2.36 ] ) to a system of ordinary differential equations @xmath139 as before the equation for @xmath140 can be solved analytically , @xmath141 while the equations for @xmath142 and @xmath143 can be solved by quadratures or numerically .
a numerical approach to the solution for @xmath133 and @xmath144 in ( [ 2.36 ] ) is also possible . as in the previous subsection
we consider again a disk of radius @xmath82 and let @xmath69 .
the general steady state solution for @xmath8 is given by ( [ 2.32 ] ) .
assuming no singularities in @xmath8 ( ie .
no protostar at the origin ) we must set @xmath146 in this equation .
furthermore since @xmath147 it follows that we must have @xmath148 where @xmath149 is the first zero of @xmath72 .
thus @xmath150 ( where we normalized @xmath151 at @xmath152 to be @xmath82 ) . for a perturbation from this steady state in the form given by ( [ 2.35 ] )
we obtain to first order in @xmath125 the following system of equations : @xmath153 where @xmath154 is bessel function of the first kind of order @xmath82 .
the evolution of an initial perturbation @xmath155 with @xmath156 from the steady state is plotted in @xmath157 .
this figure shows that as time progresses there is an accumulation of matter near the center of the disk . at the same time there is an initial separation between the core and the rest of the disk .
we computed also the solution to the system ( [ 2.28 ] ) with an initial matter distribution @xmath158 , @xmath159 and @xmath160 $ ] .
the results of the simulation ( fig .
@xmath161 ) show that as time progresses matter is starting to build up in the vicinity of the center of the disk and around @xmath162 . at the same time there is a decrease in matter density in between these two points .
in previous publications [ 18 - 20 ] we treated only the steady states of two dimensional self gravitating fluid . in this paper
we generalized this model to include disk rotation and assessed the impact of this addition on the distribution of matter in the disk .
we were able also to address the time dependent evolution of this fluid under restrictions on its vorticity .
this enabled us to simplify considerably the equations which govern its evolution . while this is a highly idealized model in the context of astrophysical applications it may still provide some analytical insights for more elaborate models .
t. ramming and g. rein , 2013 spherically symmetric equilibria for self - gravitating kinetic or fluid models in the nonrelativistic and relativistic case , siam journal on mathematical analysis 45 , pp.900 - 914 | in this paper we develop two models for the steady states and evolution of two dimensional isothermal self gravitating and rotating incompressible gas which are based on the hydrodynamic equations for stratified fluid .
the first model is for the steady states of the gas while the second addresses the time evolution of the gas subject to some constraints .
these models reduce the initial five partial differential equations that govern this system to two for the steady state model and to three for the time dependent model .
analytical and numerical solutions of the model equations are used to study the structure of the resulting steady and time dependent states of the fluid with some possible astrophysical applications . |
first discovered in the late 1980s @xcite , extremely red objects ( eros ) are defined by their very red optical / near - infrared colors .
it has been known for some time that the redness of their color constrains these galaxies to be either early - type galaxies , starburst galaxies reddened by dust , or agn ( or a combination of these three classes ) . however , until recently , with observations limited to @xmath3-band ( or shorter wavelengths ) , it has proven extremely challenging to accurately classify the eros by type @xcite , even in combination with high - resolution hst imaging @xcite .
although eros appear to consist of a heterogeneous mix of galaxy classes , the emerging paradigm is that they may well be the high redshift counterparts and progenitors of local massive e and so galaxies . their reliable classification and study , especially at intermediate redshift ( @xmath4 )
can provide crucial constraints on the evolution of massive , starburst , dusty and/or ultraluminous infrared galaxies ( ulirgs ) known to exist at higher redshift e.g. , bzks @xcite , bx / bms @xcite , distant red galaxies ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* drgs ) , submillimeter and ir - luminous lyman break galaxies ( * ? ? ?
* ; * ? ? ?
* illbgs ) . in the restframe near - ir
, old stellar populations show a turndown at wavelengths longer than the restframe @xmath5 ` bump ' , while dusty starburst populations show emission from small hot dust grains .
agn - dominated sources display a power - law spectral energy distribution . in @xcite
we showed how data from _ spitzer _ could begin to help distinguish among different ero populations . in this paper , we extend our ero study to take advantage of the rich panchromatic dataset available from the all - wavelength extended groth strip international survey ( aegis ) . all magnitudes used in this letter
are ab , unless otherwise specified .
the _ spitzer _ irac ( @xmath6 , @xcite ) component of the extended groth strip ( egs ) survey spans an area of @xmath7 arcmin ( huang et al .
2006 , in prep : see also @xcite ) . in conducting this ero study
, we also utilized deep2 spectroscopy
( davis et al .
2006 , in prep ) , and @xmath8 ( ashby et al .
2006 , in prep ) , cfht @xmath9 @xcite , acs @xmath10 and @xmath11 , deep subaru @xmath12 ( * ? ? ?
* ab , @xmath13 ) , @xmath3 ( conselice et al . 2006 , in prep ) , and _
multiband imaging photometer ( mips , @xcite ) @xmath14 imaging .
further details may be found in @xcite .
the egs field contains @xmath15 galaxies detected at @xmath16 ( 23.9 ab , @xmath13 ) .
figure [ fig : r14 ] shows an @xmath0 $ ] versus @xmath2-[8.0]$ ] color - color diagram .
the black points show the @xmath17 @xmath18-selected galaxies with both good quality subaru @xmath12 and @xmath19 photometry . the colored tracks on figure [ fig : r14 ] show the location in color - color space as a function of redshift ( @xmath20 ) for eight non - evolving empirical templates ( four common @xcite ( cww ) templates [ e , sbc , scd , i m ] empirically extended to @xmath21 using iso data ( huang et al . , 2006 , in prep ) , a dusty starburst template [ m82 ] , a dusty starburst / ulirg template [ arp220 ] , and two agn templates [ ngc 1068 and ngc 5506 ] ) .
while we fully expect these templates to become increasingly inaccurate at high redshift , they do serve to provide simple insight onto the likely nature and redshift distribution of galaxies within this color - color diagram .
a distinctive swath of galaxies is clearly apparent curving from the left to the top of figure [ fig : r14 ] ( the plume in the lower left corner is caused by stellar contamination ) .
notice , especially , the excellent agreement between this swath of galaxies observed in color - color space and the cww e ( blue ) track to @xmath22 , when the template noticeably begins to diverge from the `` bluer '' data .
we note that the predictability of the @xmath23 -[3.6]$ ] color - redshift relation for early - type galaxies can be utilized as an effective technique for detecting high redshift clusters of galaxies e.g. , at @xmath24 in the @xmath25 spitzer first look survey field ( @xcite ; muzzin et al , 2006 , in prep ) .
the sparcs collaboration is currently utilizing an even redder @xmath26 -[3.6]$ ] color to detect and study clusters to @xmath27 in the @xmath28 spitzer swire legacy fields @xcite .
we _ define _ an ero to be a galaxy redder in color than @xmath0 = 4.0 $ ] .
this is the same criterion used in @xcite , and is a very similar selection criterion to the traditional vega @xmath29 requirement ( see @xcite for a discussion ) . from figure
[ fig : r14 ] , we might expect that only high redshift ( @xmath30 ) early - type , ( @xmath31 ) dusty starburst galaxies , and agn ( at any redshift ) would satisfy this extremely red criterion .
as we shall demonstrate in the remainder of this letter , this indeed turns out to be the case .
note that late - type cww scd ( cyan ) and irregular ( purple ) galaxies are _ never _ sufficiently red to be classified as an ero _ at any redshift _ , and one would not expect to find any cww late - type sbc ( green ) eros at @xmath32 .
there are several thousand eros in the egs field . here
, we carry out a pilot study of the 87 eros with confirmed deep2 spectroscopic redshifts .
we use spectroscopy in this letter only for redshift determination .
the solid black histogram in the far left panel of figure [ fig : hist ] shows the redshift distribution of the 87 eros in our sample .
the eros occupy a relatively narrow redshift range at @xmath33 .
the dotted black histogram shows the redshift distribution of those galaxies in the deep2 egs field with good quality spectroscopic redshift determinations ( scaled down by a factor of 15 ) .
the fact that none of the eros are located at @xmath34 is not a deep2 selection effect ; in egs , deep2 samples the full redshift range @xmath35 @xcite .
it is caused by the additional ero @xmath0 > 4.0 $ ] selection requirement , which effectively excludes low redshift galaxies ( figure [ fig : r14 ] ) .
the deep2+ero selection function also limits this sample to a fairly bright but narrow range of @xmath36 < 20 $ ] , with the one exception of 12007954 , a 17th magnitude agn @xcite .
note that by selecting only those eros with spectroscopic redshifts , we may bias our sample against inclusion of any low - redshift extremely dusty galaxies ( since optically faint galaxies will not pass the @xmath37 selection requirement of the deep2 survey ) .
it is possible to divide the eros in this sample into three broad classes by means of their irac [ 3.6]- [ 8.0 ] color . in this section ,
we introduce each of these classes , show examples of their seds and discuss the sometime subtle features which are only apparent from such broad - baseline datasets .
we also study their morphologies using acs images . from figure [
fig : r14 ] , early - type galaxies are expected to have the bluest ( @xmath2- [ 8.0 ] \simeq -1.0 $ ] ) colors , dusty starburst galaxies intermediate colors ( @xmath2- [ 8.0 ] \simeq -0.5 $ ] ) , and agn reddest colors ( @xmath2- [ 8.0 ] \simeq + 1.0 $ ] ) .
the right panel of figure [ fig : hist ] corresponds to the dashed rectangle in figure [ fig : r14 ] and shows the subsample of 87 eros with spectroscopic redshifts .
we _ classify _
@xmath2- [ 8.0 ] < -0.75 $ ] galaxies as bulge - dominated early types ( blue circles ) , @xmath38 - [ 8.0 ] > 0.0 $ ] galaxies as dusty starburst@xmath39 .
indeed , as we shall discuss in
[ sec : disc ] most of these dusty starburst eros would actually be classified as lirgs . ]
( red triangles ) and [ @xmath40 - [ 8.0 ] < 0.0 $ ] galaxies as power - law agn types ( orange boxes ) .
as we shall demonstrate , these two simple color cuts do separate the classes rather well .
this is because the redshift distribution of this particular sample is rather narrow . for a sample spanning a broader redshift distribution
, one might imagine utilizing a more complex color selection to isolate the meanderings of the different populations through color - color space in figure [ fig : r14 ] .
we classify 53 of the eros as early types .
they are located at @xmath41 and their redshift distribution is shown by the blue hatching in the middle panel of figure [ fig : hist ] .
we classify 32 eros as being dusty starburst types .
the dusty starburst population ( shown by the red hatching ) has a slightly wider redshift distribution from @xmath42 ( presumably because galaxies in this class can contain arbitrary amounts of dust ) .
we classify two eros as being predominantly power - law types .
these are 12007878 at @xmath43 and 12007954 at @xmath44 , and are shown in solid orange ( see also @xcite ) .
figures [ fig : seds_e ] and [ fig : seds_dsb ] ( plate 1 ) show examples of observed seds and @xmath45 @xmath46-band ( f814w ) postage stamp acs images . for comparison ,
also shown are cww e ( blue ) and m82 ( red ) templates ( [ sec : color - color ] ) as they would appear at the redshift of each ero . at @xmath47 ,
irac s @xmath48 channel measures any emission from small hot dust grains ( indicative of a dusty starburst galaxy ) while the @xmath16 channel measures the stellar peak .
a relative red ( or blue ) @xmath2-[8.0]$ ] color , therefore , can be used to discriminate for ( or against ) dusty starburst galaxies ( and agn ) at this redshift .
figure [ fig : seds_e ] ( plate 1 ) shows six examples of galaxies irac color - classified as early - type eros .
the cww e template approximates the 13048898 , 13019047 , 13004276 , and 12004426 seds very well .
the cww e template appears to underestimate the seds of 13019309 and 12008091 in the uv . in all cases
the acs images show clear evidence of bulge - dominated morphologies , supporting our color classification as early types .
a total of 35 of the 53 early - type eros fall within the acs footprint .
a close inspection of all of the available acs images reveals them to be predominantly bulge - dominated ( e , s0 or sa - type spirals ) . a small number ( @xmath49 )
appear to be undergoing mergers .
we conclude that choosing a blue [ 3.6]-[8.0 ] color successfully selects for @xmath50 eros with old stellar populations and against those with dusty starburst or power - law features .
figure [ fig : seds_dsb ] ( plate 1 ) shows five examples of galaxies irac color - classified as dusty starburst eros , and one power - law ero ( 12007878 ) .
the m82 template approximates the seds of sources 13042940 , 12100899 , and 12007831 very well , and that of 13004291 reasonably well .
in addition to being more mid - ir luminous than an early - type ero of similar @xmath18 magnitude , the seds of these four sources are more luminous in the optical .
their acs images show clear evidence of disturbed , peculiar , interacting or merging galaxies .
neither template approximates well the sed of 12008048 .
close inspection of the acs image reveals it to be a face - on spiral galaxy .
since a normal spiral at @xmath51 would not meet the @xmath0 $ ] redness criterion , we infer that 12008048 must be an especially dusty spiral . a total of 17 of the 32 eros we classify as dusty starbursts fall within the acs footprint .
close inspection of all of their images reveals many of them ( @xmath52 ) to be disturbed or interacting galaxies , and the remainder ( @xmath53 ) to be late - type spirals .
12007878 shows a monotonically increasing power - law sed .
its host galaxy is clearly bulge - dominated and it can be unequivocally classified as an agn - dominated source .
turning to those eros which have @xmath54 detections , there are five ( @xmath55 < 19.1 $ ] ) amongst the 53 early - type and 29 ( @xmath56 < 19.1 $ ] ) amongst the 32 dusty starburst eros .
both power - law eros have a @xmath57 detection ( 14.5 and 16.5 ) . in four out of the five cases of early - type eros with a @xmath54 detection
, an acs image is available . in all four cases ,
a merging galaxy with several close neighbors is apparent .
the mips point - spread function is @xmath58 so it is possible that the signal in each case might be associated with another galaxy entirely .
we use ( * ? ? ?
* fig . 7 ) to translate our @xmath57 flux limit ( 80ujy , 19.1 ab , @xmath13 ) into a lower limit on the total detectable ir luminosity of an ero in our redshift range . depending on the model used , at @xmath59 , we are capable of detecting sources more luminous than @xmath60 i.e. , some dusty starbursts , and all lirgs and ulirgs . at @xmath61 , we are capable of detecting sources more luminous than @xmath62 i.e. , lirgs and ulirgs . based on their actual @xmath57 magnitudes and redshifts , most of our `` starburst '' sources have intrinsic luminosities between @xmath63 and @xmath64 @xmath65 , i.e. , lirg class . both of the `` agn '' sources have intrinsic luminosities of @xmath66 , putting them in the ulirg class .
although , in this study , we utilized optical / irac colors to classify our ero sample , irac-[24 ] color selection may be a more effective discriminator at higher redshift @xcite .
putting irac ero colors in the context of perhaps more familiar restframe optical colors , figure [ fig : ub14 ] shows _ restframe _ @xmath67 color versus _ observed _
@xmath2-[8.0]$ ] color for all irac - detected galaxies with good [ 8.0 ] and good cfht bri photometry ( see willmer et al .
2006 for details of the conversion to restframe @xmath67 ) .
the symbols are as in figure [ fig : hist ] .
the filled circles in figure [ fig : ub14 ] denote the 36 eros which are also detected at @xmath57 ( see also figure [ fig : hist ] ) .
the eros we classify as early types generally tend to lie on the @xmath67 defined `` red - sequence '' and the eros we classify as dusty starburst tend to lie in the `` green valley '' and at the top of the `` blue cloud '' @xcite .
in this paper , we utilized the aegis dataset to explore the nature of 87 eros with confirmed spectroscopic redshifts .
irac imaging facilitated dividing this sample into three distinct classes using a simple [ 3.6 ] -[8.0 ] color selection technique .
we presented seds and high - resolution acs images supporting our ( early - type , dusty starburst or power - law ) classification .
we showed that the three classes of ero and their redshift distribution were broadly consistent with what one would expect from a simple color - color diagram ( figure [ fig : r14 ] ) .
we found 53 early - types , 32 dusty starbursts / lirgs , and two obvious agn .
both of these agn would be classified as ulirgs .
the selection of this particular sample of eros was subject to the spectroscopic biases discussed in
[ sec : zdist ] .
we next plan to extend this study to the several thousand irac - selected eros in the aegis _ without _ spectroscopic redshifts .
this work is based in part on observations made with the spitzer space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was provided by nasa through an award issued by jpl / caltech .
alc is supported by nasa through hubble fellowship grant hf-01182.01-a .
we thank the referee , matt malkan , for constructive comments .
gw thanks mark lacy , adam muzzin , jason surace , ian smail and mike hudson for useful discussions , and uc berkeley and uc santa cruz for their hospitality . | we study 87 extremely red objects ( eros ) , selected both to have color redder than @xmath0=4.0 $ ] , and to have confirmed spectroscopic redshifts . together
, these two constraints result in this sample populating a fairly narrow redshift range at @xmath1 .
the key new ingredient included here is deep _ spitzer _ space telescope infrared array camera ( irac ) data .
based on @xmath2-[8.0]$ ] color , we demonstrate that it is possible to classify eros into early - type , dusty starburst , or power - law ( agn ) types .
we present ultraviolet to mid - infrared spectral energy distributions ( seds ) and advanced camera for surveys ( acs ) images , both of which support our simple irac color classification . |
in the past few years , remarkable progress has been achieved in the understanding of the infrared ( ir ) singularities of massless scattering amplitudes in non - abelian gauge theories , which are characterized by an intricate interplay of soft and collinear dynamics .
while factorization proofs guarantee the absence of ir divergences in inclusive observables @xcite , an all - order formula for the ir singularities of qcd amplitudes has been lacking for a long time .
an important step toward this goal was made by catani @xcite , who correctly predicted the singularities of two - loop scattering amplitudes apart from the @xmath0 pole term .
an interesting alternative approach to the problem of ir singularities was developed in @xcite , where the authors exploited the factorization properties of scattering amplitudes along with ir evolution equations to recover catani s result at two - loop order and relate the coefficient of the @xmath0 pole term to a soft anomalous - dimension matrix .
this matrix was later calculated at two - loop order in @xcite .
its color structure was found to be of the same form as at one - loop order and it was shown that the more complicated color structure of the @xmath0-term in catani s formalism @xcite was an artifact of his subtraction scheme . in recent work @xcite , we have shown that the ir singularities of on - shell amplitudes in massless qcd are in one - to - one correspondence to the ultra - violet ( uv ) poles of operator matrix elements in soft - collinear effective theory ( scet ) @xcite .
they can be subtracted by means of a multiplicative renormalization factor , whose structure is constrained by the renormalization group .
we have argued that the simplicity of the corresponding anomalous - dimension matrix holds not only at one- and two - loop order , but is in fact an exact result of perturbation theory .
a first test of this prediction at three - loop order was performed in @xcite .
detailed theoretical arguments supporting our conjecture were presented in @xcite , where we used constraints derived from soft - collinear factorization , the non - abelian exponentiation theorem @xcite , and the behavior of scattering amplitudes in two - parton collinear limits @xcite to show that the anomalous - dimension matrix retains its simple form at least to three - loop order , with the possible exception of a single color structure multiplying a function of conformal ratios depending on the momenta of four external partons , which vanishes in all collinear limits .
some of these arguments were developed independently in @xcite .
we also showed that higher casimir invariants , which in particular would lead to a violation of casimir scaling of the cusp anomalous dimension @xcite , do not appear at four - loop order .
it is interesting and relevant for many physical applications to consider generalizations of these results valid in the case of massive partons .
the ir singularities of one - loop scattering amplitudes containing massive partons were obtained some time ago in @xcite .
we will reproduce their formula as a special case of our analysis . in the limit
where the parton masses are small compared with the typical momentum transfer among the partons , they serve as regulators for collinear singularities . in @xcite
factorization theorems for this limit were proposed , which allow one to derive massive amplitudes from massless ones .
this technique has been used to derive the massive @xmath1 scattering amplitude @xcite and the virtual corrections to heavy - quark production in the limit where all kinematic invariants are larger than the heavy - quark masses @xcite .
first steps toward solving the problem of finding the ir divergences of generic two - loop scattering processes with massive partons , without restricting oneself to the limit of small masses , have recently been taken by mitov et al .
interestingly , these authors find that the simplicity of the anomalous - dimension matrix governing the ir poles does not persist at two - loop order in the massive case .
specifically , they point out that there exist non - vanishing singularities in feynman graphs connecting three different partons .
they evaluate the corresponding contributions numerically and point out that they vanish for some special kinematic configurations . in this paper
we extend our analysis in @xcite to the general case of gauge - theory amplitudes with arbitrary numbers of massive and massless partons .
we show that the structure of the terms found in @xcite follows from simple symmetry arguments , such as soft - collinear factorization in scet and non - abelian exponentiation .
the fact that in the presence of massive partons the low - energy effective theory knows about the 4-velocities of heavy particles in addition to the light - like directions of massless partons gives rise to additional color and kinematical structures , which are absent in the case of massless partons . as a result , the structure of ir poles becomes increasingly more complicated in higher orders of perturbation theory .
we present , for the first time , a general formula for the ir singularities of dimensionally regularized two - loop scattering amplitudes with arbitrary numbers of massive and massless partons and arbitrary values of the parton masses .
it generalizes the one - loop result of @xcite . for amplitudes with @xmath2 partons ,
our result contains two new functions with certain symmetry properties , for which at present no analytical expressions are available .
we begin by considering the case where the parton masses are of the same magnitude as the typical momentum transfer between the partons . in this case
the appropriate low - energy effective theory is a combination of scet and heavy - quark effective theory ( hqet ) @xcite , which is applicable since the relative velocities of the heavy partons are of @xmath3 in this case . with the general result at hand
, we then explore the limit where the parton masses are taken to be much smaller than the hard momentum transfers between the partons .
we conclude that if the two new functions did not vanish in this limit , then the qcd factorization formula of @xcite would need to be modified .
we denote by @xmath4 , with @xmath5 and @xmath6 , a uv - renormalized , on - shell @xmath7-parton scattering amplitude with ir singularities regularized in @xmath8 dimensions .
this quantity is a function of the lorentz invariants @xmath9 and @xmath10 , where the sign factor @xmath11 if the momenta @xmath12 and @xmath13 are both incoming or outgoing , and @xmath14 otherwise .
we assume that all of these invariants are of the same order and refer to them as hard scales . for massive partons ( @xmath15 ) , we define 4-velocities @xmath16 , whose components are of @xmath3 .
we have @xmath17 and define the abbreviations @xmath18 .
we use the color - space formalism of @xcite , in which @xmath7-particle amplitudes are treated as @xmath7-dimensional vectors in color space .
@xmath19 is the color generator associated with the @xmath20-th parton and acts as a matrix on its color index .
the product @xmath21 is summed over @xmath22 .
generators associated with different particles trivially commute , @xmath23 for @xmath24 , while @xmath25 is given in terms of the quadratic casimir operator of the corresponding color representation , i.e. , @xmath26 for quarks and @xmath27 for gluons .
we have shown in @xcite that the ir poles of such amplitudes can be removed by a multiplicative renormalization factor @xmath28 , which acts as a matrix on the color indices of the partons .
this quantity obeys the renormalization - group equation @xmath29 where @xmath30 is the anomalous - dimension matrix of effective - theory operators built out of collinear scet fields for the massless partons and soft hqet fields for the massive ones .
the formal solution of this equation is @xmath31 , \ ] ] where the path - ordering symbol @xmath32 means that matrices are ordered from left to right according to decreasing values of @xmath33 .
the @xmath34-factor appearing in the renormalization of effective - theory operators describes the ir behavior of on - shell amplitudes , because these amplitudes are closely related to the bare wilson coefficients of the corresponding operators .
this connection is discussed in detail in @xcite . compared to the massless case
studied there , we encounter one complication : since virtual corrections due to heavy quarks are integrated out in the effective theory , the strong coupling constant entering the @xmath34-factor in the low - energy theory is defined in a theory with massless quark flavors only , while the massive amplitudes in qcd also receive contributions from heavy - quark loops .
the @xmath34-factor we obtain from the effective theory describes the ir singularities of massive qcd amplitudes after the coupling constant is matched onto the effective theory with massless flavors .
the corresponding matching relation will be given below .
the interactions between collinear and soft fields can be decoupled by means of a field redefinition @xcite , after which soft interactions manifest themselves as interactions between a set of light - like and time - like soft wilson lines representing the massless and massive particles , respectively . generalizing the discussion of @xcite , the relevant soft operator in the present case is @xmath35 where partons @xmath36 are massless , and the remaining @xmath37 partons are massive . from now on , we label the massive partons by capital indices @xmath38 and the massless ones by lower - case indices @xmath39 .
the anomalous - dimension matrix of the effective - theory operators can be written as a sum over soft and collinear contributions @xcite , @xmath40 where @xmath41 is the soft anomalous - dimension matrix governing the uv poles of the wilson - line operator in ( [ slines ] ) .
the collinear contributions @xmath42 only arise for massless partons and are diagonal in color space .
note that color conservation implies the relation @xmath43 when acting on color - singlet states such as color - conserving scattering amplitudes . in intermediate steps in the calculation of the anomalous - dimension matrix one needs to regularize ir divergences in the effective theory , for instance by taking the massless partons slightly off their mass shell , @xmath44 .
following @xcite , we introduce the notation @xmath45 $ ] for the associated collinear logarithms , which need to cancel in the final result ( [ decomp ] ) .
the soft anomalous - dimension matrix @xmath41 is , in the most general case , a function of the cusp angles @xmath46 , @xmath47 , and @xmath48 formed by the wilson lines belonging to different pairs of massless or massive partons . with the ir regulator
as specified above , the relations expressing these cusp angles in terms of the hard momentum transfers and particle masses read and @xmath49 , see @xcite . ]
@xmath50 the anomalous dimensions appearing on the right - hand side of ( [ decomp ] ) are functions of the cusp angles @xmath46 and the collinear logarithms @xmath51 , while that on the left - hand side only depends on the hard scales @xmath52 and @xmath53 .
we begin by considering the one- and two - particle terms in the anomalous - dimension matrix ( [ decomp ] ) . they can be written as @xcite @xmath54 and @xmath55 where the notation @xmath56 refers to unordered tuples of distinct parton indices .
the various coefficients are functions of the renormalized coupling @xmath57 and , in the case of @xmath58 , of a cusp angle @xmath59 .
the fact that only a linear dependence on the cusp angles is allowed in cases where at least one massless parton is involved has been explained in @xcite .
it is instructive to see how the dependence on the ir regulators disappears in ( [ decomp ] ) , when we combine the expressions in ( [ gamc ] ) and ( [ gams ] ) and express the cusp angles in terms of hard momentum transfers and masses as well as collinear logarithms .
we note that ( after the cusp angles have been eliminated ) @xmath60 hence , the sum of all contributions is indeed independent of the ir regulators .
note that this requirement fixes the relative strength of the terms proportional to @xmath61 and @xmath62 in ( [ gams ] ) . from ( [ decomp ] )
we then obtain @xmath63 where @xmath64 .
the anomalous - dimension coefficients @xmath65 and @xmath66 ( for @xmath67 ) have been determined to three - loop order in @xcite by considering the case of the massless quark and gluon form factors .
for example , the one- and two - loop coefficients in the perturbative series @xmath68 are @xmath69 in qcd only quarks can be massive , and the first two coefficients in the expansion of @xmath70 can be extracted by matching our result with the known form of the anomalous dimension of heavy - light currents @xmath71 in scet @xcite .
we obtain @xmath72 where we assume that the heavy - quark with velocity @xmath73 is incoming and the light quark with momentum @xmath74 is outgoing .
the sum @xmath75 was first obtained at two - loop order in @xcite .
using this result leads to the one- and two - loop coefficients @xmath76 since this anomalous dimension is connected with the soft wilson - line operator in ( [ slines ] ) , its color structures are constrained by the non - abelian exponentiation theorem @xcite .
this explains the absence of a @xmath77 term , and it implies that up to @xmath78 the corresponding anomalous dimension for a massive color - octet particle is given by @xmath79 times the coefficients shown above .
it remains to determine the velocity - dependent function @xmath58 in ( [ resu1 ] ) .
this can be accomplished by applying our general result to the special case of just two heavy quarks , where it should reproduce the velocity - dependent anomalous dimension of heavy - quark currents in hqet @xcite .
it follows that @xmath80 with @xmath81 , where @xmath73 is the incoming and @xmath82 the outgoing quark s velocity .
this anomalous dimension was calculated at two - loop order in @xcite , and it has recently been recomputed and expressed in compact form in @xcite . using the result ( [ gammaq ] ) for @xmath70 , we obtain from the latter paper @xmath83 \\ & \quad\mbox{}+ \coth\beta \left [ \mbox{li}_2(e^{-2\beta } ) - 2\beta\,\ln(1-e^{-2\beta } ) - \frac{\pi^2}{6}\,(1+\beta ) - \beta^2 - \frac{\beta^3}{3 } \right ] + \frac{\pi^2}{6 } + \zeta_3 + \beta^2 \bigg\ } + \dots \ , . \end{split}\ ] ] for small cusp angle the result can be expanded in even powers of @xmath59 , @xmath84 .
\nonumber\end{aligned}\ ] ] note that the leading terms in the expansion of @xmath58 around small cusp angle are equal to @xmath85 , so that the full anomalous dimension @xmath86 vanishes in the limit @xmath87 . in the limit of large cusp angle
one finds @xmath88 where the dots represent terms that vanish for @xmath89 .
note that no constant terms remain in this limit .
the @xmath34-factor associated with a renormalization - group equation such as ( [ rge ] ) , in which the anomalous dimension is linear in @xmath90 , was derived in @xcite . to two - loop order it reads @xmath91 + \dots
\ , , \nonumber\end{aligned}\ ] ] where @xmath92 and we have expanded @xmath93 . exponentiating the result ( [ result ] ) yields the two - loop expression for @xmath34 , which encodes the ir singularities of the massive qcd scattering amplitudes .
more precisely , we have @xmath94 for @xmath95 . the quantity @xmath96 denotes the strong coupling constant in the effective theory , which is obtained after integrating out the heavy quark flavors .
it is obtained from the coupling constant @xmath97 of full qcd via the decoupling relation @xmath98 .
to first order in @xmath96 , the matching factor appropriate for @xmath99 heavy - quark flavors reads @xcite @xmath100 .\ ] ] we have checked that at one - loop order our results ( [ resu1 ] ) and ( [ result ] ) reproduce the ir pole terms obtained in @xcite .
we have also confirmed that they correctly describe the ir singularities of the two - loop massive quark form factor calculated in @xcite .
our result ( [ resu1 ] ) has been derived under the assumption of hard parton masses , in which case the appropriate low - energy effective theory is well known .
however , being an exact result in perturbative qcd , this formula can also be applied in cases where some or all of the parton masses are much smaller than the momentum transfers between the partons , @xmath101 .
this limit is relevant to many processes , such as bhabha scattering or heavy - quark production at the lhc .
it is the limit in which all cusp angles , also those involving massive partons , become large , and @xmath102 up to power - suppressed terms . in this case
our general result ( [ resu1 ] ) implies that @xmath103 , \end{split}\ ] ] where @xmath104 is the anomalous - dimension matrix in the massless case , whose conjectured all - order form is given by the terms shown in the second line in ( [ resu1 ] ) @xcite . in the equation above
, @xmath105 is the massless single - parton anomalous dimension belonging to parton @xmath106 . in qcd
only quarks can be massive , and this result can be rewritten as a sum over heavy - quark anomalous dimensions @xmath107 where the one- and two - loop coefficients of the constant terms are @xmath108 the factor @xmath109 associated with ( [ gammaqeq ] ) , which is obtained after substituting the anomalous dimension @xmath110 into the general relation ( [ result ] ) , is compatible with the results of @xcite .
specifically , we find that the product @xmath111 is finite , where the quantity @xmath112 was defined in @xcite as the ratio of the massive to the massless quark form factor in the limit where the quark mass tends to zero , and without including heavy fermion loops .
note that our derivation assumed that the massive partons are heavy enough to be integrated out in the low - energy theory using ( [ eq : decouple ] ) .
if this is not the case , then the treatment of the heavy - flavor contribution is more complicated @xcite .
it was observed in @xcite that in the case with massive partons the anomalous - dimension matrix ( [ decomp ] ) has a more complicated structure than in the massless case , and that at two - loop order non - abelian diagrams connecting three partons give rise to non - vanishing contributions .
the additional terms were found to vanish if two of the three partons are massless , or if any pair of the three kinematic invariants formed out of the parton momenta are equal . we will now show that these observations have a simple explanation .
[ r]@xmath113 graphical representation of the two three - particle terms in the anomalous - dimension matrix ( [ resu2 ] ) .
double lines represent massive partons , single lines show massless ones.,title="fig : " ] graphical representation of the two three - particle terms in the anomalous - dimension matrix ( [ resu2 ] ) .
double lines represent massive partons , single lines show massless ones.,title="fig : " ] adapting the diagrammatic analysis of our paper @xcite to the case with non - zero parton masses , we find that additional structures arise from two - loop order on , the reason being that the 4-velocities of the massive partons are known to both the full and the effective theories . in hqet the velocities appear as labels on the effective heavy - quark fields @xcite . in the full theory , they are simply given by @xmath16 .
while for massless partons the rewriting from hard to soft variables always introduces collinear logarithms , this is not true for massive partons , as shown in ( [ betasmrel ] ) . at two - loop order ,
the non - abelian exponentiation theorem then allows additional structures involving three partons .
they are absent in the massless case , because it is impossible to form a totally anti - symmetric function of three cusp angles @xmath46 , @xmath114 , @xmath115 that is independent of collinear logarithms upon the substitution shown in the first line in ( [ betasmrel ] ) @xcite .
this would violate soft - collinear factorization .
however , with massive partons this argument no longer applies .
in fact , in principle the soft anomalous - dimension matrix can contain the structures @xmath116 the function @xmath117 must be totally anti - symmetric in its arguments , while @xmath118 ( @xmath119 ) must be anti - symmetric in the last ( first ) two arguments .
soft - collinear factorization enforces that after elimination of the cusp angles using ( [ betasmrel ] ) the result ( [ f3 ] ) must be independent of collinear logarithms .
this in turn requires that @xmath120 where @xmath121 must be an odd function of @xmath122 .
note that for @xmath119 to be independent of collinear logarithms it should be a function of the combination @xmath123 , but this is symmetric in @xmath124 and so vanishes when contracted with the anti - symmetric color structure .
hence only the two possibilities illustrated in figure [ fig : graphs ] remain , and we are led to the following additional structures in the complete anomalous - dimension matrix in ( [ decomp ] ) : @xmath125 this is the most general form possible at two - loop order , and hence the sum of ( [ resu1 ] ) and ( [ resu2 ] ) gives the complete answer for the anomalous - dimension matrix at @xmath126 .
we note that the color factor in ( [ resu2 ] ) is non - zero only if there are at least four partons involved in the scattering process . for three partons , charge conservation ,
i.e. the fact that @xmath127 , implies that @xmath128 since the color matrices entering the two - particle and three - particle terms do not commute , the path ordering in ( [ zsolu ] ) becomes important in the massive case .
however , since the three - parton color structures first enter at two - loop order , this complication arises only at @xmath78 .
it is interesting to observe that the two structures in ( [ resu2 ] ) are consistent with the constraints following from the behavior of scattering amplitudes in the limit where two ( or more ) massless partons become collinear . to see this ,
consider the case where the first two light partons become collinear , such that @xmath129 and @xmath130 with @xmath131 .
as shown in @xcite , the anomalous dimension of the matrix of splitting amplitudes @xmath132 is given by the difference of @xmath133 and @xmath134 , and it must be independent of the momenta and colors of the remaining partons . indeed , the three - parton term ( [ resu2 ] ) does not contribute to this difference , because it is at most linear in the color generators of massless partons and invariant under rescalings of their momenta .
it would be interesting to explore whether additional constraints on the anomalous dimension arise from considering quasi - collinear limits @xcite , a generalization of the usual collinear limit to the case where some of the partons involved in the splitting are massive .
our result ( [ resu2 ] ) explains the observation made in @xcite , that the three - parton correlations do not arise when two or three of the involved partons are massless ( i.e. , the fact that @xmath135 ) . moreover , when any pair of kinematic invariants are equal , these terms vanish as well . for @xmath117
this follows from the fact that it is an anti - symmetric function in all of its arguments . for @xmath136
it follows since for @xmath137 both arguments of the function vanish , but @xmath136 must be odd in its second argument .
we can thus reproduce and understand all of the observations made in @xcite based on symmetry properties and without an explicit two - loop calculation .
this also demonstrates that these observations will continue to hold in higher orders of perturbation theory . on the other hand , due to reasons similar to the ones outlined above
, it is clear that when massive particles are present more and more complicated color and momentum structures will arise in higher orders of perturbation theory .
the arguments of @xcite show that all of these structures are of non - abelian origin and involve three or more partons . an integral representation for the function @xmath117 can ,
in principle , be extracted from expressions derived in @xcite .
these authors have shown by numerical evaluation that @xmath138 for generic values of its arguments .
it is an open question , however , how the functions @xmath117 and @xmath136 behave in the limit of large cusp angles , corresponding to the case where @xmath101 .
if either one of the two functions does not vanish in this limit , then the factorization theorem for massive amplitudes proposed in @xcite would have to be modified to account for the non - factorizable three - particle terms derived from ( [ resu2 ] ) .
for example , in heavy - quark production processes such as @xmath139 or @xmath140 only the second term in ( [ resu2 ] ) contributes .
if the asymptotic behavior of the coefficient function were @xmath141 , then in the limit @xmath142 the three - parton term would contribute a @xmath0 ir pole at two - loop order proportional to @xmath143 , which would be incompatible with a simple factorization formula .
the agreement of the numerical results presented in @xcite for the @xmath139 scattering amplitude with the predictions obtained in @xcite using the factorization theorem for massive partons provides some evidence that the three - parton contributions do indeed vanish in the limit of large cusp angles , but one should check this with an explicit calculation of @xmath117 and @xmath136 .
we have derived the general form of infrared singularities of two - loop qcd amplitudes involving arbitrary numbers of massive and massless partons .
these singularities can be absorbed into a multiplicative @xmath34-factor , which fulfills a renormalization - group equation . in the purely massless case ,
the associated anomalous dimension involves only color - dipole correlations .
this was first observed by explicit calculation at the two - loop level in @xcite . in a recent paper , we have argued that factorization constraints suggest that this property persists to all orders and have analyzed the three - loop case in detail @xcite .
more recently , mitov et al .
@xcite have shown that the anomalous - dimension matrix for massive partons involves color correlations among three partons , in contrast to the massless case .
they have evaluated these contributions numerically in euclidean space and pointed out that they vanish for certain kinematical configurations .
in the present paper , we have derived the structure of the two - loop anomalous - dimension matrix relevant for the general case involving both massive and massless partons .
the factorization constraints can be studied using a combination of soft - collinear effective theory for the massless partons and heavy - quark effective theory for the massive ones .
we have first derived the form of the two - parton correlations and extracted the necessary anomalous dimensions from known results for heavy - to - heavy and heavy - to - light currents in the effective theory .
the factorization constraints are weaker for terms involving massive legs , since both the full and the effective theories know about the four - velocities of the massive partons .
in particular , the constraints do not exclude three - parton correlations if at least two of the partons involved are massive .
we have shown that two such structures appear and that their symmetry properties imply that they vanish when two four - velocities of the involved massive partons become equal , which explains the findings of @xcite .
starting from their results , one should derive explicit representations for these two functions in minkowski space , preferably in analytic form .
we have briefly discussed the limit in which the particle masses are small compared to the momentum transfers . in this case , the two - parton contribution factors into a hard part , depending on the large momentum transfers , and a sum over collinear contributions depending on the parton masses .
it would be interesting and important to study this case in more detail .
in particular , one should check whether the three - parton correlations vanish in this limit , since their presence would violate the factorization theorem proposed in @xcite .
_ acknowledgments : _ m.n . is grateful to uli haisch and andrea ferroglia for useful discussions .
the research of t.b . was supported by the u.s .
department of energy under grant de - ac02 - 76ch03000 .
fermilab is operated by the fermi research alliance under contract with the department of energy .
soon after this paper was published , the two - loop expressions for the functions @xmath117 and @xmath136 describing the three - parton correlations in ( [ resu2 ] ) have been calculated in closed analytic form @xcite . in light of these results ,
some statements made in the last two paragraphs of section [ sec:3parton ] need to be revised : 1 . the conclusion drawn in @xcite and in the present work , that the three - parton terms vanish whenever the four - velocities of two massive partons coincide , is false . due to the presence of coulomb
singularities the limit @xmath144 is singular .
the three - parton terms do not vanish but diverge logarithmically in this limit .
2 . a formal argument presented in @xcite , suggesting that the three - parton terms only receive contributions from the three - gluon graphs shown in figure [ fig : graphs ] , is invalid .
additional contributions with the same color structure arise from two - loop planar and one - loop counterterm diagrams .
the calculations presented in that paper are therefore not sufficient to extract an integral representation for the function @xmath117 .
we also note that the explicit expressions for the functions @xmath117 and @xmath136 vanish like @xmath145 in the limit of small parton masses .
the factorization theorem for amplitudes with massive partons proposed in @xcite is therefore not invalidated by their presence .
j. c. collins , d. e. soper and g. sterman , adv .
high energy phys .
* 5 * , 1 ( 1988 ) [ arxiv : hep - ph/0409313 ] .
s. catani , phys .
b * 427 * , 161 ( 1998 ) [ arxiv : hep - ph/9802439 ] .
g. sterman and m. e. tejeda - yeomans , phys .
b * 552 * , 48 ( 2003 ) [ arxiv : hep - ph/0210130 ] .
s. m. aybat , l. j. dixon and g. sterman , phys .
lett . * 97 * , 072001 ( 2006 ) [ arxiv : hep - ph/0606254 ] .
s. m. aybat , l. j. dixon and g. sterman , phys .
d * 74 * , 074004 ( 2006 ) [ arxiv : hep - ph/0607309 ] .
z. bern , l. j. dixon and d. a. kosower , jhep * 0408 * , 012 ( 2004 ) [ arxiv : hep - ph/0404293 ] . t. becher and m. neubert , phys .
lett . * 102 * , 162001 ( 2009 ) [ arxiv:0901.0722 [ hep - ph ] ] . c. w. bauer , s. fleming , d. pirjol and i. w. stewart , phys .
d * 63 * , 114020 ( 2001 ) [ arxiv : hep - ph/0011336 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
d * 65 * , 054022 ( 2002 ) [ arxiv : hep - ph/0109045 ] .
m. beneke , a. p. chapovsky , m. diehl and t. feldmann , nucl .
b * 643 * , 431 ( 2002 ) [ arxiv : hep - ph/0206152 ] .
l. j. dixon , arxiv:0901.3414 [ hep - ph ] .
t. becher and m. neubert , jhep * 0906 * , 081 ( 2009 ) [ arxiv:0903.1126 [ hep - ph ] ] .
j. g. m. gatheral , phys .
b * 133 * , 90 ( 1983 ) .
j. frenkel and j. c. taylor , nucl .
b * 246 * , 231 ( 1984 ) .
d. a. kosower , nucl . phys .
b * 552 * , 319 ( 1999 ) [ arxiv : hep - ph/9901201 ] .
e. gardi and l. magnea , jhep * 0903 * , 079 ( 2009 ) [ arxiv:0901.1091 [ hep - ph ] ] .
a. armoni , jhep * 0611 * , 009 ( 2006 ) [ arxiv : hep - th/0608026 ] .
l. f. alday and j. m. maldacena , jhep * 0706 * , 064 ( 2007 ) [ arxiv:0705.0303 [ hep - th ] ] .
l. f. alday and j. m. maldacena , jhep * 0711 * , 019 ( 2007 ) [ arxiv:0708.0672 [ hep - th ] ] .
s. catani , s. dittmaier and z. trocsanyi , phys .
b * 500 * , 149 ( 2001 ) [ arxiv : hep - ph/0011222 ] .
a. a. penin , nucl .
b * 734 * , 185 ( 2006 ) [ arxiv : hep - ph/0508127 ] .
a. mitov and s. moch , jhep * 0705 * , 001 ( 2007 ) [ arxiv : hep - ph/0612149 ] .
t. becher and k. melnikov , jhep * 0706 * , 084 ( 2007 ) [ arxiv:0704.3582 [ hep - ph ] ] .
a. a. penin , phys .
lett . * 95 * , 010408 ( 2005 ) [ arxiv : hep - ph/0501120 ] .
m. czakon , a. mitov and s. moch , phys .
b * 651 * , 147 ( 2007 ) [ arxiv:0705.1975 [ hep - ph ] ] . m. czakon , a. mitov and s. moch , nucl . phys .
b * 798 * , 210 ( 2008 ) [ arxiv:0707.4139 [ hep - ph ] ] .
a. mitov , g. sterman and i. sung , phys . rev .
d * 79 * , 094015 ( 2009 ) [ arxiv:0903.3241 [ hep - ph ] ] . for a review , see : m. neubert , phys .
rept . * 245 * , 259 ( 1994 ) [ arxiv : hep - ph/9306320 ] .
s. catani and m. h. seymour , phys .
b * 378 * , 287 ( 1996 ) [ arxiv : hep - ph/9602277 ] .
s. catani and m. h. seymour , nucl .
b * 485 * , 291 ( 1997 ) [ erratum - ibid .
b * 510 * , 503 ( 1998 ) ] [ arxiv : hep - ph/9605323 ] .
t. becher , r. j. hill , b. o. lange and m. neubert , phys .
d * 69 * , 034013 ( 2004 ) [ arxiv : hep - ph/0309227 ] .
s. w. bosch , r. j. hill , b. o. lange and m. neubert , phys .
d * 67 * , 094014 ( 2003 ) [ arxiv : hep - ph/0301123 ]
. m. neubert , eur .
j. c * 40 * , 165 ( 2005 ) [ arxiv : hep - ph/0408179 ] .
a. f. falk , h. georgi , b. grinstein and m. b. wise , nucl .
b * 343 * , 1 ( 1990 ) .
korchemsky and a. v. radyushkin , nucl .
b * 283 * , 342 ( 1987 ) .
g. p. korchemsky and a. v. radyushkin , phys .
b * 279 * , 359 ( 1992 ) [ arxiv : hep - ph/9203222 ] .
n. kidonakis , phys .
lett . *
102 * , 232003 ( 2009 ) [ arxiv:0903.2561 [ hep - ph ] ] . for a review ,
see : m. steinhauser , phys .
rept . * 364 * , 247 ( 2002 ) [ arxiv : hep - ph/0201075 ] .
w. bernreuther , r. bonciani , t. gehrmann , r. heinesch , t. leineweber , p. mastrolia and e. remiddi , nucl .
b * 706 * , 245 ( 2005 ) [ arxiv : hep - ph/0406046 ] .
w. bernreuther , r. bonciani , t. gehrmann , r. heinesch , t. leineweber , p. mastrolia and e. remiddi , phys .
* 95 * , 261802 ( 2005 ) [ arxiv : hep - ph/0509341 ] .
w. bernreuther , r. bonciani , t. gehrmann , r. heinesch , t. leineweber and e. remiddi , nucl .
b * 723 * , 91 ( 2005 ) [ arxiv : hep - ph/0504190 ] .
h. georgi , phys .
b * 240 * , 447 ( 1990 ) .
s. catani , s. dittmaier , m. h. seymour and z. trocsanyi , nucl .
b * 627 * , 189 ( 2002 ) [ arxiv : hep - ph/0201036 ] .
m. czakon , phys .
b * 664 * , 307 ( 2008 ) [ arxiv:0803.1400 [ hep - ph ] ] .
a. ferroglia , m. neubert , b. d. pecjak and l. l. yang , arxiv:0907.4791 [ hep - ph ] .
a. ferroglia , m. neubert , b. d. pecjak and l. l. yang , arxiv:0908.3676 [ hep - ph ] . | a formula for the two - loop infrared singularities of dimensionally regularized qcd scattering amplitudes with an arbitrary number of massive and massless legs is derived .
the singularities are obtained from the solution of a renormalization - group equation , and factorization constraints on the relevant anomalous - dimension matrix are analyzed .
the simplicity of the structure of the matrix relevant for massless partons does not carry over to the case with massive legs , where starting at two - loop order new color and momentum structures arise , which are not of the color - dipole form .
the resulting two - loop three - parton correlations can be expressed in terms of two functions , for which some general properties are derived .
this explains observations recently made by mitov et al . in terms of symmetry arguments . |
cavity quantum electrodynamics ( cqed ) provides experimental access to the fundamental physics of light - matter interactions @xcite .
such systems are a useful tool in interferometry and spectroscopy@xcite , and quantum information storage and quantum simulation are promising applications @xcite . with few exceptions
@xcite , two - mirror fabry - prot cavities used in cqed research are probed via their fundamental tem@xmath0 mode .
however , recent interest in multimode cqed @xcite wherein the intracavity thermal or bose - condensed atoms are simultaneously coupled to many degenerate but incommensurate optical modes has encouraged an improved understanding of how cavity mode structure deviates from the ideal paraxial resonator as described in ref .
@xcite . in such experiments ,
the overlap between the atomic cloud and the cavity modes is a critical parameter that is dependent on the exact form of the cavity mode functions , the spectrum of which highly depends on mirror fabrication and alignment .
these overlaps are important to characterize , as they determine , e.g. , the effective interaction length scales and atom - atom coupling strengths .
we present here a method of mode - spectrum characterization based on driving the degenerate cavity with carefully tailored mode functions .
this enables the future engineering of atom - atom interactions via shaping the pumped light fields and the measurement of the quantum system s dynamic response @xcite to tailored drive fields .
specifically , we show in this paper how a digital micromirror device ( dmd ) , functioning as a binary spatial light modulator ( slm ) , can be used to generate holographic light fields that selectively couple to each cavity mode .
such a device has recently been used to couple light into the modes of a ring resonator @xcite .
we show that these characterization tasks can be accomplished in a manner that is rapid and re - configurable using the dmd , enabling the versatile _ in - situ _ manipulation of one mode versus another .
furthermore , we study how the mode structure changes as a result of aberration mixing when the cavity is tuned to a degenerate configuration .
we demonstrate that the spatial properties of the intracavity field at multimode resonance can be controllably pumped .
iterative procedures are employed to better couple to individual cavity modes ; this paves the way for the use of optimization algorithms to automatically determine the cavity mode structure .
the spectroscopic properties of an ideal fabry - prot cavity are well described by paraxial wave theory @xcite .
the results of such analyses are two - fold : first , the modes themselves are hermite - gauss ( hg ) beams . indeed in practice , the fundamental mode of a cavity is nearly indistinguishable from an ideal gaussian beam , even when using spherical , rather than parabolic , mirrors .
second , the frequencies ( in hz ) of these modes are given by @xmath1 where @xmath2 is the speed of light in the cavity , @xmath3 is the length of the two - mirror resonator , @xmath4 , and @xmath5 is the radius of curvature ( roc ) of the @xmath6 mirror in the @xmath7 direction .
these denote orthogonal cartesian axes .
this formula captures many of the salient features of optical resonators . in a symmetric cavity where all rocs are equal
, the cavity may exhibit degenerate configurations at particular ratios of cavity length to roc wherein a class of spatially incommensurate modes have identical resonant frequencies . specifically
, this degeneracy can occur when the gouy phase @xmath8 is , e.g. , 0 for planar cavities , @xmath9 for confocal cavities , and @xmath10 for concentric cavities .
the ideal confocal cavity is one well - known example of a degenerate two - mirror resonator ; a confocal cavity s length is equal to the roc of its mirrors @xmath11 .
the mode - spectrum separates into two degenerate families alternating every half free spectral range ( fsr ) , where an fsr is equal to @xmath12 .
one degenerate family consists of all the even modes with @xmath13 , while the other consists of the odd modes with @xmath14 .
it is worth noting that the azimuthal invariance in ideal confocal cavities means that the eigenmodes are also well - described by laguerre - gauss functions , complex admixtures of the real - valued hg functions .
the practicalities of mirror fabrication and mounting often break the mode degeneracy ; even if the mirrors are perfect paraboloids , they are unlikely to have the same four curvatures and same axial alignments . in these realistic cases ,
the spectroscopy of the near - confocal cavity is marked by two energy scales for the hg modes .
the first , @xmath15 , is the frequency difference between families of different @xmath16 , where @xmath17 and @xmath18 . here , @xmath19 , where @xmath20 is the roc of mirror 1 along direction @xmath21 , etc .
the second energy scale is @xmath22 , which is the energy offset between modes within the same family .
this picture is complicated by the fact that the cavity mirrors may not be mounted in such a way that the principle axes of the two mirrors are co - planar . moreover , mixing of the ideal hg modes into a non - uniform @xmath23 spectrum within modes of a family arises from aberrations in the shape of the mirror , e.g. , spherical aberration , beyond - paraxial effects in the propagation of the light , and defects on the mirror surface .
this mixing increases as the modes begin to overlap in frequency as degeneracy is reached .
1140 pixels are grouped into 1,440 rectangular regions , each of which is 19@xmath2438 pixels in extent .
b ) the interference pattern generated in the fourier plane by a dmd programmed with the above mask .
the phase of the fringe pattern reflects the @xmath25 accumulated due to aberrations .
c ) the real and d ) imaginary parts of the numerical fourier transform of the interference pattern shown in panel b ) .
the function describing the fringes in panel b ) is @xmath26 , meaning that in the fourier space of panels c ) and d ) , the spots are at @xmath27 and @xmath28 .
the orientation of the fringes in each spot in c ) and d ) reflects the real space orientation of the dmd grating in panel a ) . for ease
, the phase @xmath25 is found not by analysing the fringes in panel b ) , but by taking the difference in phase between the pair of fringes marked by white circles in panel c ) and d).,width=377 ]
the dmd is positioned at one plane of a 2@xmath29 imaging system ; in the paraxial limit , the electric field in this plane is connected to that in another plane by a fourier transform .
see fig .
[ fig : opticsdiagram ] .
this transform scales one set of spatial dimensions by @xmath30 with respect to the other , where @xmath31 is the wavenumber of the laser field . as such
, a delta function in one plane @xmath32 corresponds to a plane wave in the other plane @xmath33 $ ] , where @xmath34 and @xmath35 are pixel coordinates in the dmd and image planes , respectively .
it is this duality which allows us to use the dmd to generate arbitrary field profiles in the image plane using one of several algorithms known from digital holography @xcite .
a necessary step before the dmd can be used to accurately produce field profiles is the measurement of aberrations in the optical system , including the dmd itself @xcite .
the protective glass of the dmd is known to be under stress from manufacturing and thus exhibits non - uniform strain .
it is possible to characterize these aberrations by measuring the phase shift imparted onto a set of orthogonal basis functions generated at the dmd plane @xcite . locally adding phase shifts in the dmd plane provides full phase control of the holographic image for either aberration removal , phase - sensitive image construction , or both simultaneously . for a binary amplitude mask like a dmd , one way to add phase to the field in the image plane
is to include a grating into the mask design .
this is accomplished by using the dmd mask to modulate the amplitude of the field at the dmd plane via the grating function @xmath36 , where @xmath37 contains all the phase information of the mode and @xmath38 and @xmath39 . for simplicity , we choose approximate delta functions as our set of nearly - orthogonal basis functions since they are both orthogonal in the dmd plane and in any downstream plane .
we approximate delta functions by partitioning the 912@xmath241140 mirror - pixels of the dmd into 1,440 individual 48@xmath2430 rectangular regions .
examples of two such windows are shown in fig .
[ fig : calibration_figure]a .
the finite size of these window regions provides space within which to overlay phase - coherent grating functions .
the specific size is chosen as a balance between making them small enough to provide many partitions of the full dmd array thereby providing high resolution for aberration correction while also ensuring that each window is not too small , such that the transform - limited spot - size is not too large in the fourier - image plane .
setting @xmath40 for the present discussion , each orthogonal basis function has the approximate form of @xmath41 at the dmd plane , where @xmath42 is the rectangle function , @xmath43 is the center of the region , and @xmath44 and @xmath45 are the width and height of the region , respectively . turning on any two windows defined by differing @xmath43yields two interference patterns in the image plane centered at @xmath46 , as shown in fig .
[ fig : calibration_figure]b .
each of these has the form @xmath47 , where @xmath48 and @xmath49 are the separation in @xmath50 and @xmath51 in the dmd plane between the two centers of the window regions .
aberrations in the imaging system are determined by measuring the phase shift in the image plane acquired by light reflected from these window regions in the dmd plane . in the presence of aberrations in the optical path , the interference pattern shown in fig .
[ fig : calibration_figure]b is modified to be @xmath52 , where the phase @xmath25 is that which is accumulated due to the particular optical aberration encountered by the light propagating from the window region to the image plane .
the phase of this pattern is easily obtained by taking the fourier transform of the interference pattern , which produces three complex - valued intensity features .
these can be split into a real - valued image and an imaginary - valued image , as shown in figs .
[ fig : calibration_figure]c and d. the phase @xmath25 is found by taking the difference in phase between the pair of fringes marked by white circles in figs . [
fig : calibration_figure]c and d. the @xmath25 is then measured for all windows tiling the dmd pixel area .
we employ this procedure by fixing one window in the center of the dmd pixel array while another window is sequentially tiled over all other window regions .
together this allows one to determine a phase mask that is a full representation of the linear aberrations of the optical system .
the inverse of such a mask can be superimposed onto any hologram mask programmed into the dmd for the purpose of cancelling these aberrations . in this fashion
we are able to produce high - fidelity representations of desired field mode profiles .
the dmd used is a texas instruments dlp4500nir consisting of 912@xmath241140 individually addressable square aluminum mirrors , which measure @xmath5310.8 @xmath54 m on the diagonal .
they are arranged on a square lattice and can be independently controlled by sending an array of 1 s ( ` on ' ) and 0 s ( ` off ' ) to the controller .
physically , these values correspond to the mirrors changing their angle by @xmath55 with respect to the substrate on which they are fabricated .
light reflected from ` off ' mirrors is directed into a beam - blocker , while light reflected from the ` on ' mirrors is used to form the desired hologram and cavity in - coupling light .
we generate a set of dmd masks for each experiment we wish to perform ; the rate at which we can update holograms is presently limited by the 60-hz refresh rate of the software driver . the laser beam incident on the dmd surface
is magnified to a @xmath56 radius of 8 mm , quite a bit larger than the 7-mm extent of the 45@xmath57-projected dmd surface .
this insures that the dmd is exposed to an homogeneous light intensity across its surface , which we verify by imaging the surface .
the 2-cm cavity mirrors are glued onto a rotationally symmetric brass mount for temperature and vibration stability .
the cavity length can be adjusted via the turning of an 80-turns - per - inch screw that holds one of the two mirrors .
this screw provides course control of the cavity length : @xmath58 . glued in between the screw and the mirror is a single - crystal pzt that can be voltage - tuned through @xmath59 .
the lens connecting the two fourier planes has a focal length of @xmath60 , chosen to maximize the dmd area used in producing the hg fields that will be mode - matched to the cavity .
figure [ fig : opticsdiagram ] presents a diagram of the setup used to couple holograms generated by the dmd into the cavity .
the finesse of the cavity is @xmath53@xmath61 at the wavelength we use for this work , 780 nm , resulting in a cavity linewidth of @xmath53300 khz . .
the red transmission curve results from in - coupling an ideal hg@xmath62 mode into the cavity .
the imperfect overlap induces a coupling to the adjacent modes in addition to the targeted ( 3,1 ) .
the blue curves show the result of increasing optimization of the dmd mask , resulting in a hologram that couples exclusively to the cavity ( 3,1 ) mode , which is shown in the inset .
we determine that the cavity ( 3,1 ) mode roughly consists of 8% hg@xmath63 , 8% hg@xmath64 , and 84% hg@xmath62 .
b ) same is in panel a ) , but with respect to the ( 2,1 ) mode .
we determine the cavity ( 2,1 ) mode roughly consists of 4% hg@xmath65 , 4% hg@xmath66 , and 92% hg@xmath67 .
, width=529 ] the hg modes are a good basis set for dmd studies because they are eigenfunctions of the fourier transform . that is , if the field created by the mask at the dmd plane has the form of an hg mode , then the field in the associated fourier image plane will be a scaled version of the same hg mode @xcite .
to couple light to the cavity , it is necessary to create hg modes with the appropriate scale using the dmd .
we first choose a target hg mode @xmath68 normalized such that @xmath69 .
the phase of the mode @xmath70 is superimposed with the aberration - correcting phase such that the phase of the hologram to be programmed into the dmd is @xmath71 .
both phases are normalized to run between 0 and @xmath72 . including the grating structure necessary to create the hg mode in the fourier image plane , the final field to be produced by the dmd mask is @xmath73 + \phi_m(x , y)\right)^2,\ ] ] where @xmath74 and @xmath75 determine the periodicity of the grating in the dmd plane and thus the center location of the mode in the image plane at @xmath46 .
@xmath76 is a focusing parameter and @xmath77 is a focusing field that is in the form of a paraboloid that corrects for the 45@xmath57-angle of the dmd face .
the function in eqn .
[ field_dmd ] is sampled using floyd - steinberg dithering with a two - color palette to numerically develop a pixel - by - pixel representation of this field at the dmd surface for programming into the controller .
a final reshaping is necessary to generate the file that correctly maps pixel values to dmd mirrors .
this file is then loaded into the dmd controller .
the whole process takes only 30 s per mode - image using a commercial laptop running code written in python . to couple a mode into the cavity , we first choose @xmath78 and @xmath79 to be large enough to sufficiently separate the grating orders in the fourier image plane .
the first - order peak contains the mode image to be coupled and is directed into the cavity while the other orders are blocked .
transmission through the cavity is maximized using two mirrors placed immediately before the cavity .
the gaussian waist of the incoming mode and @xmath76 are optimized through manual iteration to maximize the coupling of the mode . for simplicity
, we typically use the fundamental hg@xmath0 mode for this procedure , which fixes all but one of the cavity parameters .
the indices are the number of nodes in the @xmath21 and @xmath80 axes , respectively .
the remaining parameter , the azimuthal angle of the nodal axes in the transverse plane , is determined manually by observing which modes in a set of non - azimuthally symmetric modes maximize coupling . for simplicity
, we use a collection of hg@xmath81 beams at different azimuthal angles @xmath82 , finally choosing the @xmath82 corresponding to the hg@xmath81 mode exhibiting the best coupling efficiency .
this procedure would , in principle , fully determine the hg mode - basis of the cavity if the cavity conformed to the ideal paraxial model .
however , intrafamily mixing is present , as mentioned above , and the actual cavity modes do not always correspond to ideal hg - mode shapes .
an example of this can be seen in the insets of fig .
[ fig : rectification ] . to label the modes comprising the actual mode spectrum , we number each mode of an @xmath83 family by indices @xmath84 , where @xmath85 and @xmath86 no longer necessarily correspond to the number of nodes along a direction @xmath21 and @xmath80 , respectfully . however , the condition @xmath87 remains fixed for each mode family @xmath88 .
is a good approximation .
] moreover , we choose to denote the highest - frequency mode in a family the @xmath89 mode , rather than the @xmath90 mode .
nodes along only one direction typically possess the highest resonance frequency . ]
the single - mode character of the cavity is probed by setting the length of the cavity to be 1-mm - longer than the confocality condition .
while most modes in a given family possess a spatial structure very close to an ideal hg - mode , we do find at least one per family that significantly deviates from the hg - modal structure .
for example , in the @xmath91 and @xmath92 families , the @xmath93 and @xmath94 modes both show significant deviation from the ideal hg@xmath67 and hg@xmath62 modes , respectively , as can be seen from the insets of fig .
[ fig : rectification ] .
we determine the hg - modal decompositions through a 2d fitting routing , and the fraction of other intrafamily hg modes is listed in the captions .
figure [ fig : rectification ] shows the results of an iterative procedure used to probe the @xmath93 and @xmath94 modes . in each case , attempting to couple to the mode with an ideal hg resulted in cross - coupling to the immediately adjacent modes .
the optimized masks consist primarily of the hg mode in question with small ( 5%-10% ) admixtures of the adjacent hg modes .
each mask was rotated to an angle @xmath82 optimal for mode coupling .
though there can be much mixing of hg modes , we are able to successfully generate holograms that couple to the @xmath93 and @xmath94 cavity modes with high specificity . as seen in fig .
[ fig : rectification ] , all transmission in non - targeted modes are below the level of our detection signal - to - noise . although optimizing by hand is quite straightforward , an automated procedure can be applied to determine the eigenmodes of the cavity more quickly .
other than the ( 2,1 ) and ( 3,1 ) modes , all other modes in these families were efficiently coupled to by dmd masks employing the ideal hg mode functions .
the result of a spectroscopic scan of the low - order odd and even mode families is shown in fig .
[ fig : single ] .
each trace corresponds to a separate dmd mask designed to couple to a specific single mode of the cavity .
our manual optimization routine ensures that unwanted cross - coupling is minimal and indeed not apparent in these traces .
all modes emitted out of the cavity nearly identical to ideal hg modes , with the exception of the two aforementioned mixtures .
we note that the rapidity of the frequency sweep results in an asymmetric lineshape for each mode due to ringdown effects .
mode added to an hg@xmath0 mode .
each image is the light emitted from the cavity and is the coherent addition of the two modes at a single probe frequency .
b ) same as panel a ) , but with an increasing amount of an hg@xmath66 mode added to an hg@xmath81 mode.,width=340 ] we experimentally probe the basic properties of the near - degenerate cavity at hand , because a numerical characterization @xcite is impractical due to the need to minutely characterize the mode mixing arising from mirror defects , etc . to proceed ,
the cavity is tuned to a near - degenerate configuration wherein the length was adjusted until the nearest second - order mode ( 0,2 ) became degenerate with the fundamental ( 0,0 ) .
figure [ fig : all - modes ] illustrates the character of the degeneracy in this configuration . to allow the observation of the frequency spacing of modes ,
the cavity was pumped with a broad beam slightly off - center .
the modes are seen to arrange into clusters in this near - confocal configuration .
it is possible to study some aspects of cavity degeneracy in this configuration since many modes appear with separations of less than a few linewidths .
the transmission of the cavity was measured for each ideal hg mask as before .
figure [ fig : mm ] shows the cavity transmission for each mask . as expected , there is a now a greater degree of mode mixing , especially in the higher order modes , as can be seen by the prevalence of side - peaks at many distinct probe frequencies . nevertheless , each ideal hg mask couples to a small collection of cavity modes within its spectral vicinity .
we note that the two lowest - order even and odd modes blue traces at the bottom of each figure can be selectively addressed by distinct , orthogonal hg masks . finally , to demonstrate that the near - degenerate cavity can be pumped by tailored field modes , we show in fig .
[ fig : mix ] that any linear combination of , e.g. , hg@xmath0 with hg@xmath95 or hg@xmath81 with hg@xmath66 , can be coupled into the near - degenerate cavity .
the ability of cavity fields to mediate atomic interactions with high flexibility is an enticing new direction in quantum simulation . the dmd has allowed us to demonstrate an important step in this direction , namely the identification and specific probing of modal content within overlapping mode spectra of single - mode and near - degenerate confocal fabry - pert cavities .
the authors thank j. keeling and t. ciobanu for helpful discussions . funding for this effort
was provided by the army research office ( aro ) , and b.l.l .
also thanks the david and lucile packard foundation for support . | digital micromirror devices ( dmd ) provide a robust platform with which to implement digital holography , in principle providing the means to rapidly generate propagating transverse electromagnetic fields with arbitrary mode profiles at visible and ir wavelengths .
we use a dmd to probe a fabry - prot cavity in single - mode and near - degenerate confocal configurations .
pumping arbitrary modes of the cavity is possible with excellent specificity by virtue of the spatial overlap between the incident light field and the cavity mode . 10 d. f. walls and g. j. milburn , _ quantum optics _
( springer - verlag berlin heidelberg , 2008 ) .
d. meiser , j. ye , d. r. carlson , and m. j. holland , `` prospects for a millihertz - linewidth laser , '' phys .
rev .
lett . * 102 * , 1636014 ( 2009 ) .
h. ritsch , p. domokos , f. brennecke , and t. esslinger , `` cold atoms in cavity - generated dynamical optical potentials , '' rev .
mod .
phys . *
85 * , 553601 ( 2013 ) .
m. gangl , p. horak , and h. ritsch , `` cooling neutral particles in multimode cavities without spontaneous emission , '' journal of modern optics * 47 * , 27412753 ( 2000 ) .
v. vuleti , h. w. chan , and a. t. black , `` three - dimensional cavity doppler cooling and cavity sideband cooling by coherent scattering , '' phys .
rev .
a * 64 * , 0334057 ( 2001 ) .
p. horak , h. ritsch , t. fischer , p. maunz , t. puppe , p. w. h. pinkse , and g. rempe , `` optical kaleidoscope using a single atom , '' phys .
rev .
lett . * 88 * , 0436014 ( 2002 ) .
p. maunz , t. puppe , t. fischer , p. w. h. pinkse , and g. rempe , `` emission pattern of an atomic dipole in a high - finesse optical cavity , '' opt .
lett .
* 28 * , 4648 ( 2003 ) .
a. t. black , h. w. chan , and v. vuleti , `` observation of collective friction forces due to spatial self - organization of atoms : from rayleigh to bragg scattering , '' phys .
rev .
lett . * 91 * , 2030014 ( 2003 ) .
h. w. chan , a. t. black , and v. vuleti , `` observation of collective - emission - induced cooling of atoms in an optical cavity , '' phys .
rev .
lett . * 90 * , 0630034 ( 2003 ) .
s. gopalakrishnan , b. l. lev , and p. m. goldbart , `` emergent crystallinity and frustration with bose
einstein condensates in multimode cavities , '' nat phys * 5 * , 845850 ( 2009 ) .
s. gopalakrishnan , b. l. lev , and p. m. goldbart , `` atom - light crystallization of bose - einstein condensates in multimode cavities : nonequilibrium classical and quantum phase transitions , emergent lattices , supersolidity , and frustration , '' phys .
rev .
a * 82 * , 04361229 ( 2010 ) .
s. gopalakrishnan , b. l. lev , and p. m. goldbart , `` frustration and glassiness in spin models with cavity - mediated interactions , '' phys .
rev .
lett . * 107 * , 2772015 ( 2011 ) .
p. strack and s. sachdev , `` dicke quantum spin glass of atoms and photons , '' phys .
rev .
lett . * 107 * , 2772025 ( 2011 ) .
s. gopalakrishnan , b. l. lev , and p. m. goldbart , `` exploring models of associative memory via cavity quantum electrodynamics , '' philosophical magazine * 92 * , 353361 ( 2012 ) .
m. mller , p. strack , and s. sachdev , `` quantum charge glasses of itinerant fermions with cavity - mediated long - range interactions , '' phys
. rev . a * 86 * , 02360416 ( 2012 ) .
m. buchhold , p. strack , s. sachdev , and s. diehl , `` dicke - model quantum spin and photon glass in optical cavities : nonequilibrium theory and experimental signatures , '' phys .
rev .
a * 87 * , 06362224 ( 2013 ) .
e. g. d. torre , s. diehl , m. d. lukin , s. sachdev , and p. strack , `` keldysh approach for nonequilibrium phase transitions in quantum optics : beyond the dicke model in optical cavities , '' phys . rev .
a * 87 * , 023831 ( 2013 ) .
a. j. kollr , a. t. papageorge , k. baumann , m. a. armen , and b. l. lev , `` an adjustable - length cavity and boseeinstein condensate apparatus for multimode cavity qed , '' new j. phys .
* 17 * , 043012 ( 2015 ) .
n. schine , a. ryou , a. gromov , a. sommer , and j. simon , `` synthetic landau levels for photons , '' arxiv:1511.07381 ( 2015 ) .
a. sommer and j. simon , `` engineering photonic floquet hamiltonians through fabry - prot resonators , '' arxiv:1511.00595 ( 2015 ) .
j. ningyuan , a. georgakopoulos , a. ryou , n. schine , a. sommer , and j. simon , `` observation of cavity rydberg polaritons , '' arxiv:1511.01872 ( 2015 ) .
a. e. siegman , _ lasers _ ( university science books , 1986 ) .
r. landig , f. brennecke , r. mottl , t. donner , and t. esslinger , `` measuring the dynamic structure factor of a quantum gas undergoing a structural phase transition , '' nat . commun .
* 6 * , 7046 ( 2015 ) .
w .- h .
lee , `` binary synthetic holograms , '' appl . opt .
* 13 * , 1677 ( 1974 ) .
d. stuart , o. barter , and a. kuhn , `` fast algorithms for generating binary holograms , '' arxiv:1409.1841 ( 2014 ) .
t. cizmar , m. mazilu , and k. dholakia , `` in situ wavefront correction and its application to micromanipulation , '' nature photon .
* 4 * , 388394 ( 2010 )
. i. m. vellekoop and a. p. mosk , `` focusing coherent light through opaque strongly scattering media , '' opt .
lett . *
32 * , 23092311 ( 2007 ) .
p. p.
j. zupancic , `` dynamic holography and beamshaping using digital micromirror devices , '' master s thesis , ludwig - maximilians - universitt , munich ( 2013 ) . j. y. vinet and p. hello , `` matrix simulation of optical cavities , '' journal of modern optics * 40 * , 19811993 ( 2007 ) . |
coherent elastic neutrino - nucleus scattering is a well - predicted standard model interaction that is of wide - spread interest .
characterizing the process can provide a probe of beyond the standard model physics , such as through non - standard neutrino interactions @xcite , and can contribute to our understanding of supernova dynamics . although experiments have been proposed to search for this channel @xcite , neutrino - nucleus coherent elastic scattering has never been observed .
this paper shows that , for existing proposed layouts of neutrino sources and ton - scale dark matter detectors , discovery of coherent neutrino scattering can occur with a 2 ton@xmath0year exposure under an optimistic detection scenario . along with coherent neutrino physics ,
the observation of rare , wimp - like events with a well predicted flux ( shape and absolute normalization ) and cross section in a well - known time window can provide a cross - check of the sensitivity and efficiency of dark matter detectors .
furthermore , in the case that a substantial event sample is collected with a dedicated experiment close to the neutrino source , sensitivity to physics beyond the standard model and a unique probe of @xmath1 can be obtained through a coherent neutrino scattering cross section measurement .
we discuss the physics importance of the coherent neutrino interaction in the next section and detection methods in section [ sec : detection ] .
section [ sec : discovery ] provides our assumptions for a few example experiments and the raw rates in these detectors with various exposures and as a function of envisioned energy thresholds and baseline lengths .
the effect of coherent events as a background for wimp interactions is also considered .
then , we consider the scenario in which a detector module is brought within tens of meters of the neutrino source and used to obtain a high statistics sample of events for coherent neutrino physics ( section [ sec : closedetector2 ] ) .
the coherent scattering cross section , @xmath2 , depends on the number of neutrons , @xmath3 , and protons , @xmath4 , of the target material with mass @xmath5 .
if @xmath6 is the recoil energy of the interaction and the incoming neutrino has energy @xmath7 , @xmath8 in this equation , @xmath9 and @xmath10 are the precisely known fermi constant and weak charge [ @xmath11 , respectively .
the form factor , @xmath12 , dominates the @xmath135% cross section uncertainty @xcite .
as the cross section is well predicted , coherent elastic neutrino - nucleus scattering is an ideal source to search for new physics in the neutrino sector .
a cross section measurement with @xmath1310% uncertainty will result in an uncertainty on @xmath1 of @xmath135% @xcite . while this uncertainty is large compared to existing and planned precision atomic parity violation and mller scattering measurements , a discrepancy from the standard model prediction already observed in the neutrino sector by the nutev experiment @xcite motivates more neutrino - based measurements .
notably , a @xmath1 measurement with coherent neutrino - nucleus scattering would be at @xmath14 gev / c , well away from all previous neutrino scattering measurements ( including nutev s at @xmath15 gev / c ) .
a coherent neutrino - nucleus scattering cross section measurement agreement within 10% uncertainty of the standard model prediction will result in limits on non - standard neutrino interactions ( nsi ) which improve on the present ones by more than an order of magnitude @xcite . the low-@xmath16 existing and planned precision measurements mentioned above are not sensitive to new physics unique to neutrino interactions .
nsi terms can enter the standard model lagrangian through an extra term , @xmath17 where @xmath18 is a first generation standard model fermion , @xmath19 , @xmath20 , or @xmath21 and @xmath22 @xcite .
these @xmath23 terms can appear due to a range of sources , including incorporating neutrino mass into the standard model @xcite and supersymmetry @xcite .
the @xmath24 and @xmath25 terms , in particular , are poorly constrained by existing measurements . measuring a coherent scattering cross section in disagreement with the standard model expectation
could be an indication of nsi . in the case that a cross section discrepancy is observed , multiple nuclear targets could be employed in order to disentangle effects from nsi , a @xmath1 anomaly ( e.g. consistent with nutev ) , and/or nuclear physics . characterizing neutrino coherent scattering is also essential to the understanding of supernova evolution as the energy carried away by neutrinos comprises @xmath1399% of the supernova s total energy and the coherent channel s cross section exceeds all others by at least an order of magnitude in the relevant energy region . in a stellar core collapse
, the density of the electron / nucleus plasma at the core can reach @xmath26@xmath27 @xmath28 . at these densities ,
a 20 mev neutrino s mean free path is on the order of 0.5 km @xcite with the opacity in nucleus - rich regions dominated by coherent scattering .
along with being relevant for supernova evolution in general , the coherent cross section may also affect the supernova neutrino signals expected on earth .
the coherent process is important for supernova burst neutrino detection as well , providing information about all flavors of neutrinos not just @xmath29/@xmath30 . for coherent neutrino - nucleus scattering specifically , unlike other channels of flavor - blind neutral current interactions , the nuclear recoil energy is proportional to neutrino energy due to the elastic nature of the interaction .
such information could be combined with charged current @xmath29/@xmath30 interaction information from other sources to develop a complete picture of oscillations with supernova neutrinos .
note that approximately seven neutrino - nucleus coherent events in one ton of ar during a ten second window for a galactic core - collapse supernova at 10 kpc are expected with a recoil energy threshold of 5 kev @xcite .
although the detectors discussed below are probably too small to provide a sizable sample of supernova burst neutrino - nucleus coherent scatters ( unless the supernova is very close ) , an accelerator - based measurement to confirm the predicted interaction cross section would prove valuable to a next - generation coherent neutrino scattering experiment s supernova burst neutrino measurement .
the coherent neutrino - nucleus cross section favors very low recoil energies , in the few - to - tens of kev range .
this is well below the threshold of the most sensitive recent and existing large - scale low energy neutrino detectors , like sno @xcite , borexino @xcite , and kamland @xcite , which explains why this relatively high cross section process has not yet been observed .
dark matter detectors , on the other - hand , have energy thresholds in the @xmath1310 kev range and lower . as such , these detectors are potentially ideal coherent neutrino scattering detectors if given a sufficiently large target mass and neutrino flux
. a ton - scale dark matter detector at its nominal depth underground of 1 - 2 km , in combination with an intense decay - at - rest ( dar ) neutrino source , could discover the coherent neutrino scattering process .
it is worth noting that dark matter detectors could observe @xmath31b solar neutrino coherent events , as has been pointed out in ref .
the @xmath31b rate depends on the detector threshold and material , with more events for lighter target nuclei and lower thresholds .
this solar neutrino signal becomes negligible for high-@xmath32 targets when the low - energy recoil threshold of the experiment is between 5 and 10 kev or more .
there are zero solar coherent events expected with a 5/10 kev threshold for the targets ( xe , ge / ar ) and exposures considered here . in contrast , the dar accelerator source described in section [ sec : discovery ] will produce a significant number of events within a well - defined time window when the accelerator is on .
these events can therefore only be considered a wimp background during those times ( see sec .
[ sec : wimpsearch ] ) . in terms of a coherent neutrino physics measurement
, the accelerator - based case has the luxury of an _ in - situ _ background measurement ( when the source is off ) in addition to the higher - energy , above - threshold nuclear recoils .
in order to be reasonably concrete , we study a set of experimental designs inspired by proposals for the deep underground science and engineering laboratory ( dusel ) .
we note that the detector designs are not very different from those under consideration at other underground laboratories and that the results can be easily scaled . for the neutrino source
, we assume a dar configuration produced by high intensity cyclotrons which are now under development @xcite@xcite and proposed for dusel @xcite .
dar neutrinos are known to be an excellent source for neutrino - nucleus coherent scattering experiments @xcite .
the neutrinos are produced with relatively low energies ( @xmath3352.8 mev ) , a range where coherent neutrino scattering dominates all other cross sections by about an order of magnitude .
@xcite calls for a design with 800 mev protons , accelerated via high intensity cyclotron(s ) , which impinge on a carbon target and facilitate the production and eventual decay of pions : @xmath34 , followed by @xmath35 .
a dar source flux profile is shown in fig .
[ flux ] .
, title="fig:",width=336 ] high intensity dar sources are being proposed for @xmath36 violation searches involving gd - doped ultra - large water cerenkov detectors at underground science laboratories @xcite .
the design for this search @xcite calls for multiple accelerator sites at varying distances from ultra - large water cerenkov detector(s ) .
the dae@xmath37alus cyclotron - based near accelerator is proposed to run with a duty factor between 13% @xcite and 20% @xcite , with an average of 1 mw of power in either case .
each 1 mw accelerator will provide @xmath38 neutrinos of each flavor per year produced as an isotropic flux within the time window , expected to be about 67 ms out of 500 ms for the near accelerator with the current design @xcite .
the absolute normalization of the neutrino flux , determined with electron - neutrino elastic scattering ( @xmath39 ) as measured by ultra - large water cerenkov detector(s ) , will have a systematic uncertainty of 1% with dominant contributions from the cross section and energy scale uncertainties @xcite .
the statistical uncertainty on the flux depends on the run period , but is expected to be on the order of 1% as well .
the near accelerator site is envisioned at or near the surface of the laboratory with the other cyclotrons located many kilometers away .
note that the far sites will produce insignificant coherent rates due to the 1/@xmath40 dependence of the flux .
however , the near accelerator can provide a significant event rate , during the 13% beam - on time , for detectors which are sufficiently close and large .
examples of other physics opportunities with this near accelerator are discussed in refs .
@xcite .
in order to provide realistic calculations , we examine three dark matter experiments which are drawn from the designs of geodm @xcite , lz @xcite , and max @xcite .
these experiments use germanium , xenon , and argon as their targets , respectively . note that neon is also commonly considered as an alternative target medium in the noble liquid detectors mentioned @xcite .
we assume that the accelerator and beam dump are located at or near the surface .
as geodm is proposed for the dusel 7400 ft level and lz / max are proposed for the 4800 ft level , we simply consider baseline lengths of 2.3 km and 1.5 km , respectively . the rates for each target are calculated for a ton@xmath0year fiducial exposure since the design of each detector is still under consideration . as discussed above
, the coherent neutrino - nucleus interaction takes place at very low recoil energies . fig .
[ recoils ] shows the distribution of recoil energies for a dar source with @xmath41ne , @xmath42ar , @xmath43ge , and @xmath44xe .
the experimental rates will strongly depend upon the recoil energy threshold for reconstructed events , @xmath45 . as the exact values of this cut for the various detectors are unknown , we consider five possible values of @xmath45 .
the rates with each cut are obtained by summing eq .
[ coherent ] in bins of recoil energy @xmath6 , starting at @xmath45 . for the aforementioned targets
, we find the rates given in table i , where we assume 100% efficiency for detecting events in the time - window above the threshold @xmath45 .
[ ratesa ] .coherent neutrino scattering events / ton / year ( with the accelerator running at 1 mw with a 13% duty factor ) for various detector layouts and thresholds .
the rates reported assume 100% detection efficiency . @xmath46the
present plan is for the geodm ( @xmath43ge - based ) baseline to be 2.3 km although 1.5 km is included for completeness . [
cols="^,^,^,^,^,^,^ " , ] using the baseline efficiency and energy threshold , we find that a coherent neutrino rate of 0.8 events / ton / year is expected in a @xmath43ge - based detector at a baseline of 2.3 km .
geodm expects an overall leakage rate of about 1 background surface leakage event with a 4.5 ton@xmath0year exposure , obtained by adjusting efficiency / fiducial volume cuts to get to that number .
in addition , there is a neutron - induced nuclear recoil background ( absolute rate convolved with detection efficiency ) of 0.13 events/(4.5 ton@xmath0year ) and 0.29 events/(4.5 ton@xmath0year ) for the baseline and optimistic scenarios , respectively .
the total expected background rate during the 13% of beam - on time would therefore be 0.15 events/(4.5 ton@xmath0year ) for the baseline and 0.17 events/(4.5 ton@xmath0year ) for the optimistic scenario .
the assumptions for geodm as a coherent neutrino detector are shown in table iii .
based on these considerations and assuming no wimp background " , one can see that an experiment like geodm could find evidence for coherent scattering in a @xmath134.5 ton@xmath0year exposure with 3 - 4 signal events above a background expectation of 0.15 events .
the probability for 4 observed events to be completely due to background , with a background expectation of 0.15 events , is @xmath130.002% .
the signal rate and evidence / discovery timeline would be quickly improved in the case that the baseline efficiency estimate , especially in the low energy region ( and possibly below 10 kev ) , is too conservative . with the optimistic energy threshold and efficiency scenario ,
we expect a coherent rate of 2.0 events / ton / year . under the same background assumption as above
, we find that geodm will discover coherent scattering in a @xmath132 ton@xmath0year exposure with 4 signal events above a background expectation of 0.07 events .
the probability for 4 observed events to be completely due to background , with a background expectation of 0.07 events , is @xmath130.00009% .
it is worth noting that the absolute ( 100% efficiency and 100% on - time ) solar coherent neutrino interaction rate on a @xmath43ge target above 5 kev is expected to be 0.079 events / ton / year @xcite . in either efficiency scenario and with 13% beam - on time
, the solar coherent background " rate is negligible .
[ deep_params ] scenarios considered & baseline " and optimistic " + @xmath47 source & @xmath48 @xmath47/flavor / year w/ 13% duty factor + @xmath47 flux uncertainty & 2% + distance from @xmath47 source&2.3 km + exposure&4.5 ton@xmath0year + since coherent neutrino scattering is an irreducible background for wimp searches , the presence of a neutrino source near a dark matter experiment will reduce the sensitivity of the experiment . in calculating a wimp - nucleon cross section limit , one can either use data from only the 87% of the time when the neutrino source is off or data from the 87% beam - off time and the 13% beam - on time .
since background events reduce the sensitivity of a limit in the optimum interval method @xcite , using the combined exposure may result in a worse limit than using only the beam - off exposure .
[ fig : wimplimitsbaseline ] and fig .
[ fig : wimplimitsoptimistic ] present a comparison of limits for the baseline and optimistic geodm detector scenarios .
the limits assume a geodm - style detector with a raw exposure of 4.5 ton@xmath0year before efficiency cuts and the same assumptions as in the previous section .
in particular , we use the hypothetical efficiency curves shown in fig .
[ efficiency ] , a neutron rate of @xmath49 events / ton / year before efficiency convolution , and a total surface - event leakage of 1 event per 4.5 ton@xmath0year raw exposure .
we use the recoil energy distribution from fig .
[ efficiency ] for neutrino events in the 13% beam - on time and a detector at 2.3 km from a dar neutrino source . using these parameters
, we randomly generate 100 realizations of the background events and compute the average of the upper limits on the wimp - nucleon cross section for each .
in this study , the beam - off data actually has greater sensitivity than the combined beam - on and beam - off data . using the beam - off data with 87% exposure only ,
the mean upper limit at maximum sensitivity is @xmath50@xmath51 ( baseline ) or @xmath52@xmath51 ( optimistic ) .
the combined data give a limit of @xmath53@xmath51 ( baseline ) or @xmath54@xmath51 ( optimistic ) .
the results for the 87% exposure obviously also have worse sensitivity than 100% exposure without any beam - on time .
if there were no beam - on time , the upper limit at maximal sensitivity would be @xmath55@xmath51 ( baseline ) or @xmath56@xmath51 ( optimistic ) .
note that when the number of background events is zero , the sensitivity to wimps scales as the reciprocal of the exposure .
this occurs because the expected number of events in an experiment is proportional to the product of the cross section @xmath2 and the exposure @xmath57 , @xmath58 a limit at confidence level @xmath59 is obtained by determining the expected number of events @xmath60 such that there is a probability @xmath59 of observing zero events .
assuming poisson statistics the expected number of events is @xmath61 .
it is thus apparent from eq .
( [ eqn : crosssectionscaling ] ) that the cross section limit @xmath2 corresponding to confidence level @xmath59 , scales as the reciprocal of the exposure @xmath57 . in the presence of a nonzero background that is proportional to exposure ,
the limit will scale more slowly than the reciprocal of the exposure .
fixed backgrounds set by fiducial cuts may produce more complicated scaling behavior .
a suitable detector within tens of meters of the stopped pion source could gather a rather large sample of elastic neutrino coherent scatters for physics studies .
the 300 ft adit at dusel could provide direct tunnel access to an envisioned experimental site for such a detector .
the cyclotron would then be located just outside the tunnel in a building against the cliff face .
as discussed earlier , a coherent cross section measurement is sensitive to a number of physics possibilities . for simplicity
, we consider a flux - integrated total cross section measurement as our figure of merit .
however , the shape of the cross section as a function of energy is also interesting , especially in the case that a measured total cross section is inconsistent with expectation . although many low - threshold nuclear - recoil - sensitive detector technologies could work ( including a noble liquid detector @xcite or other dark matter detector technology ) , for concreteness we consider the specific example of a set of geodm - derived detectors with 16.7 kg raw mass , 20 m away from the stopped - pion neutrino source . with the optimistic efficiency estimate ( figure [ efficiency ] ) and a fiducial mass of 10 kg , we expect a detected coherent rate of 0.74 events/(10 kg@xmath0day ) within the 13% beam - on time window .
the background rate design goal for such an experiment would be @xmath330.1 events/(10 kg@xmath0day ) within the same window .
this rate seems reasonable with @xmath13300 ft of rock shielding along with modest passive / active shielding immediately surrounding the detector for prompt cosmic - ray - induced background attenuation / tagging .
the radiogenic background is assumed negligible , at a rate consistent with geodm deep underground .
the uncertainty on the non - beam related background estimate will easily be statistics - dominated with an @xmath62 background measurement during beam - off .
we also assume that there are no background events from neutrons produced by the dar neutrino source at a 20 m baseline with 17 m rock and 3 m fe shielding .
this assumption was justified by performing a geant4 @xcite simulation of an isotropic neutron source along a 20 m baseline , fitting the flux at various distances from the neutron source to the functional form @xmath63 where @xmath32 and @xmath64 are fit parameters and @xmath65 is the distance along the baseline .
the rate from this fit was extrapolated to 20 m distance and a 50 kg@xmath0year exposure , and the number of neutron events was found to be negligible .
the neutron flux and spectrum used were taken from the sns source , which is similar to the dar source discussed here @xcite .
[ close_params ] scenario considered & optimistic " + source & @xmath48 @xmath47/flavor / year w/ 13% duty factor + @xmath47 flux uncertainty & 2% + distance from @xmath47 source&20 m + exposure&50 kg@xmath0year + background rate & 0.1 events/(10 kg@xmath0day ) in beam window + a 50 kg@xmath0year exposure with the previously described experimental design would yield about 1350 coherent events , assuming a coherent cross section consistent with the standard model . with an optimistic 1% uncertainty on the target mass , 0.1 background events/(10 kg@xmath0day ) with statistical - only error ,
2% absolute flux normalization uncertainty , and a 0.5% uncertainty on the energy resolution near threshold , a flux - averaged total cross section measurement with @xmath335% ( statistical and systematic ) uncertainty would be achieved .
the assumptions that went into the event rate and cross section measurement uncertainty estimates are summarized in table iv .
coherent elastic neutrino - nucleus scattering has never been observed .
relevant for supernova evolution , supernova - burst neutrino detection , probing non - standard neutrino interactions , and measuring @xmath1 with neutrinos at low-@xmath16 , among other topics , the process is very well predicted by the standard model and confirmation of the @xmath135% precision theoretical cross section prediction is needed .
dark matter detectors can double as coherent neutrino scattering experiments as the products of wimp and coherent scattering interactions are predicted to be nearly identical . in the case of a decay - at - rest neutrino source and a suite of dark matter experiments at the same site , the deep underground detectors there would merely need to receive a beam timing signal in order to participate in the coherent search .
furthermore , these detectors would receive a free dark matter detection consistency check in the form of non - wimp rare events in a well - known time window all with a modest cost to the wimp analysis / exposure .
the power of this consistency check is strongly dependent on the number of coherent neutrino events collected , a value which is expected to be fairly low in most configurations . in both optimistic and baseline detection scenarios , the best limit on the wimp - nucleon cross section uses only data from the period when the dar source is off . in the optimistic scenario ,
the cross section limit is only about 12% weaker than if no neutrino source were present .
a coherent neutrino interaction discovery in geodm could be achieved with a 2 ton@xmath0year exposure .
about 2.0 detected coherent neutrino events / ton / year over a background of 0.03 events / ton / year are expected in a geodm - style detector at a 2.3 km baseline , given optimistic assumptions for energy threshold and detection efficiency . even in a conservative ( baseline )
scenario , with energy threshold and efficiency reasonably consistent with cdms ii , evidence for coherent neutrino scattering could be obtained with a 4.5 ton@xmath0year exposure .
in addition , a 10 kg fiducial mass geodm - derived detector brought within 20 m of the neutrino source could collect about 1350 events with a 50 kg@xmath0year exposure .
such a sample would be good for a @xmath335% flux - averaged total cross section measurement uncertainty and significant tests of the standard model .
the authors thank jocelyn monroe for discussions , and chuck horowitz for providing the form factors for use in our calculations .
j. s. thanks bonnie fleming for support .
we thank the national science foundation for support .
j. barranco , o. g. miranda and t. i. rashba , jhep * 0512 * , 021 ( 2005 ) .
j. barranco , o. g. miranda and t. i. rashba , phys .
d * 76 * , 073008 ( 2007 ) .
s. davidson , c. pena - garay , n. rius , and a. santamaria , jhep * 03 * , 011 ( 2003 ) .
d. n. mckinsey _ et al .
_ , j. phys .
* 304 * , 012026 ( 2010 ) . c. galbiati and r. purtschert , j. phys .
ser . * 120 * , 042015 ( 2008 ) .
d. n. mckinsey and k. j. coakley , astropart .
* 22 * , 355 ( 2005 ) .
z. ahmed _ et al .
_ [ cdms collaboration ] , science * 327 * , 1619 ( 2010 ) .
z. ahmed _ et al .
_ [ cdms collaboration ] , phys . rev . lett .
102 , 011301 ( 2009 ) . | coherent elastic neutrino- and wimp - nucleus interaction signatures are expected to be quite similar .
this paper discusses how a next generation ton - scale dark matter detector could discover neutrino - nucleus coherent scattering , a precisely - predicted standard model process . a high intensity pion- and muon- decay - at - rest neutrino source recently proposed for oscillation physics at underground laboratories would provide the neutrinos for these measurements . in this paper , we calculate raw rates for various target materials commonly used in dark matter detectors and show that discovery of this interaction is possible with a 2 ton@xmath0year geodm exposure in an optimistic energy threshold and efficiency scenario .
we also study the effects of the neutrino source on wimp sensitivity and discuss the modulated neutrino signal as a sensitivity / consistency check between different dark matter experiments at dusel .
furthermore , we consider the possibility of coherent neutrino physics with a geodm module placed within tens of meters of the neutrino source . |
after its introduction in fluid dynamics in order to characterize the statistical properties of turbulence ( see the book @xcite and references therein ) , the notion of multifractality has turned out to be relevant in many areas of physics ( see for instance @xcite ) , in particular at critical points of disordered models like anderson localization transitions @xcite or in random classical spin models @xcite .
more recently , the wavefunctions of manybody quantum systems have been found to be generically multifractal , with many studies concerning the shannon - rnyi entropies of the ground states of pure quantum spin models @xcite .
the understanding of shannon - rnyi entropies of excited states in disordered models is also essential to characterize the many - body localization transition @xcite . in the present paper ,
we analyze via real - space renormalization the multifractal properties of the ground state of the quantum ising chain @xmath0 where the ferromagnetic couplings @xmath1 and the transverse fields @xmath2 are either uniform or random . for the pure chain , we compare with the previous results of refs @xcite in order to test the validity of the rg approach . for the random case , we are not aware of previous works concerning the multifractality of the ground state . the paper is organized as follows . in section [ sec_multif ] ,
we recall the multifractal formalism for quantum wavefunctions . in section [ sec_rg ]
, we derive the real space renormalization rule for the shannon - rnyi entropies .
the application to the pure and to the random quantum ising chain are described in sections [ sec_pure ] and [ sec_random ] respectively .
our conclusions are summarized in section [ sec_conclusion ] .
for an hilbert space of size @xmath3 ( growing exponentially with the volume ) , the expansion of a given wave - function @xmath4 onto a basis of m vectors @xmath5 @xmath6 involves @xmath3 coefficients @xmath7 subjected to the global normalization @xmath8 the multifractal formalism allows to characterize the statistical properties of these @xmath3 weights @xmath9 as we now recall .
the inverse participation ratios @xmath10 define the exponents @xmath11 @xmath12 the related rnyi entropies involve the generalized fractal dimensions @xmath13 @xmath14 with the simple relation @xmath15 for @xmath16 , eq .
[ renyi ] corresponds to the standard shannon entropy @xmath17 among the @xmath3 configurations , the number @xmath19 of configurations @xmath20 having a weight of order @xmath21 defines the multifractal spectrum @xmath18 @xmath22 the saddle - point calculus in @xmath23 of the inverse participation ratios of eq .
[ ipr ] yields @xmath24 where the saddle point value @xmath25 is determined by the condition @xmath26 i.e. @xmath11 and @xmath18 are related via the legendre transform @xmath27 this yields the following parametric representation of @xmath18 @xmath28 @xmath29 let us now mention some important special values of the index @xmath30 . for @xmath31 , the ipr of eq .
[ ipr ] simply measures the size of the hilbert space @xmath32 corresponding to @xmath33 and @xmath34 and to @xmath35 : this means that there exists an extensive number @xmath36 of configurations having a weight @xmath37 , so that @xmath38 is called the typical exponent . for the shannon value @xmath16 , the ipr of eq .
[ ipr ] is normalized to unity , so that @xmath39 and @xmath40 is called the information dimension that characterizes the shannon entropy of eq . [ shannon ] . in the limit @xmath41 , the ipr of eq .
[ ipr ] is dominated by the configuration having the highest weight @xmath42 , i.e. @xmath43 represents the smallest exponent @xmath44 and the corresponding singularity spectrum vanishes @xmath45 for the quantum ising chain of eq .
[ h1d ] containing @xmath46 spins , the expansion of the wave function in the @xmath47 basis @xmath48 involve @xmath49 coefficients @xmath50 . in the ferromagnetic limit where all transverse fields vanish @xmath51 ,
the ground state is simply the linear combination of the two ferromagnetic states ( both for periodic or free boundary conditions ) @xmath52 so that among the @xmath49 coefficient @xmath53 , only two are non - vanishing and equal to @xmath54 . as a consequence the ipr of eq .
[ ipr ] reduces to @xmath55 and the corresponding rnyi entropies all take the simple finite value @xmath56 i.e. all generalized fractal dimensions of eq .
[ renyi ] vanish @xmath57 in the paramagnetic limit where all ferromagnetic couplings vanish @xmath58 , the ground state is simply given by the tensor product of the ground state of @xmath59 for each spin ( both for periodic or free boundary conditions ) @xmath60 i.e. all @xmath49 coefficients are equal to @xmath61 so that the ipr of eq .
[ ipr ] read @xmath62 and the rnyi entropies all have the same extensive value @xmath63 i.e. all generalized fractal dimensions of eq .
[ renyi ] are equal to unity @xmath64 apart from the two extreme cases just discussed , one expects that the ground state wavefunction is characterized by a non - trivial multifractal spectrum . for the pure case where @xmath65 and @xmath66 , exact expressions for the generalized dimension @xmath13 as a function of the control parameter @xmath67 exist in the limit @xmath68
@xcite @xmath69 \label{dqinftyexact}\end{aligned}\ ] ] and thus also for @xmath70 as a consequence of the duality relation @xcite @xmath71 these functions are continuous in the control parameter @xmath72 , but the derivatives are singular at the critical point @xmath73 @xmath74 \nonumber \\ & & = \mp \frac{1 } { \pi ( \ln 2 ) ( 1+k ) } { \cal k } \left(\frac{4 k}{(1+k)^2 } \right ) + \frac{1 } { 2 k \ln 2 } { \rm sgn } ( k-1 ) \label{dqinftyexactderi}\end{aligned}\ ] ] where @xmath75 is the elliptic integral of the first kind displaying the logarithmic singularity @xmath76 for @xmath77 . as a consequence , the first term corresponds to a logarithmic divergence , whereas the second term corresponds to a discontinuity @xmath78 other values of @xmath30 , especially the shannon value @xmath16 , have been studied numerically @xcite . in the following sections ,
we describe how the fernandez - pacheco real - space renormalization allows to derive a simple approximation for the multifractal properties , both for the pure and for the random chain .
the pure quantum ising chain is the basic model to study quantum phase transitions at zero - temperature @xcite : it is in the same universality class as the exactly solved 2d classical ising model .
the random quantum ising chain has been solved by daniel fisher @xcite via the asymptotically exact strong disorder renormalization procedure : the transition is governed by an infinite disorder fixed point and presents unconventional scaling laws with respect to the pure case . from the point of view of block - renormalization for quantum models
, there exists a simple self - dual procedure introduced by fernandez - pacheco @xcite , which is able to reproduce the exact critical point @xmath79 and the exact correlation length exponent @xmath80 of the pure chain @xcite , even if it is not able to reproduce the exact magnetic exponent @xcite .
its application to the disordered chain also reproduces correctly the location of the critical point , the activated exponent @xmath81 and the two correlation length exponents @xcite , even if it is not able to reproduce the properties dominated by rare events @xcite . in this section , we describe how it leads to simple renormalization rules for the shannon and rnyi entropies of the ground state wavefunction . the goal is to replace each block of two spins @xmath82 by a single renormalized spin @xmath83 .
the idea of fernandez - pacheco @xcite , written here for the case of arbitrary transverse fields and arbitrary ferromagnetic couplings @xcite , is to decompose the hamiltonian of eq .
[ h1d ] as the sum @xmath84 with the following choice @xmath85 \nonumber \\ h_{extra } & & = \sum_{i } \left [ - h_{2i } \sigma_{2i}^x - j_{2i } \sigma_{2i}^z \sigma_{2i+1 } ^z \right ] \label{h1intra1d}\end{aligned}\ ] ] since @xmath86 is diagonal in the @xmath47 basis for the even spins @xmath87 , we have to diagonalize @xmath86 independently in each subspace @xmath88 .
in addition ,
the odd spins @xmath89 are independent of each other and submitted to the one - spin hamiltonian @xmath90 the two eigenvalues are independent of the value of @xmath91 @xmath92 with the following corresponding eigenvectors @xmath93 projecting onto the ground state for the odd spins yields the renormalized hamiltonian for the even spins @xmath94 \label{h1drfin}\end{aligned}\ ] ] with the renormalized transverse fields @xmath95 and the renormalized couplings @xmath96 from the point of view of the ipr of eq .
[ ipr ] , the real - space renormalization yields @xmath97 where @xmath98 represents the ipr of the wavefunction @xmath99 ( eq . [ blocklambda1eigen ] ) of the odd spin @xmath100 in the field of the even spin @xmath101 @xmath102^{q } + \left [ \frac{1-\frac{j_{2i-1}}{\sqrt{h^2_{2i-1}+j^2_{2i-1 } } } } { 2 } \right]^{q } \label{yq}\end{aligned}\ ] ] since @xmath98 is actually independent of the value @xmath91 , and only depends on the ratio @xmath103 , it is convenient to introduce the auxiliary function @xmath104^{q } + \left [ \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } \right]^{q } \label{yqpurk}\end{aligned}\ ] ] then eq . [ iprrg ] can be rewritten as @xmath105 where @xmath106 is the ipr of the renormalized system of the @xmath107 even spins with the renormalized parameters given in eqs [ rgh1d ] and [ rgj1d2x ] . the multiplicative rg rule of eq .
[ iprrgfin ] translates into the following additive rg rule for the shannon - rnyi entropies of eq .
[ renyi ] @xmath108 where @xmath109 is the entropy of the renormalized system of the @xmath107 even spins with the renormalized parameters given in eqs [ rgh1d ] and [ rgj1d2x ] . in the following sections , we analyze the consequences of the rg rule of eq .
[ rglogipr ] for the pure chain and for the random chain respectively .
if the initial parameters are @xmath110 on the whole chain , the fernandez - pacheco renormalization yields after one rg step the renormalized parameters ( eqs [ rgh1d ] and [ rgj1d2x ] ) @xmath111 so that the ratio @xmath112 evolves according to the simple rule @xmath113 the disordered attractive fixed point @xmath114 and the ferromagnetic attractive fixed point @xmath115 are separated by the unstable fixed point @xmath73 characterized by the correlation length exponent @xmath116 obtained by @xmath117 .
so the location of the critical point @xmath73 and and the correlation length exponent @xmath116 are in agreement with the exact solution @xcite . the factors @xmath118 in eq .
[ yq ] are all the same , and depend only on the control parameter @xmath119 so that the rg rule of eq . [ rglogipr ] for the shannon - rnyi entropies reads @xmath120 where the function @xmath121 has been introduced in eq .
[ yqpurk ] .
the iteration yields @xmath122 where the @xmath123 term corresponds to the contribution of the @xmath123 rg step , where @xmath124 spins coupled via @xmath125 are eliminated , and where @xmath124 spins survive with the renormalized coupling @xmath126 . from eq .
[ rglogiprpureiter ] , one obtains the generalized fractal dimensions @xmath127 of eq . [ renyi ] as a series over the rg steps @xmath128 @xmath129 in terms of the auxiliary function @xmath130 obtained from eq .
[ yqpurk ] @xmath131^{q } + \left [ \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } \right]^{q } \right ) } { ( 1-q ) \ln 2 } \label{auxdqpurk}\end{aligned}\ ] ] that becomes for the shannon value @xmath16 @xmath132 \ln \left [ \frac{1+\frac{k}{\sqrt{1+k^2 } } } { 2 } \right ] + \left [ \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } \right ] \ln \left [ \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } \right ] } { \ln 2 } \label{auxdq1purk}\end{aligned}\ ] ] as it should , one recovers at the paramagnetic fixed point @xmath114 where @xmath133 the value of eq .
[ dqpara ] @xmath134 and at the ferromagnetic fixed point @xmath135 where @xmath136 the value of eq .
[ dqferro ] @xmath137 + ( a ) generalized fractal dimension @xmath138 as a function of the rnyi index @xmath30 + ( b ) corresponding multifractal spectrum @xmath18 as a function of @xmath23 .
, title="fig:",height=226 ] + ( a ) generalized fractal dimension @xmath138 as a function of the rnyi index @xmath30 + ( b ) corresponding multifractal spectrum @xmath18 as a function of @xmath23 . , title="fig:",height=226 ] at the critical point @xmath73 , the generalized fractal dimensions @xmath139 of eq .
[ dqk ] simply coincide with the auxiliary function of eq .
[ auxdqpurk ] @xmath140^{q } + \left [ \frac{1+\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } \right ) } { ( 1-q ) \ln 2 } \label{dqcriti}\end{aligned}\ ] ] this function is plotted on fig .
[ figmultifcriti ] ( a ) for the rnyi index in the interval @xmath141 $ ] .
let us quote some special values @xmath142}{\ln 2 } = 0.228447 \label{dqcritiplusinfinity}\end{aligned}\ ] ] in order to compare them with the explicit exact values for @xmath70 and @xmath143 @xcite and the numerical values for @xmath16 and @xmath144 @xcite @xmath145 our conclusion is thus that the rg yields a very good approximated value for the shannon value @xmath16 , whereas the approximation is clearly worse for the other rnyi indices @xmath146 .
this seems to indicate that the fernandez - pacheco rg procedure is well suited for the initial hamiltonian corresponding to @xmath16 , whereas it should probably be modified for @xmath146 in order to take into account the deformation of the measure with respect to the initial quantum problem corresponding to @xmath16 .
indeed from the point of view of the equivalent 2d classical ising model , the case where @xmath147 is an integer corresponds to the ising book ( see fig . 4 of @xcite ) where @xmath147 half - planes are glued together .
this clearly changes the measure of the spin distribution and should be taken into account within the rg procedure in order to obtain better values for @xmath146 . on fig .
[ figmultifcriti ] ( b ) , the corresponding multifractal spectrum @xmath148 is plotted as a function of @xmath23 ( see section [ sec_falpha ] for a reminder on its meaning ) from the parametric form of eqs [ alphaq ] and [ falphaq ] @xmath149^{q } \ln \left [ \frac{1+\frac{1}{\sqrt{2 } } } { 2 } \right ] + \left [ \frac{1-\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } \ln \left [ \frac{1-\frac{1}{\sqrt{2 } } } { 2 } \right ] } { \left [ \frac{1+\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } + \left [ \frac{1-\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } } \right ) \nonumber \\
f_{k_c=1}(\alpha_q ) & & = q \alpha_q(k_c=1 ) + \frac { \ln \left ( \left [ \frac{1-\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } + \left [ \frac{1+\frac{1}{\sqrt{2 } } } { 2 } \right]^{q } \right ) } { \ln 2 } \label{falphaqrescriti}\end{aligned}\ ] ] in particular for @xmath31 , the typical exponent of eq .
[ alphazero ] takes the simple value @xmath150 + ( a ) the generalized fractal dimension @xmath151 is continuous a function of the control parameter @xmath67 + ( b ) the derivative @xmath152 as a function of the control parameter @xmath67 displays a logarithmic singularity at the critical point @xmath73 . , title="fig:",height=226 ] + ( a ) the generalized fractal dimension @xmath151 is continuous a function of the control parameter @xmath67 + ( b ) the derivative @xmath152 as a function of the control parameter @xmath67 displays a logarithmic singularity at the critical point @xmath73 . ,
title="fig:",height=226 ] the values @xmath153 are not universal , but the derivatives of @xmath127 with respect to the control parameter @xmath72 are expected to display universal singularities at the critical point .
the first derivative of eq .
[ dqk ] @xmath154 clearly diverges at @xmath73 .
since this divergence is due to the region of large rg - step @xmath155 , we may evaluate the form of the singularity by replacing the series over the integer @xmath155 by the corresponding integral over the real variable @xmath155 , and then perform a change of variable from @xmath155 to @xmath156 @xmath157 i.e. the second derivative displays a pole singularity @xmath158 the singularity of the first derivative is thus given by the following logarithmic divergence @xmath159 this behavior is in agreement with the logarithmic singularity of eq . [ dqinftyexactderising ] concerning the limit @xmath160 .
finally , the function itself is continuous with the following logarithmic singularity @xmath161 for the shannon index @xmath16 , the generalized fractal dimension @xmath151 and its derivative @xmath152 are plotted as a function of the control parameter @xmath67 on fig .
[ figdq1k ] .
the corresponding coefficient of the logarithmic singularity on these figures is @xmath162 as a final remark , let us mention that the rg result of eq .
[ dqk ] satisfies the exact duality relation quoted in eq .
[ duality ] indeed for the index value @xmath70 , the auxiliary function of eq .
[ auxdqpurk ] reads @xmath163^{\frac{1}{2 } } + \left [ \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } \right]^{\frac{1}{2 } } \right)^2 } { \ln 2 } % \nonumber \\ & & = \frac{\ln \left ( \frac{1+\frac{k}{\sqrt{1+k^2 } } } { 2 } + \frac{1-\frac{k}{\sqrt{1+k^2 } } } { 2 } + 2 \sqrt { \frac{1-\frac{k^2}{1+k^2 } } { 4 } } \right ) } { \ln 2 } \nonumber \\ & & = \frac{\ln \left ( 1 + \frac{1}{\sqrt{1+k^2 } } \right ) } { \ln 2 } \label{auxdqpurkdemi}\end{aligned}\ ] ] whereas for the index value @xmath143 , it reads @xmath164 so that the duality relation of eq .
[ duality ] is satisfied term by term in the series representation of eq .
in terms of the ratios @xmath165 the rg rules of eq .
[ rgh1d ] and [ rgj1d2x ] read @xmath166 that generalize the pure rg rules of eq .
[ krpur ] . since this corresponds to a simple addition in log - variables @xmath167 one obtains after @xmath155 rg steps in terms of the initial variables @xmath168 \label{inffp}\end{aligned}\ ] ] the central limit theorem yields the asymptotic behavior @xmath169 + \sqrt{2^k } \sqrt { \left [ var[\ln j_{i } ] + var[\ln h_i ] \right ] } u \label{inffpclt}\end{aligned}\ ] ] where @xmath170 is a gaussian random variable of zero mean and of variance unity .
so for large rg step @xmath155 , the probability distribution @xmath171 of the renormalized coupling @xmath172 is log - normal @xmath173 with the notations @xmath174 + var[\ln h_i ] \label{lognormalnot}\end{aligned}\ ] ] the first term of eq .
[ inffpclt ] yields that the critical point corresponds to the condition @xmath175 and that the typical correlation length exponent with respect to the length scale @xmath176 is @xmath177 outside criticality , the competition between the first and the second term of eq .
[ inffpclt ] shows that the finite - size correlation exponent is @xmath178 at criticality where the first term vanishes , the second random term of order @xmath179 corresponds to an infinite disorder fixed point of exponent @xmath180 all these conclusions of eqs [ criti1d ] , [ ntyp1d ] , [ nav1d ] , [ psi1d ] obtained via the application of the fernandez - pacheco renormalization to the random quantum ising chain @xcite , are in agreement with the fisher strong disorder renormalization exact results @xcite . averaging eq .
[ rglogipr ] over the disorder yields the following rg rule for averaged shannon - rnyi entropies @xmath181 where @xmath182 is the averaged entropy of the renormalized system of the @xmath107 even spins with the renormalized parameters given in eq [ rgruleskr ] .
note that whenever multifractality appears in disordered systems ( see for instance the review @xcite on anderson localization transitions ) , one should make the distinction between the averaged multifractal sprectrum based on the disorder - average of the ipr , and the typical multifractal sprectrum based on the disorder - average of the logarithm of the ipr .
here we only consider the disorder - averaged entropies of eq .
[ rglogiprav ] , corresponding to the disorder - average of the logarithm of the ipr that define the typical multifractal spectrum .
+ ( a ) information dimension @xmath151 as a function of the control parameter @xmath183 + ( b ) the derivative @xmath152 as a function of the control parameter @xmath72 displays a cusp singularity at the critical point @xmath73 . , title="fig:",height=226 ]
+ ( a ) information dimension @xmath151 as a function of the control parameter @xmath183 + ( b ) the derivative @xmath152 as a function of the control parameter @xmath72 displays a cusp singularity at the critical point @xmath73 .
, title="fig:",height=226 ] so in terms of the auxiliary function @xmath130 of eq .
[ auxdqpurk ] , the iteration of eq .
[ rglogiprav ] yields the generalized fractal dimension @xmath184 where @xmath171 is the probability distribution of the renormalized variable @xmath185 after @xmath155 rg steps ( eq . [ inffp ] ) .
of course the values @xmath13 of eq .
[ dqkrandom ] are non - universal as in the pure case , since they depend on the contribution of the first rg steps .
in particular they thus depend upon the initial disorder distribution @xmath186 . on fig .
[ figdq1krandom ] , we show @xmath151 and its derivative @xmath187 as a function of the control parameter @xmath72 of eq .
[ lognormalnot ] for the special case where the initial disorder distribution @xmath186 is itself given by the log - normal form of eq .
[ lognormal ] with the value @xmath188 , so that eq .
[ lognormal ] is valid for all rg steps @xmath155 @xmath189 on fig .
[ figdq1krandom ] , it is clear that @xmath151 and its derivative @xmath187 are both continuous at the critical point @xmath73 , but that the derivative displays a cusp ( see the difference with the corresponding figure [ figdq1k ] concerning the pure case ) . to understand the origin of this singularity , we may take into account the infinite disorder nature of the critical point and the strong disorder nature of the critical region .
more precisely at large rg step @xmath155 , the renormalized coupling @xmath190 is either very small ( if @xmath191 ) or either very big ( if @xmath192 ) with the following leading contributions of the auxiliary function of eq . [ auxdqpurk ] @xmath193 let us thus make the approximation that the singular part is given in the critical region by @xmath194 since the singularity is due to the region of large rg - step @xmath155 , we may evaluate the first derivative by replacing the series over the integer @xmath155 by the corresponding integral over the real variable @xmath155 , and then perform a change of variable from @xmath155 to @xmath195 @xmath196 \label{dqkransingderi1}\end{aligned}\ ] ] is continuous at the critical point @xmath73 , but displays a cusp as a consequence of the second term containing the factor @xmath197 near criticality . for the random quantum ising chain , the strong disorder renormalization procedure has been shown to be asymptotically exact by daniel fisher @xcite so that it can be used to derive all universal critical properties ( see @xcite and the review @xcite ) , as well entanglement properties ( see @xcite and the review @xcite ) .
it is thus interesting to consider its application to the shannon - rnyi entropies , even if only the singularities will be meaningful , whereas the non - universal values dominated by the first rg steps will of course not be reproduced faithfully . in the strong disorder rg procedure @xcite , the strongest term @xmath198 among the ferromagnetic couplings @xmath199 and the transverse fields @xmath200 @xmath201
is decimated iteratively . if the strongest term is a coupling @xmath199 , the two spins linked by this coupling
are coupled ferromagnetically to form a new renormalized spin ( the two kept states are @xmath202 and @xmath203 ) .
so the contribution of this decimation to the shannon - rnyi entropies is actually zero @xmath204 if the strongest term is a transverse field @xmath200 , the corresponding spin is in the eigenstate of @xmath59 , so that the contribution of this decimation to the shannon - rnyi entropies is the maximal value @xmath205 @xmath206 as a consequence , the generalized fractal dimension is simply @xmath207 where @xmath208 is the total number of @xmath209-decimations along the rg flow , and @xmath210 is the total number of @xmath211-decimations along the rg flow .
we refer to refs @xcite for the renormalization rules and for the asymptotic probability distributions of the renormalized couplings and of the renormalized transverse fields as a function of the rg scale @xmath212 and of the usual control parameter @xmath213 @xcite related to the previous notations @xmath214 of eq .
[ lognormalnot ] @xmath215 + var[\ln h_i ] } = \frac { \ln k}{v } \label{delta}\end{aligned}\ ] ] here we only need the probability @xmath216 to decimate a coupling in the interval @xmath217 $ ] and the probability @xmath218 to decimate a transverse field in the interval @xmath217 $ ] , that read @xcite @xmath219 in particular , the density @xmath220 of surviving spins at scale @xmath221 evolves according to @xcite @xmath222 n_{\delta}(\gamma ) \label{dndelta}\end{aligned}\ ] ] and the solution starting from the initial condition @xmath223 at the initial scale @xmath224 reads @xcite @xmath225 the fraction of @xmath209-decimations is then given in the thermodynamic limit by @xmath226_{\gamma=\gamma_0}^{\gamma=+\infty } \label{fractionhdecim}\end{aligned}\ ] ] at criticality @xmath227 , the fraction of @xmath209-decimation is of course @xmath228 by symmetry . in the ordered phase @xmath229
, it is given by @xmath230 whereas in the disordered phase @xmath231 it reads @xmath232 in particular the derivative @xmath233 is continuous but displays a cusp at criticality @xmath227 , in agreement with the result [ dqkransingderi1 ] of the fernandez - pacheco renormalization .
in this paper , we have studied via real space renormalization the shannon and the rnyi entropies of the ground state wavefunction in the pure and in the random quantum ising chain .
our main conclusion is that the leading extensive term presents the following singularities at the quantum phase transition : the derivative with respect to the control parameter is logarithmically divergent in the pure case , and displays a cusp singularity in the random case .
this cusp singularity for the random case has been also confirmed via the strong disorder renormalization approach .
although the logarithmic divergence for the pure quantum chain is reminiscent of the logarithmic divergence of the specific heat of the 2d classical ising model , the cusp singularity found for the random quantum chain is different from the specific heat singularity of the corresponding 2d mccoy - wu model @xcite .
we should stress that the leading extensive terms of the shannon and the rnyi entropies do not depend on the boundary conditions . as other extensive thermodynamic observables , their values
are dominated by the small scales in the bulk and are thus non - universal , whereas their singularities at the critical point are dominated by the large scales and are thus universal .
the subleading non - extensive terms of the shannon and the rnyi entropies for finite chains of size @xmath46 are of course also very interesting . in the pure chain ,
these subleading contributions are universal , but depend upon the boundary conditions as follows : \(i ) for the periodic chain , the subleading term is of order @xmath234 and is non - trivial only at the critical point @xmath73 and for the shannon value @xmath16 @xcite , whereas the direct application of the real space rg described in the text would produce a non - trivial @xmath234 term only at the critical point @xmath73 , but for all value of the rnyi index @xmath30 .
this suggests , as already mentioned after eq .
[ dqcritipluq1 ] , that the rg procedure probably needs to be improved for @xmath146 , to take into account the deformation of the measure with respect to the initial quantum problem corresponding to @xmath16 .
\(ii ) for the open chain , the subleading term at criticality is logarithmic and @xmath30-dependent @xcite , as predicted from the presence of corners in the conformal field theory of the replicated associated statistical physics model
. it would be interesting to better understand the origin of these logarithmic terms directly in the framework of the quantum ising chain , and to reproduce them by some appropriate real - space rg procedure .
it is a pleasure to thank grgoire misguich and fabien alet for very useful discussions . | the shannon and the rnyi entropies of the ground state wavefunction in the pure and in the random quantum ising chain are studied via the self - dual fernandez - pacheco real - space renormalization procedure . in particular , we analyze the critical behavior of the leading extensive term at the quantum phase transition : the derivative with respect to the control parameter is found to be logarithmically divergent in the pure case , and to display a cusp singularity in the random case .
this cusp singularity for the random case is also derived via the strong disorder renormalization approach .
# 1#2 # 1#2 |
chemically peculiar ( cp ) stars are a- and b - type stars exhibiting peculiar chemical abundances and slow rotation . with effective temperatures ranging between 10 000 k and 15 000 k ,
the late b - type hgmn stars belong to the third group of cp stars defined by preston ( 1974 ) .
they are characterized spectroscopically by large overabundances of hg ( possibly up to 6 dex ; heacox 1979 , cowley et al .
2006 ) and mn ( possibly up to 3 dex ; aller 1970 ) .
overabundances greater than 2 dex of be , ga , y , pt , bi , xe , eu , gd , w and pb are frequently observed in these stars , while he , n , mg , al , ni and zn often are underabundant by more than 0.5 dex ( takada - hidai 1991 ) . the elements c , o , si , s , ca , v , cr and fe usually show only modest peculiarities , with abundances within 0.5 dex of their solar abundances ( classified as `` normal type '' elements by takada - hidai 1991 ) .
hgmn stars rotate slowly in comparison to `` normal '' late - b stars , with @xmath6 .
their average rotation velocity is estimated to be 29 km s@xmath7 ( abt et al .
these stars present a high rate of binarity : gerbaldi et al .
( 1985 ) estimate that 51% are members of binary systems .
most of these binary systems have relatively small separations , with 41% belonging to sb1 systems , while 59% belong to sb2 systems . some authors ( mathys & hubrig 1995 ; hubrig et al .
1999 ; hubrig & castelli 2001 ) have reported the presence of a weak magnetic fields in the atmospheres of some hgmn stars .
however , attempts to confirm these claims for other hgmn stars have been unsuccessful ( landstreet 1982 ; shorlin et al .
2002 ; wade et al .
2006 ) . in light of the lack of convincing evidence , we assume for the purposes of this study that hgmn stars are non - magnetic objects . because of their high effective temperatures , the hgmn stars should not host hydrogen convective zones in their envelopes , and their low he abundances may reduce the importance of he convection zones as well . indeed , adelman ( 1994 )
showed that hgmn stars have small microturbulent velocities .
atmospheric velocity fields are estimated to be less than 1 km s@xmath7 for stars with @xmath8 ranging between 10 200 k and 12 000 k ( landstreet 1998 ) . | using high resolution , high - s / n archival uves spectra , we have performed a detailed spectroscopic analysis of 4 chemically peculiar hgmn stars ( hd 71066 , hd 175640 , hd 178065 and hd 221507 ) . using spectrum synthesis , mean
photospheric chemical abundances are derived for 22 ions of 16 elements .
we find good agreement between our derived abundances and those published previously by other authors .
for the 5 elements that present a sufficient number of suitable lines , we have attempted to detect vertical chemical stratification by analyzing the dependence of derived abundance as a function of optical depth . for most elements and most stars
we find no evidence of chemical stratification with typical @xmath0 upper limits of @xmath1 dex per unit optical depth .
however , for mn in the atmosphere of hd 178065 we find convincing evidence of stratification .
modeling of the line profiles using a two - step model for the abundance of mn yields a local abundance varying approximately linearly by @xmath2 dex through the optical depth range log @xmath3 to 2.8 .
[ firstpage ] stars : abundances - stars : atmospheres - stars : chemically peculiar - stars : individual : hd 71066 ( @xmath4 vel ) , hd 175640 ( hr 7143 ) , hd 178065 ( hr 7245 ) , hd 221507 ( @xmath5 scl ) |
atomic laser cooling has become recently one of the most active research fields in atomic physics @xcite .
several techniques @xcite have been developed to cool free atoms up to fractions of the recoil temperature ( i.e. the temperature associated with the recoil of a single photon ) .
the sub - recoil laser - cooling methods are based on the crucial concept of dark states .
these states can not absorb the laser light , and therefore can not be excited , but however can receive population via incoherent pumping , i.e. via spontaneous emission , and therefore these states behave as trapping states .
the dark - state is a general concept , and can be obtained by several mechanisms , for example the destructive interference between the absorptions from the two ground states of a three level @xmath0-system , as in velocity selective coherent population trapping ( vscpt ) @xcite . on the other hand , laser cooling of trapped atoms to the ground state of a trap has been achieved for single trapped ions using sideband cooling @xcite in the lamb - dicke limit ( ldl ) @xcite .
this limit corresponds to the case in which the size of the ground state of the trap @xmath1 is very small compared with the wavelength of the applied laser @xmath2 .
more rigorously , ldl occurs when @xmath3 , where @xmath4 is the so - called lamb - dicke parameter defined as @xmath5 .
sideband cooling method only works properly in this limit , and also requires the levels of the trap to be well resolved , which is achieved in the so - called strong - confinement limit , i.e. when @xmath6 , where @xmath7 is the trap frequency and @xmath8 is the spontaneous emission rate ( the level width ) .
this last condition can be easily fulfilled by using a metastable excited state and quenching its population using a third fast - decaying level @xcite .
however , the problem appears in the regime beyond the ldl , in which the recoil of the spontaneously emitted photons provides a heating mechanism , and therefore new cooling strategies must be proposed .
the most effective techniques have been those originally designed to cool free atoms up to subrecoil temperatures , i.e. dark - state cooling techniques @xcite .
a recent paper @xcite , proposes a new cooling scheme for trapped atoms beyond the ldl which combines the ideas of sideband and raman @xcite cooling and dark - state cooling , to confine a single trapped atom in the ground state of the trap . in this series of papers ,
we study laser cooling techniques similar to that of @xcite which allow the confining of an atomic sample into just one state of the trapping potential , either the ground state or an excited state .
this first paper is devoted to the one - atom problem while the second one @xcite analyses the new physics appearing for the many - atom case .
the one - atom problem is interesting as such in the context of quantum computation with ions @xcite , and also in the production of non - classical states of motion of single atoms or ions @xcite . also the one - atom problem is a good starting point to understand clearly the cooling processes in the many - atom case , which we discuss in detail in @xcite
; therefore this paper is a good first step to analyse the possibilities to achieve the bose - einstein condensation @xcite with all - optical means .
the paper is organized as follows . in sec .
[ sec : model ] the cooling scheme is described , as well as the difficulties to cool beyond the ldl . in sec .
[ sec:1d ] we study the one - dimensional case , analysing the cooling into the ground state of the trap ( basically summarizing @xcite ) in sec . [ sec:1dg ] .
the one - dimensional cooling into an excited of the trap is studied in sec .
[ sec:1de ] , where the dark - state conditions are presented . in section sec .
[ sec:2dnew ] and sec .
[ sec:2dre ] the new physics which appears in the two - dimensional case is considered . in sec .
[ sec:2dint ] a new interference mechanism between the lasers in different directions is presented . in sec .
[ sec:2dg ] the cooling into the ground state is analysed , whereas in sec .
[ sec:2de ] , we employ the dark - state conditions of sec .
[ sec:1de ] and also the interferential dark - states of sec .
[ sec:2dint ] , to confine the atom in an excited two - dimensional state .
we conclude in sec .
[ sec : conclusions ] .
we consider laser cooling in a three - level @xmath9-system as that presented in fig .
[ fig:1 ] .
the level @xmath10 is the ground state , @xmath11 is a metastable state , and @xmath12 is a third fast - decaying auxiliary state .
a laser , or stricktly speaking two lasers excite coherently the resonant raman transition @xmath13 ( with some associated effective rabi frequency @xmath14 ) , while a repumping laser in or off - resonance with the transition @xmath15 pumps optically the atom into @xmath10 . with this three level scheme ,
one obtains an effective two - level system with an effective spontaneous emission rate @xmath8 , which can be controlled by varying the intensity or the detuning of the repumping laser @xcite .
the atom is assumed to be confined in an harmonic ( isotropic ) potential of frequency @xmath7 .
we denote the states of the harmonic oscilator as @xmath16 , or @xmath17 when treating the two - dimensional case .
we assume @xmath6 , i.e. we suppose the so - called strong - sonfinement limit , which is also called festina - lente limit , and turns to be a crucial point in the many atom case in order to avoid the reabsorption problem @xcite .
we also assume @xmath18 and @xmath19 , which allows to adiabatically eliminate the excited state @xmath20 in the expressions below . in the lamb - dicke limit ( ldl ) , i.e. when the dimensions of the trap are very small compared with the laser wavelength @xmath2 , it is possible to use standard sideband cooling techniques to pump the atom into the ground state of the harmonic trap . in this cooling mechanism
the laser is assumed red - detuned , with detuning @xmath21 .
then , the absorption of a laser photon induces the transition @xmath22 . in the ldl , @xmath23 , where @xmath24 is the recoil energy .
then , in the spontaneous emission the photon recoil is not large enough to produce a jump between the states of the harmonic trap ; therefore only the transition @xmath25 is possible .
hence , the global effect of a complete absorption - spontaneous emission cycle is a transition @xmath26 is a dark state .
the latter statement is only true if we operate in the regime @xmath6 . beyond the ldl , @xmath27 , and
hence in the spontaneous process , the atom undergoes transitions @xmath28 where @xmath29 ranges from @xmath30 to @xmath31 .
this process introduces a heating of the atomic distribution which prevents the confinement of the atom into the ground state of the trap . in this paper
we study different techniques which allow us to overcome this problem , recovering the results of usual sideband cooling , and even obtaining new physical effects .
the previous paper @xcite studied the problem of laser cooling of a single trapped atom beyond the ldl into the ground state of a one - dimensional trap . in this section
we briefly review that paper , and discuss a cooling mechanism which avoids the heating effects of the usual sideband cooling by employing the two step process : * in the first step the atoms are confined in the levels @xmath32 .
this is achieved by using pulses with detunings @xmath33 , where @xmath34 is the closest integer to @xmath35 .
the absorption of a laser photon produces a transition @xmath36 , and the spontaneous emission @xmath37 , where @xmath29 ranges from @xmath30 to @xmath38 . as a result ,
in average some energy is lost in each cycle .
obviously , the states with @xmath39 remain dark , and therefore the atoms are finally confined in these states . due to this
, we call these pulses confinement pulses .
* using pulses of selected frequencies one can empty the trap levels @xmath40 without affecting the population of @xmath41 .
then , in each cooling cycle there is some probability to pump into @xmath41 .
there is also some probability of heating , and therefore we must repeat the confinement step , and so on .
we call the pulses with selected frequency fine - cooling pulses . as the result of the cooling process
, the atom can be confined in the ground state of the trap , as in the usual sideband cooling .
since the present cooling mechanism is based on sequences of different pulses , we call this cooling method dynamical cooling . the dynamics of the system can be well described , after adiabatic elimination of the @xmath20 states , by the following rate equation : @xmath42 where @xmath43 is the probability of the atom to be in the level @xmath44 of the trap .
the transition rates are : @xmath45 where @xmath46 is the laser wavevector and @xmath47 is the angular dipole pattern of the spontaneously emitted photons . using closure relations , assuming a laser detuning @xmath48 with @xmath49 an integer value , and neglecting the non - resonant terms , it is possible to write the rates to empty the level @xmath50 ( called hereafter empty rates ) in a very simple form in terms of the franck - condon factors : @xmath51 monte carlo simulations of eq .
( [ rateeq1d ] ) yield similar results to those of @xcite . in fig .
[ fig:2]b the cooling dynamics for the case of @xmath52 , @xmath53 is shown , where the initial distribution was taken to be thermal with the mean @xmath54 .
we use a sequence of four pulses of detunings @xmath55 and durations @xmath56 to confine the atom with almost total probability in the ground state of the trap .
pulses @xmath57 and @xmath58 are confining pulses , and @xmath59 and @xmath60 are fine - cooling pulses .
the reason of applying two different confinement pulses slightly detuned one respect to the other , is the oscilatory character of the franck - condon factors , which produces , as observed in fig . [ fig:2]a , that the empty rates have quasi - zero minima for each detuning . by applying two sligthly detuned lasers this problem
is avoided because the states that one laser does not empty the other laser does , and vice versa .
the same reasoning explains the use of two fine - cooling pulses . in the present section we present a new mechanism that allows the confining of the atom into an excited state of the trap .
this cooling mechanism is based on the properties of the franck - condon factors appearing in eq .
( [ erates1d ] ) , which are of the form : @xmath61}}\frac{(-1)^{l}\eta^{2m+s-2l}}{l!(m+s)!(m+s - l)!}. \label{fkfactors}\end{aligned}\ ] ] for a particular excited state @xmath50 of the trap , one can prove that for certain values of the detuning and of the lamb - dicke parameter , the sum in the last line of eq .
( [ fkfactors ] ) cancels out . if this occurs , and following eq .
( [ erates1d ] ) , one can observe that this level @xmath50 is not emptied , and therefore becomes a dark state .
the dark - state conditions for the states @xmath62 and @xmath63 of the trap , are respectively : @xmath64 certainly , these conditions have some limitations .
first of all , as the detuning must be integer ( i.e. @xmath49 must be an integer number ) , only particular values of @xmath4 are valid in order to satisfy the dark - state conditions .
this is not a major problem because the trap frequency can be externally controlled .
however , one must note that for the case of one - atom problem these conditions are certainly rather restrictive .
dynamical cooling of a single atom is critically sensitive to the deviations from the dark - state conditions .
we show in @xcite that fortunately this is not the case in the many atom problem with bosonic atoms .
the limitations are overcome due to quantum statistical effects ( bosonic enhancement effect ) .
using the dark - state condition it is possible to design a cooling mechanism that allows the confinement of the atom into an excited state of the trap .
the cooling mechanism consists as in the ground - state cooling in a sequence of several pulses : * \1 .
confinement pulse : exactly the same as in the ground - state cooling .
the use of two sligthly detuned confinement pulses is also recommended .
dark - state cooling pulse : which fullfills the dark - state condition for a selected excited state .
auxiliary pulses that avoid the oscilatory character of the empty rates of the dark - state cooling pulse , but do not affect the excited state in which the confinement is to be produced . in figs .
[ fig:3 ] the case of @xmath52 , @xmath65 is analyzed .
we have used cycles of four pulses of detunings @xmath66 and durations @xmath56 to confine the atom with almost total probability in the first excited state of the trap .
pulses @xmath57 and @xmath58 are confinement pulses , the pulse @xmath59 fulfills eq.([dcond1 ] ) and is therefore the dark - state pulse for the level @xmath62 , and the pulse @xmath60 is an auxiliary pulse . in fig
. [ fig:3]a the empty rates corresponding to pulses @xmath59 and @xmath60 are depicted .
note that the both pulses do not empty the level @xmath62 , and that it is the only level not emptied by them .
thus , the level @xmath62 becomes a trapping state .
fig.[fig:3]b shows the dynamics of confinement in the state @xmath62 for the considered sequence of pulses , where the initial distribution was taken to be thermal with the mean @xmath67 .
the atom can also be confined in higher levels of the trap .
[ fig:4 ] shows the dynamics of the confinement into level @xmath63 , for the case of @xmath68 , @xmath65 , the initial thermal distribution with the mean @xmath69 , and cycles of four pulses of detunings @xmath70 and durations @xmath56 .
as previously , the pulses @xmath57 and @xmath58 are confinement pulses , the pulse @xmath59 fulfills eq .
( [ dcond2 ] ) and is therefore a dark - state pulse for the level @xmath71 , and the pulse @xmath60 is an auxiliary pulse .
in this section we consider the single atom cooling in two dimensions .
the two dimensional problem contains essentially all of the main new physical features that appear when one considers dimensions higher than one , namely : * the number of levels involved is larger ( squared for the two - dimensional case ) .
therefore the atom can perform a random walk in a larger phase space , which means that the cooling is neccesarly slowlier than in the one - dimensional case . *
cooling in one direction means heating in the other directions .
this fact makes neccesary the use of lasers in different directions .
therefore , we will consider two lasers propagating in directions @xmath72 and @xmath73 , providing an electric field + @xmath74 + where the parameter @xmath75 accounts for a possible difference of intensities or dephasing between both lasers . as we indicate below this parameter
turns to be important under certain conditions .
* confinement pulses are no longer pulses of detuning @xmath76 .
this fact can be easily understood if we consider the case in which the atom absorbs a photon in direction @xmath72 . assuming a detuning @xmath77 , after an absorption the atom makes a transition @xmath78 , and after the spontaneous emission @xmath79 and @xmath80 range from @xmath30 to @xmath38 .
therefore the two - dimensional quantum number @xmath81 changes into @xmath82 where @xmath29 ranges between @xmath83 and @xmath84 , and therefore an average heating is produced .
for the two - dimensional problem , the true confinement pulses are those with detuning @xmath85 ( in general for @xmath86 dimensions @xmath87 ) , as can be easily proved by using the same reasoning as above .
* new cooling mechanisms can be designed , as we show below .
the cooling dynamics is regulated by a set of rate equations as that of eq .
( [ rateeq1d ] ) .
the transition rates in the two - dimensional laser arrangement considered are of the form : @xmath88+i\gamma } \nonumber \\ & & \qquad \times \langle n_{x}|e^{ik \sin{\theta } \cos{\phi } x}|m_{x}\rangle \delta_{l_{x},m_{x } } \biggr | ^{2 } \label{rates2d}\end{aligned}\ ] ] where @xmath89 . proceeding as in the one - dimensional case , it is possible to obtain a simple form for the empty rates : @xmath90 let us consider the case in which @xmath91 , i.e. @xmath92 .
let us also assume that both lasers in directions @xmath72 and @xmath73 have the same intensity but are @xmath93-dephased , i.e. @xmath94 .
then , eq.([erates2d ] ) becomes : @xmath95 eq .
( [ erates2ds0 ] ) shows that in the case of an isotropic trap , all the states in which @xmath96 , satisfy @xmath97 , i.e. are not emptied .
i d est , the destructive interference of the absorption of the two lasers produces that the diagonal levels become dark . as we show in the following section , this fact can be employed to cool the atom into the ground state of the two - dimensional trap with more efficiency than the simple extension of the one - dimensional method to two dimensions . moreover , if we set @xmath98 where @xmath99 is a particular state of the two - dimensional trap , then from eq .
( [ erates2d ] ) it is easy to observe that this particular level becomes a dark state .
therefore , we can specifically design the factor @xmath75 to regulate the interference between the two lasers in such a way that a selected state @xmath100 becomes dark . as always
, this fact can be employed to design a cooling mechanism to pump the atom into the level @xmath100 .
an example of this selective cooling is discussed in section iv.e .
also one can use eq .
( [ am0n0 ] ) to generalize the dark - state condition appearing from eq .
( [ erates2ds0 ] ) for the more general case of an anisotropic trap . additionally , the dark - state condition for the excited states presented in sec .
[ sec:1dec ] , remains valid for the two - dimensional case , allowing the cooling into excited states of the trap , as we show in sec .
[ sec:2de ] . in this section
we demonstrate how the previously presented destructive interference between the lasers can be employed to improve the cooling into the ground state of the trap . as in one dimension ,
the cooling mechanism consists of several pulses : * \1 .
confinement pulses : with the same function as in the one - dimensional case but , instead of pulses with detuning @xmath101 , we use , following the reasonings of sec .
[ sec:2dnew ] , @xmath102 .
the use of two sligthly detuned confinement pulses is also needed .
dark - state cooling pulses : with detuning @xmath91 , and with ratio between the laser amplitudes @xmath94 . following the discussion of the previous section the diagonal states
@xmath96 are dark for these pulses .
sideband cooling pulses : of course the dark - state pulses induce trapping in the diagonal states , and not only in @xmath103 . in order to avoid that , a usual sideband cooling pulse is neccesary , i.e. a pulse with detuning @xmath104 . *
pseudo - confining pulses : as pointed out in sec iv.a .
the atom performs its random walk in a larger phase space , and therefore with the only use of pulses @xmath57 , @xmath59 and @xmath58 the cooling is neccesarly very slow . in order to improve the cooling time we use pulses with @xmath105 and pulses with @xmath106 ( closest integer to this value ) .
these pulses pseudo - confine the atom below @xmath107 and @xmath108 , respectively . due to this fact we call these pulses pseudo - confinement pulses . fig .
[ fig:5 ] shows the cooling dynamics for the case of @xmath52 , @xmath109 , and an initial thermal distribution with mean @xmath69 .
we have used cycles of eight pulses of detunings @xmath110 , all of them with duration @xmath111 . in solid - line
is represented the case of @xmath94 for all the pulses , whereas in dashed - line the case @xmath112 is depicted .
note that due to the form of eq .
( [ erates2d ] ) , only for @xmath91 it is relevant if @xmath112 or @xmath94 .
pulses @xmath57 and @xmath113 are confinement pulses .
pulse @xmath60 is the dark - state pulse
( for the case @xmath94 ) .
pulse @xmath114 is the sideband cooling pulse .
pulses @xmath59 , @xmath58 , @xmath115 and @xmath116 are pseudo - confinement pulses .
one can observe that for the case of @xmath94 the cooling is very much effective than for the case of @xmath112 .
therefore , the interference mechanism which appears in two dimensions ( and in general in dimensions higher than one ) between the absorption amplitudes of lasers in different directions can be employed to improve the cooling technique . in this section ,
we employ the two different kinds of dark - states , i.e. those produced by the special form of the franck - condon factors ( sec .
[ sec:1dec ] ) , and those produced by the interference of lasers in different directions ( sec .
[ sec:2dint ] ) , to confine the atom into an excited state of the harmonic trap . as in the previous section
the use of confinement and pseudo - confinement pulses is in this case also neccesary .
let us analyse in first place the cooling using the franck - condon properties .
[ fig:6 ] shows in solid line the dynamics of the probability to occupy the level @xmath117 of the trap , for the case of @xmath52 , @xmath118 , the initial thermal distribution with the mean @xmath69 , and a sequence of pulses of detunings : @xmath119 , all of them with duration @xmath111 . as previously , pulses @xmath57 and @xmath113 are confinement pulses , and pulses @xmath59 , @xmath58 , @xmath115 and @xmath116 are pseudo - confinement pulses .
pulse @xmath60 fullfills eq .
( [ dcond1 ] ) , and from eq .
( [ erates2d ] ) it is trivial to see that for an isotropic trap the level @xmath117 of the trap becomes a dark - state for this pulse . finally pulse @xmath114 avoid the formation of undesired unemptied states . one can observe that an almost complete confinement of the atom into the level @xmath117 can be obtained .
[ fig:6 ] shows in dashed line the dynamics of the probability to occupy the level @xmath120 of the trap , for the case of @xmath121 , @xmath118 and a sequence of pulses of detunings : @xmath122 , all of them with duration @xmath111 .
the pulses play the same role as in the previous case .
one can observe that the atom can be confined with large probability into the level @xmath120 .
finally , let us consider the cooling mechanism using the dark - states produced by the interference between the two lasers .
[ fig:7 ] shows the cooling dynamics for the case of @xmath52 , @xmath118 , the initial thermal distribution with the mean @xmath69 , and a sequence of pulses of detunings : @xmath123 , all of them with duration @xmath124 , except pulse @xmath60 which has a duration @xmath125 .
for all the pulses we consider the case of @xmath94 , except pulse @xmath60 for which we use @xmath126 , satisfying eq .
( [ am0n0 ] ) for the level @xmath127 . as previously , pulses @xmath57 and @xmath113 are confinement pulses , and pulses @xmath59 , @xmath58 , @xmath115 and @xmath116 are pseudo - confinement pulses .
pulse @xmath60 is a dark - state pulse for the level @xmath127 , and pulse @xmath114 avoids the formation of undesired unemptied states .
one can observe that the atom can be confined with large probability into the level @xmath127 .
we have shown that the destructive interference between the lasers in both directions allows the improvement of the cooling in the ground state of the harmonic trap , and also provides a new mechanism to cool the atoms in a selected excited state of the trap .
unfortunately this interference mechanism is not valid for large values of the lamb - dicke parameter @xmath4 .
the reason is the following .
the interferential mechanism is based on pulses with detuning @xmath91 , i.e. pulses whose resonant absorption produces a jump between one state of the trap and itself .
for sufficiently large @xmath4 the debye - waller factor @xmath128 , which appears in the franck - condon factors ( eq . ( [ fkfactors ] ) ) , becomes very small , and therefore the franck - condon factors connecting one state with itself are very small .
this fact implies that the empty rates practically vanish for all the states , i.e. all the states of the trap become dark , whatever the value of @xmath75 , and therefore the interference mechanism is no more valid .
the interference mechanism can be employed , roughly speaking , for @xmath4 not larger than 4 .
the cooling dynamics of a single atom trapped in one - dimensional and two - dimensional harmonic potentials has been analysed . using a particular sequence of confinement plus fine - cooling pulses the atom
can be cooled into the ground state of the one - dimensional trap . moreover , by using the special properties of the franck - condon factors , a selected excited state of the trap can be made dark , and then a cooling mechanism can be designed to cool into it .
we have determined the dark - state condition , and proved via monte carlo simulations that the atom can be confined with almost total probability in the selected excited state .
however , the formation of the dark - state is very sensititive to the deviations from the dark - state condition .
this condition is by far very much less restrictive for the case of many atoms in the trap as we show in @xcite . the difficulties to cool in higher dimensions han been analysed on studying the two - dimensional case .
in particular one needs a new laser arrangement with two orthogonal lasers and confinement pulses with twice the detuning of the one - dimensional confinement pulses .
the difference between the amplitudes of the orthogonal lasers acts as a new degree of freedom that can be used for lamb - dicke parameters @xmath129 to produce a destructive interference between the absorption amplitudes of both lasers .
this destructive interference can be used to create dark - states which , as we have shown numerically , allow the improvement of the efectivity of the cooling into the ground - state .
also this interference mechanism can be used to cool into selected excited states of the trap . for larger lamb - dicke parameters
@xmath130 the interference method is no longer valid . in that regime
, the only possibility seems to be the extension of the one - dimensional results of @xcite ( basically based on the use of blue - detuned lasers to empty the states different than the ground - state ) to higher dimensions , by using a two - laser arrangement , and confinement and pseudo - confinement pulses .
fortunately , the dark - state condition based on the properties of the franck - condon factors remains effective for large lamb - dicke parameters , allowing the excited - state cooling also in this regime .
* acknowledgements * partial support from the spanish direccin general de investigacin cientfica y tcnica ( grant no .
pb95 - 0955 ) and from the junta de castilla y len ( grant no sa 81/96 ) is acknowledged .
l.s . wants to thank m. l. for hospitality during his stay in the institut fr theoretische physik of hannover .
this work has been supported by the deutsche forschungsgemeinschaft under sfb 407 .
a. aspect , r. kaiser , n. vanteenkiste and c. i. westbrook , physica scripta * t58 * , 69 ( 1995 ) .
a. aspect , e. arimondo , r. kaiser , n. vanteenkiste and c. cohen - tannoudji phys .
* 61 * , 826 ( 1988 ) .
m. kasevich and s. chu , phys .
lett . * 69 * , 1741 ( 1992 ) .
f. dietrich , j. c. berquist , w. m. itano and d.j .
wineland , phys .
. lett . * 62 * , 403 ( 1989 ) ; c. monroe , d. m. meekhof , b. e. king , s. r. jefferts , w. m. itano , d. j. wineland and p. gould , phys .
rev . lett . * 75 * , 4011 ( 1995 ) .
s. stenholm , rev .
phys . * 59 * , 699 ( 1986 ) .
i. marzoli , j. i. cirac , r. blatt and p. zoller , phys .
a * 49 * , 2771 ( 1994 ) .
r. dum , p. marte , t. pellizari and p. zoller , phys .
lett . * 73 * , 2829 ( 1994 ) .
g. morigi , j. i. cirac , m. lewenstein and p. zoller , europhys .
lett * 39 * , 13 ( 1997 ) .
l. santos and m. lewenstein , ( in preparation ) .
j. i. cirac and p. zoller , phys .
lett . * 74 * , 4091 ( 1995 ) .
j. i. cirac , a. s. parkins , r. blatt and p. zoller , adv . at .
37 , 237 ( 1996 ) .
m. h. anderson , j. r. ensher , m. r. matthews , c. e. wieman and e. a. cornell , science * 269 * , 198 ( 1995 ) ; k.b .
davis , m. o. mewes , m. r. andrews , n. j. van drutten , d. s. durfee , d. m. kurn and w. ketterle , phys .
* 75 * , 3969 ( 1995 ) .
y. castin , j. i. cirac and m. lewenstein , phys .
* 80 * , 5305 ( 1998 ) ; see also j. i. cirac , m. lewenstein , and p. zoller , europhys .
lett . * 35 * , 647 ( 1996 ) . | we study the laser cooling of one atom in an harmonic trap beyond the lamb - dicke regime . by using sequences of laser pulses of different detunings
we show that the atom can be confined into just one state of the trap , either the ground state or an excited state of the harmonic potential .
the last can be achieved because under certain conditions an excited state becomes a dark state .
we study the problem in one and two dimensions . for the latter case a new cooling mechanism is possible , based on the destructive interference between the effects of laser fields in different directions , which allows the creation of variety of dark states . for both ,
one and two dimensional cases , monte carlo simulations of the cooling dynamics are presented . |
a quantum system , subjected to external driving , can experience resonant transitions between its energy levels .
conservation of total energy assumes absorption or emission of several photons of the driving field .
such multiphoton processes play an important role in atomic and molecular systems interacting with electromagnetic field @xcite .
for example , the multiphoton resonant spectroscopy is one of the methods to probe the structure of atoms and molecules @xcite .
this technique has the advantage of observing highly excited states by using relatively low frequencies .
the concept of another application , the multiphoton excitation microscopy , is based on the multiphoton excitation of the fluorescent dyes molecules @xcite .
this technique allows imaging biochemical objects with high spatial resolution .
recent development of fabrication and measurement techniques enabled a study of the wide spectrum of quantum phenomena in superconducting structures . during the past years
it has been clearly shown that specially designed macroscopic superconducting circuits , which include josephson junctions , behave quantum mechanically similar to a quantum particle in a potential well . under certain conditions ,
these objects demonstrate the coherent superposition between their macroscopically distinct quantum states .
it is important to note that this is a pure quantum effect which has no classical analogue and can be used for a number of intrigued applications .
if the circuit s dynamics can be described in the frame of the two - level approximation , such two - level quantum system is called a qubit . the advance in the study of different phenomena in superconducting qubits can be found in the reviews @xcite .
in general , superconducting josephson circuits can be described as multilevel quantum systems . by analogy
, such systems are called artificial atoms , while coupled qubits systems behave as artificial molecules .
an interesting problem is how phenomena , known from atomic physics , will appear for these artificial atoms and molecules .
note that the following features differ these mesoscopic - size quantum systems from their microscopic counterparts : a high level of controllability by electronic means , coupling to the macroscopic - size read - out devices , and unavoidable dissipative environment . for characterization and controlling the states of superconducting qubits
the one - photon spectroscopy was done by using relatively weak driving nakamura99 , friedman00 , vanderwal00 , vion02 , yu02 , martinis02 , chiorescu03 , born04 .
matching of the ground and higher states with the one - photon energy was exploited to probe the upper levels of the josephson - junction circuits @xcite . with increasing driving power ,
the multiphoton excitations were used to study the features of the artificial atoms both for the two - level dynamics @xcite and when the upper levels were involved @xcite . for strong
driving , the width of the resonance lines periodically tends to zero , which can be described as the destructive landau - zener - stckelberg interference @xcite .
respective interferograms displayed double - periodical dependence of the upper - level occupation probability on the energy bias and the driving amplitude @xcite .
two and more coupled qubits can be treated as artificial molecules .
being excited by a resonant microwave field , they display one - photon transitions @xcite .
alternatively , at smaller frequencies , the two - qubit systems can experience multiphoton transitions @xcite . in this article
we review the observations of the multiphoton transitions in single and coupled superconducting qubits probed by a classical resonator , and also we present the respective theory . having the purpose of presenting and describing specific results for the multiphoton transitions , our consideration is limited to the josephson - junction qubits .
we note however that similar phenomena can be studied in different quantum objects , which can be described as two- or multi - level systems , such as quantum wires and dots @xcite , nitrogen vacancy centers in diamond @xcite , ultracold atoms zenesini10 , plotz10 , zhang11 , nanomechanical and optomechanical setups @xcite , electronic spin systems , two - dimensional electron gas , and graphene @xcite .
the paper is organized as follows . in sec .
2 we use the method of an asymptotic expansion for the qubit - resonator system in order to obtain the resonator characteristics .
this formalism allows us to separate the dynamics of the relatively slow resonator and fast qubit .
then , in sec . 3 , we consider the multiphoton dynamics of an isolated two - level system .
later the formulas of those two sections will be applied for the description of the experimentally observed multiphoton excitations in single qubits ( sec . 4 ) and in coupled qubits systems ( sec . 5 ) .
for characterization of a quantum system different techniques can be applied .
one of the possible solutions is to use the so - called parametric transducer @xcite .
a key element in any parametric transducer is an optical or a radio - frequency auto - oscillator .
a transducer , coupled to the quantum system of interest , is constructed so , that quantum system dynamics causes a change of the phase or / and the amplitude of its oscillations .
a phase ( amplitude ) shift provides information about the dynamics of a quantum system . in particular ,
for probing the qubit s state , several types of oscillators have been already used : an @xmath0 tank circuit @xcite , a nanomechanical one @xcite , and a transmission line resonator @xcite .
if the resonator quantization energy @xmath1 is smaller than the thermal excitation energy @xmath2 , the resonator can be considered as a classical oscillator .
then the qubit - resonator system can be treated semiclassically : here a qubits quantum system is driven by a classical field and probed by a classical oscillator .
it is important to note that the similar approach is well known in quantum optics - many phenomena in the atom - light system can be described by making use of this semiclassical model @xcite . in this work we present the semiclassical description of some observed effects for the resonator - qubits systems .
we will not consider here the situation of coupling the qubits systems to a high - frequency resonator , which can be realized as a transmission line resonator .
the quantum properties of this qubit - resonator system are not described by the semiclassical model . for recent works in this field
see e.g. @xcite and references therein and also refs .
@xcite , where the multiphoton excitations were used to drive transitions between the multiple energy levels of the qubit - resonator system in the strong coupling regime .
another note here should be made about the term multiphoton processes . in the context of the semiclassical approach
, it relates to the energy of several photons which is absorbed or emitted by the quantum system . in the broader sense
the term multiphoton can relate to other processes employing the quantum nature of the electromagnetic field , see ref .
@xcite for a review of the non - classical phenomena in entangled multi - photon systems .
this section is devoted to the properties of the qubit - resonator system .
it will be shown that in the frame of the semiclassical approach the influence of the qubit on the resonator can be described by the renormalization of the oscillator constants .
for instance for a mechanical resonator it can be quantified by introducing the equivalent qubits - state - dependent elasticity coefficient and damping factor . in the case of inductive / capacitive coupling , the qubit s impact on the resonator can be described by introducing the qubits - state - dependent effective inductance / capacitance , while the losses can be described by the effective resistance . for concreteness , we will consider two realistic systems : the flux qubit inductively coupled to the tank circuit shevchenkopg08 and the charge qubit capacitively coupled to the nanomechanical resonator @xcite .
first let us consider the mechanical resonator as a spring with the elasticity @xmath3 , the damping factor @xmath4 ( which is assumed to be small ) , and loaded with mass @xmath5 , as shown in fig .
[ fig : effectivel&c](a ) . the oscillator has eigenfrequency @xmath6 and the quality factor @xmath7 .
its state is influenced by the qubit through the force @xmath8 and is driven by the probe periodical force @xmath9 . here
the small parameter @xmath10 is introduced explicitly to emphasize the small qubit - resonator coupling as well as the amplitude of the external harmonic force @xmath11 , which enables us to make use of the asymptotic expansion method .
the external nonlinear force is assumed to depend on the variable @xmath12 and its derivative only , @xmath13 . and @xmath14 , and with the energy difference @xmath15 .
the resonator is demonstrated as the spring oscillator with the elasticity coefficient @xmath3 .
as described in the main text , influence of the qubit on the resonator can be described by introducing the effective elasticity coefficient @xmath16 , which includes the qubits - state - dependent ( or , parametric , for brevity ) elasticity coefficient @xmath17 .
( b ) the flux qubit coupled via the mutual inductance @xmath18 to the @xmath0 resonator .
this can be described by introducing effective qubits - state - dependent inductance @xmath19 , which includes the parametric inductance @xmath20 in parallel to the tank s inductance @xmath21 .
( c ) the impact of the charge qubit on the nanomechanical resonator s state can be described by introducing the effective qubits - state - dependent capacitance @xmath22 , which includes the parametric capacitance @xmath23 in parallel to the resonator s capacitance @xmath24.,width=10 ] the displacement @xmath12 is the solution of the motion equation @xmath25the oscillations in the nonlinear system described by eq .
( [ eq4x ] ) can be reduced to oscillations in an _ equivalent linear system _ by making use of the krylov - bogolyubov technique of asymptotic expansion bogolyubov .
specifically , in the first - order approximation with respect to a qubit - resonator coupling parameter and close to the principal resonance , @xmath26 , the equivalent linear system is characterized by the effective amplitude - dependent elasticity coefficient @xmath27 and the effective damping factor @xmath28 ( see chapter 7 in ref .
@xcite):@xmath29@xmath30where @xmath31 . note that in eq .
( [ linearized ] ) both @xmath32 and @xmath33 are time - dependent values . here ,
in equations ( [ k_eff ] ) and ( [ lambda_eff ] ) , we have introduced the _ parametric _
elasticity coefficient @xmath17 and damping factor @xmath34 . in this context
the adjective _ quantum _ is sometimes used instead of parametric to emphasize that it is the qubit - state - dependent , i.e. it is defined by the quantum properties of the coupled system . in what follows , by simply changing the notations
we will see that the parametric elasticity coefficient gives either parametric inductance or parametric capacitance _
_ , _ _ when coupling is inductive or capacitive respectively , while the parametric damping factor will give us the parametric resistance
. note that in equations ( [ k_eff ] ) and ( lambda_eff ) the parametric terms @xmath17 and @xmath35 are of the first order in the small parameter of the problem @xmath10 .
this linearization procedure allows to obtain important information even without solving equations of motion .
in particular , the effective resonance frequency of the linearized system @xmath36 gives the expression for the frequency shift@xmath37 for physical interpretations it is important to emphasize that the application of the linearization technique resulted in the substitution of the nonlinear force by the linear one:@xmath38this latter parametric force describes the work done by the quantum system over the resonator ; the respective energy transfer during one period is the following@xmath39this , in dependence on the sign of the parametric damping factor @xmath40 , describes periodical extraction or pumping of the energy by the quantum system out of or into the resonator .
this is known as the _ sisyphus damping and amplification _ @xcite .
the solution of equation ( [ linearized ] ) in the first approximation in @xmath41 is given by the expression ( [ x ] ) with the amplitude @xmath42 and the phase shift @xmath43 slowly varying in time .
for these values the asymptotic expansion method gives the following system of equations ( see chapter 15 in ref .
@xcite)@xmath44 in the regime of stationary oscillations : @xmath45 , and we obtain equations for the amplitude @xmath32 and the phase shift @xmath46 , which can be written in the form@xmath47 in what follows it will be demonstrated that the phase shift @xmath46 and the amplitude @xmath32 can be directly observed experimentally , which gives the information about the quantum system through the values of the parametric elasticity coefficient @xmath17 and damping factor @xmath40 .
now we consider as an illustrative case the system of a flux qubit ( with geometrical inductance @xmath49 and average current @xmath50 ) coupled inductively to the @xmath48 tank circuit , as shown in fig .
[ fig : qb+lcr ] .
the approach , presented here , is the development of the theory in refs .
rifkin76 , greenberg02b , smirnov03 .
the quantum system is considered to be weakly coupled via a mutual inductance @xmath18 to the classical tank circuit .
the circuit consists of the inductor @xmath21 , capacitor @xmath24 , and the resistor @xmath51 connected , for the specification , in parallel .
the tank circuit is biased by the current @xmath52 , and the voltage on it @xmath53 can be measured . * * ( tank ) circuit . * * the flux qubit is pierced by the magnetic flux @xmath54 induced by the current in the controlling coil and by the current in the tank s inductor .
the qubit is coupled via the mutual inductance @xmath18 to the tank circuit .
the resonant tank circuit consists of the inductor @xmath55 , capacitor @xmath24 , and resistor @xmath51 ; the circuit is biased with an rf current @xmath52 .
the tank voltage @xmath56 is the measurable value.,width=264 ] the flux qubit can be described by the pseudospin hamiltonian @xcite@xmath57where the diagonal term @xmath58 is the energy bias , the off - diagonal term @xmath59 is the tunneling amplitude between the wells ( which corresponds to the definite directions of the current in the loop ) and @xmath60 are the pauli matrices . to obtain the equation for the tank circuit voltage , we write down the system of equations for the current in the three branches , namely , through the inductor ( @xmath61 ) , the capacitor ( @xmath62 ) , and the resistor ( @xmath63 ) ( in particular , for systems with superconducting elements see e.g. ref .
@xcite):@xmath64where @xmath65 is the flux through the inductor of the tank circuit .
this flux is the response of the quantum system to the flux , induced in it by the current @xmath61 .
it follows that the voltage @xmath66 in the current - biased tank circuit ( @xmath67 ) is described by the following nonlinear equation @xmath68 the external flux @xmath65 is assumed to be proportional to the coupling parameter @xmath69 and to depend on time via the voltage @xmath56 and its time derivative @xmath70 .
equation ( [ eq_for_v ] ) for the voltage @xmath56 coincides with the nonlinear equation ( [ eq4x ] ) for the variable @xmath12 with obvious change of the notations .
thus , the formalism presented in the previous subsection is directly applicable for the given problem .
specifically , in the first order approximation with respect to the coupling parameter @xmath71 and close to the principal resonance ( @xmath72 ) , the equivalent linear system is characterized by the effective resistance @xmath73 and inductance @xmath19 as following @xmath74 @xmath75 here @xmath76 is the quality factor of the unloaded tank circuit ( at @xmath77 ) and the _ parametric _ ( qubits - state dependent ) _ resistance _ @xmath78 _ _ _ _ and _ _ inductance _ _
@xmath20 are given by the formulas@xmath79where @xmath80 .
the resonant frequency @xmath81 becomes amplitude - dependent and is shifted by @xmath82 the phase shift @xmath46 and the amplitude @xmath32 depend on the probing frequency detuning @xmath83 and the qubit state ( via @xmath20 and @xmath78 ) . in the stationary regime
they are given by the solution of the system of equations @xmath84 which can also be rewritten alternatively in terms of the effective quality factor @xmath85 and effective frequency shift @xmath86 /\omega _ { 0}$ ] .
thus , the observable values the amplitude @xmath32 and the phase shift @xmath87 are defined by equations ( [ with_l_r_eff ] ) , which depend on the response of the measurable system , @xmath88 . as we discussed above , strictly speaking , the dynamics of the tank circuit has to be considered jointly with the dynamics of the qubit ( corresponding calculations see e. g. in @xcite ) . however , in what follows we consider two illustrative limiting cases , when the dynamics of the qubit can be treated separately from the dynamics of the tank circuit . for simplification
we introduce phenomenologically the relaxation time @xmath89 which is caused by the coupling to the environment and to the tank as well . _ _ 1 .
low - quality qubit ( _ _ @xmath90 _ ) : phase shift probes the parametric inductance of qubit . _ first case which allows to detach the equations for the qubit and resonator , is when all the qubit s characteristic times , and in particular the relaxation time @xmath89 , are smaller than the tank s period @xmath91 .
then the equations for the tank voltage can be averaged over the period of fast oscillations .
then the time derivative of the flux @xmath92 , induced by the qubit in the tank circuit can be described as @xmath93where @xmath94 is the flux in the qubit s loop , which consists of the time - independent part @xmath95 and of the flux , induced by the current @xmath61 in the tank s inductor .
this can be rewritten by introducing the effective inductance of the qubit , @xmath96 , and the characteristic inductance value @xmath97 .
then @xmath98 and for the tank voltage we have @xmath99 . in the first approximation in @xmath71 in the expression
@xmath100 we can insert @xmath61 found from this equation @xmath101 then from eqs .
( [ rq]-[lq ] ) we have @xmath102 ( hence @xmath103 ) and @xmath104where the qubit s effective inductance is defined by the total flux @xmath105 , piercing the qubit s loop@xmath106then for the phase shift @xmath46 and the voltage amplitude @xmath32
we obtain @xcite@xmath107which is the generalization of the result of ref .
@xcite for the case when the qubit can be in the superpositional state , which is taken into account here by the expectation value of the current @xmath50 .
if the bias current amplitude @xmath108 is small enough to be ignored in eq .
( [ jos_induct ] ) , then@xmath109at the resonant frequency @xmath110 , the phase shift @xmath46 is proportional to the inverse inductance of the qubit @xmath111 .
here it is worthwhile to emphasize the expression for the parametric inductance , which is expressed via the derivative of the expectation value of the current in the qubit s loop @xmath112,@xmath113 _ _ 2 .
higher - quality qubit ( _ _ @xmath114 _ ) : parametric resistance due to qubit s lagging .
_ another illustrative situation , where the qubit s dynamics can be considered separately from the resonator s one , is the case when the qubit relaxation time @xmath89 is of the same order as the tank s
period @xmath115 , namely , @xmath116 .
the qubit s response to the resonator probing signal can be phenomenologically described by introducing the lagging time @xmath117 , so that instead of eq .
( [ dfie ] ) we have@xmath118 in this way , the qubit s response depends on the current in the tank @xmath119 , which is given by @xmath120with @xmath121 and @xmath122 . for the small bias current eqs .
( [ reff]-[leff ] ) and ( fi_e_-[prime ] ) result in the following expressions for the parametric inductance and resistance @xmath123by analogy with eq .
( [ lq2 ] ) , the latter phenomenological equation can be rewritten in the form explicitly demonstrating its quantum character,@xmath124by making use of eq .
( [ en_trasfer ] ) , we obtain that the energy transferred from qubit into the resonator ( or , out of the resonator , for the opposite sign ) during one period is@xmath125 we emphasize here that both the parametric inductance and resistance in eq .
( [ lq&rq ] ) are proportional to the qubit s inductance @xmath126 .
then , one obtains equations for @xmath46 and @xmath32 , which are simplified in the first approximation in @xmath127 . in this case for the probing frequency equal to the resonant one , @xmath110 , the resulting formulas are@xmath128note that both the phase shift and amplitude are related to the qubit s effective inductance @xmath126 , which explains their similar behavior in experiment .
these equations are useful for the analysis of the experimental results , as it will be demonstrated in section 4 .
consider now the charge qubit capacitively coupled to a resonator . in this case , like in the one considered above , the resonator can be the tank circuit .
alternatively , the resonator can be a nanomechanical resonator ( nr ) , as in ref .
@xcite . for the illustrative purpose ,
we consider here this latter case .
the split - junction charge qubit ( shown in red in fig .
[ fig : qb+nr ] ) consists of a small island between two josephson junctions ( also called cooper - pair box ) , whose state is controlled by the magnetic flux @xmath105 and the gate voltage @xmath129 . here
@xmath130 is the dc voltage used to tune the energy levels of the qubit and @xmath131 is the microwave signal used to change the energy - level occupations .
the driven cooper - pair box is described in the two - level approximation by the hamiltonian in the charge representation , eqs .
( [ h1qb]-[e(t ) ] ) , where the tunnel splitting @xmath59 is equal to the josephson energy controlled by the magnetic flux @xmath105 : @xmath132 .
the charging energy and the driving amplitude are the following @xmath133 and @xmath134 , where the coulomb energy @xmath135 is defined by the total capacitance @xmath136 and the effective josephson capacitance is introduced @xmath137 , the dimensionless driving amplitude @xmath138 .
the dimensionless polarization charge @xmath139 is the fractional part of the respective polarization charges in two capacitances : @xmath140 and @xmath141 . and the gate voltage @xmath129 .
the resonator probing the qubit s state here is the nr , which is characterized by the displacement at the midpoint @xmath12 .
the voltage - biased nr is measured through its resonance frequency shift @xmath142 .
@xcite , width=8 ] the cooper - pair box here is formed by four capacitors , @xmath143 , @xmath144 , @xmath145 , and @xmath146 .
one of the plates of the latter capacitor is formed by the nr .
the displacement of the nr @xmath12 is much smaller than the distance @xmath147 between the plates .
then the capacitance between the nr and the qubit reads @xmath148 here @xmath146 stands for the capacitance value at the zero displacement .
the displacement of the nr influences the qubit through the changes in the polarization charge ; to make it significant , a large dc voltage @xmath149 is applied . on the other side ,
the nr is biased by dc and rf voltages @xmath150 and @xmath151 through the capacitance @xmath152 .
one of the approaches to describe the system qubit - resonator is to introduce the parametric capacitance as following ( for more details see ref .
shevchenko11 ) .
let us introduce the effective capacitance , as it is demonstrated in fig .
[ fig : effectivel&c](c ) , by differentiating the charge @xmath153 of the capacitor @xmath146 sillanpaa05 , duty05 , johansson06 : @xmath154 . then , for the charge @xmath155 with the island s voltage given by @xmath156 , we obtain @xmath157 , which consists of the _ _ parametric capacitance__@xmath158and the geometric capacitance @xmath159@xmath160where the approximations are valid for @xmath161 and @xmath162 respectively
. then one can consider the force @xmath163 , which acts on the nr from the left electrode , as the electrostatic force from the effective capacitance [ see fig .
[ fig : effectivel&c](c ) ] : @xmath164
. then the term with the parametric capacitance , in which @xmath165 , results in the following resonance frequency shift of the nr@xmath166 we would like to note that the results obtained for the system qubit - nr can be definitely extended to other systems . for example the charge qubit can be coupled to a tank circuit instead of a nr .
in contrast to the inductive coupling , considered in the previous subsection , here we mean capacitive coupling . then it is straightforward to obtain the expression for the measurable value , the tank circuit phase shift at resonance frequency , @xmath167 , @xcite@xmath168cf .
( [ a_and_v ] ) , where the phase shift probes the parametric inductance . in section 4
it will be demonstrated how these expressions can be used for the description of the realistic system .
application of the semiclassical theory , presented in the previous subsection , to the description of the qubits - resonator system makes possible to separate the slow dynamics of the resonator from the fast dynamics of the qubits system .
this allows to consider first the dynamics of a qubit or a system of qubits .
then , the resonator can monitor the state of the system of qubits . in this section
we will outline the description of the multiphoton processes in a qubit , while the presentation of the specific results is the subject of the next two sections .
initialization and manipulation of the qubit s systems require certain external signals .
the principal features of the driven system are captured for the harmonic driving , eq .
( [ e(t ) ] ) , to which we limit our consideration .
different theoretical approaches can be used for a driven two - level system , which is described in the books and reviews @xcite .
the choice of the formalism depends on the formulation of a problem and on the parameters of the system , such as the bias offset @xmath169 , driving amplitude @xmath170 and frequency @xmath171 .
the clear description can be given for the temporal dynamics in the so - called adiabatic - impulse model , where the driven evolution is considered adiabatic far from the avoided - level crossings with the impulse - type landau - zener transitions , when the energy distance is minimal @xcite .
as the result of this theory , the overall dynamics is described by the long - time rabi - type oscillations of the level occupation probabilities with the step - like features due to the landau - zener transitions .
another technique , which can be more convenient for the resonant driving , is the rotating wave approximation ( rwa ) @xcite .
it consists in neglecting the rapidly oscillating ( non - resonant ) terms .
the common approach for making use of this approximation is taking small driving amplitudes , @xmath172 .
then , the first - order consideration gives usual rabi oscillations of the level occupation probabilities close to the position of the one - photon resonance , where @xmath173 . in the @xmath174-th approximation
, the resonant excitation appears close to the parameters , where the energy of @xmath174 photons matches the qubit s energy distance @xcite@xmath175the time evolution is described by the multiphoton rabi oscillations shirley65 , while the time - averaged upper - level occupation probability has the lorentzian shape with the maximum equal to @xmath176 at the exact resonance defined by eq .
( [ multi ] ) . with increasing the driving amplitude the resonances shift @xcite from their positions given by the perturbation theory and defined by the exact multiphoton relation ( [ multi ] ) .
the first - order correction to the position of the resonances is the so - called bloch - siegert shift coh - tan ; it was demonstrated for the superconducting qubits in ref .
thus , in general , the position of the multiphoton resonances is amplitude - dependent . for the description of the strongly driven qubits , another formulation of the rwa can be used . there , the minimal energy level splitting @xmath59 is the small parameter , namely , it is assumed @xmath177 lopezcastillo92 , oliver05 , ashhab07 .
then the @xmath174-photon excitation appears close to the resonant parameters , given by the relation @xmath178 .
there , the upper - level occupation probability @xmath179 oscillates with the frequency @xmath180 with the renormalized splitting @xmath181 ; @xmath182 is the bessel function .
the time - averaged probability in the vicinity of the @xmath174-th resonance is given by@xmath183 being time averaged , the rabi oscillations are described by the lorentzian dependence of the upper - level occupation on the system s parameters ( the bias or the driving frequency ) @xcite . here arises an interesting and important problem of distinction of the respective quantum oscillations from their classical counterparts , which are the parametric resonances .
this was the subject of refs .
@xcite .
the most straightforward approach for the numerical description of the dynamics of a two - level system is the solution of the schrdinger equation @xcite .
then , the influence of the dissipation can be taken into account phenomenologically by introducing energy and phase relaxation times , @xmath89 and @xmath184 , and solving the respective bloch equation @xcite . instead , in the more general approach
, the dissipative environment can be described as an ensemble of oscillators , which would result in the bloch - redfield equation for the reduced density matrix leggett87 , goorden05 .
this latter formalism will be demonstrated in sec .
5 being applied to the specific case of the two - qubit system .
note that the multiphoton transitions can also be driven by the bichromatic field , when the energy level distance @xmath15 is matched by the energy of several photons of one ( say , microwave- ) frequency plus several photons of another ( say , radio- ) frequency .
such transitions were studied both in microscopic systems @xcite , and in the josephson - junction qubits @xcite . also for the case of a flux qubit it was demonstrated that the persistence of rabi oscillations can be supported by either the low - frequency signal @xcite or induced by noise @xcite .
let us get back to the qubit - resonator systems . in the previous section
we have discussed a modification of the qubit states ( and therefore its observables ) under different types of excitations .
a natural next step is to analyze the corresponding ( via qubits ) change of the resonator properties . in this section
we demonstrate this by presenting respective theoretical results for different realizations of the qubit - resonator systems , making use of the theory presented in the previous two sections .
the emphasis is made on demonstrating the consistency of the theoretical results with the experimental ones .
consider a qubit biased with a dc flux @xmath95 and driven with an ac flux @xmath185 , introducing @xmath186 and @xmath187 in order to get the effective inductance @xmath126 , as defined by eq .
( [ jos_induct ] ) , we have to calculate the average current in the qubit : @xmath188 , where @xmath189 is the current operator defined with the amplitude @xmath190 and the pauli matrix @xmath191 .
we calculate the reduced density matrix @xmath192 with the bloch equations @xcite which include phenomenological relaxation times , @xmath89 and @xmath184 .
it is convenient to express the density matrix in the energy representation : @xmath193 , where @xmath194 are the pauli matrices for this basis and @xmath195 stands for the unity matrix .
the value @xmath196 is equal to the difference between the populations of the ground and excited states .
let us find now the explicit expressions for the effective qubit s inductance for both the interferometer - type ( split - junction ) charge qubit @xcite and flux qubit @xcite .
for the interferometer - type charge qubit , as considered in detail in ref .
shnyrkov06 , the circulating current @xmath197 is flux - dependent and eqs .
( [ qimt ] ) show that there are two terms contributing in the tank circuit s phase shift,@xmath198 in a classical system ( where the current has a definite direction ) or in the ground state , the difference between the energy level s populations is constant , @xmath199 , and the second term in eq .
( [ ch_qb ] ) is zero .
in contrast , for the quantum system the interplay between these two terms is essential . at this point
it is worthwhile to notice that the second term can dominate at resonant excitation , as it was the case in the work shnyrkov06 ( see also below ) .
this means that the second ( quantum ) term can significantly increase the sensitivity of the impedance measurement technique , as compared to the classical situation described by the first term in eq .
( [ ch_qb ] ) .
consider now the case of a flux qubit .
the current operator is defined in the flux basis @xcite , @xmath189 , where @xmath200 stands for the amplitude value of the persistent current , and hence the value @xmath201 defines the difference between the probabilities of the clockwise and counter - clockwise current directions in the loop : @xmath202 . then with eqs .
( qimt ) we obtain@xmath203 in the energy representation we rewrite eq .
( [ flux_qb1])@xmath204here @xmath205 is the distance between the stationary energy levels . after the time - averaging over the driving period @xmath206
, this expression is written as following @xmath207 if a qubit is resonantly excited with the driving frequency @xmath171 , then the partial energy levels occupation probability @xmath208 has the lorentzian - shape dependence on @xmath209 .
it follows that the derivative @xmath210 takes the shape of a hyperbolic - like structure , i.e. it changes from a peak to a dip in the point of the resonance at @xmath211 . for the description of the measurement of a _ flux qubit in the thermal equilibrium
one has to put @xmath212 and @xmath213 in eq .
( [ tan(a)_full ] ) , @xmath214 the _ _
ground - state measurement _ _ at @xmath215 is described with @xmath212 and @xmath216 , which means replacing the hyperbolic tangent in eq .
( [ alfagr ] ) with the unity . the formula ( [ alfagr ] ) for the ground state obtained by differentiating the probability @xmath217 , eqs .
( [ flux_qb1]-[tan(a)_full ] ) , coincides with the earlier obtained results ( see eqs .
( 3 - 4 ) in ref .
the resulting tank phase shift is shown in fig .
[ fig : gr_st_vs_t ] for the following parameters taken from ref .
@xcite : @xmath218 ghz , @xmath219 ghz , @xmath220 mhz , @xmath221 , @xmath222 , @xmath223 , @xmath224 .
the accurate account of @xmath208 in eq .
( [ alfagr ] ) allows to describe both the suppression and widening of the zero - bias dip ( that is at @xmath225 ) as it was experimentally demonstrated in ref .
indeed , the suppression of the zero - bias dip ( at @xmath226 ) is described by the first term in eq .
( [ alfagr ] ) .
the widening is due to the second term that comes from differentiating the hyperbolic tangent ; this term becomes relevant for temperatures larger than @xmath59 , and results in the exponential rise of the width for @xmath227 , as demonstrated in the inset in fig .
[ fig : gr_st_vs_t ] . , when the qubit is thermally excited .
the curves are plotted for @xmath228 , @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 , and @xmath234 ghz .
left inset : corresponding experimental results grajcar04 .
right inset : temperature dependence of the width @xmath235 of the dip at half - depth in the phase shift , shown in the main panel .
@xcite , width=302 ] in ref .
@xcite the resonant excitation of the interferometer - type ( split - junction ) charge qubit was demonstrated experimentally and described theoretically . in accordance with the formula ( [ ch_qb ] ) one expects the resonances to appear differently when either first or the second term is dominated . to demonstrate this , in fig .
[ fig : shsh ] we plot the dependence of the tank circuit phase shift @xmath46 both as the function of the dimensionless bias voltage @xmath236 and of the dimensionless magnetic flux detuning @xmath209 . for the former case
the value @xmath226 was taken , where @xmath237 .
this results in disappearance of the second term in eq .
( [ ch_qb ] ) , and the resonant excitation of the qubit is visualized with the lorentzian peaks in fig .
[ fig : shsh](a , b ) . when the second term is dominant , the multiphoton transitions in the qubit result in the peak - and - dip structures in the dependence of the phase shift @xmath46 on the flux , fig .
fig : shsh(c , d ) . of the tank circuit
coupled to the charge qubit , calculated theoretically ( left ) and measured ( right ) .
panels ( a ) and ( b ) show the dependence on the gate voltage , while in ( c ) and ( d ) the dependence on the flux is demonstrated .
black and gray arrows in ( c ) demonstrate the positions of 1- and 2-photon resonant transitions , and the arrows in ( d ) mark 1- , 2- , and 3-photon excitations .
@xcite , width=321 ] theoretical fitting of the experimental graphs , as for example shown in fig .
[ fig : shsh ] , allows for defining the qubit s parameters , which is the _ multiphoton spectroscopy_. the parameters found were the following : the josephson energies for the two junctions @xmath238 ghz and @xmath239 ghz , the island s coulomb energy @xmath240 ghz ; the relaxation and decoherence rates @xmath241 and @xmath242 , which correspond to the following relaxation and decoherence times : @xmath243
ns and @xmath244 ns . figure [ fig : shsh ] also demonstrates how the position of the resonances depend on the driving frequency @xmath171 and how the multiphoton resonances appear with increasing the driving power @xmath245 .
namely , first , in fig . [ fig : shsh](a ,
b ) the varied parameter is the frequency @xmath246 , which from the bottom to top curves is @xmath247 , @xmath248 , @xmath249 , and @xmath250 ghz ; the driving power is the same for all figures @xmath251 and the flux was fixed at @xmath252 . and , second , in fig .
[ fig : shsh](c , d ) the curves correspond to the varied parameter driving power : in experiment being power of excitation ( from bottom to top : @xmath253 , @xmath254 , @xmath255 db ) and in theory being amplitude @xmath245 ( from bottom to top : @xmath256 , @xmath257 , @xmath258 ) ; the frequency there was fixed , @xmath259 ghz .
as we have seen in section 3 , both the tank voltage phase shift @xmath46 and amplitude @xmath32 can be used to monitor the resonant excitation of a superconducting qubit . in fig .
[ fig : shsh ] we demonstrated this with the observation of the phase shift @xmath46 of the tank circuit coupled to the charge qubit .
now we consider one- and multi - photon resonant excitations of a flux qubit , and the nonmonotonic dependence of the tank voltage amplitude @xmath260 will visualize the resonant transitions in the qubit . matched by the driving at frequencies shown by the numbers and the arrows of the respective length .
( b ) and ( c ) theoretically calculated and experimentally measured amplitude of the tank voltage @xmath32 versus flux detuning @xmath261 for different driving frequencies .
( the upper curves are shifted vertically . )
the one - photon excitations at @xmath262 , @xmath263 , and @xmath264 ghz , demonstrated in ( b ) and ( c ) , are explained by the arrows to the left in the energy diagram ( a ) , while the arrows to the right of the length @xmath265 ghz explain the multiphoton resonances in fig .
[ fig : multiph_1qb ] .
shevchenkopg08 , izmalkov08 , width=302 ] consider first the _ spectroscopical measurement _
, where the flux qubit is driven with the low - amplitude ac flux .
we expect resonant excitation of the qubit when the driving frequency matches the qubit s energy difference , @xmath266 . in the experimental case the positions of these resonances at a given driving frequency allow to determine the energy structure of the measured qubit izmalkov08 . in fig .
[ fig : volt_ampl](b , c ) we demonstrate the dependence of the tank voltage amplitude @xmath32 on the bias flux @xmath209 at @xmath267 for different driving frequencies : @xmath268 @xmath269 and @xmath270 ghz , which is explained by the energy diagram in fig : volt_ampl(a ) .
the results of the related experiment , ref .
izmalkov08 , are presented in fig .
[ fig : volt_ampl](c ) .
the parameters for calculations were taken as following : the tunneling amplitude @xmath271 ghz , the energy bias @xmath272 ghz , the temperature @xmath273 ghz , the relaxation rate @xmath274 ghz , the dephasing rate @xmath275 ghz , and the value which describes the coupling between the qubit and the tank circuit @xmath276 .
the curves were plotted for the driving amplitudes @xmath277 @xmath278 and @xmath279 from bottom to top .
the phenomenological lagging parameter was taken @xmath280 .
figure [ fig : volt_ampl ] demonstrates the effect described in section 3 : for @xmath281 both the phase shift @xmath46 and the amplitude @xmath32 depend on the qubit s inductance @xmath111 , which results in the alternation of peak and dip around the location of the resonances . in fig .
[ fig : multiph_1qb](a , b ) we present the calculated phase shift @xmath33 and the amplitude @xmath32 as functions of the probe current frequency @xmath282 and the flux detuning @xmath209 with the phenomenological lagging parameter @xmath283 for the _ strongly - driven flux qubit _ with the parameters being the same as for fig . [
fig : volt_ampl ] and with the values for the driving amplitude and frequency : @xmath284 and @xmath285 ghz .
the top panel presents theoretical calculations , which is in good agreement with the experimental observations , presented in the bottom panel , fig .
[ fig : multiph_1qb](c , d ) .
the dashed white line shows the tank resonance frequency @xmath286 mhz .
the positions of the multiphoton resonances is explained by the arrows to the right in the energy diagram , fig .
[ fig : volt_ampl](a ) , at @xmath287 with @xmath288 , @xmath232 , @xmath279 , and @xmath289 . ( a ) and the amplitude @xmath32 ( b ) on the bias current frequency @xmath290 and the flux detuning @xmath209 .
( c , d ) experimentally measured phase shift @xmath291 and the amplitude @xmath32 . shevchenkopg08,width=321 ] note that for the lagging parameter close to @xmath231 ( here @xmath280 ) the changes in the phase shift in fig .
[ fig : multiph_1qb](a ) are small at the resonance frequency ( along the dotted line at @xmath292 ) while the voltage amplitude in fig .
[ fig : multiph_1qb](b ) changes substantially , see formulas ( [ alpha&v ] ) . and this is actually demonstrated in fig . [
fig : volt_ampl](b , c ) .
such changes of the tank effective resistance or , equivalently , quality factor were studied in ref .
@xcite for the fully quantum - mechanical model of the qubit - resonator system .
we note that this can be alternatively described with the semiclassical model , presented here .
this model gives results consistent with the experimental ones , e.g. figs .
[ fig : volt_ampl ] and [ fig : multiph_1qb ] , which imply the energy transfer between the qubit and resonator according to eq .
( [ en_trasfer2 ] ) .
more details about this energy transfer , known as the _ sisyphus damping and amplification _ , can be found in refs .
@xcite , @xcite . then , in fig . [ fig : lzsi ] we present the dependence of the tank voltage phase shift @xmath46 on the microwave amplitude @xmath293 and the dc flux bias @xmath209 .
this double quasi - periodical dependence ( on both the energy bias and the driving amplitude ) is called the _ landau - zener - stckelberg interferogram _ @xcite .
the parameters were taken the same as for fig .
[ fig : volt_ampl ] and @xmath294 ghz . the left panel in fig .
[ fig : lzsi ] presents the theoretical interferogram from ref .
@xcite while the right panel is the experimental one , ref . @xcite . in fig .
[ fig : lzsi ] the multiphoton resonances at discrete dc bias @xmath209 ( which controls the distance between energy levels ) are clearly visible .
these resonances appear when the energy of @xmath174 photons matches the qubit s energy levels , @xmath295 .
the quasi - periodical character of the dependence on the ac flux amplitude @xmath296 is known as stckelberg oscillations .
the comparison of such graph to the experimental analogue allows the relation of the microwave power to the ac flux amplitude @xmath293 to be determined , which is the calibration of the power . for this , either the estimation of the period of stckelberg oscillations , shown by the black arrow , or adjusting the interference pattern slope , shown by the white line , can be used . and on the driving flux amplitude @xmath293 .
@xcite , title="fig:",width=226 ] and on the driving flux amplitude @xmath293 .
@xcite , title="fig:",width=226 ] the formalism developed in sec .
3.3 allows to describe the system of the driven qubit coupled to the nr . as it was demonstrated in ref .
shevchenko11 , two different approaches , called direct and inverse lzs interferometry , are of interest . in the direct interferometry
the qubit state is probed via the nr s frequency shift , as in ref .
@xcite , while in the inverse interferometry the impact of the nr s state on the qubit s hamiltonian is studied .
the direct lzs interferometry was calculated in ref .
@xcite as the resonator s frequency shift @xmath297 versus the energy bias @xmath298 and the driving amplitude @xmath299 .
the agreement with the experimental result of ref .
@xcite demonstrated that the semiclassical formalism is valid for a description of the measurable quantities . in ref .
@xcite it is also demonstrated how the analogous interferogram can be calculated for the qubit - tank circuit system in relation to the experiment of ref .
such a description allows to correctly find the position of the resonance peaks in the interferogram and to demonstrate the sign - changing behaviour of the parametric capacitance , which relates to the measurable quantities . for
the formulation of the inverse problem , let us consider the qubit s bias @xmath169 as a function of the nr s displacement @xmath12 . for small @xmath300 we have the expansion ( [ cnr(x ) ] ) , which results in the decomposition of the bias @xmath301 , where @xmath302 and @xmath303 .
the hamiltonian of the qubit ( [ h1qb ] ) with the parameter - dependent bias @xmath304 allows to consider the following problem .
let us assume that the qubit s state ( its wave function , upper level occupation probability , rabi frequency , etc . )
is known ( i.e. this is measured by a device , which we do not consider here for simplicity ) . given the known qubit s state , the aim is to find the hamiltonian s parameters .
particularly interesting is the parameter - dependent bias @xmath305 , which can give the information about the position and amplitude of the oscillations of the nr . and
now , in the general context , the reverse engineering problem in the spirit of refs .
[ ] can be studied , where one is interested in finding the driving hamiltonian for a given ( desired ) final state .
on the other hand , in ref .
@xcite the authors provide the basis for measuring the nr s position @xmath12 by means of probing the qubit s state , while @xmath306 is considered a slow time - dependent function . there ,
the emphasis was made on finding optimal driving and controlled offset ( @xmath307 ) parameters for the resolution of the small bias component @xmath308 .
it was assumed that the dynamics of the parameter @xmath309 is slow enough not to be considered during either certain period of the qubit s evolution or even during the setting the stationary qubit s state .
the aim was to find a sensitive probe for small @xmath310 .
as the ultimate sensitivity , the essential changes of the qubit s state for small changes of @xmath310 were required .
the problem , formulated in this way , was solved in ref .
@xcite for different illustrative driving regimes : one- , double- , and multiple - passage regimes .
the effective hamiltonian of the system of @xmath311 coupled flux qubits is @xmath312where @xmath313 is the coupling energy between qubits , and @xmath314 , @xmath315 are the pauli matrices in the basis @xmath316 of the current operator in the @xmath317-th qubit .
the current operator is given by : @xmath318 with @xmath319 the absolute value of the persistent current in the @xmath317-th qubit ; then the eigenstates of @xmath320 correspond to the clockwise ( @xmath321 ) and counterclockwise ( @xmath322 ) current in the @xmath317-th qubit .
the tunneling amplitudes @xmath323 are assumed to be constants .
the biases @xmath324 are controlled by the dimensionless magnetic fluxes @xmath325 through @xmath317-th qubit .
these fluxes consist of three components , @xmath326here @xmath327 is the adiabatically changing magnetic flux , experimentally applied by the coil and additional dc lines .
the second term describes the flux induced by the current @xmath61 in the tank coil , to which the @xmath328-th qubit is coupled with the mutual inductance @xmath329 .
and @xmath330 is the harmonic time - dependent component driving the qubit , typically applied by an on - chip microwave antenna . equation ( [ h2qbs ] ) can be reduced to the two - qubit system .
this system is shown in fig .
fig : scheme_2qbs .
w lines , as well as to an unavoidable dissipative environment .
the convenient model for description of the environment is the bath of harmonic oscillators .
the system of two coupled qubits is also assumed to be coupled to the measuring resonant circuit ( which is not shown here ) , as in fig .
fig : qb+lcr .
@xcite , width=264 ] to describe the two - qubit system , it is convenient to present the density matrix in the following form @xmath331which was shown to be suitable for both the definition and the calculation of the entanglement and other characteristics in multi - qubit system , e.g. @xcite . here
@xmath332 and @xmath333 ; the summation over twice repeating indices is assumed .
the two vectors @xmath334 and @xmath335 , so - called coherence vectors or bloch vectors , determine the properties of the individual qubits , while the tensor @xmath336 ( the correlation tensor ) accounts for the correlations .
the important characteristic of the state of the coupled - qubits system is its entanglement .
there are different approaches to the quantification of the entanglement @xcite .
one of the often used possibilities is the so - called concurrence @xcite .
another convenient for calculations approach is to introduce the measure of entanglement as following schlienz95 @xmath337this entanglement measure fulfills certain requirements , in particular , @xmath338 for any product state and @xmath339 for any pure state with vanishing bloch vectors @xmath334 and @xmath335 , corresponding to maximum entangled states . to describe dynamics of the density matrix
we will first disregard the relaxation processes .
this can be described by the liouville equation , @xmath340 $ ] , which is generally speaking a complex equation . to deal with the liouville equation , it is convenient to use the parametrization of the density matrix as described by eq .
( dens_mat ) . due to the hermiticity and normalization of the density matrix ,
@xmath341 are real numbers and @xmath342 . then the liouville equation can be written in the form of the system of @xmath343 equations for @xmath344 @xcite @xmath345where @xmath346 and @xmath347 is the levi - civita symbol .
consider now the measurable value , which is the resonator s phase shift . as we discussed in sec .
2 , it relates to the effective inductance of qubits system .
the formula obtained for single qubits can be generalized for the two - qubit system @xcite .
then for the case of low - quality qubits , when their characteristic times are smaller than the tank s period , at the resonance frequency ( @xmath110 ) , expression for the phase shift @xmath46 in terms of the parametric inductances @xmath348 can be written as following@xmath349 in what follows this expression will be used to calculate the phase shift @xmath33 , which maps the qubits state . in sec .
4 we have considered how the measurements of the single qubits allow to determine their parameters : the tunneling amplitudes @xmath350and the persistent currents @xmath190 .
it was demonstrated @xcite that for defining the parameters of single and multiple - qubits systems both the ground - state measurements and excited - state spectroscopy can be used ; the consistency of the results of the two approaches was shown .
now we will demonstrate this for the case of the system of two coupled flux qubits described by the hamiltonian ( [ h2qbs ] ) .
first , the one - qubit parameters are defined .
for this , suppose qubit @xmath351 is the one biased far from its degeneracy point in such a way that @xmath352 is large in comparison with the other energy variables .
then , qubit @xmath351 has a well defined ground state with averaged spin variables @xmath353 and @xmath354 which can be averaged out of the two - qubit hamiltonian ( [ h2qbs ] ) reducing it to : @xmath355 .
apart from the offset in the bias term due to the coupling , this is identical to the single qubit hamiltonian .
this offset can be easily compensated and measured allowing the determination of the coupling energy @xmath356 @xcite .
the qubit parameters , @xmath357 and @xmath358 , are determined from either the ground - state measurement or the excited - state spectroscopy , as it is described in sec .
analogously , biasing qubit @xmath359 far from the degeneracy point the parameters for qubit @xmath351 , @xmath360 and @xmath361 , can be determined .
next , the coupling energy @xmath356 was determined from the offset of the qubit dips from the @xmath362 lines , visible in the pure ground - state measurements presented in fig .
fig : spectroscopy_2qbs(a ) . on the flux biases
@xmath363 and @xmath364 : ( a ) ground - state measurement ( without microwave excitation ) .
( b , c , d ) with weak microwave excitation at the driving frequencies @xmath365 @xmath366 , @xmath367 and @xmath368 ghz .
@xcite , width=302 ] then the qubits were driven by magnetic fluxes @xmath369 with weak driving amplitudes and various driving frequencies . there , we expect the position of the resonant transitions from energy level @xmath370 to an overlying level @xmath371 determined by the one - photon relation : @xmath372 , which appears when the distance between the energy levels @xmath373 is matched by the photon energy @xmath374 . in fig .
fig : spectroscopy_2qbs(b ) a frequency in - between both qubit gaps ( @xmath375 ) was used and therefore only the transitions to the first excited state are visible . for higher frequencies , also the second and third excited states become visible as can be seen in subfigures ( c ) and ( d ) .
the theoretically calculated contour lines are superposed in fig .
[ fig : spectroscopy_2qbs](b - d ) for three different frequencies for which the condition , @xmath376 , is fulfilled ; the energy levels @xmath377 were found by diagonilizing the hamiltonian . from the fitting procedure
the following parameters were found : the tunneling amplitudes @xmath378 ghz , the energy biases @xmath379 ghz [ @xmath380 na ] , the inter - qubit coupling @xmath381 ghz , and the value which describes the coupling between the qubits and the tank circuit @xmath382 , where @xmath383 .
we now consider the multiphoton excitations of a system of two strongly driven coupled flux qubits .
we will describe the effects of resonant excitation in the system in terms of its energy structure , entanglement measure , and the observable tank circuit phase shift .
then we will present results for the multiphoton excitation of two types : direct ( when multiple photon energy @xmath384 matches the energy level difference @xmath385 ) and ladder - type ( when the transition happens via an intermediate level ) .
we will demonstrate how this can be used for creating the inverse population in the dissipative two - qubit system . to describe the system of two qubits subjected to the strong driving
, the following values were calculated : the energy levels ( by diagonalizing the stationary hamiltonian ) , the density matrix @xmath192 ( by solving the liouville equation ) , the observable tank circuit phase shift @xmath46 ( which is defined with the effective inductance of the qubits ) , and the entanglement measure @xmath386 by making use of eqs .
( [ e]-alpha_4_2 ) . in this way graphs in fig .
[ fig : multiph_2qbs ] were calculated for the set of parameters of the two - qubit system realized in ref .
@xcite : @xmath387 ghz , @xmath388 ghz , @xmath389 @xmath390 ghz , @xmath391 ghz , @xmath392 , and the driving frequency was taken @xmath393 ghz ; also the change of the dc flux here was assumed symmetrical : @xmath394 . for simplicity here
the relaxation processes were ignored ( and we will pay special attention to this below ) and we consider the case when the characteristic measurement time @xmath395 is larger than the characteristic times of the dynamics of the qubit
. then the tank circuit actually probes the incoherent mixture of qubit s states and the time - averaged values of phase shift and entanglement should be considered .
when the energy of @xmath174 photons ( @xmath384 ) matches the energy difference between any two levels @xmath396 and @xmath371 , the resonant excitation to the upper level is expected .
respectively , the arrows of the length @xmath289 , @xmath234 , and @xmath397 ghz show the places of possible one- , two- , and three - photon excitations . the time - averaged total probability of the currents in two qubits to flow clockwise , @xmath398 ,
is shown in fig .
[ fig : multiph_2qbs](b ) to experience resonant excitation .
the resonances appear as peak - and - dip structures in the phase shift dependence in fig .
[ fig : multiph_2qbs](c ) .
the time - averaged entanglement measure @xmath399 in a resonance increases due to the formation of the superposition of states , fig .
[ fig : multiph_2qbs](d ) ; this provides a method to control and probe the entanglement .
calculated and plotted as functions of the bias @xmath209 are four energy levels ( a ) , total probability of the currents in two qubits to flow clockwise @xmath208 ( b ) , the tank circuit voltage phase shift @xmath400 ( c ) , the entanglement measure @xmath386 ( d ) .
shevchenkopg08,width=321 ] the experimental study of the strongly driven system of two coupled flux qubits is presented in fig .
[ fig : multiphoton - multilevel ] .
the left panel is the measured voltage amplitude of the tank as a function of qubit biases @xmath401 and @xmath402 .
the driving frequencies from top to bottom were @xmath403 , @xmath404 , and @xmath405 ghz .
the multiphoton resonances at @xmath406 are visualized with the ridge - trough lines .
we note that the resonance ridge - trough lines are disturbed with increasing or decreasing the signal ; some of these changes are shown with white circles .
this means changing the effective josephson inductance in these points .
the experimental results can be clearly understood by comparing them with the energy contour lines , calculated by diagonalizing hamiltonian ( [ h2qbs ] ) and presented in the right panel of the figure .
there , numbers @xmath407 next to the lines mean that the line relates to the energy difference @xmath408 .
consider now these multiphoton features in more details . in fig .
fig : multiphoton - multilevel(b ) the black and red lines show the positions of the expected resonant excitations from the ground state to the first and to the second excited states respectively ; the blue and orange lines are the contour lines for the possible excitations from the first and from the second excited state to the third excited state . in fig .
fig : levels_and_populations_2qbs(a ) the energy levels are plotted at the fixed value of the bias flux through qubit @xmath351 , @xmath409 , as a function of the bias flux through qubit @xmath359 , @xmath402 .
the arrows are introduced to match the energy levels with the driving frequencies @xmath410 ghz and @xmath404 ghz .
the black and red arrows in both fig . [ fig : levels_and_populations_2qbs](a ) and fig .
fig : multiphoton - multilevel(b ) show the position of one - photon transitions to the first and the second excited levels .
the double green and blue arrows in fig .
[ fig : multiphoton - multilevel ] show the position of the two - photon processes , where the excitation by the first photon creates the population of the first and the second levels and the second photon excites the system to the upper level .
these two - photon excitations happen via intermediate levels ; compare the position of these expected resonances in fig .
fig : multiphoton - multilevel(b ) shown with the blue circle and green square . the orange triangle in fig .
[ fig : multiphoton - multilevel](b ) points the ladder - type three - photon excitation , with one photon to the first excited level and then with two photons to the upper level .
analogous considerations allow to see in fig .
fig : multiphoton - multilevel(c ) and ( d ) one- and two - photon resonant excitations to the first excited level for the driving frequency @xmath411 ghz .
the two - photon resonant excitation is direct and happen without any intermediate level .
the higher level excitations via the first excited state appear due to three- and four - photon excitations , as shown with orange triangles and pink asterisk . in fig .
fig : multiphoton - multilevel(e ) the response of the two - qubit system at @xmath412 ghz exhibits 1- to 4-photon excitations to the first excited state , which can be recognized by comparing with the black lines in fig .
[ fig : multiphoton - multilevel](f ) .
numerous upper level excitations via the first excited level appear as the changes of the signal along these lines .
the transition rates can be quantified by the absolute value of the matrix element of the perturbation between the states @xmath413 and @xmath414@xmath415divided by the factor @xmath416 .
the transition matrix elements in fig .
fig : levels_and_populations_2qbs(b ) explain the ladder - type excitations in fig .
[ fig : multiphoton - multilevel](b ) .
two points , marked by the vertical dashed green and blue lines in fig . [ fig : levels_and_populations_2qbs ] describe respectively two interesting situations . to the right ( see along the blue line )
the transition element between the higher two levels ( @xmath417 and @xmath418 ) is smaller than between the lower two levels ( @xmath419 and @xmath420 ) , @xmath421 .
in contrast , to the left ( see along the green line ) the transition element between the higher two levels ( @xmath422 and @xmath418 ) is larger than between the lower two levels ( @xmath419 and @xmath422 ) , @xmath423 . in both cases
the probability of the direct excitation to the highest level is very small , which means that the transitions are induced due to the ladder - type mechanism .
( at @xmath424 ) : the energy levels ( a ) , transition matrix elements @xmath425 ( b ) , the occupation probabilities @xmath426 ( c )
. ilichev10,width=226 ] the ladder - type transitions and the population inversion can be also illustrated by calculating the energy level occupation probabilities by solving the bloch - redfield equation ( see the next subsection for more details ) ; figure [ fig : levels_and_populations_2qbs](c ) was calculated with the driving frequency @xmath403 ghz and amplitude @xmath427 .
first , the ladder - type resonant excitation takes place to the left , where the upper level occupation probability @xmath428 is of the same order as the intermediate level occupation probability @xmath429 .
second , the inverse population _ _ _ _ appears to the right , where the upper level occupation probability @xmath429 is larger than the ground state probability @xmath430 , see also refs .
@xcite for the study of the population inversion in the systems with single josephson - junction qubits .
these two phenomena are similar to those which exhibit atoms in the laser field @xcite .
furthermore , the expectation value of the current in @xmath317-th qubit is calculated with the reduced density matrix : @xmath431 .
the results of the calculations are also presented as the color insets in fig .
[ fig : multiphoton - multilevel](f ) for the following parameters : the strength of dissipation @xmath432 and the driving amplitude @xmath433 .
consider now the influence of the dissipation on the dynamics of a two - qubit system .
for this the bloch - redfield formalism will be used .
the strong dependence of the inter - level relaxation rates on the controlling magnetic fluxes will be demonstrated for the realistic system .
this allows to propose several mechanisms for lasing in this four - level system @xcite . for identification of the level structure and understanding different transition rates
it is instructive to start from considering the case of two non - interacting qubits , that is @xmath434 . in this simplified situation , the energy levels of the system of two qubits consist of the pair - wise summation of single - qubit levels , @xmath435 in fig .
[ fig : levels_2qbs](a ) the energy levels are plotted as a function of the partial bias in the second qubit @xmath364 , fixing the bias in the first qubit @xmath436 .
then the single - qubit energy levels appear as the horizontal lines at @xmath437 and as the parabolas at @xmath438 . for the lasing the hierarchy of the relaxation times
is required . for this
it is natural to assume that the relaxation in the first qubit is much faster than in the second qubit .
this allows to consider three- and four - level lasing schemes in fig .
[ fig : levels_2qbs](b , c ) . .
( a ) one - qubit and two - qubit energy levels as functions of the magnetic flux @xmath439 at fixed flux @xmath436 .
the arrows show the fastest relaxation , which is assumed to relate to the qubit @xmath351 .
( b ) and ( c ) schemes for three- and four - level lasing at @xmath440 and @xmath441 . the driving magnetic flux pumps ( p ) the upper level ; fast relaxation ( r ) creates the population inversion ; the two operating levels can be used for lasing ( l ) .
@xcite , width=234 ] as a next step , the interaction of the qubits , @xmath442 , should be considered . to describe the relaxation in this system ,
the operators are converted to the basis of eigenstates of the unperturbed hamiltonian . in this representation @xmath443
is the diagonal matrix ; the unitary matrix @xmath283 consists of eigenvectors of the unperturbed hamiltonian ; the excitation operator @xmath444 is converted as following @xmath445 the dissipative environment can be described as the thermostat , for which the convenient model is the bath of harmonic oscillators , see fig .
fig : scheme_2qbs . within the bloch - redfield formalism , the liouville equation for the quantum system interacting with the bath
is transformed into the master equation for the reduced system s density matrix @xmath446 .
then the master equation for the density matrix of our driven system can be written in the energy representation as following @xcite @xmath447 _ { ij}+\delta _ { ij}\sum_{n\neq j}\rho _ { nn}w_{jn}-\gamma _ { ij}\rho _ { ij } , \label{m_eqn}\]]where @xmath448 , and the relaxation rates @xmath449re@xmath450 and @xmath451are defined by the relaxation tensor @xmath452 , which is given by the fermi golden rule . as it was shown in refs .
@xcite , the noise from the electromagnetic circuitry can be described in terms of the impedance @xmath453 from a bath of @xmath0 oscillators , described by the hamiltonian of interaction @xmath454 in terms of the collective bath coordinate @xmath455 . here
@xmath456 stands for the magnetic flux in the @xmath174-th oscillator , which is coupled with the strength @xmath457 to the qubits .
it follows that the relaxation tensor @xmath452 is defined by the noise correlation function @xmath458@xmath459the correlator @xmath458 was calculated in refs .
@xcite within the spin - boson model and it was shown that the relevant real part of the relaxation tensor @xmath460 \label{reg}\]]is defined by the environmental ohmic spectral density @xmath461 and is cut off at some large value @xmath462 , where @xmath463 is a parameter that describes the strength of the dissipative effects . from the above equations the expression for the relaxation rates from level @xmath464 to level @xmath465 follows@xmath466 .
\label{wmn}\]]in ref .
@xcite these relaxation rates were calculated as functions of the partial flux biases @xmath436 and @xmath364 .
it was demonstrated that the fastest transitions are those between the energy levels corresponding to changing the state of the first qubit and leaving the same state of the second qubit .
such a difference in the relaxation rates creates a sort of _ artificial selection rules _ for the transitions , similar to the selection rules studied e.g. in refs .
liu05 , degroot10 , niemczyk11 . to describe the hierarchy of the relaxation rates ,
consider them in the simplified case , ignoring the interaction between the qubits ; then the single - qubit relaxation rates follow from eqs .
( [ wmn ] ) and ( [ gmn ] ) @xcite@xmath467 in particular , in the vicinity of the point @xmath468 in fig .
[ fig : levels_2qbs](a ) , where @xmath469 , we obtain @xmath470 .
if @xmath471 is chosen , consequently the first qubit relaxes much faster .
after the parametrization of the density matrix , @xmath472 , the system s dynamics is described by the equations @xcite [ eqs ] @xmath473 _ { ii}+\sum_{r\neq i}w_{ir}x_{rr}-w_{ii}x_{ii},\text { } i=1,2,3 ; \\ \dot{x}_{ij}=\omega _ { ij}y_{ij}-\frac{1}{\hbar } \left [ v^{\prime } , y\right ]
_ { ij}-\gamma _ { ij}x_{ij},\text { } i > j ; \\ \dot{y}_{ij}=-\omega _ { ij}x_{ij}+\frac{1}{\hbar } \left [ v^{\prime } , y\right ] _ { ij}-\gamma _ { ij}y_{ij},\text { } i > j;\end{gathered}\]]@xmath474 , @xmath475 ; @xmath476 , @xmath477 . when discussing fig .
[ fig : levels_2qbs ] we pointed out that in the system of two coupled qubits there are two ways to realize lasing , making use of the three or four levels to create the population inversion between the operating levels . in ref .
@xcite the lasing in the two - qubit system was demonstrated by solving numerically the bloch - type equations ( eqs ) . besides demonstrating the population inversion between the operating levels , an additional signal with the frequency matching the distance between the operating levels was applied , to stimulate the transition from the upper operating level to the lower one .
so , the driving was considered to be , first , the monochromatic signal @xmath478 to pump the system to the upper level and to demonstrate the population inversion .
then another signal stimulating transitions between the operating laser levels is applied @xmath479 with @xmath480 . solving the system of equations ( [ eqs ] ) ,
one obtains the population of @xmath481-th level of our two - qubit system , @xmath482 .
the results of the calculations are presented in fig . [ 18 ] , where the temporal dynamics of the level populations is given for two situations . and @xmath483 ;
then the stimulating signal @xmath484 is turned on .
@xcite , width=264 ] as shown in the inset schemes in fig .
[ 18 ] , the fastest ( dominating ) relaxation transitions are @xmath485 and @xmath486 .
the system is excited by either one- or two - photon transitions , with @xmath487 in fig . [ 18](a ) or with @xmath488 in fig . [ 18](b ) .
this creates the population inversion between the levels @xmath489and @xmath483 .
note that analogous competition of the driving and relaxation can lead to the population inversion in other multi - level systems @xcite .
fast relaxation , @xmath486 , helps creating the population inversion between the laser levels @xmath490and @xmath483 , which is the advantage of the four - level scheme @xcite .
then the transition @xmath491 is stimulated by another signal with a frequency matching the laser operating levels ( @xmath492 ) .
figure [ 18 ] was calculated for the following realistic parameters @xcite : @xmath493 ghz , @xmath494 ghz , @xmath495 ghz , @xmath496 ghz , @xmath381 ghz , @xmath497 ghz ; and also @xmath498 ghz , @xmath499 with the driving frequency @xmath500 ghz for ( a ) and @xmath501 ghz for ( b ) . for the realization of such lasing schemes , the system of two qubits should be put in a quantum resonator , e.g. by coupling to a transmission line resonator , as in ref . .
then the stimulated transition between the operating states , demonstrated in fig . [ 18 ] , will result in transmitting the energy from the qubits to the resonator as photons .
here we presented the experimental and theoretical results of the study of driven single and coupled superconducting qubits .
the multiphoton transitions in both charge and flux qubits were studied in details .
those processes are important for both demonstrating the fundamental quantum phenomena in mesoscopic systems and for developing controlling mechanisms for perspective devices .
the system of qubits , coupled to the controlling electronics and measuring resonator , can be described within the semiclassical approach . after presenting this formalism in application to probing the qubit systems
, we have shown some specific experimental results , which were accompanied by the calculated counterparts .
the agreement between them shows contemporary possibility to demonstrate and describe quantum phenomena in mesoscopic systems .
the results presented here were obtained together with many our colleagues who were our co - authors in the respective publications .
we are grateful to all of them for their contributions .
we thank s. ashhab for useful comments .
sns acknowledges the hospitality of ipht during his visit .
this work was partly supported by nas of ukraine ( project no .
04/10-n ) , dknii ( project no .
m/411 - 2011 ) , bmbf ( ukr 10/001 ) , eu project ( iqit ) .
n. b. delone and v. p. krainov , _ atoms in strong light fields _ , springer ser .
28 , springer , berlin heidelberg ( 1985 ) [ also in russian : _ atom v silnom svetovom pole _ ,
atomizdat , moscow ( 1978 ) ] .
g. oelsner , s. h. w. van der ploeg , p. macha , u. hbner , d. born , s. anders , e. ilichev , h .-
meyer , m. grajcar , s. wnsch , m. siegel , a. n. omelyanchouk , and o. astafiev , phys .
b * 81 * , 172505 ( 2010 ) .
m. grajcar , a. izmalkov , e. ilichev , th .
wagner , n. oukhanski , u. hbner , t. may , i. zhilyaev , h. e. hoenig , ya . s. greenberg , v. i. shnyrkov , d. born , w. krech , h .-
meyer , a. maassen van den brink , and m. h. s. amin , phys .
b * 69 * , 060501 ( 2004 ) .
m. grajcar , a. izmalkov , s. h. w. van der ploeg , s. linzen , e. ilichev , th .
wagner , u. hbner , h .-
meyer , a. maassen van den brink , s. uchaikin , and a. m. zagoskin , phys .
b * 72 * , 020503(r ) ( 2005 ) . | two basic physical models , a two - level system and a harmonic oscillator , are realized on the mesoscopic scale as coupled qubit and resonator .
the realistic system includes moreover the electronics for controlling the distance between the qubit energy levels and their populations and to read out the resonator s state , as well as the unavoidable dissipative environment .
such rich system is interesting both for the study of fundamental quantum phenomena on the mesoscopic scale and as a promising system for future electronic devices .
we present recent results for the driven superconducting qubit - resonator system , where the resonator can be realized as an @xmath0 circuit or a nanomechanical resonator .
most of the results can be described by the semiclassical theory , where a qubit is treated as a quantum two - level system coupled to the classical driving field and the classical resonator .
application of this theory allows to describe many phenomena for the single and two coupled superconducting qubits , among which are the following : the equilibrium - state and weak - driving spectroscopy , sisyphus damping and amplification , landau - zener - stckelberg interferometry , the multiphoton transitions of both direct and ladder - type character , and creation of the inverse population for lasing . |
devices based on conventional lenses can not offer resolution better than half of the operating wavelength due to diffraction limit as well as rayleigh criterion of half of the operating wavelength @xcite .
imaging below the half wavelength resolution ( @xmath0 ) or in subwavelength scale requires capturing evanescent waves which vanishes exponentially as it travels away from the source and requires time - consuming point - by - point scanning by near - field scanning microscopy @xcite .
thus , in order to overcome these restrictions , several imaging techniques were proposed in last decades including perfect lens , silver super - lens , hyper - lens , stimulated emission depletion ( sted ) fluorescence microscope etc . @xcite . besides these , metallic wire media as subwavelength imaging system was also proposed in @xcite .
however , a complete procedure to model such a metallic nanowire based lens for imaging in the visible range was discussed in @xcite . + in the case of subwavelength imaging at optical frequency , the main focus was on manipulating the length of nanowire array ( @xmath1 ) to improve resolution of the image @xcite . even though , the concept of finding optimal parameters to achieve sub - wavelength imaging was investigated by optimizing the length of the rod array @xcite .
kosulnikov _ et . al .
_ drew attention to the effect of filling factors ( defined as @xmath2 , where,`@xmath3 ' is the radius of the nanorods and ` @xmath4 ' is the lattice constant ) at infrared range @xcite .
these types of investigations are even more necessary in the optical range since it is very difficult to maintain accurate dimensions of nanowire array as the radius of the wire medium is much smaller than the skin depth ( @xmath5 ) .
thus in the context of practical realization , where the unit is counted in nanometer , the values of optimal length and radius ensure the ultimate limit of error in the manufacturing process . in this paper , we have calculated the optimal filling ratios ( ratio of the rod cross - sectional area with the lattice cross - sectional area ) and hence the range of radius allowed to achieve imaging . through full wave electromagnetic simulations
we have established our findings using infinitesimal silver nanorod array in equilateral triangular or , hexagonal lattice arrangement with two different cross - sectional geometries .
nanorod array with cylindrical @xcite , triangular @xcite and square @xcite porous matrices for fabrication was realized using different substrate material as shown in fig .
[ fig:1 ] which can be employed for experimental demonstrations in the near future .
-[],width=328 ] in our analysis , we have considered filling ratio , @xmath6 for cylindrical shaped wire array and @xmath7 for triangular shaped wire array . here
the parameter ` @xmath4 ' is known as periodicity or lattice constant of the array and ` @xmath3 ' is the radius of circular cross - section of cylindrical shape fig.[fig:2](d ) and also represents the distance of the polished edged equilateral triangular cross - section fig.[fig:2](e ) . in this manner
we compute the allowable filling ratios or radius for both cross - sectional geometries and found triangular rod array offers higher filling ratio tolerances than the cylindrical one while other parameters like periodicity ` @xmath4 ' and length ` @xmath1 ' were kept unchanged .
the schematic view of the unit cell and silver nanorod array with triangular lattice arrangement are shown in fig . [ fig:2 ] which has been numerically modeled using commercial full - wave electromagnetic simulator cst^tm^ studio suite @xcite .
the rod length is set to be 80 nm for both the geometries with the same periodicity of 40 nm .
we have used drude dispersion model for the permittivity of _ ag _ which was defined as @xmath8 where , @xmath9 = 4.9638 , @xmath10 rad / s and @xmath11/s @xcite .
we have investigated the imaging capabilities of this device with discrete coherent source which were formed by six in - phase point sources polarized along the rod length ( @xmath12-axis ) and placed on the rod axis forming a hexagon , as shown in fig .
[ fig:2 ] ( b , c ) .
the importance of using such a source was discussed in @xcite . to @xmath13 ) of the rod , ( a )
hexagonal point source ( b ) source plane and ( c - k ) image plane for @xmath3=6 - 14 nm respectively.,width=328 ] in order to show the impact of filling ratio , we have examined the @xmath14-field distributions calculated at the image plane ( 10 nm away from the back interface ) at 614thz for a range of filling ratios ( @xmath15 or , @xmath16 ) as shown in fig .
[ fig:3 ] . as observed from fig .
[ fig:3 ] the radii , _ r _ = 9 , 10 , 11 and 12 nm , the image plane field distributions matches with the one obtained in the source plane .
however for other values of ` @xmath3 ' field distributions at the image plane gives us misleading views of the source plane .
it is evident from the figure that for a rod radius larger than 12 nm the image plane field distribution gives an impression of single point source located at the center of the hexagon , whereas for a rod radius smaller than 9 nm extremely inconsistent images are produced .
in order to validate the results obtained in the previous section we have calculated the transmission coefficients of such an array applying transverse magnetic ( tm ) polarization . the architecture of the unit cell
is illustrated in fig .
[ fig:2](a ) for radii ( filling ratio ) between @xmath17 .
here periodic boundary conditions have been applied in both @xmath18 and @xmath19 directions .
the length ` @xmath1 ' of _ ag _ nanowire has been fixed at 80 nm as per ref .
@xcite because they successfully proved that the best image is achieved at 614thz corresponds to the aforementioned rod length .
the periodicity ` @xmath4 ' of the nanorod array has been set to a constant value of 40 nm as in @xcite .
transmission and reflection coefficients for various filling ratio , ` @xmath20 ' have been calculated by changing the radius from 6 nm to 15 nm with an increment of 1 nm . + the transmission coefficients calculated as a function of @xmath21 ( transverse wave vector along @xmath19-axis ) for various radii of the nanorod are plotted in fig .
[ fig:4 ] . from our previous studies
@xcite we are able to establish that the shape of the transmission coefficient curves plotted in fig .
[ fig:4 ] is reasonable enough to produce satisfactory images .
however , images have been observed only at the selected range of radii ( filling ratios ) as shown in fig .
[ fig:3 ] .
this apparent contradiction can be eliminated by studying the phase of the transmission coefficients .
it was demonstrated in ref .
@xcite that a good imaging capability requires a constant transmission phase in addition to a flat transmission coefficient magnitude over the range of transverse wave vector corresponding to the evanescent wave region . + while the radius varied from 6 nm to 14 nm with 1 nm resolution ( a - i).,width=328 ] fig .
[ fig:5 ] shows the reflection and transmission phase for different radii varying from 7 to 14 nm .
it is evident that rod radii from 9 to 12 nm produce constant phase while the rest demonstrates significant variations over the ranges of evanescent harmonics .
a constant transmission phase ensures that all evanescent harmonics propagate through the rod array and reach the image plane simultaneously .
any variation appearing in the transmission phase leads to distortion and hence a misleading image of the source . + from fig .
[ fig:5 ] we see that for the rod radius of 7 , 8 , 13 and 14 nm the phases of the transmission coefficients exhibit considerable variations and thus prevents the formation of an @xmath14-field distributions at the image plane which matches with that of the source plane .
note that in contrast to the phases of transmission coefficients , the reflection coefficient phases exhibits sudden jumps in the evanescent wave region .
+ from fig . [ fig:3 ] , [ fig:4 ] and [ fig:5 ] we can see that for 9 - 12 nm rod radius we get identifiable images .
however it is observed beyond this range @xmath14-field distributions of image plane does not match with that of the source plane .
the radii that failed to produce a recognizable image at 614thz have been found to work well at other frequencies as shown in fig .
[ fig:6 ] and [ fig:7 ] . for a rod radius of 8 nm ,
the lens works best at 550 thz ( fig .
[ fig:6 ] ) , whereas , for the radius of 13 nm the best image was observed at around 600thz ( fig . [ fig:7 ] ) .
finally we calculated the corresponding operational bandwidth for 8 nm ( @xmath22 ) and 13 nm ( @xmath23 ) which is 16.82% and 12.39% respectively .
in this section , we have made an effort to change the cross - sectional geometry of the cylindrical nanowire and observed its impact on the imaging capability of the lens .
we have employed nanorod having triangular cross - section ( with polished edge ) and examined its performance .
the @xmath14-field distributions , taken at the source and the image plane in 614thz demonstrate that the nano - lens offers good fidelity for the rod radii of 10 nm and 13 nm shown in fig .
[ fig:8 ] .
thus it is understandable that for any radius between 10 - 13 nm inclusive ( filling ratio : @xmath24 ) , the device would be operated satisfactorily .
the imaging capability of the nano - lens , however , is hampered for radius below 8 nm and above 15 nm as the @xmath14-field distributions produce misleading vision of the source .
+ = 0.20 , 0.031 , 0.053 and 0.070 respectively.,width=328 ] = 8 , 10 , 13 and 15 nm respectively ( a - d),width=328 ] the imaging behavior of this lens can again be explained with the help of transmission characteristics .
the magnitudes of transmission coefficients plotted in fig .
[ fig:9 ] shows reasonably acceptable behavior for the radii of 8 , 10 , 13 and 15 nm .
the phases , however , exhibit superior behavior , as shown in fig .
[ fig:10 ] , only for 10 and 13 nm radii .
they demonstrate non - varying characteristics that actually enable the device to function properly .
the improper images of the source obtained at image planes for 8 and 15 nm are attributed to the varying nature of the phases that dilute the image quality and preclude the device from functioning properly .
note that the @xmath14-field distribution at the image plane obtained with 15 nm radius is associated with undesired fields and thus resulting in a very noisy image that is hardly recognizable , whereas , for the radius of 8 nm the image is characterized by uneven field intensity for the point sources thus giving a different definition of the source than the actual one .
+ similar to the circular cross - section , we have also noticed for triangular cross - section that the filling ratios those failed to produce any image at 614thz have been found to work at other frequencies as demonstrated in fig .
[ fig:11 ] and [ fig:12 ] .
we have summarized our studies in table [ table1 ] that contains information about the operating band for different radii and hence the filling ratios .
note that the quality of the image becomes poor at higher filling ratio for both the circular and triangular geometries and may be attributed to the increased coupling between surface plasmons of adjacent rods .
the idiosyncratic shift of operating frequency range with the change in filling ratio can also be due to the varying degree of coupling with the changing radius .
thus the behavior of nanolens in the visible domain is much more complex than it is in the microwave regime .
+ c0.2 c0.2 c0.2 c0.2 + radius , ` @xmath3 ' ( ) & frequency range ( ) & center frequency ( ) & bandwidth ( % ) + 8 & 490 - 580 & 535 & 16.82 + 9 & 590 - 650 & 620 & 9.68 + 10 & 550 - 630 & 590 & 13.56 + 11 & 590 - 640 & 615 & 8.13 + 12 & 560 - 660 & 605 & 16.53 + 13 & 530 - 600 & 565 & 12.39 + + 8 & 450 - 540 & 595 & 15.13 + 10 & 530 - 620 & 575 & 15.65 + 13 & 540 - 630 & 585 & 15.38 + 15 & 540 - 610 & 575 & 12.17 + [ table1 ]
we have demonstrated the impact of filling ratios on subwavelength imaging capabilities of a nanolens formed by an array of silver nanorod .
we have observed that impact for two different cross - sectional geometries by analyzing image quality and calculating transmission and reflection coefficients . from our analysis
we have found that having proper filling ratio is imperative to obtaining the right imaging performance of the lens .
our study in this paper confirms that the lens can be operated at different filling ratios and therefore offers a leeway from the fabrication point of view .
our study also confirms that subwavelength imaging capability of such a lens is independent of the cross - sectional geometry and thus offers another degree of freedom .
s. y. kosulnikov , e. a. yankovskaya , s. i. maslovski , p. a. belov and y. s. kivshar ,
_ optimal filling factor of nanorod lenses for subwavelength imaging _ " , phys .
065801 , ( 2011 ) .
a. atrashchenko , a. nashchekin , m. mitrofanov , v. p. ulin , and v. p. evtikhiev ,
_ wire metamaterial based on semiconductor matrices _ " , phys .
status solidi ( rrl)-rapid research letters phys . , pp . 13 , ( 2014 ) .
a. v. atrashchenko , v. n. katz , v. p. ulin , v. p. evtikhiev , and v. p. kochereshko , _ fabrication and optical properties of porous inp structures _ " , physica e. low - dimensional systems and nanostructures , vol . 44 ( 7 ) , pp .
13241328 , ( 2012 ) . | metallic nano - structured lens has strong potential application in transporting subwavelength image information .
this be achieved by manipulating the length of the nanorod and the periodicity of the rod array . in this paper , we have demonstrated the impact of filling ratio on the subwavelength imaging capabilities of such lenses . through full - wave electromagnetic simulation , we have demonstrated that the imaging performance of silver ( ag ) nanorod array does not only depend on the length and periodicity but also on the filling ratios or the radius of the wire medium .
we have investigated this impact in two different geometries for nanorod array in the form of cylindrical and triangular rod and also examined their performances for different filling ratios .
subwavelength imaging , metallic nanowires , near - field scanning microscopy , surface plasmons . |
the enormous advancements of nano - science and technology over the last few decades have allowed us to investigate electron transport in low - dimensional model quantum systems in a very tunable way .
the precision instruments bestowed by nanotechnology have given indulgence to such low - dimensional systems with tailor - made geometries to become promising candidates for different nano - electronic devices .
one of such geometries is a mesoscopic ring which looks very simple but has very high potential from the application perspective .
several quantum mechanical phenomena have been observed in this closed loop structure specially in presence of magnetic flux @xcite .
the existence of dissipationless current , the so - called persistent current , in a mesoscopic normal metal ring pierced by an ab flux @xmath0 is one of such remarkable effects which reveals the importance of phase coherence of electronic wave functions in low - dimensional systems . in 1983 ,
et al . _
@xcite initially predicted this significant phenomenon through some theoretical arguments , but , its actual realization came much later after the nice experiment done by levy _
et al . _
they have observed the oscillations with period @xmath4 ( @xmath5 , the elementary flux - quantum ) while measuring persistent current in an ensemble of @xmath6 independent cu rings .
similar @xmath4 oscillations were also reported not only for an ensemble of disconnected @xmath7 ag rings @xcite but also for an array of @xmath7 isolated gaas - algaas rings @xcite . later many other theoretical @xcite as well as experimental @xcite attempts
have also been made to reveal the actual mechanisms of persistent current in single - channel rings and multi - channel cylinders .
similar to persistent current in such isolated closed geometries , the non - decaying charge current is also observed in open systems @xcite like , a mesoscopic ring with side attached electrodes . though a wealth of literature has already been generated involving the analysis of persistent current in conventional mesoscopic rings , cylinders , connected rings , ring with multi - arm structures ,
there is still need to look deeper into the problem to address several important issues those have not been explored earlier . in the present article
we undertake an in - depth study of magneto - transport through a binary alloy mesoscopic ring in presence of the magnetic flux @xmath0 within a single - band non - interacting tight - binding framework .
we analyze our results in two parts .
in the first part , we describe the energy band structure and persistent current for an ordered binary alloy ring .
we perform these results analytically . in the second part
, we describe the magneto - transport properties in presence of external electrodes which include transmission probabilities , persistent currents and related issues .
quite interestingly we examine that if we include some new additional atomic sites , called as impurities , those are not necessarily to be random , in any part of the binary alloy ring , some quasi - localized energy levels appear depending on the number of impurity sites within the band of extended regions .
the positions of these nearly - localized energy levels can also be tuned with respect to the extended regions by means of some external gate voltages .
it leads to the possibility of using such a system as a @xmath2-type or an @xmath3-type semiconductor by fixing the fermi level at appropriate energy .
in what follows , we describe the model quantum systems and the results . in section
ii we present the results and related discussions for an ordered binary alloy ring .
section iii illustrates the model and the theoretical approach to calculate transmission probability , persistent current and related issues in presence of external leads , and the numerical results are described in section iv . finally , in section v we draw our conclusions
* :* the model quantum system is illustrated schematically in fig .
[ ring1 ] . an ordered binary alloy ring , subjected to an ab flux @xmath0 , composed of two different types of atoms , namely , @xmath8 and @xmath9 those are placed alternately in a regular pattern .
we use a non - interacting single - band tight - binding ( tb ) hamiltonian to describe the model quantum of @xmath8 and @xmath9 sites ) binary alloy ring the tb hamiltonian reads , @xmath10 where , @xmath11 is the on - site energy and @xmath12 refers to the nearest - neighbor hopping strength .
for the site @xmath8 , @xmath11 is identical to @xmath13 , while it is @xmath14 for the site @xmath9 .
the phase factor @xmath15 appears into the hamiltonian due to the magnetic flux @xmath0 threaded by the ring which is measured by the elementary flux - quantum @xmath16 .
@xmath17 ( @xmath18 ) is the creation ( annihilation ) operator for an electron at the site @xmath19 .
0.2 cm * :* in order to explore the characteristic features of persistent current in an ordered binary alloy ring , let us begin with the energy band structure .
we do it analytically .
the energy dispersion relation for the ordered binary alloy ring can be expressed mathematically as , @xmath20 where , @xmath21 is the lattice spacing and @xmath22 is the wave vector . in presence of the ab flux
@xmath0 , the quantized values of @xmath22 are obtained from the relation , @xmath23 where , @xmath3 is an integer and it is restricted within the range : @xmath24 . as illustrative example , in fig .
[ band ] we plot the energy levels as a function of flux @xmath0 , obtained from eq .
[ equ2 ] , for a @xmath25-site binary alloy ring considering @xmath26 and @xmath27 .
it shows that two different sets of energy levels are obtained , forming two quasi - bands , and they are separated by a finite energy gap .
this gap , on the other hand , is tunable by the parameter values describing the tb hamiltonian eq .
the origin of two different sets of energy levels is also clearly visible from eq .
[ equ2 ] . from the energy spectrum ( fig .
[ band ] ) we see that at half - integer or integer multiples of flux - quantum , energy levels have either a maximum or a minimum and it results vanishing nature of persistent current at these specific values of @xmath0 , since the current is obtained by taking the first order derivative of energy @xmath28 with respect to flux @xmath0 @xcite .
all these energy levels vary periodically providing @xmath16 flux - quantum periodicity .
0.2 cm * :* once we get the energy eigenvalues as a function of flux @xmath0 , we can easily calculate persistent current for individual energy eigenstates .
it is simply the first order derivative of energy with respect to flux @xmath0 .
therefore , for an @xmath3-th energy eigenstate we can write the expression for the current as , @xmath29 } } { \sqrt{1 + 4 t_r^2 cos^2{\left [ \frac{2 \pi}{n a } \left(n+\phi/\phi_0\right ) \right]}}}\ ] ] where , @xmath30ve or @xmath31ve sign in the current expression appears depending on the choice of @xmath3 i.e. , in which sub - band the energy level exists . at absolute zero temperature , total persistent current @xmath32 for a particular filling @xmath1
is obtained by taking the sum of individual contributions from the lowest @xmath1 energy eigenstates .
thus , we can write , @xmath33 as representative examples , in fig . [ oddeven ]
we display the variation of persistent current for an ordered @xmath34-site binary alloy ring considering @xmath26 and @xmath27 in the different filled band cases . for odd number of electrons the results
are shown in fig .
[ oddeven](a ) while , in fig .
[ oddeven](b ) the results are given for even number of electrons .
both for the odd and even @xmath1 , persistent current shows saw - tooth like variation as a function of flux @xmath0 , similar to that of traditional single - channel mesoscopic rings .
the sharp transitions at half - integer ( for odd @xmath1 ) or integer ( for even @xmath1 ) multiples of flux - quantum ( @xmath16 ) in persistent current appears due to the crossing of energy levels at these respective values of @xmath0 .
quite interestingly , we also examine that the current shows always diamagnetic in nature irrespective of the filling factor .
in the rest of the present article we explore the essential features of magneto - transport properties for a binary alloy ring in presence of external leads i.e. , for an open system . 0.2 cm * :* the model quantum system is depicted in fig . [ ring2 ] where a binary alloy ring , threaded by a magnetic flux @xmath0 , is attached to two semi - infinite one - dimensional metallic electrodes .
the ring is composed of @xmath35 identical pairs of @xmath8-@xmath9 sites and @xmath36 identical foreign atomic sites , labeled as @xmath37 sites , those are embedded in a small portion of the ring .
these sites ( @xmath37 sites ) are often called as impurity sites .
we label the atomic sites of two side - attached leads in a particular way , as shown in fig .
[ ring2 ] , and they are connected to the ring via the sites @xmath38 and @xmath39 . a single - band non - interacting tight - binding framework is used to describe the entire system . for the full system we can write the total hamiltonian as a sum of three terms like , @xmath40 where , @xmath41 , @xmath42 and @xmath43 correspond to the hamiltonians for the ring , leads ( left and right ) and tunneling between the ring and leads , respectively .
the first term ( @xmath41 ) looks exactly similar as in eq .
[ hr ] , but here @xmath11 has three possibilities for three different atomic sites ( @xmath8 , @xmath9 and @xmath37 ) .
we can also tune the site energy @xmath44 by means of some external gate voltage @xmath45 , and thus , we can write it as @xmath46 , where @xmath47 is the site energy in absence of any external potential .
the other two terms of eq .
[ part ] , @xmath42 and @xmath43 , can also be expressed in a similar fashion as , @xmath48 and , @xmath49 the site energy dependent term in @xmath42 is omitted as we set the site energy for the identical sites in the leads to zero .
@xmath50 ( @xmath51 ) corresponds to the creation ( annihilation ) operator of an electron at the site @xmath52 of the leads and @xmath53 represents the nearest - neighbor hopping integral in these leads .
@xmath54 is the coupling strength between the left lead and the ring , while it is @xmath55 for the other case .
0.2 cm * :* to find transmission probability across the ring and also to calculate persistent current in such an open ring system we adopt wave - guide theory @xcite . here
we describe it very briefly .
let us start with the scrdinger equation @xmath56 , where @xmath57 is the stationary wave function of the entire system . in the wannier basis it ( @xmath57 ) can be written as , @xmath58 where , the co - efficients @xmath59 and @xmath60 correspond to the probability amplitudes in the respective sites .
assuming that the electrons are described by a plane wave we can express the wave amplitudes in the left and right leads as , @xmath61 and @xmath62 where , @xmath63 and @xmath64 are the reflection and transmission amplitudes , respectively .
@xmath22 is the wave number and it is related to the energy @xmath65 of the incident electron by the expression @xmath66 .
the lattice spacing @xmath21 is set equal to @xmath67 .
now to find out the transmission amplitude @xmath64 , we have to solve the following set of coupled linear equations .
@xmath68 where , the co - efficients @xmath69 , @xmath70 , @xmath71 and @xmath72 can be easily expressed in terms of @xmath63 and @xmath64 by using eqs .
[ equ10 ] and [ equ11 ] , and they are in the form : @xmath73 thus , for a particular value of @xmath65 we can easily solve the set of linear equations and find the value of @xmath64 . finally , the transmission probability across the ring becomes @xmath74 now , to calculate persistent current between any two neighboring sites in the ring we use the following relation , @xmath75
throughout the calculations we set @xmath26 , @xmath76 , @xmath27 , @xmath77 and @xmath78 . the energy scale is measured in unit of @xmath12 and we choose the units where @xmath79 . in fig .
[ spectrum ] we show the variation of transmission probability @xmath80 ( red color ) as a function of injecting electron energy @xmath65 for some typical binary alloy rings considering different number of impurity sites .
the average density of states ( ados ) is also superimposed in each spectrum . here ,
we fix @xmath81 and take three different values of @xmath36 those are presented in figs
. [ spectrum](a ) , ( b ) and ( c ) , respectively . in all these cases
the magnetic flux @xmath0 is set equal to zero .
the results are quite interesting and important also .
it is observed that in absence of impurity sites electron can transmit through the ring for two wide band of energies ( see fig .
[ spectrum](a ) ) .
this transmission spectrum exactly overlaps with the ados profile ( @xmath82-@xmath65 spectrum ) which ensures that electronic transmission takes place through all the energy eigenstates of the binary alloy ring and they are extended in nature .
the band splitting for the ordered binary alloy ring is easily understood from our previous discussion ( sec .
ii ) and the gap between these two bands are also tunable by the parameter values describing the system .
the situation becomes really interesting when some additional impurities ( @xmath37 sites ) are introduced in any part of the binary alloy ring . due to the inclusion of such atomic sites some energy levels
appear within the band of extended regions those are no longer extended , but they are almost quasi - localized and do not contribute to the electronic transmission .
this behavior is well explained in figs .
[ spectrum](b ) and ( c ) , where @xmath80-@xmath65 and @xmath82-@xmath65 spectra are superimposed with each other for two different numbers of impurity sites ( @xmath36 ) .
the transmission of electrons takes place mainly along the two band edges , while in the band centre , several energy levels are there which provide zero transmission probability .
the number of these almost localized energy levels between the extended regions gets increased with the increase of impurity sites , and for large enough impurity sites they almost form a quasi - energy band of localized states .
the location of the almost localized energy band can be shifted towards the edge of extended regions simply by tuning the site energy of these foreign atoms , and here we do it by means of applying an external gate voltage @xmath45 .
this phenomenon can be utilized to use such a geometry as a @xmath2-type or an @xmath3-type semiconductor by fixing the fermi level in an appropriate place . to explore this fact in fig .
[ pntype ] we present the variation of transmission probability in addition to the ados for a much larger ring size where we consider @xmath83 identical pairs of @xmath8-@xmath9 sites and @xmath84 impurity sites .
two different cases are considered , one for the case when @xmath44 is set at @xmath85 ( fig .
[ pntype](a ) ) , and the other is shown for @xmath86 ( fig .
[ pntype](b ) ) . for a quite large number of impurity sites we get almost quasi - energy bands ( ados spectra ) and
quite nicely we see that when @xmath44 is fixed at @xmath85 , a localized energy band for a wide range of energy is formed along the left edge of the extended region ( fig .
[ pntype](a ) ) .
now if we choose the fermi level around @xmath87 , then many electrons in the localized region below the fermi level can jump , even at much low temperature since the energy gap is almost zero , to the extended regions and can contribute to the current .
thus , we get a large number of excess electrons , depending on the localized energy levels and also the available extended energy states , in the conduction region which behave as n - type carriers .
an exactly opposite behavior is obtained when we set @xmath44 to @xmath88 . in this case the wide band of localized states is formed in the right edge of the extended region ( see fig .
[ pntype](b ) ) . therefore ,
if we set the fermi energy level around @xmath89 , then the electrons from the filled extended levels below the fermi level hop to the nearly empty localized levels , and these electrons do not contribute anything to the current .
but some holes are formed in the extended regions which can carry current and we get @xmath2-type carriers .
so , in short , we can emphasize that by fixing the fermi level in appropriate places the geometry can be utilized as a @xmath2-type or an @xmath3-type semiconductor .
before we end this analysis , we would like to point out that here we present the results for a particular set of parameter values and establish how such a geometry can be utilized as a @xmath2-type or an @xmath3-type semiconductor upon the movement of the fermi level . for a model calculation
we choose some specific values of the parameters used in the numerical calculation , but all these physical phenomena are exactly invariant with the change of the parameter values .
only the numerical values will be altered .
these features are also exactly valid even for a non - zero value of magnetic flux @xmath0
. the physical picture will be much more appealing if we consider larger rings with more impurity sites and it gives us the confidence to propose an experiment in this line . in a recent work bellucci _
et al . _
@xcite have done a detailed study of magneto - transport properties in quantum rings considering tunnel barriers in the presence of magnetic field and established how metal - to - insulator transition takes place in such a geometry .
it has been shown that by controlling the strength and the positions of the barriers , the energy shift can be done in a tunable way and they have also proposed this model for experimental realization .
this is quite analogous to our present study , and therefore , an experiment in this regard will be challenging .
the behavior persistent current @xmath90 as a function of energy @xmath65 in the upper arm of an ordered binary alloy ring is presented in fig .
[ curr1 ] , where ( a ) and ( b ) correspond to @xmath91 and @xmath92 , respectively .
the ring is symmetrically coupled to the side attached leads i.e. , the upper and lower arms have identical length , and accordingly , the current @xmath93 in the lower arm becomes exactly identical to the current obtained in the upper arm . from the spectra we observe that the current is available for two wide range of energies , separated by a finite gap , associated with the energy levels of the ring those are clearly visible from the ados profiles ( red color ) .
all these energy eigenstates are extended in nature since they provide a non - zero current , but the nature of the current is quite different for the two different cases i.e. , when @xmath91 ( fig .
[ curr1](a ) ) and @xmath94 ( fig . [ curr1](b ) ) . for @xmath91 ,
persistent current shows @xmath30ve value for all the energy levels which predicts that the currents are in the same phase . while , for a finite value of @xmath0 , the phase of the current changes alternatively .
so by measuring persistent current we can directly estimate the nature of the current i.e. , whether it is paramagnetic or diamagnetic in nature and also predict the characteristics of energy eigenstates , which are somewhat interesting in the study of electron transport .
to explore the effect of impurities on persistent current in such a geometry , in fig .
[ curr2 ] , we display the results for a typical binary alloy ring ( @xmath95 ) in presence of some impurity sites ( @xmath96 ) , where ( a ) and ( b ) represent the results for the upper and lower arms of the ring , respectively .
the impurity sites are added in the upper arm of the ring . with the inclusion of impurity sites ,
the symmetry between the two arms is broken , and therefore , the currents in the upper and lower arms are no longer identical to each other as shown from the spectra ( figs .
[ curr2](a ) and ( b ) ) .
unlike to the ordered binary alloy ring ( fig . [ curr1 ] )
, here all the energy eigenstates are not extended in nature .
some localized energy levels appear in the band of extended energy states due to the presence of impurity sites in the ring .
this is clearly visible from the spectra since for these impurity levels no current is available .
thus , be calculating persistent current we can emphasize the nature of energy eigenstates very nicely , and this idea can be utilized to reveal the localization properties of energy eigenstates in any complicated geometry .
analogous to our present model , few works @xcite have also been done very recently in some mesoscopic rings in presence of tunnel barriers or in presence of an artificial crystal .
the nature of energy band structures and the behavior of persistent currents have been described in detail in these samples , and , we hope our results provide some extensions in this regard .
in the present article we make a detailed study of magneto - transport through a binary alloy ring in presence of magnetic flux @xmath0 within a single - band non - interacting tight - binding framework .
the essential results are analyzed in two parts . in the first part
, we discuss the band structure and persistent current of an isolated ordered binary alloy ring .
these results are done analytically . in the other part
, we explore the magneto - transport properties of a binary alloy ring in presence of external leads .
the effect of impurities are also addressed .
quite interestingly we see that in presence of some foreign atoms , those are not necessarily be random , in any part of the ring , some quasi - localized energy levels appear within the band of extended energy levels .
the locations of these almost localized energy levels can also be regulated by means of some external gate voltage .
this leads to a possibility of using such a system as a @xmath2-type or an @xmath3-type semiconductor by fixing the fermi level in appropriate places . | magneto - transport properties are investigated in a binary alloy ring subjected to an aharonov - bohm ( ab ) flux @xmath0 within a single - band non - interacting tight - binding framework . in the first part ,
we expose analytically the behavior of persistent current in an isolated ordere binary alloy ring as functions of electron concentration @xmath1 and ab flux @xmath0 . while , in the second part of the article , we discuss electron transport properties through a binary alloy ring attached to two semi - infinite one - dimensional metallic electrodes . the effect of impurities is also analyzed . from our study
we propose that under suitable choices of the parameter values the system can act as a @xmath2-type or an @xmath3-type semiconductor . |
as one of the most promising solid - state candidates for quantum information processing [ 1,2 ] , the physical system composed of circuit cavities and superconducting qubits is of particular interest .
this is because : ( i ) superconducting qubits and microwave resonators ( a.k.a .
cavities ) can be fabricated using modern integrated circuit technology , and their properties can be characterized and adjusted in situ , ( ii ) superconducting qubits have relatively long decoherence times [ 3,4 ] , and ( iii ) a superconducting microwave cavity or resonator plays a role of quantum bus which can mediate long - range and fast interaction between distant superconducting qubits [ 5 - 7 ] .
moreover , the strong coupling between the cavity field and superconducting qubits , which is difficult to implement with atoms in a microwave cavity , was earlier predicted in theory [ 1,8 ] and has been experimentally demonstrated [ 9,10 ] . because of these features , circuit qed has been widely utilized for quantum information processing . based on circuit qed
, many theoretical proposals have been presented for realizing two - qubit gates [ 5,11 - 18 ] and multiple qubit gates [ 19 - 20 ] with superconducting qubits . moreover ,
experimental demonstration of two - qubit gates [ 6,7,21,22 ] and three - qubit gates [ 23 - 25 ] with superconducting qubits in circuit qed has been also reported .
the interest in quantum computation is stimulated by the discovery of quantum algorithms [ 26,27 ] which can solve problems of significance much more efficiently than their classical counterparts . among important quantum algorithms ,
there exist the deutsch algorithm [ 28 ] , the deutsch - jozsa algorithm [ 29 ] , the shor algorithm [ 30 ] , the simon algorithm [ 31 ] , the quantum fourier transform algorithm , and the grover search algorithm [ 32 ] . as is well known , the deutsch algorithm and the deutsch - jozsa algorithm were the first two that make use of the features of quantum mechanics for quantum computation . compared with other quantum algorithms , these two algorithms are easy to be implemented and thus have been considered as the natural candidates for demonstrating power of quantum computation .
we note that with superconducting qubits coupled to a circuit cavity , a _ two - qubit _ _ deutsch - jozsa _ quantum algorithm and a two - qubit grover search quantum algorithm were previously demonstrated in experiments [ 7 ] .
however , after a thorough investigation , we note that how to implement a _ three - qubit _ deutsch - jozsa ( dj ) quantum algorithm with superconducting qubits or qutrits in circuit qed has not been reported in both theoretical and experimental aspects .
as is known , the experimental realization of the three - qubit dj algorithm with a cavity - superconducting - device system is important because it would be an important step toward more complex quantum computation in circuit qed . in this paper
, we propose a protocol to construct the 35 @xmath0-controlled phase gates of a three - qubit refined dj algorithm , by using single - qubit @xmath2 gates , two - qubit cp gates , and two - target - qubit cp gates .
it should be noted that a two - target - qubit cp gate consists of two sequential controlled @xmath3 gates , which have a common control qubit but a different target qubit . the protocol is quite general and can be applied to implement the three - qubit dj algorithm in various of physical systems .
based on this protocol , we further discuss how to implement the three - qubit refined dj algorithm with superconducting transmon qutrits resonantly coupled to a single cavity .
our numerical calculation shows that implementation of this quantum algorithm is feasible within the present circuit qed technique .
the paper is organized as follows . in sec .
ii , we review the refined dj algorithm . in sec .
iii , we present a protocol to construct the 35 @xmath0-controlled phase gates . in sec .
iv , we discuss how to implement this dj algorithm with superconducting transmon qutrits coupled to a cavity via resonant interaction , and analyze the experimental feasibility . a concluding summary is given in sec .
the dj algorithm is aimed at distinguishing the constant function from the balanced functions on @xmath4 inputs .
the function @xmath5 takes either @xmath6 or @xmath7 for the constant function , the function values are constant ( @xmath6 or @xmath8 ) for all @xmath4 inputs .
in contrast , for the balanced function , the function values are equal to 1 for half of all the possible inputs , and @xmath6 for the other half . using the dj algorithm , whether the function is constant or balanced
can be determined by only one query . however
, a classical algorithm would require @xmath9 queries to answer the same problem , which grows exponentially with input size .
-qubit deutsch - jozsa algorithm.,width=12 ] the refined dj algorithm was proposed by collins _
et al . _ in 2001 [ 33 ] , which is illustrated in fig . 1 .
this refined dj algorithm is described below : \(i ) each input query qubit is prepared in the initial state @xmath10 \(ii ) perform a hadamard transformation @xmath11 on each qubit , resulting in @xmath12 and @xmath13 ) . as a result ,
the @xmath14-qubit initial state @xmath15 changes to the state @xmath16 ( denoted as @xmath17)@xmath18 \(iii ) apply the @xmath0-controlled phase shift @xmath19 , described by @xmath20 which leads the state @xmath21 to the state @xmath22 ( denoted as @xmath23 ) .
\(iv ) perform another hadamard transformation @xmath11 on each qubit . as a result ,
the state @xmath23 becomes @xmath24 \(v ) measure the final state of the @xmath14 qubits .
if the @xmath14 qubits are measured in the state @xmath25 then @xmath26 is constant . however , if they are measured in other @xmath14-qubit computational basis states , then @xmath5 is balanced .
this is because the amplitude @xmath27 of the state @xmath28 is given by @xmath29 which is @xmath30 for a constant @xmath5 while @xmath6 for a balanced @xmath5 .
one can see that when compared with the original dj algorithm [ 29 ] , this refined dj algorithm does not need an auxiliary working qubit .
hence , it requires one qubit fewer than the original dj algorithm .
consequently , its physical implementation requires one fewer two - state system .
for a @xmath14-qubit dj algorithm , there are a total of @xmath31 balanced functions , among which only @xmath32 balanced functions are nontrivial if the symmetry is taken into account .
for the three - qubit dj algorithm , i.e. , @xmath33 there thus exist @xmath34 nontrivial balanced functions @xmath35 ( see table i ) . in this section ,
we show how to construct the 35 @xmath0-controlled phase gates , by using single - qubit @xmath36 gates , two - qubit cp gates , and two - target - qubit cp gates . a single qubit @xmath36 gate results in the transformation @xmath37 while @xmath38 a two - qubit cp gate @xmath39 on qubits @xmath40 and @xmath41 considered here is described as follows @xmath42 which implies that if and only if the control qubit @xmath40 ( the first qubit ) is in the state @xmath43 , a phase flip happens to the state @xmath43 of the target qubit @xmath41 ( the second qubit ) , but nothing happens otherwise . in addition , a two - target - qubit cp gate @xmath44 with the control qubit @xmath40 and the two target qubits @xmath41 and @xmath45 is defined below @xmath46 which shows that if and only if the control qubit @xmath40 ( the first qubit ) is in the state @xmath47 a phase flip happens to the state @xmath43 of the target qubit @xmath41 ( the second qubit ) and the state @xmath43 of the target qubit @xmath48 ( the last qubit ) .
the construction for each of the 35 @xmath0-controlled phase gates is listed in table ii .
one can see from table ii that the 35 @xmath0-controlled phase gates @xmath49 are classified into the following four types : ( i ) type 1 includes seven @xmath0-controlled phase gates each constructed with single - qubit @xmath36 gates only ; ( ii ) type 2 contains twelve @xmath0-controlled phase gates each constructed with single - qubit @xmath36 gates and one two - qubit cp gate ; ( iii ) type 3 has twelve @xmath0-controlled phase gates each constructed by using single - qubit @xmath36 gates and a two - target - qubit cp gate ; and ( iv ) type 4 involves four @xmath0-controlled phase gates each implemented with single - qubit @xmath36 gates , a two - qubit cp gate , and a two - target - qubit cp gate at most .
using the protocol , we now discuss how to implement the three - qubit refined dj algorithm with three superconducting transmon qutrits ( @xmath50 ) each having three levels ( i.e. , the ground @xmath51 the first excited @xmath52 and the second excited level @xmath53 ) .
we then estimate the fidelity of the operation , which is performed in a setup composed of three phase qutrits and a one - dimensional coplanar waveguide resonator .
since the @xmath0-controlled phase gates belonging to type 1 are constructed using single - qubit gates only , their implementation does not require entanglement and thus can be realized in a classical way .
therefore , in the following we focus on the @xmath0-controlled phase gates belonging to type 2 , type 3 , and type 4 . without loss of generality , let us consider the three @xmath0-controlled phase gates @xmath54 ( belonging to type 2)@xmath55 @xmath56(belonging to type 3)@xmath55 and @xmath57 ( belonging to type 4 ) . by comparing them with other @xmath0-controlled phase gates in the same types , it can be found that these three unitary gates @xmath58 and @xmath57 contain the same number of two - qubit cp gates or / and two - target - qubit cp gates but a greater or equal number of single - qubit gates than the other @xmath0-controlled phase gates in the same types .
hence , if the three @xmath0-controlled phase gates @xmath54 , @xmath59 , and @xmath57 can be implemented , then other @xmath0-controlled phase gates in the same types can be achieved with a higher or equal fidelity . in this sense , to see how well the proposal works , it would be sufficient to explore the implementation feasibility of the following three joint quantum operations , described by @xmath60 where the gate operation sequence is from right to left .
the @xmath61 , @xmath62 , and @xmath63 here are constructed , according to fig . 1 and the decomposition of @xmath54 , @xmath59 , and @xmath57 given in table ii .
from eq .
( 4 ) , one can see that @xmath64 and @xmath63 are implemented through the single - qubit @xmath2 and @xmath11 gates , two - qubit cp gates and two - target - qubit cp gates . _
_ a.1 implementing single - qubit gates__the single - qubit hamardard @xmath11 gate or @xmath3 gate on qutrit @xmath40 can be realized by applying a pulse resonant with the @xmath65 transition of qutrit @xmath40 ( @xmath66 ) . to eliminate the leakage into the level @xmath67
one can employ the drag pulse [ 34,35 ] , which can reduce the gate error by an order of magnitude relative to the state of the art , all based on smooth and feasible pulse shapes [ 34 ] .
in addition , to shorten the gate time , the three joint hamardard gates @xmath68 in eq . ( 4 ) are performed simultaneously , which can be achieved by turning on and off the pulses applied to the three qutrits at the same time . in the same manner ,
the three @xmath3 gates involved in @xmath61 ( the two @xmath3 gates in @xmath62 and @xmath63 ) are performed simultaneously .
detailed discussion of how to implement the @xmath11 gate or @xmath3 gate is omitted here since implementing a single - qubit gate depends on the use of the pulse shapes and is straightforward in experiments .
transition of qutrit @xmath40 with a coupling constant @xmath69 .
( b ) the cavity is resonant with the @xmath70 transition of qutrit @xmath41 with a coupling constant @xmath71.,width=10 ] _ a.2 implementing a two - qubit cp gate _ we define @xmath72 ( @xmath73 ) as the resonant coupling constant between the cavity mode and the @xmath74 ( @xmath75 ) transition of qutrit @xmath40 ( @xmath41 ) . the cavity is initially in the vacuum state @xmath76 the procedure for realizing @xmath39 is listed as follows : step ( i ) . adjust the level spacings of qutrit @xmath40 such that the @xmath77 transition is on resonance with the cavity [ fig .
after an interaction time @xmath78 , the state @xmath79 changes to @xmath80 while nothing happens to the state @xmath81 ( e.g. , see [ 19 ] ) . step ( ii ) . adjust the level spacings of qutrit @xmath41 such that the @xmath82 transition is on resonance with the cavity [ fig .
after an interaction time @xmath83 the state @xmath84 becomes @xmath85 while the states @xmath86 @xmath87 and @xmath88 remain unchanged .
step ( iii ) . adjust the level spacings of qutrit @xmath40 such that the @xmath74 transition is on resonance with the cavity [ fig .
after an interaction time @xmath89 , the state @xmath90 changes to @xmath91 while nothing happens to the state @xmath92 one can check that the state @xmath93 remains unchanged while the state @xmath94 changes to @xmath95 after the above operations . on the other hand , the initial states \{@xmath96 } remain unchanged during the entire operation above .
these results show that a two - qubit cp gate @xmath97 described by eq .
( 2 ) , was achieved with qutrit @xmath40 ( the control ) and qutrit @xmath41 ( the target ) after the above process , while the cavity returns to its original vacuum state . _
a.3 realizing a two - target - qubit cp gate _ by carefully examining the procedure described above for implementing @xmath39 , we note that a two - targe - qubit cp gate @xmath44 described by eq .
( 3 ) can be realized using four operational steps only : steps ( i ) and ( ii ) : the operations for these two steps are the same as those for steps ( i ) and ( ii ) described above .
step ( iii ) : adjust the level spacings of qutrit @xmath48 such that the @xmath75 transition is on resonance with the cavity . after an interaction time
@xmath98 ( where @xmath99 is the coupling constant between the cavity mode and the @xmath100 transition of qutrit @xmath48 ) , the state @xmath101 becomes @xmath102 while the states @xmath103 @xmath104 and @xmath105 remain unchanged .
step ( iv ) : the operation for this step is the same as that for step ( iii ) . during performing single - qubit - gate operations ,
all three superconducting phase qutrits @xmath106 @xmath107 and @xmath108 need to be decoupled from the cavity mode ; and during performing a two - qubit cp gate or a two - target - qubit cp gate , irrelevant qutrits need to be decoupled from the cavity mode .
this requirement can be met by a prior adjustment of the level spacings of the qutrits .
note that for superconducting qutrits , the level spacings can be rapidly adjusted by varying external control parameters ( e.g. , magnetic flux applied to phase , transmon , or flux qutrits ; see , e.g. , [ 3,36,37 ] ) . as a final note
, it should be mentioned that the method described above for implementing _ a two - qubit cp gate _ via resonant interaction is not new , which was previously proposed [ 38,39 ] .
we would like to stress that our focus is to take the resonant interaction as an example to explore the possibility of implementing the three - qubit dj algorithm with superconducting transmon qutrits coupled to a single cavity , by using the protocol presented in the previous section . or qutrit @xmath41 ( @xmath109 or @xmath108 ) .
for the details , see subsection 4.2.,width=10 ] let us now study the fidelity of the operation .
since the resonant interaction is used in the implementation of the single - qubit @xmath11 gates or @xmath2 gates , these basic gates can be completed within a very short time ( e.g. , by increasing the pulse rabi frequency ) .
in addition , as mentioned previously , one can apply the drag pulses to eliminate the leakage into the level @xmath110 thus , the single - qubit gate error is negligibly small . in this case , decoherence of the system would have a negative impact on the operation of implementing a two - qubit cp gate as well as a two - target - qubit cp gate , _ _
_ _ due to the population of the cavity photons during the operation .
as discussed above , the implementation of these cp gates involves two basic operations : \(i ) the first one requires that during performing @xmath39 and @xmath111 the cavity mode is resonant with the @xmath112 transition of the control qutrit @xmath113 in realistic case , the interaction hamiltonian for this basic operation is given by @xmath114 where @xmath115 is the photon creation operator of the cavity mode , and the second term represents the unwanted off - resonant coupling between the cavity mode and the @xmath116 transition , with a coupling constant @xmath117 and a detuning @xmath118 [ fig .
3(a ) ] .
\(ii ) the second one requires that during performing @xmath39 and @xmath111 the cavity mode is resonant with the @xmath119 transition of the target qutrit @xmath120 the interaction hamiltonian for this basic operation is given by @xmath121 where the second term represents the unwanted off - resonant coupling between the cavity mode and the @xmath65 transition , with a coupling constant @xmath122 and a detuning @xmath123 [ fig .
3(b ) ] .
as discussed previously , the cavity mode needs to be resonant with the @xmath124 transition of the target qutrit @xmath48 during performing @xmath125 note that the hamiltonian governing this basic operation is the same as @xmath126 with a replacement of the index @xmath41 by @xmath127 it should be mentioned that the term describing the pulse - induced or the cavity - induced coherent @xmath128 transition for each qutrit is not included in the hamiltonians @xmath129 and @xmath126 , since this transition is negligible because of @xmath130 ( @xmath131 ) ( fig . 3 ) . for each of the two basic types of operations described above , the dynamics of the lossy system , composed of three qutrits ( @xmath50 ) and the cavity ,
is determined by @xmath132 + \kappa \mathcal{l}\left [ a% \right ] \notag \\ & & + \sum_{j=1}^3\left\ { \gamma _ { j21}\mathcal{l}\left [ s_{j21}^{-}\right ] + \gamma _ { j20}\mathcal{l}\left [ s_{j20}^{-}\right ] + \gamma _ { j10}\mathcal{l}% \left [ s_{j10}^{-}\right ] \right\ } \notag \\ & & + \sum_{j=1}^3\gamma _ { j,\varphi 2}\left ( s_{j22}\rho
s_{j22}-s_{j22}\rho /2-\rho s_{j22}/2\right ) \notag \\ & & + \sum_{j=1}^3\gamma _ { j,\varphi 1}\left ( s_{j11}\rho
s_{j11}-s_{j11}\rho /2-\rho s_{j11}/2\right ) , \end{aligned}\ ] ] where @xmath133 is the @xmath129 or @xmath126 above , @xmath40 represents qutrit @xmath40 ( @xmath134 ) , @xmath135
@xmath136 @xmath137 @xmath138 = \lambda \rho \lambda ^{+}-\lambda ^{+}\lambda \rho /2-\rho \lambda ^{+}\lambda /2 $ ] with @xmath139 .
in addition , @xmath140 is the decay rate of the cavity mode@xmath55 @xmath141 @xmath142 and @xmath143 are , respectively , the energy relaxation rates of the level @xmath144 of qutrit @xmath40 for the decay paths @xmath145 , @xmath146 , and @xmath147 and @xmath148 ( @xmath149 ) is the dephasing rate of the level @xmath53 ( @xmath150 ) of qutrit @xmath40 .
the fidelity of the operation is given by [ 40 ] @xmath151where @xmath152 is the output state of an ideal system ( i.e. , without dissipation and dephasing ) after a joint operation @xmath153 @xmath154 or @xmath155 is performed on the qutrit system initially in the state @xmath156 and the cavity mode initially in the vacuum state @xmath157 , which is given by @xmath158while @xmath159 is the final density operator of the whole system when the gate operations are performed in a realistic physical . is the length of the resonator , and @xmath160 is the wavelength of the resonator mode .
the two ( blue ) curved lines represent the standing wave magnetic field in the @xmath161-direction.,width=10 ] . here ,
@xmath162 , and the red squares , green diamonds , and blue circles correspond to the joint operations @xmath163 and @xmath155 given in eq .
( 4 ) , respectively . refer to the text for the parameters used in the numerical calculation.,width=9 ] we now numerically calculate the fidelity of the joint operations @xmath164 @xmath165 and @xmath155 , for a setup shown in fig .
4 . without loss of generality ,
let us consider three identical transmon qutrits . in this case , we can drop off the first subscript ( @xmath166 ) for the detunings , rabi frequencies , and coupling constants . for simplicity , we assume that @xmath167 one has @xmath168 for the transmon qutrit here [ 41 ]
. choose @xmath169 mhz , which can be reached for a superconducting transmon qutrit coupled to a one - dimensional standing - wave cpw ( coplanar waveguide ) resonator [ 42 ] .
other parameters used in the numerical calculation are as follows : @xmath170 @xmath171s , @xmath172 @xmath171s , @xmath173 @xmath171s [ 43 ] , @xmath174 @xmath171s , and @xmath175 @xmath171s . define @xmath176 and @xmath177 . for simplicity , we choose @xmath178 for the parameters chosen above , the fidelity versus @xmath179 is shown in fig . 5 , from which one can see that for @xmath180 @xmath181 a high fidelity @xmath182 @xmath183 and @xmath184 can be achieved for the joint operations @xmath185 @xmath154 and @xmath186 respectively .
we remark that the fidelity can be further increased by improving system parameters . for @xmath187 and @xmath188
we have @xmath189 ghz , @xmath190 ghz , which is achieved in experiments [ 44 ] .
@xmath191 and @xmath192 can be made to be on the order of @xmath193 @xmath171s for state - of - the - art superconducting transom devices [ 4 ] . for superconducting transmon qutrits
, the typical transition frequency between two neighbor levels is between 4 and 10 ghz [ 6,21,23,24 ] .
as an example , let us consider a cavity with frequency @xmath194 ghz .
thus , for the @xmath195 used in the numerical calculation , the required quality factor for the cavity is @xmath196 .
note that superconducting cpw resonators with a loaded quality factor @xmath197 have been experimentally demonstrated [ 45,46 ] , and planar superconducting resonators with internal quality factors above one million ( @xmath198 ) have also been reported recently [ 47 ] .
our analysis given here demonstrates that implementation of the three - qubit refined dj algorithm is feasible within the present circuit qed technique .
we have proposed a protocol for constructing the 35 @xmath0-controlled phase gates of a three - qubit refined dj algorithm , by using single - qubit @xmath199 gates , two - qubit cp gates and two - target - qubit cp gates . using this protocol ,
we have discussed how to implement the three - qubit refined dj algorithm with superconducting transmon qutrits resonantly coupled to a cavity .
our numerical calculation shows that implementation of this quantum algorithm is feasible for the current circuit qed .
finally , it is noted that this protocol is quite general and can be applied to implement the three - qubit refined deutsch - jozsa algorithm in various of physical systems .
c.p.y . was supported in part by the national natural science foundation of china under grant nos .
11074062 and 11374083 , the zhejiang natural science foundation under grant no .
lz13a040002 , and the funds from hangzhou normal university under grant nos . hsqk0081 and pd13002004 .
q.p.s . was supported by the zhejiang provincial natural science foundation of china under grant no .
this work was also supported by the funds from hangzhou city for the hangzhou - city quantum information and quantum optics innovation research team .
99 a. blais , r. s. huang , a. wallraff , s. m. girvin , and r. j. schoelkopf , cavity quantum electrodynamics for superconducting electrical circuits : an architecture for quantum computation , phys .
a * 69 * , 062320 ( 2004 ) .
j. q. you and f. nori , atomic physics and quantum optics using superconducting circuits , nature ( london ) * 474 * , 589 ( 2011 ) ; p. d. nation , j. r. johansson , m. p. blencowe , and f. nori , colloquium : stimulating uncertainty : amplifying the quantum vacuum with superconducting circuits , rev .
phys . * 84 * , 1 ( 2012 ) ; z. l. xiang , s. ashhab , j. q. you , and f. nori , hybrid quantum circuits : superconducting circuits interacting with other quantum systems , rev .
phys . * 85 * , 623 ( 2013 ) .
j. b. chang _
et al . _ ,
improved superconducting qubit coherence using titanium nitride , appl .
. lett . * 103 * , 012602 ( 2013 ) ; h.
paik _ et al .
_ , observation of high coherence in josephson junction qubits measured in a three - dimensional circuit qed architecture , phys .
lett . * 107 * , 240501 ( 2011 ) ; j. m. chow _
et al . _ , implementing a strand of a scalable fault - tolerant quantum computing fabric , nature communications * 5 * , 4015 ( 2014 ) .
c. p. yang , s. i. chu and s. han , possible realization of entanglement , logical gates , and quantum - information transfer with superconducting - quantum - interference - device qubits in cavity qed , phys .
a * 67 * , 042311 ( 2003 ) .
c. p. yang , s. i. chu and s. han , quantum information transfer and entanglement with squid qubits in cavity qed : a dark - state scheme with tolerance for nonuniform device parameter , phys .
lett . * 92 * , 117902 ( 2004 ) .
c. p. yang , s. i. chu and s. han , simplified realization of two - qubit quantum phase gate with four - level systems in cavity qed , phys .
a * 70 * , 044303 ( 2004 ) ; j. of phys .
: cond . mat .
* 16 * , 1907 ( 2004 ) ; x. l. he _
et al . , _ quantum logical gates with four - level superconducting quantum interference devices coupled to a superconducting resonator , phys .
a * 82 * , 024301 ( 2010 ) .
p. zhang , z. d. wang , j. d. sun , and c. p. sun , holonomic quantum computation using rf superconducting quantum interference devices coupled through a microwave cavity , phys .
rev . a * 71 * , 042301 ( 2005 ) . k. h. song , z. w. zhou , and g. c. guo , quantum logic gate operation and entanglement with superconducting quantum interference devices in a cavity via a raman transition , phys .
a * 71 * , 052310 ( 2005 ) .
c. p. yang , y. x. liu , and f. nori , phase gate of one qubit simultaneously controlling n qubits in a cavity , phys .
a * 81 * , 062323 ( 2010 ) ; c. p. yang , s. b. zheng , and f. nori , multiqubit tunable phase gate of one qubit simultaneously controlling n qubits in a cavity , phys .
a * 82 * , 062326 ( 2010 ) .
p. j. leek , s. filipp , p. maurer , m. baur , r. bianchetti , j. m. fink , m. goppl , l. steffen , and a. wallraff , using sideband transitions for two - qubit operations in superconducting circuits , phys .
b * 79 * , 180511(r ) ( 2009 ) .
p. w. shor , algorithms for quantum computation : discrete logarithms and factoring , in proceedings of the 35th annual symposium on foundation of computer science ( ieee computer society , los alamos , ca , 1994 ) , p. 124 .
d. deutsch , quantum theory , the church - turing principle and the universal quantum computer , proc .
london scr .
a * 400 * , 97 ( 1985 ) ; d. deutsch , quantum computational networks , proc .
london scr .
a * 425 * , 73 ( 1989 ) .
m. neeley , m. ansmann , r. c. bialczak , m. hofheinz , n. katz , e. lucero , a. oconnell , h. wang , a. n. cleland , and j. m. martinis , process tomography of quantum memory in a josephson - phase qubit coupled to a two - level state , nature phys . *
4 * , 523 ( 2008 ) .
j. koch , t. m. yu , j. gambetta , a. a. houck , d. i. schuster , j. majer , a. blais , m. h. devoret , s. m. girvin , and r. j. schoelkopf _
_ , _ _ charge - insensitive qubit design derived from the cooper pair box , phys . rev .
a * 76 * , 042319 ( 2007 ) .
l. dicarlo , m. d. reed , l. sun , b. r. johnson , j. m. chow , j. m. gambetta , l. frunzio , s. m. girvin , m. h. devoret , and r. j. schoelkopf , preparation and measurement of three - qubit entanglement in a superconducting circuit , nature ( london ) * 467 * , 574 ( 2010 ) .
p. j. leek , m. baur , j. m. fink , r. bianchetti , l. steffen , s. filipp , and a. wallraff , cavity quantum electrodynamics with separate photon storage and qubit readout modes , phys .
104 * , 100504 ( 2010 ) . | we propose a protocol to construct the 35 @xmath0-controlled phase gates of a three - qubit refined deutsch - jozsa ( dj ) algorithm , by using single - qubit @xmath1 gates , two - qubit controlled phase gates , and two - target - qubit controlled phase gates .
using this protocol , we discuss how to implement the three - qubit refined dj algorithm with superconducting transmon qutrits resonantly coupled to a single cavity .
our numerical calculation shows that implementation of this quantum algorithm is feasible within the present circuit qed technique
. the experimental realization of this algorithm would be an important step toward more complex quantum computation in circuit qed . |
in recent years , various observational evidences indicate that particles in snrs can be accelerated up to relativistic energies . for example , x - ray observations show that electrons can be accelerated to about 10 tev in the shell of an snr ( koyama et al .
1995 ; reynolds 1996 ; bamba et al .
2003 , 2005a , 2005b ) .
tev @xmath2-ray observations indicate that both electrons and protons can be accelerated to several tens tev in young shell - type snrs such as rx j1713.7@xmath13946 ( enomoto et al .
2002 ; aharonian et al . 2006a ) and rx j0852.0@xmath14622 ( katagiri et al . 2005 ; aharonian et al . 2005 ) .
more recently , a survey of the inner part of the galactic plane in vhe @xmath2-ray band has been performed with _
hess _ cherenkov telescope system .
aharonian et al .
( 2006b ) presented detailed spectral and morphological information of some new sources discovered in the _ hess _ survey along with some discussions on possible counterparts in other wavelength bands .
some sources discovered in the survey can be associated with old snrs .
for example , hess j1804@xmath1216 has a flux of nearly 25% of the flux of the crab nebula with a photon index of about 2.72 .
the source can be associated with either the shell - type snr [email protected] or the young vela - like pulsar psr j1803@xmath12137 ( aharonian et al .
furthermore , fatuzzo , melia , & crocker ( 2006 ) found that the snr [email protected] can plausibly account for all known radiative characteristics of hess j1804@xmath1216 .
another case is hess j1834@xmath1087 , which is an extended tev source discovered in the _ hess _ survey of the inner galaxy in vhe @xmath2-ray band .
the source , spatially coincident with the snr [email protected] , has a @xmath2-ray flux of 8% of the flux of the crab nebula above 200 gev , and the @xmath2-ray spectrum can be expressed by a power law with a photon index of @xmath3 ( aharonian et al .
the new _ xmm - newton _ observation reveals diffuse x - ray emission within the _ hess _ source and suggests an association between the diffuse x - rays and the vhe @xmath2-rays , moreover , [email protected] is estimated to be an old snr with a distance of @xmath4 kpc and an age of @xmath5 yr ( tian et al .
2007 ) .
theoretically , it is believed that the non - thermal photon spectrum for a given snr is produced through electron / positron synchrotron radiation , bremsstrahlung and inverse compton scattering as well as @xmath6 decay from the p - p interactions .
the question is , however , whether the origin of the non - thermal photons for young snrs is different from that for old ones .
sturner et al . (
1997 ) presented a temporal evolution model to calculate the non - thermal particle and photon spectra from shell - type snrs , and zhang & fang ( 2007 ) modified the model to investigate the possible hadronic contributions to tev @xmath2-ray emission from young snrs .
for the young snrs , the secondary particles produced via the p - p interaction have a negligible contribution to the non - thermal photon spectrum ( e.g. zhang & fang 2007 ) .
on the other hand , yamazaki et al . ( 2006 ) studied the emission from old snrs and found that proton acceleration could be efficient enough to emit tev @xmath2-rays both at the shock of the snr and at the giant molecular cloud .
they presented that the energy flux ratio @xmath7 can be more than @xmath8 for some snrs , and such sources may be the origins of the recently discovered unidentified tev sources . for the old snrs ,
the energies of the primary electrons accelerated by the shock wave are limited by the synchrotron cooling , then the roll - off energy of the synchrotron emission of the primary electrons is much smaller than that for the young snrs ( sturner et al . 1997 ; yamazaki et al . 2006 ; zhang & fang 2007 ) .
however , the secondary @xmath0 pairs can be produced from the p - p collisions in an snr for a long time due to the large energy loss time for the p - p interaction .
therefore , the contributions of secondary electrons and positrons to the non - thermal emission can become prominent for the old snrs .
motivated by above discussions , we study the non - thermal photon spectra from old snrs based on the time - dependent model in which the contribution of the secondary @xmath0 pairs is included ( see 2.2 ) .
it should be noted that our model is different from that given by yamazaki et al .
( 2006 ) , the latter assumes a steady particle distribution .
this paper is organized as follows .
the temporally evolving model of the non - thermal particle and photon spectra are discussed in section [ sec : model ] , including the snr evolution , shock acceleration , and various photon production mechanisms involved in the model .
the results from the applications of the model to two old snrs , [email protected] and [email protected] , are shown in section [ sec : application ] .
the main conclusions and some discussions are given in section [ sec : discussion ] .
@xcite presented a temporally - evolving model for the non - thermal particle and photon spectra at different stages in the lifetime of a shell - type snr .
zhang & fang ( 2007 ) modified the model to explain the multi - wavelength observations of the young snrs and shown that the tev @xmath2-rays from the two young shell - type snrs rx j1713.7@xmath13946 and rx j0852.0@xmath9 have hadronic origin . in this paper , we include the evolution of secondary @xmath0 pairs to model the multi - wavelength non - thermal emission from old snrs since the secondary @xmath0 pairs can have significant contributions to the non - thermal emission from old snrs . in the analytical model of the shock dynamics of an snr expanding into a uniform ambient medium with density @xmath10
, the snr evolves through three stages : the free expansion stage , the sedov stage , and the radiative stage ( sturner et al .
1997 ; yamazaki et al .
2006 ) . assuming the initial explosion energy of the snr is @xmath11 ergs , and the initial velocity of the shock is @xmath12 , the free expansion stage ends when @xmath13 yr , and the sedov stage ends at @xmath14 yr ( blondin et al .
1998 ) , where @xmath15 , @xmath16 is the hydrogen density in the local interstellar medium , @xmath17 is the mean atomic weight of the interstellar medium assuming 1 helium atom for every 10 hydrogen atoms , and @xmath18 is the mass of hydrogen . the shock velocity @xmath19 can be expressed as ( e.g. yamazaki et al .
2006 ) @xmath20 the evolution of the shock radius is estimated by @xmath21 and we assume @xmath22 cm s@xmath23 in this paper . it should be noted that the accumulated energy in non - thermal protons for an snr is usually smaller than @xmath24 of the initial explosion energy ( see details in 2.1 ) , and the energy diminishes very slowly in the radiative phase when the thermal energy in the snr experiences significant loss . at late times , the relativistic protons begin to contribute significantly to the post - shock pressure when most thermal energy of the snr is drained and then the expansion law should be revised correspondingly .
however , the expansion law will be influenced significantly by the accelerated protons at very late times and the non - thermal emission from old snrs does not depend heavily on the expansion law .
therefore , we still assume the expansion law expressed by eq .
( [ vst ] ) is always valid in this paper .
it is generally believed that the non - thermal charged particles are produced through diffusive shock acceleration ( blandford & eichler 1987 ; ellison , jones & reynolds 1990 ) . neglecting the nonlinear effects and assuming a compression ratio near 4 , the particle spectrum from the diffusive acceleration process is a power law with spectral index @xmath25 . for the particles accelerated in the shock of an snr
, the power law particle spectra do not extend to infinite energy and must be truncated by three mechanisms , namely , the finite age of the snr , energy - loss processes , and free escape from the shock region ( sturner et al .
1997 ) . the maximum kinetic energy of electrons and protons , @xmath26 and @xmath27 , can be calculated by using eqs .
( 3 ) , ( 4 ) , ( 5 ) in sturner et al . (
1997 ) , and we neglect the exact formulae here .
zhang & fang ( 2007 ) studied the non - thermal emission from three young snrs , and concluded that the model in the paper can explain the multi - wavelength observations for the three young snrs with the maximum wavelength of mhd turbulence @xmath28 cm and other appropriate parameters .
@xmath29 is also set to @xmath30 cm in this paper .
it should be pointed out that a time - dependent kinetic equation could be solved in order to obtain the energy spectrum of the accelerated particles ( berezhko et al .
1996 ; sturner at al .
1997 ) . here , following sturner at al .
( 1997 ) , we approximate the volume - averaged emissivity , @xmath31 , of the shock accelerated electrons and protons by @xmath32^{-[(\alpha+1)/2]}\\ & & \times(e+m_ec^2)\exp(-\frac{e}{e_{e,~\rm{max}}(t)})\;\;,\end{aligned}\ ] ] and @xmath33^{-[(\alpha+1)/2]}\\ & & \times(e+m_pc^2)\exp(-\frac{e}{e_{p,~\rm{max}}(t)})\;\;,\end{aligned}\ ] ] respectively , where @xmath34 is the particle kinetic energy , and @xmath35 and @xmath36 are electron mass and proton mass , the index @xmath37 , and we use @xmath38 in this paper . the function @xmath39 is given by ( sturner at al . 1997 ) @xmath40 & t\leq t_{\rm{rad } } \\ 0 & t >
t_{\rm{rad}}\;. \end{array } \right . \label{gt}\ ] ] factors @xmath41 and @xmath42 are used to normalize the particle spectra so that the total amount of the kinetic energy contained in the injected electrons and protons is @xmath43 , where @xmath44 represents the efficiency that the kinetic energy of the ejecta with initial mass @xmath45 and initial velocity @xmath12 is converted into the kinetic energy of both the electrons and the protons , @xmath46 and @xmath47 is the snr volume . in eq .
( [ etot ] ) , we need to introduce the parameter @xmath48 in order to determine @xmath41 and @xmath42 . obviously , these two quantities depend on @xmath49 and @xmath50 .
the ratio of electrons to protons , @xmath50 , is a key parameter when modeling the multi - wavelength non - thermal emission from an snr .
the measured ratio from the cosmic ray observations at the earth is about 0.013 at 10 gev ( gassier 1990 ) .
however , the measured value can not represent the ratio for a given snr due to the cosmic ray transport effect in the galaxy .
sturner et al . (
1997 ) obtained @xmath51 by assuming that the kinetic energy of the injected electrons is the same as that of the injected protons .
yamazaki et al . ( 2006 ) studied the tev emission from the old snrs in the steady state , they used the observed ratio of tev @xmath2-ray ( 110 tev ) to x - ray ( 210 kev ) energy flux , @xmath52 , for some snrs to test the values of @xmath50 , and found that the predicted @xmath52 for young snrs is similar to the observed one when @xmath53 , and that the predicted values of @xmath52 for the old snrs are not changed even if @xmath50 varies more than one order of magnitude because the primary electrons do not contribute to tev and x - ray emission in their model .
moreover , berezhko , ksenofontov & v@xmath54lk ( 2006 ) investigated the properties of kepler s snr using the nonlinear kinetic theory of cosmic ray acceleration in the snr .
they treated @xmath50 as a parameter quantitatively determined by comparison with the synchrotron observations and obtained that @xmath55 for the snr .
in fact , the value of @xmath50 for a given source is usually uncertain now , and it can be limited by comparison of the resulting spectrum with the multi - wavelength observations for an snr ( zhang & fang 2007 ) . in this paper
, @xmath49 is set to 0.1 and @xmath50 is treated as a parameter limited by comparison of the calculated results with the observations for a given snr . following sturner et al . ( 1997 ) , assuming that an snr interior is homogeneous , with a constant density @xmath56 and a magnetic field strength @xmath57 , @xmath58 and @xmath59 are used to represent the differential densities of the accelerated electrons and protons , respectively
. we can use @xmath60 and @xmath61 to calculate the direction- and volume - averaged electron intensity @xmath62 and proton intensity @xmath63 at each moment during the snr lifetime .
fokker - planck equations for both electrons and protons in energy space are used to find solutions , which are given by ( sturner et al .
1997 ) @xmath64\\\nonumber & & { } + \frac{1}{2}\frac{\partial^2}{\partial e^2_e}\left[d(e_e , t)n_e(e_e , t)\right]\\ & & { } + q_e(e_e , t ) -\frac{n_e(e_e , t)}{\tau}\;\ ; , \label{net}\end{aligned}\ ] ] and @xmath65\\ \nonumber & & { } + \frac{1}{2}\frac{\partial^2}{\partial e^2_p}\left[d(e_p , t)n_p(e_p , t)\right]\\ & & { } + q_p(e_p , t)-\frac{n_p(e_p , t)}{\tau}\;\ ; , \label{npt}\end{aligned}\ ] ] respectively , where the terms on the right - hand sides in eqs .
( [ net ] ) and ( [ npt ] ) represent systematic energy losses , diffusion in energy space , the particle source function and catastrophic energy loss .
following sturner et al .
( 1997 ) , we solve the above equations using a crank - nicholson finite difference scheme . in eqs .
( [ net ] ) and ( [ npt ] ) , @xmath66 , @xmath67 , @xmath68 and @xmath69 can be calculated with the formulae in 2.2 in zhang & fang ( 2007 ) , the exact expressions are neglected in this paper . the secondary electrons and positrons can be produced when the accelerated protons collide with the ambient matter via the p - p interaction .
those secondary particles evolve with the snr , and can contribute to the multi - waveband non - thermal emission from the snr as the primary electrons .
so the source term for electron can be represented as @xmath70 where @xmath71 @xmath72 @xmath73 is a enhancement factor for collisions involving heavy nuclei in an snr ( dermer 1996 ; sturner et al .
1997 ) , @xmath74 and @xmath75 are the differential cross section for electrons and positrons produced via p - p interaction , respectively
. since the accelerated protons colliding with the ambient medium ( i.e. p - p interaction ) can produce pions ( @xmath76 and @xmath6 ) , in which a proton loses @xmath77 of its energy per pion - producing collision , there exists the catastrophic loss for the protons .
sturner et al . (
1997 ) treated this process as an escape from the system , and the timescale for this catastrophic loss is @xmath78^{-1}\;\ ; , \label{tpp}\ ] ] where @xmath79 is the inelastic cross section for p - p interaction , which is ( kelner et al .
2006 ) @xmath80\;\;\mbox{mb } , \label{sigmapp}\ ] ] where @xmath81 , @xmath82 is the threshold energy of production of @xmath83-mesons .
we can solve the eqs .
( [ net ] ) and ( [ npt ] ) to get the volume- and direction - averaged intensities for electrons and protons at a given time , and then the multi - waveband non - thermal photon spectra can be calculated .
the non - thermal radiation processes of the accelerated particles involved in an snr are synchrotron radiation , bremsstrahlung , inverse compton scattering for leptons including electrons and positrons , and the p - p interaction for protons
. accelerated electrons can emit photons via synchrotron radiation as interacting with the ambient magnetic field in an snr .
the emissivity for synchrotron radiation is given by @xmath84 where @xmath85 is the electron pitch angle , @xmath86 , and @xmath87 with @xmath88 , where @xmath89 is a modified bessel function of order 5/3 .
for the electron - electron and electron - nucleon bremsstrahlung radiation , the emissivity is ( sturner et al . 1997 ) @xmath90 where @xmath91 and @xmath92 are correction factors for the presence of helium ( @xmath93 and @xmath94 ) , @xmath95 and @xmath96 represent the differential electron - electron and electron - nucleon bremsstrahlung cross sections . here , we use the approximate formulae given in baring et al .
( 1999 ) .
we use the full klein - nishina cross section for the relativistic electrons to tackle the inverse compton process , the emissivity can be expressed as @xmath97 where , @xmath98 is the target photon energy , @xmath99 is the number density of soft photon component @xmath100 with energy density @xmath101 and temperature @xmath102 and is given by @xmath103 @xmath104/2 $ ] is the lowest energy that electrons can scatter a target photon with energy @xmath98 to energy @xmath105 .
function @xmath106 is given by @xmath107,\end{aligned}\ ] ] with @xmath108 , @xmath109 with @xmath110 and @xmath111 .
kamae et al .
( 2006 ) presented a series of accurate , convenient parameterized formulae to calculate the spectra of stable secondary particles ( @xmath2 , @xmath0 , @xmath112 , @xmath113 , @xmath114 , @xmath115 ) produced in the p - p interactions , which greatly facilitate calculations involving the p - p interaction in astronomical environments .
the formulae were derived from the up - to - date p - p interaction model given by kamae et al .
( 2005 ) , which incorporates the logarithmically rising inelastic cross section , the diffraction dissociation process , and the feynman scaling violation .
the functional formula reproducing the secondary particle spectra include the non - diffractive , diffractive , and resonance - excitation components . for the nondiffractive process , the differential inclusive cross section to produce a secondary particle
is given as @xmath116 where @xmath117 is the energy of the secondary particle , @xmath118 , @xmath119 is the formula representing the nondiffractive cross section , and @xmath120 is the formula to approximately enforce the energy - momentum conservation limits : @xmath121 @xmath122 + 1}\nonumber \\&&\times \frac{1}{\exp\left[w_{\rm{nd , h}}(x - l_{\rm{max}})\right]+1}\;\;,\end{aligned}\ ] ] where @xmath123 and @xmath124 are the lower and upper kinematic limits imposed and @xmath125 and @xmath126 are the widths of the kinematic cutoffs , @xmath127 for all secondary particles , and the other parameters are given in table [ lmax ] ( kamae et al .
2006 ) .
.kinematic limit parameters for the nondiffractive process ( kamae et al .
[ cols="^,^,^,^",options="header " , ] the differential inclusive cross section to produce a secondary particle for the diffractive process can be represented as @xmath128 where @xmath129 represents the diffractive cross section , and @xmath130 is a function to enforce the energy - momentum conservation : @xmath131 @xmath132 here @xmath133 , @xmath134 . for the resonance - excitation processes ,
the function is @xmath135 @xmath136 represents the cross section , and @xmath130 , which is same as that for the diffraction process , enforces the energy - momentum conservation : @xmath137 a renormalization factor , @xmath138 , must be multiplied to the final spectrum to ensure that the parameterized model reproduces the experimental @xmath6 multiplicity after the readjustment in the resonance - excitation region of @xmath139 .
the exact expressions for @xmath138 and other parameters involved in the formulae are shown in kamae et al .
finally , the differential cross section for the secondary particles can be expressed as @xmath140 the resulting particle intensities calculated with the model are shown in fig .
[ fig : dtinten ] .
the parameters we choose are @xmath141 , @xmath142 , @xmath143 , @xmath144 , for ages of 2000 , 8000 , 15000 , 30000 yr . for these parameters , the primary particle sources turn off at @xmath145 yr . the turnover at the low energy region for both electrons and protons is due to coulomb loss , and the high - energy turnover for electrons is due to synchrotron loss .
more importantly , the primary electrons dominate the secondary @xmath0 pairs for a young age , however , the secondary @xmath0 pairs become more and more important and eventually surpass the primary electrons as the age of the snr increases , the reason is that the sources of the primary particles are cut off in the radiative phase of the snr and the primary electrons experience strong synchrotron energy loss , however , the energy loss for the protons is negligible due to the small cross section for the p - p interaction , and then the secondary particles can be produced continuously for a long time . we show the resulting multi - wavelength non - thermal spectra in fig .
[ fig : dtpu ] .
the distance to the snr is chosen to be 2 kpc , and other parameters are same as those in fig . [
fig : dtinten ] .
it is obvious that the emission from radio to x - ray band for the snr comes from the synchrotron radiation of leptons , and the high - energy photons can from bremsstrahlung and inverse compton scattering of leptons , and meson decay due to the hadronic interactions . more importantly , ( 1 ) the roll - off energy of the synchrotron radiation of the primary electrons deceases quickly when the snr ages in the radiative phase , however , the synchrotron emission of the secondary @xmath0 pairs become more and more important and eventually surpass the synchrotron radiation from the primary electrons when the snr ages due to the secondary @xmath0 pairs can be produced for a long time via the p - p interaction ; ( 2 ) most vhe @xmath2-rays from the snr are of hadronic origin with those parameters , and the hadronic emission from an snr is similar during a long time when the snr is in the radiative phase because the energy loss rate for the contained protons is small . of course
, the photons with energy smaller than 10 mev from some old snrs mainly come from the radiation of the secondary @xmath0 pairs .
aharonian et al . ( 2006b ) reported a survey of the inner part of the galactic plane in very high energy @xmath2-rays observed with _ hess _
cherekov telescope system , and fourteen previously unknown sources were detected at a significance level greater than @xmath146 after accounting for all trials involved in the search .
a source detected in the survey is hess j1804@xmath1216 , which is the brightest one of the new sources , with a steep photon index of @xmath147 and a flux of nearly @xmath148 of the flux from the crab nebula above 200 gev .
the source can be associated with the south - western part of the shell of the snr [email protected] of radius 26 arc minutes with flux 80 jy at 1 ghz ( green 2006 ) , or the young vela - like pulsar psr j1803@xmath12137 with a spin - down age of 16000 yr .
fatuzzo et al .
( 2006 ) presented a simple model for the tev emission from hess j1804@xmath1216 , they found that the snr [email protected] can plausibly account for all the known radiative characteristics of hess j1804@xmath1216 , and therefore the snr [email protected] was probably the source of the tev photons originating from the direction of hess j1804@xmath1216 . by associating [email protected] with coincident hii regions , kassim & weiler ( 1990 ) estimated a distance to the snr of @xmath149 kpc , with angular size of @xmath150 arcmin , a physical size of @xmath151pc , and an age of 15000 yr .
cui & konopelko ( 2006 ) reported the high - resolution x - ray observations taken with the chandra x - ray observatory of the field that contains the tev @xmath2-ray source hess j1804@xmath1216 , and a total of 11 discrete sources were detected . among those sources , only one , cxou j180351.4@xmath1213707 ,
is the most probable x - ray counterpart of hess j1804@xmath1216 , is significantly extended with photon index @xmath152 .
we now assume that the sources cxou j180351.4 @xmath1 213707 and hess j1804@xmath1216 associate with the snr [email protected] . to make the resulting spectrum consistent with the observations by _ chandra _ with a low energy flux and a photon index about 1.2
, the snr must be in the radiative phase when the roll - off energy of the synchrotron radiation of the primary electrons decreases quickly and the x - rays detected by chandra x - ray observatory are most probably from the bremsstrahlung radiation or inverse compton scattering of leptons , then @xmath16 should be greater than @xmath153 @xmath154 for an age of 15000 yr .
we now describe the detailed process in modeling the multi - waveband spectrum for the snr .
first of all , we investigate the influence of @xmath50 on the final spectra with parameters a distance of 6 kpc , an age of 15000 yr , @xmath155 , @xmath156 , and @xmath157 for the snr [email protected] .
the calculated results are shown in fig .
[ fig : dk216 ] .
the spectrum in the x - ray band depends heavily on @xmath50 , and the value of @xmath50 must be around 0.1 in order to match the observed data in x - ray band for the snr .
[ fig : dn216 ] shows the resulting spectra from the snr for different @xmath16 and ages . for an age of 15000 yr
, the roll - off energy of the synchrotron emission is still high for @xmath158 , which conflicts with the _ chandra _ observation , and the level of the tev emission is greater than the observed one by _ hess _ both for @xmath158 and for @xmath159 ( see penal ( a ) and penal ( b ) in fig .
[ fig : dn216 ] ) .
the model results can fit the multi - band observations well with a lager age of 18000 yr , moreover , the calculated spectrum in the band 0.1 1 tev for @xmath158 fit the tev observations better than that for @xmath157 ( see penal ( c ) and penal ( d ) in fig .
[ fig : dn216 ] ) .
furthermore , the contributions to the final spectrum of different radiation components with the same parameters as those in penal ( c ) of fig .
[ fig : dn216 ] are shown in fig.[fig:216 ] .
obviously from fig.[fig:216 ] , for the snr , ( 1 ) the emission from the primary electrons dominates that from the secondary @xmath0 pairs in the entire energy band except in the narrow soft x - ray band around 0.5 kev ; ( 2 ) the detected radio emission is mainly from the synchrotron radiation of the primary electrons whereas the x - rays observed with _
are primarily produced via the bremsstrahlung of these electrons ; ( 3 ) the tev photons with energies @xmath160 tev are primarily from both the bremsstrahlung of the primary electrons and the p - p interaction of the primary protons , however , those with higher energies are mainly from the p - p interactions due to the high - energy protons collide with the ambient matter .
hess j1834@xmath1087 is another extended tev source recently discovered in the _ hess _ survey of the inner galaxy in vhe @xmath2-rays .
the source , which is spatially coincident with the snr [email protected] ( w41 ) , has a size of 12 arc minutes and a @xmath2-ray flux of 8% of the flux from the crab nebula above 200 gev , with a photon index of @xmath3 ( aharonian et al .
moreover , albert et al .
( 2006 ) presented the observations of hess j1834@xmath1087 with the major atmospheric gamma imaging cherenkov telescope ( _ magic _ ) , and the differential @xmath2-ray flux can be expressed as @xmath161 .
[email protected] is an asymmetric shell - type snr with radio and x - ray photons having been detected . in radio band ,
the snr has a spectral index of 0.5 and a flux of 70 jy at 1 ghz ( green 2006 ) .
moreover , tian et al .
( 2007 ) reported new h i observations from the vla galactic plane system and a new _ xmm - newton _ observation for hess j1834@xmath1087 .
they concluded that [email protected] is an old snr with a distance of @xmath4 kpc and an age of @xmath5 yr .
furthermore , a density of 6 @xmath154 was estimated around the snr from the vgps h i line emission associated with the snr ( tian et al .
the new _ xmm - newton _ observation indicated diffuse x - ray emission within the _ hess _ source and suggested an association between the x - ray and the vhe @xmath2-ray emission .
the intrinsic x - ray spectrum from 2 to 10 kev could be characterized as a heavily absorbed power law with a flux of @xmath162 , and a photon index of @xmath163 ( see details in tian et al . 2007 ) .
we now apply the model in this paper to the old snr [email protected] , which has been observed in radio , x - ray and vhe @xmath2-ray bands .
firstly , using the similar process as that for the snr [email protected] , we find @xmath50 must be about 0.2 for the snr [email protected] with parameters , a distance of 4 kpc , @xmath155 , @xmath164 , @xmath165 , and an age of several @xmath166 yr . fig .
[ fig : dy087 ] shows the resulting spectra for different ages . for these parameters ,
the radiative age is @xmath167 yr and then the emission in each panel of fig .
[ fig : dy087 ] is in the radiative phase .
the observations in x - ray band can be explained as the synchrotron and bremsstrahlung radiation of the primary electrons for an age of 15000 yr , however , the calculated results are significantly high above the vhe @xmath2-ray observations by _ hess _ and _ magic _ ( see panel ( a ) in fig . [
fig : dy087 ] ) .
when the radiative snr ages , the cut - off energy of the primary electrons deceases quickly and the high - energy protons undergo both catastrophic energy losses due to the p - p collisions and energy - dependent diffusion .
as a result , the roll - off energy of the synchrotron radiation of the primary electrons decreases quickly and the number of the tev photons from the snr reduces too .
we can see from panel ( d ) in fig .
[ fig : dy087 ] that the radio and vhe @xmath2-ray observations can be explained by the model with an age of 80000 yr , however , the observed flux in the x - ray band can not be reproduced . to make the model results consistent with the flux of the observations in the x - ray band
, the age of the snr should not be much bigger than the radiative age of the snr .
furthermore , a smaller @xmath16 should be chosen to make the calculation spectrum in the vhe @xmath2-ray band consistent with the _ hess _ and _ magic _ observations .
finally , @xmath16 is chosen to be 1.5 @xmath154 with other parameters an age of @xmath168 yr , @xmath155 , @xmath169 , @xmath170 to model the multi - band non - thermal emission from the snr [email protected] and the resulting spectra are shown in fig .
[ fig:087 ] .
the radiative age for these parameters is @xmath171 yr and the roll - off energy of the synchrotron radiation of the primary electrons is not reduced mostly with an age of @xmath168 yr .
it is obvious from fig .
[ fig:087 ] that ( 1 ) the emission from the primary electrons dominates that from the secondary @xmath0 pairs in all bands and the emission from radio and x - ray band is mainly from the synchrotron radiation of the primary electrons ; ( 2 ) the tev photons are produced mainly via the p - p interaction .
in this paper , by taking into account the evolution of secondary @xmath0 pairs produced via the p - p interaction due to high - energy protons collide with the ambient matter in an snr , we developed a time - dependant model of multi - waveband non - thermal particle and photon spectra both for young and for old shell - type snrs .
the primary electrons accelerated directly by the shock wave usually dominate the secondary @xmath0 pairs produced from the p - p collisions for a young snr , however , the emission from the secondary @xmath0 pairs becomes more and more important and eventually surpasses the radiation from the primary electrons as the snr grows old because the source of the primary electrons is cut off and the synchrotron radiation loss is significant for electrons when the snr is in the radiative phase , whereas the secondary @xmath0 pairs can be produced for a long time due to the small cross section for the p - p interaction ( see fig .
[ fig : dtpu ] ) .
@xmath50 , which can not be directly deduced from the observations now and is usually limited by comparison of the model results with the multi - band observations for a given snr , is also a crucial parameter to determine whether the contribution to the final non - thermal emission of the secondary @xmath0 pairs dominates that of the primary electrons . for an old snr with small @xmath50 , the non - thermal photons with energies @xmath172
are usually produced mainly via the radiations of the secondary @xmath0 pairs ( see fig . [
fig : dtpu ] ) .
the model is applied to two old shell - type snrs , [email protected] and [email protected] , whose vhe @xmath2-rays are detected in the _ hess _ survey of the inner galaxy .
[email protected] is a shell - type snr with an age of @xmath173 yr .
the snr must be in the radiative phase to make the calculated results consistent with the _ chandra _ x - ray observations .
finally , a set of parameters , a distance of 6 kpc , an age of 18000 yr , @xmath155 , @xmath164 , @xmath174 , and @xmath156 is used to model the multi - wavelength non - thermal emission from the snr [email protected] .
it can be concluded that ( 1 ) the radio emission from the snr is mainly from the synchrotron radiation of the primary electrons , whereas the non - thermal photons with energies around 0.1 kev primarily come from the synchrotron radiation of the secondary @xmath0 pairs produced from the p - p collisions ; ( 2 ) the observed x - ray photons with energies ranging from 2 to 10 kev are produced mainly via the bremsstrahlung of the primary electrons ; ( 3 ) the tev photons with energies @xmath160 tev are primarily from both the bremsstrahlung of the primary electrons and the p - p collisions of the primary protons , however , those with higher energies are produced mainly through the p - p interaction ( see fig .
[ fig:216 ] ) .
another old snr is [email protected] , which is an asymmetric shell - type snr and spatially coincide with hess j1834@xmath1087 .
tian et al . ( 2007 ) reported new h i observations from the vla galactic plane system and a new _ xmm - newton_observation for hess j1834@xmath1087 , and argued that [email protected] is an old snr with a distance of @xmath4 kpc and an age of @xmath5 yr , moreover , a density of 6 @xmath154 was estimated around the snr from the vgps h i line emission associated with the snr .
we find that , using @xmath164 and an age of @xmath5 yr with other appropriate parameters , the model result agrees with the radio and vhe @xmath2-ray observations , but the calculated x - ray flux is significantly lower than the observed one given by _
xmm - newton _ ( see fig . [
fig : dy087 ] ) . a smaller age and a smaller @xmath16
are needed to make the calculation result consistent with the multi - band observations .
finally , we use parameters a distance of 4 kpc , @xmath155 , an age of @xmath168 yr , @xmath175 , @xmath169 , and @xmath170 to model the multi - band non - thermal emission from the snr [email protected] .
the modeling results can be consistent with both the radio , vhe @xmath2-ray observations and the x - ray flux with these parameters .
moreover , the emission from the secondary @xmath0 pairs is negligible and the multi - wavelength photons with those parameters for the snr mainly come from the radiation of the primary particles , and the tev photons primarily have hadronic origin ( see fig .
[ fig:087 ] ) .
the model in this paper includes the contribution of the secondary @xmath0 pairs produced from the p - p collisions to the non - thermal emission from snrs .
the secondary @xmath0 pairs can be produced for a long time and experience energy losses and diffusion , and the contribution of the secondary @xmath0 pairs to the final non - thermal emission from an snr can become more and more important and eventually surpasses that of the primary electrons when the snr ages in the radiative phase . whether the contribution of the secondary @xmath0 pairs to the final non - thermal emission is prominent or not depends heavily on the snr age , @xmath50 and @xmath16 .
these parameters can be limited by the comparison of the model result with the multi - wavelength observations for a given snr , however , the shortage of precise observations and uncertainties such as the initial eject mass and the initial velocity of the shock lessen the reliability of the calculated results
. on the other hand , for an old snr with observed tev emission , if the tev photons primarily have hadronic origin , the emission from the secondary @xmath0 pairs is usually prominent .
the morphology of the resulting spectrum for an snr in the energy range between 10 mev to 1 tev directly relates to the primary origin of the vhe photons .
fortunately , the gamma ray large area space telescope ( _ glast _ ) have a good sensitivity and resolution in the energy range between 30 mev and 300 gev , and the observations with _ glast _ will give a direct evidence on whether the tev photons from an snr mainly have hadronic origin or leptonic origin .
obviously , _ glast _ will give more limits on the parameters in the modeling process with the model in this paper for an snr , and then we can get more insights on the acceleration and photon production processes involved in snrs .
we are grateful to h. bartko for providing the _ magic _ data points and an anonymous referee for his / her very constructive comments .
this work is partially supported by a distinguished young scientists grant from the national natural science foundation of china ( nsfc 10425314 ) , nsfc grant 10463002 , nsfc grant 10778702 , and grant from the department of education of yunnan province ( 07j51074 ) .
we appreciate the use of the high performance computer facilities at yunnan university , kunming , prc .
albert j. et al . , 2006 , apj , 643 , l53 bamba a. et al . , 2003 ,
apj , 589 , 253 bamba a. et al . , 2005a , apj , 621 , 793 bamba a. , yamazaki r. , hiraga j. s. , 2005b , apj , 632 , 294 baring m. g. et al . , 1999 ,
apj , 513 , 311 berezhko e. g. , 1996 , arep , 40 , 155 berezhko , e. g. , ksenofontov , l. t. , v@xmath54lk , h. j. , 2006 , a&a , 452 , 217 blandford r. d. , eichler d. , 1987 , phys . rep . , 154 , 1 green d. a. , 2006 , a catalogue of galactic supernova remnants ( 2006 april version ) . astrophysics group , cavendish laboratory , cambridge , united kingdom kassim n. e. , weiler , k. w. , 1990 , nature , 343 , 146 katagiri h. et al . , 2005 , apj , 313 , 842 | we study the non - thermal emission from old shell - type supernova remnants ( snrs ) on the frame of a time - dependent model . in this model , the time - dependent non - thermal spectra of both primary electrons and protons as well as secondary electron / positron ( @xmath0 )
pairs can be calculated numerically by taking into account the evolution of the secondary @xmath0 pairs produced from proton - proton ( p - p ) interactions due to the accelerated protons collide with the ambient matter in an snr .
the multi - wavelength photon spectrum for a given snr can be produced through leptonic processes such as electron / positron synchrotron radiation , bremsstrahlung and inverse compton scattering as well as hadronic interaction .
our results indicate that the non - thermal emission of the secondary @xmath0 pairs is becoming more and more prominent when the snr ages in the radiative phase because the source of the primary electrons has been cut off and the electron synchrotron energy loss is significant for a radiative snr , whereas the secondary @xmath0 pairs can be produced continuously for a long time in the phase due to the large energy loss time for the p - p interaction .
we apply the model to two old snrs , [email protected] and [email protected] , and the predicted results can explain the observed multi - wavelength photon spectra for the two sources .
[ firstpage ] radiation mechanisms : non - thermal supernova remnants
ism : individual : [email protected] , [email protected] gamma - rays : theory |
in the strong - coupling regime for matter and light , radiative decay of a material excitation gives way to coupled oscillations of the polarization of the matter and of the electromagnetic field .
the quasiparticles corresponding to such coupled modes are known as polaritons @xcite .
the classic realization of polaritons is excitons in a bulk semiconductor coupled to photons in free space , as discussed many years ago by hopfield@xcite . in this example
, wavevector conservation ensures that each exciton is coupled only to a single mode of the electromagnetic field , leading to the formation of polaritons which are superpositions of a single exciton and photon .
recently , there has been a lot of interest in polaritons formed from photons confined in cavities : such cavity polaritons have now been observed for confined photons coupled to atoms@xcite , to two - dimensional excitons in quantum wells@xcite , to bulk excitons @xcite , to excitons in films of organic semiconductors @xcite , and to charged exciton complexes @xcite
. since polaritons are photons coupled to other excitations , they are bosons , and so are candidates for bose condensation@xcite .
recent observations@xcite of bosonic behavior for cavity polaritons have renewed interest in this idea
. however , there is a conceptual difficulty with a bose condensate of polaritons : while polaritons are usually considered in the low - excitation linear regime , bose condensates are stabilized by nonlinearities @xcite . for cavity polaritons ,
there is also the following more practical difficulty .
bose condensates are characterized by coherence , and in a polariton condensate this coherence will appear in the photons . given this ,
how is a polariton condensate distinct , conceptually and observationally , from a laser ? in this paper we address these problems by developing a theory of polariton condensation in the dicke model@xcite .
this nonlinear model of confined photons coupled to matter is one of the basic models of laser physics .
it allows us to go beyond the conventional linear - response concept of a polariton , including effects due to finite excitations of the matter in the cavity . in the language of semiconductors ,
it includes a `` saturation '' or `` band - filling '' nonlinearity , produced by the fermionic internal structure of the excitons .
polaritons are not conserved particles , so there is ultimately no equilibrium condensate .
we may , however , treat polaritons as conserved particles if their lifetime is much longer than the time required to achieve thermal equilibrium at a fixed polariton number .
we will study this quasi - equilibrium regime , since it is in this regime that bose condensation is well - defined @xcite . in section [ modelsec ]
, we introduce the model , and explain how the concept of a polariton can be generalized to allow for the nonlinearity of the model .
we then present , in sections[variational ] and [ variationalresults ] , a simple variational technique for calculating the ground state of the model at a fixed density of polaritons . in section[largensec ] we investigate the thermodynamics of the model using an alternative technique based on functional integrals .
this technique demonstrates that the variational approach is essentially exact , and allows us to consider finite temperatures . in section[phasediag ]
, we use the expressions derived by the functional integral method to study the phase diagram for condensation , while in section[exspec ] , we use these expressions to calculate the excitation spectra of the model . these excitation spectra provide a physical picture of the transition to the condensed state , and determine the absorption spectrum of the cavity .
finally , in section[conclusions ] we discuss our conclusions .
the functional integral approach to the thermodynamics of our model has already been the subject of a brief report @xcite .
we extend that earlier report to allow for a distribution of the energies of the electronic excitations , i.e. inhomogeneous broadening , which is significant in many potential realizations of the polariton condensate .
the dicke model @xcite consists of a set of @xmath0 two - level oscillators coupled to a single mode of the electromagnetic field by the dipole interaction .
the two - level oscillators do not interact with one another , except through their common coupling to the electromagnetic field .
we generalize the original dicke model to include an energy distribution of the two - level oscillators . making the rotating - wave approximation(see e.g. refs . ) , we consider the hamiltonian @xmath1 here the two - level oscillators are indexed by the variable @xmath2 , which is summed over .
we use a fermionic representation for the two - level oscillators , describing each one in terms of a pair of fermions with annihilation operators @xmath3 and @xmath4 . for brevity
we suppress the index @xmath2 on the fermionic operators .
the fermions are subject to the single - occupancy constraint @xmath5 on each site .
@xmath6 is the annihilation operator for the cavity mode , @xmath7 is the energy of the @xmath8 two - level oscillator , and @xmath9 is the strength of the dipole coupling .
the hamiltonian ( [ ham ] ) is a simple model of a three - dimensional cavity ( photonic dot ) @xcite , containing localized , physically separated electronic excitations . although simplified , it is a useful starting point for many systems .
for example , each of the two - level oscillators could describe the presence or absence of a localized exciton in a given eigenstate of the disorder potential in a disordered quantum well , on a given molecule in an organic film , or trapped on a particular impurity . the restriction to singly occupied states models the hard - core repulsion produced by the fermionic structure of such excitations .
it describes spinless excitations which are localized on the scale of the ( exciton ) bohr radius .
it is straightforward to generalize our calculations to allow for a finite number of excitations on each site , i.e. traps which are bigger than the bohr radius , and so can hold several excitons . @xmath10 couples the photons to excitations of the two - level oscillators , created by the operator @xmath11 .
if @xmath12 then the excited states which are created by @xmath13 from the vacuum are eigenstates of the bare hamiltonian @xmath14 . if furthermore @xmath0 is large and the two - level oscillators are near to their ground state , @xmath15 then @xmath13 is approximately a bosonic creation operator , and the hamiltonian ( [ ham ] ) becomes two coupled boson oscillators .
polaritons are usually presented as the eigenstates of such a model .
away from the low - excitation limit ( [ lowdensity - regime ] ) , @xmath13 is not a bosonic creation operator , and the conventional description of polaritons breaks down . to go beyond the low - excitation limit
, we generalize the concept of a polariton to be the quantum of excitation of the coupled matter - light system .
the polariton number is then the total number of photons and excited two - level oscillators , @xmath16 which is a conserved quantity for the model ( [ ham ] ) .
( [ exnumber - definition ] ) defines the operator @xmath17 , which we refer to as the excitation number .
we define a corresponding excitation density @xmath18 , which is the total number of photons and electronic excitations , per two - level oscillator , minus one - half .
since the numbers of photons and electronic excitations are positive , the lowest excitation density is @xmath19 .
since the number of electronic excitations is always less than @xmath0 , the electronic contribution to @xmath20 is always less than @xmath21 .
the thermal equilibrium of the dicke model , in the absence of an externally created population of polaritons , has been studied extensively since the pioneering exact solution of hepp and lieb@xcite .
these authors showed that , even in the absence of external excitation , the dicke model has a phase transition to a bose condensed state .
such an equilibrium condensate is a static , coherent state of photons : it is a ferroelectric@xcite .
here we are interested in the thermal equilibrium of a population of polaritons : the quasi - equilibrium problem posed by ( [ ham ] ) at a fixed excitation @xmath17 .
the quasi - equilibrium condensate which we find in this regime is a time - varying generalization of the ferroelectric state discovered by hepp and lieb .
we can write down a variational state which describes the polariton condensate by noting that bose condensates are described by coherent states . this produces a variational wavefunction closely related to the bcs wavefunction used to describe superconductors and exciton condensates .
as has been stressed by comte and nozires@xcite , this class of wavefunction can describe an exciton condensate in both the low and high density limits .
it thus permits a smooth interpolation from low densities , where the excitons are simple bosons , to high densities , where their fermionic internal structure is revealed . in a similar way
, it allows us to explore the polariton condensate beyond the low - density regime in which polaritons are usually considered .
the constituents of our proposed condensate are the excitations of the whole system , counted by @xmath17 . in general
, such an excitation is a superposition of an excitation of the cavity mode and an excitation of the electronic states .
thus we take for our trial wave - function a coherent state of such a superposition : @xmath22 where the state @xmath23 has a single fermion in the lower state of each two - level oscillator .
the state ( [ becform ] ) has a finite polarization of the electronic excitations as well as a finite amplitude for the cavity field . @xmath24 and @xmath25 are the variational parameters . expanding the exponential , ( [ becform ] ) explicitly becomes a superposition of a coherent state of photons and a bcs state of the fermions , @xmath26 , and @xmath27 are the variational parameters , and @xmath28 denotes the vacuum state with no fermions in any of the levels . by construction , this variational state obeys the single - occupancy constraints ( [ constraints ] ) .
we fix the overall phase of the condensate by choosing @xmath24 to be real .
the @xmath27 have been explicitly introduced to make the @xmath29 and @xmath30 real .
they are the phase differences between the cavity field and the polarizations of the electronic states . to find the ground state of ( [ ham ] ) at fixed excitation number we minimize @xmath31 with respect to the variational parameters , subject to the normalization conditions @xmath32 .
although the overall phase of the condensate is arbitrary , the relative phases @xmath27 are not : there is only one order parameter . the relative phases @xmath27 are fixed by the last term in ( [ hamexps ] ) , the dipole coupling .
this term ensures that all the two - level oscillators which have a finite dipole moment(@xmath33 ) are mutually coherent , @xmath34 , when the energy is minimized .
it is the dipole interaction which is responsible for bose condensation , and its accompanying coherence @xcite , in the present system . setting @xmath35 and defining an intensive @xmath24 by rescaling @xmath36 , the condensate parameters are given by the real solutions with @xmath37 to @xmath38 @xmath39
was introduced as a lagrange multiplier constraining the excitation number .
it is the chemical potential for our coupled modes , and is related implicitly to the excitation density by @xmath40 eliminating @xmath41 and @xmath42 from ( [ extreqs ] ) and ( [ density ] ) we can rewrite these expressions as @xmath43 where we define @xmath44 ( [ ztextr ] ) is analogous to the bcs gap equation , with an order parameter @xmath24 .
to investigate the expressions ( [ ztextr][qpenergy ] ) , we replace the summations over sites with an integral over the energy distribution of the two - level oscillators .
we take this distribution to be a gaussian with mean @xmath45 and variance @xmath46 .
the remaining parameters in our quasi - equilibrium problem are then the excitation density , @xmath20 , and the dimensionless detuning between the energy of the cavity mode and the center of the exciton line , @xmath47 . for a gaussian density of states ,
the summation on the right of ( [ ztextr ] ) diverges as @xmath48 , and approaches zero as @xmath49 .
thus for any @xmath50 there is always a condensed solution , @xmath51 , to ( [ ztextr ] ) : the system is condensed at arbitrarily small excitation densities .
this behavior is produced by the tails of the gaussian distribution .
because of these tails , we have excitons at arbitrarily low energies , and hence also bound exciton - photon states at arbitrarily low energies .
it is impossible to populate just the excitons , because no matter how small @xmath39 is , there is always a bound state involving photons below it .
we expect that if the density of states has a lower cut - off , and is continuous at this cut - off , there would be a finite critical @xmath39 below which there is no condensed solution to ( [ ztextr ] ) .
let us investigate the dependence of @xmath39 on @xmath20 in the absence of inhomogeneous broadening , @xmath52 . at low densities , @xmath53 , @xmath39 can be obtained from ( [ ztextr ] ) . expanding this expression for small @xmath24 and comparing the leading terms , we find that @xmath39 is given by the conventional linear - response polariton energy , @xmath54 $ ] . at finite densities
we calculate @xmath39 numerically , by solving ( [ ztextr ] ) and ( [ ztexdens ] ) to determine @xmath55 .
the results are plotted in the right panel of fig .
[ zerotmurhofig ] , for @xmath56 and @xmath57 . at low densities
we are describing a condensate of conventional polaritons , and so have @xmath58 .
as the density is increased the exciton states saturate , forcing the excitations to become more photon - like .
thus the chemical potential approaches @xmath59 at high densities .
for @xmath60 the separation between the exciton - like and photon - like excitations persists to @xmath61 , where the exciton states are completely saturated .
this results in a discontinuity in @xmath39 at this point , since no further excitation can be added to the exciton states .
the dependence of @xmath39 on @xmath20 in the inhomogeneously broadened case is also illustrated in the right panel of fig.[zerotmurhofig ] .
it is qualitatively rather similar to the homogeneous case . instead of the finite intercept of the homogeneous case we now have @xmath62 as @xmath63 .
this behavior is again caused by the tails of the gaussian distribution .
to demonstrate how @xmath39 approaches the conventional polariton energy @xmath64 in the homogeneous , low - density limit , we compare the behavior of @xmath39 with the density of states for the linear - response excitations of the empty(@xmath65 ) cavity .
this density of states is the optical absorption spectrum of the cavity , and is plotted in the left panel of fig.[zerotmurhofig ] for @xmath66 and @xmath56 and @xmath57 .
we will describe how it is calculated in section [ exspec ] . at very low densities
, @xmath39 lies in the tails of the exciton distribution . with increasing density
, these states quickly saturate , producing a sharp rise in @xmath39 . as @xmath39 reaches the polariton peak , the sharp rise in the density of states for the coupled modes produces a kink in the chemical potential . in the homogeneous limit , this kink moves to zero density and corresponds to the usual polariton energy .
since the density of states at this point is infinite in the homogeneous limit , these polaritons are simple bosons .
figure [ zerotoccfig ] shows the occupation of the two - level oscillators in the polariton condensate , for @xmath67 and various densities .
the occupation number of the @xmath8 two - level oscillator is @xmath68 as is clear from the figure , this is a fermi step broadened by the interaction with the photons , just as the electronic distribution in a bcs superconductor is a fermi step broadened by the pairing interaction . the states in the broadened region of the step
have a finite dipole moment and are involved in the condensate .
the fermi step moves up through the exciton line as the excitation is increased from @xmath65 and the low - lying electronic states saturate . at very large densities
there are a large number of photons , and the fermi step is almost completely flat : rather than the electronic system completely saturating in the high density limit , it approaches half filling .
this is because the half - filled state maximizes the polarization of the electronic states and hence minimizes the dipole interaction between the excitons and the macroscopically occupied cavity mode .
careful inspection of fig .
[ zerotoccfig ] reveals that the broadening of the fermi step produced by the photons does not increase monotonically with density .
this corresponds to a non - monotonic dependence of the field amplitude , @xmath24 , on density .
this dependence is illustrated in fig.[photonoccfig ] .
the field amplitude is related to the electronic polarization by the first of the equations ( [ extreqs ] ) .
it is proportional to the electronic polarization and inversely proportional to the separation between the chemical potential and the cavity mode .
the electronic polarization depends on the density of states in the vicinity of the chemical potential(fig.[zerotoccfig ] ) ; the peak in the density of states at the center of the exciton line produces the peak in fig.[photonoccfig ] .
the variational approach of sections [ variational ] and [ variationalresults ] becomes exact in the thermodynamic limit @xmath69 .
physically , this is because it corresponds to a mean - field treatment of the interaction between electronic excitations .
this interaction , between a large number ( @xmath0 ) of electronic excitations , is mediated by a small number ( one ) of cavity modes . in a mean - field treatment of this interaction , each electronic excitation is coupled to the average field produced in the cavity by the other electronic excitations .
this becomes exact when there are a large number of electronic excitations contributing to a small number of field modes , since the fluctuations of the field are then negligible . in this section
, we develop a mean - field theory for the thermodynamics of the model ( [ ham ] ) from the functional - integral representation of the partition function . in this representation ,
the partition function can be rigorously evaluated , for large @xmath0 , using a saddle - point analysis @xcite .
from such an analysis , we derive finite - temperature generalizations of the variational expressions ( [ ztextr][qpenergy ] ) , thus demonstrating that they are rigorous in the limit of large @xmath0 .
the functional integral techniques used here have previously been used @xcite to calculate the partition function and excitation energies _ in the absence of a constraint on the polariton number _ of a simplification of the dicke model . while the hamiltonian of the model discussed in refs . is given by ( [ ham ] ) with @xmath12 , the local constraints prohibiting two fermions on the same site , ( [ constraints ] ) ,
are replaced with a global constraint .
in contrast , we retain ( [ constraints ] ) as local constraints , as well as including a distribution of @xmath70 and a constraint on the polariton number . as in sections
[ variational ] and [ variationalresults ] , we work in a grand - canonical ensemble , using a chemical potential @xmath39 to constrain the excitation number .
we consider the partition function associated with this ensemble @xmath71 the coherent - state functional - integral formalism allows us to express @xmath72 , for the model ( [ ham ] ) , as the constrained functional integral @xmath73e^{-s},\ ] ] with the action @xmath74 we have introduced a nambu spinor @xmath75 for each two - level oscillator .
the matrix @xmath76 is @xmath77 rescaling the boson field @xmath78 and transferring the fermionic integrals into the action gives @xmath79 with an effective action @xmath80 in which the @xmath81 are the matrix operators @xmath82 after rescaling the boson field , and @xmath83 denotes the trivial jacobian arising from this rescaling . for large @xmath0 ,
the dominant contribution to the partition function @xmath72 comes from those functions @xmath84 which minimize the action @xmath85 .
such functions obey the euler - lagrange equation .
for the action ( [ seff ] ) , this takes the form@xmath86 where the right - hand side of this expression is the polarization of the two - level oscillators in thermal equilibrium driven by an external field @xmath84 .
this polarization appears because the field @xmath84 modifies the eigenstates @xcite of the electronic system .
a thermal population of these new eigenstates can correspond to a finite polarization of the original fermions .
equation ( [ geneleq ] ) is a self - consistency condition : the cavity field is driven by the polarization of the fermions , which itself arises from the renormalization of the fermions produced by the photons .
assuming that the self - consistent field @xmath84 is independent of @xmath87 , we can calculate the polarization term on the right of ( [ geneleq ] ) by making a bogolubov transformation @xmath88 from the @xmath89 and @xmath90 fermions to new fermions @xmath91 and @xmath92 .
this transformation diagonalizes @xmath81 when @xmath93 and @xmath94 .
the @xmath95 and @xmath96 quasiparticles then have energies @xmath97 respectively , with @xmath98 defined by equation ( [ qpenergy ] ) . since
( [ bogt ] ) is a rotation in @xmath99 space , it preserves the single occupancy constraints .
thermally populating the new fermions in accordance with the single occupancy constraint we have @xmath100 and ( [ geneleq ] ) becomes @xmath101 equation ( [ ftextr ] ) is the finite - temperature generalization of the variational result ( [ ztextr ] ) .
this generalization is rather straightforward : we have just acquired @xmath102 factors describing the thermal occupation of the two - level oscillators .
if we remove the constraint on the polariton number , by setting @xmath103 , and set @xmath12 , then ( [ ftextr ] ) is the form originally derived by hepp and lieb @xcite for the unconstrained equilibrium of the dicke model . in that problem
, the existence of a condensate requires @xmath104 since otherwise ( [ ftextr ] ) , with @xmath103 and @xmath12 , has only the trivial solution @xmath105 .
however , it is shown in refs . and that the @xmath106 terms of the minimal - coupling hamiltonian , neglected in the model ( [ ham ] ) , modify the inequality ( [ minimumg ] ) in a way which is inconsistent with the thomas - kuhn - reich sum rule .
this sum rule requires @xmath107 , where @xmath108 is the coupling constant for the @xmath106 term , while the modified inequality ( [ minimumg ] ) reads @xmath109 since this inequality can not be satisfied , the phase transition in the unconstrained case is an unphysical artifact of the model ( [ ham ] ) .
however , we do not believe that the @xmath106 terms prevent condensation in the constrained case , because the inequality corresponding to ( [ minimumg2 ] ) will be @xmath110 and the parameter @xmath39 is not restricted by the sum rule .
let us now consider the effect of small fluctuations @xmath111 around the mean - field solution .
expanding @xmath112 to second order in a functional taylor series around the mean - field solution we have @xmath113}. \label{gaussian - flucs}\ ] ] here @xmath114 is the action evaluated on the extremal trajectory and @xmath115 is the quadratic action from the second order term in the taylor series .
@xmath115 is the effective action for small fluctuations of the electromagnetic field .
the kernel of @xmath115 , @xmath116 , is the inverse of the thermal green s function for the photons .
the integral over fluctuations in ( [ gaussian - flucs ] ) contributes a term @xmath117 to the free energy density . since the mean - field solution should be a minimum of the action
, the eigenvalues of @xmath116 should be positive .
then @xmath118 is finite as @xmath69 , there is no fluctuation contribution to the free energy density in this limit , and the mean - field theory becomes exact .
however , we have yet to check whether the solutions to ( [ ftextr ] ) are actually minima of the action or merely extrema , i.e. whether the mean - field solutions are stable against fluctuations . to check this , we will need the effective action @xmath115 , which we derive in this section . to obtain @xmath115
, we calculate the ( functional ) second derivatives of @xmath119 , and evaluate them on the extrema @xmath120 . in the frequency representation ,
the components of @xmath115 are @xmath121 d\tau , \nonumber\end{aligned}\ ] ] and @xmath122 @xmath123 and @xmath124 denote bosonic matsubara frequencies , @xmath125 is the polarization operator for the @xmath8 two - level oscillator , and the integrands are the susceptibilities of the two - level oscillators in the self - consistent field @xmath24 .
( [ qpseffinhomgeneral1 ] ) and ( [ qpseffinhomgeneral2 ] ) describe coupled fluctuations of the cavity field and the electronic polarization .
they are analogous to the dyson - gorkov - beliaev equations @xcite of the theory of superconductors and weakly - interacting bose gases . because we are working with a condensed system
there is an anomalous contribution to @xmath115 , ( [ qpseffinhomgeneral2 ] ) , from fluctuations which do not conserve the number of excitations above the condensate , i.e. those in which an excitation enters or leaves the condensate . to calculate the susceptibilities which appear in ( [ qpseffinhomgeneral1 ] ) , we rewrite them in terms of the renormalized two - level oscillators using the transformation ( [ bogt ] ) , transform to the schrdinger representation , and take thermal and quantum - mechanical averages over the renormalized eigenstates .
this gives @xmath126&=&\beta \sum_{\omega } \left ( \begin{array}{cc } \bar{\delta\psi}(\omega ) & \delta\psi(-\omega ) \end{array } \right){\mathcal{g}}^{-1}\left ( \begin{array}{c } \delta\psi(\omega ) \\
\bar{\delta\psi}(-\omega ) \end{array } \right ) , \nonumber
\\ { \mathcal{g}}^{-1}&=&\left(\begin{array}{cc } k_{1 } & k_{2 } \\
k_{2}^{\ast } & k_{1}^{\ast } \end{array } \right ) , \label{thermalinvgf } \\
k_{1}&=&i\omega+\tilde{\omega}_{c}+\frac{g^2}{n}\sum_{n}\bigg[\frac{1}{e_{n}}\tanh \left(\beta e_{n}\right ) \nonumber \\ & & \times \frac{i\tilde{\varepsilon}_{n}\omega-2\tilde{\varepsilon}_{n}^2-g^2|\lambda|^2}{\omega^2 + 4e_{n}^2}+\delta_{\omega}\alpha_{n } k_{2}&=&\frac{g^4\lambda^2}{n}\sum_{n}\bigg[\frac{1}{e_{n}(\omega^2 + 4e_{n}^2)}\tanh\left(\beta e_{n}\right ) \nonumber \\ & & + \delta_{\omega } \alpha_{n } \bigg ] , \nonumber \\ \alpha_{n}&=&-\frac{\beta}{4 e_{n}^2 } { \rm{sech}}^2\left(\beta e_{n}\right ) . \nonumber\end{aligned}\ ] ] note that , for the condensed state , @xmath116 takes different forms at @xmath127 and at finite @xmath123 .
this is because thermal fluctuations at @xmath127 include both fluctuations of the order parameter and quasiparticle excitations @xcite .
only the latter appear at finite @xmath123 . in the normal state ,
@xmath105 , the effective action simplifies to @xmath128\delta\psi ( \omega).\end{aligned}\ ] ] we now use the expressions ( [ qpseffinhom][qpseff0 ] ) to investigate the nature of the extrema when @xmath12 .
considering first a condensed solution , @xmath51 , we use the extremal equation ( [ ftextr ] ) to eliminate @xmath129 from the matrix @xmath116 .
the eigenvalues of the resulting matrix are all strictly positive provided that @xmath130 , except for a single zero eigenvalue at @xmath131 . from ( [ ftextr ] ) we see that the condensed solutions always have @xmath130 .
thus we conclude that , at a condensed solution , the action has a minimum in all but one direction , and is locally flat in this one direction .
we show in appendix [ goldstappendix ] that the single zero eigenvalue describes a change in the overall phase of the condensate .
it is the goldstone mode corresponding to the broken gauge symmetry of the condensate . because we are considering a broken symmetry state
, we should not integrate over these fluctuations when calculating the partition function .
since the other eigenvalues of @xmath116 are always positive for the condensed solutions , these solutions are stable against physical fluctuations , and the mean - field theory is exact , we give a formal demonstration that the zero mode does not contribute to the free energy density as @xmath69 , so that the presence of the zero mode does not invalidate the discussion of subsection[fluc - nocontrib - subsec ] . ] .
turning now to the normal solution , @xmath105 , we find from ( [ qpseff0 ] ) that this is a minimum of the action unless @xmath132 this is just the condition for the extremal equation ( [ ftextr ] ) to have a condensed solution .
thus we have the usual scenario of a continuous phase transition : there is a phase boundary ( [ phaseboundinstab ] ) , at which the normal state becomes unstable and a stable , condensed solution appears . as well as the mean - field equation ( [ ftextr ] ) , we need the equation relating the density @xmath20 to the corresponding chemical potential @xmath39 .
this is obtained from the partition function in the standard way , @xmath133 the asymptotic form for the partition function is @xmath134 , where @xmath114 is the minimal action .
inserting this asymptotic form in ( [ gendens ] ) gives , for the solution @xmath135 , @xmath136 which is the generalization of ( [ ztexdens ] ) to finite temperatures .
the first term in ( [ ftexdens ] ) is the contribution to the excitation density from the macroscopic electromagnetic field , while the second term is the contribution from the thermal population of renormalized electronic excitations . in the absence of a macroscopic electromagnetic field ,
@xmath105 , both the photon contribution and the renormalization of the electronic excitations disappear . the expression ( [ ftexdens ] ) is then the familiar form for the excitation of a set of two - level oscillators .
from ( [ phaseboundinstab ] ) and ( [ ftexdens ] ) we have the critical temperature for condensation , as a function of the excitation density , in the homogeneous model : @xmath137 note that the transition temperature depends logarithmically on the density , and its scale is set by the interaction strength @xmath9 .
this is in contrast with a model of propagating , weakly - interacting bosons , where the transition temperature varies as a power law of the density and its scale is set by the mass of the bosons . at low densities , ( [ homphasebound ] )
is the phase boundary separating a population of electronic excitations with energy @xmath45 from a population of conventional polaritons with energy @xmath64 . to see this , note that such a transition would occur when the chemical potential for the electronic excitations reaches @xmath64 , corresponding to a density @xmath138 , which is the low - density limit of ( [ homphasebound ] ) . for the inhomogeneous model
, we calculate the phase boundary numerically , assuming the same gaussian distribution of energies as in section [ variationalresults ] .
we obtain the critical chemical potential for condensation , @xmath139 , by demanding that ( [ ftextr ] ) have a repeated root @xmath105 , and then use ( [ ftexdens ] ) to obtain the critical density @xmath140 . in fig .
[ phasediagfigure ] we plot the homogeneous phase boundaries ( [ homphasebound ] ) , along with numerical results for the inhomogeneous model with @xmath66 and 1 . on resonance , @xmath141 ,
the transition temperature increases monotonically with density .
the system is always condensed for @xmath142 , because to exceed this density would require a chemical potential above the center of the energy distribution of the electronic excitations , and hence above the bosonic cavity mode .
while for @xmath143(not illustrated ) the phase boundary is qualitatively unchanged from the resonant case , for @xmath144 we find reentrant behavior .
this behavior is the result of the saturable nature of the electronic states .
it can be understood by considering the limits @xmath145 when @xmath146 . near the @xmath65 limit , the normal state consists of a small number of electronic excitations , weakly interacting with each other through the cavity mode .
they condense when their density exceeds a critical value set by the strength of their interaction , which is determined by @xmath147 and @xmath9 . near the @xmath61 limit
, the electronic system is constrained to be fully occupied , and the normal state consists of a small number of holes in an otherwise completely excited electronic system .
these holes again interact through the cavity mode , and so the transition occurs when the density of holes , @xmath148 , exceeds a critical value . for @xmath149
the critical densities of holes and excitons are identical , so the phase diagram is symmetric about @xmath150 . for finite @xmath147 ,
the interaction is stronger for the holes than for excitons , since they are nearer in energy to the cavity mode , and so the phase boundary becomes skewed to the forms shown . at temperatures which are high compared with the inhomogeneous broadening @xmath46 , thermal fluctuations
dominate over the inhomogeneous broadening .
thus at these temperatures the inhomogeneous broadening has little effect , as can be seen in fig.[phasediagfigure ] .
however , at low temperatures the inhomogeneous broadening suppresses condensation by increasing the energy separation between the electronic excitations and the photons , collapsing the phase boundaries towards @xmath151 .
the effects of inhomogeneous broadening are further illustrated in fig.[phaseboundinhomfigure ] , which shows the dependence of the critical density on @xmath152 at various temperatures for detunings @xmath141 and @xmath153 .
in this section , we use ( [ qpseffinhom ] ) and ( [ qpseff0 ] ) to study the excitation spectrum of the quasi - equilibrium states of the model ( [ ham ] ) .
the excitation spectra we calculate explain the form of the phase diagrams in fig .
[ phasediagfigure ] .
the excitation spectra of the two quasi - equilibrium states are different from each other , and also from the excitation spectrum of a conventional laser .
since the excitation spectrum is directly related to the optical absorption spectrum of the cavity , which is an experimentally accessible quantity , these spectra offer a clear experimental signature of polariton condensation .
the matrix @xmath116 , given by ( [ qpseffinhom ] ) , is the inverse of the thermal green s function for the photons .
we use the standard relations @xcite between thermal green s functions , retarded green s functions , and excitation spectra , to extract the latter from ( [ qpseffinhom ] ) .
we begin with the normal state of the homogeneous model .
the inverse of the normal state green s function contained in ( [ qpseff0 ] ) can be written as a sum of simple poles @xmath154 the structure of this green s function is clear : we have two excitations , with quasiparticle energies @xmath155/2,\ ] ] and corresponding weights @xmath156 these normal - state excitations are polaritons in the general sense of hopfield @xcite : coupled modes involving the linear response of the electronic system around its equilibrium state .
the gap in the spectrum is increased over the bare detuning @xmath147 owing to the dipole coupling between the excitons and the cavity mode .
the presence of excitation in the ground state , either driven by finite temperatures or by finite @xmath157 , causes the two polariton branches to attract .
this attraction is due to the decrease in the polarizability of the electronic states as their population increases and saturation occurs .
it can also be understood in terms of an angular momentum representation @xcite for the collective states of the electronic system .
in such a representation , the excitation of the electronic states corresponds to the z component of an angular momentum , while their polarization corresponds to the raising operator @xmath13 .
thus the polarizability of the electronic states is maximized at @xmath158 .
since condensation is a phase transition , we expect a qualitatively different excitation spectrum in the condensed state . from ( [ qpseffinhom ] ) and ( [ ftextr ] )
, we find for the leading component of the matrix thermal green s function @xmath159 with @xmath160 .
the interpretation of this green s function is complicated because , as we have already mentioned , it describes both quasiparticle excitations and fluctuations of the order parameter . to rigorously obtain the quasiparticle spectrum
, we should extract the contribution to ( [ thermalgf ] ) from the order parameter fluctuations , and analytically continue the remainder to obtain the retarded green s function .
rather than follow such a procedure , we propose the following physically appealing if mathematically nave interpretation of ( [ thermalgf ] ) : the kronecker delta and the @xmath161 terms in the denominator describe the condensate response , leaving quasiparticle excitations at energies @xmath162 . the @xmath161 is clearly associated with the phase mode of the condensate , discussed in appendix [ goldstappendix ] , while the kronecker delta is related to number fluctuations of the condensate@xcite .
the excitations at energy @xmath162 are coupled exciton - photon modes in the presence of the macroscopic electromagnetic field of the condensate .
@xmath163 is analogous to the pair breaking energy in a superconductor : it is the energy required to extract an exciton - photon complex from the condensate .
note that if we remove the photon contribution to this energy , by setting @xmath164 , then @xmath163 becomes the familiar expression@xcite for the energy of an electron - hole pair in the presence of a classical electromagnetic field at frequency @xmath39 . in fig .
[ homoexcitationsfigure ] , we illustrate the evolution of the excitation energies of the microcavity with increasing density .
to explain the relationship between the excitation energies and the phase diagram , we also plot the chemical potentials for the normal and condensed states on this figure .
the left panel of this figure should be compared to the @xmath165 line of the corresponding phase diagram , which is the left panel of fig .
[ phasediagfigure ] . when @xmath141 and @xmath166 the system is in the normal state
. increasing @xmath167 populates the electronic excitations , increasing the chemical potential and decreasing the polariton splitting .
eventually the chemical potential crosses the lower polariton branch from below and the system condenses . at the critical density , the lower polariton branch joins to the phase mode at the chemical potential , the upper branch joins to the `` pair breaking '' excitation , and an excitation appears below the chemical potential .
this latter excitation has zero weight at the transition .
it corresponds to an excited state to ground state transition , where an exciton - photon complex is absorbed into the condensate .
there is no corresponding excitation in the normal state green s function , because the ground state of the @xmath168 particle system(@xmath168 excitons ) can not be reached from the excited states of the @xmath0 particle system(@xmath169 excitons and 1 polariton ) by adding a photon .
the relationship between the excitation spectrum and the phase diagram is slightly different when the transition occurs for @xmath142 .
for example , in the right panel of fig.[homoexcitationsfigure ] the chemical potential crosses the lower polariton branch at @xmath170 without the condensate appearing .
it is not until the chemical potential crosses the upper polariton branch that the transition occurs .
this can be understood by considering the signs of the quasiparticle weights @xmath171 .
a positive quasiparticle weight corresponds to absorption of an external field(a particle - like transition ) , whereas a negative quasiparticle weight corresponds to gain(a hole - like transition ) . for @xmath172 ,
the lower polariton branch has a negative weight : it has become hole - like , and must be below the chemical potential for stability . at the transition it is now this lower branch which joins to the `` pair forming '' excitation of the condensate , while the upper branch joins to the phase mode and the `` pair breaking '' excitation appears above the phase mode .
since the inhomogeneous model has a distribution of excitations , we must study the spectral function @xmath173 .
@xmath173 is proportional to the imaginary part of the retarded green s function , @xmath174 it is proportional to the optical absorption coefficient of the cavity at frequency @xmath123 , i.e. the imaginary part of the dielectric susceptibility . to obtain @xmath173
we require the retarded green s function @xmath175 . in the normal state
this is given by the straightforward analytical continuation @xmath176 however , in the condensed state we face the problem , already mentioned for the homogeneous case , of separating the order parameter response from the quasiparticle response .
we circumvent this problem by simply assuming that the continuation ( [ analyticcont ] ) of the normal state green s function also applies in the condensed state .
inverting the @xmath116 contained in ( [ qpseffinhom ] ) and using ( [ specdensity ] ) and ( [ analyticcont ] ) expresses @xmath173 in terms of integrals over the distribution of energies of the two - level oscillators .
we evaluate these integrals in the limit @xmath177 by setting @xmath178 in the integrands and deforming the contour of integration around the poles of the integrand on the real axis .
the contribution to the integrals from the detour around the poles can be performed analytically , leaving a principal value integral which we evaluate numerically .
figure [ inhomexcitationsfigure ] shows the evolution of our calculated absorption spectra , @xmath173 , as we increase the density through the transition , for @xmath165 , @xmath179 , and @xmath141 .
the corresponding chemical potential is marked as the dashed line .
for the empty cavity , @xmath65 , we recover the absorption spectrum calculated by houdr et al.@xcite .
comparison with fig .
[ homoexcitationsfigure ] shows that , for these parameters , the positions of the polariton peaks are largely unaffected by the inhomogeneous broadening .
however , since the polaritons are now resonant with a significant density of electronic states they become broadened .
increasing the chemical potential , but remaining in the normal state , we see the thermal occupation factors producing gain below the chemical potential and increased absorption just above . the collapse of the polariton splitting evident in fig .
[ homoexcitationsfigure ] is hardly noticeable at these low densities . as the density
is increased still further a pole appears in @xmath173 at the chemical potential ; this marks the onset of condensation . above the critical density the coherent cavity field , oscillating at frequency @xmath39 , produces a gap of magnitude @xmath180 in the spectrum .
the peak on the high energy side of the gap connects smoothly to the upper polariton peak of the normal state , just as in the homogeneous case . in the homogeneous case we noted the appearance of an excitation below the chemical potential as we crossed the transition
this is still present in the inhomogeneous case , but for the parameters used in fig .
[ inhomexcitationsfigure ] it is far too weak to be visible .
real microcavities are far more complex than the idealized model ( [ ham ] ) . however , like our model , they consist of photons coupled to electronic excitations which are bosons at low densities , but reveal their fermionic internal structure at high densities . we have shown how the polariton condensate may be generalized to allow for the saturation nonlinearity produced by such fermionic structure .
the saturation nonlinearity can produce ( 1)a collapse of the splitting between the peaks in the absorption spectrum of the normal state with increasing density , ( 2)a shift of the chemical potential of the condensate away from the conventional polariton energy , and ( 3)an unusual reentrant phase boundary for condensation .
experimental work on cavity polaritons has concentrated on microcavities containing high - quality gaas quantum wells . in these systems ,
the excitons are weakly - bound , and rather delocalized .
thus , while the saturation nonlinearity discussed here is present for these excitations , it will be accompanied by other nonlinearities produced by the overlap of the wavefunctions of different excitons and the ionization of excitons@xcite .
these effects may well prevent condensation , but are separated from the saturation nonlinearity considered here in systems with localized , tightly - bound excitons .
note also that tightly - bound excitons have a large dipole coupling @xmath9 , and hence the transition temperature will be larger .
for real examples of localized oscillators , there will be some energy @xmath181 above which delocalized states exist .
the picture of a condensate formed from localized oscillators then only holds when @xmath182 is large compared with @xmath183 and @xmath9 . by considering fig .
[ homoexcitationsfigure ] , we deduce that to completely realize a reentrant phase diagram like that shown in fig.[phasediagfigure ] requires an energy gap @xmath184 separating the localized and delocalized excitations ; this gap must be large compared with @xmath9 and @xmath183 .
such a gap could occur in organic semiconductors @xcite . in these systems ,
excitons are strongly bound and therefore small(frenkel ) .
they readily self - trap on local lattice distortions and on impurities in these , often highly disordered , materials .
an energy gap @xmath184 could exist in inorganic quantum wells if the excitons move in a potential containing deep , well - separated traps , perhaps associated with interface islands in narrow quantum wells@xcite .
the disordered quantum wells studied by hegarty et al.@xcite provide an example of a system without a gap separating the localized and delocalized excitations .
these systems show a single inhomogeneously broadened exciton line , unlike the quantum wells of refs . .
the `` mobility edge '' @xmath181 lies near to the center of the exciton line .
one may be able to form a condensate which does not involve delocalized excitations using this type of quantum well if the inhomogeneous broadening is large compared with @xmath9 and @xmath185 and the cavity mode is placed low down in the exciton line .
the transition would then occur when the chemical potential is well separated from @xmath181 .
the polariton condensate described here is formed from a quasi - thermal population of electronic excitations which are renormalized by the coherent photons in the cavity .
this renormalization , embodied in the bogolubov transformation ( [ bogt ] ) , produces a gap of magnitude @xmath180 in the absorption spectrum of the condensate .
such renormalizations , and hence the gap , are absent in conventional semiconductor lasers @xcite , for which a quasi - thermal population of the _ bare _ electronic excitations is assumed .
thus the presence or absence of a gap allows the polariton condensate to be distinguished from a conventional laser . in a conventional laser ,
the renormalization of the electronic excitations by the photons , and hence the gap , is absent because the electronic polarization is very heavily damped .
the destruction of a gap by damping is well - known in superconductors , where it is associated with magnetic impurities .
such impurities suppress the gap , eventually to zero .
the destruction of the gap does not coincide with the destruction of the order parameter however : near to @xmath186 there is a regime of gapless superconductivity , in which there is an order parameter but no gap in the single - particle spectrum .
this regime should correspond to the conventional semiconductor laser , although the actual damping mechanisms will differ .
the non - equilibrium analog @xcite of the crossover illustrated in fig .
[ homoexcitationsfigure ] and fig.[inhomexcitationsfigure ] , from a two - peaked polariton spectrum to a `` stark triplet '' , has been observed experimentally@xcite . in that experiment ,
the gapped absorption spectrum is observed simultaneously with the excitation pulse .
thus there is no thermalization involved in producing the gapped spectrum .
it is the result of coherence in the excitation pulse , rather than the spontaneous coherence of condensation .
nonetheless , these experiments demonstrate the renormalization of the electronic states that is essential in the polariton condensate . to reach the quasi - equilibrium regime we have described requires a system where the polariton lifetime is long compared with the time required to reach thermal equilibrium at a fixed number of polaritons .
current semiconductor microcavities have lifetimes for the photons , and hence the polaritons , of the order of picoseconds .
finding an exciton system which thermalizes on this timescale may be difficult .
however , there seems no reason to suppose it is impossible , particularly beyond the linear regime , since nonlinearities can enhance relaxation @xcite .
furthermore , microcavities are available with lifetimes far greater than picoseconds . for example
, silica microspheres have confined modes with lifetimes of microseconds @xcite .
this work was supported by the engineering and physical sciences research council , u.k .
in this appendix we investigate the zero eigenvalue of @xmath116 that appeared while studying the stability of the condensate in the homogeneous case .
we first prove that the zero is also present in the inhomogeneous model , and that it describes phase fluctuations of the condensate .
it is thus the goldstone mode reflecting the degeneracy of the ground state with respect to the phase of the order parameter .
we then argue that the zero eigenvalue does not contribute to the free energy density in the thermodynamic limit .
although the physics we discuss in this appendix is well understood in general , it is particularly transparent in our simple model .
we note that , at @xmath127 , @xmath187 is real and positive .
the eigenvalues of @xmath116 are then @xmath188 .
since from the explicit forms of @xmath187 , @xmath189 and the extremal equation ( [ ftextr ] ) we have @xmath190 , as required in general by the hugenholtz - pines relation @xcite , @xmath116 has a zero eigenvalue . to illustrate that the zero eigenvalue is the phase mode of the condensate , note that since @xmath191 we can write @xmath192 the eigenvector of this matrix with zero eigenvalue is perpendicular in the complex plane to the order parameter @xmath24 . since we are considering a broken symmetry system
, we should not include states with different phases of the order parameter when calculating the partition function . thus on physical grounds
, we should discard the zero mode when computing the partition function . a formal approach which allows calculations in the presence of the zero eigenvalue
is to introduce symmetry breaking terms which are taken to zero after the thermodynamic limit .
this is the standard method for applying statistical mechanics to broken symmetry systems@xcite . the appropriate symmetry breaking terms for a bose condensed system pin the phase of the order parameter .
they are sources and sinks for the photons , and appear in the effective action @xmath193 as @xmath194 these terms do not contribute directly to ( [ qpseffinhom ] ) , but appear as a source term in ( [ ftextr ] ) .
the original zero eigenvalue of @xmath116 is now @xmath195 .
since for the equilibrium solution we must have @xmath196 , the contribution of the original zero eigenvalue to the free energy density is proportional to @xmath197 | we consider a generalization of the dicke model .
it describes localized , physically separated , saturable excitations , such as excitons bound on impurities , coupled to a single long - lived mode of an optical cavity .
we consider the thermal equilibrium of the model at a fixed total number of excitons and photons .
we find a phase in which both the cavity field and the excitonic polarization are coherent .
this phase corresponds to a bose condensate of cavity polaritons , generalized to allow for the fermionic internal structure of the excitons .
it is separated from the normal state by an unusual reentrant phase boundary .
we calculate the excitation energies of the model , and hence the optical absorption spectra of the cavity . in the condensed phase
the absorption spectrum is gapped .
the presence of a gap distinguishes the polariton condensate from the normal state and from a conventional laser , even when the inhomogeneous linewidth of the excitons is so large that there is no observable polariton splitting in the normal state . |
for more than 2 decades , one of the most important goals in high - energy nuclear physics has been the study of the quark - gluon plasma ( qgp for short ) .
the existence of this deconfined phase of quarks and gluons has been unequivocally shown in many lattice qcd ( quantum chromodynamics ) studies ( for a recent study see ref .
@xcite ) . in those studies ,
the phase transition ( cross - over ) temperature between the hadronic phase and the qgp phase is shown to be roughly about 200 mev when 3 light species ( @xmath0 ) of dynamic quarks are taken into account .
this value of transition temperature is actually somewhat of a conundrum . on the one hand , this temperature , although exceeding two trillion kelvin , is nonetheless low enough to be accessible in accelerator experiments . on the other hand ,
the energy scale corresponding to this temperature is too low for qcd to be perturbative , making analytic calculations difficult . nevertheless ,
combining the results from lattice qcd calculations , insights from perturbative thermal qcd calculations and also thermodynamics , we do know much about qualitative features of this new phase of the matter .
accordingly , many researchers have proposed many different signals of the formation of the qgp in heavy ion collisions .
many of these proposed signals utilize the fact that the energy density of qgp is extremely high . at @xmath1 ,
the energy density easily exceeds @xmath2 .
systems created at the relativistic heavy ion collider ( rhic ) can reach maximum temperature of about @xmath3 . at this temperature , the energy density of qgp ( composed of gluons and @xmath0 quarks ) can be as high as @xmath4 . not surprisingly , the three most prominent qgp signals that emerged from rhic experiments the strong elliptic flow , the quenching of the high - energy ( jet- ) particles and the emergence of the medium - generated photons
are all measures of the high energy density and corresponding high pressure . among them
, the jet - quenching phenomenon is perhaps the most direct observation of the high energy density .
intuitively , it is easy to see that it will take an extremely dense matter to stop a particle with an extremely high energy . in this proceeding , a short summary of the various theoretical concepts that go into calculating the jet quenching ( equivalently , parton energy loss ) is presented .
it is , of course , impossible to do justice to the vast amount of work performed by many different researchers in this short proceeding .
what i will do mainly is to use the mcgill - amy approach @xcite that i am most familiar with as an illustrative example and highlight the differences between this and other approaches where possible .
interested readers are directed to a more comprehensive review already in print @xcite and let me apologize here to people whose interesting works i can not fully cover here for the lack of space .
\(a ) ( b ) the idea behind the jet - quenching phenomenon is rather intuitive . as depicted in figure [ fig : schematics ] , partons with high @xmath5 ( jets ) are produced when two hard partons from colliding hadrons undergo a hard scattering . in hadron - hadron collisions ( figure [ fig : schematics]-a ) , the jets then propagate and evolve in vacuum until they produce showers of particles in the detector
. on the other hand , in heavy ion collisions jets produced by relatively rare hard collisions must propagate and evolve within the hot and dense medium ( qgp ) that is created by the rest of the system ( figure [ fig : schematics]-b ) . by measuring the difference between the outcome of these two experiments
, one can then learn about the properties of the medium .
the simplest consequence of having a dense medium is the energy - loss of the propagating parton , that is , `` jet - quenching '' .
the usual way of showing the effect of energy - loss is to show the ratio of the momentum spectrum from @xmath6 scatterings and suitably normalized momentum spectrum from @xmath7 scatterings .
this ratio is usually referred to as the `` @xmath8 '' and defined as @xmath9 where @xmath10 denotes the observed hadron species and @xmath11 is the average number of hard binary collisions at the specified impact parameter .
( as far as the present author can find out , the first use of @xmath8 appeared in 1990 in publications by m. gyulassy and m. plmer @xcite and m. gyulassy and x .-
wang @xcite . )
experimental data from phenix and star collaborations spectacularly confirm that jet quenching is real .
there is a sharp reduction of high energy particles in the away - side of the trigger particle @xcite and @xmath8 has a surprisingly small value of about @xmath12 for all measured range of the high @xmath5 in central collisions @xcite .
theoretically , the calculations of the @xmath6 spectrum and the @xmath7 spectrum proceed as follows . for the @xmath6 case ,
the jet cross - section is given by the following schematic perturbative - qcd formula @xmath13 where @xmath14 is the parton distribution function , @xmath15 is the parton level cross - section and @xmath16 is the fragmentation function . for @xmath7 collisions , we have @xmath17 where @xmath18 is the parton distribution function of the nucleus @xmath19 and the conditional probability @xmath20 contains the effect of the medium that changes the parton from @xmath21 to @xmath22 . here
@xmath23 and @xmath24 denote the temperature and the flow velocity profiles throughout the evolution .
we also need to integrate over the nuclear and collision geometry .
since the geometry is more or less fixed once the impact parameter range is known , the main task of a theoretician is then to calculate @xmath20 given the profiles of the temperature and the flow velocity .
there are , however , some caveats that go with eq.([eq : sigmaaa ] ) . the @xmath6 formula ( [ eq : sigmapp ] ) is firmly based on the factorization theorem in qcd .
in contrast , the factorization theorem that would put eq.([eq : sigmaaa ] ) on a firm setting is not yet fully proven .
( gelis , lappi and venugopalan have taken promising initial steps in this direction @xcite . )
another caveat is that the medium is finite and it is evolving all the time .
the lifetime of the qgp created in a heavy ion collision is about 5fm / c .
the size of the system is about 10fm .
the hydrodynamic time scale is about @xmath25 .
the mean free path of a particle in a qgp is also of order 1fm . none of these are very large or very small compared to the others .
full accounting of these similar yet different length and time scales is therefore not an easy task .
inevitably , one needs to make some assumptions and approximations . with these caveats , nearly every theoretical approach to
the energy - loss assumes that the above formula ( [ eq : sigmaaa ] ) is at least a good approximation . and that s what will be assumed in this paper as well .
as mentioned above , the main task of a theoretician working on parton energy loss is to calculate the in - medium modification function @xmath20 .
collisions with the thermal particles cause the changes in the propagating parton .
hence , the fundamental quantity to calculate is the scattering cross - sections and the associated collision rates .
if perturbation theory is valid , this would be a relatively straightforward calculation at least at the leading order . in a hot and dense matter , this is no - longer true : the dispersion relationship of a particle is no longer that in the vacuum .
the scattering potentials are screened .
extra divergences appear due to the frequent soft exchanges .
all these conspire to make the loop expansion invalid . for elastic scatterings , one can still regard the tree diagrams as the leading order @xcite . for radiational processes ,
the above complications make non - trivial resummations necessary even for the leading order
. one must include _ both _ the elastic and the inelastic scattering processes @xcite for phenomenology . here ,
for brevity , we only discuss the inelastic radiational process .
the first study of radiational energy loss was conducted in refs @xcite .
this method is often referred to as the bdmps - z approach .
the main thesis of this approach is as follows .
consider a medium where the temperature is high enough that perturbation theory is valid . in this case , soft exchanges between the medium and the propagating parton result in the radiation of a hard collinear gluon . at the same time , the effect of multiple collisions is reduced because within the coherence length ( or formation time ) of the emitted gluon , all soft scatterings basically count as a single one .
this landau - pomeranchuck - migdal ( lpm ) effect then necessitates the resummation of all diagrams depicted in in figure [ fig : the_big_picture2 ] in calculating the leading order gluon radiation rate . in the original bdmps - z approach
some simplifications and approximations were made .
specifically , the medium was assumed to be composed of random static scatterers . also ,
the thermal screening effect was only approximately taken into account using the approach advocated in ref.@xcite .
the analysis was then carried out in the deep lpm regime and multiple emission is treated with the poisson ansatz @xcite .
subsequent approaches to the energy loss calculations use more or less the same starting point .
one still needs to calculate and sum the diagrams depicted in figure [ fig : the_big_picture2 ] .
the difference between various approaches may be classified by the way each approach ( i ) treats the scattering centers , ( ii ) resums the diagrams , and ( iii ) deals with the evolving medium .
these differences may be summarized as follows .
the acronyms in the following list are the initials of the main authors except ht ( higher - twist ) and mcgill - amy . also , the references given are not exhaustive but just a starting point for further reading .
* scattering centers * * heavy static scattering centers ( cold medium ) bdmps @xcite , zakharov @xcite , glv , @xcite asw @xcite , * * dynamic scatterers ( hot medium ) mcgill - amy @xcite , dglv @xcite , whdg @xcite * * general nuclear medium with a short correlation length
ht @xcite * resummation schemes * * sum over diagrams with all possible soft interactions bdmps ,
mcgill - amy * * path integral representation of hard parton propagation
asw , zakharov * * reaction operator method with opacity expansion
glv , dglv , whdg * evolution schemes * * poisson ansatz
bdmps , asw , glv , dglv , whdg * * fokker - planck equation mcgill - amy * * modified dglap equation ( momentum space evolution )
ht in this proceeding , i am going to use the mcgill - amy approach as the illustrative example to show what is involved in calculating each of the above items , simply because that is the one i am most familiar with .
when the temperature is high enough , the asymptotic freedom property of the qcd makes it possible to treat qgp within the perturbation theory . the usual loop expansion , however , is no longer valid . the coherence effect ( the lpm effect )
makes it necessary to resum an infinite number of generalized ladder diagrams shown in figure [ fig : rate_calc_amy]-a .
\(a ) ( b ) technically , this comes about because connecting any of the three hard lines with a soft space - like gluon line ( labeled with @xmath26 ) introduces a pair of pinching poles in the loop - frequency space as illustrated in figure [ fig : rate_calc_amy]-b .
the resulting pinching - pole singularity , when regulated by the hard - thermal - loop self - energies cancels the factors of the coupling constants introduced by the additional interaction vertices . within thermal field theory ,
arnold , moore and yaffe rigorously proved that resummation of these generalized ladder diagrams is ncessary to get the leading order result ( see @xcite and references therein ) .
luckily , the resummation of the leading order contribution organizes itself into the schwinger - dyson type equation for the radiation vertex .
figure [ fig : linear_integral_equation_gluon ] shows a diagrammatic representation of the sd equation ; the corresponding integral equation is @xmath27 \nonumber\\ & & \qquad + ( c_{\rm a}/2)[{\bf f}({\bf h})-{\bf f}({\bf h}{+}p\,{\bf q}_\perp ) ] + ( c_{\rm a}/2)[{\bf f}({\bf h})-{\bf f}({\bf h}{-}(p{-}k)\ , { \bf q}_\perp ) ] \big\ } \end{aligned}\ ] ] where @xmath28 here @xmath29 is the original hard parton momentum and @xmath30 is the momentum of the radiated hard gluon .
the masses appearing in the above equation are the medium induced thermal masses .
the 2-d vector @xmath31 is defined to be @xmath32 where @xmath33 is the chosen longitudinal direction and is a measure of the acollinearity .
the differential cross - section @xmath34 is the result of the hard - thermal - loop resummation .
the open end of the vertex is then closed off with the bare vertex and the appropriate statistical factors to get the gluon radiation rate @xmath35 . in this approach ,
the medium consists of fully dynamic thermal quarks and gluons .
furthermore , at least within the perturbation theory , the radiation rate so calculated is fully leading order in thermal qcd which sums an infinite number of the generalized ladder diagrams .
the original amy calculation of the radiation rates was carried out in order to obtain the leading order transport coefficients in hot qgp within the kinetic theory . in ref.@xcite
this is generalized to the case of propagating hard parton and its kinetic theory equation .
the initial momentum distributions of the hard partons now evolve according to the following set of fokker - planck equations @xmath36 which also includes the effect of absorbing thermal energy from the medium . for the full phenomenological study ,
this has to be supplemented by the local temperature and the flow information by independent hydrodynamics calculations . since there is nothing intrinsically boost - invariant about our formulation ,
there is no restriction on what the underlying soft matter evolution should be .
the rates appearing above then become time and space dependent through the local temperature and the flow velocity .
finally , the resulting parton distribution at the final time is convoluted with the geometry and the vacuum fragmentation function to yield the medium modified fragmentation function @xcite : @xmath37 the resulting @xmath8 for @xmath38 is shown in figure [ fig : raa]-a which includes both the effects of the radiational energy loss and the collisional energy loss @xcite . for @xmath38 calculated within the mcgill - amy approach . the panel ( a ) is from ref.@xcite and the panel ( b ) is a preliminary result from martini an event generator being developed based on the mcgill - amy energy loss mechanism . in panel ( a ) , the red solid line is the full calculation .
the green broken line includes the effect of radiative energy loss only while the blue dash - dot line includes the effect of the collisional energy loss only . in panel ( b )
, the green line is the main result with both the collisional and the radiational energy loss . here
we set @xmath39 and the underlying soft - evolution is taken from 3 + 1 d hydrodynamics calculation of nonaka and bass @xcite .
the data points are taken from @xcite .
, title="fig:",scaledwidth=47.0% ] for @xmath38 calculated within the mcgill - amy approach .
the panel ( a ) is from ref.@xcite and the panel ( b ) is a preliminary result from martini
an event generator being developed based on the mcgill - amy energy loss mechanism . in panel ( a ) , the red solid line is the full calculation .
the green broken line includes the effect of radiative energy loss only while the blue dash - dot line includes the effect of the collisional energy loss only . in panel ( b )
, the green line is the main result with both the collisional and the radiational energy loss . here
we set @xmath39 and the underlying soft - evolution is taken from 3 + 1 d hydrodynamics calculation of nonaka and bass @xcite .
the data points are taken from @xcite .
, title="fig:",scaledwidth=50.0% ] \(a ) ( b ) one thing to notice is that @xmath40 does not validate perturbative treatment since this implies @xmath41 . at this point ,
mcgill - amy approach becomes a phenomenological model .
however , i believe that this is the best one can do with current analytical tools since it at least includes all dynamic effects such as the energy - momentum conservation , broadening , thermal push , flavor change , elastic and inelastic energy losses except the interference between the vacuum and the in - medium processes @xcite .
the last one is currently being worked on by the members of the mcgill - amy team .
in the previous section , a particular approach mcgill - amy was described and shown to yield results that compare very favorably with the experimental data .
as far as the nuclear modification factor @xmath8 is concerned , other approaches can equally well describe the experimental @xmath8 with a reasonable set of assumptions @xcite . in this sense ,
the theoretical efforts to understand the jet quenching has been successful .
however , the question remains : what do we really learn about the medium when different approaches with different assumptions about the medium can describe the experimental data equally well ? for instance , consider the temperature .
since we are dealing with a thermalized system , finding the value of the temperature goes a long way to characterize that system .
it sounds like measuring temperature ought be a simple task .
even this , however , becomes a non - trivial task among different models .
temperature appears as an explicit parameter of the calculation only in some models . in others ,
the medium is treated as dense , but cold to simplify the calculation .
the connection to the qgp property is then made through the transport coefficient @xmath42 where @xmath43 is the soft momentum exchange scale and @xmath44 is the mean free path .
when soft exchange dominates the dynamics of the propagating hard parton , this should not be a bad approximation .
however , this is still an indirect connection since @xmath43 and @xmath44 are independent free parameters instead of functions of @xmath45 and @xmath23 .
combined with the differing treatments of the multiple emissions , it makes the determination of the temperature within such models rather fragile especially since the value of @xmath8 seems to be only weakly dependent on the value of @xmath46 @xcite .
recently , a collective effort to resolve this and other issues has been initiated by the tec - hqm collaboration @xcite .
right now the collaboration is performing standardized tests to see where exactly the differences between models lie .
since the shape of @xmath8 is so featureless , i believe that the true test of models will come later when other observables such as the hard photon spectrum are calculated coherently within each approaches ( for instance the @xmath47 spectrum @xcite ) .
it is therefore quite encouraging that monte - carlo event generators based on the current works on the energy - loss mechanisms start to appear on the scene . in these proceedings ,
some results from yajem @xcite , jewel @xcite , and q - pythia @xcite are reported . in the remaining space ,
let me introduce the event generator being worked on by mcgill - amy team martini ( modular algorithm for relativistic treatment of heavy ion interactions ) .
the first , very preliminary result from martini is shown in figure [ fig : raa]-b .
martini is a modular modification of pythia 8.1 to take into account the energy loss of hard partons before they can hadronize ( fragment ) into showers .
hopefully , many more tests of the mcgill - amy approach can be performed on many different hadronic and electro - magnetic probes with this new development .
first of all , i would like to express my gratitude to the members of mcgill - amy team c. gale , g. moore , s. turbide , g. qin , j. ruppert , b. schenke .
much help from u. heinz , e. frodermann , c. nonaka , s. bass , m. mustafa , and d. srivastava is also greatly appreciated .
s. turbide , c. gale , s. jeon and g. d. moore , phys . rev .
c * 72 * ( 2005 ) 014906 g. y. qin , j. ruppert , c. gale , s. jeon , g. d. moore and m. g. mustafa , phys .
* 100 * ( 2008 ) 072301 c. gale , s. turbide , e. frodermann and u. heinz , j. phys .
g * 35 * , 104119 ( 2008 ) s. turbide , c. gale , e. frodermann and u. heinz , phys .
c * 77 * , 024909 ( 2008 ) g. y. qin , j. ruppert , c. gale , s. jeon and g. d. moore , arxiv:0906.3280 [ hep - ph ] .
b. schenke , c. gale and g. y. qin , arxiv:0901.3498 [ hep - ph ] .
s. a. bass , c. gale , a. majumder , c. nonaka , g. y. qin , t. renk and j. ruppert , phys .
c * 79 * ( 2009 ) 024901 m. gyulassy and m. plumer , phys .
b * 243 * , 432 ( 1990 ) .
x. n. wang and m. gyulassy , in bnl rhic workshop 1990:0079 - 102 j. adams _ et al . _ [ star collaboration ] , phys .
* 91 * ( 2003 ) 072304 a. adare _ et al . _
[ phenix collaboration ] , phys .
* 101 * ( 2008 ) 232301 f. gelis , t. lappi and r. venugopalan , phys .
d * 78 * ( 2008 ) 054020 f. gelis , t. lappi and r. venugopalan , phys .
d * 78 * ( 2008 ) 054019 s. mrowczynski , phys .
b * 269 * ( 1991 ) 383 .
m. g. mustafa and m. h. thoma , acta phys . hung . a * 22 * ( 2005 ) 93 [ arxiv : hep - ph/0311168 ] .
r. baier , y. l. dokshitzer , a. h. mueller , s. peigne and d. schiff , nucl .
b * 483 * ( 1997 ) 291 b. g. zakharov , jetp lett .
* 65 * ( 1997 ) 615 r. baier , y. l. dokshitzer , a. h. mueller and d. schiff , nucl .
b * 531 * ( 1998 ) 403 m. gyulassy and x. n. wang , nucl . phys .
b * 420 * ( 1994 ) 583 r. baier , y. l. dokshitzer , a. h. mueller and d. schiff , jhep * 0109 * ( 2001 ) 033 [ arxiv : hep - ph/0106347 ] . m. gyulassy , i. vitev , x. n. wang and b. w. zhang , arxiv : nucl - th/0302077 .
m. gyulassy , p. levai and i. vitev , nucl .
b * 571 * ( 2000 ) 197 n. armesto , c. a. salgado and u. a. wiedemann , phys .
d * 69 * , 114003 ( 2004 ) m. djordjevic and m. gyulassy , nucl .
a * 733 * ( 2004 ) 265 m. djordjevic and u. heinz , phys .
c * 77 * ( 2008 ) 024905 s. wicks , w. horowitz , m. djordjevic and m. gyulassy , nucl .
a * 784 * ( 2007 ) 426 a. majumder and x. n. wang , arxiv:0806.2653 [ nucl - th ] .
a. majumder , phys .
c * 75 * ( 2007 ) 021901 c. nonaka and s. a. bass , phys .
c * 75 * , ( 2007 ) 014902 k. j. eskola , h. honkanen , c. a. salgado and u. a. wiedemann , nucl .
a * 747 * , ( 2005 ) 511 the tec - hqm collaboration wiki , https://wiki.bnl.gov / techqm/. i. vitev and b. w. zhang , phys .
b * 669 * ( 2008 ) 337 t. renk , phys . rev .
c * 78 * ( 2008 ) 034908 k. zapp , g. ingelman , j. rathsman , j. stachel and u. a. wiedemann , eur .
j. c * 60 * ( 2009 ) 617 n. armesto , l. cunqueiro and c. a. salgado , arxiv:0907.1014 [ hep - ph ] . | a short summary of different approaches to the parton energy loss problem is given .
a particular attention is paid to the differences between various models . a possible solution to the problem of distinguishing competing approaches
is discussed . |
opportunistic networks are very dynamical systems , characterized by intermittent contacts among mobile nodes and frequent partitions @xcite .
the model of interaction differs from classic communication paradigms , usually based on a prolonged end - to - end connectivity @xcite . at a given instant , a complete path from a source to a destination
might not exist @xcite .
rather , nodes communicate as soon as they have the opportunity to do it .
these occasional contacts are employed to share and disseminate information , route messages towards some destination , etc .
delay tolerance is thus a main characteristics of these networks , and in general the temporary contacts among nodes must be exploited to disseminate contents , due to the uncertainty of future communications .
an open research topic is concerned with the topology of an opportunistic network . due to its evolving nature
, several works simply define the topology of opportunistic networks as unpredictable , others approximate them as small worlds , others argue that temporal connection models are better suited than spatial mobility models @xcite . as a matter of fact , at a given instant an opportunistic network appears as a classic mobile ad - hoc network ( manet ) , composed of a set of links which are constrained by the geographical location of mobile users and the limitations imposed by the signal strength of the employed wireless technology @xcite .
however , while manets consider links as connections active at a given instant , opportunistic networks have a coarse grained time model .
hence , the concept of link in an opportunistic network should reflect the fact that temporal constraints are relaxed .
this suggests to model opportunistic networks as evolving graphs where links characterize the interactions of a node during a ( non - instantaneous ) time interval @xmath0 .
thus , the set of links of a node at time @xmath1 is not simply composed of the contacts active at the instant @xmath1 , but rather , it represents the aggregate of the contacts arose during @xmath2 $ ] .
then , different specific types of `` link '' are possible . for instance
, one might consider a link between two nodes as a single contact occurred during @xmath0 , or when a quasi - continuous communication is guaranteed between the two nodes ( during @xmath0 ) .
a clear consequence of such a scenario is that links among nodes would not depend only on spatial constraints , but also on the mobility of the node and on the length of @xmath0 . in fact , certain mobile users may be more active than others ; hence , while at a given instant the number of contacts of a highly- mobile node may be similar to those of other nodes , during the interval @xmath0 such node will collect more contacts , resulting in a higher number of links . in this scenario
, it might be interesting to understand whether the network can be shaped to achieve a desired topology , based on some link management scheme , locally executed by mobile nodes .
the goal is clear : knowing the network topology corresponds to identifying its statistical characteristics .
based on these properties , it is possible to define local strategies ( e.g. which and how many links to maintain ) to optimize the opportunistic dissemination @xcite . with the aim of identifying a strategy to shape opportunistic networks based on a given desired topology , we define a method to control the number of links at nodes . in our tests
we focus on a scale - free topology .
the reasons for this choice are the following ones .
first , it has been recognized that a scale - free network has several interesting characteristics for unstructured networks ( e.g. peer - to - peer ) , such as low diameter and high resilience to faults .
second , this topology is quite different to that typical of a network with spatial constraints , such as a manet ; the achievement of such a desired topology entails that several other opportunities are possible .
a simulation environment was built to test the model and the local algorithm to be executed at mobile nodes .
experiments are based on real traces taken from @xcite , which report the typical everyday behavior of more than 20000 students of the university of singapore , equipped with a mobile terminal .
these traces represent a perfect case study for an opportunistic network .
results on these traces demonstrate that , by tuning the length of the time interval @xmath0 , we can shape the opportunistic network based on some desired topology , e.g. scale - free .
the main contributions of this work are the following ones .
we characterize opportunistic networks as evolving graphs where links correspond to active contacts during a time interval .
we present a simple scheme to shape the network based on some desired topology .
we specifically cope with scale - free topologies , but the approach can be easily generalized to other network topologies . finally , we experiment with real data - traces , which record the typical behavior of a large number of users during their everyday life @xcite .
the remainder of the paper is organized as follows .
section [ sec : alg ] presents the general approach .
section [ sec : exp ] outlines the experimental scenario , while results are reported in section [ sec : res ] .
finally , section [ sec : conc ] provides some concluding remarks .
we consider an opportunistic network as an evolving graph where links among nodes correspond to the aggregate of contacts that nodes have during a given time interval @xmath0 .
thus , at a given instant a node is in direct contact with a subset of those nodes linked to it during the @xmath0 time window .
we then present a simple scheme that lets nodes to maintain a pre - defined number of neighbours .
this enables to set a `` desired topology '' .
we employ this mechanism in the attempt to give to the network a scale - free desired topology .
opportunistic networks are characterized by nodes that frequently move and with intermittent contacts @xcite .
an opportunistic network may assume different forms and it may be employed as the underlying structure for several mobile application scenarios .
applications built on top of opportunistic networks are different to those to be executed in classic ( wireless ) networks .
examples may be found in literature , ranging from communication approaches based on ad - hoc networks to be deployed on rural and developing regions , tracking systems for wild animals , to those ( delay tolerant ) applications built over highly dynamic nets in urban areas ( e.g. mobile and vehicular ad - hoc networks ) @xcite . in general , no simultaneous multi - hop path can be guaranteed @xcite .
communication protocols allow nodes to manage intermittent and unstable contacts , cache and relay messages as soon as there is the opportunity to do it . from an implementation point of view
, there is a `` bundle''-layer , in between the classic transport and the application layers , that implements such store , carry and forward strategy .
a critical point is the definition of `` link '' between nodes .
this must certainly be based on contacts between two nodes and might depend on the type of interactions required by the applications run on top of the network itself @xcite .
for instance , a single contact is sufficient to exchange some content of common interest between two users , e.g. variations on lecture schedules among students , news on train / airport tables , and in general , content distribution through some mobile publish - subscribe like service . in this case
, we may state that a link exists between two nodes during a time interval @xmath0 , if a contact arises between the two nodes within that time interval .
we hence consider the network as an evolving graph whose links of a node @xmath3 at time @xmath1 are the aggregate of active contacts of @xmath3 during the time interval @xmath2 $ ] @xcite .
we assume that only messages received in a @xmath0 time window can be relayed to a neighbour though an active link .
this constraint reflects the fact that there is a limit on the temporal validity of messages and also to consider the limited computational capabilities of mobile nodes which can not hold messages forever .
the implementation of a link , as thought in this context , may be realized at the bundle and application layers , rather than as an open communication flow between two nodes ( as usually thought in manets or p2p architectures ) . in practice ,
a link from a node @xmath3 to another node @xmath4 means that @xmath3 stores on its memory the information related to @xmath4 , such as @xmath4 s i d and its profile , together with the contents @xmath4 is trying to disseminate , or the type of contents @xmath4 is trying to retrieve ( depending on the service running on top of the opportunistic network ) .
this approach of modeling a link reflects the fact that some content disseminated from @xmath3 to @xmath4 may be delivered afterwards to another node @xmath5 , if @xmath4 has the opportunity to interact with it in a time interval @xmath0 . due to constraints of mobile devices ,
each node would maintain a limited number of nodes as neighbours , and a node might decide to replace a neighbour with another .
hence , it may happen that as the network evolves , a node @xmath3 has an entry for @xmath4 in its `` neighbour table '' , while @xmath4 does not have any entry for @xmath3 , i.e. links are directed .
we define the desired topology of an opportunistic network by specifying the probability distribution of the degree ( i.e. number of links ) that a node may have . through this choice , it is possible to give a statistical characterization of the network , which permits to automatically estimate important metrics , such as the net diameter ( i.e. the max of the shortest paths among nodes in a net ) , the average number of neighbours at a given distance from a specific node , and so on @xcite .
the algorithm for the link management is very simple and it is as follows .
at the beginning of its interactions , each node randomly selects a desired degree , locally computed through the degree probability distribution associated to the desired topology .
note that in order to randomly selecting a degree , the node might need an estimation of the network size .
schemes exist that do it @xcite .
moreover , when the network size is very large , just an approximate value is sufficient ; it is required that , in case of a significant variation , nodes can detect it , and in this case they might change their desired degree . during the evolution of the network , based on the active contacts
a node has , the node stores entries related to the nodes it encounters ( i.e. it creates links ) , till reaching its desired degree . in general , we assume that links can be established only when the number of active contacts ( within a time interval @xmath0 ) surpasses a predefined threshold @xmath6
. this parameter can be tuned depending on the type of service to be executed on top of the opportunistic network .
for instance , a single contact ( @xmath7 ) may be employed for simple dissemination services where contents must be broadcast through the net .
higher values might be set for more sophisticated services , e.g. queries to be distributed that need answers . in this case , the communication would require multiple contacts .
the link holds for a limited time , and it is removed after a time @xmath0 of no contacts . when a novel contact arises with a non - neighbour node ( say @xmath4 ) , and
if the node has already reached its desired degree , then the novel node may become a neighbour with a certain probability @xmath8 . in this case , a random entry ( e.g. one among those with the lowest number of active contacts during the @xmath0 time window ) is replaced with @xmath4 . in the next section
we evaluate whether an opportunistic network , modeled as described above , can assume a scale - free topology .
a scale - free network possesses the distinctive feature of having nodes with a degree distribution that can be well approximated by a power law function .
hence , the majority of nodes have a relatively low number of neighbors , while a non - negligible percentage of nodes ( `` hubs '' ) exists with higher degrees @xcite .
the peculiarity of these networks is that they possess a very small diameter , thus allowing to propagate information in a low number of hops .
they are quite robust to node faults ( departures ) , something usual in a wireless network . on the other hand
, the presence of hubs might represent a drawback in the context of opportunistic networking , since it corresponds to an unbalanced load distribution .
coupling scale - free and ( opportunistic ) mobile networks is unusual , due to the issue mentioned above and to the fact that in a manet nodes connect to those which are directly reachable through the networking technology in use .
therefore , the instantaneous topology strongly depends on the geographical distribution of the nodes .
however , in this work links are considered as the aggregate of active contacts arising during a time interval @xmath0 .
hence , the role of hubs might be played by those nodes that have in time a number of contacts higher than others .
an example of a possible hub , in a real opportunistic network , might be a ticket inspector in some public transportation system ( equipped with a mobile terminal ) , a postman , or even a dedicated totem ( i.e. `` information sprinkler '' ) placed to relay contents to mobile nodes in a square or in a mall @xcite .
the idea of having dedicated nodes in points of interest , which act as hubs able to relay contents , would solve also the issue of the unfair load at hubs . in any case
, while we evaluate whether a scale - free overlay can be built over an opportunistic network , the same machinery can be employed also for other topologies .
it suffices to change the distribution to compute the desired degree .
a discrete event simulator was built to mimic the algorithm executed at mobile nodes .
nodes movements and contacts were obtained from real data traces @xcite . in what follows
, we describe these traces and explain the metrics of interest , together with the methodology to collect the results , which are provided in the next section .
we employed real data traces available from @xcite .
these are the patterns among 22341 students , inferred from their behavior during class schedules for the spring semester of 2006 in national university of singapore .
the national university of singapore is composed of colleges and departments . as reported in the description of the traces , all lessons were conducted on the main campus , spanning an area of 146 hectares @xcite .
in essence , these traces represent a wide , real scenario of typical students everyday life .
it is hence an ideal use case to test algorithms on urban opportunistic networks .
user contacts were traced in time intervals .
for instance , students meeting during a lecture where considered to be in contact during a single time interval , regardless of its duration ( which is abundantly longer than time usually needed to exchange information between two mobile nodes ) .
hence , the time interval @xmath0 exploited to manage links in the network is expressed as a multiple of these intervals .
the desired topology was generated through a probability distribution for nodes desired degree @xmath9 , following a power law function @xmath10 , with @xmath11 $ ] .
we considered a minimum and a maximum value of the desired degrees that a node might want to have , @xmath12 , @xmath13 , respectively .
we varied these parameters .
figure [ fig : dd ] shows an example of the distribution of the desired degree of network nodes , generated during our tests .
the log - log chart shows a linear curve , hence confirming that such distribution follows a power law distribution . during our tests
we varied the time interval size @xmath0 .
we varied also the minimum number of active contacts @xmath6 required so that to two nodes may become neighbours , and the value of @xmath8 that controls the probability that a node replaces a link with another , once its degree is equal to its desired degree .
for the sake of conciseness , we present only results related to a setting with @xmath14 , @xmath15 , @xmath16 , while varying the values @xmath17 , @xmath18 , @xmath19 .
similar outcomes were obtained for other values of these parameters ( not shown in this paper ) .
it goes without saying that , depending on the service to be executed on top of the opportunistic network , and most of all , on the contents to be disseminated , a degree value such as that of @xmath20 might be too high for a mobile node .
other scenarios are probably more realistic .
it is however worth mentioning that : i ) the number of such hubs would be very low and in certain application scenarios such a role might be played by those `` information sprinklers '' ; ii ) this setting is a test to see whether such a threshold might be reached using a real trace such as that employed in these tests .
figure [ fig : res_min_5 ] reports the results when @xmath21 , @xmath22 , varying the duration of @xmath0 , ( i.e. @xmath23 trace intervals ) .
each chart refers to a particular value of @xmath0 , and reports the distribution of nodes degree obtained when the proposed algorithm is executed .
the probability that a node has a certain degree is reported in a log - log scale . in this case , a linear trend may be appreciated , meaning that the distribution follows a power law function .
this confirms that the approach is able to configure the topology as a scale - free .
moreover , an higher @xmath0 ( rightmost chart ) guarantees a higher probability that nodes reach their desired degree .
only some few non - zero probabilities are obtained for some degree values lower than @xmath12 .
this means that there are certain nodes that have a few number of contacts with other nodes , whatever the duration of @xmath0 .
a similar result is reported in figure [ fig : res_min_10 ] , where @xmath24 and @xmath25 trace intervals .
figure [ fig : res_d_max_variato ] reports the degree probability obtained when @xmath26 trace intervals , @xmath21 , while varying @xmath13 .
again , it is confirmed that a scale free topology inside an opportunistic network is obtained in all cases , since there is a linear distribution of degree probabilities ( in log - log scale ) .
when @xmath27 , the probability that a node has a high degree is very low .
this is however due to the fact that only few nodes might have the opportunity to meet a number of @xmath28 nodes during a @xmath0 time window . moreover ,
such values are inappropriate as number of links a mobile node might have in a real opportunistic network ; such simulation setting serves only to demonstrate that the framework scales up to those numbers .
the reader may also notice the presence of non - zero degree probabilities lower than @xmath12 , which remain , despite the settings of the algorithm executed ( i.e. value of @xmath0 ) .
the presence of these values comes from the fact that the data traces exploited to do the simulations were the same in all the considered scenarios .
in this work we have presented a discussion on how opportunistic networks might be modeled . due to the evolving nature of the network and its delay tolerance , links among nodes
should not be considered as contacts which are active simultaneously .
rather , the aggregate of contacts arising during a time interval should be preferred . to create a link
, the duration of the contact might be sufficient for a data exchange .
such amount of time might depend on the application to be run on top of the network .
in particular , a link in the evolving graph represents the fact that a node maintains on its `` neighbour table '' some application - related contents on the behalf of its neighbour , so that such contents can be relayed as soon as there is the opportunity to do it .
we have presented a simple algorithm that allows each node to set and manage its own degree , in order to shape the net based on a desired topology .
this influences the way contents can be disseminated through the network .
the scheme has been employed on real data traces , and outcomes confirm that a desired topology ( in this case a scale - free ) can be obtained . to the best of our knowledge ,
this is the first attempt to model opportunistic networks taking into consideration contacts active in a given time interval , instead of instantaneous contacts . in future works
, the scheme will be employed on other data traces and with different desired topologies . | the coupling of scale - free networks with mobile unstructured networks is certainly unusual . in mobile networks ,
connections active at a given instant are constrained by the geographical distribution of mobile nodes , and by the limited signal strength of the wireless technology employed to build the ad - hoc overlay .
this is in contrast with the presence of hubs , typical of scale - free nets .
however , opportunistic ( mobile ) networks possess the distinctive feature to be delay tolerant ; mobile nodes implement a store , carry and forward strategy that permits to disseminate data based on a multi - hop route , which is built in time , when nodes encounter other ones while moving . in this paper
, we consider opportunistic networks as evolving graphs where links represent contacts among nodes arising during a ( non - instantaneous ) time interval .
we discuss a strategy to control the way nodes manage contacts and build `` opportunistic overlays '' .
based on such an approach , interesting overlays can be obtained , shaped following given desired topologies , such as scale - free ones .
opportunistic networks , scale - free networks , self - organization |
we know that there is no hadamard state on the kerr black hole space - time with the defining features of the hartle - hawking vacuum of respecting the isometries of the space - time and being regular everywhere @xcite .
frolov and thorne @xcite have shown that with certain non - standard commutation relations a state can be found whose feynman propagator formally has symmetry properties necessary for regularity of the state on the outer event horizon @xcite , however this state fails to be regular almost everywhere @xcite .
the concensus in the literature is that the key property of the space - time which accounts for the non - existence of a true hartle - hawking vacuum is the presence of a region in which the killing vector which is parallel to the null generators of the horizon becomes spacelike . in the present article
we take up a suggestion made in ref .
@xcite of removing this region from the space - time by enclosing the black hole within an axially symmetric , stationary mirror .
we show that when this is done , there is a well behaved state with the defining features of the hartle - hawking vacuum .
in addition , when the field is in this state the expectation value of the energy - momentum stress tensor as measured by an observer rigidly rotating with the horizon corresponds on the horizon to that of a thermal distribution at the hawking temperature rigidly rotating with the horizon . the mirror serves to remove the superradiant normal modes , as can be seen heuristically by noting that amplified waves which would otherwise escape to future null infinity are reflected by it across the future horizon . in the specific case of a mirror of constant boyer - lindquist radius , we can explicitly construct a vacuum state whose feynman propagator has the symmetry properties identified in ref .
@xcite and whose anticommutator function does not suffer from the pathology noted in ref . @xcite .
although this conclusion appears to be valid irrespective of the radius of the mirror , we generalize a result due to friedman @xcite to show that the space - time is unstable to scalar perturbations if the mirror does not remove all of the region outside the speed - of - light surface .
this instability is characterized by the existence of mode solutions of the field equation with complex eigenfrequencies which we have neglected when quantizing the field . kang @xcite has shown that the contributions made by these modes to the response function of an unruh box are not stationary but increase exponentially with time .
we therefore believe that when part of the region lying outside the speed - of - light light surface is inside the mirror , it is not possible to construct the stationary states involved in our considerations .
the layout of the paper is as follows . in sec .
[ sec : fpm ] we consider the normal mode solutions of the klein - gordon equation inside a mirror of constant boyer - lindquist radius . in sec .
[ sec : nr ] we use these solutions to numerically calculate the energy - momentum stress tensor as measured by observers rigidly rotating with the event horizon when the radius of the mirror is sufficiently small to remove all of the region outside the speed - of - light surface . in sec .
[ sec : scsf ] we consider the stability of the classical scalar field in the presence of a mirror and generalize friedman s result . finally , in sec .
[ sec : nsum ] we calculate numerically the eigenfrequencies of the unstable mode solutions of the field equation present when the black hole is inside a mirror of constant boyer - lindquist radius larger than the minimum radius of the speed - of - light surface .
we follow the space - time conventions of misner , thorne and wheeler @xcite and the notation of ref .
@xcite throughout .
we consider the right hand region of the kerr black hole enclosed within a mirror @xmath0 so that the past event horizon is a cauchy surface for the space - time .
we require that @xmath0 respects the killing isometries of the space - time since we are interested in a state which is invariant under these isometries .
the simplest mirror is the hypersurface @xmath1 since the klein - gordon equation then still admits completely separable solutions .
we take the radius of the mirror to be smaller than the minimum radius of the speed - of - light surface .
we will see in sec .
[ sec : scsf ] that if the radius is larger , there are modes of complex frequency which need to be considered in addition to those discussed in this section . for brevity , we deal only with the case of a field satisfying dirichlet conditions on this hypersurface .
we can construct normal modes by @xmath2 by convention , the ranges of our mode labels here and in what follows are appropriate to the `` distant observer viewpoint '' for modes with an ` in ' or ` out ' superscript and to the `` near horizon viewpoint '' for modes with an ` up ' or ` down ' superscript .
the terminology here is that of ref . @xcite . in the `` distant observer viewpoint '' @xmath3 while in the `` near horizon viewpoint '' @xmath4 where @xmath5 and @xmath6 is the angular velocity of the horizon with respect to static observers at infinity .
it is straightforward to check that the modes @xmath7 form an orthonormal set .
if we let @xmath8 denote the solution of the radial equation which behaves like @xmath9 close to the horizon then the asymptotic form of @xmath7 on the two horizons is @xmath10 we see that , since @xmath11 is a complex constant with unit modulus , these modes are not superradiant .
we could equally well have constructed normal modes defined by @xmath12 which also satisfy the boundary conditions on the mirror and form an orthonormal set .
it can be shown , however , that @xmath13 and hence that these upgoing and downgoing modes differ from each other only by a phase .
it follows that the vacuum state obtain by expanding the field in terms of either of these sets is the same and that this state is invariant under simultaneous @xmath14-@xmath15 reversal .
we denote it by @xmath16 . dropping the now superfluous superscript from the modes ,
the anticommutator function of @xmath16 is @xmath17 .
\label{eq : bmodhadamard}\ ] ] the modes with which the field has been expanded all have positive frequency with respect to the killing vector @xmath18 given by @xmath19 here , @xmath20 and @xmath21 are the two commuting killing vectors of the kerr space - time given in boyer - lindquist co - ordinates by @xmath22 it follows that an observer moving along an integral curve of @xmath18 makes measurements relative to the state @xmath16 .
such an observer rotates rigidly with the same velocity as the horizon with respect to static observers at infinity .
we call this observer a rigidly rotating observer ( rro )
. we can extend @xmath23 to the left hand region of the space - time by requiring it to be zero there . if we place a similar mirror in this region then we can introduce a function @xmath24 defined by @xmath25 which has unit norm , satisfies the field equation , is zero in the right region of the space - time and satisfies the boundary condition on the mirror in the left hand region .
both of the linear combinations @xmath26,\\ \tau_{\omega lm}&= \frac{1}{\sqrt{1-e^{-2\pi\tilde\omega/\kappa_{+ } } } } \left[t_{\omega lm}+ e^{-\pi\tilde\omega/\kappa_{+}}s_{\omega lm}^{*}\right],\end{aligned}\ ] ] are regular functions of @xmath27 on the past event horizon which are analytic in the lower half of the complex @xmath27-plane and are regular functions of @xmath28 on the future event horizon which are analytic in the lower half of the complex @xmath28-plane .
the vacuum state defined by expanding the field in terms of these is invariant under the simultaneous @xmath14-@xmath15 reversal and has a feynman propagator with the properties required of the hartle - hawking vacuum in ref .
we denote this state by @xmath29 .
its anticommutator function is @xmath30 .
\label{eq : hmodhadamard}\ ] ]
the measurements of an rro when the field is in the state @xmath29 can be calculated from @xmath31.\ ] ] the mode by mode cancellation of the high frequency divergences which afflict both anticommutator functions in the conincident limit makes this expression amenable to straightforward numerical analysis .
we calculated the spheroidal functions and separation constants in essentially the same way as that outlined in ref .
we calculated the radial functions by integrating eq .
( 2.6 ) of ref .
the accuracy of the integration has to be carefully considered in the case that @xmath32 is small and @xmath33 is large .
the details can be found in ref .
@xcite and will be outlined in a later article . , the expectation value of @xmath34 measured by an rro when the field is in the state @xmath29 .
the dark line gives the value on the horizon for a thermal distribution at the hawking temperature rigidly rotating with the horizon .
the units are such that @xmath35 . ] , the expectation value of the energy - momentum stress tensor measured by an rro when the field is in the state @xmath29 .
the dark line gives the value on the horizon for a thermal distribution at the hawking temperature rigidly rotating with the horizon .
the units are such that @xmath35 . ] the graphs in fig .
[ fig : phi2 ] and fig .
[ fig : t ] show the numerically calculated values of @xmath36 and @xmath37 , the expectation values measured by an rro when the field is in the state @xmath29 .
they are given for a black hole for which @xmath38 which means that @xmath39 .
in all of the graphs , the units are such that @xmath35 .
the divergent behaviour as the horizon is approached has been factored out and each graph terminates on the mirror which has been given a radius @xmath40 .
this is just smaller than the minimum radius of the speed - of - light surface , @xmath41 .
see the appendix for the details of how this is calculated . on the horizon ,
the components are compared and show close agreement with those corresponding to a thermal distribution at the hawking temperature rigidly rotating with the horizon .
these are given by @xmath42 , \label{eq : thermalt } \end{aligned}\ ] ] where @xmath43 is the local temperature and is related to the hawking temperature , @xmath44 , by @xmath45 there has been some discussion in the literature about the rate of rotation of a hartle - hawking vacuum on the kerr background , should it be possible to define such a state .
it has been suggested , for example , that the state might appear isotropic to a zamo or an observer rotating at the angular rate of the carter tetrad .
the numerical calculations for the state @xmath29 provide strong evidence that it rotates rigidly up to not just leading order but also next to leading order as the horizon is approached .
this is important because it shows that the rotation is not at the angular rate of either a zamo or the carter tetrad .
one way to see this is to consider an observer at fixed @xmath46 and @xmath47 rotating at an angular velocity of @xmath48 relative to static observers at infinity . as an orthonormal tetrad carried by the observer we can take @xmath49 , \label{eq : ortho_vecs_t}\\ { \boldsymbol{e}}_{(\varphi)}&=\frac{1}{\gamma\tilde\omega}\left[\partial_{\varphi}+ \frac{\gamma^2\tilde\omega^2}{\alpha^2}(\bar{\omega}-\omega ) \left(\partial_t+\bar{\omega}\partial_{\varphi}\right)\right ] , \label{eq : ortho_vecs_phi}\\ { \boldsymbol{e}}_{(r)}&=\sqrt{\frac{\delta}{\sigma}}\partial_r,\\ { \boldsymbol{e}}_{(\theta)}&=\sqrt{\frac{1}{\sigma}}\partial_{\theta}.\end{aligned}\ ] ] we have introduced here standard functions of the metric components @xcite @xmath50 and the function @xmath51 the function @xmath52 is the lorentz factor associated with the observer s angular velocity relative to a zamo at the same point and @xmath53 is the angular velocity of this zamo relative to static observers at infinity . given an energy - momentum stress tensor @xmath54 , the flux of energy in the direction of @xmath55 measured by the observer is @xmath56\\ & = \frac{\gamma^2\tilde{\omega}}{\alpha^3 } \left[\left(\frac{\alpha^2}{\tilde{\omega}^2}+\omega^2\right)t_{t\varphi}- \omega t_{tt}+ \bar{\omega}\left(t_{tt}+ \left(\frac{\alpha^2}{\tilde{\omega}^2}+\omega^2\right ) t_{\varphi\varphi}\right)+ \bar{\omega}^2(t_{t\varphi}+\omega t_{\varphi\varphi } ) \right]\end{aligned}\ ] ] the angular velocity at which the observer must rotate in order to measure no flux of energy is given by a root of the quadratic in @xmath57 appearing in the square brackets above . from the numerical data for @xmath37 , we can solve to find @xmath57 .
[ fig : omegalocal ] gives a graph of @xmath58 .
this is compared on the horizon with the value for an rro , a zamo and an observer rotating with the same angular velocity as the carter tetrad .
these are given by @xmath59,&\bar{\omega}=\omega .
\end{cases}\ ] ] the graph shows close agreement with the value corresponding to an rro and not with either of the other two . , where @xmath57 is the rate of rotation of the frame in which there is no flux of energy .
the black line gives the value on the horizon for @xmath60 , the blue line gives the value for @xmath61 and the green line gives the value for @xmath62 . ] it might be supposed that the behaviour of @xmath37 close to the horizon is derivable by an asymptotic analysis in the manner of candelas , chrzanowski and howard @xcite .
indeed , close to the horizon the contributions to the mode sums from the @xmath63 dominate and we find that @xmath64 the right hand side of which has been calculated in ref .
@xcite for the electromagnetic field .
we wish to point out , however , that this analysis is unfortunately in error except at the poles of the horizon .
the result given in eq .
( 3.7 ) of ref .
@xcite is obtained by noting that the dominant contributions come from the high @xmath33 modes and performing an asymptotic analysis of the @xmath65 as the horizon is approached for large @xmath33 .
the spheroidal functions are also tacitly replaced with spherical functions . in general , however , contributions come from every value of @xmath66 and since the dominant contributions to the mode sum are for @xmath67 , neither @xmath68 nor @xmath66 is necessarily small compared to @xmath33
. the spheroidal functions therefore can not always be approximated by spherical functions and the separation constant used in the analysis of the @xmath65 can not always be approximated by @xmath69 .
the poles are exceptional in that we know that only @xmath70 modes contribute there and that the analysis is therefore valid .
this can be clearly seen from the result which can be written @xmath71 where the components are on the carter tetrad .
we have corrected here for a typographical error in eq .
( 3.7 ) of ref .
@xcite in which a factor of @xmath72 has been omitted . in @xcite it was shown numerically that in fact @xmath73 which is precisely that of a thermal distribution at the hawking temperature rigidly rotating with the horizon .
these comments go equally well for the scalar field ; following the method of ref .
@xcite we would obtain values for @xmath74 and @xmath37 incorrect by the same multiplicative factor of @xmath75 .
the measurements of an rro do not correspond to those of a thermal distribution everywhere .
it is an open question whether or not their deviations from ( [ eq : thermalt ] ) are regular on the horizon .
it can be checked by transforming to the kruskal co - ordinate system that the conditions that @xmath76 , given by @xmath77 is regular on both horizons are that at worst @xmath78 these components are in the rigidly rotating co - ordinate system @xmath79 and we have placed a subscript on @xmath14 to avoid confusion with the boyer - lindquist co - ordinates .
we have not been able to verify these numerically .
it was conjectured by christensen and fulling @xcite that for the schwarzschild black hole , the measurements of a static observer when the field is in the hartle - hawking state would be exactly thermal everywhere .
this was later shown not to be the case by jensen , mclaughlin and ottewill @xcite .
although the leading order of the deviation from isotropy is zero on the horizon in agreement with the asymptotic analyses of @xcite and @xcite , even in this case it remains undetermined whether or not the total deviation is regular there .
the numerical calculations of the previous section fail if the radius of the mirror is increased to include any of the region in which @xmath18 becomes spacelike .
if any of this region lies inside the mirror , new modes of the field equation come into existence which are also regular on the horizon and satisfy the boundary conditions on the mirror .
these new modes have complex eigenfrequencies and are therefore characterized by the existence of unstable solutions of the field equation .
we will show in this section that such solutions exist for all sufficiently high @xmath66 .
kang @xcite has demonstrated how to quantize the scalar field on the kerr background in the presence of complex frequency modes .
he has shown that the contributions made by them to the response function of an unruh box are not stationary but increase exponentially with time and we expect that the anticommutator function will show similar behaviour .
it follows that neither of the stationary states @xmath16 and @xmath29 exists unless the region outside the speed - of - light surface is removed by the mirror .
the analysis presented in this section is based heavily on that of friedman @xcite .
the lagrangian for the complex scalar field is @xmath80 where @xmath81 is a hypersurface of constant @xmath14 , @xmath82 is the volume @xmath83-form induced on @xmath81 by the metric and @xmath84 is the lapse function given in eq .
( [ eq : metricfunctions ] ) associated with the past pointing unit normal to @xmath81 which we denote by @xmath85 .
the field equation is @xmath86 the total charge of the field at time @xmath14 is given by @xmath87 .
\label{eq : charge}\ ] ] the total energy of the field at time @xmath14 depends on the killing vector with which we define derivatives with respect to time .
an appropriate vector is one of the form @xmath88 of which there is no preferred choice .
the energy of the field on @xmath81 is @xmath89 where @xmath90 consider the closed surface formed by the @xmath81 , @xmath91 and the mirror as in fig .
[ fig : friedman1 ] . , @xmath91 and the mirror . ] for a solution of the field equation , @xmath92 is divergence free and so its flux over this surface is zero .
its flux over the mirror is zero by virtue of the boundary conditions and so its flux over @xmath81 is minus that over @xmath91 .
that is , @xmath93 is independent of @xmath14 .
these comments go equally well for @xmath94 and so @xmath95 is also independent of @xmath14 . consider a solution of the field equation of the form @xmath96 suppose that @xmath68 is complex so that the solution is unstable , growing exponentially either forwards or backwards in time .
we find that @xmath97 |\phi_0|^2 \right\ } , \label{eq : modeenergy}\end{aligned}\ ] ] where @xmath98,\qquad \omega_{i}=\im[\omega].\ ] ] we know that @xmath93 and @xmath95 do not depend on @xmath14 and so it must be that the integrals in eq .
( [ eq : modecharge ] ) and eq .
( [ eq : modeenergy ] ) vanish so that @xmath99 and @xmath100 .
thus , for an unstable mode @xmath101 |\phi_0|^2 \right\}&=0.\label{eq : noenergy}\end{aligned}\ ] ] now if @xmath102 can be chosen so that it remains timelike everywhere then it is clear that the integral on the left hand side of eq .
( [ eq : noenergy ] ) must be positive for a non - trivial mode and so this condition can not be met .
there are therefore no unstable modes in this case .
in particular , putting @xmath103 , we see that there are no unstable modes if the radius of the mirror is smaller than the minimum radius of the speed - of - light surface .
we could equally well have restricted the field to the region outside the mirror .
this time putting @xmath104 , we see that there are no unstable modes if the radius of the mirror is larger than the maximum radius of the stationary limit surface . on the other hand
, friedman @xcite has shown that for a star with an ergoregion , unstable solutions to the klein - gordon equation exist .
it can easily be checked that his analysis is valid for the kerr space - time with the event horizon removed by surrounding the black hole with a mirror .
we now show that there are likewise unstable solutions of the klein - gordon equation if we consider the space - time inside a mirror surrounding the black hole but not removing all of the region outside the speed - of - light surface .
let @xmath105 be a family of hypersurfaces related to the killing vector @xmath106 where the parameter @xmath107 satisfies @xmath108 .
the unit normal to each of these hypersurfaces is then given by @xmath109 the vector @xmath106 can be decomposed into a part which is parallel and a part which is orthogonal to @xmath85
. that is , @xmath110,\ ] ] where @xmath111 the region inside the speed - of - light surface is clearly the region in which @xmath112 .
we can write the metric as @xmath113 where @xmath114}n_{[\nu}m_{\sigma]}.\ ] ] the energy in the field on @xmath105 is @xmath115 where @xmath94 is defined in eq .
( [ eq : ecurrent ] ) with @xmath103 .
suppose that close to the future horizon the surfaces @xmath105 become null .
consider the region bounded by the surfaces @xmath116 , @xmath105 , a hypersurface of constant boyer - lindquist radius @xmath46 and the mirror as in fig .
[ fig : friedman2 ] . , @xmath105 , a hypersurface of constant boyer - lindquist radius and the mirror .
] since the total flux of @xmath94 over this surface is zero and its flux over the mirror is zero by virtue of the boundary conditions , we find that @xmath117 in order to obtain the above expression we have taken the limit as the radius of the hypersurface of constant @xmath46 tends to the radius of the event horizon and used the condition that the field is regular on the horizon .
this means that in the advanced co - ordinate system @xmath118 @xcite , for example , @xmath119 can be expanded in @xmath46 about @xmath120 in terms of regular functions of @xmath121 , @xmath47 and @xmath122 .
( [ eq : eevolution ] ) shows that a solution which is radiating energy across the future horizon looses energy between @xmath116 and @xmath105 . to complete the proof we need to show that if @xmath123 becomes spacelike somewhere within the mirror then there is an initial configuration of the field for which @xmath124 .
then eq .
( [ eq : eevolution ] ) implies that the field is at best marginally unstable and will be strictly unstable unless it can settle down at late @xmath107 to a non - radiative state .
furthermore , this non - radiative state state must be time - dependent . to see this note that from the field equation , @xmath125 can be rewritten @xmath126\ ] ] and hence is zero if @xmath127 is zero .
we now show that there is an initial configuration of the field for which @xmath128 if there is a region in which @xmath106 is spacelike .
it is straightforward to show that the energy can be rewritten @xmath129 + \left|m^{\mu}\nabla_{\mu}\phi\right|^2 + j^{\mu\nu}\nabla_{\mu}\phi\nabla_{\nu}\phi^ { * } \right].\ ] ] suppose that in the region outside the speed - of - light surface , @xmath116 is a hypersurface of constant @xmath14 .
it then follows that @xmath130 let @xmath131 be an open region of @xmath116 lying outside the speed - of - light surface and let @xmath132 be an open ball of radius @xmath133 contained within @xmath131 .
let @xmath134 be a function which vanishes outside a compact subset of @xmath131 , whose value and derivatives are bounded in such a way that there is a positive constant @xmath135 for which @xmath136 and which on @xmath132 is given by @xmath137 consider the field which satisfies the initial conditions @xmath138 this means that @xmath139\ ] ] there is an @xmath140 for which @xmath141 everywhere within @xmath131
. suppose that within @xmath131 , @xmath84 is bounded between @xmath142 and @xmath143 and @xmath144 is bounded above by @xmath145 .
it follows that @xmath146 where @xmath147 indicates the @xmath83-volume .
we see that @xmath124 for all sufficiently high @xmath66 .
although we can say very little analytically about the occurence of complex frequency modes , we can obtain a bounded region of the complex plane within which the frequency must lie and thereby search for them numerically . from eq .
( [ eq : nocharge ] ) we find that @xmath148 using eq .
( [ eq : nocharge ] ) , we can write eq .
( [ eq : noenergy ] ) as @xmath149|\phi_0|^2 \right\}=0.\ ] ] from the bound obtained for @xmath150 , we find that @xmath151 note from this that if a mode with complex frequency exists then the real part of its frequency is in the range which we associate with superradiance in the absence of a mirror .
it is clear that there are no axially symmetric modes with complex frequency .
it is also clear that when @xmath152 is set to zero and the space - time reduces to the schwarzschild space - time , no such modes can occur as is well known .
curiously , the bound on the complex part of the frequency obtained in a similar analysis given by detweiler and ipser @xcite does not imply this .
specializing to the case of a field satisfying dirichlet conditions on the hypersurface @xmath1 , we can now numerically search for complex frequency modes within this region .
we look for a solution of the form @xmath153 where @xmath154 this can be cast as a root - finding problem for a complex function of a complex variable .
the variable is @xmath68 and the function is @xmath155 , obtained by integrating the radial differential equation from the horizon subject to the condition ( [ eq : hbc ] ) .
close to the horizon , @xmath156 increases exponentially with increasing @xmath157 and so there are no numerical problems in maintaining the accuracy of the integration .
it is necessary to know the separation constant @xmath158 and we have calculated this by using a power series expansion in @xmath159 @xcite .
the root - finding algorithm we have used is an iterative scheme known as muller s method @xcite .
[ fig : nullmodes ] shows some of the results .
we fix the value of @xmath152 and track the movement of the frequencies found as the value of @xmath160 changes . as the radius of the mirror decreases , the frequencies approach the real axis and pass into the lower half of the complex plane before the radius of the mirror reaches the minimum radius of the speed - of - light surface . in each case
, the point at which they cross the real axis is the critical frequency of superradiant scattering in the absence of a mirror , @xmath161 .
this is expected in the case that the speed - of - light surface and the static limit surface do not cross .
the reason for this is that if @xmath68 is real then @xmath119 satisfies the conditions to be an ingoing function .
that is , @xmath162 when suitably normalized .
since @xmath163 also vanishes on the mirror , it is clear that the corresponding upgoing function , @xmath63 does also .
it is straightforward to check that @xmath63 is an unstable mode in the field theory outside the mirror .
however , we know from @xcite that there are no non - trivial modes of this type if the mirror lies entirely outside the stationary limit surface . the graphs in fig .
[ fig : nullmodes ] show that this behaviour of the complex frequencies also occurs when the black hole is rotating sufficiently quickly that the speed - of - light surface lies partly within the static limit surface and the argument given above fails .
the graphs are for the case @xmath164 , for which the relevant surfaces are shown in fig .
[ fig : speedoflightsurface ] . for a black hole with @xmath164 .
as @xmath160 decreases , the frequencies pass into the lower complex plane through the point @xmath161 before the minimum radius of the speed - of - light surface , @xmath165 , is reached .
the eigenfrequency is given in units in which @xmath35 . ]
in this paper , we investigated the pathology which afflicts the hartle - hawking vacuum on the kerr black hole space - time .
we have done this by taking up a suggestion made in ref .
@xcite of enclosing the black hole within a mirror .
the mirror serves to modify the global properties of the space - time without altering its differential geometry .
we have found that the pathology can be regarded as due to rigid rotation of the state with the horizon in the sense that ( 1 ) when the mirror removes the region outside the speed - of - light surface , there is a state with the defining features of the hartle - hawking vacuum ; ( 2 ) when the field is in this state , the expectation value of the energy - momentum stress tensor measured by an observer close to the horizon and rigidly rotating with it corresponds to that of a thermal distribution at the hawking temperature rigidly rotating with the horizon ; ( 3 ) when the mirror encloses any part of the region outside the speed - of - light surface , the field equation admits unstable solutions and it is not possible to construct any stationary state such as the hartle - hawking vacuum .
we performed the calculation of the energy - momentum stress tensor by numerically solving for the mode solutions of the field equation and performing mode sums . in a future article
, we will present a more detailed description of some of the numerical techniques used .
we will also give there results for the unruh vacuum .
the speed - of - light surface is the hypersurface other than the horizon on which @xmath106 becomes null . in boyer - lindquist co - ordinates this is given by the condition @xmath166 in terms of @xmath46 and @xmath47 this can be rewritten as @xmath167 where @xmath168 the cubic polynomial in @xmath46 which appears as a factor has only one real root @xmath169 which we can solve for in terms of @xmath47 @xcite , @xmath170 where @xmath171 it can be checked that @xmath169 has a minimum value in the equatorial plane and that this is given by @xmath172 for a black hole with @xmath38 , which is considered above , the minimum radius of the speed - of - light surface is @xmath41 . as can bee seen , it is possible for the speed - of - light surface to lie partly within the static limit surface .
when @xmath173 the two surfaces just touch in the equatorial plane and in the case of an extremal black hole , @xmath174 , the speed - of - light surface just touches the outer horizon in the equatorial plane .
we would like to thank valery frolov for a helpful discussion of unstable solutions of the klein - gordon equation .
we would also like to thank the centre for high - performance computing applications in university college dublin for extensive computing time . | we show that the pathology which afflicts the hartle - hawking vacuum on the kerr black hole space - time can be regarded as due to rigid rotation of the state with the horizon in the sense that when the region outside the speed - of - light surface is removed by introducing a mirror , there is a state with the defining features of the hartle - hawking vacuum .
in addition , we show that when the field is in this state , the expectation value of the energy - momentum stress tensor measured by an observer close to the horizon and rigidly rotating with it corresponds to that of a thermal distribution at the hawking temperature rigidly rotating with the horizon . |
high - mass stars are critical elements of galactic structure , due to their evolution in molecular clouds , the turbulent energy they inject into the interstellar medium and the metals dispersed in their death .
these stars are categorised by having mass @xmath9 and eventually explode as type ii supernovae , transforming the local interstellar medium .
the mechanism of high - mass star formation ( hmsf ) remains an important unsolved problem in astrophysics .
understanding hmsf is difficult for a number of reasons : the rarity of hmsf regions , high dust extinction within molecular clouds , rapid evolution , and large distances from our solar system @xcite all combine to hinder our understanding .
hmsf is known to begin within giant molecular clouds ( gmcs ) and eventually end with star clusters or associations @xcite . to understand the various stages of hmsf , an inventory of regions must be compiled in a range of known tracers . to date
, there have been many surveys conducted to help understand hmsf .
these have been useful in identifying hmsf regions and their characteristic spectral lines , but tend to emphasise subsections of the hmsf timeline .
examples of these surveys include methanol masers ( e.g. @xcite ) , infrared dark clouds ( irdcs ; e.g. @xcite ) and radio continuum sources ( e.g. @xcite ) .
methanol masers act as reliable signposts of an early evolutionary stage of hmsf @xcite , but do not occur at every evolutionary stage ; irdcs are only seen when they are close to us , projected against a bright infrared background and are not always associated with hmsf @xcite ; radio continuum sources are powered by main sequence high - mass stars , and are seen only when their free - free emission has penetrated through the dense surroundings .
large area surveys of molecular gas are ideal for identifying hmsf regions across a broad range of evolutionary phases . in particular , the critical density of a gas tracer can act as a probe for galactic structure and star formation .
critical density ( @xmath10 ) is a measure of the density required to produce detectable emission @xcite .
previous surveys for star formation in external galaxies , typically for low - density gas tracers ( @xmath11@xmath12 ) , suffer from poor resolution ( on the order of kpc ) and sensitivity .
untargeted galactic plane surveys avoid these issues , while simultaneously offering the ability to detect new , less obvious regions of star formation within our galaxy .
other untargeted surveys have been productive in identifying extended emission , such as in co@xcite , oh@xcite and hi@xcite .
however , hmsf occurs in regions of dense molecular gas , and so mapping high - density tracers serves well to identify target regions .
co(10 ) has a relatively low effective critical density ( @xmath13 @xmath1410@xmath15@xmath12 ) compared to other tracers , such as nh@xmath2(1,1 ) ( @xmath1410@xmath16@xmath12 ; @xcite ) and hc@xmath2n ( 43 ) ( @xmath1410@xmath17@xmath12 ; @xcite ) , which have been mapped by the h@xmath5ogalactic plane survey ( hops ; @xcite ) .
hops has been successful in identifying previously unknown sites of star formation , as well as probing the structure of the milky way s spiral arms .
another , even higher density gas tracer useful for detecting hmsf and mapping the structure of our galaxy is cs .
the ground state transition @xmath18 for cslies within the 7 mm waveband , and has an effective critical density @xmath13 @xmath19@xmath12 at 10k @xcite .
previous large scale observations towards specific regions , such as rcw 106 @xcite for csand other high - critical - density molecules have been productive in hmsf research @xcite .
untargeted observations of csover bright infrared regions such as rcw 106 and regions without infrared emission reveal the full population of star - forming candidates to a small sensitivity limit .
the 7 mm waveband also contains the 44ghz 7(0,7)6(1,6 ) a@xmath3 class i methanol maser line , which is the brightest of the class i species .
previous observations of the class i maser have almost always been targeted ( e.g. @xcite ) , which limits and biases our understanding of this astrophysical phenomenon .
the 7 mm waveband includes other spectral lines detailed in table [ tab : spectral_lines ] . to help develop our understanding of hmsf using the diagnostics in the 7 mm waveband ,
we have devised the malt-45 untargeted galactic plane survey .
cshas a slightly higher effective critical density compared to nh@xmath2(@xmath13 @xmath19@xmath12 vs. @xmath1410@xmath16@xmath12 ) , but suffers from freeze - out on to dust grains in the coldest and most dense regions @xcite .
nh@xmath2appears to be resilient to freeze - out in these regions @xcite , therefore by comparing the dense gas tracers from malt-45 and hops , we are able to identify these cold , dense regions in our galaxy by analysing the cs / nh@xmath2ratio and nh@xmath2temperature .
malt-45 simultaneously surveys c@xmath20s ( 10 ) , which is a less abundant isotopologue of cs(assumed cs / c@xmath20s ratio of 24.4 ; @xcite ) .
a comparison between csand c@xmath20s allows us to measure optical depth at these locations .
c@xmath20s data are not formally presented in this paper , but a preliminary data reduction has been performed for an optical depth investigation in section [ sec : cs_nh3 ] .
there are two varieties of ch@xmath2ohmaser : class i and ii .
the class ii variety is well known and studied ; it is radiatively excited by young stellar objects with sufficient mass to be considered hmsf .
thus , it is vitally important as it occurs only in regions of hmsf , but only at a specific evolutionary state @xcite .
in contrast , class i masers are poorly understood .
research has shown that the class i maser is collisionally excited , and associated with shocked gas @xcite , but not necessarily only from outflows , as seen in @xcite .
it is known to occur in both low - mass @xcite and high - mass regions of star formation @xcite .
most known class i masers have been discovered in targeted searches towards other sources , such as 6.7ghz class ii masers @xcite , hmsf regions with and without uchiiregions @xcite and extended green objects ( egos ; @xcite ) . because class i masers have been found towards a range of sources and evolutionary states , their conditions for excitation remain uncertain . by conducting a sensitive and untargeted survey
, malt-45 reveals the full population of class i masers to a small sensitivity limit ( @xmath21 sensitivity of 4.5jy ) .
this eliminates biases towards known tracers inherent in targeted observations , and helps understand their conditions of excitation .
siomasers are commonly associated with oxygen rich asymptotic giant branch ( agb ) or red supergiant ( rsg ) stars , occurring in the extended atmosphere within a few stellar radii @xcite .
these masers are useful in determining radial velocities to such stars to within a few @xcite .
@xcite conducted a survey for siomasers at 86ghz towards evolved stars , and demonstrated their use in probing galactic kinematics through the maser velocities .
additionally , siomasers have been detected towards star - forming regions , but despite an extensive search towards a wide range of hmsf regions , only three are known ( orion source i , w51 and sgr b2 ; @xcite ) .
siomaser emission in orion provides the best evidence of a high - mass accretion disc @xcite .
malt-45 has the potential to reveal many new sources of sio(10 ) maser emission , both towards evolved stars and star - forming regions .
along with the maser emission , malt-45 also includes sio(10 ) @xmath22 , which is thermal line emission .
thermal siois commonly found in star - forming regions , produced when silicate - bearing grains are destroyed .
thus , it is a good tracer of shocked gas and outflows @xcite . in this paper
, we focus on detailing the survey from @xmath23 , @xmath24 and present results of the global properties of cs(10 ) , class i ch@xmath2ohmasers and sio(10 ) thermal and maser emission .
this paper is one of a series ; another paper will be dedicated to maser follow - up observations , and another for clump analysis with csdata . in this paper , we focus on comparing : csemission to nh@xmath2and class i ch@xmath2ohmasers ; class i ch@xmath2ohmasers to class ii ch@xmath2oh , h@xmath5oand ohmasers , thermal sioand infrared emission , and 95ghz class i ch@xmath2ohmasers ; and siomasers to infrared emission . we emphasise that the science cases presented in this paper are used primarily to highlight a few intriguing results and discuss some qualitatively plausible explanations .
full analyses for these cases will appear in future publications .
malt-45 is an untargeted galactic plane survey for spectral lines which are commonly bright in star - forming regions at 45ghz ( 7 mm waveband ) .
we have so far observed 5 square - degrees within the region bounded by @xmath23 , @xmath24 .
malt-45 observations were conducted on the australia telescope compact array ( atca ) , which provides 2@xmath25 2048mhz broadband continuum windows for observing .
section [ sec : spectral_lines ] discusses the primary lines surveyed , and their rest frequencies dictate the positions of the broadband windows for malt-45 . within the frequency ranges of the broadband windows ,
we survey for 12 spectral lines ; see table [ tab : spectral_lines ] . [
cols="<,^,^,^,^,^,^,^,^,^ " , ] [ tab : sio ] @xmath26c - presence of a 1665mhz ohmaser from @xcite ; s - presence of a 1612mhz ohmaser from @xcite ; w - presence of a h@xmath5omaser from @xcite .
the positions of class i ch@xmath2ohmasers have been compared with the published positions of class ii ch@xmath2oh , h@xmath5oand ohmasers to establish associations . when considering the maximum distance between masers that still constitute an association , we use 60 arcsec from the class i position as suitable for defining associations .
@xcite analysed a large distribution of clumps associated with hmsf in their source sample , and find that all clumps have effective radii within 55 arcsec .
additionally , we note that increasing the maximum offset beyond 60 arcsec does not increase the number of unique associations .
these findings suggest for our purposes that 60 arcsec is sufficient to group masers within a single star - forming region . we have chosen not to define an association based on source distance or size , as it is difficult to implement across a wide range of objects
, kinematic distances may be erroneous , and maser geometry may be wide - spread .
an angle of 60 arcsec on the sky corresponds to approximately 1.5pc at a distance of 5kpc .
there are two targets where the spread of class i ch@xmath2ohmaser emission is large and these have been considered as two separate maser regions .
the weaker masers ( [email protected] and [email protected] ) near to each of these bright masers ( [email protected] and [email protected] ) possess a distinct spectral velocity component and are sufficiently spatially offset to be considered a different site ( 54 and 87 arcsec , respectively ) .
see the spectra in appendix b ( available online ) .
these sources are marked with a `` p '' in table [ tab : ch3oh ] .
integrated intensity values for csand sio@xmath22 have been included for each maser , where possible , in table [ tab : ch3oh ] .
emission was integrated @xmath2810from the peak velocity of the maser .
values were not included if the emission was not significant ( @xmath29 ) .
extended sio@xmath22 emission is detected in some regions , as shown in figure [ fig : cs_sio ] , which shows that sio@xmath22 emission tends to be associated with bright peaks of csemission .
note that emission from the sio@xmath22 is typically weaker than for cs , hence the signal - to - noise is lower for this transition ; figure [ fig : cs_sio ] shows all emission above @xmath30 .
discussion on the relation between class i ch@xmath2ohmasers and sio@xmath22 emission is contained in section [ sec : sio_ch3oh ] . the vibrationally excited sio(10 )
lines were detected towards 47regions , as shown in figure [ fig : sio_glimpse ] .
the properties of each maser region are classified in table [ tab : sio ] .
the search procedure for siomasers was the same as with ch@xmath2ohmasers ; they were identified by visual inspection of the data , and had velocities categorised by a @xmath30 cutoff from the peak velocity .
velocities are specified to 0.1 , and flux densities to two significant figures .
many masers have low signal - to - noise values , but are considered real detections based on their association in glimpse , and simultaneous detection of emission in another siomaser transition .
the positions of siomasers have been compared with ohand h@xmath5omasers ; see table [ tab : sio ] and section [ sec : sio_discussion ] .
ch@xmath2ohmaser species were searched at these positions , but none were found , implying that any class i ch@xmath2ohmaser emission associated with the siomasers has a peak flux density @xmath314.5jy .
peak intensity maps are presented in appendix c , and each un - smoothed maser spectrum can be seen in appendix d. appendices are available online .
m ) image with examples of detected siomasers towards infrared stars .
circles are centred on the peak position of maser emission , and are beam - sized ( 1 arcmin ) .
all three of these siomaser regions contain the @xmath8 and @xmath7 lines , but all other maser region variants have similar infrared colours.,scaledwidth=47.0% ] figure [ fig : sio_glimpse ] shows the locations of siomasers detected by malt-45 with respect to infrared emission from glimpse .
all siomasers are detected towards point - like , isolated sources that appear to be evolved stars , based on their infrared colours and lack of obvious associations with star formation , such as irdcs or extended infrared emission .
figure [ fig : sio_glimpse_example ] shows the relative locations of three siomasers compared to infrared stars ; these masers were chosen as they show various colours of infrared stars towards all siomasers .
it can be seen that the maser locations are not identical to the stars , but are within one beam .
therefore , we can not conclusively argue that each siomaser site is coincident with an evolved star , although the evidence we have suggests this is the case
. further higher spatial resolution observations of the maser sites will be able to demonstrate conclusively whether or not each siomaser is associated with an evolved star . of the 47 siomaser regions ,
45 contain @xmath8 emission , 37 contain @xmath7 , and only 3 have @xmath32 emission .
we have found no previously published data for any of these siomasers and consider them all to be new detections .
some siomaser regions have greater @xmath7 emission than @xmath8 , and two have no detected emission @xmath8 ; this is discussed further in section [ sec : sio_discussion ] .
there are also 12 siomaser regions with an associated ohor h@xmath5omaser ; see section [ sec : sio_discussion ] .
parameters used in determining kinematic distances are the same as those found in section [ sec : results_ch3oh ] , except for a more generous velocity uncertainty @xmath33 , and @xmath34 .
four masers are labelled at the far kinematic distance , because the position and errors for a near distance are unrealistic ; see table [ tab : sio ] . by analysing the rms noise values per pixel
distributed across the malt-45 region , we find that the noise levels may vary by up to a factor of three due to some data being flagged or insufficient time on the telescope to complete the observations . however , we determine that approximately 95 per cent are consistent with the values listed in table [ tab : spectral_lines ] , and only 5 per cent show rms levels worse than these values .
thus , we assess that to first order the survey has uniform sensitivity and only a small fraction of the area has sensitivity poorer than nominal .
tables [ tab : ch3oh ] and [ tab : sio ] include the rms value for each detected maser peak , which help to identify the completeness of the malt-45 survey .
we detect many masers at less than @xmath21 significance . however , we consider the survey complete at the @xmath21 level for 95 per cent of the survey area .
this completeness limit is equivalent to 4.5jy for class i ch@xmath2ohmasers , and approximately 4.3jy for each siomaser line .
figure [ fig : cs_hops ] shows that nh@xmath2(1,1 ) from hops is closely correlated with bright cs(10 ) emission , but with prominent exceptions ; examples include [email protected] , [email protected] and [email protected] .
we calculate that malt-45 is more sensitive to cs(10 ) emission than hops is to nh@xmath2(1,1 ) by a factor approximately 9 , therefore cswithout nh@xmath2 in figure [ fig : cs_hops ] may be due to the superior sensitivity of malt-45 over hops .
however , it is also likely attributable to the phenomena described by @xcite , who reason that csis an `` early - type '' star - forming molecule , tracing quiescent gas , while nh@xmath2is `` late - type '' , associated with dense cores .
for this reason , despite having a higher effective critical density , cswithout nh@xmath2may be tracing molecular clouds before cold , dense cores have formed .
additionally , @xcite and @xcite indicate that cstends to deplete in core collapse .
this means a relatively high abundance of nh@xmath2to csshould reflect a cold , dense core , which is a @xcite `` late - type '' scenario .
as results indicate later in this section , nh@xmath2is not persistent in even further developed stages of hmsf , as it is not detected or very faint .
we distinguish `` late - type '' sources from these evolved sources through bolometric luminosities and the presences of infrared features , such as stars or hiiregions .
we produced a map of the ratio between the integrated intensities of csand nh@xmath2 in order to highlight the differences between their distributions . before performing the quotient of integrated intensity
, the csdata were smoothed to the hops nh@xmath2resolution ( @xmath142arcmin ) .
we used a cutoff at 0.5k(@xmath36 ) of the less sensitive nh@xmath2data ; csdata is @xmath37 everywhere nh@xmath2is @xmath37 .
integration of emission was taken between -100 and -40 , to eliminate subtle baseline effects which decrease the robustness of analysis ; this velocity range is where most of the emission in this part of the galaxy lies ( see figure [ fig : cs_thrumms ] for the longitude - velocity plot ) .
quotients use units of main - beam temperature for both csand nh@xmath2 . see figure [ fig : ratio ] for the full ratio map . moving inward from the clump edges ,
the cs / nh@xmath2ratio quickly changes .
we observe the amount of both csand nh@xmath2to increase , with the nh@xmath2intensity increasing more `` quickly '' .
edge values are mostly about 2 , while clump centres are usually between 1 and 0.5 , but can be as low as 0.4 .
we now discuss the ratio map qualitatively , reserving radiative transfer modelling for future work . in almost every clump ,
the centre is dominated by nh@xmath2over cs .
this could be due to any or a combination of enhanced nh@xmath2 , depleted cs , or both species being enhanced but csbecoming optically thick .
the hyperfine line structure of nh@xmath2means that it remains optically thin at higher column densities than cs . to test for optically thick cs
, we compared cswith malt-45 c@xmath20s .
we conducted the analysis of csagainst c@xmath20s in a similar manner to that of csagainst nh@xmath2 .
both integrated maps were smoothed to the same resolution , the velocity range of integration was -100 to -40 , and emission was limited to detectable c@xmath20s ( 0.15k ) .
a preliminary analysis of the c@xmath20s data was undertaken only to investigate the csoptical depth within these clumps .
figure [ fig : c34s_glimpse ] shows contours of the ratio of c@xmath20s / csagainst glimpse , with contours of cs / nh@xmath2 .
progressively higher levels of c@xmath20s / csindicate csself - absorption , but appear to be slightly offset from the lowest cs / nh@xmath2ratios .
this implies that the relatively fainter csemission in the presence of an abundance of nh@xmath2is due to both depletion and optical thickness . additionally , the higher ratios of c@xmath20s / csappear to be associated with bright infrared emission , and correlate well thermal sioemission .
we speculate that high c@xmath20s / csratios may be due to high csoptical depth as a result of high csrelative abundance in the presence of an outflow or shocked gas .
c@xmath20s is not detected in other regions of these clumps ; aside from having poor signal - to - noise , this is likely due to the isotopologue depleting similar to cs , and emission is only apparent in hotter , shocked regions where these molecules have been released into the gas phase . a similar analysis to the one we have conducted was performed in @xcite towards two cold , dense cores .
their results indicate significant depletion of csand c@xmath20s towards the core centres , while nh@xmath2peaks in abundance .
we have not undertaken a detailed analysis of the temperature or mass of the malt-45 molecular gas clumps in this paper ; this will be investigated in a future publication focusing on the csemission . in future work
, we will present radiative transfer modelling ( including depletion , time - dependant chemistry and the relative sensitivity limits of the various surveys ) of the dust , csand nh@xmath2emission to quantitatively understand the physical reason for these effects .
the following subsections briefly discuss three unusual clumps , which have the largest cs / nh@xmath2ratios .
this source is one of the brightest in csacross the malt-45 survey region ( peak of 1.5k ) , and is the brightest in thermal sioemission ( peak of 0.17k ) .
the source also contains all the common star formation region masers : class i and ii ch@xmath2oh , h@xmath5oand oh . the csspectrum has a broad profile of @xmath1410fwhm , possibly due to a powerful molecular outflow .
a strong outflow would not be surprising given the strength of the thermal siodetected , relative to all other siosources in malt-45 .
@xcite have targeted this source which has a high iras far - infrared luminosity , as well as detection of cs(21 ) by @xcite .
@xcite present data for sio(21 ) and note that of the four sources they discuss , this source is brightest in sio , and conclude that an outflow is present due to the spectral profiles of csand sio .
@xcite also discuss the requirement of ob stars to produce the ionised gas seen in other observations .
the evolved nature of the source is the likely reason for the faint nh@xmath2 , because it is being destroyed or dispersed by outflows or young stellar objects .
the cs / nh@xmath2ratio of this source is approximately 3 . of the three unusual sources discussed in this section ,
this is the only one without detected thermal sioemission .
a small region ( @xmath1430 arcsec ) of the clump has a constant c@xmath20s / csratio of 0.08 , and is coincident with a h@xmath5oand class ii ch@xmath2ohmaser , as well as what appears to be a star in glimpse .
a new class i ch@xmath2ohmaser is detected by malt-45 , but is slightly offset from this position ( @xmath1430 arcsec ) .
this source has been targeted by @xcite to determine disc properties through coobservations , and is listed with a very high bolometric luminosity ( @xmath38 l@xmath39 ) . the red msx source ( rms ; @xcite )
survey also lists a _
k_-band magnitude of 5.9 .
these values suggest that the source is not embedded . as a star is visible in glimpse and the source
is not embedded , it may be assumed that the star ( and the region ) are post star - forming .
however , we are confident that the associated star is not evolved ( beyond the main - sequence ) , as a near - infrared ( nir ) spectrum of the source contains br@xmath40 , indicating the presence of an hiiregion . the spectrum is taken from the rms survey , although the data remain unpublished ; @xcite present detailed analysis of the nir spectroscopy , but note that it does not cover this specific source .
we conclude that this source is somewhat evolved because of its exposure from natal clouds , but is pre - main - sequence because of the class ii ch@xmath2ohmaser emission .
again , the evolved nature of this region is the likely reason for the relatively faint nh@xmath2 .
this region is the brightest hiiregion within the malt-45 survey , and appears to harbour evolved sources .
c@xmath20s emission is detected within this source , and the c@xmath20s / csratio varies from 0.09 to 0.14 .
thermal sioemission is also detected towards the source . as well as a previously known class i ch@xmath2ohmaser
, this region is associated with 1612- , 1665- and 1667-mhz ohmasers @xcite , and many h@xmath5omasers from hops @xcite .
the peak velocity of the class i ch@xmath2ohmaser is @xmath41 , @xmath42for oh , and @xmath43 to @xmath44for the h@xmath5omasers .
malt-45 csalso has a double - peak profile in this region ; a strong , broad peak is associated with @xmath45and a weaker one at @xmath46 , both at the same position ( 16@xmath4722@xmath4806@xmath49 , @xmath2750@xmath5006@xmath5117@xmath52 j2000 ) .
@xcite conducted observations of h90@xmath53 towards this source , which also reveal two peaks of emission , and reason that it contains many o - type stars .
the peak velocity of each h90@xmath53 component varies greatly over the region , from @xmath54 to @xmath55and @xmath56 to @xmath57 , where the second peak is weaker and not present across the observed area .
note that the position of peak malt-45 csis outside of the data presented in @xcite ; it is possible that the spatial resolution of our csdata is limiting our ability to interpret the physics of this region .
however , it is interesting that the peak velocities of csare quite different to h90@xmath53 ; this may indicate complex structure of the source , and indeed this is discussed in @xcite .
the weak presence of nh@xmath2(1,1 ) emission is mysterious .
the high cs / nh@xmath2ratio may be attributable to the region being at a higher temperature , preventing csdepletion .
@xcite attempted a nh@xmath2(1,1 ) and ( 2,2 ) analysis of this clump , but were unable to detect significant emission .
they suggested that the relatively faint nh@xmath2emission could be due to extreme self - absorption .
@xcite comment that in general , cs(21 ) and hco@xmath58 ( 10 ) have similar velocity structure , except at this location ; the hco@xmath58 has a strong self - absorption feature .
however , given that the source is clearly evolved , it is consistent with the other sources discussed in having little nh@xmath2emission .
the cause for the high ratios ( cs / nh@xmath2@xmath59 3 ) in our analysis appears to stem from evolved regions of hmsf .
the other clumps are more difficult to classify - are they less evolved ?
do they have significantly different temperatures or column densities ?
more analysis is needed , and will be presented in a future malt-45 paper with csclump analysis .
ohand csintegrated intensities .
units of integrated intensity are jyfor class i ch@xmath2oh , and kfor cs .
there is no significant correlation observed in either subpopulation , or the total population.,scaledwidth=47.0% ] table [ tab : ch3oh ] includes the integrated flux density of cs(10 ) in each maser region .
the kinematic distance calculated for each source is used to determine the luminosities .
figure [ fig : ci_cs ] compares the integrated csintensity with the integrated class i ch@xmath2ohmaser intensity .
figure [ fig : cs_ch3oh ] shows that most class i ch@xmath2ohmasers appear to be associated with a bright peak of csemission ( 73/77 masers with @xmath590.11k of peak intensity cs ) .
this is quantified ( using integrated intensity ) in figure [ fig : ci_cs ] .
however , there is no correlation observed between the relative intensity of a class i ch@xmath2ohmaser and its associated cs .
a class i ch@xmath2ohmaser with csemission suggests association with a star - forming region , because sufficient gas and shocks exist to power their emission .
however , without being able to associate these lines with other diagnostics of star formation , it is not certain that the masers are stimulated by star formation ; cloud - cloud collisions have been found to stimulate class i ch@xmath2ohemission @xcite . while the possibility of being a non - star - forming maser extends to all candidates without additional tracers ( such as class ii ch@xmath2ohmasers ) , we expect that these are exceptional cases , and that the majority if not all class i ch@xmath2ohmasers in the malt-45 sample are associated with star formation . [
fig : cs_ch3oh_vel_hist ] a comparison of class i and class ii ch@xmath2ohmasers was undertaken by @xcite , and found a broad gaussian distribution of velocity differences .
we have performed a similar analysis with the peak velocities of both csand class i ch@xmath2ohmasers from the malt-45 data .
the resulting histogram of the velocity difference of ch@xmath2ohwith respect to csis found in figure [ fig : cs_ch3oh_vel_hist ] .
the distribution is approximated by a gaussian with a mean velocity difference [email protected] standard deviation [email protected] .
the median velocity difference is -0.1 .
the fwhm of the population presented by @xcite is approximately 7.8 , while ours is approximately 3.5 .
because csis a dense gas tracer , it will occur in cloud cores and thus trace systemic motions accurately . given the resulting gaussian of velocity differences
, this means that class i ch@xmath2ohmasers are also good indicators of systemic velocities .
@xcite discuss class i ch@xmath2ohmasers having similar velocities to the quiescent gas , due to a small amount of gas actually being shocked in these regions . as is noted in @xcite ,
@xcite indicate that ch@xmath2ohemission arising without thermal sioemission is due to mild shocks , destroying dust grain mantles but not dust grain cores .
hence the dispersion in class i ch@xmath2ohmaser velocities is thought to be small , and is reflected here .
this also indicates that the large dispersion seen by @xcite is likely due to the large class ii ch@xmath2ohmaser velocity spread .
this is readily explained by class ii ch@xmath2ohmaser emission radiating from close to a high - mass young stellar object , and therefore having a higher dispersion .
the histogram in figure [ fig : cs_ch3oh_vel_hist ] appears to have a slight blueshifted component .
this is also discussed in @xcite as perhaps due to maser emission being easier to detect on the near - side of a cloud to the observer .
while the median velocity offset is negative , it is statistically insignificant .
a larger data set is required to determine if the blueshift phenomena is real or not . as discussed in section [ sec : spectral_lines ] , sio@xmath22 is a good tracer of shocked gas and outflows .
regions of thermal sioand positions of class i ch@xmath2ohmasers correlate weakly ; 23 of 77 masers ( 30 per cent ) have at least a @xmath60 peak of sioemission . as class
i masers are collisionally excited , it is perhaps not surprising to find some association with thermal sio .
however , the majority of masers are detected without sio@xmath22 emission .
this could be due to : insufficient sensitivity to extended sioemission ; class i ch@xmath2ohmasers having stronger intensity than thermal sio ; or not all class i ch@xmath2ohmasers being associated with sio@xmath22 emission .
@xcite found that thermal ch@xmath2ohemission may be triggered by less energetic shocks , which destroy dust grain mantles but not their cores , which explains a lack of sioemission . in more energetically shocked regions ( such as a jet or molecular outflow ) ,
ch@xmath2ohmay not survive , which may explain sioemission without ch@xmath2oh .
three regions with strong sioemission ( 0.04k ) are not associated with any maser species : [email protected] , [email protected] and [email protected] .
each of these regions has a cspeak of at least 0.4k ( @xmath61 ) , but no obvious infrared emission .
table [ tab : ch3oh ] lists 28 class i ch@xmath2ohmasers with detected sioemission , implying that these three regions without masers are in the minority .
this could be due to the variability of any masers making them undetectable during observations , or that conditions in these regions are unfavourable for masers to exist .
it is clear that sensitivity to ch@xmath2ohand / or siowill play a part in the detectability of each .
however , the fact that we see regions with both ch@xmath2ohand sio , regions with only ch@xmath2ohmasers and regions with only sioemission suggests that there are multiple conditions under which each of these will exist .
if we assume that all class i ch@xmath2ohmasers are associated with star formation , then given that the majority of masers are not associated with thermal sioemission , these masers may be tracing weaker shocks that do not produce detectable sioemission .
30 per cent of ch@xmath2ohmasers also show thermal sioemission .
it is not clear whether the masers and sioare physically connected , although a reasonable interpretation is that the siooriginates in a jet , where as the masers are associated with the surrounding outflow . in a subsequent paper for malt-45
, maser follow - up observations will be conducted at high spatial resolution .
we will simultaneously observe siowith greater sensitivity , and using this data we will more thoroughly investigate the association of class i ch@xmath2ohmasers and thermal sioemission .
ohmaser with associated `` green '' emission .
this region is not classified as an ego by @xcite , but may have similar properties .
the circle represents 15 arcsec centred on the peak of the class i ch@xmath2ohmaser emission , which is the expected positional error.,scaledwidth=47.0% ] glimpse mapped the galactic plane in four infrared bands ( 3.6 , 4.5 , 5.8 and 8.0@xmath62 m ) .
these images are useful for investigating potential relationships between infrared emission and malt-45 detections .
a common example of how glimpse data is used in star formation research is by combining the 3.6 , 4.5 and 8.0@xmath62 m bands in a three - colour image ( blue , green and red , respectively ) to reveal egos ( extended green objects ) .
as discussed in @xcite , egos have a strong correlation with class i ch@xmath2ohmaser regions .
the catalogue provided by @xcite was used to investigate the presence of egos over the survey region .
the catalogue covers @xmath63 and @xmath64 , @xmath65 , and lists 22 candidate egos within the malt-45 survey region .
of the 77 class i masers found by malt-45 , 12 are associated with an ego .
association is credited for any spatial overlap within 60 arcsec .
of the 12 ego - associated class i ch@xmath2ohmasers , 11 were previously known ( [email protected] is new ) .
egos are classified by visual inspection and as such are intrinsically subjective .
however , it is clear that many of the new class i ch@xmath2ohmaser detections are associated with glimpse sources with `` green '' ( i.e. 4.5@xmath62 m excess ) emission ; see figure [ fig : ch3oh_green ] for an example source .
@xcite explains that the green emission band of glimpse contains both h@xmath5 and co ( 10 ) lines , which are excited by shocks .
indeed , some of the new masers detected by malt-45 have similar `` green '' emission as with egos , although less prominently so .
this shows a close association between shocks traced by egos and class i ch@xmath2ohmasers .
@xcite found a relationship between the peak flux density of the 95 and 44ghz class i methanol masers ( which come from the same transition family ) .
they found that the peak flux density of 44ghz masers is approximately three times greater than the 95ghz masers .
ohmasers from malt-45 with known 95ghz class i ch@xmath2ohmaser emission .
there are 19 masers with both 44 and 95ghz emission .
the line of best fit for these data is plotted alongside the line of best fit presented by @xcite .
vertical lines indicate the minimum and maximum recorded values for 95ghz masers , if available , on top of the plotted average .
units of peak flux density are in jy.,scaledwidth=47.0% ] information for 95ghz class i ch@xmath2ohmasers towards malt-45 44ghz masers are obtained from @xcite and @xcite . in total , we have 19 class i maser sources with both 44 and 95ghz peak flux densities .
when faced with more than one peak flux density over various epochs , we use the average of all values . analysing the distribution in figure [ fig:95ghz_comparison ]
, we find that our results agree with those of @xcite .
our line of best fit is @xmath66 , with an @xmath67-value of 0.98 , while the @xcite fit is @xmath68 , with an @xmath67-value of 0.53 .
the uncertainty of our fitted slope is @xmath69 .
note that our fit does not use the 44ghz maser with 120jy ; if we do include it , our fit becomes @xmath70 , with an @xmath67-value of 0.91 and uncertainty @xmath71 .
this is not significantly different , but our slope closely agrees with @xcite when the outlier is omitted . in this section , we discuss other maser species and transitions and their association with the class i ch@xmath2ohmasers detected in malt-45 .
association with each maser is credited if the emission in question is within 60 arcsec of the class i maser position . , using peak flux densities.,scaledwidth=47.0% ] in the past
, class i methanol masers have been identified from targeted searches towards class ii maser sites @xcite .
for the first time , we are able to make an unbiased comparison of the rate at which each class of methanol maser occurs and their degree of overlap . within the malt-45 survey region ,
54 6.7ghz class ii methanol masers have been detected by the mmb @xcite . to compare against malt-45 class i maser regions ,
7 class ii masers have been grouped together into `` clusters '' , as they are close together ( within 60 arcsec ) .
the clusters containing grouped masers are [email protected] , [email protected] , [email protected] , [email protected] and [email protected] .
while the velocities of each maser site grouped together may be different , there is significant overlap in the range of emission ( not greater than 9apart ) . without high - resolution positions , inferring which class ii maser is associated with the class i emission is difficult , and it is more practical to only compare their clustered position , especially as a class i ch@xmath2ohmaser is almost certainly associated with the class ii source .
therefore , we compare 47 class ii methanol masers positions against the 77 class i sources detected in malt-45
. there are 28 class i ch@xmath2ohmasers associated with the clusters , and 16 of these class i masers are new ; see table [ tab : ch3oh ] .
an association exception is made for [email protected] , where the class ii ch@xmath2ohmaser is slightly further offset than 60 arcsec from the malt-45 class i ch@xmath2ohmaser position , but is well studied and known to be related @xcite .
additionally , the next closest class i and class ii ch@xmath2ohmaser pair is approximately 100 arcsec , which aids to rule out mis - association .
thus , 60 per cent of class ii masers have a class i counterpart ( 28/47 ) , while approximately 36 per cent of class i masers have a class ii counterpart ( 28/77 ) .
@xcite reports a 38 per cent association of class ii sources with a 95ghz class i maser counterpart .
given that the 44ghz species is typically stronger than the 95ghz transition , and that the malt-45 sensitivity is better , we perhaps expect our reported class ii - to - class i association to be greater .
as discussed in section [ sec:95_masers ] , @xcite finds the relative peak flux density of 95ghz class i ch@xmath2ohmasers to 44ghz class i ch@xmath2ohmasers to be approximately one - third .
assuming @xcite had a sensitivity cutoff for 95ghz masers at 3jy , we can limit the results of the 44ghz population by a 9jy cutoff .
there are 17 44ghz masers with a peak flux density greater than 9jy associated with a class ii ch@xmath2ohmaser ( 36 per cent ) , which is only slightly lower than the proportion of associations found by @xcite .
therefore , our results are reasonably consistent . considering that the malt-45 sensitivity to class i ch@xmath2ohmasers is better than that of @xcite by approximately a factor of 2 , and
that we search for 44ghz masers instead of 95ghz masers , we are assured that there is a higher association rate than previously reported . the luminosities of class i and class ii ch@xmath2ohmasers were compared , but no correlation was found ( @xmath67-value of 0.03 ) ; see figure [ fig : classii_maser_scatter_lum ] .
the relative populations for class i masers with and without a class ii source are compared in figure [ fig : classii_maser_hist ] ; the histogram reveals that class ii sources are associated with a wide range of class i brightnesses , which reaffirms the lack of correlation . with sensitive ,
targeted 44ghz observations of class ii masers , more data can be collected for a better analysis . note that the search radius should be at least an arcmin to cover the range of separations between class i and class ii masers ; see section 4.2 of @xcite . omasers .
this plot shows that h@xmath5omasers tend to be associated with brighter class i methanol masers.,scaledwidth=47.0% ] water masers are another maser species that are collisionally excited .
therefore , we might expect a close association between class i methanol and water masers .
hops has found h@xmath5omasers in an untargeted way , and catalogues their findings into three association groups : star - forming , evolved star , and `` unknown '' @xcite .
@xcite also presents a catalogue of h@xmath5omasers , typically associated with class ii ch@xmath2ohmasers and/or ohmasers .
hops has detected a total of 48 h@xmath5omasers within the malt-45 survey region @xcite .
of these , hops has classified 10 as being associated with an evolved star , leaving 38 as star - forming or `` unknown '' . including the @xcite detections brings the total to 71 , but there is a large overlap with the hops catalogue , as well as many h@xmath5omasers near to others ( within 60 arcsec ) .
similar to the mmb class ii ch@xmath2ohmasers , after grouping masers into clusters , we compare malt-45 class i ch@xmath2ohmasers against a total of 43 h@xmath5omaser clusters . of these 43 h@xmath5omaser clusters ,
27 are associated with a class i ch@xmath2ohmaser ( 63 per cent ) .
it is possible that the class i ch@xmath2ohpopulation have more h@xmath5omaser associations , but have not been found yet .
@xcite targeted observations towards known sources , and hops has a lower - bound sensitivity between 5 and 10jy .
follow - up observations for h@xmath5omasers towards new class i ch@xmath2ohmasers may be productive in identifying new h@xmath5omasers . in the hops catalogue recently published @xcite , the water masers [email protected] and [email protected] were not redetected in the high - resolution follow - up survey .
the [email protected] water maser is associated with a malt-45 class i ch@xmath2ohmaser , but this was a known class i ch@xmath2ohmaser region ( along with other maser species ) .
the [email protected] water maser , however , is slightly offset from a new class i ch@xmath2ohmaser .
the offset is enough to not be included as an association , but it lends credence that this variable water maser is likely highlighting star - forming activity . @xcite discuss the positioning of class i methanol maser emission on a hmsf evolutionary timeline , especially compared to other masers .
class ii methanol masers have been characterised well in terms of their position in a hmsf timeline , associated with millimetre sources , but typically not hiiregions . unlike class
ii masers , class i masers are currently thought to be associated with both early- and late - type hmsf .
this stems from finding class i masers associated with outflows , but also with ohmasers . to date
, no unbiased comparison between ohand class i methanol masers has been performed . with the first unbiased sample of class i masers
, we are able to analyse the relative populations and perhaps refine class i masers on an evolutionary timeline .
@xcite mentions that class i masers associated with ohmasers are known primarily due to the class ii methanol maser emission present , as most class i sources have been found toward class ii sources .
@xcite discuss how ohfavoured maser regions have a greater number of associated hiiregions than class ii masers , indicating that ohmasers typically signpost later stages of an evolutionary timeline than class ii masers . combining the ohmasers found by @xcite and @xcite
, there are 16 star - forming ohmasers within the malt-45 survey region .
the other ohmasers in @xcite and @xcite are associated with evolved oh / ir stars , determined from simbad and individual source descriptions within @xcite .
94 per cent ( 15/16 ) of star - forming ohmasers are associated with a class i ch@xmath2ohmaser . of these 15 ,
only 2 masers are without a 6.7ghz class ii methanol maser ( 13 per cent ; [email protected] and [email protected] ) .
it is difficult to contrast maser regions containing ohbut not class ii methanol , as almost every ohmaser in the survey region is associated with a class ii ch@xmath2ohmaser . if a population of ohmasers without class ii ch@xmath2ohwas isolated , a better comparison with class i ch@xmath2ohcould be established .
the malt-45 survey results merely indicate that class i ch@xmath2ohis associated with all maser species found .
there are 30 class i ch@xmath2ohmaser regions with associated class ii ch@xmath2ohor ohmaser emission , indicating late stages of star formation .
the remaining 47 masers are unaccounted for ; if these are all associated with early - type star - formation , then class i ch@xmath2ohmasers appear to be more closely associated with earlier stages in an evolutionary timeline .
ohcomponent , particularly with class ii ch@xmath2ohas well , but there is a large population of class i masers without any associations.,scaledwidth=47.0% ] a venn diagram of the maser associations can be seen in figure [ fig : maser_venn ] . there appears to be a strong overlap between class i and class ii ch@xmath2ohmasers , with and without the presence of h@xmath5oand / or ohmasers .
this may suggest that the majority of class ii ch@xmath2ohmasers are associated with class i ch@xmath2ohmasers because of an abundance of ch@xmath2oh .
the smaller population of class ii ch@xmath2ohmasers without class i ch@xmath2ohemission ( 40 per cent ; 19/47 ) may indeed have no association , or even weaker class i ch@xmath2ohemission than is detected by malt-45 .
there is a large population of class i ch@xmath2ohmasers without any other maser counterpart . as discussed , these may highlight non - star - forming phenomena , young regions of hmsf , or low mass star - forming regions .
however , given the strong association with csseen in figure [ fig : cs_thrumms ] and with other hmsf masers seen in figure [ fig : maser_venn ] , the majority of class i ch@xmath2ohmasers are likely associated with hmsf .
figures [ fig : classii_maser_hist ] , [ fig : water_maser_hist ] and [ fig : oh_maser_hist ] show the luminosity population of class i ch@xmath2ohmasers as stacked histograms .
generally , the presence or absence of other maser transitions does not appear to correlate with the class i ch@xmath2ohmaser luminosity .
@xcite discuss the placement of h@xmath5omasers within an evolutionary timeline for hmsf , and concluded that they are not likely to be associated with any specific phase of a timeline , but rather dependant on the environment .
as h@xmath5omasers are collisionally excited like class i ch@xmath2ohmasers , and class i masers are associated with a wide range of other species ( with and without class ii ch@xmath2oh , oh ) , we tend to agree that the environment plays perhaps the predominant role in determining the likelihood of a region having an associated class i ch@xmath2ohmaser rather than the evolutionary phase .
investigation of the peak flux densities and luminosities of class i ch@xmath2ohmasers versus class ii ch@xmath2oh , h@xmath5oand ohmasers does not reveal any relation .
investigation will be performed again in a subsequent malt-45 maser follow - up paper , where maser regions can be better associated and flux density calibration is able to be more accurately determined .
malt-45 has detected 47regions with at least one of the vibrationally excited lines @xmath6 of sio(10 ) maser emission .
of these , 4 are associated with a ohmaser ( all 1612mhz , [email protected] has an additional 1665mhz ) , and 8 are associated with a h@xmath5omaser .
ohand h@xmath5omasers are common companions with evolved infrared stars @xcite , and indeed glimpse reveals that all siomaser regions have an infrared star associated within 30 arcsec of the peak maser position .
there is no overlap of siomasers associated with both an ohand h@xmath5omaser .
the population presented in this paper associated with ohand h@xmath5omasers is insufficient to perform correlation tests upon .
none of the siomasers detected have been reported previously ; thus , malt-45 is a productive survey for identifying evolved stars not found in other searches .
the sio@xmath8 line generally appears to be the strongest of the three maser lines , and is the most common detection .
there are eleven cases where the @xmath7 line is stronger than @xmath8 , and two regions contain only @xmath7 emission .
this result is somewhat unexpected , as generally the lower @xmath72 transitions should be brighter than the higher @xmath72 transitions .
@xcite predict that siomasers should have brightness decreasing with @xmath6 , and the @xmath7 is often weaker due to overlap with a water line .
neither of the two regions with only sio@xmath7 emission have other masers associated , but the sub - population of sioregions with brighter @xmath7 than @xmath8 have three h@xmath5omasers and two ohmasers associated .
this is almost half of all masers associated with sioregions .
however , we currently can not explain why some @xmath7 masers are stronger than their @xmath8 counterparts .
only three @xmath32 lines are detected , and these are all quite weak , especially compared to the @xmath73 lines within the same region ; it appears that this line is difficult to detect without sensitive observations .
the only siomaser with nearby ch@xmath2ohmasers is [email protected] , but is separated by 33 arcsec .
the same siomaser is 13 arcsec from a sevenster 1612mhz oh maser , which is cospatial with a star in glimpse . as the nearby class
i and class ii ch@xmath2ohmasers are only 4 arcsec apart but separated from the star , we consider both pairs of masers to not be associated .
malt-45 has successfully probed the galactic plane for cs(10 ) , 44ghz class i methanol masers , and the sio(10 ) family of @xmath4 lines .
this is the first survey to produce a large scale unbiased map of cs , and is the first untargeted search of class i ch@xmath2ohmasers and sioemission .
further publications will reveal the full 7 mm star - forming environment , through the remaining lines mapped by this survey .
with time , malt-45 could be extended to map more of the galactic plane , revealing more about the nature of star formation within our galaxy . with the current survey , we have determined : \(i ) across the survey region , the cs / nh@xmath2ratio is low in the centres of many clumps , but higher at the edges .
this is likely due to a combination of increased nh@xmath2abundance and csbeing depleted and/or becoming optically thick in the centres of clumps ; \(ii ) a large population of class i methanol masers are not associated with other masers ( 55% ) ; \(iii ) class
i methanol masers are good indicators of systemic velocities ; \(iv ) class i methanol masers are found towards most known class ii methanol , water and hydroxyl masers within the survey region .
thus , positioning class i methanol on a maser timeline of hmsf is difficult , and the occurrence of class i methanol masers is perhaps more likely due to environmental factors ; \(v ) silicon monoxide masers associated with evolved stars are most commonly detected in their @xmath8 vibrational mode .
when a region contains more than one vibrational mode , the intensity typically descends with vibrational mode ( @xmath6 ) .
however , we find two cases where we only detect @xmath7 and eleven cases where both @xmath8 and @xmath7 are detected , but the @xmath7 is stronger .
the authors sincerely thank the referee , anita richards , for timely and thorough comments that have improved the manuscript .
c. h. jordan thanks the staff of the paul wild observatory for the many hours of assistance lent to the malt-45 project .
the australia telescope compact array is part of the australia telescope national facility which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
shari breen is the recipient of an australian research council decra fellowship ( project number de130101270 ) .
pjb acknowledges support from nsf grant ast-1312597 and the university of florida .
this research has made use of the simbad database , operated at cds , strasbourg , france and nasa s astrophysics data system .
miriad , livedata , gridzilla and asap are software packages managed and maintained by csiro astronomy and space science . | we present the first results from the malt-45 ( millimetre astronomer s legacy team - 45ghz ) galactic plane survey .
we have observed 5 square - degrees ( @xmath0 , @xmath1 ) for spectral lines in the 7 mm band ( 4244 and 4849ghz ) , including cs(10 ) , class i ch@xmath2ohmasers in the 7(0,7)6(1,6 ) a@xmath3 transition and sio(10 ) @xmath4 .
malt-45 is the first unbiased , large - scale , sensitive spectral line survey in this frequency range . in this paper
, we present data from the survey as well as a few intriguing results ; rigorous analyses of these science cases are reserved for future publications . across the survey region
, we detected 77 class i ch@xmath2ohmasers , of which 58 are new detections , along with many sites of thermal and maser sioemission and thermal cs .
we found that 35 class i ch@xmath2ohmasers were associated with the published locations of class ii ch@xmath2oh , h@xmath5oand ohmasers but 42 have no known masers within 60 arcsec .
we compared the malt-45 cswith nh@xmath2(1,1 ) to reveal regions of csdepletion and high opacity , as well as evolved star - forming regions with a high ratio of csto nh@xmath2 .
all siomasers are new detections , and appear to be associated with evolved stars from the _ spitzer _ galactic legacy infrared mid - plane survey extraordinaire ( glimpse ) . generally , within sioregions of multiple vibrational modes ,
the intensity decreases as @xmath6 , but there are a few exceptions where @xmath7 is stronger than @xmath8 .
[ firstpage ] masers surveys stars : formation ism : molecules radio lines : ism galaxy : structure |
spectroscopic observations towards quasars show a large number of intervening absorption systems .
this ` forest ' of lines is numerically dominated by systems showing only the lyman alpha transition these absorbers are called lyman alpha clouds .
earlier work suggests that the clouds are large , highly ionized structures , either pressure confined ( eg . ostriker & ikeuchi 1983 ) or within cold dark matter structures ( eg .
escud & rees 1993 ; cen , miralda escud & ostriker 1994 ; petitjean & mcket 1995 ; zhang , anninos & norman 1995 ) .
however , alternative models do exist : cold , pressure confined clouds ( eg . barcons & fabian 1987 , but see rauch et al .
1993 ) ; various shock mechanisms ( vishniac & bust 1987 , hogan 1987 ) .
low and medium resolution spectroscopic studies of the forest generally measure the redshift and equivalent width of each cloud . at higher resolutions it is possible to measure the redshift ( @xmath7 ) , h i column density ( @xmath8 , atoms per @xmath9 ) and doppler parameter ( @xmath10 , ) .
these are obtained by fitting a voigt profile to the data ( rybicki & lightman 1979 ) . using a list of @xmath8 , @xmath7 and @xmath10 measurements , and their associated error estimates ,
the number density of the population can be studied .
most work has assumed that the density is a separable function of @xmath7 , @xmath8 and @xmath10 ( rauch et al .
1993 ) .
there is a local decrease in cloud numbers near the background quasar which is normally attributed to the additional ionizing flux in that region . while this may not be the only reason for the depletion ( the environment near quasars may be different in other respects ; the cloud redshifts may reflect systematic motions ) it is expected for the standard physical models wherever the ionising flux from the quasar exceeds , or is comparable to , the background
. since the generally accepted cloud models are both optically thin to ionizing radiation and highly ionized , it is possible to correct column densities from the observed values to those that would be seen if the quasar were more remote .
the simplest correction assumes that the shape of the two incident spectra quasar and background are similar . in this case
the column density of the neutral fraction is inversely proportional to the incident ionizing flux .
if the flux from the quasar is known , and the depletion of clouds is measured from observations , the background flux can be determined . by observing absorption towards quasars at different redshifts
the evolution of the flux can be measured .
bechtold ( 1993 ) summarises earlier measurements of the ionising flux , both locally and at higher redshifts .
recently loeb & eisenstein ( 1995 ) have suggested that enhanced clustering near quasars causes this approach to overestimate the background flux .
if this is the case then an analysis which can also study the evolution of the effect gives important information . in particular , a decrease in the inferred flux might be expected after the redshift where the quasar population appears to decrease . however , if the postulated clustering enhancement is related to the turn on of quasars at high redshift , it may conspire to mask any change in the ionizing background .
section [ sec : model ] describes the model of the population density in more detail , including the corrections to flux and redshift that are necessary for a reliable result .
the data used are described in section [ sec : data ] . in section [ sec : errors ] the quality of the fit is assessed and the procedure used to calculate errors in the derived parameters is explained .
results are given in section [ sec : results ] and their implications discussed in section [ sec : discuss ] .
section [ sec : conc ] concludes the paper .
the doppler parameter distribution is not included in the model since it is not needed to determine the ionizing background from the proximity effect .
the model here assumes that @xmath8 and @xmath7 are uncorrelated .
while this is unlikely ( carswell 1995 ) , it should be a good approximation over the restricted range of column densities considered here .
the model of the population without doppler parameters or the correction for the proximity effect is @xmath11 where @xmath12 is the hubble parameter , @xmath13 is the cosmological deceleration parameter and @xmath14 is the speed of light .
correcting for the ionizing flux and changing from ` original ' ( @xmath15 ) to ` observed ' ( @xmath8 ) column densities , gives @xmath16 where @xmath17 @xmath18 and @xmath19 is the background flux , @xmath20 is the flux from the quasar ( @xmath21 ) .
the background flux @xmath22 may vary with redshift . here
it is parameterised as a constant , a power law , or two power laws with a break which is fixed at @xmath23 ( the mid point of the available data range ) .
an attempt was made to fit models with @xmath24 as a free parameter , but the models were too poorly constrained by the data to be useful .
@xmath25 a large amount of information ( figure [ fig : nz ] ) is used to constrain the model parameters .
the high resolution line lists give the column density and redshift , with associated errors , for each line . to calculate the background ionising flux the quasar luminosity and redshift must be known ( table [ tab : objects ] ) .
finally , each set of lines must have observational completeness limits ( table [ tab : compl ] ) .
malmquist bias is a common problem when fitting models to a population which increases rapidly at some point ( often near an observational limit ) .
errors during the observations scatter lines away from the more populated regions of parameter space and into less populated areas .
line blending occurs when , especially at high redshifts , nearby , overlapping lines can not be individually resolved .
this is a consequence of the natural line width of the clouds and can not be corrected with improved spectrographic resolution .
the end result is that weaker lines are not detected in the resultant ` blend ' .
both these effects mean that the observed population is not identical to the ` underlying ' or ` real ' distribution . to calculate a correction for malmquist bias we need to understand the significance of the error estimate since any correction involves understanding what would happen if the ` same ' error occurs for different column density clouds .
the same physical cloud can not be observed with completely different parameters , but the same combination of all the complex factors which influence the errors might affect a line with different parameters in a predictable way . if this idea of the ` quality ' of an observation could be quantified it would be possible to correct for malmquist bias : rather than the ` underlying ' population , one reflecting the quality of the observation ( ie . including the bias due to observational errors ) could be fitted to the data .
for example , if the ` quality ' of an observation was such that , whatever the actual column density measured , the error in column density was the same , then it would be trivial to convolve the ` underlying ' model with a gaussian of the correct width to arrive at an ` observed ' model
. fitting the latter to the data would give parameters unaffected by malmquist bias .
another example is the case of galaxy magnitudes .
the error in a measured magnitude is a fairly simple function of source brightness , exposure time , etc .
, and so it is possible to correct a flux
limited galaxy survey for malmquist bias .
it may be possible to describe the quality of a spectrum by the signal to noise level in each bin .
from this one could , for a given line , calculate the expected error in the equivalent width .
the error in the equivalent width might translate , depending on whether the absorption line was in the linear or logarithmic portion of the ` curve of growth ' , to a normal error in either @xmath8 or @xmath26 .
but in this idealised case it has been assumed that the spectrum has not been re binned , leaving the errors uncorrelated ; that the effect of overlapping , blended lines is unimportant ; that there is a sudden transition from a linear to logarithmic curve of growth ; that the resulting error is well described by a normal distribution .
none of this is likely to be correct and , in any case , the resulting analysis , with different ` observed ' populations for every line , would be too unwieldy to implement , given current computing facilities . a more pragmatic approach might be possible .
a plot of the distribution of errors with column density ( figure [ fig : ndn ] ) suggests that the errors in @xmath26 are of a similar magnitude for a wide range of lines ( although there is a significant correlation between the two parameters ) .
could the error in @xmath26 be a sufficiently good indicator of the ` quality ' of an observation ?
= 8.5 cm if the number density of the underlying population is @xmath27 then the observed population density for a line with error in @xmath26 of @xmath28 is : @xmath29 for a power law distribution this can be calculated analytically and gives an increased probability of seeing lines with larger errors , as expected . for an underlying population density
@xmath30 the increase is @xmath31 .
this gives a lower statistical weight to lines with larger errors when fitting .
for this case a power law and log normal errors the weighting is not a function of @xmath8 directly , which might imply that any correction would be uniform , with little bias expected for estimated parameters . in practice
this correction does not work .
this is probably because the exponential dependence of the correction on @xmath28 makes it extremely sensitive to the assumptions made in the derivation above .
these assumptions are not correct .
for example , it seems that the correlation between and the associated error is important .
it is possible to do a simple numerical simulation to gauge the magnitude of the effect of malmquist bias .
a population of ten million column densities were selected at random from a power law distribution ( @xmath32 ) as an ` unbiased ' sample .
each line was given an ` observed ' column density by adding a random error distributed with a normal or log normal ( for lines where @xmath33 ) distribution , with a mean of zero and a standard deviation in @xmath26 of @xmath34 .
this procedure is a simple approximation to the type of errors coming from the curve of growth analysis discussed above , assuming that errors are approximately constant in @xmath26 ( figure [ fig : ndn ] ) .
the size of the error is larger than typical , so any inferred change in @xmath35 should be an upper limit .
since a power law distribution diverges as @xmath36 a normal distribution of errors in @xmath8 would give an infinite number of observed lines at every column density .
this is clearly unphysical ( presumably the errors are not as extended as in a normal distribution and the population has some low column density limit ) .
because of this the ` normal ' errors above were actually constrained to lie within 3 standard deviations of zero .
= 8.5 cm the results ( figure [ fig : malm ] ) show that malmquist bias has only a small effect , at least for the model used here .
the main solid line is the original sample , the dashed line is the observed population . note that the results in this paper come from fitting to a sample of lines with @xmath37 ( section [ sec : data ] ) corresponding to the data shown expanded to the upper right of the figure .
lines with smaller column densities are not shown since that fraction of the population is affected by the lower density cut off in the synthetic data .
a comparison with the two reference lines , showing the slopes for a population with @xmath38 or @xmath39 ( dashed ) , indicates that the expected change in @xmath35 is @xmath40 .
the population of lines within the logarithmic region of the curve of growth appears to be slightly enhanced , but otherwise the two curves are remarkably similar .
the variations at large column densities are due to the small number of strong lines in the sample .
what other approaches can be used to measure or correct the effects of malmquist bias and line blending ?
press & rybicki ( 1993 ) used a completely different analysis of the lyman@xmath0 forest .
generating and reducing synthetic data , with a known background and cloud population , would allow us to assess the effect of blending .
changing ( sub setting ) the sample of lines that is analysed will alter the relative ( and , possibly , absolute ) importance of the two effects . the procedure used by press & rybicki ( 1993 )
is not affected by malmquist bias or line blending , but it is difficult to adapt to measure the ionizing background .
profile fitting to high resolution data is a slow process , involving significant manual intervention ( we have tried to automate profile fitting with little success ) .
an accurate measurement of the systematic error in the ionizing background would need an order of magnitude more data than is used here to get sufficiently low error limits . even if this is possible
the analysis would need a prohibitive amount of cpu time it would be sufficient work for a separate , major paper ( we would be glad to supply our software to anyone willing to try this ) . taking a sub set of the data is not helpful unless it is less likely to be affected by the biases described above .
one approach might be to reject points with large errors , or large relative errors , in column density since these are more affected by malmquist bias .
however , this would make the observations incomplete in a very poorly understood way .
for example , relative errors are correlated with column density ( as noted above ) and so rejecting lines with larger relative errors would preferentially reject higher column density lines .
there is no sense in trying to measure one bias if doing so introduces others .
unlike rejecting lines throughout the sample , changing the completeness limit does not alter the coverage of the observations ( or rather , it does so in a way that is understood and corrected for within the analysis ) .
raising the completeness limits should make line blending less important since weaker lines , which are most likely to be blended , are excluded from the fit . whether it affects the malmquist bias depends on the distribution of errors . for blended lines , which tend to be weak , raising the completeness limit should increase the absolute value of @xmath35 since the more populous region of the ( hypothetical ) power law population of column densities
will no longer be artificially depleted . the effect on @xmath41
is more difficult to assess since it is uncertain whether the completeness limits are correct at each redshift .
if the limits increase too rapidly with redshift , for example , then raising them further will reduce blending most at lower redshifts , lowering @xmath41 .
but if they are increasing too slowly then the converse will be true .
until either profile fitting is automated , or the method of press & rybicki ( 1993 ) can be modified to include the proximity effect , these two sources of uncertainty malmquist bias and line blending will continue be a problem for any analysis of the lyman@xmath0 forest .
however , from the arguments above , it is likely that the effect of malmquist bias is small and that , by increasing the completeness limit , we can assess the magnitude of the effect of line blending .
extinction within our galaxy reduces the apparent luminosity of the quasars and so lowers the estimate of the background .
since the absorption varies with frequency this also alters the observed spectral slope .
observed fluxes were corrected using extinction estimates derived from the h i measurements of heiles & cleary ( 1979 ) for q2204573 and stark et al .
( 1992 ) for all the other objects .
h i column densities were converted to @xmath42 using the relationships : @xmath43 where the first value comes from diplas & savage ( 1994 ) and the second value , which compensates for the presence of h@xmath44 , is the first scaled by the ratio of the conversions given in bohlin , savage & drake ( 1978 ) .
a ratio @xmath45 of 3.0 ( lockman & savage 1995 ) was used and variations of extinction with frequency , @xmath46 were taken from cardelli , clayton & mathis ( 1989 ) . the correction to the observed index , @xmath47 , of the power law continuum , @xmath48 was calculated using @xmath49 which , using the notation of cardelli , clayton & mathis ( 1989 ) , becomes @xmath50 two quasars are known to have damped absorption systems along the line of sight ( wolfe et al .
the extinction due to these systems is not certain , but model * e * includes the corrections listed in table [ tab : damp ] .
these have been calculated using the smc extinction curve in pei ( 1992 ) , with a correction for the evolution of heavy element abundances taken from pei & fall ( 1995 ) .
the smc extinction curve is most suitable for these systems since they do not appear to have structure at 2220 ( boiss & bergeron 1988 ) , unlike lmc and galactic curves . .
the damped absorption systems and associated corrections ( at 1450 in the quasar s rest
frame ) for model * e*. [ cols= " < , < , < , < , < " , ]
parameter values for the different models are given in table [ tab : fpars ] .
they are generally consistent with other estimates ( lu , wolfe & turnshek 1991 ; rauch et al .
including the proximity effect increases @xmath41 by @xmath51 . although not statistically significant , the change is in the sense expected , since local depletions at the higher redshift end of each data set are removed .
figure [ fig : var_d ] shows the change in population parameters as the completeness limits for the observations are increased .
the most likely values ( curve peaks ) of both @xmath35 and @xmath41 increase as weaker lines are excluded , although @xmath41 decreases again for the last sample .
the value of @xmath35 found here ( @xmath52 ) is significantly different from that found by press & rybicki ( 1993 ) ( @xmath53 ) using a different technique ( which is insensitive to malmquist bias and line blending ) . the value of @xmath35 moves still further away as the column density completeness limits are increased .
this is not consistent with malmquist bias , which would give a smaller change in @xmath35 ( section [ sec : malm ] ) , but could be a result of either line blending or a population in which @xmath35 increases with column density . the latter explanation is also consistent with cristiani et al .
( 1995 ) who found a break in the column density distribution with @xmath35 = 1.10 for log(@xmath8)@xmath54 , and @xmath35 = 1.80 above this value .
later work ( giallongo et al . 1996 , see section [ sec : prevhi ] ) confirmed this .
recent work by hu et al .
( 1995 ) , however , using data with better signal to noise and resolution , finds that the distribution of column densities is described by a single power law ( @xmath55 ) until @xmath56 , when line blending in in their sample becomes significant .
it might be possible that their sample is not sufficiently large ( 66 lines with @xmath57 , compared with 192 here ) to detect a steeper distribution of high column density lines .
the change in @xmath41 as completeness limits are raised may reflect the decrease in line blending at higher column densities .
this suggests that the value here is an over estimate , although the shift is within the 95% confidence interval .
no estimate is significantly different from the value of 2.46 found by press & rybicki ( 1993 ) ( again , using a method less susceptible to blending problems ) .
the likelihood ratio statistic ( equivalent to the ` f test ' ) , comparing model * a * with any other , indicates that the null hypothesis ( that the proximity effect , as described by the model here , should be disregarded ) can be rejected with a confidence exceeding 99.9% .
note that this confirmation is based on the likelihood values in table [ tab : fpars ] .
this test is much more powerful than the k
s test ( section [ sec : finalq ] ) which was only used to see whether the models were sufficiently good for the likelihood ratio test to be used . to reiterate : if model * a * and model * b * are taken as competing descriptions of the population of lyman@xmath0 clouds , then the likelihood ratio test , which allows for the extra degree of freedom introduced , strongly favours a description which includes the proximity effect .
the model without the proximity effect is firmly rejected .
this does not imply that the interpretation of the effect ( ie .
additional ionization by background radiation ) is correct , but it does indicate that the proximity effect , in the restricted , statistical sense above , is ` real ' ( cf .
rser 1995 ) . if the assumptions behind this analysis are correct , in particular that the proximity effect is due to the additional ionising flux from the quasar , then the average value of the background is 100@xmath2 j@xmath3 ( model * b * ) .
if a more flexible model for the background ( two power laws ) is used the flux is consistent with a value of 120@xmath58 j@xmath3(model * d * at @xmath59 ) .
five sources of systematic error are discussed here : malmquist bias and line blending ; reddening by damped absorption systems ; increased clustering of clouds near quasars ; the effect of gravitational lensing . the constraints on the background given here may be affected by malmquist bias and line blending ( sections [ sec : malm ] and [ sec : dispop ] ) .
the effects of line blending will be discussed further in section [ sec : gerry ] , where a comparison with a different procedure suggests that it may cause us to over estimate the flux ( by perhaps @xmath60 dex ) .
malmquist bias is more likely to affect parameters sensitive to absolute column densities than those which rely only on relative changes in the observed population .
so while this may have an effect on @xmath35 , it should have much less influence on the inferred background value .
attenuation by intervening damped absorption systems will lower the apparent quasar flux and so give an estimate for the background which is too low .
this is corrected in model * e * , which includes adjustments for the known damped systems ( section [ sec : dampcor ] , table [ tab : damp ] ) .
the change in the inferred background flux is insignificant ( figure [ fig : evoln ] , table [ tab : fpars ] ) , implying that the magnitude of the bias is less than 0.1 dex .
if quasars lie in regions of increased absorption line clustering ( loeb & eisenstein 1995 ; yurchenko , lanzetta & webb 1995 ) then the background flux may be overestimated by up to 0.5 , or even 1 , dex .
gravitational lensing may change the apparent brightness of a quasar in general the change can make the quasar appear either brighter or fainter .
absorption line observations are made towards the brightest quasars known ( to get good quality spectra ) .
since there are more faint quasars than bright ones this will preferentially select objects which have been brightened by lensing ( see the comments on malmquist bias in section [ sec : malm ] ) .
an artificially high estimate of the quasar flux will cause us to over estimate the background .
unfortunately , models which assess the magnitude of the increase in quasar brightness are very sensitive to the model population of lensing objects . from pei ( 1995 )
an upper limit consistent with observations is an increase in flux of about 0.5 magnitudes , corresponding to a background estimate which is too high by a factor of 1.6 ( 0.2 dex ) . the probable effect
, however , could be much smaller ( blandford & narayan 1992 ) .
if bright quasars are more likely to be lensed we can make a rudimentary measurement of the effect by splitting the data into two separate samples . when fitted with a constant background ( model * b * )
the result for the five brightest objects is indeed brighter than that for the remaining six , by 0.1 dex .
the errors , however , are larger ( 0.3 dex ) , making it impossible to draw any useful conclusions .
the effects of malmquist bias , line blending and damped absorption systems are unlikely to change the results here significantly .
cloud clustering and gravitational lensing could be more important in each case the background would be over estimated .
the magnitude of these last two biases is not certain , but cloud clustering seems more likely to be significant .
more complex models allow the background flux to vary with redshift .
if the flux does evolve then these models should fit the data better .
however , there is no significant change in the fit when comparing the likelihood of models * c * to * e * with that of * b*. nor are @xmath61 or @xmath62 significantly different from zero .
so there is no significant evidence for a background which changes with redshift .
the asymmetries in the wings of the posterior distributions of @xmath61 or @xmath62 for model * d * ( figure [ fig : alphas_d ] ) are a result of the weak constraints on upper limits ( see next section ) .
the box plots in figure [ fig : flux_both ] illustrate the range of evolutions that are possible .
while there is little evidence here for evolution of the background , the upper limits to the background flux diverge more strongly than the lower limits at the lowest and highest redshifts .
also , the posterior probability of the background is extended more towards higher values .
the background was measured by comparing its effect with that of the quasar .
if the background were larger the quasar would have less effect and the clouds with @xmath63 would not need as large a correction to the observed column density for them to agree with the population as a whole .
if the background was less strong then the quasars would have a stronger influence and more clouds would be affected .
the upper limit to the flux depends on clouds influenced by the quasar .
figure [ fig : prox ] shows how @xmath64 changes with redshift and proximity to the background quasar . from this figure
it is clear that the upper limit is dominated by only a few clouds .
however , the lower limit also depends on clouds near to , but not influenced by , the quasar .
this involves many more clouds .
the lower limit is therefore stronger , more uniform , and less sensitive to the amount of data , than the upper limit .
other procedures for calculating the errors in the flux have assumed that the error is symmetrical ( the only apparent exception is fernndez
soto et al .
( 1995 ) who unfortunately had insufficient data to normalize the distribution ) .
while this is acceptable for @xmath35 and @xmath41 , whose posterior probability distributions ( figure [ fig : beta_gamma_d ] ) can be well approximated by gaussian curves , it is clearly wrong for the background ( eg . figure [ fig : flux_both ] ) , especially where there are less data ( at the lowest and highest redshifts ) . an estimate based on the assumption that the error is normally distributed will be biased in two ways .
first , since the extended upper bound to the background has been ignored , it will underestimate the ` best ' value .
second , since the error bounds are calculated from the curvature of the posterior distribution at its peak ( ie . from the hessian matrix ) they will not take account of the extended ` tails ' and so will underestimate the true range of values .
in addition , most earlier work has calculated errors assuming that the other population parameters are fixed at their best fit values .
this will also under estimate the true error limits .
all these biases become more significant as the amount of data decreases .
the first of these biases also makes the interpretation of the box
plots ( eg . figures [ fig : flux_both ] and [ fig : evoln ] ) more difficult .
for example , the curves in the left hand plot in figure [ fig : flux_both ] and the data in table [ tab : modeld ] show that the value of the flux with highest probability at @xmath65 is @xmath66 j@xmath3 ( for model * d * ) .
in contrast the box
plot on the right shows that the median probability is almost twice as large ( @xmath67 j@xmath3 ) .
neither plot is ` wrong ' : this is the consequence of asymmetric error distributions .
= 15.cm fernndez
soto et al .
( 1995 ) fitted high signal to noise data towards three quasars . for @xmath68
they estimate an ionizing background intensity of 32 j@xmath3 , with an absolute lower limit ( 95% confidence ) of 16 j@xmath3 ( figure [ fig : evoln ] , the leftmost cross ) .
they were unable to put any upper limit on their results .
cristiani et al .
( 1995 ) determined a value of 50 j@xmath3 using a sample of five quasars with a lower column density cut off of log(@xmath8 ) = 13.3 .
this sample was recently extended ( giallongo 1995 ) .
they find that the ionizing background is roughly constant over the range @xmath69 with a value of 50 j@xmath3 which they considered a possible lower limit ( figure [ fig : evoln ] , the middle lower limit ) . while this paper was being refereed giallongo et al .
( 1996 ) became available , extending the work above .
using a maximum likelihood analysis with an unspecified procedure for calculating errors they give an estimate for the background of @xmath70 j@xmath3 .
they found no evidence for evolution with redshift when using a single power law exponent .
williger et al .
( 1994 ) used a single object , q1033033 , which is included in this sample , to give an estimate of @xmath71 j@xmath3(figure [ fig : evoln ] , the rightmost cross ) .
the error limits are smaller than those found here , even though they only use a subset of this data , which suggests that they have been significantly underestimated . if the errors in williger et al .
( 1994 ) are indeed underestimates then these measurements are consistent with the results here .
however , the best fit values are all lower than those found here .
this may be , at least partly , because of the biases discussed in section [ sec : lims ] .
williger et al .
( 1994 ) used a more direct method than usual to estimate the background .
this gives a useful constraint on the effect of line blending in the procedures used , and is explored in more detail below .
the measured value of the background , 80@xmath72 j@xmath3 ( model * d * at @xmath73 ) , is larger than an earlier estimate using a subset of this data ( williger et al .
1994 , q1033033 , @xmath71 j@xmath3 ) .
as has already been argued , it is difficult to understand how a procedure using much less data could have smaller error limits than the results here , so it is likely that the error was an underestimate and that the two results are consistent .
however , it is interesting to see if there is also a systematic bias in the analyses used .
the correction for galactic absorption is not very large for this object ( about 20% ) .
more importantly , the procedures used differ significantly in how they are affected by blended lines .
these are a problem at the highest redshifts , where the increased lyman@xmath0 cloud population density means that it is not always possible to resolve individual clouds .
williger et al .
( 1994 ) added additional lines ( @xmath74 , @xmath10 = 30 ) to their @xmath7 = 4.26 spectra of q1033033 and found that between 40% and 75% would be missed in the line list .
as the lower column density limit is raised williger et al .
( 1994 ) find that the observed value of @xmath41 also increases .
the resulting stronger redshift evolution would make the deficit of clouds near the quasar more significant and so give a lower estimate of the background .
although not significant at the 95% level , there is an indication that @xmath41 also increases with higher column density in this analysis ( section [ sec : dispop ] , figure [ fig : var_d ] ) . while it is possible that @xmath41 varies with column density the same dependence
would be expected if line blending is reducing the number of smaller clouds . to understand how line blending can affect the estimates
, we will now examine the two analyses in more detail .
line blending makes the detection of lines less likely . near
the quasar lines are easier to detect because the forest is more sparse . in the analysis used in this paper the appearance of these ` extra ' lines
reduces the apparent effect of the quasar .
alternatively , one can say that away from the quasar line blending lowers @xmath41 .
both arguments describe the same process and imply that the estimated background flux is too large .
in contrast , williger et al .
( 1994 ) take a line list from a crowded region , which has too few weak lines and correspondingly more saturated lines , and reduce the column densities until they agree with a region closer to the quasar .
since a few saturated lines are less sensitive to the quasar s flux than a larger number of weaker lines , the effect of this flux is over estimated ( and poorly determined ) , making the background seem less significant and giving a final value for the background flux which is too small .
this method is therefore biased in the opposite sense to ours and so the true value of the background probably lies between their estimate and ours . the comparison with williger et al .
( 1994 ) gives one estimate of the bias from line blending .
another can be made by raising the completeness limits of the data ( section [ sec : malm ] ) .
this should decrease the number of weak , blended lines , but also excludes approximately half the data . in figure
[ fig : flux_both ] the flux estimates from the full data set are shown together with those from one in which the limits have been raised by @xmath75 .
there is little change in the lowest reasonable flux , an increase in the upper limits , and in increase in the ` best fit ' values .
the flux for @xmath76 is almost unconstrained by the restricted sample ( section [ sec : lims ] explains the asymmetry ) .
an increase of 0.5 in @xmath26 is a substantial change in the completeness limits . that the lower limits remain constant ( to within @xmath77 dex )
suggests that line blending is not causing the flux to be significantly over estimated .
the increase in the upper limits is expected when the number of clouds in the sample decreases ( section [ sec : lims ] ) . in summary , the total difference between our measurement and that in williger et al .
( 1994 ) is 0.7 dex which can be taken as an upper limit on the effect of line blending .
however , a more typical value , from the constancy of the lower limits when completeness limits are raised , is probably @xmath78 dex .
bechtold ( 1994 ) analysed lower resolution spectra towards 34 quasars using equivalent widths rather than individual column density measurements .
she derived a background flux of 300 j@xmath3(@xmath79 ) , decreasing to 100 j@xmath3 when a uniform correction was applied to correct for non systemic quasar redshifts . with low resolution data a value of @xmath35
is used to change from a distribution of equivalent widths to column densities . if @xmath35 is decreased from 1.7 to a value closer to that found for narrower lines ( see section [ sec : dispop ] ) then the inferred background estimate could decrease further .
the evolution was not well constrained ( @xmath80 ) .
no distinction was made between the lower and upper constraints on the flux estimate , and it is likely that the wide range of values reflects the lack of strong upper constraints which we see in our analysis .
it is not clear to what extent this analysis is affected by line blending .
certainly the comments above that relatively more clouds will be detected near the quasar also apply .
the background intensity presented in this paper is much larger than the 8 j@xmath3 upper limit at @xmath81 found by vogel et al .
kulkarni & fall ( 1993 ) obtain an even lower value of 0.6@xmath82 j@xmath3 at @xmath83 by analysing the proximity effect in hst observations .
however , even an unevolving flux will decrease by a factor of @xmath84 between @xmath65 and @xmath85 , so such a decline is not inconsistent with the results given here .
quasars are likely to provide a significant , if not dominant , contribution to the extragalactic background .
an estimate of the ionizing background can be calculated from models of the quasar population .
figure [ fig : evoln ] shows the constraints from models * d * and * e * and compares them with the expected evolution of the background calculated by fall & pei ( 1995 ) .
the background can take a range of values ( the shaded region ) , with the lower boundary indicating the expected trend for a dust free universe and larger values taking into account those quasars that may be hidden from our view , but which still contribute to the intergalactic ionizing flux .
the hypothesis that the flux is only from visible quasars ( the unobscured model in fall & pei 1995 ) is formally rejected at over the 95% significance level since the predicted evolution is outside the 95% bar in the box plots at higher redshift .
although our background estimate excludes a simple quasar dominated model based on the observed number of such objects , the analysis here may give a background flux which is biased ( too large ) from a combination of line blending ( section [ sec : gerry ] ) and clustering around the background quasars . from the comparison with williger et al .
( 1994 ) , above , there is an upper limit on the correction for line blending , at the higher redshifts , of 0.7 dex .
however , an analysis of the data when column density completeness limits were increased by @xmath75 suggests that a change in the lower limits here of @xmath77 dex is more likely . a further change of up to between 0.5 and 1 dex is possible if quasars lie in regions of increased clustering ( section [ sec : syserr ] ) .
these two effects imply that at the highest redshifts the flux measured here could reasonably overestimate the real value by @xmath86 dex .
this could make the measurements marginally consistent with the expected flux from the observed population of quasars .
there is also some uncertainty in the expected background from quasars since observations could be incomplete even at the better understood lower redshifts ( eg .
goldschmidt et al . 1992 ) and while absorption in damped systems is understood in theory ( fall & pei 1993 ) its effect is uncertain ( particularly because the distribution of high column density systems is poorly constrained ) .
the highest flux model ( largest population of obscured quasars ) from fall & pei ( 1995 ) is consistent with the measurements here ( assuming that the objects used in this paper are not significantly obscured ) . the background appears to be stronger than the integrated flux from the known quasar population .
can star formation at high redshifts explain the discrepancy ?
recent results from observations of low redshift starbursts ( leitherer et al .
1995 ) suggest that very few ionizing photons ( @xmath87% ) escape from these systems
. if high redshift starbursts are similar in their properties , then the presence of cool gas in these objects would similarly limit their contribution to the ionizing background .
however , madau & shull ( 1996 ) estimate that if star formation occurs in lyman@xmath0 clouds , and a significant fraction of the ionizing photons ( @xmath88 ) escape , then these photons may contribute a substantial fraction of the ionizing background photons in their immediate vicinity . as an example , at @xmath89 they estimate that @xmath90 if star formation sets in at @xmath91 .
this flux would dominate the lowest ( no correction for obscuration ) quasar background shown in figure [ fig : evoln ] and could be consistent with the intensity we estimate for the background at this redshift , given the possible systematic biases discussed above and in section [ sec : syserr ] .
a model has been fitted to the population of lyman@xmath0 clouds .
the model includes the relative effect of the ionising flux from the background and nearby quasars ( section [ sec : model ] ) .
the derived model parameters for the population of absorbers are generally consistent with earlier estimates .
there is some evidence that @xmath35 , the column density power law population exponent , increases with column density , but could also be due to line blending ( section [ sec : dispop ] ) .
the ionising background is estimated to be 100@xmath2 j@xmath3(model * b * , section [ sec : resbackg ] ) over the range of redshifts ( @xmath92 ) covered by the data .
no strong evidence for evolution in the ionizing background is seen over this redshift range . in particular , there is no significant evidence for a decline for @xmath93 ( section [ sec : noevoln ] ) .
previous results may have been biased ( too low , with optimistic error limits section [ sec : lims ] ) .
constraints on the evolution of the background are shown in figure [ fig : evoln ] .
the estimates are not consistent with the background flux expected from the observed population of quasars ( section [ sec : source ] ) . however , two effects are likely to be important .
first , both line blending and increased clustering of clouds near quasars lead to the measured background being overestimated .
second , a significant fraction of the quasar population at high redshifts may be obscured . since their contribution to the background
would then be underestimated this would imply that current models of the ionizing background are too low .
both of these would bring the expected and measured fluxes into closer agreement .
it is also possible that gravitational lensing makes the measurement here an overestimate of the true background .
the dominant source of errors in our work is the limited number of lines near the background quasar ( eg . figures [ fig : nu1_d ] and [ fig : prox ] ) .
systematic errors are smaller and become important only if it is necessary to make standard ( unobscured quasar ) models for the background consistent with the lower limits presented here .
further data will therefore make the estimate here more accurate , although observational data are limited by confusion of the most numerous lower column density systems ( @xmath94 ) so it will remain difficult to remove the bias from line blending .
an improvement in the errors for the highest redshift data points , or a determination of the shape of the ionizing spectrum ( e.g. from he ii / h i estimates in lyman@xmath0 clouds ) would help in discriminating between current competing models for the ionizing background .
finally , a determination of the background strength in the redshift range @xmath95 is still needed .
we would like to thank yichuan pei for stimulating discussions and for making data available to us .
tom leonard ( dept . of statistics , edinburgh ) gave useful comments and guidance on the statistics used in this paper .
we would also like to thank an anonymous referee for helpful and constructive comments . | the decrease in number density of lyman@xmath0 clouds near the background quasar is an observational result which is often called the ` proximity ' or ` inverse ' effect .
it is thought that , for nearby clouds , the quasar s flux dominates the background radiation field , increasing the ionization state of the clouds and reducing the ( observed ) h i column density . in this paper
we analyse a sample of 11 quasars from the literature for which accurate column density estimates of the lyman@xmath0 lines exist .
we confirm , to a significance level of more than 3 standard deviations , that the proximity effect exists . if it is related to the background flux then the intensity and evolution of the background have been constrained .
using a maximum likelihood method , we determine the strength of the extragalactic ionizing background for @xmath1 , taking account of possible systematic errors in our determination and estimating the effect of biases inherent in the data .
if the background is constant we find that it has an intensity of 100@xmath2 j@xmath3 , where j@xmath3 is defined as 10@xmath4 ergs @xmath5 hz@xmath6 sr@xmath6 .
there is no significant evidence for a change in this value with redshift . _
_ cosmology : diffuse radiation quasars : absorption lines galaxies : evolution |
the solar neutrino problem @xcite may allow for new break through in our understanding of the standard model @xcite .
the precise knowledge of the astrophysical @xmath1-factor of the @xmath2be(p,@xmath3)@xmath4b reaction at the gamow peak , about 20 kev , is of crucial importance for interpreting terrestrial measurements of the solar neutrino flux @xcite .
this is particularly true for the interpretation of results from the homestake , kamiokande , superkamiokande and sno experiments @xcite which measured high energy solar neutrinos mainly or solely from @xmath4b decay .
eight direct measurements were reported for the @xmath2be(p,@xmath3)@xmath4b reaction @xcite .
since the cross section at @xmath5 20 kev is too small to be measured , the @xmath1-factors at low energies have to be extrapolated from the experimental data with the help of theoretical models ( see e.g. johnson _ et al . _
@xcite , jennings _ et al . _
@xcite ) . but also in the energy range above @xmath5 100 kev , where measurements can be performed , experimental difficulties , mainly connected with the determination of the effective target thickness of the radioactive @xmath2be target , hamper the derivation of accurate results .
these difficulties are reflected in the fact that the six precise measurements @xcite can be grouped into two distinct data sets which agree in their energy dependence but disagree in their absolute normalization by about 30% . since
this discrepancy is larger than the error of adopted @xmath1 in the standard solar model , experimental studies with different methods are highly desirable for improving the reliability of the input to the standard solar model @xcite . in fig .
1 we show the world data including our new gsi measurement of @xmath1 @xcite .
it is clear from this figure that if we are to quote @xmath6 with an accuracy of @xmath7 , many more experiments using different methods are required .
world data on @xmath1 as reported by our gsi collabortion @xcite .
the discrepancy in the measured absolute value of the cross section of the 7pg reaction is quite possibly best addressed with a @xmath8 radioactive beam and a hydrogen target , allowing for a direct measurement of the beam - target luminosity as we propose to do at lln .
however , additional uncertainty exists in the theoretical extrapolation of the measured cross section to solar energies ( approximately 20 kev ) .
a few theoretical studies suggest an extrapolation procedure that is accurate to approximately @xmath9 @xcite . without discussing these rather strong statements we consider a similar situation that haunted nuclear astrophysics a few years back the s - factor of the @xmath10 reaction .
it was assumed that in this case d - waves dominate and no nuclear structure effects should play a role at very low energy , as low as 100 kev .
much in the same way , it is stated today that s - waves dominate the 7pg reaction and we do not expect nuclear structure effects to play a role at low energies in the 7pg reaction . in fig .
2 we show fowler s extrapolated d - wave s - factor that is a mere factor of 32 smaller than measured , due to a small non d - wave component in the d + d interaction @xcite . a small nuclear structure effect , namely the d - wave component of the ground state of @xmath11 , gives rise to a change by a factor of 32 in the predicted astrophysical s - factor .
similarly we may ask whether a small non s - wave component in the low energy interaction of p + 7 could alter the extrapolated @xmath6 value by considerably more than one percent .
a measurement of @xmath6 with an accuracy of @xmath7 mandates that the cross section be measured at low energies , as low as possible , so as to also test the extrapolation procedures .
extrapolation of d - wave s - factor of the @xmath10 reaction@xcite .
note the presence of small non d - wave components that yield a discrepancy from fowler s extracted s - factor by a factor of 32 .
the coulomb dissociation ( cd ) @xcite is a primakoff @xcite process that could be viewed in first order as the time reverse of the radiative capture reaction . in this case instead of studying for example the fusion of a proton plus a nucleus ( a-1 ) , one studies the disintegration of the final nucleus ( a ) in the coulomb field , to a proton plus the ( a-1 ) nucleus .
the reaction is made possible by the absorption of a virtual photon from the field of a high z nucleus such as @xmath13 . in this case since @xmath14 for a photon is approximately 1000 times larger than that of a particle beam , the small cross section is enhanced . the large virtual photon flux ( typically 100 - 1000 photons per collision )
also gives rise to enhancement of the cross section .
our understanding of the coulomb dissociation process @xcite allows us to extract the inverse nuclear process even when it is very small .
however in coulomb dissociation since @xmath15 approaches unity ( unlike the case in electron scattering ) , higher order coulomb effects ( coulomb post acceleration ) may be non - negligible and they need to be understood @xcite .
the success of cd experiments @xcite is in fact contingent on understanding such effects and designing the kinematical conditions so as to minimize such effects .
hence the coulomb dissociation process has to be measured with great care with kinematical conditions carefully adjusted so as to minimize nuclear interactions ( i.e. distance of closest approach considerably larger then 20 fm , or very small forward angles scattering ) , and measurements must be carried out at high enough energies ( many tens of mev / u ) so as to maximize the virtual photon flux .
the rikenii measurement @xcite of detailed angular distributions for the coulomb dissociation of 8 allowed us to extract the e2 amplitude in the cd of 8 .
the same data also allowed us to extract values for @xmath1 as recently published @xcite .
the @xmath16pb target and @xmath4b beam properties in this experiment were as in ref .
@xcite , but the detector system covered a large angular range up to around 9@xmath17 to be sensitive to the e2 amplitude .
the e1 and e2 virtual photon fluxes were calculated @xcite using quantum mechanical approach . the nuclear amplitude is evaluated based on the collective form factor where the deformation length is taken to be the same as the coulomb one .
this nuclear contribution results in possible uncertainties in the fitted e2 amplitude .
nevertheless , the present results lead to a very small e2 component at low energies , below 1.5 mev , of the order of a few percent , even smaller than the low value predicted by typel and baur @xcite and considerably below the upper limit on the e2 component extracted by gai and bertulani @xcite from the rikeni data .
these low values of the e2 amplitude were however recently challenged by the msu group @xcite that used interference between e1 and e2 components to observe an asymmetry of the longitudinal - momentum distribution of @xmath2be @xcite .
they @xcite claim : @xmath18/@xmath19 at 0.63 mev .
these results are considerably larger than reported by kikuchi _
@xcite , and just below the upper limit reported by gai and bertulani @xcite .
a recent analysis of the rikenii data @xcite by bertulni and gai @xcite confirmed the small e2 extracted by kikuchi et al .
@xcite as well as the negligible nuclear contribution .
note that in this analysis @xcite the acceptance of the rikenii detector is taken into account using the matrix generated by kikuchi _
recently a possible mechanism to reduce the e2 dissociation amplitude was proposed by esbensen and bertsch @xcite .
an experiment to measure the coulomb dissociation of @xmath4b at a higher energy of 254 @xmath20 mev was performed at gsi @xcite .
the present experimental conditions have several advantages : ( i ) forward focusing allows us to use the magnetic spectrometer kaos @xcite at gsi for a kinematically complete measurement with high detection efficiency over a wider range of the p-@xmath2be relative energy ;
( ii ) because of the smaller influence of straggling on the experimental resolution at the higher energy , a thicker target can be used for compensating the weaker beam intensity , ( iii ) effects that obscure the contribution of e1 multipolarity to the coulomb dissociation like e2 admixtures and higher - order contributions are reduced @xcite .
the contribution of m1 multipolarity is expected to be enhanced at the higher energy , but this allows to observe the m1 resonance peak and determine its @xmath3 width .
a @xmath4b beam was produced by fragmentation of a 350 @xmath20 mev @xmath21c beam from the sis synchrotron at gsi that impinged on a beryllium target with a thickness of 8.01 g/@xmath22 .
the beam was isotopically separated by the fragment separator ( frs ) @xcite by using an aluminum degrader with a thickness of 1.46 g/@xmath22 with a wedge angle of 3 mrad .
the beam was transported to the standard target - position of the spectrometer kaos @xcite .
the average beam energy of @xmath4b in front of the breakup target was 254.5 @xmath20 mev , a typical @xmath4b intensity was 10@xmath23 /spill ( 7s / spill ) .
beam - particle identification was achieved event by event with the tof-@xmath24 method by using a beam - line plastic scintillator with a thickness of 5 mm placed 68 m upstream from the target and a large - area scintillator wall discussed later placed close to the focal plane of kaos .
about 20 % of the beam particles were @xmath2be , which could however unambiguously be discriminated from breakup @xmath2be particles by their time of flight . an enriched @xmath16pb target with a thickness of 199.7 ( @xmath25 0.2 ) mg/@xmath22 was placed at the entrance of kaos . the average energy at the center of the target amounted to 254.0 @xmath20 mev .
the reaction products , @xmath2be and proton , were analyzed by the spectrometer which has a large momentum acceptance of @xmath26 % and an angular acceptance of 140 and 280 mrad in horizontal and vertical directions , respectively . for scattering - angle measurement or track reconstruction of the two reaction products ,
two pairs of silicon micro - strip detectors were installed at about 14 and 31 cm downstream from the target , respectively , measuring either x- or y - position of the products before entering the kaos magnets .
each strip detector had a thickness of 300 @xmath27 m , an active area of 56 @xmath28 56 mm@xmath29 , and a strip pitch of 0.1 mm .
the measured complete kinematics of the breakup products allowed us to reconstruct the p-@xmath2be relative energy and the scattering angle @xmath30 of the center - of - mass of proton and @xmath2be ( excited @xmath4b ) with respect to the incoming beam from the individual momenta and angles of the breakup products .
to evaluate the response of the detector system , monte - carlo simulations were performed using the code geant@xcite .
the simulations took into account the measured @xmath4b beam spread in energy , angle , and position at the target , as well as the influence of angular and energy straggling and energy loss in the layers of matter .
losses of the products due to limited detector sizes were also accounted for .
further corrections in the simulation are due to the feeding of the excited state at 429 kev in @xmath2be .
we used the result by kikuchi _
et al . _
@xcite who measured the @xmath3-decay in coincidence with coulomb dissociation of @xmath4b at 51.9 @xmath20 mev .
the monte carlo simulations yielded relative - energy resolutions from the energy and angular resolutions of the detection system to be 0.11 and 0.22 mev ( 1@xmath31 ) at @xmath32 and 1.8 mev , respectively . the total efficiency calculated by the simulation
was found to be larger than 50% at @xmath33 mev . due to the large acceptance of kaos .
( a , b , c ) yields of breakup events plotted against @xmath30 , the scattering angle of the excited @xmath4b , for three relative energy bins .
the full histograms show the results of a simulation taking into account the measured angular spread of the incident @xmath4b beam ( shown in d ) and assuming e1+m1 multipolarity .
the dashed histograms show the simulated results for e2 contribution according to the calculations of bertulani and gai@xcite . in fig .
3 , the experimental yield is plotted against the scattering angle of the excited @xmath4b for three relative - energy bins , [email protected] ( a ) , [email protected] ( b ) , and [email protected] ( c ) mev , together with results of the monte - carlo simulations assuming e1 excitation .
the m1 transition also contributes to the [email protected] mev bin with the same angular dependence .
since we did not measure the incident angle of @xmath4b at the target , the experimental angular - distributions represent the @xmath30 distributions folded with the angular spread of the incident beam shown in fig .
the angular resolution was estimated to be 0.35@xmath35 ( 1@xmath31 ) , smaller than the observed widths . as seen in fig .
3 the experimental distributions are well reproduced by the simulation using only e1 multipolarity , in line with the results of kikuchi _ et al . _
@xcite . to study our sensitivity to a possible e2 contribution
, we have added the simulated e2 angular distribution from the e2 coulomb dissociation cross section calculated by bertulani and gai @xcite .
note that nuclear breakup effects are also included in the calculation . by fitting the experimental angular distributions to the simulated ones
, we obtained 3@xmath31 upper limits of the ratio of the e2- to e1-transition amplitude of the @xmath2be(p,@xmath3)@xmath4b reaction , @xmath18/@xmath36 of @xmath37 , @xmath38 and @xmath39 for @xmath40 [email protected] , [email protected] and [email protected] mev , respectively .
these numbers agree well with the results of kikuchi _ et al . _
@xcite , and suggest that the results of the model dependent analysis of davids _ et al .
_ @xcite needs to be checked . in conclusion we demonstrated that the coulomb dissociation ( when used with `` sechel '' ) provides a viable alternative method for measuring small cross section of interest for nuclear - astrophysics .
first results on the cd of 8 are consistent with the lower measured values of the cross section and weighted average of all thus far data is : @xmath41 , see table 1 .
the accuracy of the extracted s - factors are now limited by our very understanding of the coulomb dissociation process , believed to be approx .
the value of the e2 s - factor as extracted from both the riken and gsi experiments are consistent and shown to be very small , @xmath42 of the order of @xmath43 or smaller , see table 1 . measured s - factors in coulomb dissociation experiments .
= = = + + riken1 @xmath44 @xmath45 + + riken2 @xmath46 @xmath47 + + gsi1 @xmath48 @xmath49 + + msu @xmath50 + + @xmath51 @xmath49 +
we propose to measure the cross section of the 7pg reaction with the use of 7 beams produced at louvain - la - neuve ( lln ) @xcite .
we intend to perform two experiments , both with 7 beams at louvain - la - neuve . the first one is already approved ( ph120 ) with the use of the old cyclotron for three data points measured at 1.0 - 0.5 mev . in a second experiment we propose to use the new cyclotron at lln for additional measurements below 0.5 mev .
it is however clear that a test of the extrapolation procedures , as discussed above , will require yet a third experiment with an implanted 7 and low energy proton beams .
the experimental setup of the uconn - lln experiment ( ph120 ) to measure the 7pg reaction with 7 beams @xcite . the experimental detector setup for the lln experiment is shown in fig .
4 . by measuring the energy of the emerging 7 beam ,
we ensure that the 8 produced in the middle of the target is stopped at the end of the first aluminum catcher foil . due to the finite target thickness ,
the recoil 8 emerge with a ( step ) distribution of energies with widths approximately 0.7 mev , and a stopping spread in aluminum of approximately 0.5 @xmath52 .
thus the stopped 8 are designed to be equally spread over the two aluminum catcher foils ( 0.5 @xmath52 each ) .
the beta - delayed alpha - particle emission of 8 will be measured by measuring coincidence between the two back to back equal energy alpha - particles detected in a pair of detectors , see fig .
4 . the large diameter of the central detector ( more than twice that of front and back detectors ) yield a large coincidence efficiency ( better than 98% ) , as shown by our extensive monte carlo simulations . by requiring
that the count rates in the front and back pair of detectors are similar we will ensure that the 8 are spread equally on both aluminum foils and thus no 8 escapes the catcher foils and none get stopped in the degrader , yielding nearly 100% collection efficiency . in the target region ,
two monitors measure beam intensity by measuring the elastic scattering off a thin au foil ( evaporated onto a very thin carbon backing ) and the recoil protons off the target . we plan to measure @xmath53 for p + 7 with high precision .
thus the cross section of the 7pg reaction will be measured relative to the elastic scattering , thereby removing several systematic uncertainties related to beam - target composition .
the hydrogen component of the target will be continuously monitored by measuring the recoil protons from the target .
these measurements will allow us to continuously measure and monitor the beam - target luminosity .
since two alpha - particles are associated with that decay we expect a very large detection efficiency , approximately 50% of @xmath54 .
our extensive monte carlo simulations yield a large ( 98% ) coincidence efficiency and thus approximately 50% total coincidence efficiency for two equal energy correlated back to back alpha - particles . for a 8 transfer time of 0.07 sec ,
every 0.5 sec , we obtain a total alpha - particle detection efficiency of approximately 25% . the closed detection geometry ( 50% of 4@xmath55 ) with a front and back detectors ( a - la calorimetry style ) also ensures that the total alpha detection efficiency is nearly independent of the exact location of the collection foils , as long as the two foils remain parallel and at constant distance and the recoil 8 nuclei are spread equally on both catcher foils . a beam intensity of @xmath56 and a 250 @xmath57 target ( @xmath58 ) containing @xmath59 yield a luminosity of @xmath60 .
with expected cross sections of @xmath61 and @xmath62 , at @xmath63 and @xmath64 mev , respectively , and alpha - particle detection efficiency of 25% , we obtain count rates of approximately 5 , 4 , and 2 counts per hour . thus experiments lasting two to three days at @xmath63 and @xmath64 mev , respectively , will yield a total count of 240 , 192 and 144 counts and statistical uncertainties of 6.4% , 7.2% and 8.3% , respectively . with approved 9 days of experiment
we plan to adjust the length of runs to achieve 5% precision at each data point .
i would like to acknowledge the work of n. iwasa , t. kikuchi , k. suemmerer , f. boue and p. senger on the data analyses of the cd data and ralph h. france iii and james e. mcdonald on the data analysis of the 7pg reaction .
i also acknowledge discussions and encouragements from professors j.n .
bahcall , c.a .
bertulani , g. baur , and th .
delbar .
m. gai , j.e .
mcdonald , r.h .
france iii , j.s .
schweitzer , c. angulo , ch .
barue , s. cherubini , m. cogneau , th .
delbar , m. gaelens , p. leleux , m. loiselet , a. ninane , g. ryckewaert , k.b .
swartz , d. visser ; bull .
. soc . * 44,ii*(1999)1529 . | the 7pg reaction is one of the major source of uncertainties in estimating the 8 solar neutrino flux and is critical for undertsanding the solar neutrino problem and neutrinos .
the main source of uncertainty is the existence of conflicting data with different absolute normalization .
attempts to measure this reaction rate with 7 beams are under way by the uconn - lln collaboration , and we discuss a newly emerging method to extract this cross section from the coulomb dissociation of the radioactive beam of 8 .
we discuss some of the issues relevant for this study including the question of the e2 contribution to the coulomb dissociation process which was measured to be small .
the coulomb dissociation appears to provide a viable alternative method for measuring the 7pg reaction rate , with a weighted average of the riken1 , riken2 and gsi1 published results of @xmath0 ev - b . |
various observations have revealed that clusters of galaxies are pervaded by the strong magnetic field of microgauss strength ( e.g. , see @xcite ; @xcite for reviews and references therein ) .
the multi - frequency faraday - rotation measurements of polarized radio sources inside or behind a cluster have been used to estimate the magnetic field strength , combined with the x - ray data ( @xcite ; @xcite ) .
recently , clarke , kronberg , & bhringer ( 2001 ) have drawn a firm conclusion that an intracluster hot plasma is universally magnetized by @xmath8 g fields under the assumption of @xmath9kpc field coherent scales , using @xmath10 normal low-@xmath11 _ rosat _ cluster sample selected to be free of unusual strong radio halos , widespread cooling flow and recent merger activity . however , except for a few cases such as some background rotation measures per square degree in the coma cluster ( @xcite ) , it is generally difficult to measure the angular profile of the magnetic field strength inside a cluster because of the lack of the number of radio sources available per cluster .
the observations of cluster - wide diffuse radio halos have also led to the evidence of the microgauss field , which is believed to be synchrotron radiation by relativistic electrons accelerated in the shock wave . moreover , in this case hard x - ray emission could be produced by the same relativistic electron population through the inverse compton scattering of cosmic microwave background ( cmb ) photons .
the combined observation of hard x - ray emission and radio halos is therefore very attractive in the sense that it allows us to directly estimate the magnetic field strength without further restrictive assumptions for the coherent length and the electron distribution .
the recent detections of hard x - ray emission have led to an independent estimation of @xmath12 g fields even in the outer ( @xmath13 ) envelopes of clusters ( @xcite ; @xcite ; @xcite ) .
however , the origin of the intracluster magnetic fields still rests a mystery in cosmology .
the following two scenarios have been usually investigated in the literature .
one is based on the idea that the cluster field is related to the fields generated by dynamo mechanism in individual galaxies and subsequent wind - like activity transports and redistributes the magnetic fields in the intracluster medium ( e.g. @xcite and references therein ) .
the general prediction of galactic dynamo mechanism is that the galactic magnetic field could arise from an exponential amplification of a small _ seed _ field during a galactic lifetime and energy of the mean magnetic field
could grow up to equipartition level with the turbulent energy of fluid ( @xcite ; @xcite ; @xcite ) .
it is not still clear , however , that the galactic dynamo theory can explain the detection of microgauss magnetic fields in high-@xmath11 damped ly@xmath14 absorption systems , which are supposed to be protogalactic clouds ( @xcite ; @xcite ) . on the other hand ,
an alternative scenario is that the intracluster magnetic fields may grow via adiabatic compression of a primordial field frozen with motions of cosmic plasma , where the primordial field is assumed to be produced somehow during the initial stages of cosmic evolution ( e.g. , @xcite ; @xcite ; and also see @xcite for a recent review ) .
recently , using the cosmological , magneto - hydrodynamic simulations of galaxy clusters , dolag , bartelmann , & lesch ( 1999 ) have quantitatively shown that , if starting with the initial magnetic field of @xmath15 g strength , the final intracluster field can be amplified by the gravitationally induced collapsing motions to the observed strength of @xmath16 g irrespective of uniform or chaotic initial field configurations motivated by scenarios of the primordial or galactic - wind induced initial seed fields on mpc scales , respectively . it is undoubtedly clear that the knowledge of the intracluster magnetic field will lead to a more complete understanding of the physical conditions of the intracluster medium and of the possible dynamical role of magnetic fields ( e.g. , @xcite ; @xcite ) .
furthermore , the magnetic field should play an essential role to the nonthermal processes such as the synchrotron and high energy radiations and possibly the cosmic ray production ( @xcite ; @xcite ; @xcite ) .
hence , it is strongly desirable to perform further theoretical and observational investigations on these issues in more detail . in this paper
we investigate how the faraday rotation effect due to the intracluster magnetic field causes a secondary effect on the cmb polarization fields . while thomson scattering of temperature anisotropies on the last scattering surface generates linearly polarized radiation at the decoupling epoch ( @xcite ; @xcite , b ; @xcite ; @xcite , b ) , the plane of linear polarization should be rotated to some extent when the cmb radiation propagates through the magnetized intracluster medium at a low redshift .
the great advantage is that , since the cmb polarization field is a continuously varying field on the sky , we can measure an angular profile of the rotation measure in a cluster in principle , which can not be achieved by any other means .
furthermore , thanks to the frequency dependence of the faraday rotation , one will be able to separate this effect from the primary signal and other secondary signals such as that induced by the gravitational lensing effect due to the large - scale structure ( @xcite ) . in this paper , based on the press - schechter theory for the cluster abundance under the cold dark matter ( cdm ) structure formation scenario and a plausible isothermal @xmath0-model of the intracluster gas distribution ,
we compute angular power spectra of the cmb polarization fields including the faraday rotation mixing effect caused by clusters at low redshifts . as for the unknown field configuration ,
we assume the uniform field of @xmath17 strength across a cluster that is consistent with the recent faraday rotation measurements ( clarke et al . 2001 ) .
this model allows us to estimate the impact of this effect in the simplest way .
our study thus proposes a new empirical opportunity to measure or constrain the intracluster magnetic field in the average sense that properties of the magnetic field could be extracted through changes of the statistical quantities , cmb angular power spectra .
so far previous works have focused mainly on investigations of effects of the primordial magnetic fields on the cmb temperature and polarization anisotropies ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; and see also @xcite for a review ) .
both the amplitudes of the cosmological primordial magnetic field and the mean baryonic density rapidly increase with redshift scaled as @xmath18 and @xmath19 , respectively . the dominant contribution to the faraday rotation effect
is therefore imprinted before decoupling ( @xmath20 ) , unless the structure formation at low redshifts causes a significant amplification of the magnetic field .
kosowsky , & loeb ( 1996 ) found that , if the current value of the primordial field is of the order of @xmath21 g on mpc scales corresponding to the microgauss galactic field under the adiabatic compression , the effect on the cmb polarization is potentially measurable by satellite missions _ map _ and _ planck surveyor _ ( see also @xcite ; @xcite ) .
it is also shown that such a stochastic magnetic field in the early universe affects the temperature fluctuations through the induced metric vector perturbations and thus current measurements can put an upper limit of @xmath22 g on the current field amplitude ( @xcite ) .
this paper is organized as follows . in section [ rm
] we first construct a model to describe the magnetized hot plasma in a cluster by assuming the uniform magnetic field configuration for simplicity .
we then derive the angular power spectrum of the faraday rotation angle based on the press - schechter description of the cluster formation and the @xmath0-model of the gas distribution . in section [ formalism ] , we present a formalism for calculating the angular power spectra of the cmb polarization fields including the faraday rotation mixing effect .
section [ results ] presents the results for the currently favored cdm models . in section [ disc ]
we briefly present the discussions and summary . unless stated explicitly , we assume the favored @xmath5cdm cosmological model with @xmath23 , @xmath24 , @xmath25 , @xmath26 , and @xmath27 as supported from observations of cmb anisotropies and large - scale structure ( e.g. @xcite ) , where @xmath28 , @xmath29 , and @xmath30 are the present - day density parameters of non - relativistic matter , baryon and the cosmological constant , respectively
, @xmath31 is the hubble parameter and @xmath32 denotes the rms mass fluctuations of a sphere of @xmath33mpc radius .
we also use the @xmath34 unit for the speed of light .
if monochromatic radiation of frequency @xmath35 is passing through a plasma in the presence of a magnetic field @xmath36 along the propagation direction @xmath37 , its linear polarization vector will be rotated through the angle @xmath38 ( e.g. , @xcite ) : @xmath39 where @xmath40 and @xmath41 denote the charge and mass of electron , respectively , @xmath42 is the number density field of electron along the line of sight , and @xmath43 is the observed wavelength . note that the factor @xmath44 comes from the relation of @xmath45 and the sign of @xmath38 could be negative depending on the orientation of the magnetic field and the reversal scale of magnetic field .
the equation above clearly shows that the magnitude of the rotation measure depends on the electron number density , the magnetic field strength and the characteristic scale length of the field , and is larger for longer observed wavelengths because of the dependence @xmath46 .
thus , as a possible source of the faraday rotation , in this paper we consider the magnetized hot plasma in clusters of galaxies , motivated by the fact that the deep gravitational potential well of dark matter makes the intracluster gas dense and highly ionized and many observations have revealed the existence of the strong magnetic field with @xmath47 strength(e.g .
, clarke et al . 2001 ) . to calculate the faraday rotation measure for the magnetized intracluster hot plasma
, we have to assume how the magnetic field is distributed in a cluster along the line of sight .
there are two possibilities considered ; the uniform field configuration and , more realistically , the tangled field , where the former field has constant strength and direction through the entire cluster while the latter field has reversal scales along the line of sight . for the coma cluster , kim et al .
( 1990 ) suggested @xmath48 for the field reversal scale from the observed rotation measures of some sources close in angular position to the cluster , and feretti et al .
( 1995 ) discovered the magnetic field structure with smaller scales down to @xmath49 from the rotation measures of radio halo in the coma , leading to the stronger field of @xmath50 on the scale . on the other hand
, there is observational evidence of coherence of the rotation measures across large radio sources , which indicates that there is at least a field component smoothly distributed across cluster scales of @xmath51 ( e.g. , @xcite ) .
thus , the magnetic field structure is still unknown because of the lack of the number of sources available per cluster for measuring the magnetic field .
the angular profile of the rotation measure compiled from all sample of radio sources for 16 different normal clusters ( clarke et al .
2001 ) yields the field strengths between @xmath52 and @xmath53 for a uniform field model , while @xmath54 for a simple tangled - cell model with a constant coherence length @xmath55 . in this paper
, we consider a uniform magnetic field model for the sake of revealing the impact of the faraday rotation effect on the cmb polarization in the simplest way .
we then assume that the amplitude of magnetic field at the cluster formation epoch is universally constant , namely it does not depend on redshift ; @xmath56 .
next let us consider a model to describe the gas distribution in a cluster .
motivated by the x - ray observations , we adopt the spherical isothermal @xmath0-model as for the gas density profile of each cluster expressed by @xmath57^{-3\beta/2 } , \label{eqn : beta}\ ] ] where @xmath58 is the central gas mass density and @xmath59 is the core radius .
the observed values of @xmath0 , obtained from x - ray surface brightness profiles , range from @xmath60 to @xmath61 ( e.g. see @xcite ) . in the following discussion we assume @xmath62 for simplicity . note that makino , sasaki and suto ( 1997 ) showed that the universal density profile of dark matter halo ( @xcite ) can reproduce the @xmath0-model for massive clusters under assumptions of the hydrostatic equilibrium and isothermality . in the spherical collapse model , the virial radius @xmath63 for the halo with mass @xmath64
is obtained from @xmath65 where @xmath66 is the overdensity of collapse and the fitting formula is given by bryan & norman ( 1998 ) , and @xmath67 is the critical matter density at redshift @xmath11 defined by @xmath68 .
once the uniform magnetic field and the gas density profile are given , we can easily calculate the angular profile of the faraday rotation angle for a cluster projected on the sky similarly as the procedure to calculate the profile of the thermal sunyaev - zeldovich ( sz ) effect ( @xcite ; @xcite ) .
using the gas density profile ( [ eqn : beta ] ) and equation ( [ eqn : rm ] ) , the angular profile can be calculated as @xmath69 where @xmath70 , @xmath71 is the relative angle between the direction of the magnetic field and the line of sight , @xmath72 is the central number density of electron , and @xmath73 is the angular separation between the line of sight and the cluster center .
the angular core size @xmath74 is related to @xmath59 via @xmath75 , where @xmath76 is the angular diameter distance to the cluster at redshift @xmath11 . note that @xmath77 , given as the ratio between the viral and the core radius , is not a parameter but a function of mass @xmath64 and redshift @xmath11 in a general case .
we also have to relate the quantities @xmath78 and @xmath79 to the total mass @xmath64 and redshift @xmath11 . assuming the full ionization for intracluster gas , the central electron number density is given by @xmath80 , where @xmath81 is the helium mass fraction and we fix @xmath82 in this paper .
it is known that the formation of cluster core regions contains large uncertainties .
we here employ the _ entropy driven _ model ( @xcite ; @xcite ) to describe the core radius , motivated by the fact that this model can explain observations such as the x - ray luminosity function better than the simplest self - similar model ( @xcite ) does . in this model
the core is supposed to be in a minimum entropy phase @xmath83 , where @xmath84 denotes the central electron temperature , @xmath85 is the specific heat capacity of the gas at constant volume , and @xmath86 is the ratio of specific heats at constant pressure and constant volume .
if using @xmath87 for @xmath88 , we can obtain @xmath89^{1/2}\nonumber\\ & \propto & \frac{t_{e0}^{1/[2(\gamma-1)]}(m , z)}{\delta_c^{1/2}(z ) } ( 1+z)^{-3/2 } , \label{eqn : rcore}\end{aligned}\ ] ] where @xmath90 is the gas mass fraction for a cluster and we assume that @xmath90 is equal to the cosmological mean ; @xmath91 ( @xcite ) .
the adiabatic index is fixed at @xmath92 throughout this paper .
moreover , we assume the isothermality and that the central electron temperature @xmath84 is equal to the virial temperature @xmath93 given by @xmath94 with @xmath95 .
the quantity @xmath96 is a fudge factor of order unity that should be determined by the efficiency of the shock heating of the gas ( @xcite ) , and we fix @xmath97 . if the proportional constant to calculate @xmath78 in equation ( [ eqn : rcore ] ) is determined by assuming @xmath98 , we can obtain the core radius for an arbitrary cluster with mass @xmath64 and at redshift @xmath11 .
this cluster model then yields @xmath99 , @xmath100 , and @xmath101 for @xmath102 at @xmath103 in the @xmath5cdm model .
using these typical values of physical quantities for a cluster with @xmath104 , equation ( [ eqn : angrm ] ) tells us the magnitude of the rotation measure at the cluster center as @xmath105 thus , if the line - of - sight component of the uniform magnetic field has a strength of @xmath106 and we can observe at low frequencies such as @xmath107 , the magnitude of the rotation measure angle could be of the order of @xmath108 for such a massive cluster .
the observed scatter of the normalized rotation measure ( @xmath109 $ ] ) is @xmath110 at the radius in the range of @xmath111 from the cluster center ( kim et al .
1991 ; clarke et al .
the value estimated in equation ( [ eqn : orderest ] ) corresponds to @xmath112 at the cluster center for the typical case of @xmath113 .
our model thus seems consistent with the observations .
this can be more quantitatively verified as follows .
using our model of the rotation measure angle ( [ eqn : angrm ] ) and the press - schechter mass function @xmath114 ( @xcite ) ( see equation ( [ eqn : ps ] ) ) , we can compute the variance of the rotation measure for present - day clusters against the radius from the cluster center : @xmath115 ^ 2\right/ \int^{m_{\rm max}}_{m_{\rm min}}\!\!dm\frac{dn}{dm}. \label{eqn : averm}\ ] ] in the calculation of the equation above , we use @xmath116 for the ensemble average of angle @xmath71 between the magnetic field direction and the line of sight , since the field orientation is considered to be random for each cluster . the solid line in figure [ fig : averm ]
shows the average profile @xmath117 for @xmath118 in the mass range of @xmath119 .
note that this mass range corresponds to @xmath120 for the @xmath121-ray luminosity in the rosat @xmath122 band , where only the thermal bremsstrahlung emission is considered .
thus , this @xmath121-ray luminosity range is roughly consistent with the rosat sample of the faraday rotation measurements used by clarke et al .
( 2001 ) . for comparison ,
square symbols represent the observational data from clarke et al .
( 2001 ) with @xmath123 errors arising from the faraday contributions due to our galaxy .
it is clear that our model with @xmath118 can roughly reproduce the observed magnitude and dispersion of the rotation measures as a function of the radius . in this sense
, it is likely that we can at least estimate a correct magnitude of the faraday rotation effect on the angular power spectra of cmb polarization fields , even though the uniform magnetic field is an unrealistic model , because the contribution to the angular power spectra comes from the second moments of the rotation measure angle by definition ( see equation ( [ eqn : rmclb ] ) ) .
furthermore , there could be an uncertain selection effect in the observational data .
for example , one cluster sample with the large rm value of @xmath124 at the radius of @xmath125 has the @xmath121-ray luminosity of @xmath126 or the mass of @xmath127 , which is smaller than the lowest mass of @xmath128 used for the theoretical prediction shown by the solid line . to illustrate the sensitivity of the lowest mass to the theoretical prediction
, the dashed line in figure [ fig : averm ] shows the result of equation ( [ eqn : averm ] ) for @xmath129 ( @xmath130 ) and @xmath131 , and it clearly underestimates the observational result . for the similar reason ,
it is thought that clarke et al .
( 2001 ) derived a higher value of @xmath132 for the uniform magnetic field model from their observational data .
in addition to these facts , there is possible contamination of line emissions from the intracluster metal to the observed @xmath121-ray luminosity .
therefore , our model of @xmath133 often used in the following discussion could lead to the minimum magnitude of the faraday rotation effect on the cmb polarization power spectra for a realistic effect . as will be shown in detail ,
the faraday rotation effect generates secondary patterns of the cmb polarizations expressed as @xmath134 and @xmath135 in terms of the stokes parameters @xmath136 and @xmath137 , where @xmath138 and @xmath139 are those parameters of the primary polarization generated via thomson scattering at the decoupling ( @xmath140 ) .
hence , equation ( [ eqn : orderest ] ) implies that , if we observe the cmb polarization through a cluster at longer wavelengths such as @xmath141 ( @xmath142 ) , the secondary signals could be comparable to the primary signals for massive clusters with @xmath143 . in this case , we could directly measure the faraday rotation effect in an individual cluster like the sz effect on the cmb temperature fluctuations .
this would be also true for particular clusters that have an unusual stronger magnetic field such as @xmath144 .
moreover , since the primary cmb polarization field has very smooth structure on angular scales of @xmath145 because of the silk damping ( @xcite ; and also see @xcite in more detail ) , the angular profile of the rotation measure could be directly reconstructed in principle from the secondary signals generated on small scales even for a single observed frequency .
these interesting possibilities are now being investigated in detail , and the results will be presented elsewhere ( ohno et al .
2001 in preparation ) .
we are interested in the secondary effect on the angular power spectra of the cmb polarization due to the faraday rotation effect caused by clusters .
we thus need to compute the angular power spectrum of the rotation measure angle @xmath146 , which gives the strength of @xmath38 in the average sense for the all - sky survey at two - point statistics level .
the angular fourier transformation of the rotation measure ( [ eqn : angrm ] ) is given by @xmath147 where @xmath148 , @xmath149 is the zero - th order bessel function , and we have used the small - angle approximation ( bond & efstathiou 1987 ) .
since clusters are discrete sources , contributions to the power spectrum of the rotation measure are divided into the poissonian contribution , @xmath150 , and the contribution from the spatial correlation between clusters , @xmath151 ( peebles 1980 ) . from equation ( [ eqn : angrm ] )
the correlation contribution @xmath151 depends on the ensemble average @xmath152 , where @xmath153 and @xmath154 are the relative angles between the magnetic fields and the line of sights for those two clusters , respectively , while @xmath150 depends on @xmath155 , leading to the non - vanishing contribution for any distribution of @xmath71 .
if the origin of the magnetic field is primordial , the correlation between two intracluster magnetic fields could be non - vanishing even on separation scales between clusters like that between the density fluctuation fields , and in this case we have some contribution from @xmath151 because @xmath156 .
however , it is known that the poissonian contribution always dominates the correlation one because clusters are rare objects ( @xcite ) . for this reason ,
we here assume @xmath157 and consider only the poisson contribution @xmath150 for simplicity . following the method developed by cole & kaiser ( 1988 ) and limber s equation , we can obtain an integral expression of @xmath151 ; @xmath158 where @xmath159 and @xmath160 are the comoving volume and the comoving distance , respectively , and @xmath161 is the comoving number density for halos with mass in the range of @xmath64 and @xmath162 at redshift @xmath11 .
we compute @xmath114 using the press - schechter model ( @xcite ) , and in this case we have @xmath163 where @xmath164 denotes the threshold , @xmath165 is the present - day rms mass fluctuation within the top - hat filter corresponding to the mass scale of @xmath64 , @xmath166 is the threshold overdensity of the spherical collapse model , and @xmath167 is the linear growth factor of density fluctuations .
figure [ fig : clrm ] shows the angular power spectra of the rotation measure through clusters for @xmath168 ( solid line ) and @xmath6 ( dashed ) assuming the observed frequency of @xmath169 in the @xmath5cdm model . note that we have again used @xmath116 .
we have confirmed that dominant contribution to the power spectrum comes from the massive clusters with mass larger than @xmath170 at low redshifts such as @xmath171 , similarly as the results of sz effect due to clusters ( e.g. , @xcite ) .
in this section we develop a formalism for calculating the secondary faraday rotation effect induced by magnetized intracluster hot plasma on the cmb polarization fields and their angular power spectra .
the properties of cmb polarization fields can be described in terms of the stokes parameters @xmath136 and @xmath137 ( @xcite ; @xcite , b ; @xcite ) .
let us consider quasi - monochromatic waves propagating in the z - direction , in which the electric and magnetic fields vibrate on the @xmath172-@xmath173 plane .
we can then express the electric field @xmath174 in terms of the complex orthogonal unit vectors of circularly polarized waves as @xmath175 , where @xmath176 and @xmath177 and @xmath178 are the unit vectors of right - hand and left - hand polarized waves , respectively . using those quantities , the stokes parameters @xmath136 and @xmath137 measured in the fixed @xmath179-coordinates
can be expressed as @xmath180=2e_{\rm r}e_{\rm l } \sin(\delta_{\rm r}-\delta_{\rm
\label{eqn : quparas}\end{aligned}\ ] ] if there is an intervening magnetic field in the ionized plasma between the last scattering surface and the present epoch , the right - hand and left - hand polarized waves travel with different phase velocities by the faraday rotation effect . therefore , from equation ( [ eqn : quparas ] ) , we can derive the following equation to govern the time evolution of @xmath136 and @xmath137 in the presence of the magnetized plasma ; @xmath181 where @xmath182 , @xmath183 is the cosmic time and we have ignored other effects associated with the thomson scattering . note that @xmath184 is also related to the rotation measure angle @xmath38 defined by equation ( [ eqn : rm ] ) as @xmath185 .
equation ( [ eqn : boltz ] ) clearly shows that the faraday rotation causes the mixing between @xmath136- and @xmath137-type polarizations . more importantly , even if the primary cmb anisotropies generate only the @xmath136-type polarization in the statistical sense ( @xcite ) , the faraday rotation mixing induces the @xmath137-type polarization on the observed sky from the primary @xmath136 mode .
since we are interested in the secondary effect on the cmb polarizations , the solutions of equation ( [ eqn : boltz ] ) for @xmath136 and @xmath137 along the line of sight with direction of @xmath186 can be approximately solved up to the first order of @xmath184 as @xmath187 where @xmath139 and @xmath138 denote the primary @xmath137- and @xmath136-polarization fluctuations generated at the decoupling epoch ( @xmath140 ) , respectively , and we have neglected other secondary polarization fields generated at low redshifts that have insignificant power compared to the primary signal on relevant angular scales for a realistic reionization history ( e.g. , @xcite ) .
the quantities @xmath188 and @xmath189 are the cosmic time at the decoupling and the present time , respectively . in
what follows quantities with tilde symbol denote quantities at @xmath188 . the stokes parameters @xmath136 and
@xmath137 depend on the choice of coordinates and are not invariant under the coordinate rotation .
for this reason , it is useful to introduce the new bases of rotationally invariant polarization fields , called the parity - even electric mode ( @xmath2 ) and the parity - odd magnetic mode ( @xmath1 ) ( @xcite ; @xcite , b ; @xcite , b ) .
any polarization pattern on the sky can be decomposed into these modes .
since the secondary effect due to clusters of galaxies on the cmb polarization is important only on small angular scales such as @xmath190 , we can safely employ the small - angle approximation ( @xcite ) .
based on these considerations , @xmath136 and @xmath137 fields can be expressed using the two - dimensional fourier transformation in terms of the electric and magnetic modes as @xmath191e^{i{\mbox{\boldmath$l$}}\cdot{\mbox{\boldmath$\theta$}}},\nonumber\\ u({\mbox{\boldmath$\theta$}})&=&\int\!\!\frac{d^2{\mbox{\boldmath$l$}}}{(2\pi)^2 } \left[e_{{\mbox{\boldmath$l$}}}\sin2\phi_{{\mbox{\boldmath$l$}}}+b_{{\mbox{\boldmath$l$}}}\cos2\phi_{{\mbox{\boldmath$l$}}}\right]e^{i{\mbox{\boldmath$l$ } } \cdot{\mbox{\boldmath$\theta$}}},\end{aligned}\ ] ] where @xmath192 is defined by @xmath193 , and @xmath194 and @xmath195 are the two - dimensional fourier coefficients for the @xmath2- and @xmath1-modes , respectively .
the fourier components @xmath196 and @xmath197 for the primary cmb polarization fields satisfy @xmath198 where @xmath199 and @xmath200 are the angular power spectra of primary @xmath2- and @xmath1-modes , respectively .
the primary @xmath1-mode polarization is generated by the quadrupole temperature anisotropies associated with the tensor ( gravitational wave ) and vector perturbations . if we assume the standard inflation - motivated model with a spectral index of @xmath201 for the scalar perturbations , the tensor and vector perturbations are sufficiently smaller than the scalar perturbations . in the following discussion
we ignore the primary @xmath1-mode power spectrum for simplicity ; @xmath202 . from equation ( [ eqn : magpol ] )
the two - point correlation functions of @xmath136 and @xmath137 fields including the faraday rotation mixing effect can be computed as @xmath203 with @xmath204 where we have employed a reasonable assumption that the primary fields ( @xmath138 and @xmath139 ) on the last scattering surface and the rotation measure field @xmath184 induced by low-@xmath11 clusters are statistically uncorrelated .
note that @xmath205 is the two - point correlation function of the rotation measure angle and can be expressed in terms of the angular power spectrum of the rotation measure as @xmath206 where @xmath207 is given by equation ( [ eqn : clpoiss ] ) .
we are now in the position to derive the angular power spectra of @xmath2 and @xmath1 modes in the observed cmb sky . following the method developed by zaldarriaga , & seljak ( 1998 ) , @xmath208 and @xmath209
can be expressed as @xmath210,\nonumber\\ c_{bl}&=&\pi\int^\pi_0\!\theta d\theta \left [ ( c_q(\theta)+c_u(\theta))j_0(l\theta)-(c_q(\theta)-c_u(\theta))j_4(l\theta ) \right].\end{aligned}\ ] ] therefore , from equation ( [ eqn : magtwo ] ) we have @xmath211 where @xmath212 and @xmath213 are the window functions defined by @xmath214 c^{\rm rm}(\theta),\nonumber\\ w^b_{ll'}&\equiv & 2\int_0^\pi\!\theta d\theta \left[j_0(l\theta)j_0(l'\theta)+j_4(l\theta)j_4(l'\theta)\right ] c^{\rm rm}(\theta ) .
\label{eqn : window}\end{aligned}\ ] ] equations ( [ eqn : rmclb ] ) clearly show that the faraday rotation mixing generates the non - vanishing angular power spectrum of the @xmath1-mode polarization from the primary @xmath2-mode .
throughout this paper , we employ the @xmath5cdm model with @xmath215 , @xmath24 , @xmath25 , @xmath26 and @xmath27 as supported from observations of the cmb anisotropies and the large - scale structure . as for the power spectrum of dark matter used in the calculation of the press - schechter mass function , we adopted the harrison - zeldovich spectrum and the bbks transfer function ( @xcite ) with the shape parameter from sugiyama ( 1995 ) .
figure [ fig : window ] shows the window functions of @xmath1- and @xmath2-modes per logarithmic interval in @xmath55 , @xmath216 and @xmath217 , for a given mode @xmath218 as derived by equation ( [ eqn : window ] ) .
we consider the magnetic field strength @xmath168 and the observed frequency @xmath107 in the @xmath5cdm model .
one can see that the window functions for the mixing between @xmath2 and @xmath1 modes are well localized in @xmath55 space . in figure
[ fig : clpol ] we show the angular power spectra of @xmath2- and @xmath1-mode polarizations on the observed sky including the faraday rotation mixing effect due to clusters for the observed frequency of @xmath219 .
the dot - dashed line shows the @xmath1-mode spectrum for @xmath220 that corresponds to the value estimated from the observational data for a model of uniform intracluster magnetic field ( clarke et al .
2001 ) , while the dotted line shows the result of @xmath118 . for comparison , the @xmath2-mode power spectra with and without the secondary effect of @xmath118
are also shown by solid and dashed lines , respectively .
one can readily see that the generated @xmath1-mode power spectrum has a peak with magnitude of @xmath221 at @xmath4 . on smaller scales of @xmath222 ,
the spectra of @xmath2 and @xmath1 modes coincide , and this is similar as the secondary effects caused by the peculiar velocity field of the ionized medium in the reionized universe ( @xcite ; @xcite ) .
these results imply that , if the uniform magnetic field has a strength of @xmath223 and we can observe at lower frequencies such as @xmath7 , the magnitude of @xmath1-mode could be comparable with the primary polarization amplitude of @xmath224 .
the detectability of the induced @xmath1 mode polarization by the sensitive satellite mission planck can be estimated following the method developed by zaldarriaga & seljak ( 1998 ) .
the relative error on the overall amplitude of the induced @xmath1 component @xmath0 can be estimated as @xmath225 where @xmath226 is the fraction of observed sky and we here fix @xmath227 . as for the detector noise and the beam width ,
we assume @xmath228 and @xmath229 for the @xmath230ghz channel on planck , where @xmath231 is the full - width at half - maximum angle .
note that the @xmath230ghz channel we here consider is the lowest frequency channel on planck . from the dependence @xmath232 of the rotation measure , the magnitude of the @xmath1-mode angular power spectrum at @xmath230ghz becomes smaller by @xmath233 than the results estimated in figure [ fig : clpol ] .
accordingly , for the magnetic field of @xmath168 , we have @xmath234 .
this means that this effect can not be detected by the planck mission for @xmath118 .
conversely , to get @xmath235 at the @xmath230ghz channel , the uniform magnetic field component must have @xmath236 strength , and this seems an unrealistic value . however , we stress that even a null detection of this effect can set a new constraint on the magnetic field strength .
it is also worth noting that this signal would be detected more effectively by a ground based experiments at lower frequency bands such as @xmath237 observing a small patch of the sky for a sufficiently long integration time to reduce the noise per pixel .
in this paper , we have investigated the secondary effect on the cmb polarization fields induced by the faraday rotation effect of the magnetic field in an intracluster hot plasma at a low redshift . to illustrate the impact of this effect in the simplest way , we employed the simple model that the magnetic field has a uniform field strength such as @xmath238 across a cluster universally when the cluster formed , which is consistent with observations of the rotation measures in clusters ( clarke et al .
as shown in figure [ fig : averm ] , our model can roughly reproduce the observed scatter of the rotation measures as a function of the distance from the cluster center ( kim et al .
1991 ; clarke et al .
2001 ) . in this sense , since the faraday rotation effect on the angular power spectra of cmb polarizations comes from the second moments of the rotation measure angle ( see equation ( [ eqn : rmclb ] ) ) , it is likely that our model can at least estimate a correct magnitude of the effect , even though detailed shapes of those spectra could be different for a realistic intracluster magnetic field as discussed below .
we showed that the parity - odd @xmath1-mode angular power spectrum is generated on the observed sky by the faraday rotation mixing effect as a new qualitative feature , when if the primary cmb anisotropies includes the parity - even @xmath2-mode only such as predicted by the standard inflation - motivated scenarios with pure primordial scalar perturbations .
we estimated that the generated @xmath1-mode power spectrum has a peak with amplitude of @xmath239 at @xmath4 under the plausible scenario of the cluster formation in the currently favored @xmath5cdm model .
it was also shown that the lowest frequency @xmath240 channel of planck can be used to set an upper limit of @xmath241 for a uniform component of the intracluster magnetic field .
detection or even null detection of the predicted @xmath1-type polarization will therefore be a new empirical tool to provide a calibration of an uncertain magnetic field in a cluster , combined with measurements of the gas distribution from the x - ray or sz datasince .
it is known that there are other secondary sources on the cmb polarization fields caused in the low redshift universe , and these nonlinear effects generally induce the @xmath1-type polarization pattern from the coupling with the primary @xmath2-mode .
the main source is the gravitational lensing effect of the large - scale structure , leading to the @xmath1-type polarization that has a peak of @xmath242 at @xmath4 and then a power of @xmath243 at @xmath244 for the simliar @xmath5cdm model as considered in this paper ( @xcite ) .
one of authors ( n.s . )
has quantitatvely shown that the secondary effect due to the peculiar motion of ionized medium in the large - scale structure also generates the @xmath1-mode polarization with @xmath245 at @xmath246 for a realistic patchy reionization model of the universe ( @xcite ) . from these results , if the uniform magnetic field has a larger strength of @xmath247 as a common feature of clusters and we observe at lower frequencies such as @xmath7 , the generated @xmath1-mode amplitude due to the faraday rotation effect would be comparable with the primary polarization amplitude and larger than the other secondary signals at @xmath248 , although observations at such low frequencies would suffer from large foreground contamination of the syncrhrotron radiation from our galaxy ( @xcite ) .
even if the uniform magnetic field is not as strong as @xmath6 , the sensitive multi low - frequency measurements will allow one to separate this signal from primary and those other secondary signals in principle thanks to the frequency dependence of @xmath249 on @xmath209 , which is analogous to measurements of secondary temperature fluctuations by the sz effect .
there are possible contributions to affect our results that we have ignored in this paper .
the intracluster tangled magnetic fields , as implied by the rotation measures of some sources inside the coma cluster ( @xcite ) , would provide contributions to our results in addition to the contribution due to the uniform field . however , the coherent scale of tangled fields in a cluster is still unknown .
the recent complete measurements of the rotation measures ( clarke et al . 2001 ) imply the @xmath250 strength of the tangled field under the assumption of the coherent scale of @xmath251 .
it is also known that the strong radio halos embedded within a ( cooling flow ) cluster are pervaded by stronger magnetic fields on smaller scales such as @xmath252 for hydra a ( @xcite ; @xcite ) .
although the simplest tangled - cell model with a constant coherence length @xmath55 has often been considered in the literature , the possible important mechanism to affect our results is the deporlarization effect ( @xcite ) , leading to the cancellations to some extent between the faraday rotations in independent cells with random orientations of the magnetic fields within a cluster .
the finite beam size will also lead to the artificial cancellations between the faraday rotations of cells covered within one beam . from these considerations , the depolarization effect may lead to an intuitive result that a feature of the @xmath1-mode spectrum can be sensitive to the characteristic angular scale , @xmath253 , originating from the coherent length .
this investigation is now in progress , and will be presented elsewhere ( ohno et al .
recently , it has been suggested that outflow or jet acitivities of quasars or black holes leave behind an expanding magnetized bubble in the itergalactic medium ( igm ) resulting in the possibility that the igm is filled by magnetic fields to some extent ( @xcite ; @xcite ) .
even thgouh the igm field has a smaller strength than @xmath254 on @xmath255 scales , a large volume coverage of the ionized plasma might lead to non - negligible contribution connected to the reionization history of the universe . finally , we comment on implications of our results to the nonthermal processes in a cluster
. the observations of diffuse synchrotron halos and nonthermal hard x - ray emission have implied the existence of nonthermal relativistic electrons in a cluster ( e.g. , @xcite ) .
these electrons are produced by the shock acceleration in an intracluster medium , and it will allow us to derive additional information on the physical conditions of the intracluster medium environment , which can not be obtained from the thermal plasma emission only .
the intracluster magnetic field should then play an important role to these nonthermal processes .
in particular , it is being recognized that the nonthermal process can be a unique probe of measuring the dynamically forming clusters before thermalization of the intracluster gas .
recently , waxman & loeb ( 2000 ) pointed out that the synchrotron radiation from those forming clusters could be foreground sources of the cmb temperature fluctuations at low frequency observations such as @xmath256
. those clusters could also produce the high energy gamma ray emission through the inverse compton scattering of cmb photons by relativistic electrons and thus be identified as _ gamma ray clusters _ by the future sensitive telescope _ glast _
, which are difficult to detect by x - ray or optical surveys because of their extended angular size of about @xmath257 ( @xcite ) . based on these ideas , it will be very interesting to investigate cross - correlations between the synchrotron radiation or the gamma ray sources and the faraday rotation signal in cmb polarization map .
this investigation is expected to provide complementary constraints on the nonthermal physical processes in the forming clusters .
m. t. is grateful to prof .
t. futamase and o. aso .
this work was initiated from the conversation with prof .
t. futamase about the original idea of o. aso .
they and m. hattori have also done the similar work as this paper .
m. t. also thanks m. chiba , m. hattori , n. seto and h. akitaya for useful discussions and comments .
we thank u. seljak and m. zaldarriaga for their available cmbfast code .
m. t. acknowledges support from the japan society for promotion of science ( jsps ) research fellowships for young scientist .
this work was supported by japanese grant - in - aid for science research fund of the ministry of education , science , sports and culture grant nos . 01607 and 11640235 , and sumitomo fundation .
adams , j. , danielsson , u. h. , grasso , d. , & rubinstein , h. , phys . lett .
b , 388 , 253 atrio - barandela , f. , & mcket , j. p. , 1999
, , 515 , 465 bagchi , j. , pislar , v. , & neto , g. b. l. 1998 , , 296 , l23 bardeen , j. m. , bond . , j. r. , kaiser , n. , & szalay , a. s. 1986 , , 304 , 15 bower , r. g. , 1997 , , 288 , 355 bond , j. r. , & efstathiou , g. p. 1987
, , 226 , 655 bryan , g. l. , & norman , m.,l .
1998 , , 495 , 80 carilli , c. l. , & taylor , g. b. 2001 , astro - ph/0110655 clarke , t. e. , krongerg , p. p. , & bhringer , h. 2001 , , 547 , l111 cole , s. , & kaiser , n. 1988 , , 233 , 637 cooray , a. 2000 , , 62 , 103506 dolag , k. , bartelmann , m. , & lesch , h. 1999 , , 348 , 351 eke , v. r. , navarro , j. f. , & frenk , c. s. 1998 , , 503 , 569 evrard , a. e. , & henry , j. p. 1991
, , 383 , 95 feretti , l. , dallacasa , d. , giovannini , g. , & tagliani , a. 1995 , , 302 , 680 furlanetto , s. r. , & loeb , a. 2001 , astro - ph/0102076 fusco - femiano , r. et al . 1999 , , 513 , l21 grasso , d. , & rubinstein , h. r. 2000 , to appear in phys
. rep . , astro - ph/0009061 harari , d. d. , hayward , j. d. , & zaldarriaga , m. 1997 , , 55 , 1841 hu , w. 2000 , , 529 , 12 hu , w. , & white , m. 1997 , , 56 , 596 hu , w. , & white , m. 1997 , newa , 2 , 232 hu , w. , & sugiyama , n. 1995 , , 444 , 489 jones , c. & forman , w. 1984 , , 276 , 38 kaiser , n. 1986 , , 222 , 323 kaiser , n. 1991 , , 383 , 104 kamionkowski , m. , kosowsky , a. , & stebbins , a. 1997 , , 78 , 2058 kamionkowski , m. , kosowsky , a. , & stebbins , a. 1997 , , 55 , 7368 kim , k .-
kronberg , p. p. , dewdney , p. e. , landecker , t. l. 1990 , , 355 , 29 kim , k .-
t . , tribble , p. c. , & kronberg , p. p. 1991
, , 379 , 80 komatsu , e. , & kitayama , t. 1999 , , 526 , l1 kosowsky , a. 1996 , ann .
phys . , 246 , 49 kosowsky , a. , & loeb , a. 1996 , , 469 , 1 kronberg , p. p. 1994 , rep.prog.phys . , 57 , 325 kronberg , p. p. , lesch , h. , & hopp , u. 1999 , , 511 , 56 kronberg , p. p. , dufton , q. w. , li , h. , & colgate , s. a. 2001 , astro - ph/0106281 lesch , h. , & chiba , m. 1997 , fundam .
cosmic phys . , 18 , 273 liu , g .- c . , sugiyama , n. , benson , a. j. , lacey , c. g. , & nusser , a. , submitted to apj , astro - ph/0101316 loeb , a. , & mao , s. 1994 , , 435 , l109 loeb , a. , & waxman , e. 2000 , , 405 , 156 mack , a. , kahniashvili , t. , & kosowsky , a. 2001 , astro - ph/105504 makino , n. , sasaki , s. , & suto , y. 1998 , , 497 , 555 navarro , frenk , & white 1997 , , 490 , 493 netterfield et al .
2001 , astro - ph/0104460 ohno , h. , takada , m. , dolag , k. , bartelmann , m. , & sugiyama , n. 2001 , in preparation parker , e. n. 1979 , cosmical magnetic fields ( oxford : clarendon ) peebles , p. j. e. 1980 , the large scale structure of the universe ( princeton ; princeton univ . press ) press , w. h. , & schechter , p. 1974 , , 187 , 425 rees , m. 1987 , qjras , 28 , 197 rephaeli , y. , gruber , d. e. , & blanco , p. 1999 , 511
, l21 rybicki , g. b. , & lightman , a. p. 1979
, radiative processes in astrophysics ( new york : john wiley & sons ) scannapieco , e. s. , & ferreira , p. g. , 1997 , , 56 , r7493 seshadri , t. r. , & subramanian , k. , 2001 , in press , astro - ph/0012056 silk , j. 1968 , , 151 , 459 sugiyama , n. 1995 , , 100 , 281 taylor , g. b. , perley , r. a. 1993 , , 416 , 554 taylor , g. b. , govoni , f. , allen , s. a. , & fabian , a. c. 2001 , astro - ph/0204223 tegmark , m. , eisenstein , d. j. , hu , w. , & de oliveira - costa , a. o. 2000 , , 530 , 133 totani , t. , & kitayama , t. 2000 , 545 , 572 tribble , p. c. 1991 , , 250 , 726 valle , j. p. , macleod , j. m. , & broten , n. w. 1986 , , 156 , 386 vikhlinin , a. , markevitch , m. , & murray , s. s. 2001 , , 549 , l47 waxman , e. , & loeb , a. 2000 , , 545 , l11 welter , g. l. , perry , j. j. , & kronberg , p. p. 1984
, , 279 , 19 white , s. d. m. , navarro , j. f. , evrard , a. e. , & frenk , c. s. 1993 , , 366 , 429 wolfe , a. m. , lanzetta , k. , & oren , a. l. 1992 , , 388 , 17 zaldarriaga , m. , & seljak , u. 1997 , , 55 , 1830 zaldarriaga , m. , & seljak , u. 1998 , , 58 , 023003 zeldovich , ya .
b. , ruzmaikin , a. a. , & sokoloff , d. d. 1983 , magnetic fields in astrophysics , the fluid mechanics of astrophysics and geophysics , vol .
3 ( new york : gordon & breach ) | the observed magnetic field of microgauss strength in clusters of galaxies should induce the faraday rotation effect on the linearly polarized cosmic microwave background ( cmb ) radiation when the cmb radiation propagates through a cluster at low redshift .
the faraday rotation arises from combined contributions of the magnetic field strength , the electron density , the cluster size , and the characteristic scale of the magnetic field along the line of sight . employing the press - schechter prescription for the cluster abundance under the cold dark matter ( cdm ) scenario and a plausible isothermal @xmath0-model for the gas distribution , we compute angular power spectra of the cmb polarization fields including the faraday rotation mixing effect under the simple assumption of uniform magnetic field configuration across a cluster . as a result
, we find that a _ parity - odd _
@xmath1-type polarization pattern is statistically generated on the observed sky , even when the primary polarization only contains the _ parity - even _
@xmath2-mode component , such as in the case of pure scalar perturbations .
the generated @xmath1-type polarization has a peak with amplitude of @xmath3 at angular scales of @xmath4 for the currently favored adiabatic @xmath5cdm model .
this result also implies that , if the magnetic field has a @xmath6 strength and we observe at lower frequencies such as @xmath7 , the secondary signal due to the faraday rotation effect could be comparable to the magnitude of the primary polarization .
the frequency dependence of the faraday rotation can be then used to discriminate the effect from primary and other secondary signals on the cmb polarizations .
our results therefore offer a new empirical opportunity to measure or constrain the intracluster magnetic field in the average sense , combined with measurements of the intracluster gas distribution through the x - ray and sz data . |
the interstellar medium ( ism ) in the disk of the milky way ( and other spirals ) is made up of several thermal `` phases , '' each with a characteristic density and temperature depending on the prevailing physical conditions ( field , goldsmith , & habing 1969 ; mckee & ostriker 1977 ) .
the principal phases of the neutral material in the disks of spiral galaxies , the cold neutral medium ( cnm ) and the warm neutral medium ( wnm ) , are expected to exist in rough pressure equilibrium , embedded in the warm and hot ionized phases of the ism . under the right conditions ,
the cnm can include a molecular component from which young stars form .
the conditions in the equilibrium neutral media are dictated by the ambient pressure and the detailed heating and cooling processes in the media ( see , e.g. , mckee & ostriker 1977 ; wolfire et al .
this is even the case in the self - gravitating molecular clouds in spirals given the external pressure term in the virial equation ( see elmegreen 1999 ) . in the `` standard '' models of thermal phase equilibrium in the neutral media ,
the most important heating source is the photoejection of electrons from the surfaces of interstellar dust grains ( bakes & tielens 1994 ; wolfire et al .
grains are also important as potential coolants ( e.g. , by recombination of electrons and ions on their surfaces ) , and in certain conditions may significantly affect the thermal balance of a medium by the removal of heavy elements from the gas - phase . thus interstellar grains play an important role in the physics of the ism in the thin gaseous disks of galaxies .
their influence on the thermal state of the gas helps determine the phase structure of the medium , and grains go on to play an important role in the star formation process within the dense ism .
our understanding of the physical conditions of neutral clouds in the halos of galaxies is less well developed .
from a theoretical standpoint , clouds far from the planes of spiral galaxies should behave similarly to clouds in the disk , with the same heating and cooling mechanisms and possibly similar phase structure ( wolfire et al .
1995a , b ) . given the importance of dust grains to the thermal physics of clouds at small heights , @xmath3 , from the midplane , knowledge of the dust content of high-@xmath3 material is essential if we are also to understand clouds in the halos and thick disks of spiral galaxies .
much of the material in the halo and thick disk of a spiral galaxy is thought to originate in the thin interstellar disk .
the processes that circulate matter between the thin disks and halos of spiral galaxies operate on both gas _ and _ dust .
however , depending on the method of ejection of matter from the thin interstellar disk and the resiliency of dust grains to destruction by various means ( jones et al .
1994 ) , the dust content of halo material may be different than that of the disk .
this can have important implications for the thermal phase structure and physics of high-@xmath3 gas ( wolfire et al.1995a , b ) . in our own galaxy
there is evidence from gas - phase abundances that individual h@xmath10i clouds at relatively large distances from the plane contain dust ( sembach & savage 1996 ; savage & sembach 1996 , and references therein ) .
there is also evidence for the existence of dust grains in the thickened warm ionized medium or `` reynolds layer '' of the galaxy ( howk & savage 1999a ; lagache et al . 1999 , 2000 ) .
it would therefore seem that the processes responsible for lifting the observed gas clouds out of the plane of the galaxy also affect the dust . and
though the dust in the galactic halo may be to some extent processed , or partially destroyed , the evidence suggests this destruction is not total .
observations of external galaxies have revealed extensive thickened layers of ionized gas traced by its h@xmath0 emission in several edge - on spiral systems ( e.g. , rand , kulkarni , & hester 1990 , 1992 ; dettmar 1990 ; rand 1996 ; ferguson , wyse , & gallagher 1996 ; hoopes , walterbos , & rand 1999 ) .
these extraplanar layers of so - called diffuse ionized gas ( dig ) are thought to be similar to the reynolds layer of warm ionized material in the galaxy ( reynolds 1991 ) . among the most spectacular examples of extraplanar dig
is seen in the nearby ( @xmath12 mpc ) edge - on sb galaxy ngc 891 ( dettmar 1990 ; rand et al.1990 ; keppel et al .
1991 ; rand 1997 , 1998 ) .
the observed h@xmath0emission from this galaxy implies an ionized gas layer with multiple electron scale - heights of @xmath13 and @xmath14 kpc ( hoopes et al.1999 ; rand 1997 ) .
the dust effects the observed quantity and distribution of emission from gas and stars in the thick disk of ngc 891 . ] in howk & savage ( 1997 , hereafter paper i ) we intiated a program to study extraplanar dust in edge - on spiral galaxies .
we presented high - resolution optical images of the nearby galaxy ngc 891 taken with the wiyn 3.5-m telescope at kitt peak national observatory .
these images reveal extensive amounts of highly - structured dust seen in absorption against the background stellar light of the galaxy .
the high - resolution wiyn images show hundreds of individual dust - bearing clouds observable to heights @xmath15 kpc from the midplane along the entire observed length of the galaxy . to denote the distance from the midplane . unless the value in question refers to a single structure , we will be implicitly assuming @xmath16 in our discussion , meaning that the value @xmath3 refers to distances from the midplane to either side of the galaxy . ]
a very simple analysis of the radiative transfer suggests these dust structures are relatively opaque ( @xmath17 to 2.0 ) .
assuming gas - to - dust relationships appropriate for the disk of the milky way ( bohlin , savage , & drake 1978 ) are applicable to these clouds , the inferred hydrogen column densities were in excess of @xmath18 atoms @xmath19 , implying quite large masses ( @xmath20 to 10@xmath21 m@xmath6 ) .
estimates for the potential energies of the observed dusty clouds relative to the midplane are in the range @xmath22 ergs , similar to the energy estimates derived for galactic supershells ( heiles 1979 ) .
the implied total gas mass associated with the ensemble of dusty high-@xmath3 clouds is roughly similar to that estimated for the mass of the extraplanar dig in ngc 891 ( @xmath23 m@xmath6 ; dettmar 1990 ) .
the images presented in paper i clearly show a substantial amount of dust is present in the thick disk of ngc 891 .
subsequent observations by alton et al .
( 1998 ) using the scuba submillimeter bolometer have detected @xmath24 m emission from extraplanar dust to heights @xmath25 kpc in ngc 891 .
the mere presence of the dust features discussed in paper i imply that the mechanism(s ) responsible for transporting material from the thin interstellar disk into the more extended thick disk of ngc 891 does not destroy the grains .
we refer the reader to paper i for a detailed discussion of the various mechanisms that may be at work in lifting gas and dust out of the thin disks of spirals . in our second paper of this series ( howk & savage 1999b , hereafter paper ii )
, we studied the frequency of such extraplanar dust features in normal spiral galaxies .
based on a small wiyn imaging survey of all the massive , edge - on spirals within @xmath26 mpc observable from the north , paper ii showed that the presence of extraplanar dust is relatively common in normal spiral galaxies , with more than half of the galaxies in our final sample exhibiting prominent extraplanar dust structures .
furthermore , in the galaxies of this survey that had previously been searched for extraplanar dig , paper ii showed a one - to - one correlation between the presence ( or absence ) of high-@xmath3dust structures and the presence ( or absence ) of high-@xmath3 dig .
thus galaxies showing extended regions of h@xmath0 emission far from the plane also showed highly - structured dust clouds visible in our broadband images .
the interpretation of the statistical correlation between the high-@xmath3dust structures and the dig is not clear . in general
those galaxies showing high-@xmath3 dust seemed to be relatively uniform in the properties of the dust - bearing clouds , while the h@xmath0 morphologies and brightnesses showed significant variation ( see rand 1996 and pildis , schombert , & bregman 1994 ) .
the discussion presented in paper ii argued that the extraplanar dust features seen in these edge - on spirals likely trace a dense phase of the ism at high-@xmath3 , separate and physically distinct from the extraplanar dig .
several lines of evidence were used to argue for such a configuration , though this conclusion was primarily based on the rough column density estimates for the dusty structures and on the general lack of morphological similarity between the dust - bearing and h@xmath0-emitting material .
the implied column densities with the observed dimensions of the structures yielded crude estimates of the particle density that were irreconcilable with the expected densities of the extraplanar dig .
the dig at @xmath27 kpc is expected to have local electron densities of order @xmath28 assuming a volume filling factor of @xmath29 ( rand 1997 ) .
the dust features are only detectable in these images because they are significantly denser than their surroundings .
estimates of the particle densities in the extraplanar clouds traced by the dust absorption in the galaxies observed in paper iiare in the range @xmath30 @xmath31 . while there is now evidence for a statistical relationship between the the presence of extraplanar h@xmath0 emission and high-@xmath3 dust absorption , there is as of yet little information on any physical relationship between the dig and dust within an individual galaxy .
are these observational tracers measuring the same high-@xmath3 material , or do they each give insight into a separate phase of the thick disk or halo ism ? in this paper we present much deeper observations of ngc 891 in the bvi bands than the data of paper i. we also present deep , high - resolution narrowband ( h@xmath0 + [ ] ) images of ionized gas emission from this galaxy .
the purposes of these observations are to probe the known dust structures in this galaxy as far from the midplane as possible , and to directly investigate the physical relationship between the dust features and the dig in the disk - halo interface of ngc 891 .
our presentation is arranged as follows . in [ sec : observations ] we present a summary of our observations and data processing .
this includes a description of our unsharp masking technique .
the results of our broadband imaging , and the implications for the extraplanar dust structures , are presented in [ sec : thickdisk ] .
our h@xmath0 images are discussed in [
sec : halpha ] , including a direct comparison of the h@xmath0emission from ngc 891 and the morphology of the absorbing dust structures .
the implications of the observations , and in particular the comparison of the h@xmath0 and dust morphologies , are discussed in [ sec : discussion ] .
that comparison shows that these two observational tracers , the dust absorption and the h@xmath0 emission , are revealing two different interstellar phases in the multiphase halo of ngc 891 .
we summarize the major results of this work in [ sec : summary ] .
all of the observations presented here were obtained with the wiyn 3.5-m telescope at kitt peak national observatory in the fall of 1997 , though we have also included some of the data from paper i in our analysis , which were taken 1996 december 4 ( ut ) .
a log of our observations is given in table [ table : log ] .
this table shows the filter used , the date of observation , and the seeing - limited resolution , expressed as the fwhm ( in arcseconds ) of gaussian fits to the stellar images using the iraf routine imexam , for each exposure .
the images presented here were taken under non - photometric conditions .
our analysis depends only on measures of the relative intensity and does not require photometric calibration ( see paper i ) .
the wiyn imager in use at the time of our observations is a thinned 2048@xmath322048 stis ccd with 21 @xmath33 m pixels .
placed at the f/6.5 nasmyth focus of the wiyn telescope , the imager has a @xmath34 field of view , with each pixel subtending @xmath35 on the sky .
the images have been bias - subtracted and flat field - corrected in the usual manner within iraf .
the flat fields were derived from observations of the `` great white spot '' in the telescope dome . in principle subtle variations in the flat - field of the image could mimic dust absorption in ngc 891 ( though very few of the dust features we observe can be referred to as `` subtle '' ) .
our experience suggests that the dome flats can adequately remove small - scale structures associated with the ccd . in the course of our observing we moved the center of the ccd relative to the galaxy by @xmath36 with each new exposure .
this shifts any instrumental flat field features relative to the galaxy .
each individual image was registered to a common reference frame using the positions of foreground stars .
the rms errors in the alignment of the images are typically @xmath37 of a pixel . in the resulting coadded data ,
the effects of any artifacts associated with the flat field structure of the ccd should be greatly diminished .
we have found no real evidence , when comparing our final coadded images with the individual exposures , for confusion of flat - field features with absorbing dust structures .
table [ table : finalimages ] gives the properties of our final coadded images , including the final angular resolution of the images .
each image was smoothed with a gaussian to the worst resolution in a given band before coaddition . for comparison , at the distance of ngc 891 ( @xmath38 mpc ; tully 1988 ) , @xmath39 corresponds to @xmath40 pc .
the narrowband images will be described more fully in
[ subsec : narrow ] .
removal of cosmic rays in the images was performed using standard sigma - clipping algorithms within iraf .
we have derived the astrometric plate solution for our images using a grid of stars whose coordinates were measured in the digitized sky survey .
the rms error in using the plate solution to derive coordinates is approximately 10 .
the sky background in each frame is estimated by fitting a planar ( or in a few cases quadratic ) surface to regions of the images free from light of the galaxy .
stars in these regions have been removed using a median smoothing technique .
this approach is somewhat problematic given the galaxy fills such a large fraction of the ccd area .
the sky - subtraction is not perfect , but we have endeavored to make as reliable a fit as possible .
all of the results presented in this work require an effective yet accurate method of displaying our high - quality images .
we will rely very heavily on unsharp masked displays of our data to show the multitude of structures present at high-@xmath3 in ngc 891 .
our approach to producing quality unsharp masked displays is discussed in some detail in paper ii , but we present here the principal points of that discussion . figure [ fig : vfull ] shows our final v - band data for ngc 891 , as well as the unsharp masked version of the data . the v - band image ( top ) is displayed to show the distribution of stellar light along the length of the galaxy . with this display the light from the bulge is saturated , making it difficult to identify high-@xmath3 dust structures .
however , the unsharp mask of the v - band data ( bottom ) allows us to show the absorbing structures along the entire length of the galaxy in one display .
this masked version of the image shows the complexity of the absorbing structures that thread through the bulge area .
we derive the unsharp masked versions of our images by dividing the original image by a version smoothed with a gaussian kernel having a fwhm of 35 pixels ( @xmath41 ) . with this approach
we remove the large - scale gradients in the background light of the galaxy , allowing us to display all parts of the galaxy in a uniform manner .
the procedure tends to accentuate structures on scales smaller than smoothing kernel ( @xmath42 pc at the distance of ngc 891 ) .
this can be both an advantage and a disadvantage .
most of the structures we find have minor axis scales similar or smaller than this , though not all .
in some sense this offsets the natural tendency to have one s eye drawn towards very large structures .
although we tend to identify structures only in computer displays of the science images ( i.e. , not the unsharp masks ) , the reader s eye may be drawn more to the small - scale structures in our unsharp masks . as in all of our papers , the unsharp masked images are only used for display , never for quantitative measurements .
bright stars near ngc 891 can produce large artifacts in unsharp masked images , destroying information over a disproportionately large area .
we have produced our unsharp masks by dividing the original data by a smoothed image with such problem stars replaced by a two dimensional fit to the surrounding background light before the smoothing process .
the replaced area is usually chosen to be a circular aperture of several times the fwhm of the seeing disk . for fainter examples , the aperture radius was typically twice the size of the seeing disk , while the brighter stars often required apertures in excess of five times the fwhm of the seeing disk .
we have also removed the effects of ccd blooming near a few bright stars . by dividing the original image by the smoothed image with bright stars removed , we are able to keep such stars from influencing large regions of the galaxy , while at the same time showing where these stars lay with respect to the target .
artifacts can be present in our unsharp masked images very near the positions of the brightest stars .
faint halos ( white in this display ) can be seen around a few of the background galaxies and faint stars in figure [ fig : vfull ] .
while there are potential problems associated with the interpretation of the unsharp masks displayed here , this approach is much more effective at showing the reader the true extent and complexity of the dust features we see at high-@xmath3 in ngc 891 .
we are confident that , with the exception of some structures visible very near the brightest stars , the dust features seen in our unsharp masked images are real and not artifacts of the masking procedure
. we will present narrowband images of ngc 891 to study the distribution of ionized gas in this galaxy . to derive an image of the emission lines ( h@xmath0 + [ ] ) from this galaxy
, we have taken images through a narrowband filter centered on the h@xmath0 emission line , and filters that cover continuum regions on each side of the h@xmath0wavelength region .
the characteristics of the on- and off - band filters used are given in table [ table : narrowfilters ] .
the filter off 1 is from the kitt peak filter set ( kp 809 ) , while the h@xmath0on - band filter and the filter designated off 2 are from the wiyn filter library ( w015 and w016 , respectively ) . the h@xmath0 on - band filter contains emission from the h@xmath0 and nearby [ ] and the nearby forbidden [ n@xmath43ii ] lines .
for brevity s sake we will hereafter refer to our image as an h@xmath0image , though the reader should be aware that the bandpass also contains the [ n@xmath43ii ] transitions .
] lines plus a stellar continuum contribution .
to produce a pure emission line image we must subract off the contribution of the stellar continuum .
our two off - band filters cover wavelength ranges shortward and longward of the h@xmath0emission line . to derive an appropriate continuum image we have coadded the off - band data with an appropriate weighting factor .
each of our off - band images was normalized to one second exposure time and combined with a weighting factor , w , defined as @xmath44^{-1},\ ] ] where @xmath45 is the average transmission of the filter , @xmath46 is the transmission - weighted average wavelength of the filter , and @xmath47 is the rest - wavelength of the h@xmath0 emission line .
the ratio of the weighting factors for the two off - band filters is @xmath48 . after combining the two off - band filters to produce a single continuum image , we have performed aperture photometry of a grid of @xmath49 stars in the sky - subtracted on- and off - band images .
the fluxes derived from these stars were used to determine the relative scaling between the on - band and continuum images .
the continuum image was multiplied by this scale factor and subtracted from the on - band image to produce the final emission line image .
the use of this scale factor assures that the foreground stars in our image ( which should be pure continuum sources ) are completely subtracted .
stars which are saturated , however , do not subtract completely .
we reiterate that the size of the galaxy relative to the wiyn imager is large , and our sky subtraction is not perfect . in particular the 5 kpc layer detected by rand ( 1997 , 1998 ) can not be reliably characterized in our images given its extent relative to the size of the ccd .
we do detect emission to @xmath50 kpc in our h@xmath0 image , but the true amount and distribution of such high-@xmath3 gas is confused by the background subtraction uncertainties .
figures [ fig : vmid ] through [ fig : idsw ] show close - up views of the disk in three sections .
figures [ fig : vmid ] , [ fig : vne ] , and [ fig : vsw ] show the unsharp masked views of the central , northeast , and southwest sections of ngc 891 , respectively .
each figure shows a @xmath51 or @xmath52 kpc section of the disk .
figures [ fig : idmid ] , [ fig : idne ] , and [ fig : idsw ] show these same sections with several individual dust features identified . following paper ii we assign each structure an identification of the form ngc 0891:d @xmath53xxx@xmath53zzz .
the `` d '' denotes a dust cloud .
the values xxx and zzz are the projected distances ( in arcsec ) of the features from the center of the galaxy traced along the major and minor axes , respectively .
north and east are positive in this reference frame .
we assume the center of ngc 891 to be @xmath54 , @xmath55 ( a mean of the de vaucouleurs et al .
1991 and ned coordinates ) and a position angle for the disk @xmath56 east of north ( garca - burillo et al .
the designations are only meant to be approximate , and the equatorial coordinates should be used for making detailed comparisons with other wavebands .
hereafter we will typically abbreviate these identifications @xmath53xxx@xmath53zzz .
table [ table : observed ] contains the equatorial coordinates , @xmath3-heights , and approximate dimensions of each structure identified in figures [ fig : idmid ] , [ fig : idne ] , and [ fig : idsw ] .
we have roughly outlined each of the identified structures to help the reader identify the regions used in our calculations . in a few cases we mark with dashed lines regions of absorption that may be associated with the identified structure but are not used in our measurements of the structure . in paper
i we characterized dust structures at high @xmath3 in ngc 891 by their `` apparent extinction , '' @xmath57 .
we defined the apparent extinction in a waveband @xmath58 as @xmath59 where @xmath60 is the surface brightness measured towards a dust cloud , and @xmath61 is the surface brightness of the local background ( see paper i ) .
the light measured towards a given dust feature contains an extincted component of starlight originating behind the feature along the line of sight as well as an unextincted component emitted in front of the feature .
the apparent extinction in a waveband , @xmath57 , is a lower limit to the true extinction , @xmath62 , given the unextincted star light and the unknown contribution of scattering into the beam ( see discussion in
[ subsec : properties ] ) . table [ table : observed ]
contains average values of @xmath63 , @xmath64 , and @xmath65 for each feature ( in magnitudes ) , measured as described in paper i. the combined statistical uncertainties and spatial variations in the dust structures give dispersions typically of order @xmath66 mag for the average @xmath57 values given in table [ table : observed ] .
the systematic errors in the @xmath63 and @xmath64 determinations could be larger . in the b and v images
there is significant overlap between the individual dust structures .
this overlap often makes the definition of a true continuum or background difficult : at low @xmath3 the dust so completely covers the galaxy that the definition of a continuum or background for making these measurements is uncertain .
measurements for clouds much below @xmath67 kpc can be difficult to make , and the errors on the shorter wavelength apparent extinctions can be great . at larger @xmath3 ,
where the dust features become less common and typically show lower @xmath57 values , the systematic uncertainties are likely less important than the measurement error and spatial variations within an individual cloud .
the overlap of the multitude of dust features is less problematic in the i - band , making the @xmath65 measurements more secure than the shorter wavelength apparent extinction values .
our new broadband images reveal a wealth of absorbing structures stretching to very large distances from the midplane of ngc 891 .
figure [ fig : vfull ] shows our v - band image of ngc 891 and the unsharp masked view of these data . though we only show the v - band images here , the structure of the dust in the other two wavebands is much the same as that seen in the v - band .
the absorbing structures are more prominent in the b - band and less so in the i - band .
the v - band images represent a good compromise between the larger opacity at shorter wavelengths and the generally higher signal to noise in the longer wavelength observations . the primary difference between the current dataset and that presented in paper i is the depth of the observations .
the new broadband data are a factor of @xmath68 , on average , deeper than the data presented in paper i , though the resolution of the current dataset is slightly worse .
this allows us to identify structures against the background stellar light to larger heights from the midplane .
we find firm evidence for structures at @xmath69 kpc along most of the length of the galaxy .
for example , figure [ fig : filament1 ] shows a close - up view of the structure identified as @xmath70 in figure [ fig : idsw ] , which will be discussed in more detail below .
material that may be associated with this structure can be traced to at least @xmath71 kpc , if not higher
. another discrete feature at quite large @xmath3 is identified as @xmath72 in figure [ fig : idmid ] .
this structure is seen prominently against the light of a background elliptical galaxy ( approximately e0 , centered at @xmath73 , @xmath74 ) at a height @xmath71 kpc .
there are a few individual structures that can be identified when examining the one dimensional light distribution at heights as large as @xmath75 kpc .
the dust structures are much less prevalent at heights beyond @xmath76 kpc than at lower @xmath3 .
though the background stellar light is rapidly dimming with height , it is clear there are no structures at heights in excess of 2.0 kpc like those seen at heights @xmath77 kpc .
our images show the clear signature of the stellar thick disk ( see morrison et al .
1997 and morrison 1999 ) in all three wavebands . to confirm the apparent thick disk light in our images
was not simply an artifact of the extended wings of the telescope point spread function ( psf ) , we have fit several bright stars with a moffat function ( moffat 1969 ) to approximate the instrumental psf .
we have then constructed several realistic thin disk model light distributions and convolved these with the instrumental psf .
we find at most a few percent of the light observed to heights @xmath78 kpc could be the result of scattering of light from the thin disk ngc 891
. we have found similar results when including a lower scale - height thick disk than that fit by morrison et al .
( 1991 ) , e.g. , by adopting scale heights @xmath79 kpc .
thus there is detected background light against which we could view the dust features to large @xmath3-heights ( e.g. , @xmath80 kpc in the v - band ) .
while a few features can be identified in the range @xmath75 kpc , the number of structures at these heights is much less than at lower heights , and the apparent extinctions are very small ( typically @xmath81 ; often unmeasurable in the other two bands ) .
color maps of ngc 891 derived from our bvi images give a similar picture : we find ngc 891 gets _ bluer _ with increasing height until @xmath82 kpc or so .
the effects of individual dust clouds on the vertical light distribution become significant in one - dimensional cuts through our data at heights in the range @xmath83 kpc .
this is consistent with a phase change of the material , where dust at heights in excess of this value is more diffusely distributed and hence not visible in our images ( see [ sec : discussion ] ) . although , the lack of dust structures at heights beyond @xmath84 kpc may simply be tracing the maximum extent to which the material was ejected . we believe the paucity of structures at very high @xmath3 ( @xmath85 kpc ) is a real effect . there is not a large amount of highly - structured ( clumped ) dust visible in our images at these heights ; however , we can not rule out the existence of a more smoothly distributed component of the dust at very high @xmath3 .
our images reveal only the dusty structures more opaque than their surroundings .
the visible structures are likely tracing significantly denser regions than the surrounding material .
the decreasing number counts and apparent extinctions of features at very high @xmath3 do not imply there is not dust at these heights , but it is clear there are very few of the overdense , dusty clouds seen at low @xmath3 to be found at heights in excess of @xmath86 kpc . in paper
i we found the number of dust features on either side of the midplane of ngc 891 was roughly similar .
this suggests the dust structures do not arise in a warp along the line of sight , which would cause an asymmetry in the distribution of absorbing structures .
the present images show that though the number of structures is similar for each side of the midplane , there is an interesting difference in the morphology of the highest-@xmath3 absorbing structures from one side of the galaxy to the other .
the northern side of the galaxy ( up in our images ) seems to show an extended series of structures oriented roughly parallel to the plane of the galaxy at @xmath87 kpc .
the structure @xmath88 in figure [ fig : idmid ] is indicative of the dominant orientation of features at these heights .
to the south of the midplane , however , the structures are oriented primarily vertically near @xmath89 kpc .
these vertically - oriented structures show an open geometry to high @xmath3 , with a few cometary - like structures ( e.g. , @xmath90 and @xmath91 in figure [ fig : idmid ] ) .
the open structures on the southern side of ngc 891 are possibly tracing galactic outflow or inflow , though other causes may be viable as well .
it is possible that the `` flattening '' of structures on the northern side of the galaxy is caused by the ram pressure from the motion of ngc 891 through an intragroup medium .
the ngc 1023 group to which ngc 891 belongs is dominated , in mass and luminosity , by three disk galaxies ( tully 1980 ) : ngc 1023 ( sb0 ) , ngc 891 ( sb / sbb ? ) , and ngc 925 ( sbc ) .
the total extent of the group is @xmath92 mpc .
ngc 891 is 0.78 mpc projected distance from ngc 1023 , similar to the separation between the milky way and andromeda , and 1.5 mpc from ngc 925 .
the group as a whole is likely bound ( tully 1980 ; hart , davies , & johnson 1980 ) , and ngc 891 lies near its periphery .
it is not unreasonable to expect that a group composed of three moderately massive disk galaxies with an associated population of dwarf galaxies ( at least 10 ; see tully 1980 ) should contain an intragroup medium .
the communication of ngc 891 s movement through a hypothetical medium could propagate through the higher-@xmath3 material to form the horizontally - oriented structures on the northern side of the galaxy .
other causes for the flattening of the dust structures on the northern side of ngc 891 may be viable as well , such as a slight warp of the disk along the line of sight .
however , swaters , sancisi , & van der hulst ( 1997 ) have ruled out a prominent line of sight warp in ngc 891 by comparing models with their h@xmath10i channel maps . though the morphological differences between the two sides of this galaxy may be interpreted in many ways , the possibility that the flattening of the high-@xmath3 structures on the northern side is an intriguing possibility .
the directly observable physical properties of a subset of dust features are given in table [ table : observed ] . from these observables
we can infer rough information on the column density , mass , and gravitational potential energies of the individual structures . in paper
i the radiation transfer through an individual dust structure was treated very simply .
we assumed some fraction @xmath93 of the stellar light of the galaxy resided between the observer and the dust structure .
therefore , a fraction @xmath94 of the background stellar light was extincted by the dust feature . in this case
the ratio of the observed surface brightness towards a dust cloud , @xmath95 , to that of the local background @xmath61 , could be written @xmath96 , \label{eqn : radtrans}\ ] ] where @xmath97 is the estimated optical depth in a waveband @xmath58 .
if @xmath98 is chosen correctly and scattering is not important , then @xmath99 , where @xmath100 is the `` true '' absorption optical depth . using observations in three wavebands and assuming @xmath98 to be constant with wavelength ( @xmath101 )
, we derived rough corrections to account for the stellar light arising in front of the dust structures .
we will denote these corrected values @xmath102 , where we have derived some fraction @xmath93 of the stellar light to arise in front of the dust structures .
the corrected apparent extinctions @xmath103 were derived in paper i assuming the ratios of the optical depths in the three wavebands followed a cardelli , clayton , & mathis ( 1989 ) parameterization of the interstellar extinction curve .
the first - order corrected @xmath103 values are lower limits to the true extinction , @xmath104 , given the unknown geometry , and hence the unknown contribution from scattering .
the corrected @xmath105 values derived in this way are therefore more correctly called an `` attenuation '' ( e.g. , ferrara et al . 1999 ; meurer , heckman , & calzetti 1999 ) .
if @xmath93 is roughly correct , we have a relationship @xmath106 .
including the observed values of @xmath65 for the current set of dust features in this approach , however , does not yield a consistent solution for @xmath93 and @xmath105 for reasonable values of @xmath107 .
we believe the reasons for this are ( at least ) two - fold .
the aforementioned differences between the structure of the features when viewed in the b- and v - band data versus that seen in the i - band data imply that the @xmath57 measurements of individual dust features may be probing different regions of the clouds at different wavelengths .
further , the assumptions made in the very simple approach to the radiation transfer presented in equation ( [ eqn : radtrans ] ) are likely not valid .
the effects of scattering may be important .
previously the scattered light could be approximately treated by incorporating it into the value of @xmath93 .
however , it is our belief that the large wavelength range covered by the current set of observations shows the assumed wavelength independence of @xmath93 is erroneous , at least as formulated in equation ( [ eqn : radtrans ] ) .
the values of @xmath108 given in table [ table : observed ] are firm lower limits to the true extinction , @xmath104 , through the individual dust clouds . in table
[ table : derived ] we show the magnitude of the implied correction by giving @xmath103 for values of @xmath109 0.25 , and 0.5 .
the @xmath110 measurements are firm lower limits to the true extinction , and we use these in all of our subsequent calculations .
we can use the lower limits @xmath111 to roughly estimate several other physically interesting properties of the individual dust features .
assuming the observed relationship between total hydrogen ( neutral plus molecular ) column density and color excess derived by bohlin et al .
( 1978 ) for the galactic disk is appropriate for the high-@xmath3 clouds in ngc 891 , we can estimate the hydrogen column density , @xmath112 , of the high-@xmath3 dust features with : @xmath113 } \label{eqn : columndensity}\ ] ] ( e.g. , paper ii ) . with an estimate for @xmath112 and the projected surface area of a cloud
, we can estimate the enclosed mass in the observed dust structures .
the results of these calculations are given in table [ table : derived ] .
the outlines of the dust clouds in figures [ fig : idmid ] , [ fig : idne ] , and [ fig : idsw ] identify the areas used for calculating the mass of each structure .
we include a factor of 1.37 correction to the mass estimates for helium and the heavy elements .
assuming rough values for the distribution of mass in ngc 891 we estimate the gravitational potential energy , @xmath114 , of the structures relative to the midplane .
we assume the mass is distributed in an isothermal sheet [ mass density @xmath115 with a mass scale height @xmath116 kpc ( e.g. , the near infrared observations of aoki et al .
1991 which roughly agree with the results of xilouris et al.1998 , who fit exponential disk models to the light distribution ) and mass surface density @xmath117 m@xmath6pc@xmath118 ( derived for the solar neighborhood by bahcall 1984 ) .
the potential energies of the structures relative to the midplane are then calculated as in paper i , using the @xmath3-heights and masses from tables [ table : observed ] and [ table : derived ] .
these estimates are order of magnitude values and should be viewed with some caution .
the values @xmath119 given in table [ table : derived ] are in the range 0.2 to 0.8 mag .
it is unlikely that the true values of the extinction are anywhere near this small .
however , even adopting the smallest allowed values of @xmath105 , the estimates of @xmath112 , mass , and energy are still large .
these values are derived assuming the gas - to - dust relationships of the galactic _ disk _ are appropriate for the high-@xmath3 structures seen in ngc 891 .
as discussed in paper i , the gas - to - dust relationships could be altered by dust destruction in hydrodynamical processes ( jones et al .
1994 ) or by the separation of the gas and dust through the effects of radiation pressure ( ferrara et al . 1991 ; davies et al . 1998 ) .
our images suggest that the destruction of dust grains can not be too severe , given that high-@xmath3 dust is visible in any form . though radiation pressure could play a role in the shaping of the visible dust structures , they are opaque enough to make such a mechanism relatively inefficient . the column densities given in table [ table : derived ] are roughly consistent with @xmath120 @xmath19 , or greater .
the extinction values and column density estimates suggest that molecular material may be present in these structures ( e.g. , savage et al . 1977 ; see also elmegreen 1985 and mckee 1989 for theoretical viewpoints ) .
the dust features identified in ngc 891 may contain large amounts of mass , of order @xmath121 m@xmath6 , assuming galactic gas to dust relationships .
the masses are reminiscent of the galactic giant molecular clouds ( gmcs ; e.g. , solomon & sanders 1985 ) .
the implied potential energies are also quite large given these masses and large heights above the plane .
sandage ( 1961 ) also suggests that the support of an extended , dusty layer of material in ngc 891 must entail large amounts of energy .
it should be pointed out that while detailed radiative transfer models are beyond the scope of this paper , some of the more sophisticated approaches to radiative transfer through a clumpy medium ( e.g. , kuchinski et al .
1998 ; witt & gordon 1996 ) could perhaps help answer some of the more important questions regarding the properties of the observed high-@xmath3 dust .
for example , detailed radiative transfer models might help shed light on the total mass of the dust at high-@xmath3and on the effects of high-@xmath3 dust opacity on optically - derived scale heights ( e.g. , for determining the scale height of dig emission ) .
several of the absorbing structures identified in figures [ fig : idmid ] , [ fig : idne ] , and [ fig : idsw]warrant further discussion .
we briefly comment here on a few of the more spectacular structures visible in our images .
ngc 0891:d @xmath70 : as identified in figure [ fig : idsw ] this structure is an individual cloud at @xmath122 kpc .
however , there are several individual clouds aligned vertically that may make up a seemingly coherent structure traceable from @xmath123 to @xmath124 kpc .
figure [ fig : filament1 ] shows a close up view of the unsharp masked v - band image centered on the region around @xmath70 .
several very high-@xmath3 dust structures can be seen in this region .
if @xmath70 represents the upper - most clump of an irregular filament , the low-@xmath3 anchor is a ring - like structure at @xmath125 pc whose diameter is similar to the width of the filament near its base ( @xmath126 pc ) .
our h@xmath0 images , which will be discussed below , show an h@xmath10ii region coincident with the inside of the dust ring . the irregular absorbing clump in the center of the dust ring seems to lie in front of the h@xmath10ii region .
we find no evidence for h@xmath0 emission associated with the filament itself . like many of the dusty structures seen in our images @xmath70 and the lower-@xmath3 portions of the filament are seen in absorption against the diffuse h@xmath0 emission of the ionized thick disk ( see below ) .
haffner , reynolds , & tufte ( 1998 ) have found an ionized filament in the dig of the milky way that seems to stretch to @xmath127 kpc as a coherent feature ( as evidenced by its velocity structure ) .
however , this filament is _ only _ seen in h@xmath0 emission , with h@xmath10i , x - ray , and ir emission being absent .
the suggested column density is @xmath128 @xmath19 , and the stringent limits on the h@xmath10i column density suggest the structure is almost fully ionized .
the haffner et al .
filament therefore has significantly different physical conditions than the extended filament tentatively identified with @xmath70 .
if the clouds visible beneath @xmath70 are physically associated , making a coherent , though irregular , filament more than 1.8 kpc in length , it is a uniquely intriguing structure .
ngc 0891:d @xmath90 : this structure ( identified as feature 7 in paper i ) is clearly cometary shaped with its apex pointed towards the center of the galaxy and a tail of lower column density material extending to high @xmath3 .
we show a close - up of this structure in figure [ fig : cometary1 ] .
the center of the head of the cometary structure is at @xmath129 kpc , but the trailing material can be traced to almost @xmath130 kpc .
this structure is exemplary of the vertically - oriented structures at high @xmath3 on the southern side of ngc 891 .
we find no evidence for h@xmath0 emission associated with this structure .
ngc 0891:d @xmath131 & @xmath132 : together these two structures ( identified as features 11a and 11b in paper i ) seem to trace the walls of a supershell centered near @xmath133 pc with a diameter of @xmath134 pc ( see figure [ fig : idne ] ) .
as discussed in paper i and below ( [ subsec : comparison ] ) , there is h@xmath0 emission loosely associated with the dusty shell walls .
the h@xmath0 emission from this structure has been discussed also in pildis et al .
our images show no `` cap '' to the structure in the dust distribution , suggesting a supershell open to high @xmath3 .
the h@xmath0 images show a patch of diffuse emission at @xmath135 pc that resides near the center of the shell , but this seems not to be the top of the structure , given that the walls traced by the h@xmath0 emission can be found to larger @xmath3-heights .
perhaps this structure represents a supershell that has experienced `` blowout . ''
we will address this issue in more detail below when we discuss the h@xmath0 emission line images .
ngc 0891:d @xmath136 : though this structure is interesting for its height ( @xmath122 kpc ) , perhaps its most interesting attribute is that it obscures a background spiral galaxy ( see figure [ fig : idmid ] ) .
the southern edge of this spiral , centered at @xmath137 , @xmath138 , is completely obscured in the b - band .
while the v - band shows slightly more of the low-@xmath3 end of this galaxy , there still appears to be a significant amount of the galaxy that is simply missing .
the analysis of the light from this galaxy could in principle give us a better estimate of the opacity though a high-@xmath3dust structure .
unfortunately such an analysis is complicated by the slightly irregular , asymmetric structure of the galaxy .
furthermore , an unresolved source is seen at the apex of the curve traced by the dust structure ( @xmath139 , @xmath140 ) , and directly over the expected emission from the background spiral , in the i - band .
this source is not present in either the v- or b - band images .
the archival _ hst _ image presented in paper i , which is taken through a very broad filter that encompasses much of the v and r bandpasses , clearly shows an unresolved source as well .
this object could conceivably be an intermediate age or globular cluster lying behind @xmath136 , and hence being extincted by this dust structure .
it is likely not a supernova associated with the background spiral given its relatively constant appearance in images taken over several epochs .
if it is a star forming region in the background galaxy , it is much brighter than any other such regions that might exist in the galaxy .
it is clear from visual inspection that this background spiral is very heavily extincted in the b and v images .
emission from extraplanar dig in ngc 891 was first detected by dettmar ( 1990 ) and rand et al .
( 1990 ) through narrowband imaging .
spectroscopic studies of the dig have more thoroughly characterized its distribution and properties ( dettmar & shultz 1992 ; keppel et al.1991 ; pildis et al .
1994 ; rand 1997 , 1998 ) .
the dig layer has been fit by a two - component exponential distribution including a thick disk with an electron scale height @xmath141 kpc and a more extended halo with a more uncertain scale height @xmath142 kpc ( hoopes et al .
1999 ; rand 1997 ) .
it should be noted that these fits have made no correction for the dust that is clearly visible in our broadband images .
fits constrained to match only the light in excess of @xmath143 kpc should be reasonably secure .
below @xmath144 kpc , the covering factor of the dust features is greater than unity , and below @xmath145 there is virtually no information on the intrinsic distribution of optical light from ngc 891 .
the power required to ionize the dig can only be comfortably met by the ionizing radiation from early type stars ( e.g. , rand et al .
1990 ) . the strength of forbidden line emission from [ ] and [ ] relative to h@xmath0 increases with height above the plane ( dettmar & shultz 1992 ; rand 1997 , 1998 ) , as does the strength of [ ] to h@xmath146 ( rand 1998 ) .
pure photoionization models for the dig ( e.g. , sembach et al . 2000 ; domgrgen & mathis 1994 ) with very dilute ionizing radiation fields can match the high [ ] and [ ] to h@xmath0 ratios , though the very highest [ ] and [ ] to h@xmath0 ratios may require heating mechanisms beyond those in low - density ionized nebulae ( reynolds , haffner , & tufte 1999 ) .
the increase in the [ ] /h@xmath146 ratio suggests a secondary source of ionization and/or heating may be at work ( see discussion in rand 1998 ) .
given the very high [ ] and [ ] to h@xmath0 ratios observed in the extraplanar dig , light scattered from low-@xmath3 nebulae off of high-@xmath3 dust grains can not dominate the observed high-@xmath3 h@xmath0 emission ( see ferrara et al .
1996 ; wood & reynolds 1999 ) .
figure [ fig : hafull ] shows our continuum - subtracted h@xmath0 image of ngc 891 .
the display in the top panel of figure [ fig : hafull ] shows the faintest diffuse gas , particularly the often faint high-@xmath3material , while saturating the brightest emission .
the display in the bottom panel is intended to show the structure of only the brightest h@xmath0 , including the very bright structures present above the plane in the northeast section of the disk .
we comment briefly here on a few of the important aspects of the emission line images presented in figure [ fig : hafull ] .
our primary interest in the h@xmath0 images is a comparison of the extraplanar dig and dust morphologies ( see [ subsec : comparison ] ) .
thus we will not detail the many aspects of the ionized gas layer ( see dettmar 1990 and rand et al .
1990 for descriptions of the dig distribution ) .
however , we will briefly discuss the properties of several individual ionized filaments in [ subsec : filaments ] .
our h@xmath0 images clearly show emission from the dig at large distances from the midplane .
we detect h@xmath0 emission to heights exceeding @xmath147 kpc , though our sky subtraction difficulties do not allow us to accurately characterize the highest-@xmath3 emission .
some of the h@xmath0emission is in the form of discrete large - scale filaments , most of which seem to run roughly perpendicular to the midplane .
there is also a more diffuse background of h@xmath0 emission visible along most of the disk and to high @xmath3 .
several large h@xmath10ii regions on the front side of the galaxy dominate the light near the midplane . as noted by rand et al .
( 1990 ) and dettmar ( 1990 ) , there is a strong asymmetry in the strength of the emission between the northeast and southwest sides of the disk ( left and right , respectively , in our images ) ; this asymmetry is also seen in the radio continuum observations of dahlem , dettmar , & hummel ( 1994 ) and in the far infrared ( 50 @xmath33 m ) observations of wainscoat , de jong , & wesselius ( 1987 ) .
we also see an asymmetry in the light distribution in our broadband images , with the northeast section of the disk being brighter than the southwest section at most heights @xmath148 kpc .
this appears to be due to a greater amount of intervening high-@xmath3 absorbing material on the southwest side of ngc 891 .
the h@xmath0 asymmetry is much stronger than that seen in the broadband images , and the radio continuum measurements are not affected by dust , implying a real asymmetry in the amount of star formation between the two sides of ngc 891 . aside from the known extraplanar dig
, we also detect what appear to be several discrete h@xmath10ii regions at high @xmath3 in ngc 891 .
these candidate nebulae are visible in figures [ fig : hadustmid ] , [ fig : hadustne ] , and [ fig : hadustsw ] which show the middle , northeast , and southwest sections of the disk , respectively , in h@xmath0and dust . these figures will be discussed in more detail below . in figure
[ fig : hadustmid ] an unresolved source in the h@xmath0 image is seen in the northern side of the disk ( in the upper right corner of this display ) corresponding to an unresolved continuum source in the broadband images .
this object lies at @xmath129 kpc above the midplane .
a similar object is seen in figure [ fig : hadustne ] , also on the northern side of the disk ( above the disk , near the middle in this display ) .
this object is again coincident with an unresolved source in the broadband images at @xmath149 kpc .
several such unresolved knots of h@xmath0 emission can be seen in figure [ fig : hadustsw ] at heights @xmath150 pc , all of which are coincident with continuum sources .
these objects are not among the list of the planetary nebulae identified in ngc 891 by ciardullo , jacoby , & harris ( 1991 ) .
these candidate h@xmath10ii regions and their associated continuum sources will be discussed in a future publication ( howk & savage , in preparation ) . in this section we very briefly summarize the properties of several of the filamentary structures observed atop the more diffuse component of the dig in ngc 891 .
we do this both because it allows the reader to compare the general properties and distribution of the ionized gas filaments with those traced by the dust , and because no previous catalog of such structures exists in the literature , though there seems to be a general consensus that this filamentary component is important ( rand 1997 , 1998 ; rand et al .
1990 ; dettmar 1990 ) .
figure [ fig : filaments ] shows another view of our high - resolution h@xmath0 observations of ngc 891 .
we have applied an unsharp masking procedure to the h@xmath0 data displayed in the top panel of figure [ fig : filaments ] , while the bottom panel shows the v - band unsharp masked image from figure [ fig : vfull ] for comparison . to produce the unsharp masked h@xmath0 image shown in figure [ fig : filaments ] ,
we used a gaussian smoothing kernel with fwhm@xmath151 pixels ( @xmath41 ) , the same used in producing the v - band unsharp masked images .
we have identified and labelled a small number of filamentary structures in figure [ fig : filaments ] .
we summarize the rough properties of the identified filaments in table [ table : filaments ] .
these properties include coordinates , @xmath3-heights , and dimensions .
we use a naming scheme like that adopted for the dust structures in [ subsec : nomenclature ] , identifying each filament using the form ngc 0891:dig @xmath53xxx@xmath53zzz . as with our discussion of individual dust structures , our selection of h@xmath0-emitting features
is biased towards the more spectacular examples at high @xmath3 ( where the confusing influence of dust absorption is lessened ) .
figure [ fig : filaments ] shows much less small - scale structure in the h@xmath0 unsharp - masked image than the unsharp - masked images of the v - band data , which accentuate the absorption by dust . in part the smaller number of filaments in the h@xmath0 images is caused by the sensitivity of our observations to interstellar matter : where the dusty filaments are present in galaxies , they are much easier to observe than the low - intensity h@xmath0-emitting filaments .
however , it is clear from figures [ fig : hafull ] and [ fig : filaments ] that the filamentary structures in the dig do not show as great a contrast compared with the background emission as the dust structures , which are visible only because they are much more opaque than their surroundings . not only are the h@xmath0-emitting filaments less numerous ( and less prominent ) than the dust structures seen in absorption , but much of the h@xmath0 emission seems to be in the form of a more smoothly - distributed medium . however , the appearance of this `` smooth '' component is made more complicated by the absorption caused by the dust features seen in our broadband images .
we have avoided identifying any apparent structures that may be caused by absorption due to intervening dust .
the distribution of dust - laden filaments is more symmetric about the center of ngc 891 than the distribution of h@xmath0 filaments .
for example , figure [ fig : vsw ] shows the southwest region of the disk contains a number of prominent dust features .
though a few h@xmath0-emitting structures seem to be visible in this region most of them are the result of absorption by the irregularly - distributed dust - bearing clouds .
the relatively small number of h@xmath0 filaments in the southwest region of the disk compared with the northeast region is consistent with the general asymmetry in the h@xmath0 emission across the center of ngc 891 ( rand et al . 1990 ; dettmar 1990 ; dahlem et al.1994 ) . in general the filaments
we have identified in figure [ fig : filaments ] are oriented roughly perpendicular to the plane of ngc 891 .
most are singular arcs of emitting material , though a few are in the form of shell - like structures ( e.g. , ngc 0891:dig @xmath152 and the pair @xmath153 and @xmath154 ) .
the faint ionized structure ngc 0891:dig @xmath152 is one of the more spectacular features in our h@xmath0 image .
as identified in figure [ fig : filaments ] ( see also figure [ fig : hafull ] ) , it forms a bubble - like structure approximately 650 pc in diameter , centered @xmath155 pc above the midplane .
this structure is closed on itself , and if it traces a superbubble there is no evidence for break - out .
however , it is interesting as much for its location above the plane of the galaxy as its morphology .
even single supernova explosions in low - density halos of galaxies can effect their environments on very large scales ( e.g. , shelton 1998 ) , and this structure could conceivably be the result of a high-@xmath3 supernova .
the shell traced out by the pair of filaments ngc 0891:dig @xmath153 and @xmath154 , which are associated with the dust structures ngc 0891:d @xmath131 and @xmath132 , respectively ( see below ) , may also be tracing a large ( @xmath134 pc diameter ) shell in the ism , though at lower @xmath3 .
this shell , centered at @xmath156 pc , has no obvious `` cap '' or top , though there is a diffuse patch of emission immediately between these filamentary shell walls .
these ionized filaments , along with ngc 0891:dig @xmath157 are discussed in some detail by pildis et al .
( 1994 ) .
a few of the h@xmath0-emitting filaments seem to be more diffuse ( thicker ) than the typical dust structures ( e.g. , ngc 0891:dig @xmath152 and @xmath158 ) , though most of those listed in table [ table : filaments ] are of comparable size to the absorbing dust clouds .
the emission measures towards the brightest filaments identified in table [ table : filaments ] are 23 times that of the nearby diffuse component of the h@xmath0 emission , while the fainter are only @xmath159 brighter .
the brightest filaments are only present above the region of vigorous star formation in the ne section of the disk .
there is no evidence in any of our images for absorption that is directly associated with the filaments observed in our h@xmath0 images .
this does not imply these structures lack dust .
if we assume that the filaments have emission measures approximately the same as the average at @xmath160 kpc , i.e. , @xmath161 pc @xmath162 ( e.g. , rand et al . 1990 ) , and assume densities of @xmath163 @xmath31 , then the hydrogen column densities in these filaments should be @xmath164 @xmath19 . if the gas - to - dust ratio in the filaments is similar to gas in the galactic disk , these structures should not show detectable absorption in our images .
we are not sensitive to extinction in the more diffuse component given our method of identifying dust in our images .
we discuss in detail the relationship between the observed dust - laden filaments and the morphology of the h@xmath0 emission in the next subsection .
the h@xmath0 images presented in figure [ fig : hafull ] show that the extraplanar dig in ngc 891 contains many filamentary structures , some stretching to quite large @xmath3 .
however , comparing the h@xmath0distribution with that of the dust seen in figure [ fig : vfull ] suggests that the h@xmath0-emitting material is more smoothly distributed than the matter traced by the dust absorption .
are these two views of the high-@xmath3 ism in ngc 891 tracing the same structures ?
the answer to this question is almost certainly no .
there are very few regions in the thick disk of this galaxy where the dig and dust structures are associated with one another in our images , and in those few regions the association is only a loose one .
there are a large number of regions where the h@xmath0 emission is obscured by the dust features , usually in much the same amount as the stellar continuum emission seen in our broadband images .
studies of h@xmath0 emission from ngc 891 must systematically underestimate the ionized gas emission measures at heights @xmath165 kpc , though the effect is most important for @xmath166 kpc .
the extinction due to the dust may also cause the observed distribution of light to differ from the intrinsic distribution ; emission scale heights derived from fits to @xmath167 kpc may need to be reconsidered given the observed absorption of the h@xmath0 emission by the dust structures in our images .
figures [ fig : hadustmid ] , [ fig : hadustne ] , and [ fig : hadustsw ] show a comparison of the dust and h@xmath0 morphology for three regions , one near the center of the galaxy ( figure [ fig : hadustmid ] ) , one @xmath168 kpc projected distance northeast of the center ( figure [ fig : hadustne ] ) , and the last centered @xmath169 kpc to the southwest of the center ( figure [ fig : hadustsw ] ) .
each of these figures shows grayscale images of the ionized gas emission as well as the v - band unsharp masked view from a @xmath170 ( @xmath171 kpc ) region of ngc 891 ( top and bottom two panels , respectively ) . in the right two panels ,
contours of the h@xmath0 emission are overlaid on top of these grayscale images .
the h@xmath0 contours can be slightly misleading given they show a small number of quantized levels of emission .
they do , however , offer a reasonable way of displaying one dataset on top of another .
one should use both the grayscales and contours for comparing in detail the distribution of dust and ionized gas and dust in these figures .
it is clear from figures [ fig : hadustmid ] , [ fig : hadustne ] , and [ fig : hadustsw ] that any direct association of the extraplanar ionized gas and dust at high @xmath3 is very weak . towards the center of the galaxy , shown in figure [ fig : hadustmid ] ,
a significant amount of the structure seen in the h@xmath0 images is simply caused by the absorption of h@xmath0 photons by foreground dust features .
the h@xmath0 contours are often just offset from the dust features .
but the general impression is not one of h@xmath0 brightening at the edges of the dust features , but rather the dimunition of the more smoothly distributed background emission .
the southwest section of the disk displayed in figure [ fig : hadustsw ] gives a very similar impression .
in particular one can quite clearly see evidence for absorption of h@xmath0 photons by the low-@xmath3 portions of the dust structures @xmath70 and @xmath172 ( see figure [ fig : idsw ] ) .
one - dimensional cuts of the light distribution of the h@xmath0 emission reinforce this conclusion that the h@xmath0 light is being diminished by the absorbing dust structures .
typical measurements of the apparent extinction values towards a number of dust features in our h@xmath0 images yield results intermediate between the @xmath64 and @xmath65 values .
there is typically no increase in the brightness of the h@xmath0 emission as one nears the location of a dust feature , but rather a dimunition of the light from ionized gas .
there are certainly filaments in the h@xmath0 images that are not simply caused by dust absorption effects , some of which are discussed in the previous subsection .
some of the filaments have also been observed by rand ( 1998 ) to have slightly different forbidden emission line intensities relative to h@xmath0 .
pc above the midplane .
] this suggests these structures may be physically distinct from the general background emission and hence not simply the result of dust absorption . in the northeast section of the disk , shown in figure [ fig : hadustne ] ,
we see several dust structures that appear to have been shaped by the effects of star formation ( e.g. , @xmath173 , @xmath131 , and @xmath132 ; see figure [ fig : idne ] ) .
this region of the disk was imaged by pildis et al .
( 1994 ) , who described the h@xmath0emission from two ionized `` supershells . ''
this side of the disk of ngc 891 is experiencing vigorous star formation as traced by various means ( e.g , .
rand et al .
1990 ; dettmar 1990 ; dahlem et al .
1994 ) .
a few of the dust structures in the northeast section of the disk , particularly those with morphologies suggestive of shells and cones , show a loose association of the dig emission and the dust absorption structures ; an association that can likely not be caused by the simple attenuation of background emission . examining figure
[ fig : hadustne ] we see the twin walls of a shell traced by the dust structures @xmath131 and @xmath132 on the southern side of the plane ( below the plane in this figure ) .
the region immediately interior of the shell walls traced by these dust features is bright in h@xmath0emission .
however , if one examines the position of the h@xmath0 contours with respect to the dust absorption , the alignment of the walls traced by the dust absorption and the h@xmath0 emission is not perfect . in particular the northern - most wall visible in absorption ( @xmath132 )
is not particularly well aligned with the corresponding h@xmath0 arc , and may even be present in absorption against the h@xmath0 emission at high @xmath3 . interestingly , though there is an irregular clump of h@xmath0-emitting material near the center of this shell , there is little evidence for a top or cap to this shell in the h@xmath0 images .
another structure showing a loose association between the h@xmath0emitting material and that traced by the dust absorption is the feature identified as @xmath173 in figure [ fig : idne ] .
this structure , appearing as a vertical cone open to high-@xmath3 in the dust absorption , appears to be edge - brightened in h@xmath0 .
it is seen just to the right of center , and below the plane , in figure [ fig : hadustne ] .
both the dust absorption and associated h@xmath0emission can be traced to @xmath174 kpc from the midplane .
though a few of the dust structures seen in figure [ fig : hadustne ] show evidence for loosely associated h@xmath0 emission , many of the dust features have no clear - cut association with the extraplanar dig .
several of the high-@xmath3 absorbing features seen in figure [ fig : hadustne ] seem to be attenuating the h@xmath0 emission .
this attenuation of the h@xmath0 emission by foreground dust structures is seen along much of the length of the galaxy .
the dust is seen to be much more highly clumped than the dig emission in our images .
when viewing our h@xmath0 images in detail , one is left with the impression that the dig emission may in fact be quite smooth , with a few filaments within this smooth layer .
all of this discussion regarding an absence of a direct correlation between the observed dust features and the dig traced by the h@xmath0emission does not imply the dig of ngc 891 is free of dust ( see [ subsec : filaments ] ) . in our own galaxy , there is increasing evidence for dust in the warm ionized medium ( lagache et al .
1999 , 2000 ; howk & savage 1999a ) .
however , our broadband absorption detection method would not be sensitive to the expected column densities and smoother structure of the dig in ngc 891 .
the deep images presented in this work have shown that h@xmath0 emission from the extraplanar dig in ngc 891 is not generally correlated with the dust - bearing clouds seen in absorption against the background stellar light , although there is a very strong statistical correlation between the presence or absence of these two phases of the ism in spiral galaxies ( paper ii ) .
our observations also show the highly - structured , or clumped , dusty clouds are prevalent only at heights @xmath175 kpc from the midplane of ngc 891 while h@xmath0 emission is seen to much larger @xmath3 distances .
many of these dust clouds have column densities in excess of @xmath176 @xmath19 , and enclose masses @xmath177 m@xmath6 .
we estimate the _ average _ densities of these high-@xmath3 clouds are roughly @xmath178 @xmath31 .
the lack of association between the h@xmath0 emission from high-@xmath3 dig and the highly - structured dusty clouds seen in our broadband images implies that the h@xmath0 emission and the dust absorption are not tracing the same interstellar material .
our images suggest an arrangement of the high-@xmath3 ism in ngc 891 where dusty dense clouds , visible through their absorption of the background stellar light , coexist with a more diffuse ionized medium .
we suggest that all of these aspects of the high-@xmath3 dust features can be understood in the context of a multiphase ism in the thick disk of ngc 891 , possibly in a quasi - stable configuration .
the dusty clouds seen far above the disks of spirals likely represent a dense phase of the ism at high-@xmath3 . in this case
they may be the result of thermal instabilities in the high-@xmath3 ism and may perhaps be in pressure equilibrium with either the dig or hot , x - ray emitting gas ( e.g. , wolfire et al .
1995a , b ) .
our images show the ionized gas is certainly more smoothly distributed than the material traced by the dust clouds .
this suggests the volume filling factor of the dense phase is lower than that of the dig , in accord with theoretical expectations ( e.g. , mckee & ostriker 1977 ) .
however , given the lack of association between the dig and the dense material seen in absorption , it is clear that some portion of the warm ionized medium is not associated with cool cores of neutral material , e.g. , as envisioned by mckee & ostriker ( 1977 ) . in paper
ii the multiphase nature of the ism in the thick disks of many nearby spiral galaxies was discussed in the context of a quasi - stable configuration of several thermal `` phases . ''
the calculations of the allowable equilibrium thermal states of a multiphase ism by wolfire et al .
( 1995a , b ) were used as guidelines for estimating the requisite conditions for such a configuration .
these authors calculate detailed thermal and ionization balance of a multiphase medium ( see also mckee & ostriker 1977 ; field et al .
1969 ) and give the results of these calculations for a range of physical conditions of the medium .
a minimum pressure requirement must be met to support a stable dense phase of the ism . in the wolfire
( 1995a ) models this pressure for high column density clouds ( @xmath179 @xmath19 ) with conditions appropriate for the disk of the milky way is @xmath180 k @xmath31 . larger column densities will tend to lower the required pressure , as will a decrease in the ambient radiation field or in the dust to gas ratio relative to those appropriate for the milky way disk .
thus the minimum pressure required to support the high-@xmath3 dust clouds in ngc 891 could be slightly lower than this value .
if we assume the observed structures are cylindrically symmetric , so that their depth along the line of sight is similar to their minor axis lengths , the column density estimates given in table [ table : derived ] suggest these structures have densities @xmath181 @xmath31 .
this is much greater than the expected density of the dig at _ any _ distance from the plane , further suggesting that the h@xmath0 emitting structures are tracing a separate medium than that traced by the dust .
this gross density estimate is consistent with the average densities predicted for dense clouds in the wolfire et al .
( 1995a ) models ( see also paper ii ) .
our rough calculations in paper ii suggested the average pressures in the dig or hot ionized medium traced by x - rays were sufficient to support a dense phase of the ism .
for example , the average pressure provided by the dig is @xmath180 k @xmath31 at @xmath182 kpc ( assuming the electron distribution of rand 1997 and @xmath183 k ) .
the x - ray emitting gas likely has a larger pressure ( bregman & houck 1997 ) . the paucity of clouds beyond @xmath184 kpc is understandable in this scenario : at heights in excess of @xmath185 kpc , the pressure of the ambient medium may be too low to support a dense phase of the ism . in this case dust may still exist at larger heights above the midplane , but our images are not sensitive to a diffuse component of dust .
we are only able to see those regions that are denser ( i.e. , more opaque ) than their surroundings . given the uncertainties in the distribution , temperature , filling factor , and ionization fraction of the ambient gas , we believe the pressure requirements for confinement of a dense medium are easily met in ngc 891 at heights @xmath186 kpc .
other theoretical and observational lines of evidence support our conclusion that the dust structures trace a dense medium at high-@xmath3 .
the three - dimensional numerical simulations of gaseous disk dynamics by de avillez ( 1999 ) suggest that the formation of dense interstellar clouds at great @xmath3-heights may be a relatively common by - product of the circulation of material between the disk and halo of a galaxy .
his simulations follow the temperature , density , and dynamics of parcels of gas in a model galactic disk undergoing star formation , and hence experiencing energy input by supernovae .
the combined effects of multiple supernovae produce a fountain - like flow in these models , with gas continually flowing up from the disk and raining down upon it from above .
these detailed simulations show cloud formation at high-@xmath3 . in these simulations
the bulk of the clouds are formed between @xmath187 kpc , and de avillez suggests the formation of dense clouds is much more efficient in regions compressed by shock waves . in particular
he finds the intersections of shocks at high-@xmath3 can serve to compress the gas , which then cools rapidly .
garca - burillo , combes , & neri ( 1999 ) have observed co emission from the edge - on galaxy ngc 4013 at high angular resolution ( @xmath188 ) .
they find about 10% of the emission in their maps comes from high-@xmath3 material .
our own wiyn images of this galaxy ( paper ii ) show a high-@xmath3 dust distribution similar to that seen in ngc 891 .
several of the high-@xmath3 dust features in our images , which have properties similar to those identified in ngc 891 ( howk & savage 1999c ; paper ii ) correspond to high-@xmath3 co emission in the maps of garca - burillo et al .
the direct observation of molecular emission associated with the extraplanar dust structures in ngc 4013 supports our suggestion in paper ii that the high-@xmath3 absorbing dust structures observed in several edge - on galaxies represent a dense , possibly molecular phase of the thick disk ism .
garca - burillo et al . (
1992 ) and sofue & nakai ( 1993 ) have also given evidence for a co - emitting component of the thick disk of ngc 891 , though at much lower resolution .
several candidate h@xmath10ii regions can be identified in our h@xmath0 images of ngc 891 at heights @xmath189 to 2.0 kpc from the midplane ( howk & savage 1999c ; howk & savage , in preparation ) .
associated with these candidate h@xmath10ii regions we find faint continuum sources ( @xmath190 to 23 mag ) .
if spectroscopically confirmed , the presence of h@xmath10iiregions at such high-@xmath3 suggests the underlying stars were formed in the thick interstellar disk of ngc 891 .
this would require the existence of molecular clouds far from the midplane , and hence lend implicit support to our conclusion that a dense phase of the ism is present at high-@xmath3 , visible as the absorbing dust structures seen in our images .
these nebulae will also be a useful probe of the physical conditions of the thick disk ism in ngc 891 .
for example , our wiyn images constrain the radii of these nebulae to be @xmath191 pc .
if ionized by b0 or earlier stars , this implies an ambient electron density of at least @xmath192 @xmath31 ( osterbrock 1989 ) .
thus the density of at least one phase of the interstellar thick disk of ngc 891 is quite dense , even at rather large heights above the plane ( @xmath184 kpc ) .
density estimates for the dig at these heights are factors of @xmath193 smaller than the lower density limit implied the h@xmath10ii regions , further suggesting large variations in the density of the interstellar thick disk .
we note that the dust filaments seen in our images likely can not be caused by a gaseous warp in this galaxy , and they are unlikely to be associated with a flare in the outer galaxy .
ngc 891 shows no evidence for a warp in its h@xmath10i distribution ( swaters et al .
1997 ; sancisi & allen 1979 ) , and the relative numbers of the dust features on either side of the midplane are very similar ( paper i ) , suggesting these features can not come from a warp along the line of sight .
while a flared gaseous layer can not be ruled out , it seems unlikely the observed dust features are associated with a flare . at least _ some _ of the features are clearly associated with regions of vigorous star formation in the disk .
the detection of co from similar extraplanar dust structures in ngc 4013 by garca - burillo et al.(1999 ) gives some information on the velocities of these structures .
in particular , one such dust feature lies immediately above the dynamical center of the galaxy , but is moving at @xmath194 km s@xmath195 relative to the systemic velocity of the galaxy .
this velocity is clearly incompatable with the location of the structure in a smoothly rotating flared gas layer .
several other features can be identified with velocities inconsistent with their presence in a galactic flare .
we do not believe a flared or warped gas layer is causing the dust structures we see in our images .
we believe a self - consistent picture of the high-@xmath3 ism in ngc 891 can be drawn in which the thick disk ism is made up of a multiphase medium , including a cold , dense phase , a warm ( neutral and ionized ) phase , and a hot phase .
the estimated masses of the warm neutral , warm ionized , and cold , dense phases of the ism above @xmath196 pc are all of order a few@xmath8 m@xmath6 .
the high-@xmath3 phases of the ism together represent @xmath197 of the total mass of the ism in ngc 891 ( swaters et al .
1997 ; dettmar 1990 ; garca - burillo et al . 1992 ; paper i ) .
the mass of the hot ism at high-@xmath3 is estimated to be @xmath198 m@xmath6 ( bregman & houck 1997 ) .
the presence of the dense phase , in particular , requires a relatively large pressure to large heights above the midplane . the available data on the hot and warm ionized media are consistent with the required pressures in the thick disk of ngc 891 .
such high pressure in this transition region between the thin disk and the more extended halo of the galaxy is required to support the weight of the higher-@xmath3 gas . however , the origin of this high-@xmath3 gas , and hence the origin of the high pressure confining the dense clouds , is likely ultimately tied to star formation in the disk of the galaxy . and
though the existence of interstellar material in the thick disks and halos of galaxies was once thought to be a relatively rare phenomenon , our previous work ( paper ii ) implies the transfer of matter from the thin interstellar disks to the more extended halos may be a common property of spiral galaxies .
furthermore , our images imply that this transfer does not destroy the dust grains .
this is important since it can therefore be argued that the more extended halos of galaxies may not be completely devoid of dust ( see , e.g. , zaritsky 1994 ) .
we have presented deep , high - resolution ground based images of the edge - on spiral ngc 891 obtained with the wiyn 3.5-m telescope .
these images were used to study the extensive distribution of extraplanar dust and ionized gas in this galaxy , building on the work presented in paper i. the major results of our work are as follows . 1 .
our deep broadband images show an extensive web of inhomogeneously distributed ( clumped ) dust far from the midplane of ngc 891 .
many of these structures were visible in the images presented in paper i. these new images presented here show individual absorbing dust structures to heights @xmath199 kpc from the midplane .
the number and apparent extinctions of these dust structures decreases significantly beyond @xmath200 kpc .
structures like those seen at @xmath201 kpc would be easily observable to heights @xmath202 kpc if they existed at such heights .
2 . we identify a small subset of the high-@xmath3 absorbing structures and derive their physical properties ( see tables [ table : observed ] and [ table : derived ] ) .
these structures are characterized by sizes of order 50 - 300 pc , with optical extinctions exceeding @xmath203 mag .
assuming galactic gas to dust relationships the extinction values imply the column density of associated gas in these structures is @xmath204 @xmath19 .
the implied mass of the associated gas is @xmath205
m@xmath6 per cloud , and the gravitational potential energies relative to the midplane are large , @xmath206 ergs .
the masses and average extinction values derived for these clouds are reminiscent of the galactic gmcs , and in general we conclude that the dust structures are tracing the dense cold neutral phase of the high-@xmath3 ism .
a few of the more spectacular absorbing dust structures are discussed in some detail .
these include a cometary structure of width @xmath207 pc and length @xmath208 pc at a height @xmath129 kpc from the midplane ; a 600-pc diameter supershell centered at @xmath174 kpc ; and a filament that may stretch from @xmath123 to @xmath209 kpc .
our deep , high - resolution h@xmath0 emission line images recover the high-@xmath3 dig studied in earlier works ( e.g. , rand et al .
1990 ; dettmar 1990 ; pildis et al . 1994 ) .
these data show a filamentary component of extraplanar ionized gas as well as an apparently more diffuse component . in general
the dig emission is much more smoothly distributed than the interstellar material traced by the dust absorption .
the northeast section of ngc 891 shows significantly more emission , both in and out of the plane , than the opposite side of the disk .
though filaments are indeed present in the distribution of the dig , they are much less numerous and prominent than the filamentary dust - bearing clouds .
the excess emission in the filaments is between 30% and 100% the brightness of the background dig emission . at low @xmath3 ( @xmath79 kpc )
much of the structure seen in the h@xmath0 images is the result of absorption by the same dust - bearing clouds visible in our broadband images .
we identify several ionized , filamentary structures and summarize their rough properties .
our catalog of structures includes a spectacular supershell of diameter @xmath210 pc that is centered @xmath155 pc above the plane of ngc 891 .
a direct comparison of the dig morphology with that of the complicated extraplanar dust distribution shows little correspondence between the two .
_ there are many regions where the h@xmath0 is extincted by the dust structures in our images , but very few that show a direct physical relationship between these two tracers of high-@xmath3 material . in the northeast section of the disk , a region showing evidence for very vigorous star formation , there are a few structures that show a loose association between the thick disk dig and dust .
the morphologies of these structures strongly suggest they are connected to the energetic effects of star formation in the underlying disk .
the high-@xmath3 extinction structures seen in our images represent a dense , likely cold , _ phase _ of the high-@xmath3multiphase ism .
this conclusion is based on the lower limits to the dusty cloud extinctions , the mass and density estimates , and the clear separation between the high-@xmath3 medium traced by the dust structures and that traced by emission from the dig .
secondary support for this conclusion is provided by the possible existence of star formation , evidenced by several h@xmath10ii regions visible in our h@xmath0 data , at large distances from the plane ; and the recent detection of co emission from similar dust structures in the galaxy ngc 4013 by garca - burilloet al .
the thermal pressure of the hot and warm ionized gas is sufficient to maintain a stable two - phase neutral medium ( warm+cold ) at large distances from the plane ( wolfire et al .
1995a ) , the densest portions of which we are seeing in our images .
the estimated masses of the cold dense , warm neutral , and warm ionized phases at @xmath211 pc are comparable and of the order a few@xmath212 m@xmath6 .
the presence of h@xmath10ii regions , their underlying young stellar populations , and the natal molecular clouds from which these stars have formed in the thick disk of ngc 891 provides an opportunity to study the physics of star formation in an extreme environment .
our broadband images give a direct visual means of studying the dense high-@xmath3 clouds in this galaxy with superb spatial resolution ( for studies of this phase of the ism ) .
it is a pleasure to thank the many people who have made the wiyn telescope a reality .
we extend special thanks to those who maintain and operate observatory .
jch recognizes support from a nasa graduate student researcher fellowship under grant number ngt-5 - 50121 .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
clcc b & 1997 august 30 & 1200 & 0.8 b & 1997 august 30 & 900 & 0.6 b & 1997 august 30 & 900 & 0.7 b & 1997 august 31 & 1200 & 0.8 b & 1997 august 31 & 1200 & 0.8 v & 1996 december 04 & 120 & 0.7 v & 1997 august 30 & 1200 & 0.7 v & 1997 august 30 & 1200 & 0.8 v & 1997 august 31 & 900 & 0.6 i & 1997 august 30 & 840 & 0.7 i & 1997 august 30 & 1500 & 0.7 i & 1997 august 31 & 1200 & 0.7
i & 1997 november 28 & 820 & 0.8 h@xmath0 & 1997 august 31 & 1200 & 0.7 h@xmath0 & 1997 august 31 & 300 & 0.8 h@xmath0 & 1997 november 26 & 1800 & 0.6 h@xmath0 & 1997 november 26 & 1800 & 0.6 h@xmath0 & 1997 november 26 & 1800 & 0.6 h@xmath0 & 1997 november 26 & 1800 & 0.7 h@xmath0 & 1997
november 29 & 1200 & 0.6
h@xmath0 & 1997 november 29 & 1800 & 0.8 h@xmath0 & 1997 november 29 & 1200 & 0.8 h@xmath0 & 1997 november 29 & 1200 & 0.7 off 1 & 1997 november 26 & 1800 & 0.6 off 1 & 1997 november 26 & 1800 & 0.6
off 1 & 1997 november 29 & 1200 & 0.8 off 2 & 1997 august 30 & 1200 & 0.9 off 2 & 1997 august 31 & 900 & 0.7 lcccccccl @xmath213 & 02 22 30.8 & + 42 18 55 & 870 & @xmath214 & 0.37 & 0.22 & 0.15 & flaring cloud @xmath172 & 02 22 31.2 & + 42 19 06 & 880 & @xmath215 & 0.50 & 0.35 & 0.19 & irr .
cloud @xmath70 & 02 22 26.9 & + 42 19 45 & 1750 & @xmath216 & 0.28 & 0.25 & 0.19 & irr .
cloud @xmath88 & 02 22 29.1 & @xmath217 20 23 & 1450 & @xmath218 & 0.77 & 0.56 & 0.30 & elongated cloud @xmath90 & 02 22 35.2 & @xmath217 20 29 & 1300 & @xmath219 & 0.36 & 0.32 & 0.17 & cometary cloud @xmath136 & 02 22 29.8 & + 42 20 57 & 1660 & @xmath220 & 0.32 & 0.21 & 0.16 & arc @xmath91 & 02 22 36.2 & + 42 20 47 & 1430 & @xmath221 & 0.27 & 0.24 & 0.13 & cometary cloud @xmath222 & 02 22 32.5 & + 42 21 14 & 730 & @xmath223 & 1.02 & 0.83 & 0.49 & column @xmath72 & 02 22 30.8 & + 42 21 40 & 1950 & @xmath224 & & & & irr .
cloud @xmath173 & 02 22 39.0 & + 42 22 34 & 750 & @xmath225 & 0.50 & 0.40 & 0.27 & vertical cone @xmath131 & 02 22 38.9 & + 42 22 39 & 730 & @xmath226 & 0.63 & 0.56 & 0.31 & shell wall @xmath132 & 02 22 39.6 & + 42 22 50 & 845 & @xmath227 & 0.40 & 0.35 & 0.23 & shell wall ccccccc @xmath228 & 0.22 & 0.30 & 0.50 & @xmath229 & @xmath230 & @xmath231 @xmath232 & 0.35 & 0.50 & 0.87 & @xmath233 & @xmath234 & @xmath235 @xmath70 & 0.25 & 0.35 & 0.58 & @xmath236 & @xmath237 & @xmath238 @xmath239 & 0.56 & 0.84 & 1.79 & @xmath240 & @xmath230 & @xmath241 @xmath90 & 0.32 & 0.45 & 0.78 & @xmath242 & @xmath230 & @xmath241 @xmath136 & 0.21 & 0.29 & 0.47 & @xmath229 & @xmath243 & @xmath244 @xmath91 & 0.24 & 0.33 & 0.55 & @xmath236 & @xmath245 & @xmath235 @xmath222 & 0.83 & 1.35 & & @xmath246 & @xmath247 & @xmath244 @xmath173 & 0.40 & 0.57 & 1.04 & @xmath248 & @xmath249 & @xmath250 @xmath131 & 0.56 & 0.84 & 1.78 & @xmath240 & @xmath251 & @xmath244 @xmath132 & 0.35 & 0.50 & 0.87 & @xmath233 & @xmath237 & @xmath241 @xmath158 & 02 22 31 & + 42 21 30 & 1570 & @xmath252 @xmath152 & 02 22 39 & + 42 22 02 & 1510 & @xmath253 @xmath254 & 02 22 35 & + 42 22 45 & 1100 & @xmath255 @xmath153 & 02 22 39 & + 42 22 40 & 820 & @xmath256 @xmath157 & 02 22 36 & + 42 22 58 & 660 & @xmath257 @xmath154 & 02 22 39 & + 42 22 52 & 510 & @xmath258 @xmath259 & 02 22 40 & + 42 23 02 & 1010 & @xmath260 | we present new , deep optical images ( bvi+h@xmath0 ) of the interstellar medium ( ism ) far above the plane of the edge - on sb galaxy ngc 891 .
these sub - arcsecond ( @xmath1 ) ground - based images give a direct visual view of two physically distinct `` phases '' of the thick interstellar disk of this galaxy . a dense , likely cold , phase of the thick disk ism is observed in our bvi images as highly - structured dust - bearing clouds viewed in absorption against the background stellar light of the galaxy .
these dusty structures are traceable to heights @xmath2 kpc from the midplane .
though our data are deep enough to identify such clouds at higher @xmath3 , very few highly - structured dust features are present at @xmath4 kpc . if the more prominent dust structures have galactic gas - to - dust ratios , then they have gas masses in excess of @xmath5 m@xmath6 , each having visual extinctions well in excess of unity .
a warm ionized phase of the high-@xmath3 ism is observed through its well - studied h@xmath0 emission .
our images of this medium , to date the highest - resolution observations of the warm ionized medium in ngc 891 , show that the diffuse ionized medium in this galaxy is relatively smoothly distributed with some filamentary structure superposed on this smooth background .
our data show very little correspondence between the h@xmath0-emitting material and the absorbing dust structures . _
these two phases of the multiphase high - z ism are physically distinct . _
the material traced by h@xmath0 emission is much more smoothly distributed than that traced by the absorbing dust clouds , and it is clear that the h@xmath0 emission is being heavily extincted in many places by the dense dust - bearing medium .
the high-@xmath3 dust clouds in ngc 891 are a major component of the multiphase halo medium .
the mass found in the cold dense , warm neutral , and warm ionized phases are for @xmath7 kpc are comparable and of order a few@xmath8 m@xmath6 .
our discussion of the physics of the multiphase medium at high-@xmath3 relies heavily on the theoretical work of wolfire et al . , and we suggest the dense phase of the thick disk ism exists to @xmath9 kpc because the thermal pressures of the warm and hot phases in this region are sufficient for its maintenance .
our h@xmath0 observations show evidence for several discrete h@xmath10ii regions at large distances from the midplane ( latexmath:[$0.6 \la the thick disk of ngc 891 suggests that on - going star formation may be present in some of the dense , high-@xmath3 clouds visible in our images . |
pulsar - like compact stars as the densest observable objects in the universe , are not only important in understanding diverse phenomena in astronomy , but also significant in fundamental physics .
the answer to the question whether they are neutron or quark stars would have profound implications on the physics of condensed matter , the nature of strong interaction as well as the qcd phase diagram.@xcite investigating the state of condensed matter inside pulsar - like compact stars by first principles is extremely difficult .
however , it is meaningful to consider some phenomenological models to understand observational properties of pulsar - like compact stars .
we conjectured that pulsar - like compact stars could be `` quark - cluster stars '' which are composed of quark - clusters with almost equal numbers of up , down and strange quarks , and could be in a solid state if the kinetic energy of quark clusters ( order of @xmath2 , with temperature @xmath3 ) is lower than the interaction energy between the clusters . from a physical point of view , we knew that strong coupling between quarks is still very strong even in the hot quark - gluon plasma,@xcite so in cold quark matter at realistic baryon densities ( @xmath0 ) , quarks would be expected to be also coupled strongly and to condensate in position space , rather than only in momentum space , to form quark clusters . from an astronomical point of view , the stiff equation of state of quark - cluster matter could naturally explain the @xmath4 solar mass neutron star,@xcite and the solid state star could provide gravitational , phase transition and elastic energies to understand bursts of anomalous x - ray pulsars , soft gamma - ray repeaters and even the shallow decay of gamma - ray burst afterglow.@xcite more observational hints come from the surface , since the surface of quark - cluster stars could be bare and is chromatic confined , which is quite different from that of conventional neutron stars .
one important surface hint is the drifting @xcite and bidrifting @xcite subpulses .
the vacuum gap model can explain drifting subpulses naturally,@xcite but for conventional neutron star , the binding energy is too low to form such a vacuum gap.@xcite however , for quark - cluster stars , the surface is chromatic confined and the potential barrier built by the electric field in the vacuum gap above the polar cap can prevent electrons from streaming into the magnetosphere , therefore a vacuum gap can be formed.@xcite other important surface hints are associated with the bare surface of quark - cluster stars . as discussed in xu ( 2002 ) ,
a quark star could be born to be bare .
such a bare quark - cluster star can not only provide a clean fireball for supernovae and gamma - ray bursts , but also explain the thermal featureless spectrum of neutron stars.@xcite what is more , quark - cluster stars are enveloped in thin electron layers and vortex hydrodynamical oscillations may be invoked in those electron seas , which could explain the absorption features detected in the spectra of some pulsar - like compact stars .
@xcite the outline of the paper is as following . in
@xmath5 , we briefly review the quark - cluster star model . in @xmath6 ,
we summarize different hints from the surface of quark - cluster stars . in
@xmath7 , conclusions are presented .
although pulsars have been discovery for more than @xmath8 years , the inner structure of pulsar - like compact stars is still in controversy .
conventional neutron stars and quark stars are two main types of models .
while conventional neutron stars are composed mainly of hadrons , the equation of supra - nuclear matter should be understood essentially in the level of quarks .
previously it is supposed that a perturbative strong interaction inside quark stars could make quarks to be almost free if the coupling is weak , whereas for the conventional neutron stars , the highly non - perturbative strong interaction makes quarks grouped into neutrons .
however , results of relativistic heavy ion collision experiments have shown that the interaction between quarks is very strong in hot quark - gluon plasma .
@xcite therefore , in cold quark matter at realistic baryon densities ( @xmath0 ) , quarks could also be coupled strongly and the strong interaction could make quarks to condensate in position space to form quark clusters .
we may then expect that quark matter inside compact stars could be in a `` quark clustering phase '' , where the energy scale would be high enough to allow the restoration of 3-light - favor symmetry , but may not be high enough to make the quarks really deconfined . for a realistic pulsar - like compact star whose temperature is low
, the kinetic energy of quark clusters should be much lower than the interaction energy between the clusters , and the stars should in a solid state .
the peculiar inner structure of quark - cluster stars which are very different from conventional neutron stars would result in special global and surface properties .
these properties are not only instructive for us to understand astrophysical phenomena , but also helpful to identify quark - cluster stars . globally speaking ,
quark - cluster stars have stiff equation of state with which a compact star could have a maximum mass even to be @xmath9 solar mass .
@xcite the heat capacity of quark - cluster matter is negligible , therefore the cooling behavior would be different from conventional neutron stars,@xcite and the missing compact star of sn1987a could be a quark - cluster star.@xcite gravitational and elastic energies of solid quark - cluster stars could explain bursts of anomalous x - ray pulsars and soft gamma - ray repeaters;@xcite the latent heat of phase transition of quark - cluster matter from liquid to solid could explain the shallow decay phase of gamma - ray bursts.@xcite as for the surface of quark - cluster stars , it could be bare and is chromatic confined , and would be directly associated with the broadband radiation of pulsar - like compact stars . in the next section , we may discuss the details about hints from the surface .
in addition , the conjectured quark - cluster matter with 3-flavor symmetry may also have profound implication to the research of cosmic rays .
strangelets composed by quark clusters would exist and propagate in interstellar space if the gravitational energy is not significant to form stable quark clusters with strangeness .
it is well know that the most stable hadrons are nucleons ( protons and neutrons ) , with mass of @xmath10 mev .
the simplest particle in 3-flavor symmetry is @xmath11 ( @xmath12 mev ) , but the interaction between @xmath13 s is attractive.@xcite therefore the simplest quark clusters in compact stars would be @xmath13-@xmath13 dibaryon , the so - called @xmath14-particle.@xcite a strangelet with mass per baryon @xmath15 mev ( i.e. , binding energy per baryon @xmath16 mev ) could be stable in cosmic rays , and would decay finally into nucleons when collision - induced decrease of baryon number make it unstable due to the increase of surface energy . when a stable strangelet with probably a few ( or decades of ) quark clusters bombard the atmosphere of the earth , its fragmented nuggets may decay quickly into @xmath13 s by strong interaction and further into nucleons by weak interaction .
quark - cluster stars could be bare , since proto - quark - cluster stars may be bare owing to strong mass ejection and high temperature,@xcite and accretion of supernova fallback and the debris disk are unlikely to cover the star with a crust.@xcite for bare quark - cluster stars , surface clusters are confined by strong colour interaction while the huge potential barrier built by the electric field can usually prevent electrons from streaming freely into the magnetosphere.@xcite several observational facts have indicated that such a bare , self - confined surface could be necessary for us to understand different phenomena .
these hints from the surface of pulsar - like compact stars may help us identify quark - cluster stars too .
drifting subpulses were first explained by ruderman & sutherland ( 1975 ) . in the seminal paper , a vacuum gap
was first suggested above the polar cap of a pulsar .
the sparks produced by the inner - gap breakdown result in the subpulses , and the @xmath17 drift which is due to the lack of charges within the gap , causes the observed drifting features .
however , the above model encounters the so - called `` binding energy problem '' .
calculations have shown that the binding energy of fe at the neutron star surface is @xmath18 kev,@xcite which is not sufficient to reproduce the vacuum gap .
one way to solve this binding energy problem is the a partially screened inner gap model,@xcite but a simple way would be in the bare quark - cluster star model.@xcite in yu & xu ( 2011 ) , the magnetospheric activity of bare quark - cluster star was investigated in quantitative details .
since quarks on the surface are confined by strong color interaction , the binding energy of quarks can be considered as infinity compared to electromagnetic interaction . as for electrons on the surface ,
on one hand the potential barrier of the vacuum gap prevents electrons from streaming into the magnetosphere , on the other hand the total energy of electrons on the fermi surface is none zero .
therefore , the binding energy of electrons is determined by the difference between the height of the potential barrier in the vacuum gap and the total energy of electrons .
calculations have shown that the huge potential barrier built by the electric field in the vacuum gap above the polar cap can usually prevent electrons from streaming into the magnetosphere , unless the electric potential of a pulsar is sufficiently lower than that at the infinite interstellar medium .
figure 1 shows a potential barrier of electrons on the stellar surface of a typical pulsar.@xcite we conclude that in the bare quark - cluster star model , both positive and negative particles on the surface are usually bound strongly enough to form a vacuum gap above its polar cap , and the drifting subpulses can be understood naturally .
it is well known that the radiation fireballs of gamma - ray bursts and supernovae as a whole move towards the observer with a high lorentz factor.@xcite the bulk lorentz factor of the ultrarelativistic fireball of grbs is estimated to be order of @xmath19.@xcite for such an ultrarelativistic fireball , the total mass of baryons can not be too high , otherwise baryons would carry out too much energy of the central engine energy , and this is the so - called `` baryon contamination '' . for conventional neutron stars as the central engine , the number of baryons loaded with the fireball is unlikely to be small , since neutron stars are gravity - confined and the luminosity of fireball is extremely high .
however , the baryon contamination problem can be solved naturally if the central compact objects are quark - cluster stars .
the bare and chromatic confined surface of quark - cluster stars separate baryonic matter from the photon and lepton dominated fireball .
inside the star , baryons are in quark - cluster phase and can not escape due to strong color interaction , but @xmath20-pairs , photons , neutrino pairs and magnetic fields can escape from the surface .
thus , the surface of quark - cluster stars automatically generates a low baryon condition for grbs as well as supernovae.@xcite a nascent quark - cluster star born in the center of grb or supernova would radiate planck - like thermal emission due to its ultrahigh surface temperature,@xcite and the photon luminosity is not constrained by the eddington limit since the surface of quark - cluster stars could be bare and is chromatic confined . therefore , the strong radiation pressure caused by enormous thermal emissions from quark - cluster stars might play an important role in promoting core - collapse supernovae.@xcite calculations have shown that the radiation pressure due to such strong thermal emission can push the overlying mantle away through photon - electron scattering with energy as much as @xmath21 ergs .
such photon - driven mechanism in core - collapse supernovae by forming a quark - cluster star inside the collapsing core is promising to alleviate the current difficulty in core - collapse supernovae .
an illustration of the photon - driven mechanism is shown in figure 2.@xcite the nascent quark star could be in a fluid state , but according to xu & liang ( 2009 ) , a phase transition of the quark star from liquid to solid could occur about @xmath22 s later after its birth . since the quark - cluster star in phase transition would emit thermal radiation at almost constant temperature and a solid quark star
would cool very fast due to its extremely low heat capacity,@xcite the latent heat of this phase transition not only provides a long - term steady central engine , but also shows an abrupt cutoff of energy injection when the phase transition ends .
such kind of energy injection may explain the long - lived plateau followed by an abrupt falloff observed in some afterglows.@xcite a schematic cooling behavior of a new - born quark star is shown in figure 3.@xcite after the central quark - cluster star is solidified , star - quake would occur when strains accumulate to a critical value .
the star - quake induced energy ejection would then results in the observed x - ray flares of bursts.@xcite in conventional neutron star / crusted quark stars models , an atmosphere exists above the surface of central star
. the spectrum determined by the radiative transfer in atmosphere should differ substantially from planck - like one , depending on the chemical composition , magnetic field , etc.@xcite many calculations ( e.g. , romani ( 1987 ) ; zavlin , pavlov , & shibanov 1996 ) show that spectral lines should be detectable with the spectrographs on board chandra and xmm - newton .
one expects to know the chemical composition , magnetic field of the atmosphere through such observations , and eventually to determine the mass of the compact star according to the redshift and pressure broadening of an absorption spectrum .
however , up to now , no atomic line has been observed with certainty .
although this discrepancy could be explained for some of the sources by assuming a low - z element ( hydrogen or helium ) photosphere or by adjusting the magnetic field , we address that the featureless thermal spectrum can be understood naturally in the bare quark - cluster star scenario.@xcite the best example is the detailed spectral analysis of the combined x - ray and optical data of rx j1856.5 - 3754 , which have shown that no atmosphere model can fit well the data.@xcite nevertheless , the best absorption features were detected for the central compact object ( cco ) 1e 1207.4 - 5209 in the center of supernova remnant pks 1209 - 51/52 at @xmath23 kev and @xmath24 kev.@xcite although initially these features were thought to be due to atomic transitions of ionized helium , soon thereafter it was noted that these lines might be of electron - cyclotron origin.@xcite recent timing results seems to agree with the electron - cyclotron interpretation.@xcite but this simple single particle approximation might not be reliable due to high electron density on strange stars , and xu et al .
( 2012 ) investigated the global motion of the electron seas on the magnetized surfaces .
it is found that hydrodynamic surface fluctuations of the electron sea would be greatly affected by the magnetic field , and an analysis shows that the seas may undergo hydrocyclotron oscillations whose frequencies are given by @xmath25 $ ] , where @xmath26 and @xmath27 is the cyclotron frequency .
the fact that the absorption feature of 1e 1207.4 - 5209 at @xmath28 kev is not much stronger than that at @xmath29 kev could be understood in this hydrocyclotron oscillations model , because these two lines with @xmath30 and @xmath31 could have nearly equal intensity , while the strength of the first harmonic is much smaller than that of the fundamental in the electron - cyclotron model . besides the absorption in 1e 1207.4 - 5209
, the detected lines around @xmath32 kev in the burst spectrum of sgr 1806 - 20 and those in other dead pulsars ( e.g. , radio quiet compact objects ) would also be of hydrocyclotron origin.@xcite
the nature of pulsar - like compact stars is closely related to the physics of condensed matter and the elementary strong interaction . according to the results of relativistic heavy ion collision experiments and various properties of pulsar - like compact stars , we conjecture that quarks could be coupled strongly and condensate in position space to form quark clusters in cold dense matter at realistic baryon densities ( @xmath0 ) . under this scenario , pulsars could be `` quark - cluster stars '' which are composed of quark - clusters with almost equal numbers of up , down and strange quarks , and could be in a solid state during almost all of their lives .
both global and surface properties of quark - cluster stars are different from conventional neutron stars .
globally speaking , the equation of state of quark - cluster stars is stiffer to support higher mass , while the heat capacity of quark - cluster stars is low , and solid quark - cluster stars could provide gravitational , phase transition and elastic energies to various bursts phenomena . however ,
more direct hints come from the surface of quark - cluster stars . drifting subpulses phenomena
may comfortably be understood with the vacuum gap above the polar cap , and we have noted that such vacuum gap can easily form above a quark - cluster star surface , while the binding energy of conventional neutron stars is too low to form vacuum gap .
the bare and chromatic confined surface of quark - cluster star could naturally overcome the baryon contamination problem in gamma - ray burst and supernova fireball , and the strong thermal radiation from quark - cluster star could promote core - collapse supernovae .
the non - atomic thermal spectra of some pulsar - like compact stars could also be explained by the bare surface of quark - cluster star , and we have shown that the hydro - cyclotron oscillation of the electron sea could reproduce observational frequencies of the absorption features .
all these hints indicate that the surface of pulsar - like compact star should be bare and self - confined , which is different from conventional neutron stars .
together with stiff equation of state , we think that the quark - cluster star model is reasonable to describe pulsar - like stars .
we also expect future astrophysical observations would identify quark - cluster stars along this line .
finally , it is worth noting that a high binding ( @xmath33 mev per baryon ) between quark clusters not only results in the large maximum mass of quark - cluster stars,@xcite but also favors one to conjecture that bulk quark - cluster matter in 3-flavor symmetry could be absolutely stable . compared with nuclear matter ( binding energy of @xmath34 mev per baryon )
, strange quark - cluster matter is bound strongly , indicating that the color interaction is still very strong for cold matter at a few nuclear densities .
no - detection of stellar - mass black holes with mass @xmath35 may hint that the maximum mass for quark - cluster stars is really large .
we would like to thank useful discussions at our pulsar group of pku .
this work is supported by the national natural science foundation of china ( grant nos .
10935001 , 10973002 ) , the national basic research program of china ( grant nos .
2009cb824800 , 2012cb821800 ) , and the john templeton foundation . | the matter inside pulsar - like compact stars could be in a quark - cluster phase since in cold dense matter at a few nuclear densities ( @xmath0 ) , quarks could be coupled still very strongly and condensate in position space to form quark clusters .
quark - cluster stars are chromatically confined and could initially be bare , therefore the surface properties of quark - cluster stars would be quite different from that of conventional neutron stars .
some facts indicate that a bare and self - confined surface of pulsar - like compact stars might be necessary in order to naturally understand different observational manifestations . on one hand
, as for explaining the drifting sub - pulse phenomena , the binding energy of particles on pulsar surface should be high enough to produce vacuum gaps , which indicates that pulsar s surface might be strongly self - confined .
on the other hand , a bare surface of quark - cluster star can overcome the baryon contamination problem of @xmath1-ray burst as well as promote a successful core - collapse supernova .
what is more , the non - atomic thermal spectra of dead pulsars may indicate also a bare surface without atmosphere , and the hydro - cyclotron oscillation of the electron sea above the quark - cluster star surface could be responsible for those absorption features detected .
these hints could reflect the property of compact star s surface and possibly the state of condensed matter inside , and then might finally result in identifying quark - cluster stars . |
planets form in circumstellar disks . in the standard scenario ,
formation of earth - size planets or planetary cores occurs through coagulation of small dust particles ( e.g. , weidenschilling & cuzzi 1993 ) .
thus , the total amount , radial density distribution , and size distribution of dust particles in the disks are the important initial conditions of planet formation . in many of the previous studies of planet formation ,
the dust - gas ratio of circumstellar disks is assumed to be similar to the solar value ( @xmath6 ) , and dust particles are considered to be well mixed with gas , i.e. , the dust - gas ratio is constant throughout the disk ( hayashi , nakazawa , & nakagawa 1985 ) .
however , the motion of dust particles is different from that of gas .
dust particles have a radial motion which is induced by the gas drag force .
adachi , hayashi , & nakazawa ( 1976 ) and weidenschilling ( 1977 ) studied the motion of particles in gas disks . due to radially outward pressure gradient force
, the rotation velocity of the gas is generally slower than particles on nearly keplerian circular orbits .
consequently , the gas drag force takes angular momentum from particles , resulting in their inward migration ( whipple 1972 ) . at a few au from the central star , the orbital decay time
is estimated to be @xmath7 for @xmath8 particles and @xmath9 for @xmath10 particles .
these times are much shorter than the life time of gas disks ( @xmath11 ) .
thus , dust component may evolve much faster than the gas in protostellar disks , in which case , the dust - gas ratio would decline .
it appears that the initial conditions of planet formation need not to be protostellar disks in which dust particles are well mixed with the gas and the dust - gas ratio is constant . in this paper
, we start a series of studies to determine the initial distribution of dust particles in the context of planet formation . here , we study the radial migration of dust particles in gas disks .
we focus our attention on particles smaller than @xmath12 in order to consider evolution of dust disks before the initiation of planetesimal formation .
the migration of these small particles establishes the initial density distribution of the dust component and sets the stage for planetesimal formation . in this phase of disk evolution ,
the gas component is considered to be turbulent and is accreting onto the star . the discussion in the above paragraph is based on the study of particle migration under the assumption that particles have completely sedimented on the midplane of laminar disks .
however , in turbulent disks , the sedimentation of particles is prevented , because particles are stirred up to high altitude from the midplane by turbulent gas motion . in such disks ,
particles are distributed in the vertical direction , and the radial motion of particles depends on the distance from the midplane .
thus , there is a need for studying the radial migration of particles which reside above the midplane .
there are two important factors causing the vertical variation in the migration velocity of particles .
the first is the variation of rotation velocity of the gas in the vertical direction .
as mentioned above , the gas rotation differs from the keplerian rotation because of the gas pressure gradient . as the gas density decreases with the distance from the midplane , the radial pressure gradient varies and under some circumstances even changes its sign . while at the midplane the pressure gradient force is outward
, it is inward near the disk surface because the disk thickness increases with the radius .
the gas drag force on particles also varies with the height , resulting in a variation of particle migration velocities . in [ sec : vrd ] , we will see that particles at high altitude , where the pressure gradient force is inward , flow outward .
the second factor is the variation in the radial gas flow .
small particles ( @xmath0 ) are well coupled to the gas , so that they migrate with the gas flow , as discussed in
[ sec : vrd ] below . that is
if the gas is accreted onto the star , the small particles would also be accreted , while if the gas flows outward , particle flow would also be outward .
in accreting protostellar disks , the gas does not always flow radially inward .
radial velocity of the gas varies with the distance from the midplane .
viscous stress can cause gas outflow near the midplane ( urpin 1984 ; kley & lin 1992 ) .
ryczka , bodenheimer , & bell ( 1994 ) showed that the outflow occurs if the radial pressure gradient at the midplane is steep enough .
the density ( or pressure ) gradient at the midplane is steeper than the average spatial ( or surface ) density gradient , e.g. , if the surface density varies as @xmath13 and the disk thickness varies as @xmath14 , the midplane density has a steeper variation corresponding to @xmath15 .
viscous diffusion at the midplane causes the outflow of the gas to reduce the steep density gradient . on the other hand , at high altitude from the midplane , the density gradient is shallow and the viscous torque causes the usual inward flow .
thus , in accretion disks , the outward flow at the midplane is sandwiched by the inflow at the disk surfaces ( see [ sec : vrg ] for details ) .
because small particles are carried by the gas flow , they are expected to flow outward near the midplane , and inward at higher altitude .
thus , if most of dust particles concentrate in the region where they migrate outward , i.e. , near the disk surface or around the midplane , the dust component of the disk would accumulate in total mass and expand in size .
in reality , the vertical spatial distribution of particles is determined by the degree of sedimentation , and their size distribution is regulated by the coagulation , condensation , and sublimation processes . in this paper , we calculate the vertical distribution of particles in order to identify the mass flux of the dust particles as functions of their size and their distance from the midplane and their host stars . the results of the present calculation will be used to study the long term dynamical evolution of the dust component of the disk . for the present task , we do not consider the effects of gas depletion , particles size evolution or their feedback influence on the flow velocity of the gas .
these effects will be considered in a future investigation .
there are several works on the particle migration in turbulent disks .
stepinski & valageas ( 1997 ) derived the vertically averaged velocity of migration for two limiting cases .
one model is constructed for small particles which are well mixed with the gas and the other is for large particles which are totally sedimented to the midplane .
the velocity for intermediate sized particles is calculated by the interpolation of these limiting cases .
supulver & lin ( 2000 ) investigated the evolution of particle orbits in turbulent disks numerically .
this paper gives analytical expression of the particle velocity for any size of particles , which are estimated from interpolation by stepinski & valageas ( 1997 ) .
several studies showed that vortices or turbulent eddies can trap particles within some size range ( barge & sommeria 1995 ; cuzzi , dobrovolskis , & hogan 1996 ; tanga et al .
1996 ; klahr & henning 1997 ; cuzzi et al .
2001 , see however hodgson & brandenburg 1998 ) .
particle trapping by eddies may be important for local collection of size - sorted particles . in this paper , turbulent eddies
are treated just as a source of the particle diffusion and of the gas viscosity .
the effect of particle trapping on their coagulation is not discussed here .
the plan of the paper is as follows . in 2 , the vertical variation of gas flow is derived . in 3 ,
we describe how the radial velocity of dust particles varies in the vertical direction , and then calculate the net radial migration velocity . in 4 ,
we discuss the steady distribution of particles assuming no particle growth .
we show that the size fractionation of particles occurs as a result of radial migration , and that the distribution of particles differs from that of the gas .
gas disks rotate with a slightly different velocity from the keplerian velocity .
the rotation velocity is determined by the balance between the stellar gravity , the centrifugal force , and the gas pressure gradient . in cylindrical coordinates @xmath16 ,
the balance of the forces described by the momentum equation in the @xmath17-direction is given by @xmath18 where @xmath19 is the angular rotation velocity of the gas , @xmath20 is the gravitational constant , @xmath21 is the stellar mass , @xmath22 is the gas density , and @xmath23 is the gas pressure . the density distribution in the vertical direction is determined by the balance between the @xmath24-components of the stellar gravity and the pressure gradient , which is written as @xmath25 we assume that the disk is isothermal in the vertical direction . neglecting @xmath26 and higher order terms , the vertical density distribution is obtained as @xmath27 the disk scale height @xmath28 is given by @xmath29 where @xmath30 is the isothermal sound speed , and @xmath31 is the keplerian angular velocity at the midplane .
the radial variations of physical quantities are assumed to have power law forms , such that @xmath32 where the subscript `` 0 '' means the values at @xmath33 , @xmath34 is the radius in the unit of au , and power law indices @xmath35 and @xmath36 are usually negative .
the surface density is @xmath37 where @xmath38 . to derive the rotation law of the gas
, we retain @xmath26 and lower order terms in equation ( [ eq : gaseq_r ] ) , then obtain @xmath39 \ .
\label{eq : omegag}\ ] ] the difference of the gas rotation from the kepler rotation is order of @xmath40 , and the gas at high altitude rotates slower than the gas at the midplane . in viscous disks ,
angular momentum is transfered from the inner part to the outer part of the disk under the action of viscous stress .
the azimuthal component of the momentum equation implies @xmath41 \
, \label{eq : gasang}\end{aligned}\ ] ] where @xmath42 and @xmath43 are the @xmath17- and @xmath24-components of the gas velocity , respectively , and @xmath44 is the kinetic viscosity .
the right hand side of equation ( [ eq : gasang ] ) represents the torque exerted on an annulus with unit width in the @xmath17- and @xmath24-directions .
the left hand side is the variation of the angular momentum of the annulus accompanying with the motion .
we assume that the angular velocity is always adjusted to the equilibrium ( eq .
[ [ eq : omegag ] ] ) on a dynamical time scale .
since molecular viscosity is too ineffective , it is customary to assume that angular momentum is primarily transported by the reynolds stress induced by the turbulent motion of the gas .
following conventional practice , we adopt an _ ad hoc _ but simple to use @xmath45 prescription for turbulent viscosity ( shakura & sunyaev 1973 ) , such that @xmath46 the mass conservation of the gas is @xmath47 in accretion disks considered here , the second term is the same order of the third term .
then , we see that @xmath48 . besides , from equation ( [ eq : omegag ] ) we see @xmath49 .
thus , @xmath50 is factor @xmath40 smaller than @xmath51 and can be neglected in equation ( [ eq : gasang ] ) .
using power law expressions ( [ eq : powerlaw1 ] ) and ( [ eq : nu ] ) and retaining terms of order @xmath40 , the radial velocity is reduced to @xmath52 in the following discussions , we adopt these values as being representative of a typical model : @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , and @xmath58 .
these values correspond to a minimum mass solar nebula whose surface density distribution has a power law form with index @xmath59 and has a gas mass @xmath60 inside 100au .
figure [ fig : vrg ] shows the radial velocity of the gas .
the radial velocity is outward ( positive ) near the midplane , while it is inward ( negative ) above the height @xmath61 . in the standard model ( @xmath57 ) ,
the radial velocity is independent of the distance from the star , as seen from equation ( [ eq : vrg ] ) . at the midplane ,
the radial gradient of the gas density is so steep ( @xmath56 ) that the gas receives more torque from the inner disk than the torque it exerts on the outer disk .
this radial dependence means that viscous stress acts to move the gas outward in such a steep density gradient .
the condition for the outward gas flow at the midplane is @xmath62 which is satisfied in many accretion disk models .
the rotation of the gas is slower at higher altitude as seen in equation ( [ eq : omegag ] ) .
thus , the gas at the midplane loses its angular momentum through the action of viscous stress it has with the gas at higher altitude .
however , in the standard model , this angular momentum loss is smaller than the angular momentum gain through the viscous stress in the @xmath17-direction .
the outflow zone around the midplane shrinks as the radial density gradient is reduced ( @xmath63 ) , and when the condition ( [ eq : condoutflow ] ) is violated , gas at all altitudes flows inward .
the radial density gradient is shallower at higher altitude , because the disk scale height increases with the radius .
thus , at high altitude , the gas loses more angular momentum through exerting torque on the outer disk than it is receiving from the inner disk . for such gas
, the presence of viscous torque induces it to flow towards the star .
the net radial velocity of the gas is given by averaging in the vertical direction : @xmath64 the net accretion of the gas toward the star occurs if @xmath65 we may consider two models in which a steady state is achieved .
one model is the case in which the net radial velocity is zero , i.e. , @xmath66 .
for example , if @xmath67 and @xmath57 , there is no net accretion of the disk .
a more realistic model is the case in which the mass flux of the gas is constant with the radius , i.e. , @xmath68 is constant .
we take this as our standard model , i.e. , @xmath69 and @xmath57 .
this surface density profile is expected to arise after the disk has undergone a period of initial viscous evolution .
the azimuthal velocities of dust particles are different from that of the gas .
the resulting gas drag force transfers angular momentum between the particles and the gas and moves particles in the radial direction .
if there is no gas drag force , particles would orbit with the keplerian angular velocity , which is approximated as @xmath70 we express the deviation of the angular velocity of the gas from the keplerian angular velocity as @xmath71 from equation ( [ eq : gaseq_r ] ) it is seen that + @xmath72 is the ratio of the gas pressure gradient to the stellar gravity in the radial direction . from equations ( [ eq : omegag ] ) , ( [ eq : omegak ] ) , and ( [ eq : veldiff ] ) , @xmath73 is written within the order of @xmath40 as , @xmath74 note that the pressure gradient force is outward ( @xmath73 is positive ) around the midplane , while it is inward ( @xmath73 is negative ) where @xmath75 . in the standard model , @xmath73 changes sign at @xmath76 .
dust particles near the midplane rotate faster than the gas , and particles at high altitude rotate slower than the gas .
the equations of motion of a particle are @xmath77 @xmath78 where @xmath79 and @xmath80 are the @xmath17- and @xmath81- components of the velocity , respectively , with the subscripts `` @xmath82 '' and `` @xmath83 '' distinguishing gas and dust , and @xmath84 is the keplerian velocity at the midplane .
the gas drag force is expressed through the non - dimensional stopping time @xmath85 ( normalized by the kepler time at the midplane ) .
we do not solve the equation of motion in the @xmath24-direction .
instead , we simply assume @xmath86 .
when an equilibrium in the vertical dust distribution is achieved , the dust sedimentation to the midplane is balanced by the diffusion of particles , as discussed below in [ sec : vdist ] .
the vertical velocity of a particle is zero when time averaged .
the mean free path of gas molecules is larger than @xmath10 for @xmath87 in our models ( nakagawa , sekiya , & hayashi 1986 ) .
in this paper , we consider particles smaller than @xmath10 and use epstein s gas drag law .
then , the non - dimensional stopping time @xmath85 , as given by takeuchi & artymowicz ( 2001 ) , is @xmath88 where @xmath89 is the particle internal density , @xmath90 is the particle radius , and the mean thermal velocity is @xmath91 .
is defined to be @xmath92 times the mean thermal velocity .
this definition causes a factor @xmath92 difference between equation ( [ eq : stoptime ] ) and their equation ( 10 ) .
] we take the particle internal density as @xmath93 . for particles smaller than @xmath10 , the non - dimensional stopping
time is much smaller than unity through most of the disk .
figure [ fig : stoptime ] shows the radii of particles whose stopping time is unity . at the midplane ( see the solid line ) , the non - dimensional stopping time is smaller than unity for particles of size @xmath94 at @xmath95 . at @xmath96 ( the dashed line ) , although the gas density is lower than at the midplane and the stopping time is longer , particles of size @xmath97 still have the non - dimensional stopping time smaller than unity .
these particles are well coupled to the gas and their angular velocity is similar to the gas angular velocity ( [ eq : omegag ] ) . only particles with @xmath98 which are located at high altitude and @xmath99
become decoupled from the gas .
we assume that motions of both the gas and the particles are close to keplerian , i.e. , @xmath100 , and that @xmath101 .
then , from equation ( [ eq : motion_th ] ) , we have @xmath102 using equation ( [ eq : veldiff ] ) and neglecting terms of order @xmath103 and higher , equation ( [ eq : motion_r ] ) is reduced to @xmath104 the left hand side is order of @xmath105 and is neglected if @xmath106 .
substituting equation ( [ eq : delta_vt ] ) into equation ( [ eq : motion_r_approx ] ) , we find the radial velocity of the particle to be @xmath107 in the above derivation of the particle s radial velocity , we assume that in the @xmath24-direction the particle sedimentation is balanced by the turbulent diffusion , but that in the @xmath17-direction the turbulent effects are neglected . for particles well coupled with the gas ( @xmath108 ) , the radial velocity reduces to @xmath109 where @xmath110 is the relative velocity from the gas . using equations ( [ eq : powerlaw1 ] ) , ( [ eq : eta ] ) , and ( [ eq : stoptime ] ) , @xmath111 where @xmath112 is the non - dimensional stopping time at the midplane .
the radial drift velocity increases exponentially with the height .
because the particles are strongly coupled with the gas , the gas drag force suppresses their drift velocity . at the midplane where the gas density is highest ,
the suppression of the drift is most effective .
the drift velocity at the midplane is @xmath113 which agrees with the derivations of adachi et al .
( 1976 ) and weidenschilling ( 1977 ) . for large particles decoupled from the gas ( @xmath114 ) , the radial motion of the gas does not affect the particles velocity , so that @xmath115 . in this paper
, we do not consider such large particles .
figure [ fig : vrd ] shows the radial velocity @xmath116 of particles of @xmath117 , @xmath8 , and @xmath118 at @xmath119 in the standard model disk . near the midplane , the drift velocity @xmath120 ( relative to the gas velocity shown by the dotted line ) of @xmath3 particles is small .
the particles move outward almost together with the gas .
as the altitude increases and the gas velocity decreases , the radial velocity of the particles also decreases , then becomes negative . at @xmath76 where @xmath73 changes its sign , the radial pressure gradient vanishes and the gas rotates with the keplerian velocity .
the particles co - rotate with the gas at that location , and have the same radial velocity as the gas .
the drift velocity changes its sign to become positive . at high altitude ( @xmath121 ) where the gas drag is weak , the particles drift velocity begins to increase rapidly , and then the radial velocity becomes positive again at @xmath122 .
the behavior of the radial velocity of @xmath8 particles is qualitatively similar to that of @xmath123 particles , i.e. , it is outward at the midplane , inward at intermediate altitude , and outward again at high altitude .
the coupling of @xmath8 particles to the gas is weaker , so that their radial velocity is much different from the gas velocity even at the midplane .
however , the drift velocity at the midplane is still slightly smaller than the gas outflow velocity and the particles move outward .
particles of @xmath118 move inward even at the midplane , because the inward drift velocity of such large particles is larger than the outflow velocity of the gas .
these large particles also move outward at high altitude where the gas rotates faster than the particles . at the beginning stage of star formation from molecular cloud cores , dust particles in circumstellar disks may be well mixed with the gas and the dust - gas ratio may be uniform throughout the disk .
after the termination of gas infall onto the disks , the disks become to be hydrostatic and particles begin to sediment toward the midplane because of the @xmath24-component of the stellar gravity .
the particles reach the terminal velocity , where the gravity and the gas drag balance with each other , @xmath124 the time scale of the sedimentation is @xmath125 . if this time scale is much smaller than the radial migration time scale , @xmath126 , i.e. , if @xmath127 , then the particles sediment before the large migration in the @xmath17-direction .
( at high altitudes or around the midplane , particles may drift to the opposite direction to the gas , as seen in fig .
[ fig : vrd ] .
some particles with @xmath128 may have the radial migration time scale @xmath129 much larger than the above estimate .
even for such particles , the condition @xmath127 for the fast sedimentation is still valid . ) for example , at 10 au in a disk with @xmath130 and @xmath131 , particles larger than @xmath132 sediment to the midplane without large radial movement .
we consider disks at later stages , such as the t tauri stage .
the disks are still turbulent , and the turbulent motion of the gas stirs dust particles up to high altitude to prevent dust sedimentation .
an equilibrium distribution of particles in the vertical direction is achieved by the balance between the sedimentation and the diffusion due to the turbulent gas .
the turbulent diffusion is modeled in an analogy of molecular diffusion , provided that the particles are considered to be the passive tracers of fluid , i.e. , if the particles have no influence on the gas motion and have the similar velocity of the surrounding gas [ see e.g. , chapter 10 in monin & yaglom ( 1971 ) and section 3.5.1 in morfill ( 1985 ) ] .
the equation of continuity is written as @xmath133 where @xmath134 is the particle density , and @xmath135 is the particle velocity . the diffusive mass flux @xmath136 is estimated by @xmath137 where the schmidt number @xmath138 represents the strength of coupling between the particles and the gas . for small particles , @xmath139 approaches unity and the particles have the same diffusivity as the gas , while it becomes infinite for large particles .
for intermediate particle sizes , @xmath138 can be as small as @xmath140 , which means that particle diffusion occurs effectively ( see e.g. , fig . 1 in cuzzi et al .
1993 ) . in our standard model , we use @xmath141
. if @xmath138 is not unity , the particles are not the passive tracers which completely follow the fluid motion . in this case
, the formulation according to the analogy of molecular diffusion may be inappropriate .
in addition , if the velocity of sedimentation , @xmath142 , is comparable to or larger than the turbulent velocity , the particle could not be the passive tracers .
the estimate of the diffusive mass flux by equation ( [ eq : difflux ] ) should be considered just as a `` gradient diffusion hypothesis . '' in a steady state of axisymmetrical disks , @xmath143 and @xmath144 in equation ( [ eq : dustcont ] ) is zero . in addition , for particles satisfying the condition @xmath127 ( i.e. , for particles sedimenting fast without large radial migration ) , we see that @xmath145 . therefore , the mass flux in the @xmath24-direction must be zero , i.e. , @xmath146 this equation is the same as the one derived by dubrulle , morfill , & sterzik ( 1995 ) .
solving equation ( [ eq : dustdiff ] ) with equations ( [ eq : gasdensity ] ) , ( [ eq : hg ] ) , ( [ eq : nu ] ) , ( [ eq : stoptime ] ) , and ( [ eq : termnalvel ] ) gives the particle density @xmath147 \ .
\label{eq : dustdistr}\ ] ] the first term in the exponential comes from the gas distribution and the second term represents the sedimentation . in the limit of tight coupling of the particles and the gas ( @xmath148 ) , the particle distribution is the same as that of the gas .
the surface mass density of dust particles is @xmath149 figure [ fig : verden]@xmath150 shows variation of the dust - gas ratio in the vertical direction at @xmath1 in the standard model disk .
we see the sedimentation of particles .
large particles ( @xmath2 ) sediment around the midplane , while small particles ( @xmath5 ) spread over a few scale height of the gas disk .
these small particles are stirred up to high altitude by the turbulence of the gas .
the gas drag force becomes weaker at higher altitude where the gas density decreases , so the gas can not sustain the dust particles there .
thus , the particle density drops rapidly at some altitude and the dust disk obtains a relatively sharp surface ( e.g. , at @xmath151 for @xmath3 particles ) .
the sedimentation of particles is more effective in the outer part of the disk , farther away from the star ( see fig . [
fig : verden]@xmath152 ) .
for example , @xmath8 particles at @xmath33 spread over the gas disk , while the particles at @xmath153 concentrate around the midplane . at the outer part of the disk ,
the gas density is lower , so that the gas drag is weaker and turbulence can not loft the particles to high altitudes .
dust particles are well mixed with the gas if @xmath154 i.e. , below the height @xmath155^{1/2 } \ .\ ] ] in the standard model , particles smaller than @xmath156 uniformly distribute with regard to the gas under the height @xmath157 .
as discussed in
[ sec : vrd ] , the radial velocity of particles varies with the altitude . in this subsection
, we discuss the net radial velocity averaged in the vertical direction . at the beginning stage of formation of circumstellar disks , small particles
would be distributed uniformly in the disk gas . before
the particles have undergone enough sedimentation , their net radial velocity averaged in the vertical direction is @xmath158 figure [ fig : netvrd1 ] shows the net radial velocities of small particles .
the velocity of particles approaches the gas velocity as the distance from the star becomes smaller .
near the star , the gas density is high enough to induce almost all particles to move together with the gas .
as the distance from the star increases , the gas density decreases and particles at high altitude begin to drift outward from the gas . at sufficiently large distance from the star ,
the net radial velocity can be positive , i.e. , particles move outward .
however , before they move over a large distance in the radial direction , they also sediment to the midplane .
the time scale of the outward radial motion is @xmath159 ( for outflowing particles , @xmath160 is smaller than @xmath161 and is neglected in the estimate of @xmath162 ) , while the sedimentation time scale is @xmath163 . because @xmath164 , @xmath165 .
the outward motion of those particles which are well mixed with the gas may proceed shortly after the formation of circumstellar disks , but it does not significantly modify the radial distribution of dust particles .
the evolution of radial distribution occurs primarily through the motion of sedimented particles .
the net radial velocity of sedimented particles is @xmath166 the function @xmath167 , which is proportional to the mass flux , is shown in figure [ fig : fvrd ] . because of the sedimentation to the midplane , which is efficient for larger particles , the mass flux is dominated by the particles near the midplane .
for example , the mass flux of @xmath3 particles is mainly carried by particles moving outward around the midplane ( @xmath168 ) and by those moving inward at intermediate altitude ( @xmath169 ) .
although particles at high altitude ( @xmath170 ) move outward , their contribution to the mass flux is negligibly small .
very large particles , for example @xmath118 particles , move inward even at the midplane .
thus , the sedimentation causes the vast majority of particles to move inward .
sedimentation effectively removes outflowing particles at high altitude .
this flow pattern is seen as follows .
the particles flowing outward have a drift velocity larger than the gas inflow velocity , i.e. , @xmath171 . from equations ( [ eq : vrg ] ) and ( [ eq : vrdragap ] )
, it is seen that @xmath172 for such particles .
however , from equation ( [ eq : dustdistr ] ) , we also see that the density of these particles is as small as @xmath173 .
therefore , very few particles remain in the outflow region at high altitude . if the sedimentation is so effective that most particles concentrate around the midplane ( @xmath174 ) , e.g. , for particles larger than @xmath118 at @xmath1 ( see fig .
[ fig : verden]@xmath150 ) , the inward drift velocity at the midplane is larger than the gas outflow velocity
. thus , such particles flow inward . again from equation ( [ eq : dustdistr ] ) , we see that @xmath175 for such particles . then , it is seen from equations ( [ eq : vrg ] ) and ( [ eq : vrdragap ] ) that @xmath176 ) , the radial motion of the particles would not affected by the radial motion of the gas . in this limit ,
the particles motion is similar to that deduced by adachi et al .
( 1976 ) and weidenschilling ( 1977 ) .
figure [ fig : netvrd2]@xmath150 shows the net radial velocities of sedimented particles .
these net radial velocities are negative for particles of all sizes .
particles rapidly move inward when they are at large distances from the star . in the outer regions of the disk , particles sediment and
concentrate at the midplane , and their net radial velocity , which is much faster than the gas outflow velocity @xmath42 , is approximated by the drift velocity at the midplane ( eq . [ [ eq : vrdriftmid ] ] ) .
their inward velocity decreases as they approach the star , because the suppression of inward velocity by the gas drag becomes stronger .
the time scale of their orbital decay is @xmath177 .
when they approach the location where the gas density is dense enough to make @xmath178 , the inward drift velocity becomes smaller than the gas outflow velocity at the midplane .
the particles begin to move with the gas flow . at the same time
, particles are spreaded over more than the disk scale height .
the particles net radial velocity approaches the net gas velocity ( the dashed line in fig .
[ fig : netvrd2]@xmath150 ) .
however , because the particle distribution concentrates slightly more to the midplane than the gas distribution , the number of particles riding on the outflowing gas around the midplane is larger than in the case where the particle distribution is same as the gas .
the enhancement of outflowing particles retards their net inflow velocity to values below that of the gas . at the inner part of the disk , the net inward velocity of particles
is always slower than the gas velocity . near the star ,
the particles mix more with the gas and their motion converges with that of the gas .
the particles inward velocity has a minimum magnitude , @xmath179 , which is about half of the gas velocity .
particles smaller than @xmath3 have a slower inward velocity than the gas throughout nearly the entire disk ( @xmath180 ) , while particles larger than @xmath118 have slower velocity only in the innermost part of the disk ( @xmath181 ) .
the difference in the inward velocity induces the size fractionation of particles , as discussed in
[ sec : steady_dist ] below .
the dust - gas ratio may increase during the viscous evolution of disks , because the accretion velocity of small particles is slower than that of the gas .
small particles which satisfy the condition ( [ eq : condmix ] ) are mixed well with the gas .
the net radial velocities of such small particles may be calculated by equation ( [ eq : netvelmix ] ) , even after larger particles have undergone the sedimentation .
it is seen from comparison between figures [ fig : netvrd1 ] and [ fig : netvrd2]@xmath150 that , e.g. , for @xmath3 particles , the radial velocities are about the same in both figures for @xmath182 au , while they deviate significantly between these figures for @xmath183 au . as shown in figure [ fig : netvrd1 ] , the radial velocity of @xmath123 particles at @xmath183 au is outward when the particles are mixed with the gas , i.e. , in the very first stage of disk formation . however , after the particle sedimentation , the radial velocity becomes inward even at @xmath183 au , as shown in figure [ fig : netvrd2]@xmath150 .
it is concluded that the particle sedimentation suppresses the outward flow of particles in the standard model . in figure
[ fig : netvrd2]@xmath152 , we show the case with @xmath184 where the surface density profile of the gas is flatter than in the standard model .
the velocity profiles
are qualitatively similar to those in the standard model . particles of @xmath3 move slower than the gas throughout nearly the entire disk , while @xmath118 particles rapidly migrate inward . in figure
[ fig : netvrd2]@xmath30 , the case with a steeper surface density gradient is shown . in this model
, we adopt @xmath185 which satisfies the condition ( [ eq : condaccretion ] ) for inward gas accretion flow . in this case
, we see that the net radial velocity of particles becomes positive at some locations . in these regions ,
the accretion of particles is prevented .
the particles flowing inward from large distances terminate their migration at the location where the net radial velocity becomes zero , and then accumulate at that particular orbital radius .
particles located just inside this critical radius flow outward and accumulate there also .
particles at the innermost part of the disk flow inward onto the star .
size fractionation occurs because the location of the accumulation depends on the particle size .
the accumulation continues as particles drift radially inward from large radii until particles in the outer disk are significantly depleted or the number density of accumulated particles becomes so high that the removal of particles through coagulation or collisional destruction becomes efficient .
although accretion of dust particles terminates at some locations , the gas disk continues to accrete onto the star .
the dust - gas ratio increases as the accretion of the gas proceeds .
the functions of net radial velocities shown in figure [ fig : netvrd2 ] for various sizes have self - similar forms . from equations ( [ eq : vrg ] ) , ( [ eq : vrd ] ) , ( [ eq : vrdragap ] ) , ( [ eq : dustdistr ] ) , ( [ eq : dustsurface2 ] ) , and ( [ eq : netvrd ] ) , the net radial velocity can be written as @xmath186 where @xmath187 is a function of @xmath35 , @xmath36 , @xmath138 , and @xmath188 , @xmath189 \frac{t_{s,{\rm mid}}}{\alpha } \exp z^{\prime 2 } \right\ } \nonumber \\ & \times \exp \left [ -z^{\prime 2 } - \frac{{\rm sc } t_{s,{\rm mid}}}{\alpha } \left ( \exp z^{\prime 2 } -1 \right ) \right ] dz^{\prime } \nonumber \\ & \times \left\ { \int_{- \infty}^{+ \infty } \exp \left [ -z^{\prime 2 } - \frac{{\rm sc } t_{s,{\rm mid}}}{\alpha } \left ( \exp z^{\prime 2 } -1 \right ) \right ] dz^{\prime } \right\}^{-1 } \ , \end{aligned}\ ] ] and @xmath190 .
if we vary the properties of particles , keeping @xmath188 to be constant , the function @xmath187 yields the same value .
for example , when we vary the particle size @xmath191 , then transform the orbital radius as @xmath192 } r$ ] , we have the same value of @xmath188 and the same functional form of @xmath187 . in our models , @xmath193 , so @xmath194 depends on @xmath17 only through the function @xmath187 . in this case , the profiles of the net radial velocities for different sizes are obtained by the transformation in the @xmath17-direction from one profile . in figure
[ fig : f ] , the function @xmath187 is plotted for various values of the schmidt number @xmath138 . in the standard model
, the net radial velocity of particles is inward everywhere in the disk .
this flow pattern is partially due to the removal of particles , through sedimentation , from high altitudes where they would flow outward . for particles which are sufficiently large to concentrate near the midplane , the magnitude of their inward drift velocity is larger than that of the outflowing gas . thus , the strength of the sedimentation is always moderate enough to remove outflowing particles .
the sedimentation is controlled by coupling of particles to the turbulent motion of the gas .
if coupling is relatively strong ( @xmath139 is smaller than unity ) , the sedimentation would be less effective and particles would mix more completely to the gas , while if it is relatively weak ( @xmath138 is larger ) , particles would concentrate more to the midplane .
usually @xmath138 is assumed to be unity , but the actual value may be different from unity .
some experimental data are plotted in figure 1 in cuzzi et al .
( 1993 ) which show @xmath138 as small as @xmath140 when the stokes number @xmath195 is about @xmath196 ( the stokes number is considered to be the same order of the non - dimensional stopping time @xmath85 in our models ) . for large particles ( @xmath197 ) , @xmath138 increases with the particle size .
figure [ fig : netvrd2]@xmath83 shows the net radial velocities in the case where the sedimentation is efficient ( @xmath198 ) .
the net radial velocity becomes outward at its peak .
thus , an accumulation of particles and an increase in the dust - gas ratio would occur .
because of the strong sedimentation , the dust disk is much thinner than in the standard model ( @xmath199 ) and concentrates more to the midplane where the gas flows outward . in the outer part of the disk ( for example @xmath183 au for @xmath200 particles ) , the inward drift velocity of particles ( relative to the gas ) at the midplane is faster than the outward gas velocity . as the particles drift in , the inward velocity decreases , and at some distances from the star ( @xmath201 au for @xmath8 particles ) , the inward drift velocity at the midplane becomes smaller than the outward gas velocity , and the particles around the midplane flow outward . at such distances
, the dust disk is still concentrated around the midplane , and the net radial velocity is dominated by the particles around the midplane and is outward . as the distance from the star decreases , the particles mix more with the gas and the concentration to the midplane becomes weaker .
the net radial velocity approaches the gas velocity , which is inward , at the innermost part of the disk .
if the schmidt number @xmath138 is less than unity , the sedimentation of particles is less efficient . therefore , the particles are always well mixed with the gas at where the inward drift velocity of particles is smaller than the gas outflow velocity ( i.e. , at where the particle outflow at the midplane occurs ) .
thus , the net radial velocity of particles is inward everywhere .
figure [ fig : f ] shows the function @xmath187 for various values of the schmidt number , @xmath202 , @xmath203 , and @xmath204 .
the net radial velocity has the same sign as @xmath187 ( see eq . [ [ eq : netvelapprx ] ] ) .
it is seen that the outflow of dust particles occurs if the sedimentation is efficient ( @xmath205 ) .
because the inward velocity of particles depends on their size , particles of different sizes accumulate at different locations . in this subsection , we discuss the density distribution of dust particles as a consequence of their radial flow and show how their size fractionation may occur .
we adopt the models in which the net velocity of particles is always inward ( @xmath199 and @xmath206 ) .
we assume that the distribution of particles approaches a steady state after a brief stage of initial evolution .
when a steady state is achieved , the mass flux of dust particles becomes constant in the radial direction .
we calculate the mass flux of particles as @xmath207 where @xmath208 and @xmath209 are the mass flux and the surface density , respectively , of particles smaller than size @xmath90 . for simplicity , we neglect three physical processes .
first , the evolution of the particle size through the coagulation , collisional destruction , condensation , and sublimation is neglected , i.e. , we assume the mass flux @xmath210 for each size range is constant in the radial direction .
second , in turbulent disks , the particles mass flux comes not only from the mean flow with an average velocity @xmath211 but also from the turbulent diffusion of particles , which appears as @xmath136 in equation ( [ eq : dustcont ] ) for example .
we neglect the mass flux from the turbulent diffusion .
third , we assume the structure of the gas component of the disk is entirely determined by the gas itself .
we neglect the feedback drag induced by the particles on the gas . in the limit that the spatial density of the dust component becomes comparable to that of the gas near the midplane , this effect would speed up the azimuthal velocity of the gas to the keplerian value and quench the radial migration of the dust ( cuzzi et al .
such a dust concentration requires substantial sedimentation of relatively large particles .
contributions from all of these factors will be investigated in future grain - evolution calculations . here
, we adopt the simplest assumptions to focus on showing the size fractionation of particles .
the mass flux of particles in all size range is @xmath212 where @xmath213 and @xmath214 are the minimum and maximum sizes of particles , respectively .
the surface density in the steady state is calculated from equation ( [ eq : mflux ] ) with given @xmath208 .
figure [ fig : surden]@xmath150 shows the surface density of the dust component composed of single size particles
. particles of different sizes have different density profiles . if the dust component is composed of relatively large particles , it would be concentrate at the inner region of the disk .
thus , as particles grow in size , their surface density distribution becomes more centrally concentrated .
figure [ fig : surden]@xmath152 shows the surface density , assuming the size distribution of particles is a power law with index @xmath215 , @xmath216 , and @xmath217 .
the surface density distribution of the dust particles is different from that of the gas .
the power law index of the dust distribution @xmath209 is about @xmath218 for a gas disk with index @xmath69 ( @xmath219 for a gas disk with @xmath220 ) .
the implied power law index , @xmath218 , is similar to the value anticipated from the present mass distribution of planets in the solar system ( hayashi et al .
however , note that the density distributions in figure [ fig : surden ] are derived assuming no size evolution of particles , and no turbulent diffusion in the radial direction .
the growth of particles during the radial flow adds a source term in the equation of continuity .
the evolution of the dust density profile should be investigated further by taking particle growth and turbulent diffusion into account .
the accretion velocity of dust particles is different from that of the gas .
this difference causes evolution of the dust - gas ratio . in the standard model ,
@xmath3 particles between @xmath204 and @xmath221 au have an inflow velocity which is about half of the gas velocity .
thus , if the mass of the dust disk is dominated by @xmath3 particles ( for example , if the particles have size distribution @xmath222 with maximum size @xmath3 ) , the dust - gas ratio would increase through the gas accretion .
for example , if the initial mass of the gas disk is @xmath223 and it reduces to @xmath224 after viscous accretion , the dust - gas ratio would increase to be 6 times of its initial value .
the increase in the dust - gas ratio speeds up the formation of planetesimals through mutual collisions .
their enhanced abundance may also cause the dust component to become unstable and promote the planetesimal formation through the gravitational instability ( goldreich & ward 1973 ; sekiya 1998 ) . if the particle growth proceeds to make @xmath203 mm to @xmath203 cm particles during the gas accretion phase , such large particles migrate inward rapidly and accumulate in the inner part of the disk ( see fig [ fig : surden]@xmath150 ) .
this size fractionation causes an increase in the dust - gas ratio at the inner disk , while this process may decrease the dust - gas ratio at the outer disk , resulting in a dust disk concentrated to the inner part of the gas disk .
dust particles move radially both inward and outward by the gas drag force .
particles at high altitude ( @xmath225 ) move outward because they rotate slower than the gas whose pressure gradient force is inward .
small particles ( @xmath226 at @xmath1 ) near the midplane ( @xmath174 ) are advected by the gas outflow . on the other hand , particles at intermediate altitude and large particles ( @xmath227 at @xmath1 )
move inward .
the net radial velocity , averaged in the vertical direction , is usually inward , provided that the radial gradient of the gas surface density is not too steep ( @xmath228 ) .
sedimentation removes outflowing particles from high altitudes .
small particles , which can be advected by the outflowing gas around the midplane , do not concentrate at the midplane . in the inner part of the gas disk ( @xmath229 au for @xmath3 particles ) ,
the inflow velocity of particles is smaller than the gas accretion velocity , resulting in an increase in the dust - gas ratio . \3
. the particle sedimentation would be efficient if dust - gas coupling is relatively weak ( @xmath230 ) .
if the sedimentation is so efficient ( @xmath205 ) , the number of outflowing particles around the midplane would be large , and the direction of the net radial velocity of particles would change to outward at some distances from the star .
accumulation of particles at such locations serves to increase the local dust - gas ratio .
we are grateful to the anonymous referee who gave us important criticism , especially on the improper treatment of particle diffusion in the original version of the manuscript .
his / her suggestions also considerably improved the paper .
we thank greg laughlin for careful reading of the manuscript and jeff cuzzi for discussions on the turbulent motion of dust particles .
this work was supported in part by an nsf grant ast 99 87417 and in part by a special nasa astrophysical theory program that supports a joint center for star formation studies at uc berkeley , nasa - ames research center , and uc santa cruz . | we study the radial migration of dust particles in accreting protostellar disks analogous to the primordial solar nebula .
our main objective is to determine the retention efficiency of dust particles which are the building blocks of the much larger planetesimals .
this study takes account of the two dimensional ( radial and normal ) structure of the disk gas , including the effects of the variation in the gas velocity as a function of distance from the midplane .
it is shown that the dust component of disks accretes slower than the gas component . at high altitude from the disk midplane ( higher than a few disk scale heights ) , the gas rotates faster than particles because of the inward pressure gradient force , and its drag force causes particles to move outward in the radial direction .
viscous torque induces the gas within a scale height from the disk midplane to flow outward , carrying small ( size @xmath0 at @xmath1 ) particles with it
. only particles at intermediate altitude or with sufficiently large sizes ( @xmath2 at @xmath1 ) move inward . when the particles radial velocities are averaged over the entire vertical direction , particles have a net inward flux .
the magnitude of their radial motion depends on the particles distance from the central star . at large distances , particles migrate inward with a velocity much faster than the gas accretion velocity .
however , their inward velocity is reduced below that of the gas in the inner regions of the disk .
the rate of velocity decrease is a function of the particles size . while larger particles retain fast accretion velocity until they approach closer to the star , @xmath3 particles have slower velocity than the gas in the most part of the disk ( @xmath4 au ) .
this differential migration of particles causes the size fractionation .
dust disks composed mostly of small particles ( size @xmath5 ) accrete slower than gas disks , resulting in the increase in the dust - gas ratio during the gas accretion phase .
if the gas disk has a steep radial density gradient or if dust particles sediment effectively to the disk midplane , the net vertically averaged flux of particles can be outward . in this case , the accretion of the dust component is prevented , leading to the formation of residual dust disks after their gas component is severely depleted . |
dark matter still remains one of the main unsolved problem in physics .
the common accepted paradigm is the existence of an exotic weakly interacting massive particle ( wimp ) .
such a particle has to be found in some extension of the standard model ( sm ) of particle physics .
supersymmetry and extra space - time dimensions are commonly used ingredients for a consistent theory beyond the sm .
one the most studied framework is the mssm , the minimal supersymmetric extension of the sm . in the mssm
the lightest supersymmetric particle ( lsp ) is usually a neutralino , which is a good candidate for cold dark matter ( for a recent review see @xcite ) .
the pattern of the soft supersymmetry breaking terms and the presence of extra space - time dimensions greatly affects the composition and the strength of the dominant interactions of the neutralino .
hence it is crucial to study the neutralino phenomenology in different scenarios .
the glast experiment will have a great chance to shed light on the nature of dark matter , for example through the analysis of the continuum @xmath0-ray flux supposed to come from pair wimp annihilations .
one of the most studied supersymmetric scenario is the constrained mssm ( cmssm ) in which the soft supersymmetry breaking terms are supposed to derive from a high energy supergravity theory with a common scalar mass @xmath1 and a common gaugino mass @xmath2 at the gut scale . in this framework
the neutralino is the lsp in most of the parameter space .
it can be studied through indirect detection of @xmath0-rays from pair annihilation .
the expected @xmath0-ray continuum flux at a given photon energy @xmath3 from a direction that form an angle @xmath4 is given by @xmath5 it depends on the particle physics model assumed through the neutralino mass @xmath6 , the total annihilation cross section @xmath7 and through the sum of all the photon yield @xmath8 per each annihilation channel weighted by the corresponding branching ratio @xmath9 .
the flux ( [ gammafluxcont ] ) also depends on the wimp density in the galactic halo @xmath10 .
the integral has to be performed along the line of sight ( l.o.s . ) .
the wimp density profiles @xmath10 are in general extremely cuspy towards the galactic center ( gc ) that hence is a good place where to look for an exotic signal . in our analysis we take into account the navarro , frenck and white ( nfw ) profile @xcite and the moore profile @xcite .
the usual parametrization for a dark matter halo density is @xmath11^{(\beta-\gamma)/\alpha } } \;\;. \label{profile}\ ] ] the nfw profile behaves like @xmath12 towards the gc while the moore profile goes as @xmath13 . in general
one expects also contributions coming from standard astrophysical sources . in many diffuse continuum @xmath0-ray production models @xcite in our galaxy ,
the dominant astrophysical contribution comes from @xmath14 .
this has to be considered the `` standard '' background .
we performed a statistical analysis , based on the usual @xmath15 test , in order to determine what are the models that can be disentangled against the background by glast at a given significance @xcite .
we have taken into account only the statistical errors , i.e. @xmath16 where @xmath17 is the number of background photons in each energy bin .
doing a detailed scan of the cmssm parameter space defined by @xmath1 , @xmath2 , @xmath18 , @xmath19 and @xmath20 one can find the regions that are detectable by glast .
besides the previously explained parameters @xmath1 , @xmath2 , @xmath18 denotes the ratio of the vacuum expectation values of the two neutral components of the su(2 ) higgs doublet , @xmath19 is the proportionality factor between the supersymmetry breaking trilinear couplings and the yukawa couplings while @xmath21 is determined ( up to a sign ) by imposing the electro - weak symmetry breaking ( ewsb ) conditions at the weak scale . the result , at a @xmath22 confidence level , for @xmath23 case is shown in fig .
[ tanbe55cmssm ] .
we assumed a total exposure of @xmath24 , an angular resolution ( at @xmath25 gev ) of @xmath26 sr and 4 years of data taking @xcite . ]
blue solid lines represent the glast reach for a nwf and moore profile while dashed lines represent the neutralino isomass contours ( expressed in gev ) .
gray shaded regions are parameter space regions excluded by either theoretical or experimental ( accelerator ) constraints .
the upper left region is excluded because the lightest stau is the lsp , the lower left region is excluded due the accelerator bounds on the higgs boson masses , @xmath27 , slepton and squark masses , etc . , while the right lower region is excluded because there is no electroweak symmetry breaking .
the wmap compatible region ( with relic density @xmath28 ) is the red one while in the green region the neutralino is a subdominant dark matter component ( with @xmath29 ) .
in recent times many scenarios involving extra space - time dimensions with or even without supersymmetry received great attention .
it is very interesting to see if glast is able to detect signals in this context .
we consider a very wide scenario involving extra dimensions and low energy supersymmetry @xcite . in this framework
there are three additional parameters , namely the typical size of extra dimensions @xmath30 , their number @xmath31 and the number @xmath32 of fermion generations that are allowed to have extra kaluza - klein ( kk ) states .
the most striking feature in this class of models is the presence of power - law corrections to all the couplings and masses of the mssm @xcite .
this implies that one can have a grand unification scale ( gut ) scale as low as few tev . from the point of view of indirect detection
the case @xmath33 in which some fermion generation have kk tower is disfavored .
thus in the following we concentrate on the minimal case @xmath34 . moreover the number of extra dimensions @xmath31 does not seem to be crucial for the neutralino phenomenology @xcite . in almost all the parameter space
the neutralino is still the lsp .
one of the main result is that , unlike the standard cmssm case , the neutralino is no longer bino - like but it tends to be a very pure higgsino . in general
this result holds for not too high compactification scale @xmath35 tev while going towards higher scale the neutralino tends to be a bino .
we show the @xmath22 glast reach ( along the line sketched in the previous section ) for @xmath36 tev , @xmath37 and @xmath34 and for a low value of @xmath18 .
tev , @xmath37 and @xmath34 [ glast reach extradim 104 ] ] the plot assumptions and the explanation of the different regions are the same as in fig .
[ tanbe55cmssm ] . in this scenario
a nfw profile is not enough for glast to be able to explore the cosmologically favored region in which the neutralino has a mass greater than @xmath38 gev . in order to do that a moore profile is needed .
this corresponds to an enhancement factor of about one order of magnitude in the continuum flux .
it has to be remarked that assuming some non standard cosmological evolution the green region could become a wmap compatible region .
this is due to a non thermal enhancement of the relic density ( see for example @xcite ) .
these kind of scenarios are particularly motivated in scenarios with extra dimensions . in fig .
[ glast reach extradim 108 ] we show the case with @xmath39 tev and with only one extra dimensions @xmath40 .
all the kk particles are quite heavy in this case but the effects on the low energy theory , through one loop processes , are still sizable .
tev , @xmath40 and @xmath34 [ glast reach extradim 108 ] ] in this scenario the glast reach for a nfw profile is still below the wmap allowed region but for a moore profile glast will be able to probe almost all the cosmologically favored region .
we have shown that , assuming a nfw profile , glast will be able to probe a huge part of the wmap allowed zone of the cmssm parameter space especially in the high @xmath18 case . in the scenario involving extra dimensions , with quite low compactification scale @xmath41 tev , glast will still be able to probe some part of the cosmologically favored region of the parameter space either in the case of more cuspy moore profile or in the case of an enhancement of the neutralino relic density in a non standard cosmological scenario . in the case of higher compactification scale @xmath42 tev glast will be able to cover almost all the wmap region .
[ concl ] 99 g. bertone , d. hooper and j. silk , phys .
* 405 * ( 2005 ) 279 j. f. navarro , c. s. frenk and s. d. white , astrophys .
j. * 462 * ( 1996 ) 563 b. moore , t. quinn , f. governato , j. stadel and g. lake , mon . not .
* 310 * ( 1999 ) 1147 f. w. stecker , astrophys
. j. * 212 * ( 1977 ) 60 .
a. cesarini , f. fucito , a. lionetto , a. morselli and p. ullio , astropart .
* 21 * ( 2004 ) 267 http://www-glast.slac.stanford.edu/software/is/glast_lat_performance.htm f. fucito , a. lionetto and m. prisco , jcap * 0606 * ( 2006 ) 002 k. r. dienes , e. dudas and t. gherghetta , nucl .
b * 537 * ( 1999 ) 47 t. moroi and l. randall , nucl .
b * 570 * ( 2000 ) 455 | we give an overview of the possibility of glast to explore theories beyond the standard model of particle physics . among the wide taxonomy we will focus in particular on low scale supersymmetry and theories with extra space - time dimensions .
these theories give a suitable dark matter candidate whose interactions and composition can be studied using a gamma ray probe .
we show the possibility of glast to disentangle such exotic signals from a standard production background .
address = infn and department of physics roma tor vergata + via della ricerca scientifica , 1 + italy . |
the notion of a _ quantum random walk _ has emerged as an important element in the development of efficient quantum algorithms . in particular
, it makes a dramatic appearance in the most efficient known algorithm for element distinctness @xcite .
the technique has also provided simple separations between quantum and classical query complexity @xcite , improvements in mixing times over classical walks @xcite , and some interesting search algorithms @xcite .
the basic model has two natural variants , the _ continuous _ model of childs , et al .
@xcite , on which we will focus , and the _ discrete _ model introduced by aharonov , et al . @xcite .
we refer the reader to szegedy s @xcite article for a more detailed discussion . in the continuous model , a quantum walk on a graph @xmath0 is determined by the time - evolution of the schrdinger equation using @xmath1 as the hamiltonian , where @xmath2 is the laplacian of the graph and @xmath3 is a positive scalar to which we refer as the `` jumping rate '' or `` energy '' .
in addition to being a physically attractive model , it has been successfully applied to some algorithmic problems as indicated above .
such walks have been studied over a variety of graphs with special attention given to cayley graphs , whose algebraic structure has provided immediate methods for determining the spectral resolution of the linear operators that determine the system s dynamics . once it had been discovered that quantum random walks can offer improvement over their classical counterparts with respect to such basic phenomena as mixing and hitting times , it was natural to ask how robust these walks are in the face of decoherence , as this would presumably be an issue of primary importance for any attempt at implementation @xcite .
in this article , we study the effects of a natural notion of decoherence on the hypercubic quantum walk .
our notion of decoherence corresponds , roughly , to independent measurement `` accidentally '' taking place in each coordinate of the walk at a certain rate @xmath4 .
we discover that for values of @xmath4 beneath a threshold depending on the energy of the system , the walk retains the basic features of the non - decohering walk ; these features disappear beyond this threshold , where the behavior of the classical walk is recovered .
moore and russell @xcite analyzed both the discrete and the continuous quantum walk on a hypercube .
kendon and tregenna @xcite performed a numerical analysis of the effect of decoherence in the discrete case . in this article , we extend the continuous case with the model of decoherence described above . in particular , we show that up to a certain rate of decoherence , both linear instantaneous mixing times and linear instantaneous hitting times still occur . beyond the threshold ,
however , the walk behaves like the classical walk on the hypercube , exhibiting @xmath5 mixing times .
as the rate of decoherence grows , mixing is retarded by the quantum zeno effect .
consider the continuous quantum walk on the @xmath6-dimensional hypercube with energy @xmath3 and decoherence rate @xmath4 , starting from the initial wave function @xmath7 , corresponding to the corner with hamming weight zero .
we prove the following theorems about this walk . when @xmath8 , the walk has instantaneous mixing times at @xmath9 for all @xmath10 , @xmath11 . at these times , the total variation distance between the walk distribution and the uniform distribution is zero .
this result is an extension of the results in @xcite , and an improvement over the classical random walk mixing time of @xmath5 .
note that the mixing times decay with @xmath4 and disappear altogether when @xmath12 .
further , for large @xmath4 , we will see that the walk is retarded by the quantum zeno effect . when @xmath8 , the walk has approximate instantaneous hitting times to the opposite corner @xmath13 at times @xmath14 for all @xmath10 , @xmath15 .
however , the probability of measuring an exact hit decays exponentially in @xmath16 ; the probability is @xmath17^n\enspace.\ ] ] in particular , when no decoherence is present , the walk hits at @xmath18 and it does so exactly , i.e. @xmath19 . for @xmath12 , no such hitting occurs .
this result is a significant improvement over the exponential hitting times of the classical random walk , with the caveat that decoherence has a detrimental effect on the accuracy of repeated hitting times .
finally , we show that under high levels of decoherence , the measurement distribution of the walk actually converges to the uniform distribution in time @xmath5 , just as in the classical case . for a fixed @xmath12 ,
the walk mixes in time @xmath5 . in the remainder of the introduction
, we describe the continuous quantum walk model , and recall the graph product analysis of moore and russell @xcite . in the second section , we describe our model of decoherence , derive a superoperator that governs the behavior of the decohering walk , and prove that it is decomposable into an @xmath6-fold tensor product of a small system .
we then fully analyze the small system in the third section , and use those results to draw conclusions about the general walk in 3 distinct regimes : @xmath8 , @xmath20 , and @xmath21 .
these regimes are roughly analogous to underdamping , critical damping , and overdamping ( respectively ) of a simple harmonic oscillator with damping rate @xmath4 and angular frequency @xmath22 .
a continuous quantum walk on a graph @xmath0 begins at a distinguished vertex @xmath23 of @xmath0 , the initial wave function of the walk being @xmath24 , where @xmath25 if @xmath26 and @xmath27 otherwise .
the walk then evolves according to the schrdinger equation . in our case ,
the graph is the @xmath6-dimensional hypercube .
concretely , we identify the vertices with @xmath6-bit strings , with edges connecting those pairs of vertices that differ in exactly one bit .
since the hypercube is a regular graph , we can let the hamiltonian @xmath28 be the adjacency matrix instead of the laplacian @xcite ; the dynamics are then given by the unitary operator @xmath29 and the state of the walk at time @xmath30 is @xmath31 .
the following analysis makes use of the hypercube s product graph structure ; this structure will be useful again later when we consider the effects of decoherence .
the analysis below diverges from that of moore and russell @xcite only in that we allow each qubit to have energy @xmath32 instead of @xmath33 .
the energy of the entire system is then @xmath3 .
let @xmath34 and let @xmath35 where the @xmath36th term in the sum has @xmath37 as the @xmath36th factor in the tensor product .
then we have @xmath38^{\otimes n } \\ &
= \left[\begin{matrix}\cos(kt / n ) & i~\sin(kt / n ) \\
i~\sin(kt / n ) & \cos(kt / n ) \end{matrix}\right]^{\otimes n}\enspace.\end{aligned}\ ] ] applying @xmath39 to the initial state @xmath7 , we have @xmath40^{\otimes n}\ ] ] which corresponds to a uniform state exactly when @xmath41 is an odd multiple of @xmath42 .
we begin by recalling a model of decoherence commonly used in the discrete model , with the intention of deriving a superoperator @xmath39 , acting on density matrices , which mimics these dynamics in our continuous setting . the discrete model , described in @xcite , couples unitary evolution according to the discrete - time quantum random walk model of aharonov et al .
@xcite with partial measurement at each step occurring with some fixed probability @xmath4 .
specifically , the evolution of the density matrix can be written as @xmath43 where @xmath44 is the unitary operator of the walk , @xmath45 runs over the dimensions where the decoherence occurs , and the @xmath46 project in the usual `` computational '' basis @xcite . in the continuous
setting , the unitary operator that governs the non - decohering walk is @xmath47 , where @xmath28 is the normalized adjacency matrix of the hypercube times an energy constant . to extend the above decoherence model to this setting ,
recall that the superoperator @xmath48 associated with these dynamics has the property that @xmath49\enspace;\ ] ] wishing to augment these dynamics with measurement occurring at some prescribed rate @xmath4 , we desire a superoperator @xmath50 that satisfies @xmath51[(1 - p\,dt ) \identity + pdt(\mathbf{p})]\ ] ] where @xmath52 is the operator associated with the decohering measurement . intuitively , the unitary evolution of the system is punctuated by measurements taking place with rate @xmath4 , analogous to the discrete case . letting @xmath53 , we can expand and simplify : @xmath54[(1 - pdt ) \identity + pdt(\mathbf{p } ) ] \\ & = s_t[(\identity - ihdt ) \otimes ( \identity + ihdt)][(1 - pdt ) \identity + pdt(\mathbf{p } ) ] \\ & = s_t[\identity \otimes \identity + idt(\identity \otimes h - h \otimes \identity ) - pdt \identity \otimes \identity + pdt(\mathbf{p})]\enspace.\end{aligned}\ ] ] in terms of a differential equation , @xmath55 - s_t}{dt}\\ & = s_t[i(\identity \otimes h - h \otimes \identity ) - p\identity \otimes \identity + p(\mathbf{p})]\enspace.\end{aligned}\ ] ] the solution is @xmath56t\right)\enspace.\ ] ] we now define the decoherence operator @xmath57 .
this operator will correspond to choosing a coordinate uniformly at random and measuring it by projecting to the computational basis @xmath58 let @xmath59 and @xmath60 be the single qubit projectors onto @xmath61 and @xmath62 , respectively .
we define @xmath63\ ] ] where @xmath64 with the nonidentity projector appearing in the @xmath45th place .
we define @xmath65 similarly , so that @xmath66 ignores all the qubits except the @xmath45th one , and projects it onto @xmath67 where @xmath68 .
note that @xmath69 \otimes \cdots \otimes [ \pi_j \otimes \pi_j ] \otimes \cdots \otimes [ \identity \otimes \identity]\ ] ] for @xmath68 .
the pure continuous quantum walk on the @xmath6-dimensional hypercube is easy to analyze , in part , because it is equivalent to a system of @xmath6 non - interacting qubits .
we now show that , with the model of decoherence described above , each dimension still behaves independently .
in particular , the superoperator that dictates the behavior of the walk is decomposable into an @xmath6-fold tensor product . recall the product formulation of the non - decohering hamiltonian @xmath70 where @xmath71 with @xmath37 appearing in the @xmath36th place in the tensor product .
we have given each single qubit energy @xmath32 , resulting in a system with energy @xmath3 .
this choice will allow us to precisely describe the behavior of the walk in terms of the relationship between the energy of the system and the rate of decoherence
. we can write each of the terms in the exponent of the superoperator from ( [ superop - soln ] ) as follows : @xmath72 \otimes \cdots \otimes [ \identity \otimes \sigma_x ] \otimes \cdots \otimes [ \identity \otimes \identity]\enspace , \\ h \otimes \identity & = \sum_{j=1}^n [ \identity \otimes \identity ] \otimes \cdots \otimes [ \sigma_x \otimes \identity ] \otimes \cdots \otimes [ \identity \otimes \identity]\enspace.\end{aligned}\ ] ] our decoherence operator can also be written in this form : @xmath73 \\ & = & \frac{1}{n } \sum_{j=1}^n ( [ { \identity } \otimes { \identity } ] \otimes \cdots \otimes [ \pi_0 \otimes \pi_0 ] \otimes \cdots \otimes [ { \identity } \otimes { \identity}]\\ & & + [ { \identity } \otimes { \identity } ] \otimes \cdots \otimes [ \pi_1 \otimes \pi_1 ] \otimes \cdots \otimes [ { \identity } \otimes { \identity}])\enspace.\end{aligned}\ ] ] the identity operator has a consistent decomposition : @xmath74 \otimes \cdots \otimes [ { \identity } \otimes { \identity}].$ ] we can now put these pieces together to form the superoperator : @xmath75 \otimes \cdots \otimes \mathbf{a } \otimes \cdots [ { \identity } \otimes { \identity}]\right ) \\ & = \prod_{j=1}^n [ { \identity } \otimes { \identity } ] \otimes \cdots \otimes e^\mathbf{a } \otimes \cdots [ { \identity } \otimes { \identity}]\\ & = \left[e^\mathbf{a}\right]^{\otimes n}\end{aligned}\ ] ] where @xmath76 \\ & = & \frac{t}{n}\left ( \begin{matrix } 0 & ik & -ik & 0 \\ ik & -p & 0 & -ik \\
-ik & 0 & -p & ik \\ 0 & -ik & ik & 0 \\ \end{matrix } \right).\end{aligned}\ ] ] notice that for @xmath77 , @xmath78^{\otimes n } = \left[e^{-it\sigma_x } \otimes e^{it\sigma_x}\right]^{\otimes n}$ ] , which is exactly the superoperator formulation of the dynamics of the non - decohering walk .
so far we have shown that the walk with decoherence is still equivalent to @xmath6 non - interacting single - qubit systems .
we now analyze the behavior of a single - qubit system under the superoperator @xmath79 .
the structure of this single particle walk will allow us to then immediately draw conclusions about the entire system .
the eigenvalues of @xmath80 are @xmath27 , @xmath81 , @xmath82 and @xmath83 . here
@xmath84 is a complex constant that will later turn out to be important in determining the behavior of the system as a function of the rate of decoherence @xmath4 and the energy @xmath3 .
the matrix exponential of @xmath80 in this spectral basis can then be computed by inspection . to see how our superoperator acts on a density matrix @xmath85
, we may change @xmath85 to the spectral basis , apply the diagonal superoperator to yield @xmath86 , and finally change @xmath86 back to the computational basis . at that point
we can apply the usual projectors @xmath59 and @xmath60 to determine the probabilities of measuring @xmath27 or @xmath87 in terms of time .
let @xmath88 and @xmath89 . in the diagonal basis ,
@xmath90\ ] ] and thus at time @xmath30 we have @xmath91.\ ] ] if we then change back to the computational basis and project by @xmath59 and @xmath60 , we may compute the probabilities of measuring @xmath27 and @xmath87 at a particular time @xmath30 : @xmath92 = \frac{1}{4}\left[2 + e^{\frac{-p t - \alpha t}{2n}}(1 - p/ \alpha ) + e^{\frac{-p t + \alpha t}{2n}}(1 + p/ \alpha)\right]\ ] ] @xmath93 = \frac{1}{4}\left[2 - e^{\frac{-p t - \alpha t}{2n}}(1 - p/ \alpha ) - e^{\frac{-p t + \alpha t}{2n}}(1 + p/ \alpha)\right]\ ] ] which can be simplified somewhat to @xmath92 = \frac{1}{2 } + \frac{1}{2 } e^{-\frac{pt}{2n } } \left[\cos\left(\frac{\beta t}{2n}\right ) + \frac{p}{\beta}~\sin\left(\frac{\beta t}{2n}\right ) \right]\ ] ] @xmath93 = \frac{1}{2 } - \frac{1}{2 } e^{-\frac{pt}{2n } } \left[\cos\left(\frac{\beta t}{2n}\right ) + \frac{p}{\beta}~\sin\left(\frac{\beta t}{2n}\right ) \right].\ ] ] here we have let @xmath94 for simplicity .
a quick check shows that when @xmath95 @xmath96 = \cos^2\left(\frac{kt}{n}\right)$ ] and @xmath97 = \sin^2\left(\frac{kt}{n}\right)$ ] , which are exactly the dynamics of the non - decohering walk .
the probabilities for this non - decohering case are shown in figure [ fig1 ] .
the three regimes mentioned before are immediately apparent . for @xmath8
, @xmath100 is real .
when @xmath20 , we have @xmath101 , which appears to be a serious problem at first glance . finally , for @xmath21 , @xmath100 is imaginary .
we now address each of these three situations in detail .
when @xmath8 , we recover the perhaps most interesting feature of the non - decohering walk : the instantaneous mixing time is linear in @xmath6 . to exactly determine the mixing times for our decohering walk , we solve @xmath96 = p[1 ] = \frac{1}{2}$ ] ; this amounts to determining when @xmath103\ ] ] equals zero .
clearly the exponential decay term results in mixing as @xmath104 ; our principle concern , however , is with the periodic mixing times analogous to those of the original walk .
we thus ignore the exponential term when solving the equality @xmath105 , which yields @xmath106 this equation actually has more solutions than the one we started with , because of the use of half - angle formulas for simplification .
the solutions that we want are @xmath107\ ] ] where @xmath16 ranges over the positive integers .
evidently , the mixing times still occur in linear time ; an example is shown in figure [ fig2 ] .
note also that if we let @xmath77 , we have @xmath108 , which are exactly the nice periodic mixing times of the non - decohering walk . in the decohering case , however , these mixing times drift towards infinity , and cease to exist altogether beyond the threshold of @xmath20 .
this proves theorem 1 .
we now wish to determine when our small system is as close as possible to @xmath109 .
since our large - system walk begins at @xmath110 , this will correspond to approximate hitting times to the opposite corner @xmath111 .
these times correspond to local maxima of @xmath97 $ ] ; the solutions are @xmath112 where @xmath16 ranges over the non - negative integers . at these points in time , the value of @xmath97 $ ] is @xmath113 which immediately yields theorem 2 .
we first observe that @xmath114 as @xmath115 .
hence , we do not expect to see any mixing in this case . to analyze the probabilities exactly , we take the limit of @xmath116 as @xmath115 .
the solution is @xmath117\enspace.\ ] ] indeed , since @xmath3 , @xmath30 and @xmath6 are all positive , @xmath116 is zero only in the limit as @xmath104 . the linear mixing and hitting behavior from the previous section has entirely disappeared . as in the critical damping of simple harmonic motion ,
a small decrease in the rate @xmath4 can result in drastically different behavior , in this case a return to linear mixing and hitting .
we leave the limiting mixing analysis of this case for the next section , where we develop some relevant tools .
the goal of this section is to show two interesting consequences of the presence of substantial decoherence in the quantum walk on the hypercube .
first , we will show that for a fixed @xmath12 , the walk behaves much like the classical walk on the hypercube , mixing in time @xmath5 .
second , we show that as @xmath120 , the walk suffers from the quantum zeno effect .
informally stated , the rate of decoherence is so large that the walk is continuously being reset to the initial wave function @xmath121 by measurement .
consider a single qubit .
let @xmath122 be the distribution obtained by full measurement at time @xmath30 , and @xmath44 the uniform distribution : @xmath123 where @xmath124.\ ] ] for @xmath125 , @xmath126 are the analogous product distributions in the @xmath6-dimensional case . to analyze the limiting mixing behavior of the walk
, we will consider the total variation distance @xmath127 between these distributions . in order to give bounds for total variation , we will use _
hellinger distance _
@xcite , defined as follows : @xmath128 we will make use of the following two properties of hellinger distance : @xmath129 and @xmath130 the first property makes it easy to work with product distributions .
the second gives a nice relationship between hellinger distance and total variation distance . in our case , @xmath131 and hence , by ( [ helltv ] ) , @xmath132 consider the walk with decoherence rate @xmath21 .
we have @xmath133 , where @xmath134 and @xmath4 are positive and real .
it follows that for a fixed @xmath21 , @xmath135 and @xmath136 as @xmath104 .
hence the walk does indeed mix eventually , and the measurement distribution in fact converges to the uniform distribution .
let @xmath137 where @xmath138 is a constant , and rewrite @xmath116 as follows : @xmath139.\ ] ] suppose we choose @xmath140 such that @xmath141 .
then @xmath142 , which implies that @xmath143 .
on the other hand , if @xmath144 , then @xmath145 and there exists a constant @xmath146 such that @xmath147 .
this shows that the walk mixes in time @xmath5 when @xmath21 .
notice that when @xmath20 , @xmath148 , so that the same technique easily extends to that case via equation ( [ pequals4k ] ) .
this completes the proof of theorem 3 .
recall from the previous section that the time required to mix when @xmath21 is @xmath149 which clearly increases with @xmath4 .
further , for large @xmath4 , @xmath150 tends to @xmath87 , and hence @xmath116 tends to @xmath151 notice that @xmath152 corresponds to remaining at the initial state forever .
we conclude that the mixing of the walk is retarded by the quantum zeno effect , where measurement occurs so often that the system tends to remain in the initial state .
alexander russell gratefully acknowledges the support of the national science foundation , under the grants career ccr-0093065 , ccr-0220264 , eia-0218443 , and aro grant w911nf-04-r-0009 .
the authors are grateful to viv kendon for helpful suggestions . | we study a natural notion of decoherence on quantum random walks over the hypercube .
we prove that this model possesses a decoherence threshold beneath which the essential properties of the hypercubic quantum walk , such as linear mixing times , are preserved . beyond the threshold ,
we prove that the walks behave like their classical counterparts . |
it is now widely accepted that ` feedback ' from radio - loud active galactic nuclei ( agn ) keeps the bulk of the diffuse x - ray gas hot in the cores of clusters of galaxies ( e.g. rafferty et al .
2008 ; cavagnolo et al .
although the underlying plasma microphysics are complex ( mcnamara & nulsen 2007 ) , it is also clear that agn heating is not always and everywhere 100 per cent effective .
some gas does succeed in cooling , perhaps in an episodic way when the agn is in quiescence , and accumulates in a reservoir of cool ionized and molecular gas ( e.g. edge 2001 ; edge et al .
2002 ; johnstone et al .
2007 ) , giving rise to star formation ( crawford et al . 1999 ; odea et al . 2008 ) .
the associated emission line nebulosity has been the subject of many investigations over the years , focussing variously on the kinematics of the gas , the origin of its filamentary morphology , its relationship to the radio source , the role of galaxy interactions and mergers , and the nature of the excitation source .
progress on the latter front came with the simulations of ferland et al .
( 2009 ) , who showed that extra heating by dissipative mhd processes or non - thermal excitation by high energy cosmic rays could account for the unusually strong ro - vibrational 2 lines and other emission line ratios which are at odds with the predictions of photoionization models .
an issue that has seldom been addressed is whether this cool gas reservoir plays an active role in the feedback cycle or whether it is just an interesting byproduct thereof , serving merely to show that cooling and agn feedback are taking place but of no greater significance .
it is of particular interest to know whether the cool gas actually fuels the accretion on to the supermassive black hole . for many years
, it has been hypothesized that ` hot feedback ' in the form of quasi - spherical bondi accretion of partially - cooled x - ray gas could be the mechanism at work ( e.g. allen et al .
2006 ; narayan & fabian 2011 ) .
this model has , however , encountered problems and appears to be energetically infeasible except in low - power systems ( e.g. soker 2010 ; rafferty et al .
2006 ; mcnamara et al .
2011 ) , leading to the development of the ` cold feedback ' model in which the agn accretes directly from the cool gas reservoir ( see pizzolato & soker 2005 , 2010 and references therein ) .
non - linear overdensities in the x - ray gas throughout the inner 30 are assumed to cool and flow steadily inwards .
blobs with an overdensity exceeding a critical factor @xmath4 relative to the ambient medium undergo catastrophic cooling , contract and enter a free - fall plunge over the final 10 , creating a reservoir of cool gas for star formation and black hole accretion .
pizzolato & soker ( 2010 ) demonstrated that the infalling blobs efficiently lose angular momentum via drag forces and mutual collisions , thereby removing a potential obstacle to the accretion .
the means by which the molecular gas couples to the agn output and thereby regulates the cold feedback process are , however , unclear ; outflows driven by jets or collimated winds may be involved ( e.g. soker & pizzolato 2005 ; wagner & bicknell 2011 ) . aspects of the cold feedback model are steadily gaining observational support . in ngc 1275 ,
lim , yiping & dinh - v - trung ( 2008 ) observed filaments of co emission extending out to radii of 8 , coincident with the coolest x - ray gas either side of the radio cavities .
this cool molecular gas is in gravitational free - fall and appears to feed a disk of hot 2 at radii @xmath5 ( wilman et al .
2005 ; mcgregor et al .
2007 ) , thereby closing the feedback loop . in a study with near - infrared integral field spectroscopy , wilman et al .
( 2009 ) observed three well - known local cooling flow clusters : a1664 , a2204 and pks 0745 - 191 .
the findings were consistent with the three systems being captured at distinct phases of the cold feedback cycle .
building on these studies , we now present k - band integral field spectroscopy ( ifs ) of the cental cluster galaxy in the nearby cluster 2a 0335 + 096 , focussing on the 2 line emission .
the observations were conducted with the near - infrared integral field spectrograph ( nifs ) on the _ gemini north _ telescope , used in conjunction with the altair adaptive optics facility to maximise spatial resolution in the circumnuclear regions of this low redshift system ( z=0.0349 ) . optical ifs by hatch et al .
( 2007 ) showed that the emission arises in a bar with twin peaks within 1 of the nucleus of the brightest cluster galaxy ( bcg ) .
this bar is part of a diffuse stream of extending over 20 ( 14 ) , to and beyond a companion cluster elliptical galaxy 4 away in projection , for which donahue et al .
( 2007 ) presented narrow - band imaging and long - slit spectroscopy and archival ultraviolet imaging , building on earlier and multi - colour imaging by romanishin & hintzen ( 1988 ) .
they confirmed that the redshifts of the bcg and its companion lie within 100 of each other and that the pair are probably interacting .
the estimated timescale for the interaction is comparable with the age of the radio source , suggesting that the latter was triggered by the interaction .
both galaxies exhibit co emission with interferometry by edge & frayer ( 2003 ) revealing that the bulk of it , @xmath6 , is situated within the central 3.5 of the bcg .
the latest _
chandra _ x - ray and radio observations ( sanders , fabian & taylor 2009 ) showed that the x - ray emission from the coolest 0.5 gas exhibits a close spatial correspondence with the stream , with the agn radio emission emerging perpendicular to it .
mid - infrared photometry and spectroscopy with _
spitzer _ have yielded measurements of the star formation rate in this system of 2.1 ( odea et al .
2008 ) and 0.7 ( donahue et al . 2011 ) , respectively .
observations of 2a 0335 + 096 with _
gemini_-nifs were performed in queue mode on 2009 october 10 , 11 and 13 yielding a total of 17 useable on - source exposures each of 10 minutes duration .
sky subtraction was enabled via an object - object - sky exposure nodding pattern using a sky region 30 from the bcg .
four additional on - source observations for the programme were obtained on 2009 september 8 and october 12 but were not included in our analysis due to technical problems during the observations . the altair adaptive optics ( ao )
unit was used with a laser guide star , operating under natural seeing in the range 0.20.5 .
nifs operates on the image slicing principle with a @xmath7 field of view sampled by 0.103-wide slitlets and a pixel scale of 0.043 along the slits . a position angle of 335 degrees
was used to align the slits with the major axis of the bcg .
the k - band grating was used , spanning 1.992.40 at a dispersion of @xmath8 per pixel , corresponding to a velocity resolution of 60 .
the data were processed in iraf using the gemini and nifs - specific routines .
the task nftelluric was used to correct for the atmospheric absorption via a standard star , and the task nifcube was used to generate a single datacube for each of the on - source observations with the default re - sampling to 0.05 square pixels .
the ao - corrected point spread function ( psf ) was measured using the telluric standard stars and in all cases was found to be sharply peaked with a full width at half maximum ( fwhm ) of 0.11 along the slitlets ; in the perpendicular direction the 0.103 slit - width undersamples the diffraction limit and the psf is consequently less sharply peaked , with a fwhm of approximately 0.19 .
the encircled energy fractions within radii of 0.1 and 0.2 were 43 and 63 per cent , respectively .
the individual on - source science frames were median combined to produce a single datacube .
the latter was re - sampled to 0.1 pixels for compatibility with the psf and enhanced signal - to - noise .
the spectra were not flux calibrated .
idl was used for all subsequent analysis . during the latter , residuals from sky emission line subtraction
were found across the entire field of view in two adjacent spectral pixels at 2.191 , within the profile of an 2 emission line in 2a 0335 + 096 .
the feature was present in all the individual datacubes and was mitigated by interpolating across pixels either side of the glitch . in a cosmology with @xmath9 , @xmath10 and
@xmath11 the spatial scale in 2a 0335 + 096 ( z=0.0349 ) is 700 per arcsec .
the only detectable emission lines in the nifs spectra are part of the v=1 - 0 s ro - vibrational series of 2 , namely s(1 ) , s(2 ) and s(3 ) .
the s(3 ) line is redshifted into an atmospheric absorption band so our analysis focusses on the remaining ortho transition , s(1 ) at 2.1218 ( rest ) which is redshifted to a clean region of transmission at 2.195 .
the results derived from the automated fitting of a single gaussian to the 2 v=1 - 0 s(1 ) emission line in the individual spaxels are shown in fig .
[ fig:2ah2 ] . the 2 comprises two prominent clumps straddling the nucleus some 600 to the north - west and south - east , the latter perhaps splitting into two sub - clumps .
weaker emission extends from both clumps towards the nucleus and diffuse emission envelops the entire structure , particularly on the south - western side . in fig .
[ fig:2ahst ] we overlay contours of the 2 emission on an _ hubble space telescope _ wfpc2 f606w image of 2a 0335 + 096 .
this highlights a region of dust absorption 11.5 south of the nucleus at the lower edge of the nifs field and the companion elliptical galaxy 4 away in projection .
the 2 emission is oriented roughly perpendicular on the plane of the sky to the weak lobes of 5 ghz radio emission ( sanders et al .
2009 ) which extend for 12 ( 8.4 ) either side of the nucleus .
the 2 kinematics exhibit an ordered velocity gradient of more than 300 over the nifs field of view .
the line width mostly lies in the range 200400 fwhm , but rises to a peak of 800 fwhm in the nuclear spaxel and a ridge of elevated velocity dispersion ( 400600 fwhm ) runs from there through the middle of the south - eastern clump .
a second ridge of elevated fwhm lies in the lower right of fig .
[ fig:2ah2 ] , but lies near the edge of the detected emission and is likely to be of lower statistical significance , although hints of a similar feature are apparent in the observations of hatch et al .
( 2007 ) . in fig .
[ fig:2ah2kinemcut ] we show an 2 rotation curve and a position - velocity diagram along the major axis of the 2 emission . .
lower : position velocity diagram for 2 v=1 - 0 s(1 ) along the same axis ( the dashed line indicates the position of the nucleus).,title="fig:",width=245 ] .
lower : position velocity diagram for 2 v=1 - 0 s(1 ) along the same axis ( the dashed line indicates the position of the nucleus).,title="fig:",width=245 ] there is a close morphological and kinematic correspondence between this 2 emission and the double - peaked bar of emission seen in the ifs data of hatch et al .
( 2007 ) at lower spatial resolution . similarly , there is also close agreement between the global 2 kinematics and those of the co emission : in single - dish observations of co(2 - 1 ) emission edge ( 2001 ) measured a linewidth of 390 fwhm , offset by 175 from an assumed zeropoint redshift of z=0.0338 , which translates to 145 in our assumed frame ( z=0.0349 ) i.e. the mid - point of the rotation curve in fig .
[ fig:2ah2kinemcut ] . interferometry by edge & frayer ( 2003 ) demonstrated that the co(1 - 0 ) emission arises from both the bcg and the companion galaxy , with the bulk of the emission residing in the former a mass of @xmath6 of cool molecular gas on a spatial scale @xmath12 .
given the close spatio - kinematic correspondence , it is reasonable to assume that the , 2 and co emission are all tracing the same gas system . in fig .
[ fig:2ah2cocomp ] we compare the 2 v=1 - 0 s(1 ) line profile integrated over the nifs aperture with the single dish co(10 ) profile from edge ( 2001 ) .
whilst there are some differences in detail such as the strong narrow 2 feature from the north - western clump at + 300 the two lines span the same overall range in velocity , although proportionally more co emission arises at larger velocities . the latter may originate from diffuse emission within the bcg , possibly in the form of a jet - driven outflow , as discussed further in section 4.2 .
indeed , the spatially - unresolved , compact co(10 ) emission found by edge & frayer ( 2003 ) , which is likely to correspond directly to the 2 we see with nifs , accounts for only 60 per cent of the integrated single - dish line intensity measured by edge ( 2001 ) within the 21.6 iram beam .
there may , however , be some contribution to the single - dish flux from the bcg s companion galaxy .
postage - stamp plots of the 2 line profiles on a grid of 0.5 cells arranged according to position in the nifs field are shown in fig .
[ fig:2ah2profs ] .
wilman et al . (
2009 ) demonstrated that the 2/ line ratio serves as a good discriminant between 2 emission excited by star formation ( for which 2 v=1 - 0 s(3)/ @xmath13 ) and the non - photoionization processes advocated by ferland et al .
( 2009 ) ( for which 2 v=1 - 0 s(3)/ @xmath2 ) . in the long - slit spectrum of edge et al .
( 2002 ) we measured 2 v=1 - 0 s(3)/=0.79 . lies just blueward of the spectral coverage of these nifs data , so we focus instead on at 2.1661 ( rest ) which is within the spectral range .
the line is not detected anywhere in the field of view , with a @xmath14 upper limit of /2
v=1 - 0 s(1 ) @xmath15 for the integrated spectrum .
this limit is consistent with the values of /2
v=1 - 0 s(1)=0.039 and 0.078 for the ` extra heat ' and ` cosmic ray ' cases , respectively , calculated by ferland et al .
[ fig:2ah2spec ] shows spectra within the nifs @xmath7 arcsec aperture and for the @xmath16 arcsec region of nuclear emission indicated in fig .
[ fig:2ah2 ] .
there is no discernible difference in the emission line ratios of the two spectra , suggesting that the excitation source is spatially distributed rather than nucleated . when combined with the flux ratio of 2 v=1 - 0 s(1)/=0.53 from edge et al .
( 2002 ) we infer /@xmath17 .
this is significantly below the case b recombination value of 0.083 , but closer to the values of 0.058 and 0.064 expected for the ` extra heat ' and ` cosmic ray ' filament excitation models .
the expected /2
v=1 - 0 s(1 ) ratio for star - forming regions is typically much higher . for example , in moorwood & oliva ( 1988 ) , 7 of out 8 galactic nuclei optically - classified as pure hii systems have /2
v=1 - 0 s(1 ) @xmath18 ; in composite and seyfert nuclei , the ratio is of order unity .
our limit is well below this range implying that the excitation of 2 emission in 2a 0335 + 096 can not be dominated by star formation . in any case ,
the star formation rate inferred from the mid - infrared emission in 2a 0335 + 096 by odea et al .
( 2008 ) is just 2.1 for the entire galaxy , compared to @xmath19for the systems in the wilman et al .
( 2009 ) sample .
the @xmath14 upper limit on the equivalent width of is 0.8 . with reference to the evolutionary models shown in davies et al .
( 2006 ) ( their fig .
8) the latter would imply an absence of star formation for at least the past 10 , if considered as a monolithic stellar population .
the measured equivalent width of the co(2,0 ) bandhead at 2.2935 ( rest ) is 9.9 ; although this figure is below the 12 expected for most stellar populations more than 10 old ( fig .
6 of davies et al . 2006 ) , it is likely to lie within the bounds of the systematic uncertainties arising from , for example , the form of the initial mass function in these environments .
modest dilution from non - stellar sources of continuum emission remains another possibility .
although we do not detect emission within the nifs aperture , emission is present in this region in the ifu observations of hatch et al .
( 2007 ) and in the long - slit spectrum of donahue et al .
( 2007 ) . from these works we estimate an flux within the @xmath7 arcsec nifs aperture of @xmath20 , equating to an luminosity of @xmath21 . using the kennicutt ( 1998 ) star formation rate relation ,
this would translate to a star formation rate of 0.11 if all the were due to star formation .
corrections for galactic and intrinsic dust extinction ( each approximately 1 magnitude ; donahue et al .
2007 ) would boost the estimates by a factor of 6 . as noted by donahue et al .
( 2007 ) , the extinction - corrected star formation rate for the entire galaxy inferred from the emission is 1520 and that inferred from the excess uv emission is of a similar order of magnitude .
these estimates would be compatible with the mid - infrared estimate of 2.1 ( odea et al .
2008 ) if only 10 per cent of the emission were due to star formation . for comparison , a fit by donahue et al .
( 2011 ) to the mid - infrared emission features with an hii region photodissociation region model yielded a star formation rate of 0.7 .
the low level of ongoing circumnuclear star formation corroborates the findings of romanishin & hintzen ( 1988 ) .
their analysis of the b
i colour gradient of the bcg showed that star formation is occurring at radii 430 , as evidenced by a plateau in b
i bluer than in non - cooling flow bcgs ; the colour reddens at smaller radii , suggesting a decline in the star formation rate or dust extinction .
the latter explanation is ruled out by our observations since the extinction in the k - band corresponding to the observed reddening ( @xmath22(b i ) = 0.15 ) would be completely negligible .
we conclude that star formation does not make a significant contribution to the excitation of the 2 gas within the nifs field - of - view . in conjunction with the size contraint ( @xmath12 ) on the co emission reservoir from edge & frayer ( 2003 )
, this implies that the star formation is spatially offset from the bulk of the molecular gas .
we will now address the implications of this finding . ) ; the latter is shown scaled up by a factor of 50 and both spectra were boxcar smoothed by 2 pixels in dispersion .
the prominent 2 v=1 - 0 s(1 ) , s(2 ) and s(3 ) emission lines are indicated , as is the expected position of the undetected emission line .
the portions of the spectra around 2.166 were excised due to the presence of emission in the telluric standard.,width=245 ]
based on the evidence presented in section 3 , the observed 2 emission appears to consist of a rotating structure .
this could take the form of a flared disk or a ring of 2 , or the 2 peaks could correspond to gas on highly - elliptical orbits which spends most of its time at large radii .
given its general alignment with the 14 stream of emission discussed in section 1 , it is reasonable to infer that the 2 has been accreted from this stream and may provide a fuel source for the agn . indeed , the rotational axis of the 2 aligns with the position angle of the present lobes of radio emission , at least in projection .
we now discuss this interpretation further in the context of the cold feedback model .
donahue et al . ( 2007 ) argued that the bcg and its nearby companion are interacting and probably gravitationally bound to one another .
this does not , however , imply that the observed stream was stripped from the companion . for the case of a1664 ,
kirkpatrick et al . ( 2009a ) argued that the stripping of gas from infalling companion galaxies was an unlikely origin for the cool gas seen in the cluster core and this generic argument also applies to 2a 0335 + 096 . under a stripping scenario , it would in any case be hard to understand why there should be a good spatial correspondence between the accretion stream and the coolest x - ray gas ( sanders et al .
we speculate that there are at least three possible explanations for the association of the stream and the companion galaxy .
firstly , the disturbance created by the motion of the companion could generate non - linear overdensities in the icm ( of a factor 2 or more ) which can readily cool , as proposed in the cold feedback model of pizzolato & soker ( 2005 ) .
secondly , small amounts of dense molecular gas and dust stripped from the companion may offer seeding sites onto which the surrounding icm can condense and cool , as argued by sparks et al .
( 2004 ) in the case of m87 .
thirdly , the companion could have merely disturbed the pre - existing cool gas reservoir of the bcg , as argued by wilman et al .
( 2006 ) for several other bcgs .
alternatively , the companion could be just a chance projection
. there may be some parallels with the radio galaxies 3c 31 and 3c 264 which both exhibit double - horned co profiles from @xmath23 of molecular hydrogen , possibly acquired through cannibalism with gas - rich galaxies over the past few tens of ( lim et al .
2000 ) .
once within the bcg , the accretion stream will interact with the local interstellar medium . depending on the relative velocity of impact ,
this collision may generate shocks and disturbed kinematics . the locally elevated 2 velocity dispersion in the south - eastern 2 peak and nearby dust feature ( fig . [ fig:2ahst ] ) may be signatures of just such an interaction .
hydrodynamic interaction may lead to a loss in angular momentum of the accreted material , driving gas into the circumnuclear regions as a potential fuel source for the agn .
mutual collisions among the 2 cloud ensemble in its dense inner regions provide a further mechanism for angular momentum loss , as argued by pizzolato & soker ( 2010 ) . whilst the smooth rotation curve of fig .
[ fig:2ah2kinemcut ] suggests that the 2 is rotating in the potential of the bcg , the existence of two distinct 2 peaks could suggest that the 2 clouds are on highly elliptical orbits , with the peaks corresponding to their apastrons ( see e.g. fig .
9 of pizzolato & soker 2010 ) .
the distortions of the 2 isovelocity contours in fig .
[ fig:2ah2 ] point to departures from simple circular motions , possibly in the form of disk warping and elliptical orbits .
we note a resemblance to the twin peaks of molecular gas which are commonly observed within the central kiloparsec of barred and starburst galaxies ( e.g. kenney et al .
1992 ; jogee et al . 2005 ) .
such peaks result from gas being driven into the inner lindblad resonance ( ilr ) of the barred potential , which is associated with the transition from the highly elongated family of x@xmath24 loop orbits to the smaller and more circular x@xmath25 orbit family .
the x@xmath24 orbits are aligned with the major axis of the bar , whose leading edges typically exhibit radial dust - lanes along which gas flows into the nuclear regions . whilst there is no obvious evidence for a bar in the bcg , the presence of the nearby companion galaxy , or triaxility intrinsic to the bcg itself , may induce a non - axisymmetric potential with analogous dynamical consequences for the accreted gas .
we are , however , unable to infer whether the observed 2 peaks correspond to gas flowing along the major axis of a bar in the early stages of inflow , or whether the material has already been funelled down to the ilr , and do not pursue this interpretation any further .
given that the accreted 2 appears to be a viable fuel source for the agn , we now discuss the associated timescales and energetics .
the current outburst of the radio source gave rise to the observed 8.4-long lobes in the 5 ghz image of sanders et al .
( 2009 ) . the estimated age of this radio emission is 25 - 50 ( donahue et al .
if on circular orbits , the orbital timescale for the rotating 2 structure is 25 for a radius of 600 and a velocity of 150 ( fig .
[ fig:2ah2kinemcut ] ) , neglecting projection effects .
as sanders et al .
( 2009 ) discussed , the enthalpy of the observed cavities in the cluster core ( @xmath26 ) is sufficient to offset the luminosity within the 122 x - ray cooling radius for a further @xmath27 , i.e. for a few dynamical times of the 2 . on a similar timescale , we would expect that the central supermassive black hole would be able to accrete some of the 2 in order to sustain the feedback .
the energetics of the outburst arising from accretion from the 2 reservoir are , of course , dependent on how much of the gas can be accreted by the black hole and how much is instead channelled into star formation .
to maintain strict conformity with the black hole : bulge mass scaling relation of magorrian et al .
( 1998 ) , 700 of star formation would be expected for every 1 accreted by the black hole .
applied to 2a 0335 + 096 , accretion of 1 part in 700 of the @xmath6 2 reservoir at the canonical efficiency of @xmath28 would produce an outburst of @xmath29 i.e. comparable in size to the present cavity system measured by sanders et al .
( 2009 ) which could offset cooling for @xmath30 . whilst the rates of star formation and black hole growth for a repesentative sample of clusters are in line with such expectations statistically , the sizeable scatter shows that they are not growing in precise lockstep in any given system ( rafferty et al .
2006 ) .
alternative possibilities stem from the low levels of star formation in the circumnuclear region , which may allow a larger fraction of the molecular gas to be accreted . as noted in section 3.3 , the current star formation rate within the nifs aperture , which encompasses most if not all of the @xmath6 molecular gas reservoir , appears to be no more than 1 . at this rate
, it would take at least 1 for the gas to be consumed .
the accretion of just @xmath1 per cent of the gas onto a moderately - spinning black hole would suffice to offset cooling in the cluster core over the same period .
this would imply that the agn would be essentially on for @xmath2 , either continuously or in a series of mini - outbursts .
the viability of this scenario hinges firstly on whether this current low level of star formation can be sustained , and secondly on whether the gas can be prevented from accreting much more quickly , given that the orbital timescale of the observed 2 is just 25 . indeed , from a comparison of the relative 2 line strengths in the nuclear and integrated spectra ( fig .
[ fig:2ah2spec ] ) , we infer that @xmath31 per cent of the molecular gas ( @xmath32 ) is already within 100 of the nucleus .
these two issues go to the heart of understanding the mechanisms which regulate cold feedback , on which we speculate below .
one explanation could be that the current radio outburst may have initially evacuated gas from the inner regions or otherwise suppressed star formation ; the resolved lateral extent of the current 5 ghz radio source ( i.e. perpendicular to the inferred jet direction ) in sanders et al .
( 2009 ) is @xmath33 ( 3.5 ) . the molecular gas could be supported by the associated magnetic field and thus prevented from collapsing to form stars , as argued by fabian et al .
( 2008 ) and ferland et al .
( 2009 ) for the filaments in ngc 1275 and the excitation of their line emission .
it is also possible that the current star formation rate is unrepresentative of the average value on gyr timescales ; the star formation rate could increase and consume the gas more quickly in the future .
magnetic support of the gas , if sufficiently strong , could hinder the accretion onto the black hole .
outflows driven by the radio source provide a further means to disrupt its fuel supply . in section 3.2
, we noted that some of the diffuse co emission within the bcg is found at relatively high velocity , suggestive of an outflow .
less than 1 per cent of the current outburst enthalpy would suffice to accelerate @xmath34 of molecular gas to a velocity of 600 , so the energetics are feasible and similar to the fraction of the outburst energy required to drive the metal outflows in clusters ( kirkpatrick et al . 2009b , 2011 ) .
if such mechanisms are , however , not effective in regulating the accretion and the bulk of the 2 is instead accreted rapidly by the black hole over several orbital timescales , this could generate an outburst exceeding @xmath3 ( for @xmath28 ) .
such an outburst would be comparable with most powerful currently known , in ms0735.6 + 7421 ( mcnamara et al .
2005 ) and hydra a ( wise et al .
further simulations of cold feedback including the effects of magnetic fields and outflows on the molecular gas are required to discriminate between such possibilities .
not withstanding these uncertainties , the new observations of 2a 0335 + 096 presented here , revealing an ample supply of cool molecular gas available to fuel the radio source on the required timescale of 25 - 100 and offset icm cooling , further bolster the case for cold agn feedback .
based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia e tecnologia ( brazil ) and ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) .
program i d gn-2009b - q-69 .
brm acknowledges generous financial support from the canadian space agency space science enhancement program , and from the natural sciences and engineering research council of canada s discovery grant program .
allen s.w . ,
dunn r.j.h .
, fabian a.c . , taylor g.b .
, reynolds c.s . , 2006 , mnras , 372 , 21 cavagnolo k.w . , donahue m. , voit g.m . , sun m. , 2008 , apj , 683 , l107 crawford c.s . ,
allen s.w . , ebeling h. , edge a.c .
, fabian a.c
. , 1999 , mnras , 306 , 857 davies r.i . ,
et al . , 2006 , apj , 646 , 754 donahue m. , sun m. , odea c.s . , voit g.m . ,
cavagnolo k.w . , 2007 , apj , 134 , 14 donahue m. , de messires g.e .
, oconnell r.w .
, voit g.m . ,
hoffer a. , mcnamara b.r . ,
nulsen p.e.j . , 2011 ,
apj , 732 , 40 edge a.c . , 2001 ,
mnras , 328 , 762 edge a.c . , wilman r.j . ,
johnstone r.m .
, crawford c.s . , fabian a.c . ,
allen s.w . , 2002 , mnras , 337 , 49 edge a.c . , frayer d.t . , 2003 ,
apj , 594 , l13 fabian a.c . , johnstone r.m . , sanders j.s . ,
conselice c.j .
, crawford c.s . , gallagher j.s .
iii , zweibel e. , 2008 , nature , 454 , 968 ferland g.j .
, fabian a.c . ,
hatch n.a . ,
johnstone r.m . ,
porter r.l . , van hoof p.a.m .
, williams r.j.r . , 2009 , mnras , 392 , 1475 hatch n.a . , crawford c.s . , fabian a.c . , 2007 , mnras , 380 , 33 jogee s. , scoville n. , kenney j.d.p . , 2005 , apj , 630 , 837 johnstone r.m
, hatch n.a . , ferland g.j . , fabian a.c .
, crawford c.s . , wilman r.j . , 2007 , mnras , 382 , 1246 kenney j.d.p . , wilson c.d . ,
scoville n.z . ,
devereux n.a .
, young j.s . , 1992 ,
apj , 395 , l79 kennicutt r.c . , jr . , 1998 , ara&a , 36 , 189 kirkpatrick c.c . , mcnamara b.r . ,
rafferty d.a . , nulsen p.e.j . , birzan l. , kazemzadeh f. , wise m.w . , gitti m. , cavagnolo k. , 2009a , apj , 697 , 867 kirkpatrick c.c . , gitti m. , cavagnolo k. , mcnamara b.r . , david l.p .
, nulsen p.e.j . , wise m.w . , 2009b , apj , 707 , l69 kirkpatrick c.c . , cavagnolo k. , mcnamara b.r . , 2011 , apj , 731 , l23 lim j. , leon s. , combes . , dinh - v - trung , 2000 , apj , 545 , l93 lim j. , yiping a. , dinh - v - trung , 2008 , apj , 672 , 252 magorrian j. , et al .
, 1998 , aj , 115 , 2285 mcgregor p. , dopita m. , sutherland p. , beck t. , storchi - bergmann t. , 2007 , apss , 311 , 223 mcnamara b.r , nulsen p.e.j . , wise m.w . , rafferty d.a .
, carilli c. , sarazin c.l . , blanton e.l
. , 2005 , nature , 433 , 45 mcnamara b.r , nulsen p.e.j .
, 2007 , ara&a , 45 , 117 mcnamara b.r , rohanizadegan m. , nulsen p.e.j . , 2011 ,
apj , 727,39 moorwood a.f.m . , oliva e. , 1988 , a&a , 203 , 278 narayan r. , fabian , a.c . , 2011 ,
arxiv:1105.0594 , mnras accepted odea c.p . , et al . , 2008 , apj , 681 , 1035 pizzolato f. , soker n. , 2005 , apj , 632 , 821 pizzolato f. , soker n. , 2010 , mnras , 408 , 961 rafferty d.a .
, mcnamara b.r . ,
nulsen p.e.j .
, wise m.w . , 2006 ,
apj , 652 , 216 rafferty d.a . ,
mcnamara b.r .
, nulsen p.e.j .
, 2008 , apj , 687 , 899 romanishin w. , hintzen p. , 1988 ,
apj , 324 , l17 sanders j.s .
, fabian a.c . ,
taylor g.b . , 2009 , mnras , 396 , 1449 soker n. , pizzolato f. , 2005 , apj , 622 , 847 soker n. , 2010 , mnras , 407
, 2355 sparks w.b . , donahue m. , jordan a. , ferrarese l. , cote p. , 2004 ,
apj , 607 , 294 wagner a.y . ,
bicknell g.v . , 2011 ,
apj , 728 , 29 wilman r.j . ,
edge a.c . ,
johnstone r.m . , 2005 , mnras , 359 , 755 wilman r.j . , edge a.c .
, swinbank a.m. , 2006 , mnras , 371 , 93 wilman r.j . , edge a.c . ,
swinbank a.m. , 2009 , mnras , 395 , 1355 wise m.w . ,
mcnamara b.r .
, nulsen p.e.j .
, houck j.c . , david l.p . , 2007 ,
apj , 659 , 1153 | we present adaptive optics - assisted k - band integral field spectroscopy of the central cluster galaxy in 2a 0335 + 096 ( z=0.0349 ) .
the 2 v=10 s(1 ) emission is concentrated in two peaks within 600 of the nucleus and fainter but kinematically - active emission extends towards the nucleus .
the 2 is in a rotating structure which aligns with , and appears to have been accreted from , a stream of emission extending over 14 towards a companion galaxy .
the projected rotation axis aligns with the 5 ghz radio lobes .
this 2 traces the known @xmath0 co - emitting reservoir ; limits on the emission confirm that the 2 emission is not excited by star formation , which occurs at a rate of less than 1 in this gas .
if its accretion onto the black hole can be regulated whilst star formation remains suppressed , the reservoir could last for at least 1 ; the simultaneous accretion of just @xmath1 per cent of the gas could drive a series of agn outbursts which offset x - ray cooling in the cluster core for the full @xmath2 . alternatively ,
if the regulation is ineffective and the bulk of the 2 accretes within a few orbital periods ( 25100 ) , the resulting @xmath3 outburst would be among the most powerful cluster agn outbursts known . in either case , these observations further support cold feedback scenarios for agn heating . #
1to 0pt#1 # 1 # 1 to size # 1 to size = cmr10 scaled2 = cmbx10 scaled2 = cmti10 scaled2 ps .
2 = 6truein galaxies : active galaxies : clusters : individual : 2a 0335 + 096 cooling flows intergalactic medium |
_ program title : _ ` mnpbem ` toolbox supplemented by a collection of demo files + _ programming language : _ matlab 7.11.0 ( r2010b ) + _ computer : _ any which supports matlab 7.11.0 ( r2010b ) + _ operating system : _ any which supports matlab 7.11.0 ( r2010b ) + _ ram required to execute with typical data : _ @xmath0
gbyte + _ has the code been vectorised or parallelized ? : _
yes + _ keywords : _ plasmonics , electron energy loss spectroscopy , boundary element method + _ cpc library classification : _ optics + _ external routines / libraries used : _ ` mesh2d ` available at ` www.mathworks.com ` + _ nature of problem : _ simulation of electron energy loss spectroscopy ( eels ) for plasmonic nanoparticles + _ solution method : _ boundary element method using electromagnetic potentials + _ running time : _ depending on surface discretization between seconds and hours +
plasmonics has emerged as an ideal tool for light confinement at the nanoscale @xcite .
this is achieved through light excitation of coherent charge oscillations at the interface between metallic nanoparticles and a surrounding medium , the so - called _ surface plasmons _ , which come together with strongly localized , evanescent fields .
while the driving force behind plasmonics is downscaling of optics to the nanoscale , conventional optics can not be used for mapping of plasmonic fields because to the abbe diffraction limit of light . to overcome this limit , various experimental techniques , such as scanning near field microscopy or scanning tunneling spectroscopy @xcite ,
have been employed . in recent years , electron energy loss spectroscopy ( eels )
has become an extremely powerful experimental device for the minute spatial and spectral investigation of plasmonic fields @xcite . in eels ,
electrons with a high kinetic energy pass by or penetrate through a metallic nanoparticle , excite particle plasmons , and lose part of their kinetic energy . by monitoring this energy loss as a function of electron beam position , one obtains a detailed map about the localized plasmonic fields @xcite .
this technique has been extensively used in recent years to map out the plasmon modes of nanotriangles @xcite , nanorods @xcite , nanodisks @xcite , nanocubes @xcite , nanoholes @xcite , and coupled nanoparticles @xcite ( see also refs .
@xcite for the interpretation of eels maps ) .
simulation of eel spectra and maps has primarily been performed with the discrete - dipole approximation @xcite and the boundary element method ( bem ) approach @xcite . within the latter scheme ,
the boundary of the metallic nanoparticle becomes discretized by boundary elements , and maxwell s equations are solved by attaching artificial surface charges and currents to these elements which are chosen such that the proper boundary conditions are fulfilled @xcite .
the methodology for eels simulations within the bem approach has been developed in refs .
@xcite , and has been successfully employed in comparison with experimental eels data @xcite .
the purpose of the eels software described in this paper is to allow for a simple and efficient computation of electron energy loss spectroscopy of plasmonic nanoparticles and other nanophotonic structures .
the software consists of two classes ` eelsret ` and ` eelsstat ` devoted to the simulation of eel spectroscopy and mapping of plasmonic nanoparticles , see fig .
[ fig : flowchart ] , which can be used in combination with the ` mnpbem ` toolbox @xcite that provides a generic simulation platform for the solution of maxwell s equations .
our implementation for eels simulation of plasmonic nanoparticles relies on a bem approach @xcite that has been successfully employed in various studies @xcite .
a typical simulation scenario consists of the following steps . 1 .
first , one sets up the particle boundaries and the dielectric environment within which the nanoparticle is embedded .
this step has been described in detail in our previous ` mnpbem ` paper @xcite .
2 . we next initialize an ` eelsret ` or ` eelsstat ` object which defines the electron beam .
this object stores the beam positions and the electron velocity . for a given electron loss energy
, it then returns the external scalar and vector potentials @xmath1 and @xmath2 which can be used for the solution of the bem equations @xcite .
3 . for given @xmath1 and @xmath2
, we solve the full maxwell equations or its quasistatic limit , using the classes ` bemret ` or ` bemstat ` of the ` mnpbem ` toolbox @xcite .
the solutions are provided by the surface charge and current distributions @xmath3 and @xmath4 , which allow to compute the potentials and fields at the particle boundary and everywhere else @xcite ( using the green function of the helmholtz equation ) .
finally , we use @xmath3 and @xmath4 to compute the electron energy loss probabilities , which can be directly compared with experimental eel data . rather than providing the additional classes separately ,
we have embedded them in a new version of the ` mnpbem ` toolbox which supersedes the previous version @xcite .
the main reason for this policy is that also the mie classes ` mieret ` and ` miestat ` had to be modified , which allow a comparison with analytic mie results and can be used for testing .
the new version of the toolbox also corrects a few minor bugs and inconsistencies .
however , we expect that all simulation programs that performed with the old version should also work with the new version .
we have organized this paper as follows . in sec .
[ sec : start ] we discuss how to install the toolbox and give a few examples demonstrating the performance of eels simulations .
the methodology underlying our approach as well as a few implementation details are presented in sec .
[ sec : theory ] . finally , in sec .
[ sec : results ] we present results of our eels simulations and provide a detailed toolbox description .
to install the toolbox , one must simply add the path of the main directory ` mnpbemdir ` of the ` mnpbem ` toolbox as well as the paths of all subdirectories to the matlab search path .
this can be done , for instance , through addpath(genpath(mnpbemdir ) ) ; to set up the help pages , one must once change to the main directory of the ` mnpbem ` toolbox and run the program ` makemnpbemhelp ` > > cd mnpbemdir ; > > makemnpbemhelp ; once this is done , the help pages , which provide detailed information about the toolbox , are available in the matlab help browser .
note that one may have to call ` start > desktop tools > view start button configuration > refresh ` to make the help accessible . under matlab 2012 the help pages can be found on the start page of the help browser under _ supplemental software_. the toolbox is almost identical to our previously published version @xcite .
the only major difference concerns the inclusion of eels simulations , which will be described in more detail in this paper .
the ` mnpbem ` toolbox comes together with a directory ` demoeels ` containing several demo files . to get a first impression
, we recommend to work through these demo files . by changing to the demo directory and typing > > demotrianglespectrum at the matlab
prompt , a simulation is performed where the eel spectra are computed for a triangular silver nanoparticle . the run time
is reported in table [ table : examples ] , and the simulation results are shown in fig .
[ fig : trianglespectrum ] .
one observes a number of peaks associated with the different plasmon modes of the nanoparticle .
note that through ` plot(p,'edgecolor','b ' ) ` one can plot the nanoparticle boundary . by running next ` demotrianglemap.m ` we obtain the spatial eels maps at the plasmon resonance energies indicated by dashed lines in fig .
[ fig : trianglespectrum ] . figure [ fig : trianglemap](a ) reports the map for the degenerate dipolar modes , whereas panels ( b d ) show the eels maps for higher excited plasmon modes ( see ref .
@xcite for experimental results ) . are computed . ]
lrx demo program & runtime & description + ` demomie.m ` & 26.3 sec & comparison of bem simulations with analytic mie results + ` demomiestat.m ` & 14.7 sec & same as ` demomie.m ` but for quasistatic limit + ` demodiskspectrum.m ` & 29.4 sec & eel spectra for nanodisk at selected beam positions + ` demodiskspectrumstat.m ` & 14.6 sec & same as ` demodiskspectrum.m ` but for quasistatic limit + ` demodiskmap.m ` & 105.6 sec & spatial eels maps for nanodisk at selected loss energies + ` demotrianglespectrum.m ` & 173.5 sec & eel spectra for nanotriangle at selected beam positions + ` demotrianglemap.m ` & 104.6 sec & spatial eels maps for nanotriangle at selected loss energies + at the plasmon resonances indicated by dashed lines .
the maps have been computed with the demo program ` demotrianglemap.m ` , with color axes scaled to the maxima of the respective maps ( see fig . [
fig : diskmap ] for color bar ) . ] in the following we briefly discuss the demo file ` demotrianglespectrum.m ` ( see sec .
[ sec : results ] for a detailed description of the software ) .
we first set up a ` comparticle ` object ` p ` for the nanotriangle , which stores the particle boundary and the dielectric materials , as well as a bem solver ` bem ` for the solution of maxwell s equations .
these steps have been described in some length in a previous paper @xcite .
additional information is also provided by the help pages . to set up the eels simulation , we need the impact parameters of the electron beam , a broadening parameter for the triangle integration ( see sec . [
sec : penetrate ] for details ) , and the electron velocity in units of the speed of light .
initialization is done through b = [ - 45 , 0 ; 0 , 0 ; 26 , 0 ] ; vel = 0.7 ; width = 0.2 ; exc = eelsret ( p , b , width , vel ) ; the ` exc ` object returns through ` exc(enei ) ` the external potentials for a given loss energy , which allow for the solution of the bem equations by means of artificial surface charges and currents ` sig ` @xcite . from `
sig ` we can obtain the surface and bulk loss probabilities for the electron .
sig = bem exc ( enei ) ; [ psurf , pbulk ] = exc.loss ( sig ) ; finally , the eel spectrum can be computed by performing a loop over loss energies , and eel maps can be obtained by providing a rectangular grid of impact parameters .
a more detailed description of the eels classes will be given in sec .
[ sec : results ] .
for the sake of completeness , we start by briefly summarizing the main concepts of the bem approach for the solution of maxwell s equation ( see refs .
@xcite for a more detailed discussion ) .
we consider dielectric nanoparticles , described through local and isotropic dielectric functions @xmath5 , which are separated by sharp boundaries @xmath6 . throughout , we set the magnetic permeability @xmath7 and consider maxwell s equations in frequency space @xmath8 @xcite . in accordance to refs .
@xcite , we adopt a gaussian unit system .
the basic ingredients of the bem approach are the scalar and vector potentials @xmath9 and @xmath10 , which are related to the electromagnetic fields via @xmath11 here @xmath12 and @xmath13 are the wavenumber and speed of light in vacuum , respectively .
the potentials are connected through the lorentz gauge condition @xmath14 . within each medium
, we introduce the green function for the helmholtz equation defined through @xmath15 with @xmath16 being the wavenumber in the medium @xmath17 . for an inhomogeneous dielectric environment , we then write down the solutions of maxwell s equations in the _ ad - hoc _
form @xcite [ eq : adhoc ] @xmath18 where @xmath1 and @xmath2 are the scalar and vector potentials characterizing the external perturbation . owing to eq . , these expressions fulfill the helmholtz equations everywhere except at the particle boundaries . @xmath19 and
@xmath20 are surface charge and current distributions , which are chosen such that the boundary conditions of maxwell s equations at the interfaces between regions of different permittivies @xmath21 hold .
this leads to a number of integral equations .
upon discretization of the particle boundaries into boundary elements , one obtains a set of linear equations that can be inverted , thus providing the solutions of maxwell s equation in terms of surface charge and current distributions @xmath19 and @xmath20 . through eqs .
( [ eq : adhoc]a , b ) one can compute the potentials everywhere else .
for further details about the working equations of the bem approach the reader is referred to refs .
@xcite . in the following
we consider the situation where an electron passes by or penetrates through a metallic nanoparticles , and loses energy by exciting particle plasmons .
we assume that the electron kinetic energy is much higher than the plasmon energies ( for typical electron microscopes operating with electron energies of several hundreds of kev this assumption is certainly fulfilled ) .
we can thus discard in the electron trajectory the small change of velocity due to plasmon excitation , and describe the loss process in lowest order perturbation theory .
we emphasize that our approach is correspondingly not suited for low electron energies or thick samples .
for an electron trajectory @xmath22 , with @xmath23 , the electron charge distribution reads @xmath24 here @xmath25 and @xmath26 are the charge and velocity of the electron , respectively , @xmath27 is the impact parameter in the @xmath28-plane , and @xmath29 is a wavenumber .
the potentials associated with the charge distribution of eq . can be computed in infinite space analytically ( linard - wiechert potentials @xcite ) , and we obtain @xcite @xmath30 @xmath31 is the modified bessel function of order zero , and @xmath32 . within our bem approach , we can directly insert the expressions of eq .
for the unbounded medium into eq . since the calculated surface charge and current distributions @xmath19 and @xmath20 will automatically guarantee that the proper boundary conditions at the interfaces are fulfilled @xcite .
we next turn to the calculation of the electron energy loss . ignoring the small change of the electron velocity caused by the interaction with the plasmonic nanoparticle
, the energy loss can be computed from the work performed by the electron against the induced field @xcite @xmath33\,dt=\int_0^\infty \hbar\omega \gamma_{\rm eels}(\bm r,\omega)\,d\omega\,,\ ] ] with the loss probability , given per unit of transferred energy , @xmath34\bigr\}\,dt+ \gamma_{\rm bulk}(\omega)\,.\ ] ] note that eq .
is a classical expression , where @xmath35 has been introduced only to relate energy and frequency .
@xmath36 is the induced electric field , which can be computed from the potentials originating from the surface charge and current distributions @xmath19 and @xmath20 alone .
@xmath37 is the bulk loss probability for electron propagation inside a lossy medium , see eq .
( 18 ) of ref . @xcite . within the quasistatic approximtion ,
it is proportional to the loss function @xmath38 $ ] and the propagation length inside the medium .
expressions similar to eq . but derived within a full quantum approach , based on the born approximation , can be found in ref .
@xcite .
quite generally , @xmath39 can be computed by calculating the induced electric field along the electron trajectory and evaluating the expression given in eq . .
in what follows , we describe a computationally more efficient scheme .
insertion of the induced potentials of eq . into the energy loss expression of eq .
yields @xmath40dz\,.\ ] ] here @xmath41 and @xmath42 are the entrance and exit points of the electron beam in a given medium , and @xmath43 parameterizes the electron trajectory .
we next introduce a potential - like term @xmath44 , associated with the electron propagation inside a given medium . performing integration by parts
, we can rewrite the second term in parentheses of eq . as @xmath45 the first term on the right - hand side gives , upon insertion into eq .
, @xmath46 .
the integral expression precisely corresponds to the scalar potential at the crossing points of the trajectory with the particle boundary . as the potential is continuous across the boundaries , the contributions of all crossing points sum up to zero .
thus , we arrive at our final result @xmath47\,.\ ] ] in comparison to eq . , this expression has the advantage that the integration is only performed over the particle boundary , where the surface charge and current distributions @xmath19 and @xmath20 are readily available , and we do nt have to compute the induced electric field along the electron trajectory . in the calculation of the external potential , eq . , and the eels probability of eq .
the points where the electron trajectory crosses the boundary have to be treated with care .
for small distances , the potential scales with @xmath48 .
when integrating this expression within our bem approach over a small area , we find in polar coordinates that @xmath49 remains finite .
the same is true for the surface derivative of the potential . in a computational approach
it is somewhat tedious to perform such integration properly , in particular for crossing points that are located close to the edges or corners of boundary elements .
for this reason , we suggest a slightly different approach . the main idea is to replace the delta - like transversal trajectory profile @xmath50 by a smoothened distribution .
the potential at the transverse position @xmath51 then reads @xmath52\,.\ ] ] here @xmath53 [ see eq . ] and the term in brackets is our smoothing function , with @xmath54 being a parameter that determines the transversal extension . for small arguments
, we can expand @xmath55 , where @xmath56 is the euler constant , and perform all integrations analytically to obtain @xmath57 .
this suggests replacing the potential of eq . by the smoothened function @xmath58 for large arguments this expression coincides with the linard - wiechert potential , but remains finite for small arguments .
a corresponding smoothening is also performed in the potential - like function @xmath59 of eq . .
in the quasistaic approximation one assumes that the size of the nanostructure @xmath60 is significantly smaller than the wavelength of light , such that @xmath61 .
this allows us to keep in the simulations only the scalar potential and to set in the green function @xmath62 .
we are thus left with the solution of the laplace or poisson equation , rather than the helmholtz equation , but we keep the full frequency - dependence of the permittivities @xmath63 . the calculation of eels probabilities with the bem approach has been described in some detail in ref . @xcite . in the following we briefly describe the basic ingredients .
first , we compute the external potential from the solution of poisson s equation @xmath64 with @xmath65 being the charge distribution of eq . .
we next compute the surface charge distribution @xmath66 from the solution of the boundary integral equation , which , for a nanoparticle described by a single dielectric function @xmath67 embedded in a background of dielectric constant @xmath68 , reads @xcite @xmath69 here @xmath70 is the static green function and @xmath71 denotes the surface derivative , where @xmath72 is the outer surface normal of the boundary . for materials consisting of more than one material , eq .
has to be replaced by a more general expression @xcite .
finally , we compute the electron energy loss probability from ( see also eq . ( 18 ) of ref .
@xcite ) @xmath73 where @xmath37 is the bulk loss probability ( see eq . ( 19 ) of ref .
we first discuss the demo file ` demomie.m ` that simulates the energy loss probability for an electron passing by a silver nanosphere .
figure [ fig : mie ] shows results of our bem simulations which are in good agreement with analytic mie results @xcite .
let us briefly work through the demo program . in the first lines we define the dielectric materials and the nanosphere ( for a more detailed discussion of the ` mnpbem ` toolbox see ref .
@xcite ) .
epsm = epstable ( silver.dat ) ; epstab = epsconst ( 1 ) , epsm ; diameter = 80 ; p = comparticle ( epstab , trisphere ( 256 , diameter ) , [ 2 , 1 ] , 1 ) ; we next define the excitation of the electron beam . for the solution of the full maxwell equations , the excitation and the calculation of the eels probability
is performed by the ` eelsret ` class , which is initialized through exc = eelsret ( p , impact , width , vel , propertyname , propertyvalue , ... ) here ` p ` is the previously computed ` comparticle ` object , which stores the particle boundaries and the dielectric functions at both sides of the boundary . `
impact ` is a vector ` [ x , y ] ` for the impact parameter @xmath74 of the electron beam defined in eq . .
if simulations for various impact parameters are requested , as is usually the case for the simulation of eels maps , ` impact ` can also be an array ` [ x1,y1;x2,y2 ; ... ] ` . `
width ` is the broadening parameter @xmath54 of the electron beam , see eq . , which will be discussed in more detail below , and ` vel `
is the electron velocity to be given in units of the speed of light in vacuum .
the optional pairs of property names ( ` ' cutoff ' ` , ` ' rule ' ` , or ` ' refine ' ` ) and values allow to control the performance of the toolbox , as detailed in sec .
[ sec : penetrate ] . in the `
demomie.m ` program we next set up the eels excitation with b = 10 ; vel = eelsbase.ene2vel ( 200e3 ) ; [ width , cutoff ] = deal ( 0.5 , 8 ) ; exc = eelsret ( p , [ diameter / 2 + b , 0 ] , width , vel , cutoff , cutoff ) ; note that ` eelsbase.ene2vel ` allows to convert a kinetic electron energy in ev to the electron velocity in units of the speed of light in vacuum @xmath13 . in the above example
, a kinetic energy of 200 kev corresponds to a velocity of approximately @xmath75 .
we next set up the solver ` bemret ` for the solution of the bem equations @xcite and compute the loss probabilities of eq .
for various loss energies .
bem = bemret ( p ) ; ene = linspace ( 2.5 , 4.5 , 80 ) ; psurf = zeros ( size ( ene ) ) ; for ien = 1 : length ( ene ) sig = bem exc ( ev2 nm / ene ( ien ) ) ; psurf ( ien ) = exc.loss ( sig ) ; end here ` sig ` is a ` compstruct ` object that stores the surface charges and current distributions @xmath76 and @xmath77 , inside and outside the particle boundaries @xcite , as computed for the eels excitation of eq . .
with ` exc.loss(sig ) ` we finally compute the loss probabilities according to eq . .
note that ` ev2 nm ` defined in ` units.m ` allows to convert between energies given in electronvolts and wavelengths given in nanometers , the latter being the units used by the ` mnpbem ` toolbox . and @xmath2 of eq .
are integrated over the boundary element .
the boundary element integration is controlled by the ` refine ` and ` rule ` properties , as described in more detail in the text . in the figure we set `
refine=2 ` .
( b ) eel spectra for a silver nanodisk with a diameter of 60 nm and a height of 10 nm .
the impact parameters of the electron beams for the different spectra are reported in the inset , and the beam propagation direction is the @xmath78-direction perpendicular to the shaded disk .
we investigate different ` width ` parameters of 0.1 nm ( solid lines ) , 0.2 nm ( dashed lines ) , and 0.5 nm ( dashed - dotted lines ) , finding practically no differences in the results .
the ` cutoff ` parameter is set to 10 nm.,title="fig : " ] and @xmath2 of eq .
are integrated over the boundary element .
the boundary element integration is controlled by the ` refine ` and ` rule ` properties , as described in more detail in the text . in the figure we set `
refine=2 ` .
( b ) eel spectra for a silver nanodisk with a diameter of 60 nm and a height of 10 nm .
the impact parameters of the electron beams for the different spectra are reported in the inset , and the beam propagation direction is the @xmath78-direction perpendicular to the shaded disk .
we investigate different ` width ` parameters of 0.1 nm ( solid lines ) , 0.2 nm ( dashed lines ) , and 0.5 nm ( dashed - dotted lines ) , finding practically no differences in the results .
the ` cutoff ` parameter is set to 10 nm.,title="fig : " ] to summarize , in the following we list the most important properties of the ` eelsret ` class exc = eelsret ( p , impact , width , vel ) ; pot = exc ( enei ) ; [ psurf , pbulk ] = exc.loss ( sig ) ; we emphasize that the functionality of the ` eelsret ` class is very similar to that of the ` planewaveret ` and ` dipoleret ` classes , which account for plane wave and dipole excitations .
the only major difference is that ` eelsret ` requires the particle boundaries ` p ` of the ` comparticle ` object already in the initialization .
this is because upon initialization ` eelsret ` computes the crossing points between the particle boundaries and the electron trajectories ( if the electron passes by the nanoparticle no crossing points are found ) , and these crossing points are used in subsequent calculations of the potentials and the loss probabilities to speed up the simulation .
we next investigate the situation where the electron beam passes through the nanoparticle .
the working principle is almost identical to the previous case where the electron passes by the nanoparticle , but the ` width ` parameter and the different optional properties have to be set with more care . in the following ,
we briefly discuss these quantities in more detail . * ` width ` .
we have discussed in sec .
[ sec : refine ] that the integration of the external potentials over the boundary elements can be facilitated if we use a smoothening parameter @xmath54 in the calculation of the external potentials , see eq . .
it is important to stress that the integration could be also performed for @xmath79 and that the finite @xmath54 value only facilitates the computation . in general , ` width ` should be chosen smaller than the average size of the boundary elements , as also shown in fig .
[ fig : beam](a ) . * ` cutoff ` .
the ` cutoff ` parameter determines those boundary elements where the external potentials become integrated . in more detail
, we select all boundary elements fully or partially located within a circle with radius ` cutoff ` and centered around the impact parameter @xmath80 , as shown in fig .
[ fig : beam](a ) by the red circle . ` cutoff ` should be set such that at least all direct neighbours of the boundary element crossed by the electron beam are included . * ` refine ` and ` rule ` .
the integration over the boundary elements is controlled by ` refine ` and ` rule ` . `
refine ` gives the number of integration points within a triangle .
quadfaces are divided into two triangles . on default ` rule=18 ` is used ( see ` doc triangle_unit_set ` for details ) , and we recommend to use this value throughout . with ` refine ` one can split the triangles into subtriangles . usually the default value ` refine=1 ` should give sufficiently accurate results . . ]
figure [ fig : beam](b ) shows the eel spectra of a nanodisk ( for parameters see figure caption ) for three impact parameters , which are described in the inset . in fig .
[ fig : diskmap ] we show a density plot of identical loss spectra for a whole range of impact parameters .
one observes a number of peaks , attributed to the dipolar and quadrupolar modes at 2.6 ev and 3.1 ev , respectively , a breathing mode at 3.5 ev @xcite , and the bulk losses at 3.8 ev . .
for ` width=0 ` in panel ( a ) the computed maps show spikes at certain points , when the impact parameter @xmath27 is too close to a collocation point . for the moderate ` width ` parameters of panels ( b , c )
the results are sufficiently smooth and almost independent on the chosen value , whereas for too large parameters the eels map become smeared out , see panel ( d ) . ]
how should one chose the ` width ` , ` cutoff ` , and ` refine ` parameters ?
quite generally , the results depend rather unsensitively on the chosen parameters . in fig . [
fig : beam ] we show results for various ` width ` parameters listed in the figure caption , which are almost indistinguishable . in fig .
[ fig : diskbeam ] we depict eels maps for the dipolar disk mode at 2.6 ev and for various ` width ` parameters . for ` width=0 ` in panel ( a ) one observes for certain impact parameters spikes ( some of them indicated with arrows ) , where the loss probabilities becomes significantly enhanced or reduced in comparison to neighbour points .
this indicates that the impact parameter is located too closely to the collocation point of the boundary element , and the numerical integration fails . for ` width ` parameters of 0.1 or 0.2 nm , panels ( b , c ) , these spikes are absent and the results are almost indistinguishable .
finally , in panel ( d ) we report results for a too large smoothening parameter with a significant smearing of the features visible in panels ( a c ) . thus , ` width ` should be chosen significantly smaller than the size of the boundary elements but large enough to avoid spikes in the computed eels maps .
let us finally briefly discuss the simulation of eels maps for a nanotriangle , as computed with the demo program ` demotrianglemap.m ` .
results are shown in fig .
[ fig : trianglemap ] .
first we set up an array @xmath81 of impact parameters and initialize the ` eelsret ` object .
[ x , y ] = meshgrid ( linspace ( - 70 , 50 , 50 ) , linspace ( 0 , 50 , 35 ) ) ; impact = [ x ( : ) , y ( : ) ] ; vel = eelsbase.ene2vel ( 200e3 ) ; [ width , cutoff ] = deal ( 0.2 , 10 ) ; exc = eelsret ( p , impact , width , vel , cutoff , cutoff ) ; note that in the initialization of ` exc ` we pass a matrix ` [ x(:),y ( : ) ] ` of impact parameters .
as for the bem solver , we recommend to use for the boundary element integration the same or larger ` ' cutoff ' ` and ` ' refine ' ` values as for the ` eelsret ` object ( see ref .
@xcite and toolbox help pages for further details ) .
op = green.options ( cutoff , 20 , refine , 2 ) ; bem = bemret ( p , [ ] , op ) ; finally , once the loss probabilities are computed one should reshape ` psurf ` and ` pbulk ` to the size of the impact parameter mesh .
p = reshape ( psurf + pbulk , size ( x ) ) ; -direction for dipolar disk mode .
the simulation and disk parameters are identical to those of fig .
[ fig : beam ] , the fields are computed with ` demodiskfield.m ` . ] in some cases it is useful to plot the electromagnetic fields induced by the electron beam .
we briefly discuss how this can be done .
a demo program is provided by ` demodiskfield.m ` , and the simulation results are shown in fig .
[ fig : diskfield ] .
we first set up the ` eelsret ` object and the bem solver , following the prescription given above , and compute the surface charge and current distributions .
exc = eelsret ( p , [ b , 0 ] , width , vel , cutoff , cutoff ) ; bem = bemret ( p , [ ] , op ) ; sig = bem exc ( enei ) ; next , we define the points where the electric field should be computed , using the ` compoint ` class of the ` mnpbem ` toolbox , and define a green function object ` compgreen ` .
z = linspace ( - 80 , 80 , 1001 ) . ; pt = compoint ( p , [ b + 0 * z , 0 * z , z ] , mindist , 0.1 ) ; g = compgreen ( pt , p , op ) ; field = g.field ( sig ) ; e = pt ( field.e ) ; in the last two lines we compute the electromagnetic fields , and extract the electric field .
the command ` e = pt(field.e ) ` brings ` e ` to the same form as ` z ` , setting fields at points too close to the boundary ( which we have discarded in our ` compoint ` initialization with the parameter ` mindist ` @xcite ) to ` nan ` .
figure [ fig : diskfield ] shows simulation results .
for the electron beam passing through the nanodisk , the field amplitude @xmath82 increases strongly in vicinity of the nanoparticle , which we attribute to evanescent plasmonic fields , and @xmath82 is very small inside the nanodisk because of the efficient free - carrier screening inside conductors . with the ` mnpbem ` toolbox it is also possible to compute the light emitted from the nanoparticles , the so - called _
cathodoluminescence_. to this end , we first set up a ` spectrumret ` object for the calculation of scattering spectra , determine the electromagnetic fields at infinity , and finally compute the scattering spectra .
spec = spectrumret ; field = farfield ( spec , sig ) ; sca = scattering ( spec , field ) ; in the initialization of ` spec ` one could also use a sphere segment rather than the default unit sphere , e.g. , to account for finite angle coverages of photodetectors .
efficient parallelization can be achieved for typical energy loops of the form : for ien = 1 : length ( enei ) sig = bem exc ( enei ( ien ) ) ; [ psurf ( : , ien ) , pbulk ( : , ien ) ] = exc.loss ( sig ) ; end we can replace this loop with : matlabpool open ; parfor ien = 1 : length ( enei ) sig = bem exc ( enei ( ien ) ) ; [ psurf ( : , ien ) , pbulk ( : , ien ) ] = exc.loss ( sig ) ; end the important point is that all computations inside the loop can be performed independently , as is the case for the bem simulation as well as the calculation of the external potentials and loss probabilities . , but computed within the quasistatic limit using the demo program ` demodiskspectrumstat.m ` . the solid lines report simulation results for the full maxwell equations ( same as fig .
[ fig : beam ] ) , the dashed lines report results for the quasistatic approximation . for the dipolar mode at lowest energy ,
the peak position and width somewhat differ due to retardation effects and radiation damping . ]
the implementation of the quasistatic limit within the ` eelsstat ` class closely follows the retarded case .
the demo program ` demodiskspectrumstat.m ` is very similar to ` demodiskspectrum.m ` discussed in sec .
[ sec : penetrate ] .
we first set up a disk - like nanoparticle and specify the electron beam parameters .
next , we initialize an ` eelsstat ` object by calling exc = eelsstat ( p , b , width , vel , cutoff , cutoff , refine , 2 ) ; the definition of the various parameters is identical to the retarded case . finally , we set up the quasistatic ` bemstat ` or ` bemstateig ` bem solver @xcite , and compute the surface charge distribution and the energy loss probability using the equations presented in sec .
[ sec : quasistatic ] bem = bemstat ( p , [ ] , op ) ; for ien = 1 : length ( enei ) sig = bem exc ( enei ( ien ) ) ; [ psurf ( : , ien ) , pbulk ( : , ien ) ] = exc.loss ( sig ) ; end simulation results are shown in fig .
[ fig : diskspectrumstat ] .
we observe that the results of the full and quasistatic simulations are very similar , and only for the dipolar mode at lowest energy the peak position and width somewhat differ due to retardation effects and radiation damping
i am grateful to andi trgler for most helpful discussions , and thank him as well as toni hrl , jrgen waxenegger , and harald ditlbacher for numerous feedback on the simulation program .
this work has been supported by the austrian science fund fwf under project p24511n26 and the sfb nextlite . | within the ` mnpbem ` toolbox , we show how to simulate electron energy loss spectroscopy ( eels ) of plasmonic nanoparticles using a boundary element method approach .
the methodology underlying our approach closely follows the concepts developed by garca de abajo and coworkers [ for a review see rev .
mod .
phys .
82 , 209 ( 2010 ) ] .
we introduce two classes ` eelsret ` and ` eelsstat ` that allow in combination with our recently developed ` mnpbem ` toolbox for a simple , robust , and efficient computation of eel spectra and maps .
the classes are accompanied by a number of demo programs for eels simulation of metallic nanospheres , nanodisks , and nanotriangles , and for electron trajectories passing by or penetrating through the metallic nanoparticles .
we also discuss how to compute electric fields induced by the electron beam and cathodoluminescence .
plasmonics , metallic nanoparticles , boundary element method , electron energy loss spectroscopy ( eels ) |
pros@xmath0sb@xmath1 is the first pr - based heavy - fermion superconductor and attracts considerable interest because of unusual superconducting properties.@xcite one of the most interesting topics is what is the role of the multipole moments of pr @xmath7 electrons for the heavy - fermion nature and unconventional superconductivity in pros@xmath0sb@xmath1.@xcite long range magnetic dipole order does not coexist with superconductivity at zero field.@xcite instead , the existence of a magnetic field - induced ordered phase has been reported above the upper critical field @xmath8 for the superconductivity.@xcite in our previous study antiferromagnetic reflections with _ * q*_=(100 ) were observed in the field - induced ordered phase for _ * h*_@xmath2[001].@xcite the antiferromagnetic moment of 0.02@xmath3/pr at 8 t along the [ 010 ] direction perpendicular to the applied field was significantly smaller than the field - induced ferromagnetic moment parallel to _ * h*_. this small dipole moment and the @xmath9-@xmath10 phase diagram can be well explained by the mean - field calculation based on the crystal field level scheme of the @xmath5 singlet ground state with @xmath11 triplet excited state and assuming an antiferroquadrupolar ( afq ) interaction.@xcite the field - induced ordered phase appears as a result of the level crossing of the zero field ground and excited state which comes from the field split triplet state .
the level scheme and its field response have been experimentally clarified by means of neutron inelastic scattering under magnetic fields.@xcite the overall aspect of the field - induced ordered phase indicates that the antiferromagnetic dipole is not a primary order parameter
. generally speaking , there are many possibilities for the primary multipole order parameters .
the ordered phase is stable only under applied magnetic field , where the time reversal symmetry has been broken , thus , the antiferromagnetic dipole order appears simultaneously with the coupled order parameters such as quadrupole , octupole , and hexadecapole in the order parameter space with given @xmath12 symmetry.@xcite since neutrons exhibit no cross section for an electric quadrupole moment , the observation of the field - induced antiferromagnetic peak by neutron is not a direct proof for the primary afq order parameter . in this sense ,
our previous study did not rule out possibilities of other high - rank multipoles.@xcite actually , it is shown that the qualitative features for _ * h*_@xmath2[001 ] , the induced antiferromagnetic moment perpendicular to the applied field and the phase diagram can be reproduced in terms of octupole as well as quadrupole@xcite .
recent intensive studies on skutterudites report an important role of higher rank multipole , octupole@xcite and hexadecapole@xcite , in some compounds .
therefore , it is indispensable to clarify strictly the dominant interaction in pros@xmath0sb@xmath1 . for this purpose ,
further information is required : field direction and pressure dependencies and inelastic neutron scattering .
careful magnetization measurements on pros@xmath0sb@xmath1 clarified that the field - induced ordered phase also exists for [ 110 ] and [ 111 ] with remarkable anisotropy in the @xmath9-@xmath10 phase diagram.@xcite the existence of the field - induced ordered phase for these different field directions also supports the singlet - triplet crystal field level scheme .
it has been pointed out that the anisotropy in the phase diagram is intrinsic to @xmath12 symmetry and the predominance of quadrupolar interaction.@xcite the field - induced ordered states for both _ * h*_@xmath2[110 ] and [ 111 ] are also considered to be the afq ordered phase and not to be the dipolar one .
the order parameter for _ * h*_@xmath2[110 ] was theoretically predicted to be @xmath4@xcite , but has not been confirmed so far .
if this is the case , one could discriminate the primary order parameter of non - magnetic quadrupole from coupled magnetic octupole since the induced antiferromagnetic moment direction is expected to be different between them .
therefore the study for _ * h*_@xmath2[110 ] is worthwhile in order to clarify a dominant interaction in pros@xmath0sb@xmath1 . besides
, it motivates us that the magnetization measurements revealed the existence of some additional anomalies within the field - induced ordered phase for _ * h*_@xmath2[110 ] , indicating possible transitions between multipolar ordered states .
@xcite the purpose of the present neutron scattering study is to clarify the order parameter of the field - induced ordered phase for _ * h*_@xmath2[110 ] and to investigate the origin of anomalies in this ordered state .
we found that the induced antiferromagnetic moment is parallel to the field direction , which is the strong evidence for the @xmath4-type quadrupole as the primary order parameter . a photograph of the single crystalline sample of pros@xmath0sb@xmath1 used for the present neutron scattering study.,width=151 ]
( a ) the field dependence of the 221 superlattice reflection profile of pros@xmath0sb@xmath1 measured at 0.29 k for _ * h*_@xmath2[110 ] .
the solid lines denote gaussian fitting .
( b ) the field dependence of the 221 reflection intensity measured at various temperatures .
( c ) the 110 reflection intensity plotted as a function of applied magnetic field taken at 0.27k .
the inset gives its differential curve , d@xmath13/d@xmath9.,width=302 ] neutron scattering experiments have been carried out on the cold neutron triple - axis spectrometer ltas installed in the guide hall of the research reactor jrr-3 in japan atomic energy agency , jaea .
the instrumental setup was just the same as our previous experiment.@xcite the use of a cold neutron with triple - axis mode was crucial for taking high quality data to reduce the contamination of elastic spectra by the low - energy excitations .
we used a liquid he - free 10 t superconducting magnet and @xmath14he-@xmath15he dilution refrigerator both developed by jaea .
@xcite the vertical field was applied along [ 110 ] , perpendicular to the ( @xmath16@xmath17@xmath18 ) scattering plane .
we observed superlattice peaks at @xmath19 , while ferromagnetic scattering is superposed on nuclear bragg peak at @xmath20 , for example ( 110 ) . a large single crystalline sample with a mass of 6 g has been grown by the antimony - self - flux method .
the details of the sample preparation have been published elsewhere.@xcite figure [ f1 ] shows the picture of the sample , which is composed of small crystallites typically 1 mm@xmath14 distributed within 1 degree of the crystallographic axes .
the four - circle x - ray crystal structure analysis on the small piece of this sample confirms the filled skutterudite structure with the space group @xmath21 ( @xmath22 , # 204 ) .
the sample was characterized via magnetic susceptibility measurements ; we observed clear curie - weiss behavior at high temperature and the maximum at 3.6k , which is identical to the high - quality sample used in the de haas van alphen study.@xcite the resistivity was also very similar to the previous results and our sample showed superconducting transition at the onset of 1.88k .
these results guarantee the high quality of the present sample .
fortunately there was no domain structure concerning the two - fold symmetry in the basal plane , thus for example , ( @xmath23 ) reflection is generally different from ( @xmath24 ) .
we do not know the mechanism for the single domain growth process , nevertheless the single domain sample can be reproducibly obtained with our sample growth condition , which is rather mysterious .
( a ) the temperature dependence of the 221 superlattice reflection of pros@xmath0sb@xmath1 measured under the magnetic field of 8 t applied along the [ 110 ] direction .
the solid lines denote the result of fitting with gaussian and linear background for each temperature .
( b ) the temperature dependence of the 221 reflection intensity measured at various fields .
the solid lines are guides for the eyes.,width=302 ] the @xmath9-@xmath10 phase diagram of pros@xmath0sb@xmath1 for _ * h*_@xmath2[110 ] .
the closed and open circles are fields corresponding to the onset and the maximum of the field - induced antiferromagnetic peak , respectively .
the closed triangles are the phase boundary determined from the temperature dependence .
, width=302 ]
figure [ f2](a ) shows the representative data for the scattering profile of the 221 superlattice peak measured at 0.29 k with magnetic fields along the [ 110 ] direction .
an application of the magnetic field of 8 t induces a clear resolution - limited superlattice reflection at ( 221 ) where no trace of peak was observed at zero field . the observed field - induced superlattice peak positions are the same as those for _ * h*_@xmath2[001 ] , namely the propagation vector is _ * q*_=(100 ) . a marked decrease to @xmath251/10 of the integrated intensity occurs between 8 and 10 t whereas the magnetic bragg peaks maintain their positions and widths .
the field dependence of the 221 peak intensity was measured at various temperatures as shown in fig .
[ f2](b ) . applying the field at 0.27k
, the peak intensity appears at 5 t ( = @xmath26 ) , reaches a maximum around 8 t ( = @xmath27 ) , and decreases steeply above 8 t .
@xmath26 increases as the temperature rises : 6 t and 7 t at 0.51k and 0.71k , respectively .
in contrast , the maximum around 8 t is almost temperature independent .
no trace of superlattice peak was found above 1k .
figure [ f2](c ) displays a field variation of the peak intensity at 0.27k taken at _
* q*_=(1@xmath280 ) .
an increase of 1@xmath280 reflection intensity originating from the uniform magnetic moment was clearly observed .
the inset gives the differential curve , d@xmath13/d@xmath9 , which corresponds to the differential curve of the square of the ferromagnetic moment @xmath29 . with applied fields ,
the intensity exhibits a gradual increase and changes its slope around 5.5 t and 8 t , which can be clearly seen in the inset .
these inflection fields are consistent with the onset ( @xmath26 ) and maximum fields ( @xmath27 ) in the superlattice reflection intensity as well as the result of magnetization.@xcite .
the small hump around 1.5 t might arise from the breaking of the heavy - fermion superconductivity since it is very close to the upper critical field @xmath8 . the product of the magnitude of the magnetic moment and the angle factor derived from the observed magnetic bragg reflection intensity .
the open and closed circles indicate the data for @xmath308 t and 10 t , respectively .
the dotted lines are the fitting results for the reflection in the ( @xmath31 ) scattering plane by assuming the antiferromagnetic moment parallel to the applied field , namely , @xmath32.,width=302 ] the temperature dependence of the 221 superlattice peak profile under the applied field of 8 t is shown in fig .
[ f3](a ) .
the 221 superlattice reflection becomes weak without significant peak shift and/or broadening as temperature increases , and no trace of peak was observed at 1.4k .
figure [ f3](b ) shows peak intensity at ( 221 ) as a function of temperature taken under the field of 3 , 8 and 10 t . with increasing temperature from 0.27k at 8 t ,
the peak intensity showed the monotonous decrease and disappeared around @xmath33=0.8k .
the feature for 10 t is quite similar to that for 8 t except for the absolute intensity .
no peak was observed in the result for 3 t .
no additional anomaly at 8 t and 10 t is consistent with the fact that @xmath27 is almost temperature independent .
the present results for * _ h_*@xmath2[110 ] of pros@xmath0sb@xmath1 are summarized in the @xmath9-@xmath10 magnetic phase diagram as shown in fig .
the closed and open circles are the field corresponding to the onset and the maximum of the field - induced antiferromagnetic peak , respectively .
the closed triangles are the phase boundary determined from the temperature dependence of the antiferromagnetic intensity .
the overall aspect of the phase boundary , determined from the present diffraction experiments , is consistent with the result of magnetization measurements.@xcite the field - induced superlattice reflection with _ *
q*_=(100 ) exists only in the region of the field - induced ordered phase , in other words , the ordered state with _
* q*_=(100 ) is the basis of the field - induced ordered phase for * _ h_*@xmath2[110 ] which is the same as * _ h_*@xmath2[001 ] . the obtained magnetic structure for the field - induced ordered phase of pros@xmath0sb@xmath1 with application of the magnetic field along _ * h*_@xmath2[110].,width=302 ] in order to determine the magnetic structure in the field - induced ordered phase
, the integrated intensity of the superlattice peaks on the ( @xmath34 ) scattering plane were measured . the magnetic bragg reflection intensity for unpolarized neutron diffraction
can be written , @xmath35 ^ 2 \label{eq : nint}\ ] ] where @xmath36 is the scale factor , @xmath37 is the lorentz factor , @xmath38 is the antiferromagnetic moment , @xmath39 is the magnetic form factor and @xmath40 is the magnetic structure factor which is unity for the present case .
@xmath41 is the angle between the ordered magnetic moment and the scattering vector @xmath42 .
the square of the product of angle factor and antiferromagnetic moment @xmath38 can be given as follows ; @xmath43 ^ 2}}. \label{eq1}\]]figure [ f5 ] shows the square root of integrated intensity divided by scale factor , lorentz factor , structure factor and magnetic form factor of the pr@xmath44 free ion . therefore the vertical axis in fig .
[ f5 ] is the product of the antiferromagnetic moment @xmath38 and the angle factor @xmath45 . as clearly seen in fig .
[ f5 ] , this quantity is almost isotropic within the present scattering plane .
although the data obtained at 10 t is much weaker , it is also isotropic .
this isotropic nature of the left hand side of equation eq.([eq1 ] ) indicates that the antiferromagnetic moment is perpendicular to the ( @xmath34 ) scattering plane at both 8 t and 10 t within our experimental accuracy .
thus the induced dipole moment observed by neutron is parallel to the direction of the applied field along [ 110 ] . using this angle factor , @xmath45=1 , the induced antiferromagnetic moment for 8 t and 10 t
is deduced to be @xmath46 and @xmath47 , respectively .
the antiferromagnetic moment exhibits a strong reduction in the magnitude from 8 t to 10 t .
there is a large induced ferromagnetic moment parallel to the field , which is roughly ten times larger than that of the field - induced antiferromagnetic moment .
@xcite the sum of the ferromagnetic and antiferromagnetic moment gives the magnetic structure in the field - induced ordered phase as shown in fig .
the moment size is different for the each magnetic sublattice due to the antiferromagnetic component parallel to the field direction ; the magnetic moment at the corner of the unit cell is larger than that at the center , or vice versa .
we present a mean - field calculation for the present results with _
* h*_@xmath2[110 ] . in pros@xmath0sb@xmath1
, the @xmath5 singlet ground state with the @xmath6 first excited triplet state at 8k is well separated from the other excited state.@xcite this means that the low - lying singlet - triplet levels are responsible for low temperature physical properties .
hence , we move on our discussions to this low - lying singlet - triplet subspace .
active multipoles for the singlet - triplet system with _
* j*_=4 in the @xmath48 and @xmath12 symmetries are classified in ref . and .
according to the symmetry lowering from @xmath48 to @xmath12 , the triplet states in @xmath12 are represented by linear combinations of those in @xmath48 , whereas the @xmath5 singlet is not influenced .
the mixing is parameterized by @xmath49 , the coefficient in the crystal field hamiltonian reflecting the effect of @xmath12 , where @xmath50 corresponds to @xmath48.@xcite the possible multipole order parameters under magnetic fields which cause admixture are classified by symmetry analysis .
there are two possible symmetries for _ * h*_@xmath2[110 ] , @xmath5 and @xmath51 . for @xmath5 ,
the magnetic dipole @xmath52 is mixed with @xmath53 depending on the magnitude of @xmath49 .
the induced antiferromagnetic moment in the field - induced ordered phase was revealed to be @xmath52 in the present experiment .
this result indicates the order parameter for _ * h*_@xmath2[110 ] to be the @xmath5 symmetry which is consistent with the theoretical prediction.@xcite furthermore , the induced moment parallel to applied fields within the experimental accuracy suggests the small @xmath49 parameter which is also consistent with the previous results on neutron diffraction for _ * h*_@xmath2[001 ] and a small anisotropy in the @xmath9-@xmath10 phase diagram . the field dependence of the field - induced ferromagnetic and antiferromagnetic moment .
the open circles are obtained from the neutron diffraction , whose value corresponds to the left axes .
the results of mean - field calculation are represented by lines corresponding to the right axes .
, width=302 ] hereafter , we study to what extent the simple quadrupolar interaction model can reproduce the present observed results .
our analysis is based on the hamiltonian consisting of crystal field potential , zeeman energy and afq interaction , @xmath54 it is shown that the mechanism of field - induced afq order in this model can be clarified with a pseudo - spin representation analogous to the dimer - spin systems.@xcite with the use of two pseudo - spins , @xmath55 and @xmath56 with @xmath57 , which are defined in the singlet - triplet subspace , the local part of the hamiltonian is described as @xmath58where @xmath59 is the singlet - triplet energy splitting and @xmath60 is a scaled magnetic field .
the appearance of an effective staggered field ( @xmath61 ) to the two spins is characteristic of the @xmath12 system .
it has been shown that the interaction of @xmath4-type quadrupoles is mapped to the biquadratic pseudo - spin hamiltonian , @xmath62 , \end{aligned}\]]where@xmath63see ref . and for details on the dependence of @xmath64 and @xmath65 on the @xmath12 parameter @xmath49 .
an advantage of this mapping is to clarify that the predominant and isotropic contribution to the field - induced order is given by the part of a vector product @xmath66 , whereas the remaining part related with @xmath67 produces an anisotropic correction due to the @xmath12 symmetry . in the following ,
we discuss the mean - field solutions of this model in the field direction _ * h*_@xmath68 $ ] , in which the stable phase is shown to be of @xmath4 .
the calculation has been carried out by using the same parameter set as in ref . .
namely , we assume @xmath69 , @xmath70k , @xmath71 and @xmath72 , where @xmath73 is the number of the nearest neighbor bonds , and @xmath74 are the crystal field parameters
. the results of magnetic field dependence of the ferromagnetic ( @xmath75 ) and antiferromagnetic moment ( @xmath76 ) are shown in fig .
they are compared with the present experimental results that are deduced by normalizing @xmath75 and @xmath76 at 8 t to be 1.2@xmath77@xcite and 0.16@xmath77 , respectively .
one finds that the observed overall field response of both ferromagnetic and antiferromagnetic moments is well explained by this model analysis .
these agreement clearly indicates that the staggered moment should be induced by the primary afq order which remains inaccessible to the neutron probe .
it is also shown that the calculated value of antiferromagnetic moment is larger than that for @xmath78 $ ] , and this tendency is qualitatively consistent with the observed results .
note , however , that there exist quantitative differences between theory and experiment , probably due to fluctuation effects or contribution of other multipole interactions .
calculated angle @xmath79 between the field - induced antiferromagnetic moment and the field direction [ 110 ] as a function of relative magnetic interaction strength @xmath80 for @xmath81=1.2 , corresponding to @xmath827t.,width=302 ] in order to clarify the unique properties of the afq model more definitely , let us study the effects of possible magnetic interaction .
it is shown that the magnetic interaction of dipoles and octupoles is expressed as an isotropic bilinear spin hamiltonian.@xcite although the generic form of the interaction is somewhat complex with a few parameters , we introduce here the simplest form to capture the essential physics , @xmath83 this symmetric model in the two spins corresponds to a mixed interaction of dipoles and octupoles .
note , however , that the relevant part is the octupolar interaction , because they have much larger matrix elements between singlet and triplet states .
it should be stressed that the total hamiltonian in the limit , @xmath84 and @xmath85 , reduces to the conventional dimer - spin model@xcite , in which a similar field - induced order takes place by the magnetic interaction .
thus , it is quite instructive to see the difference between quadrupolar and magnetic interactions by interpolating the two limits . in the framework of the mean - field theory ,
we have calculated the antiferromagnetic moment direction for the total hamiltonian @xmath86 in the field direction _ * h*_@xmath68 $ ] .
taking the ratio of the interactions as @xmath87 and @xmath88 , the angle from the applied field direction , @xmath79 , is plotted in fig .
[ f8 ] as a function of relative magnetic strength @xmath89 with a fixed @xmath90 . in the case that the magnetic interaction is small , the moment direction is almost parallel to the field . whereas the magnetic interaction becomes dominant , the moment orients perpendicular to the field , as expected in the dimer - spin systems .
such a remarkable change of the antiferromagnetic structure that depends on the interaction is characteristic in the field direction _ * h*_@xmath68 $ ] . in the case of
_ * h*_@xmath91 $ ] , both interactions result in the antiferromagnetic moment perpendicular to the field.@xcite thus , the present experiment for _ * h*_@xmath68 $ ] is crucial in distinguishing the type of multipolar interactions . in this sense ,
the observed antiferromagnetic moment parallel to the field is regarded as strong evidence of the predominant afq interaction .
the field - induced antiferromagnetic moment parallel to the field for _ * h*_@xmath2[110 ] is different from the perpendicular antiferromagnetic moment for _ * h*_@xmath2[001 ] , whereas the propagation vector @xmath92=(100 ) is the same . in the latter case
the antiferromagnetic moment is parallel to the [ 010 ] direction , and not to the [ 100 ] direction ; a clear peak was observed for _ * q*_=(100 ) but no trace of the antiferromagntic peak was observed for _ * q*_=(010 ) .
this characteristic moment direction reflecting the lack of four - fold symmetry gives the important information on the possible order parameter .
variation in the orientation of the field - induced antiferromagnetic moment according to the applied field direction is characteristic to the quadrupolar ordered phase .
both results indicate the predominance of the @xmath4-type afq interaction in pros@xmath0sb@xmath1 .
a recent inelastic neutron scattering study clarified that the crystal field excitations from the @xmath5 ground state to the @xmath11 triplet first excited state have a dispersion ; the observed excitation softens at _ * q * _ = ( 100 ) which is the same as the ordering vector of the field - induced afq ordered state.@xcite in addition , the intensity at _ * q*_=(100 ) is weaker than that at zone center .
though the magnetic and non - magnetic interaction gives the same energy dispersion , the _ * q * _ dependences of the intensity are opposite;@xcite namely , magnetic interaction should lead to the stronger intensity at the zone boundary .
thus , the observed excitation spectra indicate the dominance of antiferroquadrupolar interaction which is quite consistent with the present result for _ * h*_@xmath2[110 ] . together with the anisotropy in the @xmath9-@xmath10 phase diagram , the predominance of afq interaction is clarified in pros@xmath0sb@xmath1 .
furthermore , the excitation observed in the inelastic scattering study gives an interaction strength .
note that the quadrupolar coupling constant obtained from the inelastic scattering is almost consistent with that used in the present study .
there is a slight difference in the definition ; @xmath93 in the present paper can be connected with @xmath94 in ref . with @xmath95 where @xmath96 is the off - diagonal matrix elements of quadrupolar moments between singlet and triplet .
taking this difference into account , the coupling constant @xmath97 - 2.7k derived from inelastic spectra is consistent with @xmath982.4k used in the present study .
in addition , the quadrupolar coupling constant @xmath99 obtained from the ultrasound measurements of @xmath100 is also consistent with the present value.@xcite in other words , the coupling constant used in the present study corresponds with those obtained in the other measurements .
the maximum in the induced antiferromagnetic moment , @xmath27 , corresponds well to the @xmath101 anomaly in the magnetization@xcite , as shown in fig .
however , no change in the antiferromagnetic structure was found on passing through @xmath27 . as shown in fig .
[ f7 ] , the induced @xmath52 appears to vanish slightly above 10 t .
this critical field is well below the reported critical field for the field - induced ordered phase of @xmath2512 t but rather close to the @xmath102 anomaly.@xcite in other words , there is another field - induced ordered state above @xmath102 .
a possible explanation would be the switching of the order parameter from @xmath5 to @xmath51 representation at @xmath102 .
it is theoretically predicted that the energy difference of these two ordered states is very small.@xcite the coupled magnetic dipole order parameter of @xmath51 representation is @xmath103 for _ * h*_@xmath2[110 ] .
thus a change of moment direction from [ 110]@xmath104 to [ 001]@xmath105 should be accompanied by the quadrupole - quadrupole transition . this reorientation of the antiferomagnetic moment has not been observed in our present neutron scattering experiment up to 10 t . in order to clarify the origin of the anomalies observed in magnetization
, higher field neutron scattering experiments should be carried out . in conclusion , we observed the field - induced antiferromagnetic order of heavy - fermion superconductor pros@xmath0sb@xmath1 for _ * h*_@xmath2[110 ] .
the antiferromagnetic moment was parallel to the field direction , which is direct evidence for the primary quadrupolar order parameter of the field - induced ordered phase .
together with our previous study for _ * h*_@xmath2[001 ] , we can conclude that the @xmath4-type quadrupolar interaction is intrinsic to the many body interaction between @xmath7 electron based quasiparticles in heavy - fermion superconductor pros@xmath0sb@xmath1 .
we are very grateful to j. -m .
mignot and a. gukasov for communication of unpublished data obtained at llb saclay .
we also wish to thank h. sugawara for helpful advice on sample preparation .
this work was supported by grants - in - aid for scientific research , young scientist ( b ) ( no .
16740212 ) and in priority area `` skutterudite '' ( no .
18027012 , 18027013 and 18027015 ) of the ministry of education , culture , sports , science and technology , japan . in the case of the @xmath48 symmetry ( @xmath50 ) ,
the dipoles and the octupoles in the singlet - triplet system are simply expressed with the pseudo spins as @xmath106 and @xmath107.@xcite when the @xmath12 term is introduced ( @xmath108 ) , they are weakly mixed with each other . concerning the mean - field solution of the interaction ( 7 ) , it is known that the antiferromangetic order parameter is transverse to the field direction@xcite . if the @xmath49 component of the spins , or equivalently that of the dipoles and the octupoles , is chosen as the order parameter in @xmath109 $ ] , one can expect @xmath110 due to the positive @xmath59 term in eq .
this leads to a tiny antiferrmagnetism because @xmath111 for a small @xmath12 term , which is compatible to the experimental results for @xmath109 $ ] .
since the model is isotropic for the field directions , this transverse antiferromagnetism does not account for the present experimental results in @xmath112 $ ] . | neutron scattering experiments have revealed a field - induced antiferroquadrupolar order parameter in the pr - based heavy - fermion superconductor pros@xmath0sb@xmath1 . we observed the field - induced antiferromagnetic dipole moment with the propagation vector _ * q * _ = ( 100 ) for the applied field direction both _ * h*_@xmath2[110 ] and [ 001 ] . for _ * h*_@xmath2[110 ] at 8 t , it should be noted that the induced antiferromagnetic moment of 0.16(10 ) @xmath3/pr orients parallel to the field within our experimental accuracy .
this observation is strong evidence that the @xmath4 electric quadrupole to be the primary order parameter for _ * h*_@xmath2[110 ] .
a mean - field calculation , based on @xmath5 singlet and @xmath6 excited triplet with the @xmath4 quadrupolar interaction , reproduces the induced moment direction and its field response .
these facts indicate the predominance of @xmath4-type antiferroquadrupolar interaction in pros@xmath0sb@xmath1 . |
the run ii at the tevatron will define a new level of precision for qcd studies in hadron collisions .
both collider experiments , cdf and d0 , expect to collect up to @xmath1 of data in this new run period .
the increase in instantaneous luminosity , center - of - mass energy ( from 1.8 tev to 2 tev ) and the improved acceptance of the detectors will allow stringent tests of the standard model ( sm ) predictions in extended regions of jet transverse momentum , @xmath2 , and jet rapidity , @xmath3 .
the hadronic final states in hadron - hadron collisions are characterized by the presence of soft contributions ( the so - called _ underlying event _ ) from initial - state gluon radiation and mutiple parton interactions between remnants , in addition to the jets of hadrons originated by the hard interaction .
a proper comparison with pqcd predictions at the parton level requires an adequate modeling of these soft contributions which become important at low @xmath4 . in this letter ,
a review of some of the most important qcd results from run ii is presented .
the measurement of the inclusive jet production cross section for central jets constitutes one of the pillars of the jet physics program at the tevatron .
it provides a stringent test of perturbative qcd predictions over almost nine orders of magnitude and probes distances up to @xmath5 .
thanks to the increase in the center - of - mass energy in run ii the jet production rate has been multiplied ( by a factor of five for jets with @xmath6 gev ) and the first measurements have already extended the @xmath7 coverage by 150 gev compared to run i. in addition , both cdf and d0 experiments explore new jet algorithms following the theoretical work that indicates that the cone - based jet algorithm employed in run i is not infrared safe and compromises a future meanful comparison with pqcd calculations at nnlo .
figure [ fig1 ] shows the measured inclusive jet cross section by cdf using the longitudinally invariant @xmath8 algorithm @xcite and based on the first 145 @xmath9 of run ii data .
measurements have been performed using values for the d parameter in the @xmath8 expression , @xmath10 equal to 0.5 , 0.7 and 1.0 .
the measurements are compared to pqcd nlo calculations @xcite using cteq6 @xcite parton density functions in the proton and antiproton and the renormalization and factorization scales set to @xmath11 .
the measured cross section is reasonably well described by the predictions for @xmath12 within the present uncertainties .
the systematic errors on the data are dominated by the uncertainty on the jet energy scale determination while the theoretical predictions suffer from our limited knowledge of the gluon distribution at high @xmath13 . at lower @xmath7
, the data is systematically above the predictions and the effect increases as d increases ( see figure [ fig2 ] ) .
this indicates the presence of soft - gluon contributions and fragmentation effects that have not been taken into account yet .
figure [ fig3 ] shows the measured inclusive jet cross section by d0 based on the first 143@xmath9 of run ii data .
the new midpoint@xcite jet algorithm has been used with a cone size r=0.7 .
this algorithm constitutes an improved version of the cone - based algorithm used in run i and it is shown to be infrared safe when used in fixed - order parton - level calculations .
the data is in good agreement with the pqcd nlo predictions using cteq6 parton density functions and @xmath14 .
however , the measurement is dominated by a relatively large uncertainty on the absolute jet energy scale .
figure [ fig4 ] shows the measured cross section by d0 as a function of the dijet invariant mass in dijet production of central jets .
this measurement is particularly sensitive to the presence of narrow resonances decaying into jets of hadrons up to masses of 1.3 tev .
the data is well described by pqcd nlo predictions .
nowadays , the tevatron high-@xmath2 jet data is used , together with prompt - photon data from fixed target experiments , to constrain the gluon distribution at high-@xmath13 .
jet measurements at large rapidities are important because they constrain the gluon density in a region in @xmath7 where no effect from new physics is expected .
the d0 experiment has already extended the jet cross section measurements to the forward region for jets with @xmath15 ( see figure [ fig5 ] ) . at the moment , the results are affected by large systematic errors . in the near future
the experiments will highly reduce their uncertainties and precise cross section measurements will allow to further constrain the gluon distribution and thus enhance their sensitivity to new physics at very high @xmath7 .
as mentioned in previous section , the hadronic final states at the tevatron are characterized by the presence of soft underlying emissions , usually denoted as _ underlying event _ , in addition to highly energetic jets coming from the hard interaction .
the underlying event contains contributions from initial- and final - state soft gluon radiation , secondary semi - hard partonic interactions and interactions between the proton and anti - proton remnants that can not be described by perturbation theory .
these processes must be approximately modeled using monte carlo programs tuned to describe the data .
the jet energies measured in the detector contain an underlying event contribution that has to be subtracted in order to compare the measurements to pqcd predictions .
hence , a proper understanding of this underlying event contribution is crucial to reach the desired precision in the measured jet cross sections . in the analysis presented here ,
the underlying event in dijet production has been studied by looking at regions well separated from the leading jets , where the underlying event contribution is expected to dominate the observed hadronic activity .
jets have been reconstructed using tracks with @xmath16 gev and @xmath17 and a cone algorithm with r=0.7 .
the @xmath18 space around the leading jet is divided in three regions : _ towards _ , _ away _ and _ transverse _ ( see figure [ fig6]-left ) , and the transverse region is assumed to reflect the underlying event contribution
. figure [ fig6]-right shows the average track density in the transverse region as a function of @xmath19 of the leading jet for the dijet inclusive sample and for events where the leading jets are forced to be back - to - back in @xmath18 , in order to further reduce extra hard - gluon radiation .
the observed plateau indicates that the underlying event activity is , to a large extend , independent from the hard interaction .
the measurements have been compared to the predictions from pythia @xcite and herwig @xcite monte carlo programs including leading - order qcd matrix elements plus initial and final parton showers .
the pythia samples have been created using a special tuned set of parameters , denoted as pythia - tune a , which includes an enhanced contribution from initial - state soft gluon radiation and a tuned set of parameter to control secondary parton interactions .
it was determined as a result of similar studies of the underlying event performed using cdf run i data @xcite .
pythia - tune a describes the hadronic activity in transverse region while herwig underestimates the radiation at low @xmath19 .
similar measurements in z+jet(s ) events would allow to explore the univesality of the underlying event contribution in events with a very different colour configuration in the final state .
the internal structure of jets is dominated by multi - gluon emissions from the primary final - state parton .
it is sensitive to the relative quark- and gluon - jet fraction and receives contributions from soft - gluon initial - state radiation and beam remnant - remnant interactions .
the study of jet shapes at the tevatron provides a stringent test of qcd predictions and tests the validity of the models for parton cascades and soft - gluon emissions in hadron - hadron collisions .
the cdf experiment has presented results on jet shapes for central jets with transverse momentum in the region
@xmath20 gev @xmath21 gev , where jets are searched for using the midpoint merging fraction has been used instead of the default @xmath22 . ]
algorithm and a cone size @xmath23 .
the integrated jet shape , @xmath24 , is defined as the average fraction of the jet transverse momentum that lies inside a cone of radius @xmath25 concentric to the jet cone : @xmath26 where @xmath27 denotes the number of jets .
the measured jet shapes have been compared to the predictions from pythia - tune a and herwig monte carlo programs .
in addition , two different pythia samples have been used with default parameters and with and without the contribution from multiple parton interactions ( mpi ) between proton and antiproton remnants , the latter denoted as pythia-(no mpi ) , to illustrate the importance of a proper modeling of soft - gluon radiation in describing the measured jet shapes .
figure [ fig7](left ) presents the measured integrated jet shapes , @xmath28 , for jets with @xmath29 gev , compared to herwig , pythia - tune a , pythia and pythia-(no mpi ) predictions .
in addition , figure [ fig7](right ) shows , for a fixed radius @xmath30 , the average fraction of the jet transverse momentum outside @xmath31 , @xmath32 , as a function of @xmath33 where the points are located at the weighted mean in each @xmath33 range .
the measurements show that the fraction of jet transverse momentum at a given fixed @xmath34 increases ( @xmath32 decreases ) with @xmath33 , indicating that the jets become narrower as @xmath33 increases .
pythia with default parameters produces jets systematically narrower than the data in the whole region in @xmath33 .
the contribution from secondary parton interactions between remnants to the predicted jet shapes ( shown by the difference between pythia and pythia-(no mpi ) predictions ) is relatively small and decreases as @xmath33 increases .
pythia - tune a predictions describe all of the data well .
herwig predictions describe the measured jet shapes well for @xmath35 gev but produces jets that are too narrow at lower @xmath33 .
figure [ fig8](left ) shows the measured integrated jet shapes , @xmath28 , for jets with @xmath29 gev , compared to pythia - tune a and the predictions for quark- and gluon - jets plane ) its directions with that of one of the outgoing partons from the hard interaction . ]
figure [ fig8](right ) shows the measured @xmath32 , @xmath30 , as a function of @xmath33 .
the monte carlo predictions indicate that the measured jet shapes are dominated by contributions from gluon - initiated jets at low @xmath33 while contributions from quark - initiated jets become important at high @xmath33 .
this can be explained in terms of the different partonic contents in the proton and antiproton in the low- and high-@xmath33 regions , since the mixture of gluon- and quark - jet in the final state partially reflects the nature of the incoming partons that participate in the hard interaction . for a given type of parton - jet in the monte carlo ( quark- or gluon - jet ) ,
the observed trend with @xmath33 shows the running of the strong coupling constant , @xmath36 .
jet shape measurements thus introduce strong constrains on phenomenological models describing soft - gluon radiation and the underlying event in hadron - hadron interactions .
similar studies with b - tagged jets will be necessary to test our knowledge of b - quark jet fragmentation processes in hadronic interactions , which is essential for future precise top and higgs measurements .
the d0 experiment has employed the dijet sample to study azimuthal decorrelations , @xmath37 , between the two leading jets . the normalized cross section :
@xmath38 is sensitive to the spectrum of the gluon radiation in the event .
the measurements has been performed in different regions of the leading jet @xmath7 starting at @xmath39 gev and the second jet is requied to have at least @xmath40 gev .
figure [ fig9 ] shows the measured cross section compared to lo and nlo predictions from nlojet++ program @xcite .
the lo predictions , with at most three partons in the final state , is limited to @xmath41 , for which the three partons define a _ mercedes - star _ topology .
it presents a prominent peak at @xmath42 corresponding to the soft limit for which the third parton is collinear to the direction of the two leading partons .
the nlo predictions , with four partons in the final state , describes the measured @xmath37 distribution except at very high and very low values of @xmath37 where additional soft constributions , corresponding to a resummed calculation , are necessary . a reasonable approximation to such calculations is provided by parton shower monte carlo programs .
figure [ fig10 ] presents the measured cross section compared to pythia - tune a , pythia and herwig predictions in different regions of @xmath7 .
pythia with default parameters underestimates the gluon radiation at large angles .
pythia - tune a predictions , which include an enhanced contribution from initial - state soft gluon radiation and secondary parton interactions , describe the azimuthal distribution .
herwig also describes the data although tends to produce less radiation than pythia - tune a close to the direction of the leading jets .
both cdf and @xmath0 experiments have carried out measurements of the inclusive jet production cross section in run ii at the tevatron using different jet algorithms .
the measurements are in agreement with pqcd nlo predictions
. dedicated studies of the underlying event , jet shapes and azimuthal decorrelations in dijet final states allowed to establish the validity of the monte carlo models used to describe the soft - gluon contributions in the final state .
i thank the organizers for their kind invitation to the conference and for the exciting program of talks and discussions they made possible .
stephen d. ellis , davision e. soper , phys.rev .
* d 48 * 3160 - 3166 ( 1993 ) .
giele , e.w.n .
glover , d.a .
kosower , nucl .
phys.*b 403 * 633 ( 1993 ) .
j. pumplin et al . , jhep * 0207 * 012 ( 2002 ) .
g. c. blazey , et al . , hep - ex/0005012 .
ellis , j. huston and m. toennesmann , hep - ph/0111434 .
bengtsson and t. sjstrand , comp .
comm . * 46 * 43 ( 1987 ) .
g. marchesini et al . ,
. comm . * 67 * 465 ( 1992 ) .
d. acosta et al . ,
cdf collaboration , phys . rev .
* d65 * , 092002 ( 2002 ) .
zoltn nagy , zoltn trcsnyi , phys.rev.lett .
* 79 * 3604 ( 1997 ) . | in this contribution , a number of new qcd results on jet production from the cdf and @xmath0 experiments in run ii are discussed in detail .
= 11.6pt |
early - type galaxies are found in virtually all environments from the field , to small groups , to rich clusters ( hubble & humason 1931 ; oemler 1974 ; dressler 1980 ) . in the highest density environments , ellipticals and lenticulars
are known to dominate the overall fraction of bright galaxies , @xmath16 , with the precise contribution depending on local galaxy density and redshift ( smith 2004 ; postman 2005 ) . in the virgo cluster ,
the rich cluster nearest to our own galaxy , @xmath17 for galaxies brighter than @xmath18 or @xmath19 ( julian et al .
1997 ) . mag ( tonry 2001 ; mei 2005b ) . ]
if one considers not just giant galaxies , but also the much more common dwarfs , then the dominance of early - type galaxies is even more pronounced : i.e. , among the confirmed members of virgo with unambiguous morphological classifications , the early - type fraction is @xmath1 0.8 ( reaves 1983 ; binggeli , tammann & sandage 1987 , hereafter bts87 ) .
it has long been recognized that early - type galaxies , both in virgo and elsewhere , often show compact nuclei near their centers . in their landmark study of the virgo cluster ,
bts87 carried out a visual search for nuclei using wide - field , blue - sensitive photographic plates from the 2.5-m du pont telescope .
of the 1277 members and 574 probable members in their virgo cluster catalog ( hereafter vcc ) , a total of 1192 were classified as non - nucleated dwarfs ( des or ds0s ) while an additional 415 dwarfs ( predominantly de , ns ) were found to be nucleated .
thus , roughly 25% of the dwarf galaxies in virgo were found by bts87 to have a discernible nucleus , although the precise fraction was also found to depend on galaxy luminosity and position within the cluster ( see figure 8 of sandage , binggeli & tammann 1985 and figure 19 of bts87 , respectively ) .
unfortunately , progress toward understanding the nature of these nuclei has been limited because of several factors : @xmath20 , ground - based studies must contend with the contamination from the underlying galaxy light , it is difficult to de - couple the brightness profiles of the nuclei from those of their host galaxies , and the nuclei are sufficiently compact that they appear unresolved in even the sharpest ground - based images ( caldwell 1983 , 1987 ; durrell 1997 ) . while the photographic survey of bts87 remains a landmark study of nucleated galaxies in the local universe , there are reasons to believe that a modern survey of the nuclei belonging to early - type galaxies would be advisable preferably one which capitalizes on the high angular resolution afforded by the _ hubble space telescope _ ( _ hst _ ) . first , _ hst _
imaging of late - type galaxies has revealed that 5070% of these systems have compact stellar clusters at or near their photocenters ( phillips et al .
1996 ; carollo , stiavelli & mack 1998 ; matthews et al 1999 ; bker et al . 2002
; bker et al .
second , if the early - type members of the local group are any guide , then one may expect estimates for the fraction of nucleated galaxies to increase as better imaging becomes available .
for instance , in recent years a number of local group dwarfs have been found to contain previously unrecognized central substructures and/or nuclei , including sagittarius ( layden & sarajedini 2000 ; monaco et al . 2005 ) , ursa minor ( kleyna et al . 2003 ; palma et al .
2003 ) , andromeda ii ( mcconnachie & irwin 2006 ) and fornax ( coleman et al . 2004 ) .
third , in their wfpc2 survey of dwarfs in the virgo and fornax clusters , lotz et al .
( 2004 ) found that six of the 30 non - nucleated " dwarf ellipticals in their sample actually contained nuclei which had gone unnoticed in the ground - based surveys ( binggeli , tammann & sandage 1985 , hereafter bts85 ; bts87 ; ferguson 1989 ) .
very recently , grant , kuipers & phillipps ( 2005 ) used imaging from the wide field camera on the isaac newton telescope to show that faint nuclei in virgo dwarfs were frequently missed in photographic surveys .
these results suggest that there may be significant incompleteness in our census of nuclei in early - type galaxies .
indeed , in their photographic study of virgo , bts87 cautioned that most nuclei in the luminous e and s0 galaxies were probably missed due to [ the ] high surface brightness [ of the underlying galaxy . ] " in addition to this surface - brightness selection effect , bts87 state explicitly that any nuclei with @xmath21 23 ( @xmath22 ) would fall below their plate detection limits and hence be missing from their catalog .
the central regions of early - type galaxies have been favorite targets for _ hst _ since its launch in 1990 . for the most part
, such surveys have tended to focus on the core structure of the galaxies .
however , several studies reported the discovery of compact nuclei in ( predominantly bright ) samples of early - type galaxies , beginning with pre - refurbishment ( wfpc ) imaging ( crane 1993 ; lauer 1995 ) and continuing with imaging from wfpc2 ( rest 2001 ; lauer 2005 ) and nicmos ( ravindranath 2001 ) .
these studies , which are primarily based on single - filter imaging of samples of ( 3377 ) galaxies with distances between 1 and 100 mpc , have confirmed that some early - type galaxies do contain compact nuclei , but there is disagreement over their overall frequency ( with estimates ranging from 13% to @xmath1 50% ) , whether or not they are resolved structures , and their classification as stellar or non - thermal ( agn ) sources .
a better understanding of the physical properties of these nuclei is important since they almost certainly hold clues to the violent processes that have shaped the central regions of galaxies , which could include star formation triggered by infalling gas , collisions and mergers of stars and star clusters , tidal disruption of clusters and the growth of stellar cusps " by central black holes , and the mechanical and radiative feedback from accreting black holes or intense nuclear starbusts .
this paper presents a homogenous analysis of the nuclei belonging to a sample of 100 early - type galaxies in the virgo cluster .
our images , taken with the advanced camera for surveys ( acs ; ford et al .
1998 ) , form the basis of the acs virgo cluster survey ( acsvcs ; ct 2004 ; hereafter paper i ) . other papers in this series have discussed the data reduction pipeline ( jordn et al .
2004a = paper ii ) , the connection between low - mass x - ray binaries in m87 ( jordn et al .
2004b = paper iii ) , the measurement and calibration of surface brightness fluctuation magnitudes ( mei et al .
2005ab = papers iv and v ) , the morphology , isophotal parameters and surface brightness profiles for early - type galaxies ( ferrarese et al .
2006a = paper vi ) , the connection between globular clusters and ultra - compact dwarf galaxies ( haegan et al .
2005 = paper vii ) , the color distributions of globular clusters ( peng et al .
2006a = paper ix ) , the half light radii of globular clusters and their use as a distance indicator ( jordn et al .
2006 = paper x ) and the discovery of diffuse star clusters in early - type galaxies ( peng et al .
2006b = paper xi ) .
there are several features of the acs virgo cluster survey which make it uniquely suited to the study of nuclei in early - type galaxies .
first , the survey itself targets a large sample of 100 early - type galaxies lying at a common distance of about 16.5 mpc so that the @xmath23 fwhm of the acs point - spread function ( psf ) corresponds to a small , and nearly constant , physical scale of @xmath24 pc .
this excellent spatial resolution , coupled with the fine plate scale of 0049 pix@xmath25 , makes it possible to measure structural parameters for any nuclei larger than a few parsecs in size .
second , with blue magnitudes in the range @xmath26 ( @xmath27 ) , our program galaxies span a wide range in luminosity so it is possible to study the phenomenon of nucleation in giant and dwarf galaxies simultaneously .
third , the images are sufficiently deep that they reveal not only the nuclei , but also the many globular clusters belonging to our program galaxies ; thus , the same images which provide information on the nucleus and host galaxy can also be used to study the associated globular cluster systems and to explore possible evolutionary links between the clusters and nuclei . and
finally , because multi - band imaging is available in two widely - separated bandpasses ( f475w and f850lp ) for each object in the survey , it is possible to use broadband colors to place rough constraints on the star formation and chemical enrichment histories of the nuclei and their host galaxies .
the organization of this article is as follows .
2 gives a brief summary of the observational material used in our analysis .
a description of the galaxy brightness profiles and the method of their analysis is presented in 3 .
4 contains a discussion of the empirical properties of the nuclei in our survey , such as their overall numbers , possible displacements from the galaxy photocenters , luminosities , colors , surface brightnesses and half - light radii . in 5 we discuss the implications of our findings for various formation scenarios .
the article concludes with a summary of the main results in 6 . a future paper in this series will discuss the broader implications of our findings for models of nucleus formation in early - type galaxies ( merritt 2006 )
our analysis is based on _ hst _ imaging for 100 early - type galaxies having morphological types e , s0 , de , de , n and ds0 . all are confirmed members of the virgo cluster based on radial velocity measurements .
images were taken with the acs instrument used in wide field channel ( wfc ) mode with the f475w and f850lp filter combination , which are roughly equivalent to the @xmath28 and @xmath2 bands , respectively , in the sloan digital sky survey photometric system .
these images form the basis of the acs virgo cluster survey , a complete description of which may be found in paper i. note that the 26 brightest galaxies in this survey constitute a complete sample of early - type members of virgo with @xmath29 , and that the full sample represents 44% of all early - type members of virgo spanning the magnitude range @xmath30 .
a customized data pipeline ( described in detail in paper ii ) produces geometrically - corrected , flux - calibrated , cosmic - ray - free images in the f475w and f850lp bandpasses .
table [ tab : data ] gives some basic information about the target galaxies , tabulated in order of increasing blue magnitude ( decreasing luminosity ) .
an identification number for each galaxy is given in the first column , followed by the identification from the vcc ( bst85 ) and other names for the galaxy in the messier , ngc , ugc or ic catlogs .
blue magnitudes , @xmath31 , from bst85 are presented in column 4 , while the fifth column records the adopted galactic reddening from schlegel , finkbeiner & davis ( 1998 ) .
columns 6 and 7 record the surface brightness of each galaxy , in both the @xmath28 and @xmath2 bandpasses , measured via spline interpolation at a geometric mean radius where @xmath32 is the semi - major axis and @xmath33 is the ellipticity . ] of @xmath34 1 ( @xmath1 80 pc ) .
this model - independent choice of surface brightness should closely approximate the galaxy s _ central _ surface brightness , but is measured at a radius large enough to ensure negligible contamination from any central nucleus .
the remaining columns of table [ tab : data ] will be described below .
coordinates , morphological classifications and other information on the program galaxies may be found in papers i and vi .
our goals in this paper include the measurement of the structural and photometric properties of the nuclei in our program galaxies , and an investigation into the relationship between these nuclei and their host galaxies .
additionally , we wish to compare the properties of the nuclei to those of the globular clusters in the program galaxies and , more generally , to the virgo ultra - compact dwarf ( ucd ) galaxies ( e.g. , drinkwater et al .
1999 ; hilker et al .
1999 ; drinkwater et al .
2000 ; phillipps et al . 2001 ) identified in the course of this survey and described in haegan ( 2005 ; hereafter paper vii ) and haegan ( 2006 ) . a companion paper in this series ( paper vi ) presents an analysis of the surface brightness profiles of the program galaxies along with a tabulation of the best - fit structural parameters , while two other articles examine the photometric ( paper ix ) and structural ( paper x ) parameters of the globular clusters .
as we make use of several results from these studies , the reader is referred to these papers for complete details . because the nuclei are always superimposed on the light of the underlying galaxy , measuring their photometric and structural properties
requires a model for the galaxy surface brightness profile . for each galaxy in our survey ,
@xmath28- and @xmath2-band azimuthally - averaged radial surface brightness profiles are available from paper vi .
these profiles were derived by fitting the isophotes with the ellipse task in iraf which , in turn , is based on the algorithm of jedrzejewski ( 1987 ) .
the @xmath28- and @xmath2-band brightness profiles were parameterized with a standard srsic ( 1968 ) model , @xmath35,$$ \end{array } \label{eq1}\ ] ] where @xmath36 is the central intensity and @xmath37 is a shape parameter which yields an @xmath38-law profile for @xmath39 ( de vaucouleurs 1948 ) and an exponential profile for @xmath40 .
the parameter @xmath41 is defined such that @xmath42 , where @xmath43 and @xmath44 are the complete and incomplete gamma functions , respectively ( e.g. , graham & driver 2005 ) . as shown by caon , capaccioli & donofrio ( 1993 ) , a convenient approximation relating @xmath41 to the shape parameter @xmath37 is @xmath45 for @xmath46 .
given this definition of @xmath41 , @xmath47 is the effective radius of the galaxy .
the @xmath28- and @xmath2-band brightness profiles for each galaxy were also fit with a `` core - srsic '' model , @xmath48^{\gamma / \alpha } \exp \biggl [ -b_n \biggl ( { r^{\alpha } + r_b^{\alpha } \over r_e^{\alpha } } \biggr ) ^{1/(\alpha n ) } \biggr ] , \end{array } \label{eq2}\ ] ] where @xmath49 \label{eq3 } \end{array}\ ] ] this model , which was first proposed by graham ( 2003 ) , consists of a power - law component in the inner region of a galaxy , which `` breaks '' to a traditional srsic profile beyond some radius , @xmath50 .
the model has a total of six free parameters : the logarithmic slope of the inner power - law ( @xmath51 ) ; the shape of the srsic function ( @xmath37 ) ; the break radius ( @xmath52 ) ; the effective half - light radius of the srsic profile ( @xmath47 ) ; the intensity at the break radius ( @xmath53 ) and a parameter ( @xmath54 ) which governs the sharpness of the transition between the inner power law and the outer srsic function .
after some experimentation , it was decided to use the modified parametrization of trujillo ( 2004 ) , @xmath55 @xmath56 \label{eq4}\ ] ] in which @xmath57 and @xmath58 is the heaviside step function .
this model produced more stable fits , with better consistency between the five remaining parameters ( @xmath53 , @xmath51 , @xmath37 , @xmath47 and @xmath52 ) measured in the @xmath28 and @xmath2 bandpasses .
equations ( 1 ) and ( 4 ) are intended to describe the profiles of galaxies which have no central nucleus .
however , it is obvious that many galaxies in our sample do indeed have compact sources at or near their centers . for such nucleated galaxies , a single - component king model ( michie 1963 ; king 1966 )
was used to represent this central component .
this introduces three additional parameters to the fit : the total intensity of the nucleus ( @xmath59 ) ; the projected half - light radius ( @xmath60 ) ; and the king concentration index ( @xmath61 ) . in other words , for _ nucleated galaxies _ , the fitted model , @xmath62 , takes the form @xmath63 where @xmath64 is either a pure srsic model ( equation 1 ) or a core - srsic model ( equation 4 ) , depending on the galaxy in question , and @xmath65 is the central king model component . for non - nucleated galaxies ,
the profiles are fit simply with models of the form of equations ( 1 ) or ( 4 ) .
a detailed justification for the choice of galaxy model ( i.e. , srsic vs. core - srsic ) is given on a case - by - case basis in paper vi .
we adopt these classifications verbatim , with the exception of three intermediate - luminosity galaxies : vcc543 ( ugc7436 ) , vcc1528 ( ic3501 ) and vcc1695 ( ic3586 ) . for all nucleated ( type ia ) core - srsic galaxies ; @xmath66 for the four galaxies in this category .
] while the _ global _ brightness profiles of these galaxies are adequately represented by srsic models , such models overpredict the amount of galaxy light on subarcsecond scales . for the purposes of measuring photometric and structural properties for the nuclei in these galaxies , we parameterize the galaxy profiles with core - srsic models in all three cases .
we note that the definition of a nucleus " invariably hinges on some assumption explicit or otherwise about the intrinsic brightness profile of the host galaxy .
our study is no exception in this regard .
choices for the galaxy profiles made by previous workers have included king models ( binggeli & cameron 1993 ) , pure exponentials ( binggeli & cameron 1993 ; stiavelli 2001 ) , nuker laws ( rest 2001 ; ravindranath 2001 ; lauer 2005 ) and sersic profiles ( durrell 1997 ; stiavelli 2001 ) .
after considerable experimentation ( paper iv ) , we opted to use the family of models represented by equations ( 1 ) and ( 2 ) because they have the great advantage they are flexible enough to provide accurate fits to the brightness profiles of both giant and dwarf galaxies ( see paper vi ) .
the use of a single ( srsic ) model to describe the full sample of galaxies also seems advisable in light of mounting evidence that , at least in terms of their _ structural parameters _ , the longstanding perception of a fundamental dichotomy between giant and dwarf ellipticals ( e.g. , kormendy 1985 ) may be incorrect ( see , e.g. , jerjen & binggeli 1997 ; graham & guzman 2003 ; paper vi ) . from a theoretical perspective
, the choice of equations ( 1 ) and ( 4 ) also seems reasonable given recent findings that the srsic law provides an accurate representation of the spatial and surface density profiles of dark matter halos in high - resolution @xmath67cdm simulations ( navarro et al .
2004 ; merritt et al .
2005 ) .
at the same time , our decision to parameterize the central nuclei with king models is motivated by high - resolution observations of the nuclei in nearby galaxies . in nucleated local group galaxies such as ngc205 and sagittarius ,
king models are found to provide accurate representations of the central components ( e.g. , djorgovski 1992 ; butler & martnez - delgado 2005 ; monaco 2005 ) .
nevertheless , for galaxies at the distance of the virgo cluster , we are working close to the limits of resolvability , so we caution that our choice of king models to parameterize the central components may not be unique , particularly for faint nuclei in the highest surface brightness galaxies .
alternative parameterizations of the central brightness cusps " in our sample galaxies will be explored in a future paper in this series ( merritt 2006 ) . as described in paper ii , our analysis of the nuclei , brightness profiles , and isophotal structure of the galaxies
is based on f475w and f850lp images in which a _ gaussian _
kernel is used to distribute flux onto the output ( drizzled ) images .
this choice of kernel has the advantage that , relative to _ lanczos3 _ kernel , bad pixels can be repaired more effectively , albeit with the penalty of a slight reduction in angular resolution . due to the compact nature of the nuclei ( even the most extended objects have effective radii @xmath68 ) , it is important that the effects of the psf are taken into account when fitting models to the observed brightness profiles .
psfs in the f475w and f850lp filters , varying quadratically with ccd position , were derived using daophot ii as described in paper ii .
briefly , archival images of the galactic globular cluster ngc104 ( 47 tucanae ) taken during programs g0 - 9656 and go-9018 were used to construct empirical psfs in the two bandpasses .
these archival images were drizzled in the same manner as the images for the program galaxies .
a total of @xmath1 200 stars in each filter were used to construct the psfs , which extend to a radius of 05 in both bandpasses .
to follow the behavior of the psfs to still larger radii , we matched our empirical psfs at a radius of 03 to those measured for high - s / n composite stars by sirianni ( 2005 ) .
these latter psfs extend to radii of 3 , and were constructed from images of 47 tuc fields taken as part of the photometric calibration of acs .
figure [ fig01 ] shows azimuthally averaged psfs for the f475w and f850lp filters measured at the position of the nucleus in vcc1303 ( ngc4483 ) the program galaxy whose center is nearest to the mean position for the full sample of program galaxies . a @xmath69 minimization scheme was used to find the models which best fit the azimuthally - averaged , one - dimensional intensity profiles for each galaxy .
minimizations were carried out using the minuit package in the cern program library ; initial determinations of the minima , obtained using a simplex minimization algorithm ( nelder & mead 1965 ) , were later refined using a variable metric method with inexact line search ( migrad ) .
following byun et al .
( 1996 ) , all points in the profile were assigned equal weight .
for both nucleated and non - nucleated galaxies , the psfs at the location of the galaxy s center were convolved with the models before fitting to the intensity profiles .
customized psfs were created for each galaxy in the survey , centered at the exact ( sub - pixel ) location of the nucleus . while , in practice , the psf convolution has little impact on the fitted srsic or core - srsic model parameters , with the exception of @xmath51 , this step is critically important when evaluating accurate structural parameters for the central nucleus .
profile fits are carried out independently in the two bandpasses , with the exception of the 11 nucleated galaxies brighter than @xmath70 ( i.e. , type ia galaxies ; see [ sec : results ] ) .
our numerical experiments suggest that in this high surface brightness regime , the profile of the underlying galaxy makes the measurement of nuclei half - light radii and total magnitudes extremely challenging ( see appendix a ) . for these galaxies ,
the composite @xmath28- and @xmath2-band profiles were first fitted simultaneously and the individual fits constrained so that the galaxy shape index parameter , nucleus concentration index and half - light radius were the same in the two bandpasses . when dust is present ( see below ) , the models are fitted to the dust - corrected surface brightness profiles if @xmath71 50% of the points along a given isophote are affected ; otherwise the dust affected regions are masked .
more details on the correction for dust obscuration are given in paper vi .
sufficiently compact nuclei will appear unresolved even in our acs images . to estimate the resolution limit of our observations
, we constructed brightness profiles for a number of likely stars which appear in our images .
these candidate stars were classified as unresolved in the object catalogs produced by kingphot , the reduction package used to measure structural and photometric parameters for the globular clusters in these fields ( see papers ii and x ) .
fitting king models to the brightness profiles of these objects gives median half - light radii of 0011@xmath720004 and 0018@xmath720005 in the f475w and f850lp bandpasses , respectively . as an additional test , we may make use of the fact that vcc1316 ( m87 = ngc4486 ) , one of the agn galaxies in our survey ( see below ) , contains a prominent non - thermal central point source .
although this source is saturated in our f475w images , a king model fitted to the central source in the @xmath2-band brightness profile gives @xmath73 . in what follows ,
we adopt a conservative upper limit of 0025 @xmath1 2 pc for the resolution limit in both bandpasses . before proceeding ,
we pause to demonstrate that the vast majority of the nuclei belonging to our program galaxies are indeed more extended than point sources . in figure
[ fig02 ] , we show @xmath28-band surface brightness profiles for a representative sample of nine nucleated galaxies , chosen to span the full range in fitted half - light radius ( with @xmath74 decreasing from left to right and from top to bottom ) . in each panel ,
the red curves show the results of fitting the nuclear component with a king model , while the blue curves show the results of fitting a central point source ; residuals from both fits are shown in the lower panel . with the exception of vcc1528 ( ic3501 ) , the central nucleus is resolved for all of the galaxies in figure [ fig02 ] . in total , six galaxies in our sample vcc1883 ( ngc4612 ) , vcc140 ( ic3065 ) , vcc1528 , vcc1695 ( ic3586 ) , vcc1895 ( ugc7854 ) and vcc1826 ( ic3633 ) have best - fit half - light radii , measured in at least one bandpasses , that fall below our nominal resolution limit of 0025 .
these half - light radii are given in parantheses in table [ tab : data ] .
they have been included in the following analysis , but we caution that they are formally unresolved in our acs images .
we shall return to the issue of these compact nuclei in [ sec : discussion ] .
additional tests on the resolution limits , possible biases in the derived photometric and structural parameters , and a discussion of measurement errors , are given in 4.1 and appendix a.
as many as eighteen of the 100 galaxies in table [ tab : data ] show evidence for dust either as isolated patches and filaments , or in the form of disks having varying degrees of regularity ( see paper vi ) . for the most part
, this dust has no impact on the identification of possible nuclei .
however , for four galaxies in our sample ( i.e. , vcc1535 = ngc4526 , vcc1030 = ngc4435 , vcc685 = ngc4350 and vcc571 ) the central dust obscuration is severe enough to make a reliable classification of these galaxies as nucleated or non - nucleated impossible .
moreover , for vcc1535 and vcc1030 , both of which harbour massive , kpc - scale dust disks , the surface brightness profiles are themselves so limited that it is not even possible to place the galaxies in the appropriate srsic or core - srsic categories . in general , the census of active galactic nuclei ( agn ) in intrinsically faint galaxies
and in the acs virgo cluster survey galaxies in particular is far from complete .
however , two of the brighter galaxies in our sample ( vcc1316 = ngc4486 , m87 , 3c 274 and vcc763 = ngc4374 , m84 , 3c 272.1 ) are known to host agns with strongly non - thermal spectral energy distributions ( e.g. , wrobel 1991 ; ho 1999 ; chiaberge , capetti & celotti 1999 ) . in both cases , the unresolved non - thermal nucleus is clearly seen in the acs images ; in neither instance , however , does there appear to be a resolved stellar nucleus . a third galaxy ( vcc1619 = ngc4550 ) , is classified as a liner by ho ( 1997 ) .
this galaxy contains some dust within the central @xmath75 25 , but there is clear evidence for a resolved stellar nucleus .
wrobel ( 1991 ) detected nuclear radio emission in three other galaxies in our survey ( vcc1226 = ngc4472 , m49 ; vcc1632 = ngc4552 , m89 ; and vcc1978 = ngc4649 , m60 ) . in both vcc1226 and vcc1632 , the innermost @xmath75 1
are slightly obscured by dust ( see paper vi ) , but once a correction for dust obscuration is performed , there is no evidence of a stellar nucleus in either case .
we see no sign of a nucleus in vcc1978 .
a search for low - level agn in our program galaxies is now underway using low- to intermediate - resolution ground - based optical spectra , the results of which will be presented in a future paper in this series .
these spectroscopic data will be useful in establishing the extent to which non - thermal sources are responsible for , or contribute to , the central luminosity excesses observed in a number of these galaxies . for the time being , table [ tab : class ] summarizes our classifications for the program galaxies , as discussed in the next section .
we begin by defining a class of galaxies ( type 0 ) in which dust obscuration ( four galaxies ) or agn emission ( two galaxies ) renders a reliable classification as nucleated or non - nucleated impossible . in what follows
, we shall limit our analysis to the remaining 94 galaxies . as a first step in the identification of nuclei in our program galaxies ,
the @xmath28- and @xmath2-band surface brightness profiles were each fitted with the appropriate galaxy model ( i.e. , either a pure srsic or core - srsic model ) outside a geometric mean radius of 05 . those galaxies with brightness profiles which lay systematically above the inward extrapolation of fitted model for @xmath76 were considered to be nucleated . because many of the nuclei are somewhat bluer than the underlying galaxies , a central excess was often more apparent in the @xmath28-band profile than in the redder bandpass .
in addition to classifying the galaxies on the basis of their brightness profiles , the f475w and f850lp images for each galaxy were carefully inspected for the presence of a distinct central excess .
using these two criteria , a total of 62 galaxies were found to show clear evidence for a central nucleus ; such galaxies are classified as type ia or ib . unfortunately , for 11 of these 62 galaxies , although the presence of a faint central component could be established from the images themselves or from the brightness profiles , the nucleus itself was too faint to allow us to recover trustworthy photometric or structural parameters from the surface brightness profiles .
such galaxies are referred to as type ib in tables [ tab : data ] and [ tab : class ] .
our analysis of the structural and photometric properties of the nuclei is therefore based on the subset of 51 nucleated galaxies for which it was possible to obtain a reliable fit to the central brightness profiles : i.e. , type ia galaxies .
the type ib galaxies are classified as nucleated for the purposes of computing the overall frequency of nucleation , but their nuclei are omitted from the analysis in 4.4 to 4.8 .
of the remaining 9462 = 32 galaxies , five may have nuclei which are offset by @xmath77 or more from the centers of the isophotes ( type ie galaxies ) .
we consider these five galaxies in more detail in 4.3 .
the remaining 94625 = 27 objects consist of galaxies which are either unquestionably non - nucleated , or galaxies with uncertain classifications .
as described in appendix a , we have carried out a series of experiments in which simulated nuclei having sizes and luminosities that obey the empirical scaling relations found in [
sec : results ] , are added to and removed from the observed brightness profiles . by re - fitting the brightness profiles obtained in this way
, we aim to refine the nuclear classifications of these galaxies .
to summarize our conclusions from these simulations , we classify 12 of these 27 galaxies as _ certainly non - nucleated _ ( type ii ) , 11 as _ possibly nucleated _ ( type i d ) and four others as _ likely nucleated _ ( type ic ) .
these classifications are reported on a case - by - case basis in table [ tab : data ] , and summarized for the entire sample in table [ tab : class ] .
figure [ fig03 ] shows f475w images for the central 10@xmath7810 regions ( @xmath1 800@xmath78800 pc ) of all 100 galaxies in the survey . each galaxy is labelled according to the classifications from table [ tab : data ] .
azimuthally - averaged surface brightness profiles in the @xmath28-band for all 100 galaxies are shown in figure [ fig04 ] .
for the 51 type ia galaxies shown in this figure , the dashed and dotted curves indicate the best - fit models for the nucleus and galaxy , while the combined profile is shown by the solid curve . for all remaining galaxies ,
the solid curve simply shows the best - fit srsic or core - srsic model .
note that no fit was possible for either vcc1535 or vcc1030 , the two galaxies with the most severe dust obscuration .
open symbols in figure [ fig04 ] denote datapoints that were omitted when fitting the galaxy profile ( e.g. , the innermost datapoints for galaxies which contain nuclei too faint to be fit reliably , outer datapoints for the close companions of luminous giant galaxies , and , occasionally , datapoints corresponding to pronounced rings , shells , or other morphological peculiarities ) .
given that independent fits of the @xmath28- and @xmath2-band brightness profiles are performed for the type ia galaxies , it is natural to ask how well the photometric and structural parameters of the nuclei measured in the two bands agree .
the first two panels of figure [ fig05 ] compare the king model half - light radii , @xmath60 , and total magnitudes , @xmath79 and @xmath80 , measured from the separate profiles ( filled circles ) .
note that for 11 of these 51 galaxies ( i.e. , those objects with @xmath81 ) , the king concentration index and half - light radii of the nuclei were constrained to be the same in the two bandpasses ; these nuclei are plotted as open stars in the first panel of figure [ fig05 ] .
in addition , we include in this figure the five galaxies with possible offset nuclei , bearing in mind that in these cases , the @xmath60 , @xmath79 and @xmath80 measurements were carried out in a rather different way ( see 4.3 for details ) .
the open circles show the nuclei of these five galaxies .
the third panel of figure [ fig05 ] compares two estimates for the color of the nuclei : i.e. , that obtained by integrating the best - fit @xmath28- and @xmath2-band king models , @xmath82 , and an aperture color , @xmath83 , obtained using a circular aperture of radius 4 pixels ( 020 @xmath1 16 pc ) applied to the nucleus of the galaxy - subtracted image .
the mean difference between the total and aperture colors is 0.018 mag , in the sense that the aperture colors are slightly redder .
the @xmath84 scatter in the measured radii and colors is found to be @xmath85 0007 and @xmath86 0.059 mag , respectively . assuming the latter uncertainty arises equally from errors in the @xmath28 and @xmath2 bands , we find @xmath87 0.041 mag for the nuclei magnitudes .
we adopt these values for the typical uncertainties on the fitted radii , colors and magnitudes , bearing in mind that additional systematic errors ( e.g. , in the photometric zeropoints or in the construction of the psfs ) may affect the measurements . in any case
, we conclude from figure [ fig05 ] that there is excellent internal agreement between the measured sizes , colors and magnitudes .
vcc classifications for our program galaxies are given in table [ tab : data ] where column ( 8) reports the classification from bst85 : y means nucleated , n means non - nucleated .
our new classifications are given in column ( 9 ) .
column 10 indicates which type of model was used to represent the galaxy : s " = srsic or
cs " = core - srsic .
the most basic property of the nuclei which might serve as a constraint on theories for their origin is the overall frequency , @xmath88 , with which they are found in our program galaxies . among the 94 galaxies which can be reliably classified as either nucleated or non - nucleated
, we find 62 galaxies , or @xmath89% of the sample to show clear evidence for a central nucleus ( types ia and ib ) .
however , we believe this estimate should be considered a firm lower limit on the frequency . including the type ic galaxies ( which are very likely to be nucleated but could not be classified as such unambiguously ) , gives @xmath90% .
if one also includes the type i d galaxies , which _ may _ be nucleated , one then finds @xmath91% .
finally , if all five galaxies with possible offset nuclei are included ( although we caution in 4.3 that the weight of evidence argues against doing so ) , the percentage of nucleated galaxies could be as high as @xmath92% .
while the true frequency probably lies between these extremes , it is nevertheless striking to think that , among our sample of 94 classifiable galaxies , in only 12 cases can the _ absence _ of a nucleus be established with any degree of certainty . among
the 100 elliptical , lenticular and dwarf galaxies in the acs virgo cluster survey , 24 dwarf galaxies ( de , ns ) and one e galaxy were classified as nucleated in the original vcc ( bst85 ; see also table [ tab : data ] of paper i ) .
the frequency of nucleation which we derive here , @xmath93 6687% , is much higher than the value of @xmath94 found using the classifications of bst85 , and represents a sharp upward revision of the nucleation frequency for early - type galaxies in this luminosity range .
there are several reasons why such a discrepancy should come as no surprise .
to the best of our knowledge , ours is the first systematic census of nuclei in early - type galaxies that includes both dwarf and giant galaxies ( spanning a factor of @xmath75 460 in blue luminosity ) .
more importantly , the studies of bst85 and bts87 , along with most of the major subsequent studies of dwarf galaxies and their nuclei ( e.g. , binggeli & cameron 1991 ; 1993 ) , were based on visual inspection of photographic plates . as pointed out in 1 and stressed by bst85 themselves ,
the vcc classifications are known to be incomplete fainter than @xmath21 23 ( @xmath22 ) and to suffer from surface brightness selection effects for the luminous e and s0 galaxies .
clearly , selection effects of this sort are less of an issue for our survey , where the identification of the nuclei is relatively straightforward thanks to the depth and high spatial resolution of the acs images .
in any case , care must be taken when comparing our measurement to previous estimates since the frequency of nucleation is known to depend on the luminosities of the galaxies under consideration ( e.g. , sandage , binggeli & tammann 1985 ) .
figure [ fig06 ] shows the luminosity functions for our sample of 62 nucleated galaxies ( types ia and ib ) as the double - hatched histogram ; the hatched histogram shows this same sample plus the 15 likely or possibly nucleated galaxies of types ic and i d ( i.e. , 77 galaxies in total ) .
for comparison , the 25 galaxies classified as nucleated by bst85 are shown by the filled histogram , and the open histogram shows the distribution of the 94 classifiable galaxies from the acs virgo cluster survey . as expected , the disagreement between our classifications and that of bst85 is quite dramatic for galaxies brighter than @xmath95 .
this happens to be the approximate dividing point between dwarf and giant galaxies in the vcc , which strongly suggests that the disagreement is the result of selection effects that made it difficult or impossible for bst85 to identify nuclei in bright , high - surface - brightness galaxies . for @xmath96 ,
there is better agreement although we still find significantly more nuclei even among these faint galaxies : i.e. , we classify 46 of 53 galaxies , or @xmath97 , of this subsample as nucleated , compared to just 25/56 ( @xmath1 47% ) using the bst85 classifications .
the luminosity dependence of @xmath88 is shown explicitly in the lower panel of figure [ fig06 ] .
a vivid demonstration of the importance of surface brightness selection effects when classifying nuclei is shown in figures [ fig07 ] and [ fig08 ] .
the first of these figures compares the distribution of galaxy surface brightnesses , measured at a geometric mean radius of 1 , for the same four samples shown in the figure [ fig06 ] .
by contrast , figure [ fig08 ] shows nucleus magnitude as a function of galaxy surface brightness measured at a geometric mean radius of 1 .
filled symbols show the 51 type ia galaxies in our sample , while the open squares show the 25 galaxies classified as nucleated by bst85 .
open circles in this figure denote the five galaxies with possible offset nuclei .
figures [ fig07 ] and [ fig08 ] leave little doubt that the survey of bst85 preferentially missed nuclei in the bright , high - surface - brightness galaxies .
we further note that a recent survey of 156 virgo dwarfs with the wide field camera on the isaac newton telescope uncovered faint nuclei in 50 galaxies previously classified as non - nucleated , consistent with our upward revision for frequency of nucleation ( grant , kuipers & phillipps 2005 ) .
of course , it is conceivable we _ too _ may be missing faint nuclei in the highest surface brightness galaxies ; it is certainly true that the 11 galaxies for which we are unable to measure reliable photometric or structural parameters for the nuclei are among the highest surface brightness galaxies in our survey .
accordingly , we stress once again that the estimate of @xmath98% from 4.1 , which is based on galaxies with unambiguous nuclei , _ is certainly a lower limit to the true frequency of nucleation among our sample of early - type galaxies_. we shall return to this point in 5.2 ( see also appendix a ) . figures [ fig09 ] and [ fig10 ] illustrate the importance of _ hst _ imaging for the identification of nuclei in these galaxies . in figure [ fig09 ]
we show a comparison between the co - added f475w image for vcc2048 ( ic3773 ) a type ia galaxy with three simulated ground - based images for this same galaxy . in these three cases ,
the co - added f475w frame has been binned @xmath99 and convolved with gaussians having dispersions of 1 , 2 and 3 pixels , corresponding to fwhm of 05 , 09 and 14 .
it is clear that seeing effects alone make the detection of faint , compact nuclei challenging under normal conditions of ground - based seeing .
this finding is all the more sobering when one considers that vcc2048 , classified as ds0(9 ) in the vcc , was thought on the basis of the original bst85 classifications to be the brightest non - nucleated dwarf galaxy in our sample .
the first two panels of figure [ fig10 ] compare the f475w image for vcc784 ( ngc4379 ) with a @xmath100-band image taken with the 2.4 m hiltner mdm telescope on 21 april , 1993 in conditions of 114 seeing . this galaxy , one of the brightest type ia galaxies in our survey , was also classified as non - nucleated in the study of bst85 . as the third panel of figure [ fig10 ] demonstrates
, there is no hint of a central nucleus in the ground - based surface brightness profile , despite the fact that the nucleus , which is clearly visible in the acs brightness profile , is among the brightest and largest in our sample .
as noted in 1 , a few _ hst _ studies had previously revealed the presence of compact nuclei in bright early - type galaxies ( e.g. , rest 2001 ; ravindranath 2001 ; lauer 2005 ) . while these programs preferentially focussed on distant , high - luminosity ellipticals and lenticulars with 80% of the galaxies in these respective surveys having absolute magnitudes brighter than @xmath101 and @xmath102 , compared to @xmath103 for the present survey there is nevertheless some overlap with our sample at the bright end due to the large number of luminous e and s0 galaxies in the virgo cluster . in this section
, we compare our nuclear classifications with those reported in these previous surveys , limiting the comparison to those galaxies in our survey which have unambiguous classifications ( i.e. , types ia , ib and ii ) . for completeness
, we also compare our classifications for three faint galaxies to those of lotz ( 2004 ) who carried out a wfpc2 snapshot survey of early - type dwarf galaxies in the virgo and fornax clusters .
table [ tab : comp ] summarizes the nuclear classifications for galaxies in common with the surveys of rest ( 2001 ) , ravindranath ( 2001 ) , lauer ( 2005 ) and lotz ( 2004 ) .
the rest ( 2001 ) study presented wfpc2 ( f702w @xmath104 ) imaging for 67 early - type galaxies between 6 and 54 mpc , with a mean distance of @xmath1059 mpc . to minimize spurious detections , rest ( 2001 ) adopted rather conservative criteria in their search for nuclei , identifying nucleated galaxies as those objects which showed a central excess , along both the major and minor axes , over the best - fit nuker " model inside a radius of 015 .
based on these criteria , they identified nuclei in 9 of their 67 galaxies ( 13% ) .
no structural and photometric parameters were measured for the nuclei .
there are six galaxies in common between their survey and ours .
we find reasonable agreement between the two studies , with the exception of vcc731 ( ngc4365 ) : rest ( 2001 ) report no nucleus in this galaxy , whereas we find a small , but definite , central brightness excess .
accordingly , we classify this galaxy as type ib ( i.e. , certainly nucleated ) . the nicmos study of ravindranath ( 2001 )
was carried out using f160w ( @xmath106 ) images from the nic2 ( fwhm = 017 , scale = 0076 ) and nic3 ( fwhm = 022 , scale = 02 ) cameras . for 33 galaxies with distances in the range 7 to 69 mpc and @xmath10714 mpc
, these authors fitted two - dimensional , psf - convolved nuker " models to their nicmos images . compact sources consisting of narrow , psf - convolved gaussians were then included for those galaxies whose one - dimensional surface brightness profiles showed evidence for a central excess ( 14 of their 33 galaxies ) .
fwhms and magnitudes for the nuclei were then obtained by @xmath69 minimization .
we find good agreement with the ravindranath ( 2001 ) classifications .
specifically , we confirm the absence of nuclei in vcc1226 and vcc881 ( m86 = ngc4406 ) . for vcc763 ( m84 = ngc4374 ) , which is classified as nucleated by these authors , we confirm the presence of a central point source , although the galaxy is classified as type 0 in table [ tab : data ] due to the presence of strong agn activity .
all of the nuclei in the study of ravindranath ( 2001 ) were found to be unresolved point sources , although this is probably a consequence of the relatively poor resolution of their images : i.e. , at the mean distance of their sample galaxies , the nicmos fwhm corresponds to @xmath75 20 pc .
the wfpc2 study of lauer ( 2005 ) was based on f555w or f606w ( @xmath108 ) imaging for 77 galaxies ; a little more than half of their galaxies ( 45 ) were also imaged in the f814w ( @xmath109 ) bandpass .
the galaxies have distances in the range 10 to 97 mpc , with mean @xmath110 , so the 007 fwhm for the pc1 ccd corresponds to a physical scale of 11 pc for the typical galaxy .
magnitudes and colors for the nuclei in their sample identified as an excess above the nuker " model which best fits the observed brightness profile were derived by direct integration of the model residuals . in only two of their 25 nucleated galaxies
did the nucleus appear resolved .
there are seven acsvcs galaxies having unambiguous nuclear classifications which are in common with lauer ( 2005 ) .
the classifications are in agreement in three cases : vcc1978 , vcc731 and vcc1903 ( m59 = ngc4621 ) . for vcc1146 ( ngc4458 ) , which lauer ( 2005 )
classify as non - nucleated , we believe the discrepancy may be due to the highly extended nature of the nucleus . with @xmath111 08 = 62 pc ,
it is largest nucleus in our sample , and would be difficult to distinguish from the underlying galaxy profile in brightness profiles of limited radial extent ; the lauer ( 2005 ) brightness profile for this galaxy covers just the inner 5 .
the three remaining galaxies vcc1226 , vcc881 and vcc1632 are listed as nucleated in lauer ( 2005 ) , but we classify each of these galaxies as type ii ( non - nucleated ) . we speculate that the detection of nuclei in vcc1226 and vcc1632 is an artifact resulting from the presence of dust in both galaxies , which partly obscures the innermost @xmath75 1 .
lauer ( 2005 ) do not correct their images for dust obscuration ; once such a correction is performed , we find no indication of a central nucleus in either galaxy ( see also paper vi ) .
in the case of vcc881 , which is classified as nucleated by lauer ( 2005 ) , a faint continuum enhancement is indeed detected in both the @xmath28 and @xmath2 bands at the central location .
this feature would certainly be enhanced by the deconvolution procedure applied by lauer ( 2005 ) to their data .
however , it is unclear whether this corresponds to a stellar nucleus .
if one assumes that vcc881 follows the scaling relation between nucleus and galaxy luminosity obeyed by the rest of the sample , then the putative nucleus would be underluminous by a factor of @xmath75 250 .
furthermore , starting around 04 , the surface brightness profile of vcc881 _ decreases _ towards the center ( carollo 1997 ) .
the origin of this central surface brightness depression is unclear ( lauer 2002 ; paper vi ) : e.g. , an intrinsic decrease in the luminosity ( or mass ) density , or perhaps obscuration by gray dust , might be responsible . since either processes could produce a modest and localized continuum enhancement such as the one seen at the nuclear location , we believe this galaxy is best classifed as non - nucleated ( type ii ) .
we further note that there is no evidence of nuclear activity in vcc881 from its optical , radio and x - ray properties ( ho 1997 ; rangarajan 1995 ; fabbiano 1989 ) .
lotz ( 2004 ) carried out wfpc2 ( f555w @xmath108 and f814w @xmath109 ) imaging for 69 des and de , ns , mostly belonging to the virgo and fornax clusters . in their analysis ,
a nucleus was identified as a bright compact object within 15 of the galaxy photocenter .
while the lotz ( 2004 ) survey tended to focus on fainter galaxies than does our survey ( i.e. , their program galaxies have absolute blue magnitudes in the range @xmath112 mag , with mean @xmath113 mag ) , there are three galaxies which appear in both studies : vcc9 ( ic3019 ) , vcc543 and vcc1948 . in the case of vcc9 and vcc1948 , the two studies agree in finding no evidence of a nucleus at the position of the galaxy photocenter .
however , we have identified both of these galaxies as possible examples of galaxies with offset nuclei ( see [ sec : offset ] ) .
although lotz ( 2004 ) do not comment on a possible offset nucleus in the case of vcc1948 , they state that :
vcc9 was originally classified as nucleated by binggeli ( 1985 ) , but its brightest globular cluster candidate is 18 from its center " . for comparison , we measure an offset of 191@xmath72007 for this object and , like lotz ( 2004 ) , conclude that it is probably a star cluster projected close to the galaxy photocenter , rather than a _ bonafide _ nucleus .
the remaining galaxy , vcc543 , appears in the list of non - nucleated galaxies in table 3 of lotz ( 2004 ) , although we find unmistakable evidence for a nucleus in this object ( see figures [ fig03 ] and [ fig04 ] ) that is offset by no more than 007@xmath72012 from the galaxy photocenter .
finally , we note that two recent papers ( de propris 2005 ; strader 2006 ) have examined the properties of nuclei belonging to subsets of the galaxies from the acs virgo cluster survey , based on the same observational material used in this paper .
a detailed comparison of the sizes , magnitudes and colors we measure for the nuclei with those reported by de propris ( 2005 ) and strader ( 2006 ) is given in appendix b. before proceeding , we pause to consider those galaxies that may have offset nuclei .
nuclei displaced from the photocenters of their host galaxies are potentially interesting since they may hold clues to the general processes which trigger and/or regulate the formation of nuclei in general . for instance , offsets may arise through the ongoing merging of globular clusters through dynamical friction ( tremaine , ostriker & spitzer 1975 ; miller & smith 1992 ) , the fading of stellar populations in dwarf irregular or blue compact dwarf galaxies as they evolve into dwarf ellipticals ( e.g. , davies & phillips 1988 ) , recoil events following the ejection of a supermassive black hole from the nucleus ( merritt 2004 ) or counter - streaming instabilities that develop in flat and/or non - rotating systems ( zang & hohl 1978 ; de rijcke & debattista 2004 ) . from an observational perspective , the identification of such nuclei is a complicated problem .
they are prone to confusion with globular clusters , foreground stars or background galaxies difficulties that are particularly serious in ground - based imaging , where the nuclei and contaminants will appear unresolved .
the most ambitious study of offset nuclei undertaken to date is that of binggeli , barazza & jerjen ( 2000 ) , who measured offsets for a sample of 78 nucleated dwarf galaxies in the virgo cluster using digitized images of blue photographic plates obtained in conditions of fwhm @xmath1 12 seeing .
they found offset nuclei to be commonplace , with @xmath114 in 45 ( 58% ) and @xmath115 in 14 of the objects ( 18% ) .
it is of interest to check these results given the small sizes of the measured offsets relative to the ground - based seeing disk , the absence of color information that might be used to identify contaminants , and the possibility of confusion with galactic stars , globular clusters and , to a lesser extent , background galaxies .
we have used our acs images to measure offsets for the nuclei of the 62 type ia and ib galaxies in our sample . in both the f475w and f850lp images for each galaxy ,
we first calculate the centroid of the nucleus and its corresponding uncertainty .
the location of the galaxy photocenter is then found by averaging the centers of ellipses fitted to the galaxy isophotes over the range 1 @xmath116 ( paper vi ) . the uncertainty on the position of the photocenter is taken to be the standard deviation about the mean ellipse center .
adding in quadrature the uncertainties for the position of the nucleus and photocenter then yields the uncertainties for the offset .
the results reported in column 17 of table [ tab : data ] are averages of the offsets measured from the f475w and f850lp images .
figure [ fig11 ] shows the measured offsets for the 62 galaxies .
offsets are shown both in arcseconds ( upper panel ) and in units of the effective radius of the galaxy , @xmath117 , taken from paper vi ( lower panel ) . in only three galaxies
do we see evidence for an offset as large as 1 . using a less restrictive criterion of @xmath118 05 , we find only five galaxies that may have offset nuclei ( i.e. , type ie galaxies ) .
these galaxies , which are shown as the open circles in figure [ fig11 ] , are : * _ vcc9_. this very low surface brightness galaxy has multiple bright sources near its photocenter ; it may be a dirr / de transition object and seems to contain a rich population of diffuse star clusters " ( paper xi ) .
in addition to the presumed nucleus , there is a second source about two magnitudes fainter which is located @xmath1 15 from the photocenter ( and a similar distance from the presumed nucleus ) .
both the color and the half - light radius of the presumed nucleus are similar to those of metal - poor globular clusters in our dwarf galaxies .
thus , it is conceivable that this galaxy has no nucleus at all . * _ vcc21 ( ic3025)_. more than a dozen bright sources are found in the inner regions of this very low surface brightness galaxy . based on its mottled appearance ,
this galaxy too should be re - classified as a dirr / de transition object .
the presumed nucleus is located @xmath1 @xmath119 from the galaxy photocenter , the smallest offset in our sample of five candidates .
there are , however , two fainter sources close by , and given the large number of sources in this galaxy , its transitional morphology , and the fact that the presumed nucleus has a very blue color of @xmath120 , we believe the evidence indicates that the nucleus " in vcc21 is probably a young star cluster . * _ vcc1779 ( ic3612)_. this highly flattened galaxy ( @xmath121 ) is noteworthy in that it contains dust filaments unusual for low- and intermediate - luminosity galaxies in our sample ( see paper vi ) . like vcc9 and vcc21
, this galaxy may be a dirr / de transition object .
the acs images reveal four bright sources , all of which may be globular clusters , near the galaxy center .
we identify the brightest of these sources , which is _ not _ the nearest to the center , as the putative nucleus .
if the nearest ( and second brightest ) source is instead identified as the nucleus , then the offset would be @xmath1 04 rather than @xmath1 05 . *
_ vcc1857 ( ic3647)_. this galaxy , another very low surface brightness object , has a very bright source located @xmath1 7from its center .
this is by far the largest offset for any galaxy in our sample , so the identification of this source as a nucleus should be viewed with some caution .
the color and half - light radius of the presumed nucleus are consistent with those expected for an otherwise normal ( metal - poor ) globular cluster . *
_ vcc1948_. the presumed nucleus in this galaxy , another very low surface brightness object , is located @xmath122 from the galaxy photocenter .
it is the brightest of several sources in the inner few arcseconds .
there also appears to be a very faint surface brightness excess " that is nearly coincident with the galaxy photocenter .
it is therefore possible that this galaxy may be a normal ( type ib ) nucleated galaxy , albeit one with an unusually faint nucleus .
if so , then the source identified as the possible nucleus may be a globular cluster .
we conclude that in every case there is considerable uncertainty regarding the nature of the presumed offset nucleus .
it is possible and we consider it likely that the offset nuclei " in all five of these galaxies are merely globular clusters residing in non- or weakly - nucleated galaxies . are there nuclei with even smaller offsets ?
the nuclei of four other galaxies ( vcc1539 , vcc2019 , vcc1895 and vcc1695 ) have offsets @xmath123 and , in fractional terms , there are four other galaxies that have nuclei offset by more than 1% of the galaxy effective radius ( vcc2019 , vcc1199 , vcc1695 and vcc1895 ) .
we remind the reader , however , that these offsets correspond to just two acs / wfc pixels , and should be confirmed using deeper , higher resolution imaging .
our only secure conclusion is that , _ at most _ , only five of the nucleated galaxies in our survey , or @xmath1 7% of the sample , have nuclei that are offset by more than 05 from their photocenters at least three times smaller than the value of @xmath75 20% found by binggeli ( 2001 ) .
moreover , we believe that most and perhaps _ all _ of the nuclei " in the type ie galaxies are probably globular clusters , so this should be considered a firm upper limit on the percentage of galaxies with nuclei offset by more than 05 . the actual percentage may in fact be zero .
a key result to emerge from the survey of bts87 was the discovery of a spatial segregration between nucleated and non - nucleated dwarf galaxies in virgo : the de , ns are more strongly concentrated to the cluster center than the des ( see , e.g. , figure 9 of bts87 ) .
a similar trend was later reported for the fornax cluster by ferguson & sandage ( 1989 ) .
our discovery of nuclei in many of the galaxies classified as non - nucleated by bst85 suggests that it is worth reconsidering this important issue . to facilitate comparison with the bts87 and bst85 results , we limit our analysis in this section to those galaxies fainter than @xmath96 , the approximate dividing point between dwarfs and giants in the vcc .
as it happens , this magnitude also divides the acs virgo cluster survey equally into two samples of 50 galaxies . in the left panel of figure [ fig12 ]
, the heavy solid curve shows the cumulative distribution of projected distances from the center of the virgo cluster for the 50 galaxies with @xmath124(j2000 ) = 12:30:49.4 and @xmath125(j2000 ) = 12:23:28 . ] using the vcc classifications , one finds this sample to be composed predominantly of dwarfs ( i.e. , 33 of 50 galaxies according to table 1 of paper i ) .
the dotted and dashed curves show the corresponding distributions for the nucleated ( 23 ) and non - nucleated ( 27 ) galaxies in this sample , once again using the vcc classifications .
a ks test confirms the visual impression from this figure that the nucleated galaxies in our survey exhibit the same trend noted by bts87 for the full sample of nucleated dwarfs in the vcc : i.e. , the de , n galaxies are more centrally concentrated than their non - nucleated counterparts . in the right panel of this figure
, we show the sample of 49 galaxies with @xmath96 which we are able to classify as nucleated or non - nucleated from our acs images ( heavy solid curve ) . excluding for the moment the five galaxies with possible offset nuclei , whose true nature is uncertain , we find 40 of 44 remaining galaxies to be nucleated ( dotted curve ) .
given the preponderance of nucleated galaxies , it is not surprising to see that a systematic difference in central concentration between the two populations is no longer apparent . of course , with just four non - nucleated galaxies in this regime , the sample falls below the minimum size needed for statistically reliable results using a ks test , but our point in showing this comparison is to stress again that the overwhelming majority of program galaxies with @xmath126 contain nuclei .
a definitive investigation into the spatial distribution of nucleated and non - nucleated galaxies in the virgo cluster would require acs imaging for many hundreds of galaxies .
nevertheless , we can speculate on the origin of the trend noted by bts87 .
it has long been known that galaxies in virgo show some segregation in terms of luminosity and morphology .
ichikawa et al .
( 1988 ) noted that the dwarf elliptical galaxies in the central regions of virgo appear to be larger and brighter than those in the cluster outskirts . at the same time , bts87 showed that the bright early - type galaxies ( e+s0 ) are less strongly concentrated to the cluster center than the faint ( de ) early - type galaxies ( see their figures 7 and 8) .
since central surface brightness is proportional to total luminosity for early - type galaxies , the bts87 finding implies that bright , high - surface - brightness dwarfs ( hsbds ) in virgo are more spatially extended than low - surface - brightness dwarfs ( lsbds ) . because the original bts87 classifications suffer from a serious surface brightness selection effect
in the sense that nuclei belonging to galaxies with central surface brightnesses @xmath127 mag arcsec@xmath128 will go undetected ; see figure [ fig08 ] the observed trend may simply be a consequence of this surface brightness selection effect .
to put this claim on a more quantitative footing , we have calculated the density profiles for hsbd and lsbd early - type galaxies in virgo , using a surface brightness of @xmath129 mag arcsec@xmath128 as a dividing point between the two subgroups . as shown in figure [ fig08 ]
, bts87 would have tended to classify lsbds as nucleated , while the hsbds would have been preferentially classified as non - nucleated .
a least - squares fit to our sample galaxies gives @xmath130 , so that @xmath127 mag arcsec@xmath128 corresponds to a total galaxy magnitude of @xmath131 .
restricting ourselves to the early - type members of virgo with @xmath132 , this leaves us with a total of 448 galaxies .
the upper limit of @xmath133 represents the approximate transition between dwarfs and giants in virgo , while the lower limit reflects the completeness limit of the bts87 survey . among this sample of 448 galaxies ,
there are 42 hsbds with @xmath134 , and 406 lsbds with @xmath135 .
figure [ fig13 ] shows the density profiles , @xmath136 , for these two populations . in calculating the profiles ,
we have excluded galaxies belonging to the m and w clouds , and discarded galaxies with declinations less than 9@xmath137 to guard against contamination from the virgo b subgroup centered on vcc1226 ( m49 ) .
fitting exponentials of the form @xmath138 gives scalelengths of @xmath139 and @xmath140 deg@xmath25 for the lsbd and hsbd populations , respectively .
in other words , there is a statistically significant tendency for the hsbd early - type galaxies the same galaxies which would preferentially be misclassified as non - nucleated in the vcc because of their high central surface brightnesses to be more spatially extended .
this is consistent with the interpretation that the differing spatial distributions for de and de , n galaxies noted by bts87 is , in fact , a consequence of their survey s surface brightness limit .
lotz ( 2004 ) and grant , kuipers & phillipps ( 2005 ) found that brighter galaxies tend to contain brighter nuclei .
the upper panels of figure [ fig14 ] plot the magnitudes of the nuclei against those of the host galaxies ; results for the @xmath28 and @xmath2 bandpasses are shown in the left and right panels , respectively . filled and open symbols show the results for 51 type ia and five type ie galaxies .
the dashed lines in these panels show the least - squares lines of best fit : @xmath141 and @xmath142 . for comparison ,
the solid lines show the best - fit relations , with ( fixed ) unity slope : @xmath143 the lower panels of figure [ fig14 ] show these same data in a slightly different form . in these panels , we plot the ratio of nucleus luminosity , @xmath144 , to host galaxy luminosity , @xmath145 , @xmath146 as a function of galaxy magnitude .
total luminosities for the nuclei were obtained by integrating the brightness profiles of the best - fit king model components ( see 3 ) .
these magnitudes are recorded in columns ( 11 ) and ( 12 ) of table [ tab : data ] .
galaxy luminosities are taken from paper vi , in which the best - fit galaxy model either srsic or core - srsic , as specified in column ( 10 ) of table [ tab : data ] was integrated over all radii to obtain the total luminosity .
the contribution of the nucleus itself was excluded in the calculation of @xmath145 .
the primary conclusion to be drawn from figure [ fig14 ] is that the nucleus - to - galaxy luminosity ratio does not vary with galaxy luminosity , although there is considerable scatter about the mean value . in terms of @xmath147 , the relations in equation 6 are equivalent to @xmath148 for the two bands , where the quoted uncertainties refer to the mean errors .
our best estimate for the mean nucleus - to - galaxy luminosity ratio is then @xmath149 this is well below previous estimates : only 5 of the 51 nucleated galaxies in figure 9 of binggeli , barazza & jerjen ( 2000 ) have nuclei with fractional luminosities smaller than this . while the discrepancy may partly be the result of different choices for the models used to parameterize the galaxy brightness profiles ( e.g. , binggeli et al .
2000 use king models for the galaxy when calculating the luminosity of the central excess ) , it is also true that the greater depth and sensitivity of the acs imaging allows us to identify fainter nuclei than is possible from the ground , while the high spatial resolution allows a more accurate subtraction of the underlying galaxy light .
the luminosity function of nuclei is one of the most powerful observational constraints on models for their formation .
for instance , one theory involves the growth of a central nucleus through mergers of globular clusters whose orbits have decayed because of dynamical friction ( tremaine , ostriker & spitzer 1975 ; tremaine 1976 ; lotz 2001 ) . while this scenario is consistent with the well known fact that the brightest nuclei have luminosities that exceed those of the brightest globular clusters ( e.g. , durrell et al .
1996 ; durrell 1997 ) , a reliable measurement for the luminosity function of the nuclei has been hard to come by due to the lack of high - resolution ccd imaging for large , homogenous samples of early - type galaxies .
the need for _ hst _ imaging in this instance is clear , since subtle differences in the subtraction of the galaxy light ( particularly the choice of model to represent the galaxy ) can lead to large differences in the inferred luminosities of the nuclei ( see , e.g. , section 5 of binggeli & cameron 1991 ) . in figure
[ fig15 ] , we plot the luminosity functions , in @xmath28 and @xmath2 , for the sample of 51 type ia nuclei given in table [ tab : data ] . a gaussian distribution , @xmath150 provides an adequate representation of the luminosity functions , although there is no physical justification for this particular choice of fitting function ( and it is likely that the luminosity function suffers from some degree of incompleteness at both the bright and faint ends ) .
moreover , if the mean luminosity of nuclei in early - type galaxies is indeed a roughly constant fraction , @xmath151 , of that of their host ( [ sec : lum ] ) , then the distribution shown in figure [ fig14 ] may largely be a reflection of the luminosities of the program galaxies .
with these caveats in mind , we overlay the best - fitting gaussian distributions in figure [ fig15 ] as the dotted curves .
fitted parameters and their uncertainties are recorded in table [ tab : lf ] . a core objective of the acs virgo cluster survey is a study of the globular cluster populations associated with early - type galaxies .
since many thousands of virgo globular cluster candidates have been identified in the course of the survey ( e.g. , papers ix , x and xi ) , it is possible to compare directly the luminosity functions of the nuclei with those of the globular clusters .
figure [ fig15 ] shows the @xmath28- and @xmath2-band luminosity functions for @xmath1 11,000 high - probability globular cluster candidates from the survey .
these objects were chosen to have globular cluster probability indices , @xmath152 , in the range @xmath153 ( see jordn 2006 for details ) .
a complete discussion of the globular cluster luminosity function is beyond the scope of this paper and will be presented in a future article . for the time being , we simply note that , brighter than the 90% completeness limits of @xmath154 mag and @xmath155 mag , the luminosity functions of the globular clusters in virgo ( which are dominated by the contributions from the brightest galaxies ) are well described by gaussian distributions with @xmath156 mag and reddening - corrected turnover magnitudes of @xmath157 mag and @xmath158 mag .
these gaussians are shown as the upper dotted curves in each panel of figure [ fig15 ] .
it is apparent that the luminosity function of the nuclei extends to higher luminosities than that of the globular clusters and that , irrespective of the functional form used to parameterize the luminosity function of the nuclei , their distribution is significantly broader than that of the globular clusters .
in addition , their distribution is displaced to higher luminosities than that of the globular clusters .
we measure this offset to be @xmath159 mag and @xmath160 mag in the two bands .
thus , on average , the nuclei are @xmath75 25 times brighter than a typical globular cluster .
we shall return to this point in 5.2 .
also shown in figure [ fig15 ] are seven probable ucd galaxies in the virgo cluster , drawn from paper vii and from haegan ( 2006 ) .
these objects were identified on the basis of magnitude and half - light radius from the same images used to study the nuclei and globular clusters .
although the ucd sample is limited in size , membership in virgo has been established for each object through radial velocity measurements , surface brightness fluctuation distances , or both .
furthermore , the mass - to - light ratios presented in paper vii demonstrate that at least some of these objects appear genuinely distinct from globular clusters .
several explanations for their origin have been proposed ; according to what is probably the leading formation scenario , they are the surviving nuclei of dwarf galaxies which have been extensively stripped by gravitational tidal fields in the host cluster ( e.g. , bassino 1994 ; bekki 2001 ) . the ucds shown in figure [ fig15 ] have luminosities which coincide with the peak of the nuclei luminosity function , which is certainly consistent with this threshing " scenario .
however , it is important to bear in mind that the luminosities of the ucds shown in figure [ fig15 ] are entirely due to the construction of the sample : i.e. , candidates from paper vii and haegan ( 2006 ) were _ selected _ to have @xmath161 mag and @xmath162 mag . prior to the launch of _ hst _
, virtually nothing was known about the sizes of the nuclei in dwarf galaxies .
a notable exception was the compact , low - luminosity nucleus of the local group dwarf elliptical galaxy ngc205 , which was measured to have @xmath163 04 = 1.4 pc by djorgovski et al .
( 1992 ) . this early estimate , which was based on deconvolved ground - based images , is in good agreement with more recent values obtained using acs surface brightness profiles ( e.g. , merritt 2006 ) .
however , measuring half - light radii for nuclei at the distance of virgo using ground - based images is impossible ( e.g. , caldwell 1983 ; sandage , binggeli & tammann 1985 ; caldwell & bothun 1987 ) .
for instance , using high - resolution cfht images for ten virgo dwarfs , durrell ( 1997 ) was only able to place an upper limit of @xmath164 0405 ( 3040 pc ) on the sizes of the nuclei .
even with the excellent angular resolution and spatial sampling afforded by _ hst_/acs , the measurement of structural parameters for the nuclei is challenging more so than for a typical virgo globular cluster because the nuclei are observed on a bright background which is varying rapidly in both the radial and azimuthal spatial directions . in their wfpc2 snapshot survey of dwarf galaxies in the virgo and fornax clusters ,
stiavelli ( 2001 ) and lotz et al .
( 2004 ) did not attempt to measure half - light radii for the nuclei , but they noted that these nuclei have sizes " less than 013 ( 10 pc ) . working from the same wfpc2 data for a subset of five nucleated dwarfs , geha , guhathakurta & van der marel ( 2002 ) derived half - light radii in the range 914 pc ( 011018 ) . with their greater depth and superior sampling of the instrumental psf ,
our acs images are better suited to the measurement of half - light radii than any previous dataset , including the wfpc2 imaging .
moreover , images are available in two filters , so it is also possible to carry out independent size measurements and identify possible systematic errors arising from uncertainties in the f475w and f850lp psfs .
as shown in figure [ fig05 ] , we have made such a comparison and find good agreement between the half - light radii measured in the different bandpasses , with a typical _ random _ measurement error of @xmath165 0007 .
we note that half - light radii measured for the nuclei of approximately two dozen of our program galaxies have recently been reported by de propris ( 2005 ) and strader ( 2006 ) .
appendix b presents a comparison of our structural and photometric parameters with those measured in these studies .
figure [ fig16 ] shows the distribution of half - light radii for the nuclei of type ia galaxies from table [ tab : data ] .
the distribution is evidently quite broad , with a peak at @xmath166 ( 4 pc ) and an extended tail to much larger radii ( @xmath167 @xmath1 62 pc ) .
the dashed line at 0025 ( @xmath1 2 pc ) in each panel shows our estimate for the resolution limit of the images used to characterize the properties of the nuclei .
00125 @xmath1 1 pc .
however , the negative lobes of this kernel makes it difficult to repair bad pixels , so the _ gaussian _ kernel is preferred for the analysis of the galaxy / nucleus surface brightness profiles . ] in both bandpasses , the median half - light of the nuclei in our sample is found to be @xmath168 ( 4 pc ) .
clearly , the nuclei have a size distribution that is different from that of the globular clusters .
in the latter case , the distribution is sharply peaked , with a typical ( and nearly constant ; paper x ) half - light radius of @xmath169 pc ( i.e. , @xmath75 30% larger than the resolution limit for the nuclei ) .
it is clear that the nuclei are not only _ brighter _ than typical globular clusters ( 4.5 ) but they are also , on average , larger .
there is , however , considerable overlap between the two distributions , and the most compact nuclei have half - light radii that are indistinguishable from those of globular clusters .
the ucds , on the other hand , have half - light radii which resemble those of the nuclei . as with the luminosity functions
, this agreement may be a consequence of the selection process : i.e. , ucd candidates were identified from the sample of sources with sizes in the range 01705 ( 1440 pc ) .
figure [ fig17 ] shows that there is a clear correlation between size and luminosity for the nuclei , in the sense that the brighter objects have larger half - light radii .
we have fit relations of the form @xmath170 to the data in figure [ fig17 ] , excluding both the offset nuclei and the 56 nuclei which fall below the nominal resolution limit of 0025 ( shown by the dashed lines in the two panels ) .
the solid lines in the two panels show the relations : @xmath171 this luminosity dependence constitutes another clear point of distinction between nuclei and globular clusters : the latter , both in the milky way ( van den bergh et al .
1991 ; paper vii ) and in our program galaxies ( paper x ) , have a near - constant size of @xmath172 pc .
this value is indicated by the arrows in figure [ fig17 ] .
nuclei fainter than @xmath173 mag and @xmath174 mag have typical half - light radii of 004 ( 3.2 pc ) , or about 20% larger than a typical globular cluster ; the brightest nuclei are an order of magnitude larger still . given their uncertain nature , it is worth noting that all five of the candidate offset nuclei from 4.3 have half - light radii close to the mean of the globular clusters .
it is interesting to see that the ucds which in figure [ fig16 ] were found to have half - light radii comparable to those of the largest nuclei are outliers in this size - luminosity plane .
compared to nuclei of comparable luminosity , the ucds are nearly three times larger , with @xmath111 0205 .
alternatively , one might consider the ucds to be @xmath75 2 mag underluminous for their size . in any event
, the fact that ucd candidates from paper vii were chosen to lie within a specific range of magnitude and half - light radius suggests that a general conclusion about systematic size differences between the two populations would be premature .
figure [ fig18 ] shows that there also exists a difference in surface brightness between the globular clusters and nuclei .
this figure plots the average surface brightness within the half - light radius , @xmath175 for these two populations . because of their near - constant size
, the globular clusters fall along a diagonal swath in this diagram .
note that the dashed line in figure [ fig18 ] is _ not _ a fit to the globular clusters , but simply the expected relation for clusters which obey equation 11 and have a constant half - light radius of @xmath176 pc .
although there is sizeable scatter , the nuclei have a mean surface brightness of @xmath177 mag arcsec@xmath128 in @xmath28 and @xmath178 mag arcsec@xmath128 in @xmath2 , although there is considerable scatter ( @xmath179 mag ) about these values .
the basic properties for the ucds and nuclei in virgo are compared in table [ tab : global ] . by virtue of their larger radii at fixed luminosity ,
the virgo ucds have surface brightnesses that are lower than those of comparably bright nuclei .
this is opposite to the claim of de propris et al .
( 2005 ) who argued that fornax ucds have _ higher _ surface brightness than the virgo nuclei .
however , these authors seem to base this conclusion on a visual comparison of the nuclei brightness profiles with that for their mean ucd " .
we have calculated the average surface brightness within the half - light radius for the four fornax ucds which have absolute magnitudes and half - light radii reported in their table 2 . in doing so
, we have transformed their @xmath100-band magnitudes into the @xmath28 and @xmath2 bandpasses using assumed colors of @xmath180 and @xmath181 , which are appropriate for old , intermediate - metallicity populations ( see table 3 of paper iii ) .
their radii have been converted from parsecs to arcseconds using their adopted fornax distance modulus of @xmath182 = 31.39 .
the resulting surface brightnesses for these four ucds are shown as the open squares in figure [ fig18 ] .
we find that the fornax and virgo ucds occupy similar locations in the diagram , and that both populations have _ lower _ surface brightness ( by @xmath75 2.5 mag arcsec@xmath128 in both bandpasses ) than comparably bright nuclei , contrary to the claims of de propris et al .
( 2005 ) .
finally , we note that the five candidate offset nuclei are observed to fall along the diagonal swath defined by the globular clusters in figure [ fig18 ] .
this strengthens the conclusion from 4.3 that these objects are globular clusters , rather than _ bonafide _ nuclei .
the first comprehensive investigation into the colors of nuclei in dwarf galaxies was the series of papers by caldwell ( 1983 ; 1987 ) and caldwell & bothun ( 1987 ) .
based on imaging of 30 dwarfs in the fornax cluster , caldwell & bothun ( 1987 ) found no evidence for a color difference between the nuclei and their host galaxies .
they did , however , find a correlation between nuclei luminosity and galaxy color , in the sense that the reddest galaxies tended to harbor the brightest nuclei . at a given luminosity
the nucleated galaxies were also found to be slightly redder than their non - nucleated counterparts .
a decade later , high - resolution cfht imaging for two virgo dwarfs ( durrell 1997 ) hinted at an apparent diversity in nuclei colors : in one galaxy ( vcc1254 ) , the nucleus was found to be significantly bluer than the galaxy , while in the case ( vcc1386 ) , the colors were indistinguishable .
recently , lotz et al .
( 2004 ) have carried out aperture photometry for the nuclei of 45 de , n galaxies in the virgo and fornax clusters using @xmath183 images from three wfpc2 snapshot programs .
they find that : ( 1 ) the nuclei are consistently bluer than the underlying galaxy light , with offset @xmath184(@xmath185 ) = 0.10.15 mag ; ( 2 ) the nuclei colors correlate with galaxy colors and luminosities , in the sense that the redder nuclei are found in the redder and brighter galaxies ; ( 3 ) and the nuclei are slightly redder than the globular clusters associated with the host galaxy .
our examination of the nuclei colors begins with figure [ fig19 ] .
the left panel of this figure shows the color - magnitude diagram for the nuclei of the 51 type ia galaxies ( filled circles ) and the five galaxies with possible offset nuclei ( open circles ) . for the type ia galaxies , the symbol size is proportional to the blue luminosity of the host galaxy . for reference ,
11 galaxies with nuclei redder than @xmath186 have been labeled .
note that one other galaxy , vcc21 , has a very blue nucleus with @xmath187 .
although it is listed in table [ tab : data ] as a possible example of a galaxy with an offset nucleus , we have argued in 4.3 , 4.6 and 4.7 that such offset nuclei " are likely to be misclassified star clusters .
in the case of vcc21 , the blue color of the object points to a young age ( i.e. , @xmath188 gyr for virtually any choice of metallicity ; see figure 6 of paper i ) .
this interpretation is consistent with the galaxy s dirr / de transitional morphology .
there are a number of noteworthy features in the color - magnitude diagram shown in figure [ fig19 ] .
first , both the colors and luminosities of the nuclei are seen to correlate with host galaxy luminosity , in the sense that the nuclei belonging to the most luminous galaxies are the brightest and reddest objects in our sample .
this finding is consistent with the trend noted by lotz et al .
even more striking , though , is the tendency for _ the nuclei themselves to follow a clear color - luminosity relation_. to the best of our knowledge , this is the first time such a trend has been detected .
the dashed line in figure [ fig19 ] shows the relation @xmath189 obtained from a least - squares fit to the 37 nuclei belonging to galaxies with @xmath190 .
while this relation provides an excellent description of the color - magnitude relation for the nuclei in faint galaxies , it appears to break down for brighter galaxies : in this regime , the nuclei not only show considerable scatter , but they lie systematically to the faint / red side of the extrapolated relation .
these red nuclei cause the histogram of nuclei colors to have secondary peak at @xmath191 1.5 ( see the right panel of figure [ fig19 ] ) .
they are found exclusively in high - surface - brightness environments , which raises the possibility of a bias in the measured colors .
however , the simulations described in appendix a in which artificial nuclei of known size , magnitude and color are added to the non - nucleated galaxy vcc1833 and their properties measured in the same way as the actual nuclei show no evidence for a significant color bias for such bright nuclei .
in addition , the colors for most of these nuclei are actually redder than the galaxies themselves , by @xmath75 0.1 mag , so it seems unlikely that contamination from the host galaxy can entirely explain their red colors . for comparison , the open stars in figure [ fig19 ] show the sample of virgo ucds from paper vii and haegan ( 2006 ) .
the agreement with the nuclei is striking : i.e. , with a mean color of @xmath192 = [email protected] mag , the ucds have colors that are virtually identical to those of comparably bright nuclei .
this constitutes yet another piece of evidence for a link between ucds and the nuclei of early - type galaxies . to better visualize how the properties of the host galaxy may affect
the relationship between nuclei color and magnitude , figure [ fig20 ] divides the sample by host galaxy magnitude into four subgroups .
these subsamples are shown in the four separate panels , with the color - magnitude relation given by equation ( 13 ) repeated in each case ( the dashed line ) .
also included in each panel are the globular clusters ( small points ) belonging to the galaxies in each of these magnitude intervals ; to reduce contamination from stars and compact galaxies , we plot only those sources with
globular cluster probabilities " ( see paper ix for details ) in the range @xmath153 .
note the clear bimodality in the colors of globular clusters belonging to these galaxies ( paper ix ) . with the exception of the very red nuclei noted above
, we conclude that the nuclei in our type ia galaxies have colors which fall within the range spanned by the bulk of the globular clusters in these same galaxies : @xmath193 mag . comparing the globular cluster colors to those of the ucds from figure [ fig19 ]
, we see that the ucds are @xmath1 0.10.2 mag redder than the population of blue globular clusters , but @xmath1 0.20.3 mag bluer than the red clusters .
this may be a point of distinction with the ucds in fornax , which mieske ( 2006 ) find to have colors similar to the red globular clusters .
figure [ fig21 ] shows how the colors of the galaxies , nuclei and globular clusters depend on the galaxy luminosity .
results are shown for the @xmath28 and @xmath2 bands in the left and right panels , respectively . a common distance of 16.5
mpc has been assumed for all galaxies ( mei 2005 ) .
galaxy colors are taken from paper vi and represent the average color in the range 1 @xmath116 , subject to the acs / wfc field view and excluding those regions with surface brightnesses 1 mag arcsec@xmath128 or more below the sky .
the majority of our galaxies show no evidence for strong color gradients , so these colors should accurately reflect their integrated colors . for the globular clusters , we plots colors for the red and blue subpopulations , as determined in paper ix , along with that of the composite cluster system . to highlight the subtle trends exhibited by these various samples , we show mean colors for the nuclei , globular clusters and galaxies in four broad bins of approximately equal width in galaxy magnitude ( @xmath75 2 mag ) .
results for the nuclei are shown for three bins containing an equal number of objects .
this figure reveals a correlation between nucleus color and galaxy luminosity that is broadly consistent with the finding of lotz ( 2004 ) for fainter galaxies .
however , the trend is relatively weak and is in fact due mainly to the @xmath75 one dozen galaxies noted above that have very red nuclei .
these galaxies make up most of the objects in the bins at @xmath194 , @xmath195 and @xmath196 , @xmath197 . in the fainter galaxies ,
the nuclei colors show essentially no correlation with galaxy luminosity . for galaxies fainter than @xmath198 , the nuclei have @xmath199
intermediate in color to the globular clusters and stars in galaxies in this luminosity regime .
in the preceding sections we have focussed on the observed properties of the nuclei found in our program galaxies .
we now turn to the broader question of what these observations may be telling us about the origin and evolution of galactic nuclei . before doing so
, we pause to briefly review some of the scenarios that have been proposed as possible explanations for stellar nuclei in early - type galaxies .
a more complete discussion of the theoretical implications of our findings will be given in merritt ( 2006 ) .
tremaine , ostriker & spitzer ( 1975 ) were the first to suggest that galactic nuclei may be the remains of merged globular clusters , which were driven inward to the galactic center by dynamical friction ( chandrasekhar 1943 ) . according to this
merger model " , the metallicity and luminosity of the nucleus should be a superposition of the metallicity and luminosity of the progenitor clusters . because dynamical friction causes the orbits of most massive globular clusters to decay most rapidly , a nucleated galaxy would be expect to show a selective depletion of _ bright _ globular clusters , at least in the inner regions where the dynamical friction timescale is short compared to a hubble time .
the contribution of globular clusters mergers to the growth of central black holes and galactic nuclei has been explored in a series of papers by capuzzo - dolcetta and coworkers ( e.g. , capuzzo - dolcetta 1993 ; capuzzo - dolcetta & tesseri 1999 ) .
recently , bekki ( 2004 ) have used numerical simulations to examine the physical properties ( e.g. , half - light radii , central velocity dispersion , mean density ) of nuclei that form in such mergers . motivated by the discovery that the de galaxies in virgo are less centrally concentrated than the de , n galaxies ( c.f .
4.4 ) , oh & lin ( 2000 ) revisited the question of how the tidal field from the virgo cluster affects the evolution of globular cluster orbits within de galaxies . they found that tidal perturbations acting on galaxies near the center of the cluster tend to be compressive , and have little net effect on the rate of decay of the globular cluster orbits . in the outer regions of the virgo cluster , tidal forces tend to disrupt galaxies , and
the resulting decrease in density leads to longer time scales for dynamical friction .
thus , tidal forces favor the formation of nuclei in galaxies which are located in the cluster core , and suppress the formation in more distant galaxies .
clearly , this will cause the relative number of nucleated and non - nucleated galaxies to vary within the cluster , with the highest fraction of nucleated galaxies in the core .
a second , broad category of models focuses on a dissipational origin for the nuclei . noting that galaxies with nuclei are typically rounder than those without , van den bergh ( 1986 )
speculated that nuclei form from the gas which collects more easily in the centers of slowly - rotating galaxies .
silk , wyse & shields ( 1987 ) argued that dwarf galaxies experience late accretion of cool gas from the intergalactic medium , leading to star formation and the growth of compact central nuclei . in a similar vein ,
davies & phillips ( 1988 ) proposed that early - type dwarfs result from the fading of stellar populations in dwarf irregular or blue compact dwarf galaxies . in this scenario ,
intermittent bursts of central star formation driven by the infalling gas continue until the gas reservoir is depleted . according to this scenario , the final star - forming event gives rise to the nucleation observed today .
babul & rees ( 1992 ) examined the impact of the local intergalactic medium on the evolution of a low - mass galaxy .
they argued that the pressure of the intergalactic medium acts as a confining agent : in a high - pressure environment , early - type dwarfs are able to retain more gas and produce a nucleus from the gas that has been prevented from escaping by the intergalactic medium . since the external pressure acting on galaxies decreases with increasing distance from the cluster center , some properties of the nuclei ( such as their luminosity and color ) should also depend on position within the cluster , with the highest frequency of nucleation in the central regions of a cluster .
gas inflow models have also been explored within the context of disk galaxy mergers .
mihos & hernquist ( 1994 ) used n - body / hydrodynamical simulations to show that such mergers are accompanied by gas dissipation and central star formation which may result in the formation of a dense stellar core , or the fueling of a pre - existing agn . following weedman ( 1983 ) , mihos & hernquist ( 1994 ) further note that the dense stellar core may itself collapse to form a supermassive black hole ( sbh ) .
the observational evidence for a possible link between such sbhs and the stellar nuclei of early - type galaxies is examined in [ sec : bh ] .
high - resolution _ hst _ imaging for sa - sd galaxies has shown that these objects frequently contain compact nuclear clusters near their photocenters ( e.g. phillips et al .
1996 ; carollo , stiavelli & mack 1998 ; matthews et al 1999 ; bker et al . 2002
; bker et al . 2004 ) .
figure [ fig22 ] compares the sample of nuclear clusters from bker ( 2004 ) to our sample of early - type nuclei . in the upper panel ,
we plot the physical sizes for both samples , where we have assumed a common distance of 16.5 mpc for the virgo galaxies ( tonry 2001 ; paper v ) .
it is clear that the nuclear clusters of bker ( 2004 ) have sizes similar to the early - type nuclei .
the lower panel of figure [ fig22 ] compares the absolute magnitudes of the two samples .
note that the observations of bker ( 2004 ) were carried out in the f814w ( @xmath59 ) bandpass . comparing the means of the samples , we find the two populations to be comparably bright , with @xmath200 and @xmath201 for the early - type nuclei , and @xmath202 for the nuclear clusters .
bker ( 2002 ) further report that 59 of 77 late - type spirals in their survey contain a nuclear cluster close to the galaxy photocenter , giving an overall frequency of nucleation of @xmath203 . for comparison ,
in 4.2 we estimated @xmath204 for our sample of early - type galaxies , counting galaxies with possible offset nuclei as non - nucleated .
thus , in all these respects , the nuclear clusters found in late - type galaxies are nearly identical to the nuclei studied here .
the lone point of distinction between the nuclear clusters and the early - type nuclei seems to be one of age : the majority of the nuclear clusters appear to have @xmath205 yr ( walcher 2005 and references therein ) , while the broadband colors rule out such young ages for _ all _ of the type ia nuclei , irrespective of metallicity ( see [ sec : stellpop ] and figure 6 of paper i ) .
this difference notwithstanding , the similar properties of the nuclei and nuclear clusters and their appearance in galaxies of such disparate morphology clearly points to a rather generic formation mechanism : e.g. , one which is largely independent of local or global environmental factors , such as the gas content and present - day morphology of the host galaxy , or the density of neighboring galaxies .
the above conclusion applies equally well to the luminosity of the host galaxy : i.e. , the nuclei are not confined to just the dwarfs , but are also found with regularity in many of the giants .
in fact , half ( 21/42 ) of the galaxies brighter than @xmath206 or @xmath10 ( the approximate division between dwarfs and giants in the vcc ) have classifications of type ia or ib . the fact that nuclei are common above and below the canonical dwarf - giant boundary suggests that , at least in terms of their _ nuclear _ properties , there is no evidence for a fundamental change in galaxies at this magnitude .
this is consistent with the mounting evidence from photometric scaling relations that the dichotomy " between normal and dwarf ellipticals , as originally envisioned by kormendy ( 1985 ) and others , may not be real ( e.g. , jerjen & binggeli 1997 ; graham & guzmn 2003 ; paper vi ) . on the other hand
, there _ does _ appear to be a fundamental transition at @xmath15 in terms of nuclear properties .
brighter than this , we measure @xmath207 and , in almost all cases , the presence of a nucleus can be ruled out with some confidence ( see appendix a ) .
fainter than @xmath15 , the fraction of nucleated galaxies rises sharply , as shown in the lower panel of figure [ fig06 ] .
it has been argued ( graham & guzmn 2003 ; trujillo 2004 ; graham 2004 ; paper vi ) that this magnitude marks a transition from faint , srsic - law galaxies to bright , core - srsic - law galaxies , whose flat cores " are presumed to result from core depletion by coalescing of supermassive black holes ( ebisuzaki 1991 ; quinlan & hernquist 1997 ; faber 1997 ; milosavljevi & merritt 2001 ) .
the absence of nuclei in galaxies brighter than @xmath15 is consistent with this scenario . of course
, it is equally possible that these missing " nuclei are absent in the bright galaxies not because of the disruptive effects of mergers and black hole coalescence , but because they failed to form in such galaxies in the first place .
discriminating between these competing scenarios should prove to be a fruitful area for future theoretical study .
in almost all cases , the nuclei are found to be coincident with the photocenters of their host galaxies . in only five cases
does there appear to be a statistically significant offset of @xmath208 .
the bulk of the evidence , however , favors the view that these nuclei " are , in actuality , star clusters projected close to the galaxy photocenters . that is to say , the sizes , surface brightnesses and colors of the five possible offset nuclei more closely resemble those of globular clusters than those of the other nuclei in our sample .
interestingly , all five of the host galaxies show some characteristics of dirr / de transition objects , including blue colors , low central surface brightnesses , the presence of dust , and a mottled or irregular appearance .
this suggests that if dwarf ellipticals represent an evolutionary stage that follows gas exhaustion and stellar fading ( davies & phillips 1988 ) , ram pressure stripping ( mori & burkert 2000 ) or harrassment ( moore , lake & katz 1998 ) of gas - rich dirr / disk galaxies , then the formation of a central nucleus is not an immediate or inevitable consequence .
additional time seems to be required to grow " a central nucleus . because the theoretical framework for the globular cluster merger model is at a more mature stage than for any other model ,
we now turn our attention to the question of whether this scenario is consistent with our new observations for the nuclei .
we note that lotz ( 2001 ) have previously examined the viability of the merger hypothesis on the basis of data collected for nuclei and globular clusters in their wfpc2 snaphot survey of dwarf galaxies . apart from identifying a possible depletion of bright clusters in the innermost regions of the galaxies
, they could find no strong evidence for a merger origin of the nuclei , either from the spatial distribution of the clusters or from the measured luminosities of the nuclei .
as pointed out in 4.6 , a comparison of the luminosity functions of nuclei and globular clusters in these galaxies shows that the _ typical _ nucleus is @xmath1 3.5 magnitudes brighter than a typical globular cluster .
if cluster mergers are responsible for the formation of a central nucleus , then one might expect an average of @xmath75 25 mergers would be needed to assemble a nucleus from typical clusters . of course
, as figure [ fig20 ] shows , a single number does not tell the whole story .
the four panels of this figure plot the distribution of nuclei and globular clusters in the color - magnitude diagram . for the brightest galaxies ( shown in the first panel ) ,
the nuclei have a median luminosity @xmath1 125@xmath78 that of globular clusters at the peak of the cluster luminosity function . for the fainter galaxies ( shown in the three remaining panels ) ,
the nuclei are brighter than a typical globular clusters by factors of 29 , 15 and 17 , respectively . are these numbers feasible ? in figure [ fig23 ] we attempt to answer this question by plotting the integrated luminosity in globular clusters against the luminosity of the nucleus for type ia galaxies in our survey .
results are shown in the upper panels , with measurements made in the @xmath28 and @xmath2 bands given in the left and right panels , respectively .
as in figures [ fig19 ] and [ fig20 ] , symbol type indicates the magnitude of the host galaxy . in calculating the total luminosity in globular clusters for these galaxies
, we have simply summed the luminosities of globular cluster candidates with probabilities in the range @xmath209 .
although this approach will obviously miss any globular clusters located outside the acs field , the correction should be small for the type ia galaxies in our survey which , with @xmath210 , have @xmath211 15 or less ( paper vi ) . the correlations apparent in these panels are a consequence of the fact that both the total number of globular clusters , and the luminosity of the nucleus , scale with host galaxy luminosity .
the lower panels of figure [ fig23 ] plot the ratio of globular - to - nucleus luminosities , @xmath212 , in the two bandpasses . in both cases ,
the ratio is near unity .
this should perhaps come as no surprise since the mean nucleus - to - galaxy luminosity ratio , @xmath213% , found in 4.5 is nearly identical to the globular cluster formation efficiency of @xmath214% measured by mclaughlin ( 1999 ) for early - type galaxies .
this latter measurement is in turn based on observations of 97 early - type galaxies and represents the total mass in globular clusters normalized to the total baryonic ( stellar + gas ) mass . while the agreement may be purely coincidental , it is a remarkable empirical result that the formation of early - type galaxies results in a nearly constant fraction of the initial baryonic mass , @xmath75 0.3% , being deposited into both globular clusters and , in many cases , a central nucleus .
of course , this conclusion appears _ not _ to apply to galaxies brighter than @xmath15 , which lack nuclei either because they did not form in the first place or because they were subsequently destroyed by some as - yet - unidentified process . in any case , galaxies which lie below the dashed line at @xmath215 in figure [ fig23 ] pose a clear challenge to the merger model for the obvious reason that they simply have too few clusters to explain the luminosity of the nucleus .
the difficulty is most severe for the dozen or so red nuclei associated with the brightest type ia galaxies .
of course , this argument is based on the number of clusters contained by the host galaxy _ at the present time_. if the observed clusters are the rare survivors " descended from a much larger initial cluster population , then it may be possible to circumvent this problem .
an additional test of the merger model is possible .
if the mergers were dissipationless so that star formation and chemical enrichment can be ignored , then we can use the observed colors of globular clusters to predict colors for the nuclei . since both globular cluster color and the fraction of red globular clusters are increasing functions of host galaxy luminosity ( paper ix ) , we expect the nuclei in this model to become progressively redder in brighter galaxies .
the heavy solid curve shown in the four panels of figure [ fig20 ] shows the predicted color magnitude relation for nuclei which grow through globular cluster mergers .
this curve is based on monte carlo experiments in which the color evolution of the nuclei is followed using the observed colors of the globular clusters in these galaxies .
the thin curves show the 95% confidence limits on the relation .
although these simulations do indeed predict redder colors for the brighter nuclei ( which are found preferentially in the brighter galaxies containing a larger proportion of red clusters ) , the predicted scaling is much milder than what is observed .
bekki ( 2004 ) have used numerical simulations to investigate the physical properties of nuclei which form through repeated mergers of globular clusters .
their predicted scaling relation between half - light radius and luminosity , @xmath216 , is shown by the dotted line in figure [ fig17 ] .
the relation has a somewhat milder luminosity dependence than the observed relation , @xmath9 , but is nevertheless in reasonable agreement .
a similar conclusion applies to the luminosity dependence of surface brightness averaged within the half - light radius .
we find no strong correlation between @xmath217 and @xmath218 , but the predicted relation , @xmath219 ( shown as the dotted line in figure [ fig18 ] ) , is reasonably consistent with the observations , having only a weak luminosity dependence . in the future , it will be of interest to compare the predicted and observed relationship between luminosity and central velocity dispersion .
spectra for most of our program galaxies are now in hand and such a comparison will be the subject of a future paper in this series .
ground - based spectroscopy will also be useful for investigating the history of star formation and chemical enrichment in the nuclei , although care must be exercised when decoupling the contributions from the galaxy and nucleus .
this separation is more straightforward in the _ acs _ imaging , although in this case we are limited in our ability to measure ages and metallicities because of the well known age - metallicity degeneracy of broadband colors .
the upper panel of figure [ fig24 ] shows linear interpolations of the [ fe / h]-@xmath220 relation for simple stellar populations from the models of bruzual & charlot ( 2003 ) .
the four relations show color - metallicity relation for ages of @xmath221 = 1 , 2 , 5 and 10 gyr , although it is , needless to say , quite unlikely that a single age is appropriate for all of the nuclei in our sample . for comparison ,
the heavy dashed curve in black shows the color - metallicity relation derived from globular clusters in the milky way , m49 and m87 ( paper ix ) .
if it is assumed that the nuclei have ages similar to the globular clusters , then this empirical relation may be used to derive metallicities for the nuclei .
converting from colors to metallicities with these relations , we find the five metallicity distributions shown in the lower panel of figure [ fig24 ] .
the results are summarized in table [ tab : met ] . not suprisingly , the metallicity distributions derived from the models depend sensitively on the assumed age . for @xmath221
= 10 gyr , the colors of the bluest nuclei , with @xmath222 mag , would require very low metallicities : i.e. , [ fe / h ] @xmath223 or less .
by the same token , the reddest nuclei in our sample would require metallicities 1 - 100@xmath78 solar for an assumed age of 1 gyr .
for ages as young as @xmath205 yr , which is appropriate for many of the nuclear clusters in late - type galaxies ( see [ sec : boker ] ) no reasonable choice of metallicity can explain the colors of @xmath1 0.81.5 that are observed .
thus , to the extent that the nuclei can be characterized by a single formation epoch , they show evidence for an old to intermediate age : i.e. , @xmath224 gyr . using the globular cluster color metallicity relation
gives a mean metallicity of @xmath225}\rangle$ ] = @xmath226 dex .
firmer conclusions on the ages and metallicities of the nuclei must await the spectroscopic analysis .
spectroscopic constraints on the mix of stellar populations in the nuclei should also shed some light on what may be the most serious challenge facing the merger model : the existence of a tight correlation between nucleus luminosity and color ( figures [ fig19]-[fig20 ] ) .
such a correlation is generally thought to be a signature of self enrichment in stellar systems , and is reminiscent of the color - magnitude relation for dwarf and giant galaxies ( e.g. , bower , lucey & ellis 1992 ; caldwell 1992 ) .
that the colors correlate tightly with the luminosities of the nuclei , and less so with those of the host galaxies , suggests that the chemical enrichment process was governed primarily by local / internal factors .
the existence of a tight color - magnitude relation for the nuclei is a difficulty for the merger model as envisioned by tremaine ( 1975 ) since the clusters from which the nuclei are assembled show no color - magnitude relation themselves , and our monte - carlo experiments reveal the slope of the observed color - magnitude relation is steeper than that predicted in dissipationless cluster mergers .
we speculate that the merger model in its original form ( i.e. , dry " mergers of stellar clusters ) is an oversimplication of a process that almost certainly involves some gas dissipation .
in fact , if nuclei do indeed have stellar ages of a few gyr old or more , then they were assembled during the earliest , most gas - rich stage of galaxy evolution .
it would be interesting to revisit the merger model with the benefit of numerical simulations that include the effects of not just dark matter and stars , but also gas , to see if star formation and chemical enrichment caused by mergers / inflows are capable of producing a color - magnitude relation consistent with that shown in figure [ fig19 ] .
in a number of respects , the dozen or so bright nuclei labelled in figure [ fig19 ] appear to form a population distinct from their faint counterparts , most notably in their integrated colors ( which appear redder than the galaxies themselves ) .
these nuclei may be candidates for the dense stellar cores " which form in numerical simulations ( mihos & hernquist 1994 ) when ( chemically - enriched ) gas is driven inward , perhaps as a result of mergers .
our acs observations may also provide some clues to the origin of ucd galaxies . in terms of color , luminosity and size , the ucds from paper vii bear a strong resemblance to many of the nuclei studied here , leading credence to the galaxy threshing scenario ( bassino 1994 ; bekki 2001 ) .
however , these ucds were selected for study on the basis of luminosity and half - light radius , so it is unclear to what extent these conclusions may apply to the population of ucds as a whole .
an unbiased survey of the ucd population in virgo should be undertaken to clarify this issue , although this will be a difficult and time - consuming task as it requires high - resolution spectroscopy and hst imaging to distinguish true ucds from bright globular clusters ( see 7 of paper vii ) .
jones ( 2006 ) have recently reported the first results from a program to search for ucds in virgo using radial velocities for @xmath75 1300 faint , star - like sources in the direction of the cluster . however , lacking structural and internal dynamical information for ucd candidates found in this way , it is impossible to know to what extent their sample has been polluted " by globular clusters : either those associated with galaxies or intergalactic in nature ( e.g. , west 1995 ) . for the time being , we may use the existing sample of virgo ucds from paper vii and haegan ( 2006 ) to re - examine the threshing model in light of our findings for the virgo nuclei . specifically , we may estimate the luminosities of the ucd progenitor galaxies within the context of the threshing model : for the typical ratio of nucleus - to - galaxy luminosity found in 4.5 , @xmath227% , we expect the progenitors to have @xmath228 , with a value mean of @xmath229 .
if the threshing scenario is correct , then we might expect the surviving analogs of the ucd progenitors to resemble galaxies # 4069 in table [ tab : data ] .
it is interesting to note that only about half ( 16/30 ) of these galaxies were classified as dwarfs by bst85 , meaning that a significant fraction of _ bright _ galaxies may need to have been threshed " to explain the ucd luminosity function within the context of this model .
photometric , dynamical and structural parameters for these candidate ucd progenitor galaxies may serve as useful constraints for self - consistent numerical simulations of galaxy threshing and ucd formation .
a large body of literature now exists on the sbhs that reside in the centers of many galaxies ( see , e.g. , the review of ferrarese & ford 2005 ) . while it had been known for some time that sbh masses , @xmath230 , correlate with the bulge masses , @xmath231 , of their host galaxies ( kormendy & richstone 1995 ) , it was only after the discovery of a tight relation between @xmath230 and bulge velocity dispersion ( ferrarese & merritt 2001 ; gebhardt 2001 ) that merritt & ferrarese ( 2001 ) were able to show that the frequency function for galaxies with sbhs has a roughly normal distribution in @xmath232 .
fitting to the data available at that time , merritt & ferrarese ( 2001 ) found a mean of value of @xmath233 ( 0.13% ) and a standard deviation of 0.45 dex .
remarkably , this mean value is , to within a factor of @xmath1 two , identical to the mean fractional luminosity contributed by nuclei to their host galaxies ( [ sec : lum ] ) .
in fact , the nuclei and sbhs share a number of other key similarities that are highly suggestive of a direct connection : e.g. , they share a common location at the bottom of the gravitational potential wells defined by their parent galaxies and dark matter halos , and both are probably old components that formed during the earliest stages of galaxy evolution ( [ sec : stellpop ] ; haehnelt 1998 ; silk & rees 1998 ; wyithe & loeb 2002 ) . could it be that the compact nuclei which are found so frequently in the low- and intermediate - luminosity early - type galaxies are related in some way to sbhs dectected in the brighter galaxies ?
figure [ fig25 ] examines the connection between nuclei and sbhs in more detail . in the upper panel ,
we show the distribution of absolute blue magnitudes , @xmath234 , for the 51 galaxies in our survey that contain type ia nuclei ( solid histogram ) . this distribution should be compared to that for the early - type galaxies having sbh detections ( open squares and dotted histogram ) .
this latter sample is based on data from table ii of ferrarese & ford ( 2005 ) , which reports sbh mass measurements from resolved dynamical studies for 30 galaxies . among this sample ,
there are 23 early - type galaxies with measured sbh masses ( all based on stellar and/or gas dynamical methods ) .
it is clear from figure [ fig25 ] that the two samples have very different distributions . with the exception of m32 ( with @xmath235 mag and @xmath236 solar masses ) , the sbh galaxies are all brighter than @xmath237 , a cutoff that is thought to reflect the formidable technical challenges involved in detecting smaller sbhs in fainter early - type galaxies .
by contrast , the type ia galaxies have @xmath238 mag . note that this does _ not _ reflect the true upper boundary for nucleated galaxies , since nuclei definitely exist in galaxies brighter than this table [ tab : data ] lists 14 galaxies with certain or suspected nuclei ( i.e. , types ib , ic or i d ) having @xmath239 mag but the high surface brightness of the host galaxies do not allow a reliable measurement of the nuclei luminosities or sizes . as we have argued in [ sec : fund ] , the more fundamental cutoff seems to occur at @xmath6 mag since we find no nucleated galaxies brighter than this . before moving on , we note that four of the galaxies with sbh masses in table ii of ferrarese & ford ( 2005 ) appear in our survey . in two cases
vcc1978 ( ngc4649 ) and vcc1231 ( ngc4473 ) there is no evidence for a nucleus so we classify the galaxies as type ii . in a third case , vcc763 ( ngc4374 )
, the center of the galaxy is partly obscured by an agn ( type o ) but we see no evidence of a resolved stellar nucleus ( 4 ) .
the fourth and final galaxy , vcc1664 ( ngc4564 ) , has a reported sbh mass of 5.6@xmath240 ( gebhardt 2003 ) .
we classify this object as type ic , meaning that we see evidence for a resolved nucleus but are unable to measure its properties owing to the high surface of the galaxy . in the lower panel of figure [ fig25 ]
, we compare the frequency functions of sbhs and type ia nuclei .
bulge masses for the sbh galaxies were found by assuming a constant bulge color of @xmath241 mag and combining the magnitudes from ferrarese & ford ( 2005 ) with the mass - to - light ratio relation @xmath242 from paper vii .
for the sbh sample , we find @xmath243 whereas for the nuclei , we find @xmath244 for comparison , gaussian distributions with these parameters are shown in the lower panel of figure [ fig25 ] . in our view
, the available evidence favors the view that the compact stellar nuclei found in many of the fainter galaxies may indeed be the counterparts of sbhs in the brighter galaxies .
if this speculation is correct , then it might be more appropriate to think in terms of _ central massive objects _ either sbhs or compact stellar nuclei that accompany the formation of almost all early - type galaxies and contain a mean fraction @xmath1 0.3% of the total bulge mass .
we note that a similar conclusion has been reached independently by wehner & harris ( 2006 ) .
models for the formation of sbhs in massive galaxies would then be confronted with the additional challenge of explaining the different manifestations of central massive object formation along the galaxy luminosity function , with an apparent preference for sbh formation above @xmath245 mag .
these issues are explored in more detail in ferrarese ( 2006b ) .
using multi - color acs imaging from the _ hubble space telescope _ , we have carefully examined the central structure of the 100 early - type galaxies which make up the acs virgo cluster survey . in this paper , we have focussed on the compact nuclei which are commonly found at , or close to , the photocenters of many of the galaxies .
our main conclusions are as follows : * nuclei in early - type galaxies are more common than previously believed . excluding the six galaxies for which the presence of a nucleus could not be established , either because of dust obscuration or the presence of an agn , and counting the five galaxies with possible offset nuclei as non - nucleated
, we find the frequency of nucleation to fall in the range @xmath246% for early - type galaxies brighter than @xmath3 .
* nuclei are found in galaxies both above and below the canonical dividing line for dwarfs and giants ( @xmath10 ) .
half ( 21/42 ) of the galaxies brighter than @xmath10 are found to contain nuclei .
this is almost certainly a lower limit on the true frequency of nucleation because of incompleteness and surface brightness selection effects in these bright galaxies .
* on the other hand , galaxies brighter than @xmath15 appear to be distinct in that they do _ not _ contain nuclei at least , not those of the kind expected from upward extrapolations of the scaling relations obeyed by nuclei in fainter galaxies . whether this means
that nuclei never formed in these core - srsic " galaxies ( see paper vi and references therein ) , or were instead subsequently destroyed by violent mergers and core evacuation due to coalescing supermassive black holes , is unclear .
the absence of nuclei in galaxies brighter than @xmath15 adds to the mounting evidence for a fundamental transition in the structural properties of early - type galaxies at this magnitude . *
few , if any , of the nuclei are found to be significantly offset from the photocenters of their host galaxies . in only five galaxies do we measure offsets @xmath114 or @xmath247 . in all fives cases
, however , the available evidence ( i.e. , magnitudes , colors , half - light radii and surface brightness measurements ) suggests that such
nuclei " are probably star clusters projected close to the galaxy photocenters .
* the central nuclei are _ resolved _ , with a few notable exceptions : such as the two agn galaxies , m87 and m84 , which have prominent non - thermal nuclei , and a half dozen of the faintest galaxies with very compact , but presumably stellar , nuclei .
this observation rules out low - level agn as an explanation for the central luminosity excess in the vast majority of the galaxies . excluding those galaxies with faint , unresolved nuclei
, we find the half - light radii of the nuclei to vary with luminosity according to the relation @xmath9 . *
a gaussian distribution provides an adequate , though by no means unique , description of the luminosity function for the nuclei .
the peak of the luminosity function occurs at @xmath248 and @xmath249 , or roughly 25@xmath78 brighter than the peak of the globular cluster luminosity functions in these same galaxies .
we find the ratio of nucleus - to - galaxy luminosities to be @xmath250 0.3% , independent of galaxy luminosity but with significant scatter . *
a comparison of the nuclei to the nuclear star clusters in late - type galaxies shows a remarkable similarity in luminosity , size and overall frequency .
this suggests that a quite generic formation mechanism is required to explain the nuclei : one which is largely independent of local or global environmental factors , such as the gas content and present - day morphology of the host galaxy , or the density of neighboring galaxies . * in terms of color , luminosity and size , the ucds from paper vii bear a strong resemblance to the compact nuclei in a number of these galaxies , leading credence to the threshing " scenario in which ucds are assumed to be the tidally stripped remains of nucleated galaxies .
if this model is correct , then we argue that the ucd progenitor galaxies would if they avoided threshing " now resemble galaxies with magnitudes in the range @xmath228 .
simulations to test the viability of the threshing mechanism for such galaxies are advisable . * the colors of the nuclei are tighly correlated with their total luminosities , but only weakly with those of their host galaxies .
this may suggest that the history of chemical enrichment in the nuclei was governed by local or internal factors . *
the mean of the frequency function for the nucleus - to - galaxy luminosity ratio in our nucleated galaxies , @xmath11 dex ( @xmath12 ) , is indistinguishable from that of the sbh - to - bulge mass ratio , @xmath13 dex ( @xmath14 ) , calculated in 23 early - type galaxies with detected sbhs .
* we argue that the compact stellar nuclei found in many of our program galaxies may be the low - mass counterparts of sbhs detected in the bright galaxies , a conclusion reached independently by wehner & harris ( 2006 ) .
if this view is correct , then one should think in terms of _ central massive objects ( cmos ) _ either sbhs or compact stellar nuclei that accompany the formation of almost all early - type galaxies and contain a mean fraction @xmath1 0.3% of the total bulge mass .
in this view , sbhs would be the dominant mode of cmo formation above @xmath15 . as the nearest large collection of early - type galaxies , the virgo cluster is likely to remain , for the forseeable future , at the center of efforts to understand the physical processes that drive nucleus formation .
unfortunately , exploring the stellar dynamics of the most compact nuclei and modeling the mass distribution within the central few parsecs of the host galaxies requires integrated - light spectra with an angular resolution of @xmath75 01 or better .
thus , the virgo nuclei are obvious targets for diffraction - limited , near - ir spectrographs on 8m - class ground - based telescopes , particularly since the demise of the _ space telescope imaging spectrograph _ on _
hst_. for the time being , though , acs imaging of the nuclei should serve to guide models of their formation and evolution
. this will be the subject of a future paper in this series , in which we will examine the implications of these observations for theories of nucleus formation ( merritt 2006 ) .
we thank peter stetson for assistance with the construction of the psfs used in this study .
support for program go-9401 was provided through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
p.c . acknowledges additional support provided by nasa ltsa grant nag5 - 11714 .
. acknowledges additional financial support provided by the sherman m. fairchild foundation .
is supported by nsf grant ast-020631 , nasa grant nag5 - 9046 , and grant hst - ar-09519.01-a from stsci .
m.j.w . acknowledges support through nsf grant ast-0205960 .
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 .
the approach used to classify galaxies according to the presence or absence of a central nucleus has been described in 4 .
briefly , the classification procedure relies on both a visual inspection of the acs images and the detection of a central excess " in the brightness profile relative to the fitted srsic or core - srsic galaxy model .
the results are summarized in table [ tab : class ] .
we find a total of 62 galaxies in which the presence of a nucleus could be established with a high level of confidence ( i.e. , the type ia and ib galaxies ) .
five more galaxies ( type ie ) _ may _ contain an offset nucleus but , as we have argued above , the weight of evidence favors the view that these nuclei " are actually globular clusters .
six other galaxies ( type 0 ) contain either an agn or dust at the photocenters , making the identification of a nucleus difficult or impossible .
this leaves us with a sample of 100 62 5 6 = 27 galaxies which may be classified provisionally as non - nucleated .
of course , the faintest , most extended nuclei will go undetected in any survey , especially when superimposed on a bright galaxy background .
it therefore seems likely that at least some of these galaxies may , in fact , be nucleated . in this appendix
, we attempt to elucidate the nature of these galaxies with the aid of numerical simulations guided by our findings from 4 . for the 27 galaxies in question , figure [ fig26 ] plots residuals , over the
innermost 10 , between the _ observed _ brightness profile and the fitted models shown in figure [ fig04 ] .
because a few of these galaxies contain multiple components ( e.g. , rings , bars or shells ) , or have outer brightness profiles that are contaminated by the light of nearby giant galaxies , the profiles were sometimes fit over a restricted range in radius . in two cases where this outer fitting radius is @xmath251 ( vcc778 = ngc4377 and vcc575 = ngc4318 ) , an upward arrow shows the adopted limit . likewise , six galaxies ( vcc1664 = ngc4564 , vcc944 = ngc4417 , vcc1279 = ngc4478 , vcc355 = ngc4262 , vcc1025 = ngc4434 and vcc575 ) in which the presence of a faint central nucleus was suspected on the basis of an upturn in the central brightness profile , an upward arrow at 0203 shows the inner limit used to avoid biasing the fitted galaxy parameters . note that in most cases , the fitted srsic or core - srsic model provides a reasonably accurate match to the central brightness profile , meaning that any nuclei which may be hiding in these galaxies have had only a minor impact on the observed profiles
of course , two possibilities exist for any given galaxy : either it contains a nucleus or it does not . to test the first possibility , we use the scaling relation from 4 which links the luminosity of the galaxy to that of its nucleus ( equation 6 ) . meanwhile , equation 11 provides a link between galaxy luminosity and nucleus size ( half - light radius ) . for each of the 27 galaxies in figure [ fig26 ]
, we then subtract a nucleus of the expected size and luminosity based on the magnitude of the galaxy itself .
if the nucleus - subtracted profile of the galaxy is better represented by a sersic or core - srsic model than the original profile , this would be ( circumstantial ) evidence for the presence of a faint nucleus .
the alternative possibility is that the galaxy is truly non - nucleated . in this case
, we can _ add _ a simulated nucleus of the appropriate size and magnitude to see if it would be detectable from the images and/or the surface brightness profile .
taken together , these two experiments allow us to crudely estimate the overall completeness of our survey and to refine the provisional classifications for these 27 galaxies .
we caution , however , that the approach of adding and subtracting nuclei not only assumes that the scaling relations found in 4 are valid for all galaxies in the survey , but it ignores the significant scatter about the fitted relations . with these caveats in mind
, we present the results of this exercise in figure [ fig26 ] . in each panel ,
the small blue squares show the residuals profile obtained after adding a simulated nucleus to the image and recalculating the brightness profile .
small red squares show the profile obtained if this nucleus is instead subtracted . for four galaxies ( vcc1692 , 1664 , 944 and 1025 )
, the best - fit srsic / core - srsic model for the subtracted profile provides some improvement over the original fit .
we therefore classify these four as galaxies as type ic systems . at the same time
, we identify 12 other galaxies which can be classified unambiguously as non - nucleated ( type ii ) . in these galaxies ,
the subtracted brightness profiles show strong inward gradients in their central regions : an obviously non - physical result .
interestingly , these 12 galaxies fall into two rather distinct categories : ( 1 ) _ bright giants _ which have shallow cores " in the central few arcseconds and thus are best fit with core - srsic models ; and ( 2 ) _ faint dwarfs _ which are best fit with pure srsic models . the common feature linking these two populations is the presence of a low - surface brightness core that facilitates the detection of a central nucleus .
for this reason , we can say with some confidence that these 12 galaxies do _ not _ contain nuclei that follow the scaling relations observed in 4 for the type ia galaxies .
the final 11 galaxies remain elusive since we can neither confirm nor rule out the presence of a nucleus in these cases .
we classify these objects as possibly nucleated ( type i d ) .
figures [ fig06]-[fig07 ] clearly demonstrate that it is possible to detect nuclei in galaxies that span a wide range in luminosity and central surface brightness .
but to what extent is our ability to detect nuclei and to measure their sizes and magnitudes affected by the surface brightness of the underlying galaxy and their own luminosity or size ? needless to say ,
a complete characterization of the biases and incompleteness affecting the nuclei requires _ a priori _ knowledge of their intrinsic properties : information that we are obviously lacking .
nevertheless , we may take a first step towards answering these questions by adding simulated nuclei of known size and magnitude to the center of a non - nucleated galaxy .
for these experiments , we focus on a single non - nucleated galaxy , vcc1833 , which , as a srsic - law galaxy with a central surface brightness of @xmath252 and @xmath253 mag arcsec@xmath128 , is representative of the type ia galaxies in our survey .
nuclei that span a range in both magnitude and size were added to the galaxy photocenter .
input magnitudes covered the intervals @xmath254 and @xmath255 in 1 mag increments ; at each magnitude , nuclei were added with half - light radii of 000 , 002 , 003 , 004 , 01 , 005 , 01 , 015 and 02 .
simulations were carried out independently for the f475w and f850lp images , and for each simulation , the surface brightness profile was measured from the artificial image using the same procedure as for the real galaxy .
a nucleated srsic model was then fitted to the profile of the simulated galaxy+nucleus and the best - fit parameters for the nucleus recorded .
the results of these simulations are shown in figure [ fig27 ] .
the upper panel of this figure shows the difference between the recovered and input half - light radius , @xmath256 , as a function of input radius .
the lower panel plots the difference between the recovered and input magnitude , @xmath257 , as a function of input magnitude . in both panels ,
results are shown for the separate f475w and f850lp bandpasses ( blue and red squares , respectively ) .
the symbol size is proportional to either input magnitude ( as in the upper panel , where larger symbols correspond to brighter nuclei ) or input radius ( as in the lower panel , where larger symbols correspond to more compact nuclei ) .
there are several conclusions which may be drawn from this figure , although sweeping claims must be avoided because the results of the simulations will almost certainly depend on the central surface brightness of the galaxy , the steepness of its brightness profile , etc , so the findings are not generalizable to the other program galaxies in any straightforward way . with these caveats in mind ,
we note that nuclei brighter than @xmath258 mag or @xmath259 mag in this particular galaxy would be detected for any choice of @xmath60 in the range 0 - 02 .
conversely , nuclei fainter than @xmath260 mag or @xmath261 mag would never be detected .
there appears to be no serious bias affecting the @xmath60 measurements for nuclei with @xmath262 , at least for nuclei brighter than @xmath263 2021 mag . in this size regime
a range which encompasses half of the nuclei in table [ tab : data ] the input half - light radii are recovered to a precision of @xmath75 15% or better . for larger nuclei , with @xmath264 , there is a bias which ranges from @xmath265 10% for the brightest nuclei , to nearly a factor of two for the faintest detectable nuclei , in the sense that the recovered nuclei are smaller .
unfortunately , the intrinsic distribution of nuclei sizes is unknown , so it is not possible to apply an _ a posteriori _ correction to the measured sizes . in any case
, we note that the result from [ sec : results ] that would be most directly affected by a bias of this sort is the observed scaling between nucleus and luminosity ( figure [ fig17 ] ) , where it was found @xmath266 with @xmath267 . if we make the admittedly dubious assumption that the luminosity dependence of the bias found in the case of vcc1833 is representative of the full sample of type ia galaxies , then we would expect the exponent in equation 11 to fall to @xmath268 . for the faintest nuclei ,
the simulations reveal that completeness is a function of surface brightness , in the expected sense that , at fixed luminosity , more compact nuclei ( i.e. , higher surface brightness ) are more likely to be detected .
as the lower panel of figure [ fig27 ] shows , there is also a tendency to measure fainter magnitudes for the simulated nuclei , at least in this galaxy .
not surprisingly , the importance of this bias depends sensitively on the input magnitude ; for the brightest nuclei , the bias is less than @xmath269 mag in both filters , irrespective of the input @xmath60 . for the fainter nuclei
, the bias can be as large as @xmath75 0.5 mag , and is slightly worse in the f850lp filter .
to the extent that the simulations for vcc1833 are applicable to other galaxies in the survey , this means that the faintest type ia nuclei may have measured colors that are systematically too blue by @xmath75 0.1 mag .
two recent papers , de propris et al .
( 2005 ) and strader ( 2006 ) , have presented magnitudes , colors and half - light radii for the nuclei belonging to a subset of our program galaxies . since the same observational material that forms the basis of our analysis was used in each of these studies
, it is of interest to compare the various measurements .
based on the vcc classifications that were available when the acs virgo cluster survey was begun , 25 of the 100 program galaxies were thought to contain nuclei ( see table 1 of paper i ) . as we have shown in
4 , the actual number of nucleated galaxies in our survey is about three times larger than this , although in a number of cases the nuclei were too faint to determine reliable photometric or structural parameters ; in the final analysis , magnitudes , colors and sizes could be measured for 51 ( type ia ) nuclei .
we first consider the results of de propris et al .
( 2005 ) , who studied 18 of the 25 galaxies originally classified as nucleated dwarfs .
these authors parameterized the underlying galaxies as srsic models . after subtracting this srsic component ,
colors and magnitudes for the nuclei were determined by summing the light within a 1 aperture , while half - light radii for the nuclei were measured with the ishape software package ( larsen 1999 ) for a circular plummer profile and tiny tim psf .
we have transformed the de propris ( 2005 ) vegamag photometry onto the ab system using the zeropoints given in table 11 of sirianni ( 2005 ) .
their half - light radii were converted from parsecs to arcseconds using their adopted virgo distance of 15.3 mpc .
extinction corrections , which in both studies are based on the dirbe maps of schlegel ( 1998 ) , were applied to our photometry as described in paper ii .
the two upper panels of figure [ fig28 ] compare our magnitudes with those of de propris et al .
( 2005 ) ( filled circles ) , where the dashed lines show the one - to - one relations .
there is only fair agreement between the measured magnitudes ( the @xmath84 scatter is @xmath1 0.30 mag in both bands ) . in the lower left panel of figure [ fig28 ]
, we compare our two estimates for the nuclei colors with those of de propris et al .
( 2005 ) . whether one uses integrated or aperture colors , the agreement is fair at best ( @xmath84 scatter @xmath1 0.17 mag in either case ) .
as discussed in 4.1 , an internal comparison of our color measurements shows a an @xmath84 scatter of 0.059 mag between the integrated and aperture colors . in any case , the scatter in the comparison with the de propris et al .
( 2005 ) colors is largely driven by three galaxies vcc200 , vcc1826 and vcc2050 ( ic3779 ) which de propris et al .
( 2005 ) find to host exeptionally blue nuclei , @xmath270 mag .
for single burst stellar populations , such colors would require ages @xmath271 gyr for virtually any choice of metallicity ( see figure 6 of paper i ) .
by contrast , we measure colors in the range 0.81.2 for these three nuclei .
in addition , we find poor agreement ( @xmath84 scatter = 0056 ) between the half - light radii measured in the two studies .
in the lower right panel of figure [ fig28 ] , we plot the de propris et al .
( 2005 ) half - light radii against the mean of our measurements in the @xmath28 and @xmath2 bandpasses .
an internal comparison of our @xmath272 and @xmath2-band measurements shows good agreement , with an @xmath84 scatter of @xmath75 001 ( 4.1 ) .
unfortunately , an internal comparison of the de propris et al .
( 2005 ) is not possible since they report a single value of the half - light radius for each nucleus , and it is not clear if this value refers to a measurement made in a single bandpass , or the average of measurements in the @xmath28 and @xmath2 bandpasses .
figure [ fig28 ] also shows a comparison of our magnitudes , colors and half - light radii for 25 nuclei to those of strader ( 2006 ) ( open squares ) .
the strader ( 2006 ) measurements were also determined using the ishape package ( larsen 1999 ) , although these authors used an empirical psf and assumed a king model nucleus of fixed concentration index @xmath273 .
although there is no discussion of how the contribution from the underlying galaxy was modeled in their analysis of the nuclei , the authors do state that photometry and size measurements for the nuclei were carried out using procedures identical to the globular clusters , in which the background is usually modeled as a constant or a plane . however , near the photocenter where the nuclei are found , the galaxy light is varying rapidly in both the radial and azimuthal directions , and since the galaxy brightness profiles nearly always exhibit an inward rise , this procedure will lead to overestimates of the nuclei luminosities and sizes . from the upper panels of figure [ fig28 ] , we see that the strader ( 2006 ) magnitudes are , on average , @xmath75 0.4 mag brighter than ours .
in addition , the discrepancy rises to @xmath274 1 mag for the faintest nuclei those which should be most prone to errors in modeling the underlying galaxy light .
there is better agreement between the measured colors from the two studies , as the lower left panel shows ( @xmath84 scatter = 0.098 and 0.072 mag for the integrated and aperture colors , respectively ) . at the same time
, however , there is poor agreement between the measured half - light radii ( @xmath84 scatter = 0055 ) , where their radii are @xmath75 80% larger than ours .
unfortunately , strader ( 2006 ) tabulate a single value of the radius for each nucleus , so no internal comparison of their size measurements is possible .
babul , a. , & rees , m.j .
1992 , mnras , 255 , 346 bassino , l.p . ,
muzzio , j.c .
, & rabolli , m. 1994 , , 431 , 634 bekki , k. , couch , w.j . , & drinkwater , m.j .
2001 , , 552 , l105 bekki , k. , couch , w.j . ,
drinkwater , m.j . , & shioya , y. 2004 , apj , 610 , l13 bentez , n. , 2002 , bull .. am .
soc . , 34 , 1236 binggeli , b. , barazza , f. , & jerjen , h. 2000 , a&a , 359 , 447 binggeli , b. , & cameron , l.m .
1991 , a&a , 252 , 27 binggeli , b. , & cameron , l.m .
1993 , a&ass , 98 , 297 binggeli , b. , sandage , a. , & tammann , g.a . 1985 , aj , 90 , 1681 ( bst85 ) binggeli , b. , tammann , g.a .
, & sandage , a. 1987 , aj , 94 , 251 ( bts87 ) bker , t. , laine , s. , van der marel , r.p . ,
sarzi , m. , rix , h .- w .
, ho , l. , & shields , j.c .
2002 , , 123 , 1389 bker , t. , sarzi , m. , mclaughlin , d.e .
, van der marel , r.p . ,
rix , h .- w .
, ho , l.c . , & shields , j.c .
2004 , , 127 , 105 bower , r.g .
, lucey , j.r . , & ellis , r.s .
1992 , , 254 , 601 butler , d. j. , & martnez - delgado , d. 2005 , , 129 , 2217 byun , y.i .
, 1996 , , 111 , 1889 caldwell , n. 1983 , aj , 88 , 808 caldwell , n. 1987 , aj , 94 , 1116 caldwell , n. , & bothun , g. d. 1987 , aj , 94 , 1126 caldwell , n. , armandroff , t.e . , seitzer , p. , & da costa , g.s .
1992 , , 103 , 840 caon , n. , capaccioli , m. , & donofrio , m. 1993 , , 265 , 1013 capuzzo - dolcetta , r. 1993 , , 415 , 616 capuzzo - dolcetta , r. , & tesseri , a. 1999 , , 308 , 961 carollo , c.m . ,
franx , m. , illingworth , g.d .
, & forbes , d.a .
1997 , , 481 , 710 carollo , c.m . , stiavelli , m. , & mack , j. 1998 , , 116 , 68 chabrier , g. 2003 , , 115 , 763 chandrasekhar , s. 1943 , , 97 , 255 chiaberge , m. , capetti , a. , & celotti , a. 1999 , , 349 , 77 coleman , m. , da costa , g.s .
, bland - hawthorn , j. , martnez - delgado , freeman , k.c . & malin , d. 2004 , , 127 , 832 ct , p. , blakeslee , j.p . , ferrarese , l. , jordn , a. , mei , s. , merritt , d. , milosavljevi , m. , peng , e.w . ,
tonry , j.l .
, & west , m.j .
2004 , , 153 , 223 ( paper i ) crane , p. , 1993
, , 106 , 1371 davies , j.i . , & phillipps , s. 1988 , , 233 , 553 de propris , r. , phillipps , s. , drinkwater , m.j . ,
gregg , m.d . ,
jones , j.b .
, evstigneeva , e. , & bekki , k. 2005 , , 623 , 105 de rijcke , s. , & debattista , v.p .
2004 , apj , 603 , l25 de vaucouleurs , g. 1948 , ann .
dastrophys . , 11 , 247 djorgovski , s. , bendinelli , o. , parmeggiani , g. , & zavatti , f. 1992 , in morphological and physical classification of galaxies , ed . g. longo , m. capaccioli , & g. busarello ( kluwer , dordrecht ) , p. 439
dressler , a. 1980 , , 236 , 351 drinkwater , m.j .
, phillipps , s. , gregg , m.d . ,
parker , q.a . ,
smith , r.m . ,
davies , j.i . ,
jones , j.b . , sadler , e.m .
1999 , , 511 , l97 drinkwater , m. j. , phillipps , s. , jones , j. b. , gregg , m. d. , deady , j. h. , davies , j. i. , parker , q. a. , sadler , e. m. , & smith , r. m. 2000 , , 355 , 900 durrell , p.r .
1997 , , 113 , 531 ebisuzaki , t. , makino , j. , & okumura , s.k .
1991 , nature , 354 , 212 fabbiano , g. , gioia , i.m . , & trinchieri , g. 1989 , , 347 , 127 faber , s. , 1997 , , 114 , 1771 ferrarese , l. , & merritt , d. 2000 , 539 , l9 ferrarese , l. , & ford , h.c . 2005 , ssrv , 116 , 523 ferrarese , l. , ct , p. , jordn , a. , peng , e.w . , blakeslee , j.p . ,
piatek , s. , mei , s. , merritt , d. , milosavljevi , m. , tonry , j.l . , & west , m.j .
2006a , , in press , ( astro - ph/0602297 ) ( paper vi ) ferrarese , l. , ct , p. , dalla bont , e. , peng , e.w . , merritt , d. , jordn , a. , blakeslee , j.p . ,
haegan , m. , mei , s. , piatek , s. , tonry , j.l . , & west , m.j .
2006b , , submitted ferguson , h.c .
1989 , , 98 , 367 ferguson , h.c .
, & sandage , a. 1989 , , 346 , l53 ford , h.c . , 1998 , proc .
spie , 3356 , 234 gebhardt , k. , 2000 , , 539 , l13 graham , a.w . , & guzmn , r. 2003 , , 125 , 2936 graham , a.w . ,
erwin , p. , trujillo , i. , & asensio ramos , a. 2003b , aj , 125 , 2951 graham , a.w .
2004 , , 613 , 33 graham , a.w . , & driver , s.p .
2005 , pasa , 22 , 118 grant , n.i . , kuipers , j.a . , & phillipps , s. 2005 , , 363 , 1019 haegan , m. , jordn , a. , ct , p. , djorgovski , s.g . ,
mclaughlin , d.e . ,
blakeslee , j.p .
, mei , s. , west , m.j . , peng , e.w . ,
ferrarese , l. , milosavljevi , m. , tonry , j.l . , & merritt , d. 2005 , apj , 627 , 203 ( paper vii ) haegan , m. , 2006 , in preparation haehnelt , m.g . , natarajan , p. , & rees , m.j .
1998 , , 300 , 817 hilker , m. , infante , l. , vieira , g. , kissler - patig , m. , & richtler , t. 1999 , , 134 , 75 ho , l.c . , filippenko , a.v . , &
sargent , w.l.w .
1997 , , 112 , 315 ho , l.c .
1999 , , 516 , 672 jedrzejewski , r.i .
1987 , , 226 , 747 jones , j.b . ,
drinkwater , m.j . ,
jurek , r. , phillipps , s. , gregg , m.d .
, bekki , k. , couch , w.j .
, karick , a. , parker , q.a . , & smith , r.m .
2006 , , 131 , 312 jordn , a. , blakeslee , j.p . , peng , e. , mei , s. , ct , p. , ferrarese , l. , merritt , d. , milosavljevi , m. , tonry , j.l . , & west , m.j .
2004a , , 154 , 509 ( paper ii ) jordn , a. , ct , p. , ferrarese , l. , blakeslee , j.p .
, mei , s. , merritt , d. , milosavljevi , m. , peng , e.w . , tonry , j.l . , & west , m.j .
2005b , , 613 , 279 ( paper iii ) jordn , a. , ct , p. , blakeslee , j.p . , ferrarese , l. , mclaughlin , d.e . ,
mei , s. , peng , e.w . ,
tonry , j.l . ,
merritt , d. , milosavljevi , m. , sarazin , c.l . ,
sivakoff , g.r . ,
west , m.j .
2005 , , 634 , 1002 ( paper x ) jordn , a. , 2006 , in preparation jerjen , h. , & binggeli , b. 1997 , in asp conf .
116 , the nature of elliptical galaxies , ed .
m. arnaboldi , g. s. da costa , & p. saha ( san francisco : asp ) , 239 julian , w.h .
, kooiman , b.l . , & sanders , w.l .
1997 , , 109 , 297 king , i.r . 1966 , aj , 71 , 64 kleyna , j.t . , wilkinson , m.i . , gilmore , g. , & evans , n.w .
2003 , , 588 , l21 kormendy , j. 1985 , apj , 295 , 73 kormendy , j. , & richstone , d 1995 , , 33 , 581 larsen , s. 1999 , , 139 , 393 lauer , t.r .
, 1995 , , 110 , 2622 lauer , t.r . , 2005 , , 129 , 2138 layden , a.c . , & sarajedini , a. 2000 , , 119 , 1760 lotz , j.m . , telford , r. , ferguson , h.c . , miller , b.w . , stiavelli , m. , & mack , j. 2001 , apj , 552 , 572 lotz , j.m . , miller , b.w . , & ferguson , h.c .
2004 , , 613 , 262 matthews , l.d . , 1999 ,
aj , 118 , 208 mcconnachie , a.w . , &
irwin , m.j .
2006 , , 365 , 1263 mei , s. , blakeslee , j.p . ,
tonry , j.l .
, jordn , a. , peng , e.w . ,
ct , p. , ferrarese , l. , merritt , d. , milosavljevi , m. , & west , m.j .
2005a , , 156 , 113 ( paper iv ) mei , s. , blakeslee , j.p . , tonry , j.l . ,
jordn , a. , peng , e.w . ,
ct , p. , ferrarese , l. , west , m.j . ,
merritt , d. , & milosavljevic , m. 2005b , , 625 , 121 ( paper v ) merritt , d. , & ferrarese , l. 2001 , , 320 , 30 merritt , d. , milosavljevi , m. , favata , m. , hughes , s.a . , & holz , d.e .
2004 , apj , 607 , l9 merritt , d. , navarro , j.f . , ludlow , a. , & jenkins , a. 2005 , , 624 , l85 merritt , d. , 2006 , in preparation michie , r.w .
1963 , , 125 , 127 mieske , s. , hilker , m. , infante , l. , & jordn , a. 2006 , , in press ( astro - ph/0512474 ) mihos , j.c . , & hernquist , l. 1994 , , 437 , l47 miller , r.h . , &
smith , b.f .
1992 , , 393 , 508 milosavljevi , m. , & merritt , d. 2001 , , 563 , 34 monaco , l. , bellazini , m. , ferraro , f.r . , & pancino , e. 2005 , , 356 , 1396 mori , m. , & burkert , a. 2000 , , 538 , 559 moore , b. , lake , g. , & katz , n. 1998 , , 495 , 139 navarro , j.f . , hayashi , e. , power , c. , jenkins , a.r . , frenk , c.s . ,
white , s.d.m .
, springel , v. , stadel , j. , & quinn , t.r 2004 , , 349 , 1039 nelder , j.a . , & mead , r. 1965 , comput.j . ,
7 , 308 oh , k.s .
, & lin , d.n.c . 2000 , , 543 , 620 oemler , a. 1974 , , 194 , 1 palma , c. , majewski , s.r . , siegel , m.h . , patterson , r.j . ,
ostheimer , j.c . , & link , r. 2003 , , 125 , 1352 peng , e.w . ,
jordn , a. , ct , p. , blakeslee , j.p . ,
ferrarese , l. , mei , s. , west , m.j . ,
merritt , d.,milosavljevic , m. , & tonry , j.l .
2006a , , 639 , 95 ( paper ix ) peng , e.w . , ct , p. , jordn , a. , blakeslee , j.p . , ferrarese , l. , mei , s. , west , m.j . ,
merritt , d.,milosavljevic , m. , & tonry , j.l .
2006b , , in press ( paper xi ) phillips , a.c . , illingworth , g.d .
, mackenty , j.w . , & franx , m. 1996 , , 111 , 1566 phillipps , s. , drinkwater , m. , gregg , m. , & jones , j. 2001 , apj , 560 , 201 postman , m. , 2005 , , 623 , 721 quinlan , g.d . , & hernquist , l. 1997 , newa , 2 , 533 rangarajan , f.v.n . ,
white , d.a .
, ebeling , h. , & fabian , a.c .
1995 , , 277 , 1047 ravindranath , s. , ho , l.c . , peng , c.y . ,
filippenko , a.v . , & sargent , w.l.w .
2001 , , 122 , 653 reaves , g. 1983 , , 53 , 375 rest , a. , van den bosch , f.c . ,
jaffe , w. , tran , h. , tsvetanov , z. , ford , h.c . , davies , j. , & schafer , j. 2001 , , 121 , 2431 sandage , a. , binggeli , b. , & tammann , g.a .
1985 , aj , 90 , 1759 srsic , j .- l . 1968 , atlas de galaxias australes ( crdoba : obs .
crdova ) schlegel , d.j . ,
finkbeiner , d.p . , &
davis , m. 1998 , apj , 500 , 525 silk , j. , wyse , r.f.g . , & shields , g.a .
1987 , , 332 , l59 silk , j. , & rees , m.j .
1998 , , 331 , l1 sirianni , m. , jee , m.j . ,
bentez , n. , blakeslee , j.p . ,
martel , a.r . ,
meurer , g. , clampin , m. , de marchi , g. , ford , h.c . ,
gilliland , r. , hartig , g.f . , illingworth , g.d .
, mack , j. , & mccann , w.j .
2005 , , 117 , 1049 smith , g.p . ,
treu , t. , ellis , r.s . ,
moran , s.m . , & dressler , a. 2004 , , 620 , 78 stiavelli , m. , miller , b.w . ,
ferguson , h.c .
, mack , j. , whitmore , b.c . , & lotz , j.m .
2001 , , 121 , 1385 strader , j. , brodie , j.p . ,
spitler , l. , & beasley , m.a .
2006 , , submitted ( astro - ph/0508001 ) tonry , j.l . ,
dressler , a. , blakeslee , j.p . , ajhar , e.a . , fletcher , a.b . , luppino , g.a . , metzger , m.r . , & moore , c.b .
2001 , apj , 546 , 681 tremaine , s.d . , ostriker , j.p . , & spitzer , l. , jr .
1975 , apj , 196 , 407 tremaine , s.d . 1976
, , 203 , 72 trujillo , i. , erwin , p. , asensio ramos , a. , & graham , a.w .
2004 , aj , 127 valluri , m. , ferrarese , l. , merritt , d. , & joseph , c.l .
2005 , , 628 , 137 van den bergh , s. 1986 , aj , 91 , 271 walcher , c.j .
, van der marel , r.p . , mclaughlin , d. , rix , h .- w . ,
bker , t. , hring , n. , ho , l.c . , sarzi , m. , & shields , j.c 2005 , , 618 , 237 weedman , d.w .
1983 , , 266 , 479 wehner , e.h . , &
harris , w.e .
2006 , , submitted west , m.j . , ct , p. , jones , c. , forman , w. , & marzke , r.o .
1995 , , 453 , l77 .
wrobel , j.m .
1991 , , 101 , 127 wyiethe , j.s.b . , &
loeb , a. 2002 , , 581 , 886 zang , t.a . , & hohl , f. 1978 , apj , 226 , 521 cccccccccccccccccl 1 & 1226 & m49 , n4472 & 9.31 & 0.022 & 16.74 & 15.12 & n & ii & cs & & & & & & & & dust .
+ 2 & 1316 & m87 , n4486 & 9.58 & 0.023 & 17.58 & 15.95 & n & 0 & cs & & & & & & & & dust .
+ 3 & 1978 & m60 , n4649 & 9.81 & 0.026 & 16.93 & 15.27 & n & ii & cs & & & & & & & & agn ?
+ 4 & 881 & m86 , n4406 & 10.06 & 0.029 & 16.66 & 15.08 & n & ii & cs & & & & & & & & dust .
+ 5 & 798 & m85 , n4382 & 10.09 & 0.030 & 16.23 & 14.89 & n & ii & cs & & & & & & & & dust : .
+ 6 & 763 & m84 , n4374 & 10.26 & 0.041 & 16.47 & 14.88 & n & 0 & cs & & & & & & & & dust . agn .
+ 7 & 731 & n4365 & 10.51 & 0.021 & 16.95 & 15.34 & n & ib & cs & & & & & & & [email protected] & + 8 & 1535 & n4526 & 10.61 & 0.022 & & & n & 0 & & & & & & & & & dust .
+ 9 & 1903 & m59 , n4621 & 10.76 & 0.032 & 16.44 & 14.80 & n & i d & s & & & & & & & & + 10 & 1632 & m89 , n4552 & 10.78 & 0.041 & 16.34 & 14.68 & n & ii & cs & & & & & & & & dust .
agn ? + 11 & 1231 & n4473 & 11.10 & 0.028 & 16.45 & 14.87 & n & ii & s & & & & & & & & + 12 & 2095 & n4762 & 11.18 & 0.022 & 16.76 & 15.25 & n & ib & s & & & & & & & [email protected] & + 13 & 1154 & n4459 & 11.37 & 0.045 & 16.67 & 15.11 & n & i d & s & & & & & & & & dust .
+ 14 & 1062 & n4442 & 11.40 & 0.022 & 16.63 & 15.02 & n & i d & s & & & & & & & & + 15 & 2092 & n4754 & 11.51 & 0.032 & 16.77 & 15.15 & n & ib & s & & & & & & & [email protected] & + 16 & 369 & n4267 & 11.80 & 0.047 & 16.98 & 15.36 & n & ib & s & & & & & & & [email protected] & + 17 & 759 & n4371 & 11.80 & 0.036 & 17.50 & 15.93 & n & ii & s & & & & & & & & dust .
+ 18 & 1692 & n4570 & 11.82 & 0.022 & 16.65 & 15.06 & n & ic & s & & & & & & & & + 19 & 1030 & n4435 & 11.84 & 0.029 & & & n & 0 & & & & & & & & & dust .
+ 20 & 2000 & n4660 & 11.94 & 0.034 & 16.47 & 14.89 & n & i d & s & & & & & & & & + 21 & 685 & n4350 & 11.99 & 0.028 & 16.57 & 14.98 & n & 0 & s & & & & & & & & dust .
+ 22 & 1664 & n4564 & 12.02 & 0.033 & 16.90 & 15.27 & n & ic & s & & & & & & & & + 23 & 654 & n4340 & 12.03 & 0.026 & 17.45 & 15.95 & n & ib & s & & & & & & & [email protected] & + 24 & 944 & n4417 & 12.08 & 0.025 & 16.93 & 15.42 & n & ic & s & & & & & & & & + 25 & 1938 & n4638 & 12.11 & 0.026 & 16.97 & 15.43 & n & ib & s & & & & & & & [email protected] & + 26 & 1279 & n4478 & 12.15 & 0.024 & 17.40 & 15.90 & n & i d & s & & & & & & & & + 27 & 1720 & n4578 & 12.29 & 0.021 & 17.66 & 16.13 & n & ia & s & 18.40 & 16.82 & 1.57 & 1.57 & 0.085 & 0.085 & [email protected] & + 28 & 355 & n4262 & 12.41 & 0.036 & 16.75 & 15.15 & n & i d & s & & & & & & & & dust .
+ 29 & 1619 & n4550 & 12.50 & 0.040 & 17.69 & 16.25 & n & ia & s & 17.13 & 15.58 & 1.55 & 1.58 & 0.323 & 0.323 & [email protected] & dust . agn .
+ 30 & 1883 & n4612 & 12.57 & 0.025 & 17.20 & 15.86 & n & ia & s & 18.74 & 17.60 & 1.14 & 1.11 & ( 0.024 ) & ( 0.024 ) & [email protected] & + 31 & 1242 & n4474 & 12.60 & 0.042 & 17.59 & 16.09 & n & ia & s & 19.84 & 18.14 & 1.70 & 1.68 & 0.035 & 0.035 & [email protected] & + 32 & 784 & n4379 & 12.67 & 0.024 & 17.53 & 16.05 &
n & ia & s & 18.34 & 16.68 & 1.66 & 1.67 & 0.161 & 0.161 & [email protected] & + 33 & 1537 & n4528 & 12.70 & 0.046 & 17.30 & 15.82 & n & i d & s & & & & & & & & + 34 & 778 & n4377 & 12.72 & 0.038 & 16.94 & 15.44 & n & i d & s & & & & & & & & + 35 & 1321 & n4489 & 12.84 & 0.028 & 17.92 & 16.52 & n & i d & s & & & & & & & & + 36 & 828 & n4387 & 12.84 & 0.033 & 18.01 & 16.50 & n & ia & s & 18.53 & 16.96 & 1.57 & 1.59 & 0.208 & 0.208 & [email protected] & + 37 & 1250 & n4476 & 12.91 & 0.028 & 17.81 & 16.58 & n & ia & cs & 19.73 & 18.19 & 1.55 & 1.52 & 0.026 & 0.026 & [email protected] & dust .
+ 38 & 1630 & n4551 & 12.91 & 0.039 & 18.00 & 16.46 & n & ia & s & 17.39 & 15.72 & 1.68 & 1.71 & 0.501 & 0.501 & [email protected] & + 39 & 1146 & n4458 & 12.93 & 0.023 & 18.03 & 16.59 & n & ia & s & 15.37 & 13.95 & 1.42 & 1.47 & 0.780 & 0.780 & [email protected] & + 40 & 1025 & n4434 & 13.06 & 0.022 & 17.48 & 16.00 & n & ic & s & & & & & & & & + 41 & 1303 & n4483 & 13.10 & 0.020 & 18.00 & 16.50 & n & ib & s & & & & & & & [email protected] & + 42 & 1913 & n4623 & 13.22 & 0.022 & 18.51 & 17.02 & n & ia & s & 17.55 & 15.95 & 1.60 & 1.64 & 0.597 & 0.597 & [email protected] & + 43 & 1327 & n4486a & 13.26 & 0.023 & 17.49 & 15.90 & n & ii & s & & & & & & & & dust .
+ 44 & 1125 & n4452 & 13.30 & 0.030 & 18.57 & 17.15 & n & ia & s & 20.48 & 19.51 & 0.97 & 0.950 & 0.060 & 0.060 & [email protected] & + 45 & 1475 & n4515 & 13.36 & 0.031 & 17.85 & 16.46 & n & ib & s & & & & & & & [email protected] & + 46 & 1178 & n4464 & 13.37 & 0.022 & 17.47 & 15.99 & n & ib & s & & & & & & & [email protected] & + 47 & 1283 & n4479 & 13.45 & 0.029 & 19.11 & 17.62 & n & ia & s & 20.65 & 19.07 & 1.58 & 1.55 & 0.053 & 0.053 & [email protected] & + 48 & 1261 & n4482 & 13.56 & 0.029 & 19.71 & 18.49 & y & ia & s & 19.50 & 18.28 & 1.22 & 1.25 & 0.041 & 0.036 & [email protected] & + 49 & 698 & n4352 & 13.60 & 0.026 & 18.43 & 17.03 & n & ia & s & 19.93 & 18.61 & 1.32 & 1.29 & 0.041 & 0.041 & [email protected] & + 50 & 1422 & i3468 & 13.64 & 0.031 & 20.04 & 18.76 & y & ia & s & 20.22 & 19.00 & 1.22 & 1.26 & 0.038 & 0.035 & [email protected] & dust : .
+ 51 & 2048 & i3773 & 13.81 & 0.032 & 19.39 & 18.17 & n & ia & s & 21.45 & 20.30 & 1.15 & 1.14 & 0.037 & 0.031 & [email protected] & + 52 & 1871 & i3653 & 13.86 & 0.030 & 18.89 & 17.42 & n & ia & s & 18.73 & 17.48 & 1.25 & 1.35 & 0.125 & 0.108 & [email protected] & + 53 & 9 & i3019 & 13.93 & 0.039 & 22.11 & 20.95 & y & ie & s & 22.01 & 21.22 & 0.79 & 0.88 & 0.040 & 0.034 & [email protected] & dirr / de .
+ 54 & 575 & n4318 & 14.14 & 0.025 & 18.38 & 16.96 & n & i d & s & & & & & & & & + 55 & 1910 & i809 & 14.17 & 0.031 & 20.05 & 18.64 & y & ia & s & 19.82 & 18.63 & 1.19 & 1.19 & 0.038 & 0.038 & [email protected] & + 56 & 1049 & u7580 & 14.20 & 0.022 & 19.74 & 18.73 & n & ii & s & & & & & & & & + 57 & 856 & i3328 & 14.25 & 0.024 & 20.67 & 19.46 & y & ia & s & 18.97 & 17.85 & 1.12 & 1.17 & 0.163 & 0.153 & [email protected] & + 58 & 140 & i3065 & 14.30 & 0.037 & 19.60 & 18.37 & n & ia & s & 22.09 & 21.19 & 0.91 & 0.93 & 0.030 & ( 0.022 ) & [email protected] & + 59 & 1355 & i3442 & 14.31 & 0.034 & 21.84 & 20.59 & y & ia & s & 21.10 & 20.07 & 1.03 & 1.03 & 0.043 & 0.038 & [email protected] & + 60 & 1087 & i3381 & 14.31 & 0.027 & 20.60 & 19.27 & y & ia & s & 20.22 & 18.89 & 1.33 & 1.33 & 0.027 & 0.027 & [email protected] & + 61 & 1297 & n4486b & 14.33 & 0.021 & 17.49 & 15.93 & n & i d & s & & & & & & & & + 62 & 1861 & i3652 & 14.37 & 0.029 & 20.76 & 19.43 & y & ia & s & 20.11 & 19.06 & 1.04 & 1.19 & 0.137 & 0.119 & [email protected] & + 63 & 543 & u7436 & 14.39 & 0.031 & 20.56 & 19.29 & n & ia & cs & 22.56 & 21.20 & 1.36 & 1.19 & 0.157 & 0.196 & [email protected] & + 64 & 1431 & i3470 & 14.51 & 0.051 & 19.82 & 18.40 & y & ia & s & 19.66 & 18.54 & 1.13 & 1.17 & 0.238 & 0.233 & [email protected] & + 65 & 1528 & i3501 & 14.51 & 0.028 & 19.51 & 18.18 & n & ia & cs & 22.27 & 21.31 & 0.96 & 0.89 & ( 0.015 ) & ( 0.018 ) & [email protected] & + 66 & 1695 & i3586 & 14.53 & 0.045 & 20.34 & 19.22 & n & ia & cs & 22.61 & 21.23 & 1.38 & 1.35 & ( 0.022 ) & ( 0.024 ) & [email protected] & + 67 & 1833 & & 14.54 & 0.036 & 19.28 & 18.09 & n & ii & s & & & & & & & & + 68 & 437 & u7399a & 14.54 & 0.029 & 20.75 & 19.45 & y & ia & s & 20.00 & 19.01 & 1.00 & 1.03 & 0.089 & 0.083 & [email protected] & + 69 & 2019 & i3735 & 14.55 & 0.022 & 20.83 & 19.63 & y & ia & s & 20.31 & 19.20 & 1.12 & 1.18 & 0.037 & 0.029 & [email protected] & + 70 & 33 & i3032 & 14.67 & 0.037 & 20.79 & 19.75 & y & ia & s & 22.18 & 21.27 & 0.91 & 0.91 & 0.033 & 0.032 & [email protected] & + 71 & 200 & & 14.69 & 0.030 & 20.56 & 19.31 & y & ia & s & 22.86 & 21.94 & 0.92 & 1.06 & 0.053 & 0.038 & [email protected] & + 72 & 571 & & 14.74 & 0.022 & 20.12 & 19.11 & n & 0 & s & & & & & & & & dust .
+ 73 & 21 & i3025 & 14.75 & 0.021 & 20.42 & 18.79 & n & ie & s & 20.64 & 20.36 & 0.28 & 0.32 & 0.033 & 0.029 & [email protected] & dirr / de + 74 & 1488 & i3487 & 14.76 & 0.021 & 20.24 & 19.51 & n & ia & s & 23.71 & 22.99 & 0.72 & 0.69 & 0.025 & 0.025 & [email protected] & + 75 & 1779 & i3612 & 14.83 & 0.028 & 20.45 & 19.62 & n & ie & s & 22.41 & 21.58 & 0.83 & 0.85 & 0.021 & 0.024 & [email protected] & dust . dirr / de + 76 & 1895 & u7854 & 14.91 & 0.017 & 20.54 & 19.42 & n & ia & s & 23.48 & 22.61 & 0.88 & 0.89 & ( 0.023 ) & ( 0.021 ) & [email protected] & + 77 & 1499 & i3492 & 14.94 & 0.030 & 19.55 & 19.12 & n & ii & s & & & & & & & & dirr / de + 78 & 1545 & i3509 & 14.96 & 0.042 & 19.98 & 18.53 & n & ia & s & 21.93 & 20.88 & 1.05 & 1.16 & 0.050 & 0.037 & [email protected] & + 79 & 1192 & n4467 & 15.04 & 0.023 & 19.05 & 17.53 & n & ia & s & 19.09 & 17.90 & 1.19 & 1.22 & 0.120 & 0.121 & [email protected] & + 80 & 1857 & i3647 & 15.07 & 0.025 & 22.72 & 21.72 & y & ie & s & 21.11 & 20.27 & 0.84 & 0.86 & 0.041 & 0.041 & [email protected] & + 81 & 1075 & i3383 & 15.08 & 0.027 & 21.43 & 21.40 & y & ia & s & 21.07 & 20.11 & 0.96 & 0.97 & 0.040 & 0.039 & [email protected] & + 82 & 1948 & & 15.10 & 0.025 & 21.59 & 20.57 & n & ie & s & 24.30 & 23.21 & 1.09 & 1.04 & 0.041 & 0.054 & [email protected] & + 83 & 1627 & & 15.16 & 0.039 & 19.04 & 17.59 &
n & ia & s & 18.83 & 17.39 & 1.44 & 1.46 & 0.197 & 0.197 & [email protected] & + 84 & 1440 & i798 & 15.20 & 0.028 & 19.25 & 17.96 & n & ia & s & 19.70 & 18.38 & 1.32 & 1.33 & 0.063 & 0.056 & [email protected] & + 85 & 230 & i3101 & 15.20 & 0.028 & 21.14 & 19.91 & y & ia & s & 20.31 & 19.22 & 1.09 & 1.08 & 0.038 & 0.033 & [email protected] & + 86 & 2050 & i3779 & 15.20 & 0.023 & 21.02 & 19.82 & y & ia & s & 22.38 & 21.41 & 0.97 & 1.08 & 0.073 & 0.066 & [email protected] & + 87 & 1993 & & 15.30 & 0.025 & 20.61 & 19.34 & n & ib & s & & & & & & & [email protected] & + 88 & 751 & i3292 & 15.30 & 0.032 & 19.92 & 18.58 & n & ia & s & 21.22 & 20.17 & 1.05 & 1.22 & 0.046 & 0.035 & [email protected] & + 89 & 1828 & i3635 & 15.33 & 0.037 & 21.40 & 20.12 & y & ia & s & 21.50 & 20.50 & 1.00 & 1.04 & 0.060 & 0.057 & [email protected] & + 90 & 538 & n4309a & 15.40 & 0.020 & 20.14 & 18.98 & n & ia & s & 21.27 & 20.15 & 1.13 & 1.13 & 0.033 & 0.030 & [email protected] & + 91 & 1407 & i3461 & 15.49 & 0.032 & 20.75 & 19.50 & y & ia & s & 20.39 & 19.40 & 0.98 & 1.08 & 0.145 & 0.127 & [email protected] & + 92 & 1886 & & 15.49 & 0.033 & 22.08 & 21.12 & y & ia & s & 22.05 & 21.04 & 1.01 & 0.97 & 0.036 & 0.041 & [email protected] & + 93 & 1199 & & 15.50 & 0.022 & 19.56 & 17.97 & n & ia & s & 19.75 & 18.39 & 1.36 & 1.45 & 0.075 & 0.063 & [email protected] & + 94 & 1743 & i3602 & 15.50 & 0.019 & 21.45 & 20.31 & n & ib & s & & & & & & & [email protected] & + 95 & 1539 & & 15.68 & 0.032 & 22.01 & 20.86 & y & ia & s & 20.93 & 19.81 & 1.11 & 1.02 & 0.231 & 0.265 & [email protected] & + 96 & 1185 & & 15.68 & 0.023 & 21.89 & 20.62 & y & ia & s & 20.86 & 19.91 & 0.95 & 1.00 & 0.057 & 0.050 & [email protected] & + 97 & 1826 & i3633 & 15.70 & 0.017 & 20.52 & 19.34 & y & ia & s & 20.10 & 18.91 & 1.19 & 1.17 & ( 0.024 ) & ( 0.025 ) & [email protected] & + 98 & 1512 & & 15.73 & 0.050 & 20.27 & 19.42 & n & ii & cs & & & & & & & & dirr / de + 99 & 1489 & i3490 & 15.89 & 0.034 & 22.03 & 20.98 & y & ia & s & 22.38 & 21.51 & 0.87 & 0.83 & 0.051 & 0.058 & [email protected] & + 100 & 1661 & & 15.97 & 0.020 & 22.57 & 21.34 & y & ia & s & 20.30 & 19.25 & 1.05 & 1.02 & 0.079 & 0.082 & [email protected] & + lr no classification possible ( type 0 ) & + 0.2truein _ dust _ & 4 + 0.2truein _ agn _ & 2 + 0.2truein total & * 6 * + & + nucleated ( type i ) & + 0.2truein _ fitted nucleus _ ( ia ) & 51 + 0.2truein _ no fit possible _ ( ib ) & 11 + 0.2truein total & * 62 * + & + non - nucleated ( type ii ) & * 12 * + & + uncertain ( type i ) & + 0.2truein _ likely nucleated _ ( ic ) & 4 + 0.2truein _ possibly nucleated _
( i d ) & 11 + 0.2truein _ possible offset nucleus _ ( ie ) & 5 + 0.2truein total & * 20 * + & + all galaxies & * 100 * + cccccccc 1226 & ngc4472 & -21.87 & & no & yes & & no ( ii ) + 1978 & ngc4649 & -21.39 & & & no & & no ( ii ) + 881 & ngc4406 & -21.15 & & no & yes & & no ( ii ) + 798 & ngc4382 & -21.13 & & & & & no ( ii ) + 763 & ngc4374 & -21.01 & & yes & & & agn ( o ) + 731 & ngc4365 & -20.67 & no & & yes & & yes ( ib ) + 1632 & ngc4552 & -20.49 & & & yes & & no ( ii ) + 1903 & ngc4621 & -20.47 & no & & no & & possibly ( i d ) + 1231 & ngc4473 & -20.11 & & & no & & no ( ii ) + 2000 & ngc4660 & -19.30 & & & no & & possibly ( i d ) + 1664 & ngc4564 & -19.21 & no & & & & probably ( ic ) + 944 & ngc4417 & -19.12 & & no & & & probably ( ic ) + 1279 & ngc4478 & -19.04 & no & & yes & & possibly ( i d ) + 1242 & ngc4474 & -18.67 & yes ( ii ) & & & & yes ( ia ) + 1146 & ngc4458 & -18.26 & & & no & & yes ( ia ) + 1261 & ngc4482 & -17.65 & yes ( iii ) & & & & yes ( ia ) + 9 & ic3019 & -17.33 & & & & no & possible offset ( ie ) + 1297 & ngc4486b & -16.85 & & & no & & possibly ( i d ) + 543 & ugc7436 & -16.93 & & & & no & yes ( ia ) + 1948 & ic3693 & -16.10 & & & & no & possible offset ( ie ) + lcc @xmath60 ( pc ) & @xmath8262 & [email protected] + & [email protected] ( med . ) & + @xmath275 ( mag arcsec@xmath128 ) & [email protected] & [email protected] + @xmath276 ( mag arcsec@xmath128 ) & [email protected] & [email protected] + @xmath277 ( mag ) & @[email protected] & @[email protected] + @xmath279 ( mag ) & @[email protected] & @[email protected] + @xmath280 ( mag ) & 0.671.61 & [email protected] + & [email protected] & + lcc bruzual & charlot ( 2003 ) & 1 & @[email protected] + & 2 & @[email protected] + & 5 & @[email protected] + & 10 & @[email protected] + peng ( 2006 ) & & @[email protected] + | the acs virgo cluster survey is a _ hubble space telescope _ program to obtain high - resolution imaging , in widely separated bandpasses ( f475w @xmath0 and f850lp @xmath1 @xmath2 ) , for 100 early - type members of the virgo cluster , spanning a range of @xmath1 460 in blue luminosity .
we use this large , homogenous dataset to examine the innermost structure of these galaxies and to characterize the properties of their compact central nuclei .
we present a sharp upward revision in the frequency of nucleation in early - type galaxies brighter than @xmath3 ( @xmath4% ) and show that ground - based surveys underestimated the number of nuclei due to surface brightness selection effects , limited sensitivity and poor spatial resolution .
we speculate that previously reported claims that nucleated dwarfs are more concentrated to the center of virgo than their non - nucleated counterparts may be an artifact of these selection effects .
there is no clear evidence from the properties of the nuclei , or from the overall incidence of nucleation , for a change at @xmath5 , the traditional dividing point between dwarf and giant galaxies .
there _ does _ , however , appear to be a fundamental transition at @xmath6 , in the sense that the brighter , core - srsic " galaxies lack resolved ( stellar ) nuclei .
a search for nuclei which may be offset from the photocenters of their host galaxies reveals only five candidates with displacements of more than 05 , all of which are in dwarf galaxies . in each case , though , the evidence suggests that these nuclei " are , in fact , globular clusters projected close to the galaxy photocenter . working from a sample of 51 galaxies with prominent nuclei , we find a median half - light radius of @xmath7 pc , with the sizes of individual nuclei ranging from 62 pc down to @xmath8 2 pc ( i.e. , unresolved in our images ) in about a half dozen cases . excluding these unresolved objects ,
the nuclei sizes are found to depend on nuclear luminosity according to the relation @xmath9 .
because the large majority of nuclei are resolved , we can rule out low - level agn as an explanation for the central luminosity excess in almost all cases . on average , the nuclei are @xmath1 3.5 mag brighter than a typical globular cluster . based on their broadband colors ,
the nuclei appear to have old to intermediate - age stellar populations .
the colors of the nuclei in galaxies fainter than @xmath10 are tightly correlated with their luminosities , and less so with the luminosities of their host galaxies , suggesting that their chemical enrichment histories were governed by local or internal factors . comparing the nuclei to the nuclear clusters " found in late - type spiral galaxies reveals a close match in terms of size , luminosity and overall frequency . a formation mechanism that is rather insensitive to the detailed properties of the host galaxy
is required to explain this ubiquity and homogeneity .
the mean of the frequency function for the nucleus - to - galaxy luminosity ratio in our nucleated galaxies , @xmath11 dex ( @xmath12 ) , is indistinguishable from that of the sbh - to - bulge mass ratio , @xmath13 dex ( @xmath14 ) , calculated in 23 early - type galaxies with detected supermassive black holes ( sbhs ) .
we argue that the compact stellar nuclei found in many of our program galaxies are the low - mass counterparts of the sbhs detected in the bright galaxies .
if this interpretation is correct , then one should think in terms of _ central massive objects _ either sbhs or compact stellar nuclei that accompany the formation of almost all early - type galaxies and contain a mean fraction @xmath1 0.3% of the total bulge mass .
in this view , sbhs would be the dominant formation mode above @xmath15 . |
intermittency is well - known to be an ubiquitous phenomenon in nonlinear science .
its arousal and main statistical properties have been studied and characterized already since long time ago , and different types of intermittency have been classified as types i iii intermittencies @xcite , on off intermittency , eyelet intermittency and ring intermittency @xcite . despite of some similarity ( the presence of two different regimes alternating suddenly with each other in the time series ) ,
every type of intermittency is governed by its own certain mechanisms and characteristics of the intermittent behavior ( such as the dependence of the mean length of the laminar phases on the control parameter , the distribution of the laminar phase lengths , etc . ) of different intermittency types are distinct .
there are no doubts that different types of intermittent behavior may take place in a wide spectrum of systems , including cases of practical interest for applications in radio engineering , medical , physiological , and other applied sciences .
this article is devoted to the comparison of characteristics of type - i intermittency in the presence of noise and eyelet intermittency taking place in the vicinity of the phase synchronization boundary .
although these types of intermittency are known to be characterized by different theoretical laws , we show here for the first time that these two types of the intermittent behavior considered hitherto as different phenomena are , in fact , the same type of the system dynamics .
the structure of the paper is the following . in sec .
[ sct : relation ] we give the brief theoretical data concerning both the type - i intermittency with noise and eyelet intermittency observed in the vicinity of the phase synchronization boundary as well as arguments confirming the equivalence of these types of the dynamics .
the next sections [ sct : numericalverification][sct : vdpunderrsslr ] aim to verify the statement given in the sec .
[ sct : relation ] by means of numerical simulations of the dynamics of several model systems such as a quadratic map , rssler oscillators , etc . eventually , in sec .
[ sct : boundary ] we discuss the problem of the upper boundary of the intermittent behavior .
the final conclusions are given in sec .
[ sct : conclusions ] .
first , we consider briefly both eyelet intermittency in the vicinity of the phase synchronization boundary and type - i intermittency in the presence of noise following conceptions accepted generally . the main arguments confirming equivalence of these types of the intermittent behavior are given afterwards . the intermittent behavior of type - i is known to be observed below the saddle - node bifurcation point , with the mean length of laminar phases @xmath0 being inversely proportional to the square root of the criticality parameter @xmath1 , i.e. @xmath2 where @xmath3 is the control parameter and @xmath4 is its bifurcation value corresponding to the bifurcation point @xcite .
the influence of noise on the system results in the transformation of characteristics of intermittency , with the intermittent behavior being observed in this case both below and above the saddle - node bifurcation point @xmath4 . in the supercritical region @xcite of the control parameter values ( i.e. , above the point of bifurcation , @xmath5 ) the mean length @xmath0 of the laminar phases is given by @xmath6 where @xmath7 , @xmath8 is the intensity of a delta - correlated white noise @xmath9 [ @xmath10 , @xmath11 , with equation ( [ eq : typeilawmeanlength ] ) being applicable in the region @xmath12 of the control parameter plane @xcite . in this region
the criticality parameter @xmath13 is large enough and , therefore , the approximate equation @xmath14 ( where @xmath15 ) is used typically ( see @xcite for detail ) instead of ( [ eq : typeilawmeanlength ] ) . in turn
, the distribution @xmath16 of the laminar phase lengths @xmath17 is governed by the exponential law @xcite @xmath18 for the chaotic systems in the vicinity of the phase synchronization boundary ( if the natural frequencies of oscillator and external signal are detuned slightly ) two types of the intermittent behavior and , correspondingly , two critical values are reported to exist @xcite .
below the boundary of the phase synchronization regime the dynamics of the phase difference @xmath19 features time intervals of the phase synchronized motion ( laminar phases ) persistently and intermittently interrupted by sudden phase slips ( turbulent phases ) during which the value of @xmath20 jumps up by @xmath21 . for two coupled chaotic systems there are two values of the coupling strength @xmath22 being the characteristic points which are considered to separate the different types of the dynamics . below the coupling strength value
@xmath23 the type - i intermittency is observed , with the power law @xmath24 taking place for the mean length of the laminar phases , whereas above the critical point @xmath25 the phase synchronization regime is revealed . for the coupling strength @xmath26 the super - long laminar behavior ( the so called _ `` eyelet intermittency '' _ ) should be detected .
for eyelet intermittency ( see , e.g. @xcite ) the dependence of the mean length @xmath0 of the laminar phases on the criticality parameter is expected to follow the law @xmath27 or @xmath28 ( @xmath29 , @xmath30 and @xmath31 are the constants ) given for the first time in @xcite for the transient statistics near the unstable - unstable pair bifurcation point . the analytical form of the distribution of the laminar phase lengths has not been reported anywhere hitherto for eyelet intermittency .
the theoretical explanation of the eyelet intermittency phenomenon is based on the boundary crisis of the synchronous attractor caused by the unstable - unstable bifurcation when the saddle periodic orbit and repeller periodic orbit join and disappear @xcite .
this type of the intermittent behavior has been observed both in the numerical calculations @xcite and experimental studies @xcite for the different nonlinear systems , including rssler oscillators .
although type - i intermittency with noise and eyelet intermittency taking place in the vicinity of the chaotic phase synchronization onset seem to be different phenomena , they are really the same type of the dynamics observed under different conditions .
the difference between these types of the intermittent behavior is only in the character of the external signal . in case of the type - i intermittency with noise the stochastic signal influences on the system , while in the case of eyelet intermittency the signal of chaotic dynamical system is used to drive the response chaotic oscillator . at the same time , the core mechanism governed the system behavior ( the motion in the vicinity of the bifurcation point disturbed by the stochastic or deterministic perturbations ) is the same in both cases . to emphasize the weak difference in the character of the driving signal we shall further use the terms `` type - i intermittency with noise '' and `` eyelet intermittency '' despite of the fact of the equivalence of these types of the intermittent behavior . indeed , the phenomena observed near the synchronization boundary for periodic systems whose motion is perturbed by noise ( in other words , the behavior in the vicinity of the saddle - node bifurcation perturbed by noise ) have been shown recently to be the same as for chaotic oscillators in the vicinity of the phase synchronization boundary @xcite .
thus , both for two coupled chaotic rssler systems and driven van der pol oscillator the same scenarios of the synchronous regime destruction have been revealed @xcite .
moreover , for two coupled rssler systems the behavior of the conditional lyapunov exponent in the vicinity of the onset of the phase synchronization regime is governed by the same laws as in the case of the driven van der pol oscillator in the presence of noise @xcite .
additionally , when the turbulent phase begins the phase trajectory demonstrates motion being close to periodic both for the eyelet intermittency observed in the vicinity of the phase synchronization boundary ( see @xcite ) and for type - i intermittency with noise .
finally , the repeller and saddle periodic orbits of the same period in the vicinity of the parameter region corresponding to the intermittent behavior tend to coalesce with each other ( see , e.g. @xcite ) for both these types of the intermittent behavior .
obviously , if the phenomena observed near the saddle - node bifurcation point for the systems whose motion is perturbed by noise are the same as for chaotic oscillators in the vicinity of the phase synchronization onset , one can expect that the intermittent behavior of two coupled chaotic oscillators near the phase synchronization boundary ( eyelet intermittency ) is also exactly the same as intermittency of type - i in the presence of noise in the supercritical region .
so , if type - i intermittency with noise and eyelet intermittency taking place in the vicinity of the chaotic phase synchronization onset are the same type of the system dynamics , the theoretical equations ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : eyeletlawmeanlength ] ) obtained for these types of the intermittent behavior are the approximate expressions being the different forms of eq .
( [ eq : typeilawmeanlength ] ) describing the dependence of the mean length of the laminar phases on the criticality parameter .
therefore , eq . ( [ eq : eyeletlawmeanlength ] ) can be deduced from eq .
( [ eq : typeilawmeanlengthapprox ] ) and vice versa . as a consequence , the coefficients @xmath32 , @xmath33 and @xmath29 , @xmath30 in ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : eyeletlawmeanlength ] )
are related with each other .
obviously , the mean length of the laminar phases must obey eq .
( [ eq : typeilawmeanlengthapprox ] ) and eq .
( [ eq : eyeletlawmeanlength ] ) simultaneously , independently whether the system behavior is classified as eyelet intermittency or type - i intermittency with noise . additionally , the laminar phase length distribution for the considered type of behavior must satisfy the exponential law ( [ eq : lampahselengthdistribution ] ) . the intermittent behavior under study is considered in the coupling strength range @xmath34 . in the case of the system driven by external noise ( type - i intermittency with noise ) the lower boundary value @xmath4 corresponds to the saddle - node bifurcation point when external noise is switched off .
for the dynamical systems demonstrating the chaotic behavior ( eyelet intermittency ) the lower boundary @xmath4 may be found , e.g. , in the way described in @xcite .
as far as the choice of the upper boundary value @xmath25 is concerned , this subject is discussed in detail in sec .
[ sct : boundary ] of this paper both for chaotic and stochastic external signals . to find the relationship between coefficients in ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : eyeletlawmeanlength ] ) we introduce the auxiliary variable @xmath35 and expand @xmath36 determined by eq .
( [ eq : typeilawmeanlengthapprox ] ) ( type - i intermittency with noise ) into taylor series in the vicinity of the point @xmath37 , where @xmath38 , i.e. @xmath39 having neglected the term @xmath40 in ( [ eq : teylorseries ] ) one can write eq .
( [ eq : teylorseries ] ) in the form @xmath41 having required @xmath42 we obtain , that @xmath43 and , therefore , equation ( [ eq : typeimodified ] ) describing the dependence of the mean length @xmath0 of the laminar phases on the criticality parameter for type - i intermittency with noise coincides exactly with eq .
( [ eq : eyeletlawmeanlength ] ) corresponding to eyelet intermittency .
correspondingly , in terms of eq .
( [ eq : c2coefficientvalue ] ) the relationship between coefficients @xmath32 , @xmath33 and @xmath29 , @xmath30 in ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : eyeletlawmeanlength ] ) is the following @xmath44 simulating the theoretical law ( [ eq : typeilawmeanlengthapprox ] ) and its approximation by the curve @xmath45 corresponding to law ( [ eq : eyeletlawmeanlength ] ) for eyelet intermittency .
the parameter values are @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 .
the points corresponding to @xmath51 are shown by symbols .
the theoretical law ( [ eq : eyeletlawmeanlength ] ) for eyelet intermittency is shown by the solid line ] fig .
[ fgr : twolaws ] illustrates the relationship of two theoretical laws ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : eyeletlawmeanlength ] ) in the region @xmath52 . here
function @xmath53 simulates the theoretical law ( [ eq : typeilawmeanlengthapprox ] ) ( the critical point is supposed to be @xmath54 ) , whereas the curve @xmath45 corresponds to law ( [ eq : eyeletlawmeanlength ] ) for eyelet intermittency .
the value of coefficients @xmath32 , @xmath33 , @xmath29 and @xmath30 have been selected according to eq .
( [ eq : coefficients ] ) .
one can see that in the region of the study both curves coincide with each other .
it means that the mean length of the laminar phases obeys eq .
( [ eq : typeilawmeanlengthapprox ] ) and eq .
( [ eq : eyeletlawmeanlength ] ) simultaneously , independently whether the system behavior is classified as eyelet intermittency or type - i intermittency with noise .
to confirm the concept of the equivalence of intermittencies being the subject of this study we consider several examples of the intermittent behavior classified both as eyelet intermittency taking place in the vicinity of the phase synchronization onset ( two coupled rssler systems ) and type - i intermittency with noise ( quadratic map and driven van der pol oscillator ) . as we have mentioned above , the intermittent behavior of two coupled chaotic oscillators in the vicinity of the phase synchronization boundary is classified traditionally as _ eyelet intermittency
_ @xcite .
nevertheless , the behavior of two coupled rssler oscillators close to the phase synchronization onset was considered from the point of view of type - i intermittency with noise for the first time in @xcite , whereas the same dynamics from the position of eyelet intermittency was studied in @xcite . according to different works the mean length of laminar phases happens to satisfy both eq .
( [ eq : typeilawmeanlengthapprox ] ) ( ref .
@xcite ) and eq .
( [ eq : eyeletlawmeanlength ] ) ( ref .
recently @xcite the distribution of the laminar phase lengths has been found to obey the exponential law ( [ eq : lampahselengthdistribution ] ) corresponding to type - i intermittency with noise . to give the complete picture we replicate the consideration of two coupled rssler systems near the onset of the phase synchronization regime for the different type of coupling between oscillators and another set of the control parameter values and
show that the observed intermittent behavior may be classified both as eyelet intermittency and type - i intermittency with noise .
the system under study is represented by a pair of unidirectionally coupled rssler systems , whose equations read as @xmath55 where @xmath56 [ @xmath57 are the cartesian coordinates of the drive [ the response ] oscillator , dots stand for temporal derivatives , and @xmath58 is a parameter ruling the coupling strength
. the other control parameters of eq .
( [ eq : roesslers ] ) have been set to @xmath59 , @xmath60 , @xmath61 , in analogy with our previous studies @xcite .
the @xmath62parameter ( representing the natural frequency of the response system ) has been selected to be @xmath63 ; the analogous parameter for the drive system has been fixed to @xmath64 .
for such a choice of the control parameter values , both chaotic attractors of the drive and response systems are phase coherent
. the instantaneous phase of the chaotic signals @xmath65 can be therefore introduced in the traditional way as the rotation angle @xmath66 on the projection plane @xmath67 of each system . )
are shown by symbols `` '' .
the theoretical laws ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) are shown by the solid lines .
( _ a _ ) _ eyelet intermittency : _ the dependence of @xmath36 on the parameter @xmath68 ; @xmath69 , @xmath70 , @xmath71 .
( _ b _ ) _ type - i intermittency with noise : _ the dependence of the mean laminar phase length @xmath0 on the parameter @xmath72 , with the ordinate axis being shown in the logarithmic scale ; @xmath73 , @xmath74 , @xmath75 in fig
. [ fgr : rsslrsgraph ] one and the same result of the numerical simulation of two coupled rssler systems ( [ eq : roesslers ] ) is shown in different ways to compare obtained data with the analytical predictions ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) for eyelet intermittency taking place near the phase synchronization boundary ( fig .
[ fgr : rsslrsgraph],_a _ ) and type - i intermittency with noise ( fig .
[ fgr : rsslrsgraph],_b _ ) , respectively .
the dependence of @xmath0 on @xmath76 is shown in the whole range of the coupling parameter strength values ( fig .
[ fgr : rsslrsgraph],_a _ ) to make evident the deviation of numerically obtained data from law ( [ eq : eyeletlawmeanlength ] ) far away from the onset of the phase synchronization .
the coupling strength @xmath58 plays the role of the control parameter .
the critical point @xmath77 relates to the onset of the phase synchronization regime in two coupled rssler systems .
the point @xmath78 used in ( [ eq : typeilawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) corresponds to the saddle - node bifurcation point if the chaotic dynamics being the analog of noise could be switched off .
the value of this point has been found from the dependence of the zero conditional lyapunov exponent on the coupling strength ( see for detail @xcite ) .
one can see , that the intermittent behavior of two coupled rssler systems may be treated both as eyelet and noised type - i intermittency with the excellent agreement between numerical data and theoretical curve in both cases . moreover , the coefficients @xmath29 , @xmath30 and @xmath32 , @xmath33 of the theoretical equations ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) agree very well with each other according to eq .
( [ eq : coefficients ] ) .
it allows us to state that both these effects are the same type of the system dynamics . nevertheless , to be totaly convinced of the correctness of our decision we have to consider other examples of the intermittent behavior classified traditionally ( contrary to the previous case of two coupled rssler systems ) as type - i intermittency with noise .
the second sample dynamical system to be considered is van der pol oscillator @xmath79 driven by the external harmonic signal with the amplitude @xmath80 and frequency @xmath81 with the added stochastic term @xmath82 .
the values of the control parameters have been selected as @xmath83 , @xmath84 .
for the selected values of the control parameters and @xmath85 the dynamics of the driven van der pol oscillator becomes synchronized when @xmath86 that corresponds to the saddle - node bifurcation .
the probability density of the random variable @xmath87 is @xmath88 where @xmath89 . to integrate eq .
( [ eq : drivenvdposcillatorandnoise ] ) the one - step euler method has been used with time step @xmath90 , the value of the noise intensity has been fixed as @xmath91 . )
are shown by symbols `` '' .
the theoretical laws ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) are shown by the solid lines .
( _ a _ ) _ eyelet intermittency : _ the dependence of @xmath36 on the parameter @xmath92 ; @xmath93 , @xmath94 , @xmath95 .
( _ b _ ) _ type - i intermittency with noise : _ the dependence of the mean laminar phase length @xmath0 on the parameter @xmath96 ; @xmath97 , @xmath98 , @xmath99 on the one hand , as it has been discussed above , the intermittent behavior in this case have to be classified as type - i intermittency with noise .
the corresponding dependence of the mean length of laminar phases on the criticality parameter @xmath100 is shown in fig .
[ fgr : vdpgraph],_b_. if the amplitude @xmath80 of the external signal exceeds the critical value @xmath101 the exponential law @xmath102 is expected to be observed . to make this law evident the abscissa in fig .
[ fgr : vdpgraph],_b _ has been selected in the @xmath103-scale and the ordinate axis @xmath0 is shown in the logarithmic scale .
one can see again the excellent agreement between the numerically calculated data and theoretical prediction ( [ eq : typeilawmeanlengthapprox ] ) .
the distribution of the lengths of the laminar phases @xmath104 obtained for @xmath105 also confirms the theoretical curve ( [ eq : lampahselengthdistribution ] ) , see fig . 7 in @xcite .
on the other hand , trying to choose the corresponding values of @xmath106 for the driven van der pol oscillator ( [ eq : drivenvdposcillatorandnoise ] ) one can find out that the intermittent behavior of this system also may be identified as eyelet intermittency . indeed , in fig .
[ fgr : vdpgraph],_a _ one can see a very good agreement between the numerically obtained mean length @xmath0 of the laminar phases for the different values of the coupling parameter and theoretical law ( [ eq : eyeletlawmeanlength ] ) corresponding to the eyelet intermittency .
note also , that for the well chosen values of @xmath106 the dependence @xmath107 in the axes @xmath108 behaves in the same way as the corresponding function @xmath109 in the axes @xmath110 for two coupled rssler systems ( [ eq : roesslers ] ) .
again , as well as for two coupled rssler oscillators , the coefficients @xmath29 , @xmath30 and @xmath32 , @xmath33 of the theoretical equations ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) agree very well with each other according to eq .
( [ eq : coefficients ] ) .
the next example is the quadratic map @xmath111 where the operation of `` @xmath112 '' is used to provide the return of the system in the vicinity of the point @xmath113 , @xmath114 and the probability density of the stochastic variable @xmath115 is distributed uniformly throughout the interval @xmath116}$ ] . if the intensity of noise @xmath8 is equal to zero the saddle - node bifurcation is observed for @xmath117 .
the intermittent behavior of type - i is observed for @xmath118 , whereas the stable fixed point takes place for @xmath119 .
having added the stochastic force ( @xmath120 ) in ( [ eq : quadraticmap ] ) we suppose that the intermittent behavior must be also observed in the area of the positive values of the criticality parameter @xmath3 , with the mean length of the laminar phases obeying law ( [ eq : typeilawmeanlengthapprox ] ) . )
are shown by symbols `` '' .
the theoretical laws ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) are shown by the solid lines .
( _ a _ ) _ eyelet intermittency : _ the dependence of @xmath36 on the parameter @xmath121 ; @xmath122 , @xmath123 , @xmath124 .
( _ b _ ) _ type - i intermittency with noise : _ the dependence of the mean laminar phase length @xmath0 on the parameter @xmath125 ; @xmath54 , @xmath126 , @xmath127 , @xmath128 although in this case we deal with type - i intermittency with noise , the numerically obtained points corresponding to the mean length of laminar phases are approximated successfully both by eq .
( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) ( see fig .
[ fgr : qmapgraph ] ) , with the coefficients @xmath29 , @xmath30 and @xmath32 , @xmath33 of the theoretical equations ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) agreeing with each other according to eq .
( [ eq : coefficients ] ) .
these findings confirm our statement about identity of the considered types of the intermittent behavior .
so , having studied the intermittent behavior of different systems which ( based on the prior knowledge ) should be classified either eyelet intermittency in the vicinity of the phase synchronization boundary or type - i intermittency with the presence of noise , we can conclude that the obtained characteristics are exactly the same in all cases described above .
two next sections are devoted to the consideration of another systems to give the additional proofs of the correctness of the introduced concept .
in this section we consider van der pol oscillator driven by the chaotic signal of rssler system @xmath129 where @xmath130 , @xmath131 , @xmath132 , @xmath83 , @xmath133 are the control parameters .
the auxiliary parameters @xmath134 and @xmath135 alter the characteristics ( the amplitude and main frequency ) of the chaotic signal influencing on van der pol oscillator . from the formal point of view the behavior of system ( [ eq : vdpunderrsslr ] ) can be classified neither eyelet intermittency nor type - i intermittency with noise .
indeed , since the response oscillator is periodic there are no unstable periodic orbits embedded into its attractor to be synchronized , therefore , the system dynamics can not be considered as eyelet intermittency . alternatively , due to the presence of chaotic perturbations there is no pure saddle - node bifurcation in this system to say about type - i intermittency .
nevertheless , it is intuitively clear that this example is nearly related to all cases considered above and one can expect to observe here the same type of intermittency as before . )
are shown by symbols `` '' .
the theoretical laws ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) are shown by the solid lines .
( _ a _ ) _ eyelet intermittency : _ the dependence of @xmath36 on the parameter @xmath121 ; @xmath136 , @xmath137 , @xmath138 .
( _ b _ ) _ type - i intermittency with noise : _ the dependence of the mean laminar phase length @xmath0 on the parameter @xmath139 ; @xmath140 , @xmath141 , @xmath142 fig .
[ fgr : vdpunderrsslrgraph ] makes this statement evident .
indeed , the numerically obtained data obey both laws ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) , with the coefficients @xmath29 , @xmath30 and @xmath32 , @xmath33 of the theoretical equations ( [ eq : eyeletlawmeanlength ] ) and ( [ eq : typeilawmeanlengthapprox ] ) agreeing with each other in accordance with eq .
( [ eq : coefficients ] ) . additionally , the distribution of the lengths of the laminar phases follows the exponential law @xcite , that allows us to say that we deal here with the same type of the dynamics as in the cases of quadratic map ( [ eq : quadraticmap ] ) , driven van der pol oscillator ( [ eq : drivenvdposcillatorandnoise ] ) and two coupled rssler systems ( [ eq : roesslers ] ) considered above .
all arguments given above may be considered as the evidence of the proposed statement on the equivalence of both types of the intermittent behavior . at the same time , one great difference between type - i intermittency with noise and eyelet intermittency taking place near the onset of the phase synchronization seems to exist .
this difference is connected with the upper boundary of the intermittent behavior and this point could refute the main statement of this manuscript . indeed , for type - i intermittency with noise in the supercritical region there is no an upper threshold ( see eq .
( [ eq : type - iintermittencypowerlaw ] ) ) and the intermittent behavior may be ( theoretically ) observed for arbitrary values of the criticality parameter @xmath143 , although the length of the laminar phases may be extremely long in this case , depending on the ratio between the criticality parameter value and the noise intensity .
alternatively , the existence of the boundary of the phase synchronization regime being the upper border of the eyelet intermittency is believed to be undeniable , since there is a great amount of works where the boundary of the phase synchronization had been observed and determined .
so , this circumstance along with the arguments given above in sections [ sct : relation][sct : numericalverification ] involve a seeming contradiction .
to resolve this disagreement we consider the probability @xmath144 to observe the turbulent phase in the time realization of the system demonstrating type - i intermittency with noise in the supercritical region @xmath5 during the observation interval with the length @xmath145 . the probability to detect the turbulent phase depends on the length @xmath145 of the observation interval and the length @xmath17 of the laminar phase being realized at the beginning of the system behavior examination .
obviously , if @xmath146 the turbulent phase is detected with the probability @xmath147 and , in turn , the probability for the laminar phase with length @xmath148 to be realized is @xmath149 where @xmath16 is given by eq .
( [ eq : lampahselengthdistribution ] ) . otherwise , when @xmath150 the probability to detect the turbulent phase is @xmath151 , whereas the laminar phase with the length @xmath17 takes place with the probability @xmath152 .
correspondingly , the probability to observe the turbulent phase is @xmath153 where @xmath154 is the incomplete gamma function , @xmath155 is the mean length of laminar phases depending on the criticality parameter @xmath3 and given by eq .
( [ eq : typeilawmeanlength ] ) . and corresponding to it the level curves with the step @xmath156 .
the level curves @xmath157 demarcating the regions with the high and low probabilities to detect the turbulent phase are shown by solid lines .
( _ a _ ) the theoretical expression ( [ eq : probabilitysurface ] ) , for the simplicity the values of the control parameters in eq .
( [ eq : typeilawmeanlength ] ) are taken @xmath91 , @xmath158 , @xmath54 .
( _ b _ ) the probability surface @xmath144 for the circle map with noise ( [ eq : circlemapseries ] ) .
( _ c _ ) the probability surface @xmath144 for two coupled rssler systems ( [ eq : roesslers ] ) ] the surface @xmath144 determined by the analytical expression ( [ eq : probabilitysurface ] ) and level curves corresponding to it are shown in fig .
[ fgr : probabilitysurfaces],_a_. it is clear that the probability to detect the turbulent phase for type - i intermittency with noise during one observation grows with the increase of the examination length @xmath145 but decreases when the criticality parameter @xmath159 is enlarged . obviously , if one examines ( experimentally or numerically ) the system behavior in the time interval with the length @xmath145 varying the control parameter @xmath3 , one observes the alternation of the laminar and turbulent phases for the relatively small values of the @xmath3-parameter , where @xmath144 is close to one , and only the laminar behavior for the relatively large ones , where @xmath144 is close to zero .
having no information about the kind of intermittency ( e.g. , when the experimental study of some system is carried out ) one can suppose the presence of the boundary separating two different types ( intermittent and steady ) of dynamics and , moreover , find a value @xmath25 corresponding to the `` onset '' of the laminar behavior . evidently , this `` boundary point '' would be correspond to the low probability @xmath144 , say , e.g. @xmath160 .
in addition , one can perform `` more careful '' measurements with the increased length @xmath161 of the observation to determine the value of the boundary point @xmath25 more precisely . in this case a new value @xmath162 would be obtained ( @xmath163 ) , with it being slightly larger than the previous one .
the schematic location of the `` boundary '' curve @xmath164 on the plane @xmath165 is shown in fig .
[ fgr : probabilitysurfaces],_a _ by the solid line .
it is clearly seen , that for the @xmath166-level the length @xmath145 grows extremely rapidly with the increase of the @xmath3-value . in other words , the major extensions of the observation interval @xmath145 result in the minor corrections of the `` boundary '' point @xmath25 .
since the resources of the both experimental and numerical studies are always limited , some final value @xmath25 with the maximal possible accuracy will be eventually found .
so , despite the fact , that for the type - i intermittency with noise in the supercritical region the turbulent phases can always be observed theoretically , from the practical point of view ( in the experimental studies or numerical calculations ) the boundary point @xmath25 exists , above which only the laminar behavior is observed .
moreover , with the further development of the experimental and computational resources the additional studies would result only in the insufficient increase of the boundary value . to illustrate the drawn conclusion we consider the circle map @xmath167 in the interval @xmath168 , where @xmath3 is the control parameter , @xmath169 , @xmath9 is supposed to be a delta - correlated gaussian white noise [ @xmath10 , @xmath11 .
if the intensity of noise @xmath8 is equal to zero , the saddle - node bifurcation is observed in ( [ eq : circlemap ] ) for @xmath170 , when the stable and unstable fixed points annihilate at @xmath113 .
obviously , for the selected value of the control parameter @xmath171 the evolution of system ( [ eq : circlemap ] ) in the vicinity of the bifurcation point may be reduced to the quadratic map @xmath172 allowing an easy comparison with the results given in the previous sections .
the intermittent behavior of type - i is observed for @xmath173 , whereas the stable fixed point takes place for @xmath174 . for the added stochastic force ( @xmath175 ) in circle map ( [ eq : circlemap ] ) the intermittent behavior
is also observed in the supercritical region of the criticality parameter @xmath3 , with the mean length @xmath0 of the laminar phases and the distribution @xmath16 of the laminar phase lengths @xmath17 obeying laws ( [ eq : typeilawmeanlengthapprox ] ) and ( [ eq : lampahselengthdistribution ] ) , respectively . the surface of the probability @xmath144 to observe the turbulent phase for the circle map ( [ eq : circlemap ] ) as well as the corresponding level curves
are shown in fig .
[ fgr : probabilitysurfaces],_b_. to obtain this surface we have made @xmath176 observations for every point taken with the steps @xmath177 , @xmath178 on the parameter plane @xmath165 . the probability @xmath144 was calculated as @xmath179 , where @xmath180 is the number of observations for which the turbulent phase has been detected .
one can see the excellent agreement between the results of numerical calculations and theoretical predictions ( compare fig .
[ fgr : probabilitysurfaces],_a _ and _ b _ ) . similarly , the analogous probability surface @xmath181 and the level curves shown in fig .
[ fgr : probabilitysurfaces],_c _ have been calculated for two coupled rssler systems ( [ eq : roesslers ] ) in the vicinity of the phase synchronization boundary , where eyelet intermittency is observed . in this case
@xmath182 observations have been made for every point to be examined , with these points being taken with the steps @xmath183 and @xmath184 on the plane @xmath185 .
it is easy to see that for the eyelet intermittency the probability surface as well as the level curves are exactly the same as for type - i intermittency with noise in the supercritical region . as a consequence
, we can draw a conclusion , that the eyelet intermittency taking place in the vicinity of the phase synchronization boundary and type - i intermittency with noise in the supercritical region are the same type of the dynamics observed under different conditions .
another consequence of the made consideration is the fact , that the phase synchronization boundary point can not be found absolutely exactly , since it separates the regions with the high and low probabilities to observe the phase slips in the coupled chaotic systems with the help of the experimental and computational resources existing at the moment of study . if someone , using a more powerful tools , tried to refine , say , the value of the coupling strength corresponding to the phase synchronization boundary reported in the earlier paper
, one would obtain a new value being close to the previous one , but larger .
exactly the same situation may be found , e.g. , in the work @xcite , where two mutually coupled rssler systems have been considered . in this work the refined boundary value @xmath186
is reported with the reference to the earlier work @xcite , where the value @xmath187 was given .
having considered two types of the intermittent behavior , namely eyelet intermittency taking place in the vicinity of the phase synchronization boundary and type - i intermittency with noise , supposed hitherto to be different , we have shown that these effects are the same type of the dynamics observed under different conditions . the analytical relation between coefficients of the theoretical equations corresponding to both types of the intermittent behavior has been obtained .
the difference between these types of the intermittent behavior is only in the character of the external signal . in case of the type - i intermittency
the stochastic signal influences on the system , while in the case of eyelet intermittency the signal of chaotic dynamical system is used to drive the response chaotic oscillator . at the same time , the core mechanism governed the system behavior as well as the characteristics of the system dynamics are the same in both cases .
we thank the referees for useful comments and remarks .
this work has been supported by federal special - purpose programme `` scientific and educational personnel of innovation russia ( 20092013 ) '' and the president program ( nsh-3407.2010.2 ) .
a. s. pikovsky , g. v. osipov , m. g. rosenblum , m. zaks , j. kurths , attractor repeller collision and eyelet intermittency at the transition to phase synchronization , phys .
79 ( 1 ) ( 1997 ) 4750 .
s. boccaletti , e. allaria , r. meucci , f. t. arecchi , experimental characterization of the transition to phase synchronization of chaotic @xmath188 laser systems , phys .
89 ( 19 ) ( 2002 ) 194101 .
a. e. hramov , a. a. koronovskii , m. k. kurovskaya , a. ovchinnikov , s. boccaletti , length distribution of laminar phases for type - i intermittency in the presence of noise , phys .
e 76 ( 2 ) ( 2007 ) 026206 . | in this article we compare the characteristics of two types of the intermittent behavior ( type - i intermittency in the presence of noise and eyelet intermittency taking place in the vicinity of the chaotic phase synchronization boundary ) supposed hitherto to be different phenomena .
we show that these effects are the same type of dynamics observed under different conditions .
the correctness of our conclusion is confirmed by the consideration of different sample systems , such as quadratic map , van der pol oscillator and rssler system .
consideration of the problem concerning the upper boundary of the intermittent behavior also confirms the validity of the statement on the equivalence of type - i intermittency in the presence of noise and eyelet intermittency observed in the onset of phase synchronization .
fluctuation phenomena , random processes , noise , synchronization , chaotic oscillators , dynamical system , intermittency 05.45.xt , 05.45.tp , 05.40.-a |
the antiferromagnetic properties of the one - dimensional spin @xmath1 xxz - model with hamiltonian : @xmath2\end{aligned}\ ] ] have been studied by analytical @xcite and numerical @xcite methods .
the critical exponents @xmath3 , which govern the large distance behavior of the spin - spin correlators in the groundstate : @xmath4 are given by:@xcite @xmath5 at finite temperature @xmath6 the spin - spin correlators decrease exponentially with a rate given by the inverse correlation length:@xcite @xmath7 with a critical exponent : @xmath8 independent of @xmath9 and : @xmath10 it has been shown in ref . that the structure factors : @xmath11 are most suited to extract the critical behavior from finite systems . in refs .
we have studied the static structure factors in the following three limits : @xmath12 @xmath13 @xmath14 the transverse and longitudinal structure factors have a common form in each of these limits : @xmath15 where : @xmath16 are the `` running variables '' .
the common form reflects the fact that the structure factors scale in the critical regime : @xmath17 if we keep fixed : @xmath18 the antiferromagnetic properties of the model become visible in the singularities of the structure factors in the limit ( [ ptn ] ) .
the transverse structure factors are infinite in this limit ; they develop a `` hard '' singularity .
in contrast , the longitudinal structure factors stay finite .
their critical behavior is hidden in subleading terms which produce a `` soft '' singularity .
away from the critical regime ( [ ptn ] ) the antiferromagnetic properties are lost .
there are further possibilities to destroy antiferromagnetic ordering
e.g. by a uniform external field @xmath19 or by frustration
i.e. by switching on a next - to - nearest neighbor interaction . in this paper
we are going to study the effect of a uniform magnetic field on the zero temperature structure factors .
for this purpose we have determined the static structure factors in the groundstates @xmath20 of the hamiltonian ( [ h ] ) at given total spin @xmath21 .
the groundstates were computed with a lanczos algorithm on rings with @xmath22 sites .
we will discuss the static structure factors as function of the magnetization @xmath23 .
the known @xcite magnetization curve translates @xmath24 into the fieldstrength @xmath19 . for convenience , we list in table i the @xmath19-field values corresponding to our @xmath24-values : .magnetization and the corresponding @xmath19-fields [ cols="^,^,^,^,^,^,^",options="header " , ] the outline of the paper is as follows . in sections
ii and iii we discuss the characteristic features of the longitudinal structure factors as there are the @xmath25 and @xmath24 dependence and the finite - size effects . the same is done for the transverse structure factors in sections iv and v.
the longitudinal structure factor of the xx - model ( @xmath26 ) is known from the exact solution obtained by niemeijer @xcite in the fermion representation : @xmath27 p_3(m ) & \text{for } \quad p_3(m)\leq p \leq \pi , \end{array } \right.\end{aligned}\ ] ] where @xmath28 though the result of ref .
has been derived for the thermodynamical limit @xmath29 , ( [ sf3gpid2 ] ) turns out to be correct for all system sizes with @xmath30 .
the linear behavior in @xmath25 has been found before @xcite for the case @xmath31 .
let us next turn to the isotropic case @xmath32 .
1 presents a panoramic view on the longitudinal structure factor with @xmath33 .
the emergence of the singularity at @xmath34 is clearly visible . along the momentum axis at @xmath31
we see the logarithmic singularity : @xmath35 discussed in ref . .
along the magnetization axis : @xmath36 we observe a logarithmic singularity in @xmath24 . the same type of singularity is also found in the limit : @xmath37 where @xmath38 is given in ( [ p3 m ] ) .
the longitudinal structure factor has its maximum at @xmath39 , @xmath24 fixed . with increasing strength of the uniform field the maximum position moves from @xmath40 to @xmath41
i.e. from antiferromagnetic to ferromagnetic order
. such a behavior has been conjectured by mller@xcite _
et al._. indications for this have been found also by parkinson and bonner@xcite and ishimura and shiba@xcite on small systems ( @xmath42 ) .
johnson and fowler@xcite were able to reformulate the isotropic heisenberg model for large spins and magnetizations close to saturation in terms of a gas of magnons . by accident their prediction for the longitudinal structure factor in the limit
@xmath43 is identical with the exact result ( [ sf3gpid2 ] ) for the xx - model . in fig .
2 we compare the longitudinal structure factors for @xmath32 and @xmath26 at @xmath44 and @xmath45 . for @xmath45
the structure factors almost coincide . for smaller @xmath24 values
, however , the isotropic structure factor deviates from ( [ sf3gpid2 ] ) .
the cusp along the line ( [ p3 m ] ) becomes more pronounced with decreasing values of @xmath24 .
looking at the finite - size effects which will be analyzed in the next section we find an @xmath46 behavior away from the cusp and a slower decrease @xmath47 with @xmath48 at the cusp @xmath39 .
the change in the finite - size behavior signals the emergence of a nonanalytic behavior in the thermodynamical limit .
we have also determined the longitudinal structure factors for the anisotropies @xmath49 .
the @xmath50 dependence of the longitudinal structure factor looks similar to the isotropic case . instead of an infinity
we find a peak at @xmath34 . in the limit @xmath51 the structure factor
is adequately described by ( [ sjgya ] ) with @xmath52 . in the limit @xmath53
we find again a behavior of the form ( [ sjgya ] ) with a running variable : @xmath54 the appearence of the cusp along the line @xmath39 is indeed independent of the anisotropy parameter @xmath9 . for increasing values of @xmath9
the cusp is less pronounced .
in the critical regime @xmath55 we expect the longitudinal structure factors to obey finite - size scaling : @xmath56 in the combined limit ( [ limitpnmt ] ) we have to keep fixed @xmath57 defined in ( [ z1z2]) and : @xmath58 in ref .
we checked finite - size scaling in the combined limit : @xmath59 it was found that finite - size scaling works for the transverse structure factors at all @xmath60 values . in contrast , finite - size scaling breaks down for the longitudinal structure factors , if @xmath61 . in this section
we are going to study consequences of the ansatz ( [ s3g3 ] ) in the limits ( [ pp0 ] ) and ( [ pp3 m ] ) at zero - temperature .
the behavior of the isotropic structure factor can be read of ( [ sjgya ] ) : @xmath62 where @xmath63 stands for the two limits ( [ pp0],[pp3 m ] ) .
the isotropic structure factors on finite systems with @xmath64 sites scale in @xmath24 ( for @xmath65 ) and are linear in @xmath66 in the limit @xmath67 , as can be seen from fig .
3 . finite - size effects are small for @xmath68 but rather large for @xmath69 .
we have studied the finite - size effects in the differences : @xmath70 at fixed magnetizations : @xmath71 \frac{1}{4 } & : \quad n=4,8,12,16,20,24,28 \\[0.1 cm ] \frac{1}{6 } , \frac{2}{6 } & : \quad n=6,12,18,24 , \end{array } \right.\end{aligned}\ ] ] which can be realized on the systems with size @xmath64 .
the @xmath64-dependence of the difference ( [ djmn ] ) can be parametrized by : @xmath72 in our finite - size analysis we have included as well the magnetizations : @xmath73 \frac{1}{12},\frac{5}{12 } & : \quad n=12,24\\[0.1 cm ] \frac{1}{14},\frac{2}{14},\frac{3}{14 } , \frac{4}{14},\frac{5}{14},\frac{6}{14 } & : \quad n=14,28 , \end{array}\right.\end{aligned}\ ] ] which occur on two systems . here
we have assumed that the finite - size dependence is described correctly by ( [ d0d3 ] ) .
the extrapolations of the structure factors to the thermodynamical limit are represented by the the solid dots in fig .
so far our estimates of the thermodynamical limit are restricted to magnetizations @xmath74 due to the finiteness of our systems @xmath75 .
additional information on the structure factors for @xmath24-values : @xmath76 closer to the critical point @xmath31 can be obtained if finite - size scaling ( [ s3g3 ] ) holds for the longitudinal structure factors in the limits ( [ pp0],[pp3 m ] ) where we keep fixed @xmath77 .
this variable is just @xmath78 for the sequence ( [ mdn ] ) , and we can compute the scaling function at this value from : @xmath79 in this way we get from the finite - size scaling ansatz ( [ limitpnmt ] ) in the limits ( [ pp0],[pp3 m ] ) an estimate of the thermodynamical limit of the structure factors : @xmath80 for the sequence of @xmath24 values in ( [ mdn ] ) . the result for the differences ( [ d0d3 ] )
is marked by the open dots in figs .
4(a),(b ) .
we have repeated the finite - size analysis described above for the isotropic case for : @xmath81 . the resulting estimate of the thermodynamical limit : @xmath82 versus the variable : @xmath83 is represented in figs .
4(a),(b ) by the triangles and squares , respectively .
the most remarkable property of the transverse structure factor is its approximate constancy : @xmath84 which has been found by mller@xcite _
et al . _ on small systems for the isotropic case @xmath32 . for @xmath85 ( [ s1gpmt ] ) can be easily proven to be exact by means of the bethe ansatz solution . for @xmath86 and @xmath87
, however , ( [ s1gpmt ] ) is not exact . as an example we present in figs .
5(a),(b ) the momentum dependence of @xmath88 at @xmath89 for @xmath26 and @xmath32 , respectively .
the constancy in the regime @xmath90 is striking .
deviations from ( [ s1gpmt ] ) can be seen on a scale magnified by a factor 100 in the inset of figs .
5(a),(b ) .
these deviations follow a single scaling curve which increases monotonically with momentum @xmath25 . on the scaling curve
finite - size effects die out with @xmath91 . at @xmath92 , however , significant finite - size effects of the order @xmath93 with @xmath94 for @xmath95 and @xmath96 for @xmath26 become apparent . beyond the regime ( [ s1gpmt ] )
the transverse structure factor of the xx - model ( @xmath26 ) is linear in @xmath97 , as can be seen from fig .
therefore we find the same type of singularity for @xmath98 at @xmath99 .
in contrast to the naive expectation antiferromagnetic order is not destroyed in the transverse structure factors by an external field . in the isotropic case ( @xmath32 )
the transverse structure factor is approximately linear in @xmath100 for @xmath101 , as can be seen from fig .
this type of singularity was found for @xmath98 at @xmath31 .
the slight curvature at @xmath44 might indicate that the type of the singularity has changed here to a power behavior @xmath102 .
in the anisotropic case with @xmath103 and @xmath44 we find again the constant behavior ( [ s1gpmt ] ) and a linear increase in @xmath104 . again this type of singularity follows from ( [ sjgya ] ) with @xmath52 and @xmath31 .
let us start with the xx - model ( @xmath26 ) . fig .
6(a ) shows the transverse structure factor for @xmath40 and @xmath105 as function of @xmath24 .
in contrast to the longitudinal case we have no scaling in @xmath24 . at @xmath31 we know from ( [ sjgya ] ) with @xmath106 that the transverse structure factors diverge as @xmath107 .
the same type of divergence appears at @xmath44 , as can be seen from fig .
here we compare the @xmath64-dependence of @xmath108 for @xmath31 and @xmath44 . in both cases there
is a linear increase in @xmath107 .
let us now turn to the isotropic case ( @xmath32 ) . here
the longitudinal and transverse structure factors coincide provided that there is no external field . in the presence of a uniform field , however , they differ drastically . at @xmath40 ,
the longitudinal structure factor is infinite at @xmath31 , but becomes finite and monotonically decreasing for @xmath109 .
in contrast the transverse structure factor stays at infinity for @xmath109 .
moreover , on finite systems , the longitudinal structure factor scales for @xmath109 as was demonstrated in fig .
3 whereas the corresponding transverse structure factor ( at @xmath40 ) does not scale at all , as can be seen from fig .
more surprising , at fixed @xmath109 not too large , the transverse structure factors increase with the system size @xmath64 stronger than at @xmath31 .
a comparison of the @xmath64-dependence at @xmath31 and @xmath44 is shown in fig .
7(b ) . at @xmath44
the increase with @xmath110 is definitely steeper , which signals a strengthening of the singularity at @xmath40 .
note also that there are deviations from linearity in @xmath110 , which increase with @xmath24 .
this might indicate a change from a logarithmic behavior at @xmath31 to a power behavior for @xmath109 . in ref .
mller _ et al .
_ reported on the transverse structure factor for @xmath111 ; they found already that the dominant mode remains situated at @xmath40 independent of the field .
6(b ) tells us that the strengthening of the singularity at @xmath40 becomes more and more pronounced with increasing system size @xmath64 .
in the presence of a uniform external field in @xmath112-direction the static structure factors of the xxz - model show up the following features : 1 .
the longitudinal structure factors have a cusp along the line ( [ p3 m ] ) . in case of the xx - model ( @xmath26 ) there are no finite - size effects and the longitudinal structure factor is given by ( [ sf3gpid2 ] ) for all system sizes with @xmath113 . for smaller values of @xmath9 and @xmath24 ,
the cusp becomes sharper .
finite - size effects decrease along the cusp with @xmath47 , @xmath114 . away from the cusp we find a more rapid decrease of the order @xmath91 .
the longitudinal structure factor is finite for @xmath115 and @xmath116 , but develops a logarithmic singularity in this limit for the isotropic case ( @xmath32 ) .
this means : a uniform field * weakens * the antiferromagnetic order in the * longitudinal * structure factor for @xmath32 .
the transverse structure factor is almost contant for @xmath117 [ cf.([s1gpmt ] ) ] .
finite - size effects die out slowly with @xmath93 with @xmath118 along the line @xmath92 , but rapidly with @xmath91 away from this line .
4 . in the limit
@xmath119 fixed , we observe a singularity of the type @xmath97 in the transverse structure factor for @xmath26 . in the isotropic case ( @xmath32 ) this singularity appears to be stronger than @xmath100 .
this means : a weak uniform field * strengthens * the antiferromagnetic order in the * transverse * strucure factor for @xmath32 .
therefore the effect of a uniform external field on the longitudinal and transverse structure factors for ( @xmath32 ) is similar to the effect of switching on the anisotropy parameter @xmath9 .
the logarithmic singularity found in the isotropic structure factor at @xmath40 changes with @xmath9 .
it is strengthened in the transverse , but weakened in the longitudinal structure factor . * the longitudinal structure factor versus momentum @xmath25 and magnetization @xmath24 with a ridge along the line @xmath120 , for @xmath33 .
* comparison of the longitudinal structure factors at @xmath26 [ eq . ( [ sf3gpid2 ] ) ] and @xmath32 for @xmath121 , respectively . * scaling of the longitudinal structure factors @xmath122 with @xmath68 ( open symbols ) , @xmath69 ( solid symbols ) , with the magnetization @xmath24 .
* estimate of the thermodynamical limit for the difference @xmath123 [ eq . ( [ djmn ] ) ] at momentum @xmath68 : solid symbols represent results from the finite - size analysis ( [ d0d3 ] ) , open symbols are results from finite - size scaling ( [ s3inf ] ) . * same as fig .
4(a ) for the difference @xmath124 at the cusp @xmath120 . * the momentum dependence of the transverse structure factor of the xx model @xmath125 at fixed magnetizations @xmath126 , respectively .
the inset shows a magnification of the low momentum regime . * same as fig .
5(a ) for the transverse structure factor in the isotropic case @xmath127 . the inset shows a magnification of the low momentum regime . * the transverse structure factor of the xx model @xmath125 versus magnetization @xmath24 at momentum @xmath40 . * same as fig .
6(a ) for the isotropic model @xmath127 . *
comparison of the size dependence of the transverse structure factors of the xx model @xmath125 at @xmath40 : @xmath128 , respectively . * same as fig .
7(a ) for the isotropic model @xmath127 . 99 e - mail : [email protected] e - mail : [email protected] e - mail : [email protected] c. n. yang and c. p.
yang , phys . rev . * 150 * , 321 , 327 ( 1966 ) ; 151 , 258 ( 1966 ) ; r.baxter , _ exactly solved models in statistical mechanics _ ( academic , new york 1980 ) ; k .- j .-
b . lee and p. schlottmann , j. phys .
: condens .
matter * 1 * , 2759 ( 1989 ) ; k .- j .-
lee , j. phys . : condens .
matter * 3 * , 1679 ( 1991 ) .
bonner and m. fisher , phys . rev .
* 135 * , a640 ( 1964 ) ; r.b .
griffiths , phys .
rev . * 133 * , a768 ( 1964 ) ; m. takahashi , prog .
* 51 * , 1348 ( 1974 ) ; k. fabricius , u. lw , k .- h .
mtter and p. ueberholz , phys .
rev . b*44 * , 7476 ( 1991 ) ; k. fabricius , m. karbach , u. lw and k .- h .
mtter , phys .
rev . b*45 * , 5315 ( 1992 ) ; k. fabricius , u. lw and k .- h .
mtter , z. phys .
b*91 * , 51 ( 1993 ) .
m. karbach and k .- h .
mtter , z. phys .
b * 90 * , 83 ( 1993 ) ; m. karbach , k .- h .
mtter , and m. schmidt , to be published in phys .
b. k. fabricius , u. lw and k .- h .
mtter , submitted to phys .
b. a. luther and i. peschel , phys . rev . *
b 12 * , 3908 ( 1975 ) ; h.c .
fogedby , j. phys .
c * 11 * , 4767 ( 1978 ) ; v.e .
korepin and a.g .
izergin , commun . math .
phys . * 94 * , 67 ( 1984 ) ; * 99 * , 271 ( 1985 ) ; n.m .
bogoliubov , a.g .
izergin , and v.e .
korepin , nucl .
b * 275 * , 687 ( 1986 ) .
j. suzuki , y. akutsu , and m. wadati , j. phys .
jpn . * 59 * , 2667 ( 1990 ) ; j. suzuki , t. nagao , and m. wadati , int .
b*6 * , 1119 ( 1992 ) .
m. karbach , thesis university of wuppertal wub - dis 94 - 6 .
niemeijer , physica , * 36 * , 377 ( 1967 ) .
g. mller , h. thomas , h. beck and j.c .
bonner , phys .
b*24 * , 1429 ( 1981 ) .
j. b. parkinson and j. c. bonner , phys .
rev . b*32 * , 4703 ( 1985 ) .
n. ishimura and h. shiba , progr .
phys . * 64 * , 479 ( 1980 ) ; m. d. johnson and m. fowler , phys .
b*34 * , 1728 ( 1986 ) . | the static structure factors of the xxz model in the presence of uniform field are determined from an exact computation of the groundstates at given total spin on rings with @xmath0 sites .
in contrast to the naive expectation a weak uniform field strengthens the antiferromagnetic order in the transverse structure factor for the isotropic case . |
there has been great progress in classifying the complexity of counting problems recently .
one important success is for counting constraint satisfaction problems ( # csp ) , where a sweeping complexity dichotomy is proved @xcite .
while the landscape of exact counting becomes clearer , the complexity of approximate counting remains mysterious .
two typical classes of problems have been identified : 1 ) those that have a fully polynomial - time randomized approximation scheme ( fpras ) , and 2 ) those that are # sat - hard with respect to approximation preserving reductions ( ap - reductions ) @xcite . if @xmath0 then # sat admits no fpras @xcite , and therefore neither does any problem in the second class .
these two classes are analogous to p - time tractable vs. np - hard decision or optimization problems .
interestingly , in approximate counting , there has emerged a third distinct class of natural problems , which seems to be of intermediate complexity .
it is conjectured @xcite that the problems in this class do not have an fpras but that they are not as hard as # sat to approximate .
a canonical problem in this class has been identified , which is to count the number of independent sets in a bipartite graph ( # bis ) . despite many attempts , nobody has found an fpras for # bis or an ap - reduction from # sat to # bis .
the conjecture is that neither exists .
mossel et al .
@xcite showed that the gibbs sampler for sampling independent sets in bipartite graphs mixes slowly even on bipartite graphs of degree at most @xmath1 .
another interesting attempted markov chain for # bisby ge and tefankovi @xcite was also shown later to be slowly mixing by goldberg and jerrum @xcite . # bis plays an important role in classifying counting problems with respect to approximation .
a trichotomy theorem is shown for the complexity of approximately solving unweighted boolean counting csps , where in addition to problems that are solvable by fprases and those that are ap - reducible from # sat , there is the intermediate class of problems which are equivalent to # bis @xcite .
many counting problems are shown to be # bis - hard and hence are conjectured to have no fpras @xcite , including estimating the partition function of the the ferromagnetic potts model @xcite . moreover , under ap - reductions
# bis is complete in a logically defined class of problems , called # rh@xmath2 , to which an increasing variety of problems have been shown to belong . other typical complete problems in #
rh@xmath2 include counting the number of downsets in a partially ordered set @xcite and computing the partition function of the ferromagnetic ising model with local external fields @xcite .
the problem of counting independent sets ( # is ) can be viewed as a special case in the general framework of spin systems , which originated from statistical physics to model interactions between neighbors on graphs . in this paper , we focus on two - state spin systems .
in general such a system is parameterized by edge weights @xmath3 and a vertex weight @xmath4 .
an instance is a graph @xmath5 .
a configuration @xmath6 is a mapping @xmath7 from vertices to ( two ) spins .
the weight @xmath8 of a configuration @xmath6 is given by @xmath9 where @xmath10 is the number of @xmath11 edges given by the configuration @xmath6 , @xmath12 is the number of @xmath13 edges , and @xmath14 is the number of vertices assigned @xmath15 .
we are interested in computing the partition function , which is defined by @xmath16 the partition function is the normalizing factor of the gibbs distribution , which is the distribution in which a configuration @xmath6 is drawn with probability @xmath17 . the spin system is called _ ferromagnetic _ if @xmath18 and _ antiferromagnetic _ if @xmath19 . in particular ,
when @xmath20 , such a system is the famous _ ising model _ , and when @xmath21 and @xmath22 , it is the _ hard - core gas model _ , the partition function of which counts independent sets when @xmath23 .
the external field @xmath24 is typically referred to as the activity or fugacity of the hard - core model . approximating the partition function of the hard - core model
is especially well studied .
we now know that the complexity transition from easy to hard corresponds exactly to the uniqueness of the gibbs measure in infinite @xmath25-ary trees @xmath26 ( for details of these notions , see @xcite ) .
notice that @xmath25-ary trees are graphs of maximum degree @xmath27 , hence our use of the notation @xmath26 . on the algorithmic side
, weitz presented a fully polynomial - time approximation scheme ( fptas ) for the hard - core gas model on graphs of maximum degree @xmath27 when uniqueness holds @xcite . on the other hand ,
sly showed that the approximation problem is # sat - hard for a small interval beyond the uniqueness threshold @xcite .
building on their work , it is now established that for all antiferromagnetic 2-spin systems there is an fptas for graphs of maximum degree @xmath27 up to the uniqueness threshold @xcite ( see also @xcite ) , whereas non - uniqueness implies # sat - hardness under ap - reductions on @xmath27-regular graphs @xcite ( see also @xcite ) .
the only place that remains unclear is exactly at the uniqueness threshold .
a key feature of spin systems in the antiferromagnetic non - uniqueness region is the ability to support a gadget with many vertices whose spins are highly correlated with the phase of the gadget ( which is either @xmath28 or @xmath29 ) , but are nearly independent among themselves conditioned on that phase .
such a gadget was used by sly @xcite to show inapproximability of the partition function of the hard - core model when @xmath24 is just above the uniqueness threshold .
a different gadget with similar properties was used by goldberg et al .
@xcite to show inapproximability on a planar graph when @xmath24 is much larger .
we abstract this notion of nearly - independent phase - correlated spins .
it is this feature that enables us to reduce from # sat to approximating the partition function of antiferromagnetic two - spin systems in the non - uniqueness region .
restricted to bipartite graphs , it appears that supporting nearly - independent phase - correlated spins alone is not enough to imply # bis - hardness .
it was shown that sly s gadget is applicable to the antiferromagnetic ising model without an external field by galanis et al .
however , such a system has an fpras on bipartite graphs .
the reason is that this system is perfectly symmetric on bipartite graphs and therefore can be translated into a ferromagnetic ising system , whose partition function can be approximated using the fpras of jerrum and sinclair @xcite ( see corollary [ cor : ising : w / o ] in section [ sec : sym ] for details ) . to get around this perfectly symmetric case
, we introduce the second key concept called unary symmetry breaking .
unary symmetry breaking is about the availability of non - trivial
unary weights " which can be built in the model .
the availability of unary weights is relevant to counting complexity , for example in the context of counting csps .
unary symmetry breaking simply means that some ( any ) non - trivial unary weight can be built .
formal definitions of the two notions nearly - independent phase - correlated spins and unary symmetry breaking
can be found in section [ sec : def ] .
our main technical theorem is the following .
suppose a tuple of parameters @xmath30 with @xmath31 and @xmath32 supports nearly - independent phase - correlated spins and unary symmetry breaking
. then the partition function ( [ eq : pfdef ] ) of two - spin systems @xmath33 is # bis - hard to approximate on bipartite graphs with maximum degree @xmath27 .
[ thm : main ] previous hardness proofs for the problem # is and for the problem of estimating the partition function of antiferromagnetic 2-spin systems typically reduce from max - cut or from the problem of counting certain types of cuts @xcite .
however such a technique sheds little light in the bipartite setting as cut problems are trivial on bipartite graphs .
reductions between # bis - equivalent problems typically involve transformations that `` blow up '' vertices and edges into sets of vertices that are completely connected , so they do not apply to bounded - degree graphs either .
a key property of sly s gadget is that either phase occurs with probability bounded below by an inverse polynomial
. this bound is sufficient in sly s setting to reduce from max - cut , but it is not enough to construct ap - reductions for our use .
we resolve this issue by introducing a balancing construction .
the construction takes two copies of a gadget with nearly - independent phase - correlated spins , and produces a new gadget with similarly - correlated spins , but in the new gadget the two phases occur with probability close to @xmath34 .
the proof of theorem [ thm : main ] utilizes an intermediate problem , that is , computing the partition function of antiferromagnetic ising systems with non - uniform external fields on bipartite graphs .
a non - uniform external field means that the instance specifies a subset of vertices on which the external field acts .
a 2-spin system with a non - uniform external field is very similar to a boolean # csp with one binary symmetric non - negative valued function ( corresponding to edge weights ) and one unary non - negative valued function ( corresponding to vertex weights ) ( see , for example @xcite ) .
our reduction implements an external field , and this is where unary symmetry breaking comes into play .
as discussed earlier , the partition function of the ising model without an external field has an fpras , so the unary symmetry breaking gadget seems necessary . in fact , we show ( lemma [ lem : sym : break ] ) that unary symmetry breaking holds for all 2-spin systems except for the ising model without an external field or degenerate systems ( i.e. , systems satisfying @xmath35 ) .
we also prove ( lemma [ lem : indpt - phases ] ) that all antiferromagnetic 2-spin systems support nearly - independent phase - correlated spins in the non - uniqueness region .
taking lemmas [ lem : indpt - phases ] and [ lem : sym : break ] together with theorem [ thm : main ] yields our main result : [ thm : bis - nonuniq ] for all tuples of parameters @xmath30 with @xmath36 and @xmath19 , except for the case @xmath37 , if the infinite @xmath27-regular tree @xmath26 is in the non - uniqueness region then approximating the partition function ( [ eq : pfdef ] ) on bipartite graphs with maximum degree @xmath27 is # bis - equivalent under ap - reductions .
there is an fpras for the exceptional case ( @xmath38 ) ( corollary [ cor : ising : w / o ] ) .
let us now survey the approximability picture that this theorem helps establish . for general antiferromagnetic 2-spin models with soft constraints (
i.e. , @xmath39 ) , non - uniqueness holds if and only if @xmath40 and @xmath41 for some critical values @xmath42 and @xmath43 depending on @xmath44 , @xmath45 , and @xmath27 ( see ( * ? ? ? * lemma 5 ) ) .
hence , for all @xmath46 where @xmath19 , and all @xmath36 the following holds : 1 .
if @xmath47 , for all @xmath24 , there is an fptas to approximate the partition function for @xmath27-regular graphs @xcite ( this extends to graphs of maximum degree @xmath27 in an appropriate sense , see @xcite for details ) .
2 . if @xmath40 , then there exists @xmath48 so that : 1 . for all @xmath49 $ ] , there is an fptas to approximate the partition function for @xmath27-regular graphs @xcite ( this again extends in an appropriate sense to graphs of maximum degree @xmath27 @xcite ) .
2 . for all @xmath50 ,
it is # sat - hard to approximate the partition function on @xmath27-regular graphs @xcite .
3 . for all @xmath50 ,
it is # bis - hard to approximate the partition function on bipartite graphs of maximum degree @xmath27 ( theorem [ thm : bis - nonuniq ] in this paper ) . for the particular case of the hard - core model the critical value ( i.e. , critical activity @xmath51 )
is more easily stated . for the hard - core model ( i.e. , @xmath52 and @xmath53 ) kelly @xcite showed that non - uniqueness holds on @xmath26 if and only if @xmath54 . as a consequence we get the following corollary for the hard - core model . for all @xmath36 , all @xmath55 , it is # bis - hard to approximate the partition function of the hard - core model on bipartite graphs of maximum degree @xmath27 .
[ cor : hardcore ] we also get a corollary concerning the more general partition function as long as @xmath44 and @xmath45 are less than @xmath15 and the degree bound @xmath27 is sufficiently large . for and
@xmath44 and @xmath45 satisfying @xmath56 and any @xmath4 , there exists a @xmath27 such that @xmath33 is in the non - uniqueness region of @xmath26 ( * ? ? ?
* lemma 21.2 ) .
this implies the following corollary . for every @xmath56 and @xmath4 ,
there exists a @xmath27 such that the 2-spin system with parameters @xmath57 and with uniform or non - uniform external field @xmath24 on bipartite graphs with degree bound @xmath27 is # bis - equivalent under ap - reductions , except when @xmath20 and @xmath23 , in which case it has an fpras .
[ cor : general - graphs ] more generally , for antiferromagnetic 2-spin systems we get the following picture for the complexity of approximating the partition function on general graphs . as usual
there is a difficulty classifying the complexity of approximating the partition function at the boundary between uniqueness and non - uniqueness . to address this issue , for parameters @xmath30 ,
@xcite define a notion of _ up - to-@xmath27 unique _ which is equivalent to the parameters lying in the interior of the uniqueness region for the infinite @xmath58-ary tree @xmath59 for all @xmath60 ( see definition 7 in @xcite ) .
moreover , the parameters @xmath33 satisfy _ @xmath61-strict - uniqueness _ if it is up - to-@xmath61 unique .- strict - uniqueness is called universally unique in ( * ? ? ?
* definition 7 ) ) . ] on the other side , we say the parameters @xmath33 satisfy _ @xmath61-non - uniqueness _ if for some @xmath36 the tree @xmath26 has non - uniqueness .
the only gap between the notions of @xmath61-strict - uniqueness and @xmath61-non - uniqueness is the case when the parameters @xmath33 are at the uniqueness / non - uniqueness threshold of @xmath26 for some @xmath27 .
the following result detailing the complexity for general graphs is now established .
[ cor : dichotomy ] for all tuples of parameters @xmath33 with @xmath19 , the following holds : 1 .
if the parameters satisfy @xmath61-strict - uniqueness then there is a fptas for the partition function for all graphs ( * ? ? ?
* theorem 2 ) .
if the parameters satisfy @xmath61-non - uniqueness then : 1 .
it is # sat - hard to approximate the partition function on graphs @xcite .
2 . except when @xmath20 and @xmath23
, it is # bis - hard to approximate the partition function on bipartite graphs ( theorem [ thm : bis - nonuniq ] in this paper ) . a recent paper of liu et al .
@xcite shows that our theorem [ thm : main ] can also be used to analyse the complexity of _ ferromagnetic _ partition functions ( where @xmath62 ) . in particular , it uses theorem [ thm : main ] to show # bis - hardness for approximating the partition function for ferromagnetic 2-spin systems when @xmath63 for sufficiently large external field @xmath24 . an interesting problem that remains open is to prove # bis - hardness for the entire non - uniqueness region for ferromagnetic 2-spin systems with @xmath63 .
the remainder of the paper is organized as follows .
we begin by formally defining ap - reductions and the problems that we use in our reductions in section [ sec : prelim ] . in section [ sec : def ] we formally define nearly - independent phase - correlated spins and unary symmetry breaking , and present the main lemmas concerning these concepts .
we prove theorem [ thm : main ] in section [ sec : proof - main ] .
finally , in sections [ sec : proof - balanced - gadget ] , [ sec : sym ] , and [ sec : indpt - phases ] we prove the various lemmas about the two key concepts that are presented in section [ sec : def ] .
we are interested in the complexity of approximate counting .
let @xmath64 be a finite alphabet .
we want to approximate the value of a function @xmath65 . a _
randomized approximation scheme _ is an algorithm that takes an instance @xmath66 and a rational error tolerance @xmath67 as inputs , and outputs a rational number @xmath68 such that , for every @xmath69 and @xmath70 , @xmath71\geq\frac{3}{4}.\ ] ] a _ fully polynomial randomized approximation scheme _ ( fpras ) is a randomized approximation scheme which runs in time bounded by a polynomial in @xmath72 and @xmath73 .
note that the quantity @xmath74 can be changed to any value in the interval @xmath75 or even @xmath76 for a problem of size @xmath77 without changing the set of problems that have fully polynomial randomized approximation schemes since the higher accuracy can be achieved with only polynomial delay by taking a majority vote of multiple samples .
dyer _ et al . _
@xcite introduced the notion of approximation - preserving reductions .
suppose @xmath78 and @xmath79 are two functions from @xmath80 to @xmath81 . an _ approximation - preserving reduction _
( ap - reduction ) from @xmath78 to @xmath79 is a randomized algorithm @xmath82 to approximate @xmath78 using an oracle for @xmath79 .
the algorithm @xmath82 takes an input @xmath83 , and satisfies the following three conditions : ( i ) every oracle call made by @xmath82 is of the form @xmath84 , where @xmath85 is an instance of @xmath79 , and @xmath86 is an error bound satisfying @xmath87 ; ( ii ) the algorithm @xmath82 meets the specification for being a randomized approximation scheme for @xmath78 whenever the oracle meets the specification for being a randomized approximation scheme for @xmath79 ; ( iii ) the run - time of @xmath82 is polynomial in @xmath72 and @xmath73 . if an ap - reduction from @xmath78 to @xmath79 exists , we write @xmath78@xmath88@xmath79 , and say that @xmath78 is _ ap - reducible _ to @xmath79 .
if @xmath78@xmath88@xmath79 and @xmath79@xmath88@xmath78 , then we say that @xmath78 and @xmath79 are _ ap - interreducible _ or _ ap - equivalent _ , and write @xmath78@xmath89@xmath79 .
the problem # bis is defined as follows .
* # bis . + * instance . * a bipartite graph @xmath90 . + * output . *
the number of independent sets in @xmath90 .
+ in this paper , we are interested in 2-spin systems over bounded degree bipartite graphs parametrized by a tuple @xmath33 .
we say a real number @xmath68 is _ efficiently approximable _ if there is an fpras for the problem of computing @xmath68 . throughout the paper we only deal with non - negative real parameters that are efficiently approximable . for efficiently approximable non - negative real numbers @xmath91 and a positive integer @xmath27
, we define the problem of computing the partition function of the 2-spin system @xmath92 with external field @xmath24 on bipartite graphs of bounded degree @xmath27 , as follows .
* bi-(m-)2-spin@xmath30 .
+ * instance . * a bipartite ( multi)graph @xmath93 with degree bound @xmath27 . +
* output . *
the quantity @xmath94 notice that we also introduced a multigraph version of the same problem .
it will be useful later .
we drop the parameter @xmath27 when there is no degree bound , that is , bi-2-spin@xmath33 is the same as bi-2-spin@xmath95 .
we also found the notion of non - uniform external field useful in the reductions .
the following problems are introduced as intermediate problems .
we also introduce a multigraph version , but as intermediate problems we do not need the bounded degree variant . + * name . *
+ * instance . * a bipartite ( multi)graph @xmath93 and a subset @xmath96 . +
* output . * the quantity @xmath97
in this section we define two key concepts : nearly - independent phase - correlated spins and unary symmetry breaking . we first describe the basic setup of a certain gadget . for positive integers @xmath27 , @xmath98 and @xmath77 where @xmath77 is even and
is at least @xmath99 , let @xmath100 and @xmath101 be disjoint vertex sets of size @xmath98 and let @xmath102 be a size-@xmath103 superset of @xmath100 and @xmath104 be a size-@xmath103 superset of @xmath101 which is disjoint from @xmath102 .
let @xmath105 and @xmath106 .
let @xmath107 be the set of bipartite graphs with vertex partition @xmath108 in which every vertex has degree at most @xmath27 and every vertex in @xmath109 has degree at most @xmath110 .
we refer to the vertices in @xmath109 as `` terminals '' .
vertices in @xmath101 are `` positive terminals '' and vertices in @xmath100 are `` negative terminals '' .
when the gadget @xmath111 is drawn from @xmath107 , we use the notation @xmath112 to refer to the set of terminals .
each configuration @xmath113 is assigned a unique phase @xmath114 .
roughly in our applications of the definitions below the phase of a configuration @xmath6 is @xmath115 if @xmath116 contains more vertices with spin @xmath15 than does @xmath117 .
we define measures @xmath118 and @xmath119 .
fix some @xmath120 . for any positive integer @xmath98 , *
@xmath118 is the distribution on configurations @xmath121 such that , for every @xmath122 , @xmath123 independently with probability @xmath124 and , for every @xmath125 , @xmath123 independently with probability @xmath126 , and * @xmath119 is the distribution on configurations @xmath121 such that , for every @xmath127 , @xmath123 independently with probability @xmath124 and , for every @xmath122 , @xmath123 independently with probability @xmath126 . to give a rough sense for the values @xmath128 and @xmath129 they will correspond to the marginal probabilities of the root of an infinite tree obtained by taking limits of finite trees with appropriate boundary conditions , see section [ sec : indpt - phases ] for more details . to prove the # bis - hardness
we need a gadget where the spins of the terminals are drawn from distributions close to @xmath118 or @xmath119 conditioned on the phase @xmath28 or @xmath29 .
the following definition formalises this notion .
the definition will be crucial for obtaining our @xmath130-hardness results for antiferromagnetic models in the non - uniqueness region . while the definition is stated for general 2-spin systems ,
it is most useful for antiferromagnetic systems ; a ferromagnetic 2-spin system is not going to satisfy the conditions of the definition since one expects that for such systems @xmath131 or , equivalently , there is no notion of a binary phase " .
[ def : sly ] a tuple of parameters @xmath30 * supports nearly - independent phase - correlated spins * if there are efficiently - approximable values @xmath132 such that the following is true
. there are functions @xmath133 , @xmath134 , and @xmath135 ,
each of which is bounded from above by a polynomial in @xmath98 and @xmath136 , and for every @xmath98 and @xmath70 there is a distribution on graphs in @xmath137 such that a gadget @xmath5 with terminals @xmath109 can be drawn from the distribution within @xmath134 time , and the probability that the following inequalities hold is at least @xmath138 : 1 .
the phases are roughly balanced , i.e. , @xmath139 2 . for a configuration @xmath140 and any @xmath121 , @xmath141 in fact , given a gadget with the above property
, one can construct a gadget where the phases are ( nearly ) uniformly distributed as detailed in the following definition .
[ def : balsly ] we say that the tuple of parameters @xmath30 supports * balanced * nearly - independent phase - correlated spins if definition [ def : sly ] holds with replaced by : @xmath142 where @xmath143 is quantified as in definition [ def : sly ] .
the following lemma shows that for essentially all 2-spin systems , definition [ def : sly ] implies definition [ def : balsly ] .
hence , in the proof of theorem [ thm : main ] we assume the existence of a gadget with balanced phases .
[ lem : gettingbal ] if the parameter tuple @xmath30 with @xmath144 supports nearly - independent phase - correlated spins , then it supports balanced nearly - independent phase - correlated spins .
the main technical result for proving # bis - hardness for 2-spin antiferromagnetic systems in the tree non - uniqueness region is the following lemma , which is proved in section [ sec : indpt - phases ] .
[ lem : indpt - phases ] for all @xmath36 , all @xmath145 where @xmath146 , if the infinite @xmath27-regular tree @xmath26 is in the non - uniqueness region then the tuple of parameters @xmath30 supports balanced nearly - independent phase - correlated spins .
the second property of the gadget is the notion of unary symmetry breaking which is relatively simple .
[ def : symmetry ] we say that a tuple of parameters @xmath30 * supports unary symmetry breaking * if there is a bipartite graph @xmath147 whose vertices have degree at most @xmath27 which has a distinguished degree-@xmath15 vertex @xmath148 such that @xmath149
. we will prove in section [ sec : sym ] that unary symmetry breaking holds for all 2-spin models except for two cases .
assume @xmath150 .
the parameters @xmath30 support unary symmetry breaking unless ( i ) @xmath35 or ( ii ) @xmath20 and @xmath23 .
[ lem : sym : break ]
in this section we prove theorem [ thm : main ] .
we first show how the two notions of `` nearly - independent phase - correlated spins '' and `` unary symmetry breaking '' lead to # bis - hardness .
the goal of this section is to show that it is # bis - hard to approximate the partition function of antiferromagnetic ising models with non - uniform non - trivial external fields on bipartite graphs . for any @xmath151 , @xmath4 and @xmath152 , [ lem :
bis : ising ] by flipping @xmath153 to @xmath15 and @xmath15 to @xmath153 for each configuration @xmath6 , we see that bi - m - nonuniform-2-spin@xmath154 is in fact the same as bi - m - nonuniform-2-spin@xmath155 .
hence we may assume @xmath156 .
let @xmath157 and @xmath158 note that @xmath159 and @xmath160 , so @xmath161 .
let @xmath93 be an input to @xmath130 with @xmath162 and @xmath163 .
let @xmath164 be the number of independent sets of @xmath90 .
let @xmath70 be the desired accuracy of the reduction .
we will construct an instance @xmath165 with a specified vertex subset @xmath166 for bi - m - nonuniform-2-spin@xmath154 such that @xmath167 where @xmath168 is a quantity that is easy to approximate .
therefore it suffices to call oracle bi - m - nonuniform-2-spin@xmath154 on @xmath169 with the specified subset @xmath170 with accuracy @xmath171 and approximate @xmath168 within @xmath171 .
the construction of @xmath169 involves two positive integers @xmath172 and @xmath173 .
let @xmath172 be the least positive integer such that @xmath174 note that @xmath172 depends on @xmath77 and @xmath70 and there is a polynomial @xmath175 in @xmath77 and @xmath136 such that @xmath176 .
let @xmath173 be the least positive integer depending on @xmath77 , @xmath70 and @xmath172 such that @xmath177 once again , @xmath173 is bounded from above by a polynomial in @xmath77 and @xmath136 . given the integers @xmath172 and @xmath173 , the graph @xmath169 is constructed as follows ( see figure [ fig : toy ] for an illustrative diagram ) .
let @xmath178 for each @xmath179 where @xmath180 is the degree of @xmath181 in @xmath90 .
let @xmath182 for any @xmath179 and @xmath183 .
let @xmath184 the vertex set of @xmath169 is @xmath185 .
note that @xmath186 and @xmath187 .
we add @xmath172 parallel edges in @xmath169 between @xmath188 and @xmath181 for each @xmath189 and add edges between @xmath181 and every vertex in @xmath190 , and between @xmath191 and every vertex in @xmath192 for each @xmath179 and @xmath183 .
formally the edge set of @xmath169 is @xmath193 where @xmath194 denotes a disjoint union as a multiset of @xmath172 copies of @xmath195 , @xmath196 and @xmath197 for each @xmath181 and @xmath198 .
let @xmath199 and @xmath200 for each @xmath201 , let @xmath8 be the contribution to @xmath202 of configurations that are consistent with @xmath6 .
first consider configurations @xmath6 such that @xmath203 for all @xmath204 .
denote by @xmath64 the set of all such configurations on @xmath205 .
then for @xmath206 , we have @xmath207 to see , note that @xmath208 separates the vertices in @xmath170 from the rest of the vertices of @xmath169 . since @xmath6 specifies the spins of all vertices in @xmath208 , the weight @xmath8 is the product of : ( i ) the weights induced on edges that do not involve vertices in @xmath170 , and ( ii ) the weights induced on edges with endpoints in @xmath170 . item ( i ) is just @xmath209 . for item ( ii ) , we need to sum over all possible assignments on @xmath170 ; the factor @xmath210 gives the value of this sum as can be seen by expanding the expression @xmath211 let @xmath212 be the subset of configurations which induce an independent set on the vertices @xmath213 and @xmath214 be its contribution to @xmath202 .
let @xmath215 and @xmath216 be its contribution . if @xmath217 then @xmath218 . otherwise , @xmath219 .
it implies @xmath220 since @xmath172 satisfies eq .
( [ eqn : t1 ] ) .
next consider configurations @xmath6 on @xmath205 such that @xmath221 for at least one @xmath204 .
denote this set by @xmath222 and its contribution by @xmath223 .
then for @xmath224 , @xmath225 it implies @xmath226 since @xmath227 and @xmath173 satisfies eq .
( [ eqn : t2 ] ) . by eq .
( [ eqn : sigma ] ) and eq .
( [ eqn : sigmaprime ] ) we have @xmath228 and clearly @xmath229 .
it is also clear that @xmath168 can be approximated accurately enough given fprases for @xmath24 and @xmath230 .
this finishes our proof . in this section
we prove the following lemma .
suppose @xmath44 , @xmath45 and @xmath24 are efficiently approximable reals satisfying @xmath3 , @xmath4 and @xmath31 .
suppose that @xmath27 is either an integer that is at least @xmath231 or @xmath232 ( indicating that we do not have a degree bound ) .
if @xmath30 supports nearly - independent phase - correlated spins and unary symmetry breaking , then there exist efficiently approximable @xmath151 and @xmath233 such that @xmath234 and [ lem : main ] we prove the antiferromagnetic case first , that is , @xmath19 .
@xmath230 and @xmath235 are chosen as follows .
recall that @xmath236 and @xmath237 .
. then @xmath239 . therefore @xmath240 and
let @xmath241 .
moreover , suppose @xmath147 is the unary symmetry breaking gadget with distinguished vertex @xmath148 .
let @xmath242 where @xmath243 denote @xmath244 for spin @xmath245 and @xmath246 .
let @xmath247 , and @xmath248 .
it is easy to verify that @xmath234 as @xmath249 by the unary symmetry breaking assumption . given @xmath250 and a bipartite multigraph @xmath93 with a subset @xmath96 where @xmath251 , @xmath252 , and @xmath253 , our reduction first constructs a bipartite graph @xmath169 with degree at most @xmath27 .
the construction of @xmath169 involves a gadget @xmath111 .
since @xmath30 supports nearly - independent phase - correlated spins , by lemma [ lem : gettingbal ] @xmath30 also supports balanced nearly - independent phase - correlated spins . therefore we draw @xmath254 such that eq . (
[ eqn : balance ] ) and eq .
( [ eqn : near : ind ] ) hold with probability at least @xmath138 , where @xmath255 and @xmath256 .
assume @xmath111 satisfies them and otherwise the reduction fails .
we will construct @xmath169 such that @xmath257 where we use the abbreviated expressions @xmath258 , @xmath259 , and @xmath260 .
the lemma follows by one oracle call for @xmath261 with accuracy @xmath262 , one oracle call for @xmath263 with accuracy @xmath264 , and an approximation of other terms in the denominator with accuracy @xmath262 using fprases for @xmath126 , @xmath124 , @xmath44 , @xmath45 and @xmath24 .
the graph @xmath169 is constructed as follows . for each vertex
@xmath179 we introduce a copy of @xmath111 , denoted by @xmath265 with vertex set @xmath266 .
moreover , for each vertex @xmath267 we introduce a copy of @xmath147 , denoted by @xmath268 . whenever a terminal vertex is used in the construction once ,
we say it is occupied . for each @xmath189 , we connect one currently unoccupied positive ( and respectively negative ) terminal of @xmath269 to one currently unoccupied positive ( and respectively negative ) terminal of @xmath265 .
denote by @xmath270 all these edges between terminals .
for each @xmath267 , we identify an unoccupied positive terminal of @xmath269 with the distinguished vertex @xmath271 of @xmath268 .
we denote this terminal by @xmath272 .
the resulting graph is @xmath169 .
it is clear that @xmath169 is bipartite and has bounded degree @xmath27 .
let @xmath273 be a configuration of the phases of the @xmath265 s .
let @xmath274 be the contribution to @xmath261 from the configurations @xmath6 that are consistent with @xmath275 in the sense that , for each @xmath179 , @xmath276 .
then @xmath277 . let @xmath109 be the set of all terminals @xmath278 and @xmath121 be a configuration on @xmath109 .
let @xmath279 be the configuration @xmath280 restricted to @xmath281 .
recall that for @xmath282 , @xmath283 is the contribution to @xmath284 from configurations that have phase @xmath115 and are consistent with @xmath279 . also , @xmath285 moreover , for each @xmath267 and each spin @xmath245 , let @xmath286 be the contribution to @xmath287 from configurations @xmath6 with @xmath288 .
hence , @xmath289 we express @xmath274 as @xmath290 where @xmath291 is the contribution of edges in @xmath270 given configuration @xmath280 .
notice that we divide the last factor by @xmath24 when @xmath292 because we counted the vertex weight twice in that case .
define @xmath293 to be an approximation version of the partition function where on each @xmath281 the spins are chosen exactly according to @xmath294 .
that is , @xmath295 let @xmath296 .
( [ eqn : near : ind ] ) implies that @xmath297 and @xmath293 are close , that is , @xmath298 moreover eq . ( [ eqn : balance ] ) implies that @xmath299 notice that here @xmath284 is the same for any @xmath179 as the @xmath265 s are identical copies of @xmath111 . then we calculate the following quantity given @xmath275 @xmath300 as the measure @xmath294 is i.i.d .
, we may count the weight of each edge in @xmath270 independently .
notice that @xmath301 is the edge contribution when one end point is chosen with probability @xmath302 and the other @xmath303 . for an edge @xmath304 ,
if @xmath188 and @xmath181 are assigned the same phase @xmath28 , then an edge in @xmath270 connecting one @xmath28 terminal of @xmath269 and one @xmath28 terminal of @xmath265 gives a weight of @xmath305 and an edge connecting two @xmath29 terminals gives @xmath306 .
the total weight is @xmath307 .
similarly if @xmath188 and @xmath181 are assigned the same phase @xmath29 , the total weight is @xmath308 as well . on the other hand
if @xmath188 and @xmath181 are assigned distinct phases @xmath28 and @xmath29 , the total weight is @xmath309 .
recall that @xmath310 .
moreover , for each @xmath267 , if @xmath311 , then the contribution of @xmath268 is @xmath312 and otherwise @xmath313 .
notice that here @xmath287 is the same for any @xmath267 as the @xmath268 s are identical copies of @xmath147 .
recall that @xmath314 .
plug these calculation into eq .
( [ eqn : tildez ] ) , we have @xmath315 where @xmath316 denotes the number of edges of which the two endpoints are of the same phase given @xmath275 , and @xmath317 denotes the number of vertices in @xmath170 that are assigned @xmath28 given @xmath275 . apply eq.([eqn : ratio : balance ] ) to eq.([eqn : tildezsig ] ) , @xmath318 then we sum over @xmath275 in eq .
( [ eqn : tildez : sig ] ) , @xmath319 however notice that @xmath320 by just mapping @xmath29 to @xmath153 and @xmath28 to @xmath15 in each configuration @xmath275 .
combine eq.([eqn : ratio : z : tildez ] ) , and eq.([eqn : tildezb ] ) , @xmath321 recall that @xmath256 and we get the desired bounds .
the other case is ferromagnetic , that is , @xmath18 .
notice that in this case @xmath322 , so we choose @xmath323 and @xmath235 to be the same as the antiferromagnetic case .
the construction of @xmath169 is similar to the previous case , with the following change . for each @xmath189 , we connect one unoccupied positive terminal of @xmath269 to one unoccupied negative terminal of @xmath265 , and vice versa .
the rest of the construction is the same . with this change ,
given a configuration @xmath324 , if two endpoints are assigned the same spin , the contribution is @xmath325 and otherwise @xmath326 .
therefore the effective edge weight is @xmath159 when the spins are the same , after normalizing the weight to @xmath15 when the spins are distinct .
the rest of the proof is the same .
we can now conclude the proof of theorem [ thm : main ] .
# bis - hardness in theorem [ thm : main ] follows directly from lemma [ lem : bis : ising ] and lemma [ lem : main ] .
the other direction , # bis - easiness , follows fairly directly from theorem 47 of @xcite ( the full version of @xcite ) .
an edge in the instance graph can be viewed as a constraint of arity 2 . if @xmath327 , then the constraint on the edge is `` weakly log - supermodular '' and the vertex weight can be viewed as a unary constraint , which is taken as given in a `` conservative '' csp . if @xmath328 , then reverse the interpretation of @xmath153 and @xmath15 on one side of the bipartition of the instance graph , so that the effective interaction along an edge is given by the matrix @xmath329 .
this constraint is also `` weakly log - supermodular '' since @xmath330 . after the reversing
there are two vertex weights @xmath24 and @xmath331 , which are also allowed for `` conservative '' csp instances .
in this section we prove lemma [ lem : gettingbal ] that a gadget with nearly - independent phases can be used to construct a gadget with balanced phases . before proving the lemma
, we introduce some notation .
let @xmath332 let @xmath129 and @xmath128 be the quantities from definition [ def : sly ] and let @xmath333 the two columns of @xmath334 correspond to spin @xmath153 and spin @xmath15 .
the first row corresponds to the distribution induced on a positive terminal from @xmath335 and the second to the distribution induced from @xmath336 .
similarly the first row also corresponds to the distribution induced on a negative terminal from @xmath336 and the second to the distribution induced from @xmath335 .
notice that @xmath337 .
when the parameters @xmath44 , @xmath45 and @xmath24 are clear , we sometimes make the notation more concise , referring to the partition function as @xmath263 rather than as @xmath338 .
also , given a configuration @xmath339 and a subset @xmath340 of @xmath341 , we often use the notation @xmath342 to denote the restriction @xmath343 . for
a gadget @xmath111 drawn from @xmath137 , let @xmath344 be the contribution of phase @xmath345 to the partition function @xmath263 .
moreover , for a subset @xmath346 , suppose @xmath347 is a configuration on terminals in @xmath340 .
let @xmath348 be the contribution of configurations that are consistent with @xmath349 and belong to phase @xmath350 , that is , @xmath351 where @xmath8 is the weight of configuration @xmath6 defined in ( [ ourdefw ] ) .
it is easy to see that for @xmath345 , @xmath352 and @xmath353 we are now prepared to prove the lemma which is the focus of this section .
let @xmath70 satisfy @xmath250 . by assumption we may draw a gadget @xmath111 from @xmath354 such that it satisfies eq .
( [ eqn : rough : balance ] ) and eq .
( [ eqn : near : ind ] ) with probability at least 3/4 , where @xmath355 and @xmath356 will be specified later .
assume @xmath111 does
. otherwise the construction fails .
we consider first the antiferromagnetic case @xmath19 .
we construct a gadget @xmath357 such that @xmath357 satisfies eq .
( [ eqn : balance ] ) and eq .
( [ eqn : near : ind ] ) .
we make two copies of @xmath111 , say @xmath358 and @xmath359 .
let the terminals of @xmath360 be @xmath361 for @xmath362 .
for each @xmath345 , we add a set of edges that form a perfect matching between @xmath355 terminals in @xmath363 and @xmath355 terminals in @xmath364 .
denote by @xmath365 the edges of the two perfect matchings.@xmath357 is the resulting graph .
denote by @xmath366 the vertices of @xmath360 that are endpoints of @xmath365 .
the terminals of @xmath357 are those @xmath99 terminal nodes in @xmath367 that are still unmatched , that is @xmath368 for @xmath345 , and @xmath369 .
denote by @xmath370 the terminals of @xmath359 that are unmatched .
we define the phase of @xmath357 to be the phase of @xmath358 , that is , @xmath357 is said to have phase @xmath28 or @xmath29 if and only if @xmath358 has the same @xmath28 or @xmath29 phase regardless of the phase of @xmath359 .
let @xmath371 be a vector denoting the phases of @xmath358 and @xmath359 where @xmath372 .
then the @xmath28 phase of @xmath357 corresponds to the vector @xmath373 and the @xmath29 phase corresponds to @xmath374 . for two configurations @xmath375 and @xmath376 with @xmath377 , let @xmath378 be the joint configuration on @xmath379 .
then we have the following : @xmath380 where @xmath381 denote the contribution from edges of @xmath365 given configurations @xmath382 and @xmath383 , and @xmath384 .
moreover by eq .
( [ eqn : near : ind ] ) , for @xmath362 and any subset @xmath385 , we have @xmath386 where we have used @xmath387 to denote the probability that the configuration on terminals in @xmath340 is @xmath349 in the distribution @xmath388 . therefore by eq .
( [ eqn : zk : tau ] ) , @xmath389 we need to calculate the quantity @xmath390 recall our definitions of @xmath391 and @xmath334 .
let @xmath392 where @xmath393 means transposition . then @xmath394 .
we write @xmath395 and let @xmath396 . here
@xmath301 is the edge contribution when one end point is chosen with probability @xmath302 and the other @xmath303 .
also notice that each edge is independent under @xmath397 so we can count them separately .
then the quantity in eq.([eqn : mudef ] ) is @xmath398 plug eq .
( [ eqn : mu ] ) in eq .
( [ eqn : zk+ ] ) , @xmath399 summing over @xmath400 in eq .
( [ eqn : zk+:tau : upper ] ) we get @xmath401 similarly we get an estimate for @xmath402 : @xmath403 let @xmath404 .
notice that @xmath405 as @xmath358 and @xmath359 are identical copies .
combine eq .
( [ eqn : zk+:upper ] ) and eq .
( [ eqn : zk-:lower ] ) , @xmath406 by eq .
( [ eqn : rough : balance ] ) there is @xmath407 such that @xmath408 and @xmath407 is bounded above by a polynomial in @xmath409 and @xmath410 . to show eq .
( [ eqn : balance ] ) it suffices to show @xmath411 and @xmath412 .
clearly @xmath411 can be achieved by picking @xmath413 and @xmath414 in eq .
( [ eqn : zk : ratio ] ) . to show @xmath412 is similar and therefore omitted .
establishing that eq .
( [ eqn : near : ind ] ) still holds is easy to show
. we will show @xmath415 and the other bounds are similar . by an argument similar to eq .
( [ eqn : zk+ ] ) , we have @xmath416 moreover , summing over @xmath400 in eq .
( [ eqn : zk+ ] ) we get : @xmath417 the desired bound follows as @xmath413 .
the other case is ferromagnetic , that is , @xmath18 .
we construct @xmath357 in the same way as in the previous case , but with the following change .
to form the perfect matching @xmath365 , we match @xmath418 terminals of @xmath358 to @xmath419 terminals of @xmath359 .
the proof goes similarly but @xmath420 .
however since we made a twist in connecting @xmath358 and @xmath359 , it follows that @xmath421 therefore we continue with the new constant @xmath422 and the rest of the proof is the same .
in this section we prove lemma [ lem : sym : break ] that almost all 2-spin models support unary symmetry breaking .
consider the sequence of gadgets @xmath423 , defined as follows .
the vertex set of @xmath424 is @xmath425 , and @xmath188 is considered the attachment vertex .
the edge set of @xmath424 is @xmath426 we shall argue that if the first three graphs in the sequence , namely @xmath427 , @xmath428 and @xmath429 , all fail to be symmetry breakers then one of conditions ( i ) or ( ii ) holds . the graph @xmath427 has an isolated vertex that could clearly be removed ; we leave it in to make the calculations uniform .
note that the maximum degree of any vertex in these graphs is @xmath231 .
let @xmath430 , @xmath431 and @xmath432 .
then the effective weight of vertex @xmath188 is given by the column vector , @xmath433 where @xmath434 for @xmath435 to be a symmetry breaker we require the vector @xmath436 not to be a multiple of @xmath437 .
suppose @xmath438 , @xmath439 and @xmath440 all fail to be symmetry breakers .
then they must all lie in a one - dimensional subspace of @xmath441 .
one way this can happen is if the matrix @xmath109 is rank 1 , i.e. , if @xmath35 .
this is case ( i ) .
otherwise , @xmath442 and @xmath443 and @xmath444 lie in a one - dimensional subspace , namely the one generated by @xmath445 .
this implies @xmath446 and @xmath447 , or recasting , @xmath448 so either @xmath449 , or ( dividing the second equation by the first ) @xmath450 and @xmath23 . in either case
, substituting for @xmath451 , @xmath452 and @xmath453 in terms of @xmath44 , @xmath45 , @xmath24 , we obtain either @xmath454 ( which belongs to case ( i ) ) or @xmath20 and @xmath23 ( which is case ( ii ) ) .
there are two exceptional cases .
the first is @xmath35 .
it is well - known that in this case the 2-spin system can be decomposed and hence tractable .
the other case of @xmath20 and @xmath23 is perfectly symmetric and this symmetry can not be broken .
this system is the ising model without external fields . for this system ,
the marginal probability of any vertex in any graph is exactly @xmath34 .
regarding the ferromagnetic case , jerrum and sinclair @xcite presented an fpras for the ferromagnetic ising model with consistent external fields . on the other hand , anti - ferromagnetic ising models without external field on bipartite graphs are actually equivalent to ferromagnetic ising models .
the trick is to flip the assignment on only one part of the bipartition , which has been used before by goldberg and jerrum @xcite . for @xmath151 , bi-2-spin@xmath455 @xmath456 bi-2-spin@xmath457 .
[ lem : ising : w / o ] let @xmath458 be a bipartite graph where @xmath459 and @xmath460 are two partitions of vertices .
let @xmath252 .
then we have @xmath461 where in the second line we flip the assignment of @xmath462 . combining lemma [ lem : ising : w / o ] with the fpras by jerrum and sinclair @xcite for the ferromagnetic ising model yields the following corollary . for any @xmath463
, bi-2-spin@xmath455 has an fpras .
[ cor : ising : w / o ] as explained in the introduction , this corollary helps explain why the notion of unary symmetry breaking is necessary to achieve # bis - hardness .
in this section we prove lemma [ lem : indpt - phases ] that for 2-spin antiferromagnetic systems in the tree non - uniqueness region there is a gadget with nearly - independent phase - correlated spins .
recall that the gibbs distribution is the distribution in which a configuration @xmath6 is drawn with probability @xmath464 let @xmath465 be the infinite @xmath25-ary tree .
a _ gibbs measure _ on @xmath465 is a measure such that for any finite subtree @xmath466 , the induced distribution on @xmath109 conditioned on the outer boundary is the same as that given by ( [ eqn : gibbs ] ) .
there may be one or more gibbs measures ( see , e.g. , @xcite for more details ) .
a gibbs measure is called _ translation - invariant _ if it is invariant under all automorphisms of @xmath465 , and is _ semi - translation - invariant _ if it is invariant under all parity - preserving automorphisms of @xmath465 . in our context , the gibbs measures that will be of interest are the two extremal semi - translation - invariant gibbs measures corresponding to the all @xmath15 s and all @xmath153 s boundary conditions .
these two measures are different in the non - uniqueness region of @xmath465 .
let @xmath467 be the two marginal probabilities of the root having spin @xmath15 in the two extremal semi - translation - invariant gibbs measures .
more precisely , one can define @xmath468 as follows . for @xmath469 ,
let @xmath470 denote the probability that the root is assigned spin @xmath15 in the @xmath25-ary tree of depth @xmath471 in the gibbs distribution where the leaves are fixed to spin @xmath472 . in standard terminology , fixing the configuration on the leaves to all 1 s or all 0 s is most commonly referred to as the @xmath473 boundary conditions , respectively ( and hence the indices @xmath473 in our notation of @xmath468 ) .
it is not hard to show that @xmath474 is decreasing in @xmath471 , while @xmath475 is increasing .
let @xmath476 and let @xmath477 .
the two quantities @xmath124 and @xmath126 satisfy the standard tree recursion in the following sense .
let @xmath478 for @xmath345 , and @xmath479 . then @xmath480 sly ( @xcite , theorem 2.1 ) showed that for every @xmath36 , there exists @xmath481 such that the hard - core model @xmath482 supports nearly - independent phase - correlated spins for any @xmath24 satisfying @xmath483 .
this region is a small interval just above the uniqueness threshold . in the same paper sly also showed that @xmath484 supports nearly - independent phase - correlated spins .
the quantities defining measures @xmath397 in definition [ def : sly ] are exactly the marginal probabilities @xmath485 on @xmath486 .
galanis et al .
@xcite extended the applicable region of sly s gadget for the hard - core model to all @xmath487 for @xmath488 and @xmath489 .
the gap of @xmath490 and @xmath491 is later closed @xcite . in the later paper galanis et al .
@xcite also verified sly s gadget for parameters @xmath30 such that @xmath19 , @xmath492 , and @xmath33 is in the non - uniqueness region of infinite tree @xmath486 . using @xcite we extend the applicable region of sly s gadget to the entire non - uniqueness region for all 2-spin antiferromagnetic models .
sly s construction is based on a technical analysis of the configurations with the largest contribution to the partition function of a random bipartite @xmath27-regular graph .
this required a second moment argument involving a optimization which posed technical difficulties .
following recent developments in @xcite , we describe how to obtain the properties of sly s gadget for all 2-spin antiferromagnetic systems ( including the case with an external field ) in the non - uniqueness region .
we will give an overview of the second moment approach and how to tackle it using the technique in @xcite .
we begin by presenting the relevant definitions and describing the construction of the gadget . 1 .
@xmath494 is supported on bipartite graphs .
the two parts of the bipartite graph are labelled by @xmath473 and each is partitioned as @xmath495 where @xmath496 , @xmath497 for @xmath498 .
@xmath170 denotes the set @xmath499 and similarly @xmath208 denotes the set @xmath500 .
2 . to sample @xmath501 , sample uniformly and independently @xmath27 matchings : ( i ) @xmath25 random perfect matchings between @xmath502 and @xmath503 , ( ii ) a random perfect matching between @xmath504 and @xmath505 .
the edge set of @xmath111 is the union of the @xmath27 matchings .
thus , vertices in @xmath170 have degree @xmath27 , while vertices in @xmath208 have degree @xmath110 .
the case @xmath506 will also be important , in which case we denote the distribution as @xmath507 .
note that @xmath507 is supported on bipartite @xmath27-regular graphs .
strictly speaking , @xmath508 are supported on multi - graphs , but it is well known that every statement that holds asymptotically almost surely on these spaces continues to hold asymptotically almost surely conditioned on the event that the graph is simple .
we are now ready to give the construction of the gadget . for
constants @xmath509 , let @xmath510 .
note that @xmath511 .
first , sample @xmath111 from the distribution @xmath512 conditioning on @xmath111 being simple .
next , for @xmath513 , attach @xmath98 disjoint @xmath25-ary trees of even depth @xmath471 ( with @xmath514 and @xmath515 ) to @xmath516 , so that every vertex in @xmath516 is a leaf of exactly one tree ( this is possible since @xmath517 ) .
denote by @xmath518 the roots of the trees , so that @xmath519 .
the trees do not share common vertices with the graph @xmath111 , apart from the vertices in @xmath208 .
the final graph @xmath520 is the desired gadget , where the terminals @xmath109 are the roots of the trees . we denote the family of graphs that can be constructed this way by @xmath521 . note that the size of the construction is @xmath522 which is bounded above by a polynomial in @xmath98 when @xmath27 is a fixed constant .
moreover , any @xmath520 drawn from @xmath521 is bipartite .
the terminals of @xmath520 are @xmath523 , and @xmath101 and @xmath100 are on distinct partitions of the bipartite graph .
next we show that if @xmath91 lie in the non - uniqueness regime of @xmath26 , then the gadget @xmath520 satisfies definition [ def : sly ] .
lemma 19 in @xcite shows that this is true , assuming that a certain technical condition is true ( * ? ? ?
* condition 1 ) .
we will introduce this condition and then show that it always holds when @xmath91 lie in the non - uniqueness regime of @xmath26 .
the technical condition involves asymptotics of the leading terms of the first and second moments
. we will derive expressions for these shortly .
moreover , the condition is stated for the case when @xmath506 , as @xmath524 is relatively small and @xcite shows that it does not affect the leading terms .
suppose that @xmath525 .
we look at the contribution of configurations which assign a prescribed fraction of vertices in @xmath526 the spin 1 , with a view to identifying the phases of the gadget . here
@xmath527 and @xmath528 and @xmath529 . for @xmath530 , let @xmath531 where @xmath532 denotes the set of vertices assigned the spin @xmath15 in the configuration @xmath6 .
denote by @xmath533 the total contribution to the partition function of @xmath111 by the configurations in @xmath534 , that is , @xmath535 we will calculate the leading term in the exponent ( as a function of @xmath77 ) of the first and second moment of @xmath533 . by linearity
we have @xmath536=\sum_{\sigma\in \sigma^{\chi^{+},\chi^{-}}}{\ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n}\big[w_g(\sigma)\big]$ ] .
fix @xmath537 . for convenience ,
set @xmath538 .
let @xmath539 be the number of @xmath13 edges in a random perfect matching between @xmath540 under the configuration @xmath6 .
note that @xmath541 , where @xmath542 and @xmath543 .
it is not hard to see that @xmath544=\lambda^{n(\chi^+_n + \chi^-_n ) } \left(\sum_{\chi_n\leq x_n\leq\chi_n ' } \kappa^{\chi^{+}_n,\chi^{-}_n}_{x_n } \right)^{\delta } , \mbox { where } \kappa^{\chi^{+}_n,\chi^{-}_n}_{x_n } = \frac { \binom{n\chi^+_n}{nx_n } \binom{n(1-\chi^+_n)}{n(\chi^-_n - x_n ) } } { \binom{n}{n\chi^-_n } } \beta^ { n(1-\chi^+_n-\chi^-_n+x_n ) } \gamma^{nx_n}.\ ] ] by the linearity of expectation and symmetry , we then have @xmath545 = \big|\sigma^{\chi^+,\chi^-}\big|\ , { \ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n}\big[w_g(\sigma)\big ] , \mbox { where } \big|\sigma^{\chi^+,\chi^-}\big|=\binom{n}{n\chi^+_n}\binom{n}{n\chi^-_n}. \label{two2april}\end{aligned}\ ] ] let @xmath546 and @xmath547 .
note that as @xmath548 , @xmath549 . in section [ sec : asymptotics ] , using stirling s approximation , it is shown that @xmath550=\psi_{1;\beta,\gamma,\lambda}(\chi^+,\chi^-)+o(1),\ ] ] where , under the convention @xmath551 , @xmath552 note that all the quantities in are independent of @xmath77 . to calculate the second moment
, the approach is completely analogous though we need to introduce more variables . for @xmath553 ,
let @xmath554 let @xmath555 .
notice that @xmath556 is the contribution to @xmath557 from pairs of configurations in @xmath534 with overlap @xmath558 and @xmath559 .
then , we have @xmath560 = \sum_{\upsilon^+,\upsilon^-}{\ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n}\left[y^{\chi^+,\chi^-}_{\upsilon^+,\upsilon^-}\right],\ ] ] where the sum is over integer multiples @xmath561 and @xmath562 of @xmath563 . later
we will replace this sum with a maximum and consider general real numbers @xmath561 and @xmath562 . to unify the notation in both cases , set @xmath564 and @xmath565 .
note that the overlap @xmath566 is at most @xmath567 and at least @xmath568 and similarly for the minus side . to get an expression for
@xmath569 $ ] , fix arbitrary @xmath570 . the configurations
@xmath571 and @xmath572 divide each of @xmath573 and @xmath505 into @xmath574 subsets . let the subsets chosen for spin @xmath15 by both @xmath571 and @xmath572 be numbered @xmath15 on the @xmath28 side and the @xmath29 side . similarly , let the subsets chosen for spin @xmath15 in @xmath571 and for spin @xmath153 in @xmath572 be numbered @xmath575 and the subsets chosen for spin @xmath153 in @xmath571 and spin @xmath15 in @xmath572 be numbered @xmath231 . finally , let the subsets chosen for spin @xmath153 in both @xmath571 and @xmath572 be numbered @xmath574 .
let @xmath576 be the number of edges between subset @xmath577 on one side and subset @xmath198 on the other , and @xmath578 .
define @xmath579 observe that the @xmath580 satisfy @xmath581,\quad \sum_i y_{ijn}=r_{jn},~\text { for } j\in[4],\\ y_{ijn}\geq 0~\text { and } y_{ijn } \text { is an integral multiple of $ 1/n$}. \end{gathered}\ ] ] then we have @xmath582=\lambda^{2n(\chi^+_n + \chi^-_n)}\bigg ( \sum_{\mathbf{y_n}}\xi^{\chi^+_n , \chi^-_n}_{\upsilon^+_n,\upsilon^-_n,\mathbf{y}_n}\bigg)^{\delta},\label{longeq}\ ] ] where @xmath583 to see why is true note that a random matching may be obtained by fixing the permutation within each of the @xmath584 subsets @xmath585 of the `` @xmath29 '' side and then choosing the @xmath584 corresponding blocks of the `` @xmath28 '' side giving the denominator @xmath586
. then the numerator counts all matchings consistent with @xmath571 and @xmath572 giving @xmath587 .
the factors @xmath588 cancel , giving the first line in .
the second line is just @xmath589 . to account for the cardinality of @xmath590 , by the linearity of expectation and symmetry ,
we obtain @xmath591 = { n\choose nl_{1n},nl_{2n},nl_{3n},nl_{4n}}{n\choose nr_{1n},nr_{2n},nr_{3n},nr_{4n}}\ , { \ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n}[w(\sigma_1)w(\sigma_2)]\ ] ] using again stirling s approximation , in section [ sec : asymptotics ] it is shown that @xmath595 = \psi_{2;\beta,\gamma,\lambda}'(\chi^+,\chi^-,\upsilon^+,\upsilon^-)+o(1),\end{aligned}\ ] ] where @xmath596 where the maximum is over non - negative ( real vectors ) @xmath597
satisfying @xmath598 , \quad \sum_i y_{ij}=r_j,~\text { for } j\in[4].\\\ ] ] to obtain the leading term of the second moment @xmath599 $ ] , define @xmath600 in section [ sec : asymptotics ] , we will show @xmath601 = \psi_{2;\beta,\gamma,\lambda}(\chi^+,\chi^-)+o(1),\ ] ] where @xmath602 we want to focus on @xmath603 which maximize @xmath604 .
it is shown in ( * ? ? ?
* lemma 1 ) that when non - uniqueness holds , there exist @xmath605 with @xmath606 such that @xmath607 is the unique maximizer ( up to symmetry ) of @xmath604 .
we should note here that the values @xmath605 are the analogues of @xmath468 for the infinite @xmath27-regular tree ( observe that the infinite @xmath27-regular tree differs from @xmath26 only at the degree of the root ) .
the technical condition from @xcite is as follows . to see that our version of condition [ cond : cond ] is equivalent to the version in @xcite , use ( [ lastone ] )
our interest in condition [ cond : cond ] is justified by the following lemma which is proved in @xcite .
we will establish condition [ gsv12:condition1 ] when @xmath33 lies in the non - uniqueness region of @xmath26 . to do so
, we apply results from @xcite .
the next lemma is a specialization of theorem @xmath231 from @xcite to our setting for the case that there is no field .
for the first equation , note from the definition of @xmath615 that @xmath616 thus , from the definition of @xmath617 , we have @xmath618 similarly , from the definition of @xmath619 , we have @xmath620 and from the definition of @xmath621 , @xmath622 however by @xmath623 therefore we have @xmath624 as @xmath625 is the maximum of @xmath619 over @xmath597 and @xmath626 is the maximum of @xmath625 over @xmath627 , the second and the third equations hold .
lemma [ lem : new ] shows that modulo a constant term , the expressions @xmath617 and @xmath626 are preserved by the standard transformation on @xmath27-regular graphs whereby an external field @xmath24 is pushed into the edge interactions .
using this lemma , we can now draw conclusions about the maximisation of these quantities in our setting .
[ lem : first and second moment ] for any parameter set @xmath33 and @xmath36 , if @xmath33 lies in the non - uniqueness region of @xmath26 , then @xmath628 and @xmath629 maximizes both @xmath604 and @xmath630 .
first , using equalities ( [ ssone ] ) and ( [ sstwo ] ) , and the fact that @xmath631 \geq { \ensuremath{\mathbf{e}}}[x]^2 $ ] , @xmath632 + o(1)\\ & \geq \frac{1}{n}\log { \left ( { \ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n}\big[z_g^{p^+,p^-}\big ] \right)}^2 + o(1)\\ & = 2 \frac{1}{n}\log { \ensuremath{\mathbf{e}}}_{{\ensuremath{\mathcal{g}}}_n } \big[z^{p^+,p^-}_g\big ] + o(1)\\ & = 2 \psi_{1;\beta,\gamma,\lambda}(p^+,p^-)+o(1).\end{aligned}\ ] ] from corollary [ cor : new ] , we have that for all @xmath641 and @xmath642 and all @xmath643 , @xmath646 in particular , this holds for @xmath647 and @xmath648 . together with lemma [ lem : first and second moment ] , it gives us that for all @xmath649 , @xmath650 it implies that @xmath608 is maximized at @xmath651 , which is condition [ gsv12:condition1 ] .
the proof is a rather lengthy formalisation of the following main idea : as @xmath77 grows large , the sums in the expressions for the first and second moments are dominated by their maximum terms and the integrality conditions for the variables may be dropped introducing only @xmath652 ( additive ) error in the asymptotics of the logarithms .
we will use the following lemma .
recall that for a metric space @xmath653 , an @xmath70-cover of a set @xmath654 is a set @xmath655 such that every point in @xmath656 is within distance @xmath70 from a point in @xmath657 .
note that we do _ not _ require @xmath658 . recall also that @xmath659 is a limit point of @xmath656 if @xmath69 is in the closure of @xmath660 .
[ lem : maximalimit ] let @xmath661 be a continuous function on @xmath662 .
let @xmath663 be a compact subregion of @xmath664 . for @xmath665
, let @xmath666 be an @xmath667-cover of @xmath663 , where @xmath668 as @xmath592 .
assume further that the @xmath669 are uniformly bounded and that the set of limit points of @xmath670 is a subset of @xmath663 .
then , as @xmath592 , it holds that @xmath671 we first prove the right - most inequality in .
assume for the sake of contradiction that @xmath676 .
then , there exist sequences @xmath677 with @xmath678 and @xmath679 such that @xmath680 for some @xmath681 .
since the @xmath669 are uniformly bounded , @xmath682 are bounded , and so they have a convergent subsequence , whose limit we denote by @xmath683 . from the continuity of @xmath674
, we obtain @xmath684 .
we claim that @xmath685 , and thus obtain a contradiction to the choice of @xmath673 .
indeed , if there exists @xmath686 such that @xmath687 for all @xmath688 , we have @xmath685 by compactness of @xmath663 .
otherwise , by restricting to a subsequence of @xmath682 if necessary , @xmath683 is a limit point of @xmath670 , and hence @xmath685 . to show the left - most inequality in ,
let @xmath689 be such that @xmath690 .
since the @xmath669 are @xmath667-covers of @xmath663 and @xmath691 , there exists a sequence @xmath692 with @xmath693 such that @xmath694 .
note that @xmath695 and thus , from the continuity of @xmath674 , we obtain @xmath696 .
this finishes the proof of .
we will now prove equations , and .
we start with the first moment @xmath536 $ ] . under the convention @xmath697
, @xmath698 is continuous for @xmath699 and @xmath700 .
we will write @xmath701 for @xmath702 .
the function @xmath701 is continuous over the region @xmath703 the following asymptotic expressions are obtained from stirling s approximation for @xmath704 and @xmath705 using equations and .
@xmath706 note that the sum over @xmath707 in the expression which is used in for @xmath536 $ ] has at most @xmath77 terms , so obviously the number of terms is at most a polynomial in @xmath77 .
thus , we may approximate the sum by its maximum term to obtain @xmath708= \log(\lambda)(\chi^+_n+\chi^-_n)+(\delta-1){f}_1(\chi^+_n,\chi^-_n)+\delta \max_{x_n\in[\chi_n,\chi_n ' ] } { g}_{1}(x_n,\chi^+_n,\chi^-_n)+o(1).\ ] ] where , recall , @xmath709 . moreover , as @xmath592 , we have @xmath710}{g}_{1}(x_n,\chi^+_n,\chi^-_n)&\rightarrow \max_{x\in[\chi,\chi ' ] } { g}_{1}(x,\chi^+,\chi^-)\label{eq : ng1},\end{aligned}\ ] ] where recall that @xmath711 .
the limit follows from @xmath712 and the continuity of @xmath713 . to establish the limit we use lemma [ lem : maximalimit ] for the function @xmath615 , with @xmath714 as defined in ( [ eq : rg ] ) , @xmath715 and @xmath716 .
the @xmath669 are @xmath717-covers of @xmath663 ( with room to spare for the constant 10 ) under the euclidean metric on @xmath718 .
it is also immediate to verify that the set of limit points of @xmath719 is a subset of @xmath663 as a consequence of @xmath720 .
we next turn to the second moment @xmath722 $ ] .
the reasoning is almost identical , though carrying out the arguments is more cumbersome due to the larger number of variables .
let @xmath723 the function @xmath724 is continuous over @xmath725 .
we will also write @xmath726 instead of @xmath727 . note that @xmath726 is continuous over @xmath728 , y_{ij } \geq 0\}$ ] . by analogy
to , define @xmath729 for @xmath730 , define @xmath731 } \mid \forall i , j\in[4 ] , y_{ij } \geq 0,\sum_j y_{ij}=l _ i , \sum_i y_{ij } = r_j \big\}.\ ] ] let @xmath732 and let @xmath733 .
we are now set to compute the asymptotics of @xmath738 $ ] , for which we have derived an expression in .
approximating the sums in the expression by their maximum terms only introduces @xmath652 error , so we obtain @xmath739&=2\log(\lambda)(\chi^+_n+\chi^-_n)+(\delta-1){f}_2(\chi^+_n,\chi^-_n , \upsilon^+_n,\upsilon^-_n)+\delta \max_{\mathbf{y}_n\in \mathcal{c}_n } { g}_{2}(\mathbf{y}_n)+o(1).\end{aligned}\ ] ] taking limits as @xmath592 , we will now derive analogues of and : @xmath740 equation ( [ eq : nf2 ] ) follows from the continuity of @xmath741 .
equation ( [ eq : ng2 ] ) follows by lemma [ lem : maximalimit ] .
combining , and gives @xmath742=2\log(\lambda)(\chi^++\chi^-)+ ( \delta-1){f}_2(\chi^+,\chi^- , \upsilon^+,\upsilon^-)+\delta \max_{\mathbf{y}\in \mathcal{t}(\chi^+,\chi^- , \upsilon^+,\upsilon^- ) } { g}_{2}(\mathbf{y})+o(1).\ ] ] the right - hand side is @xmath743 , so we have established . to obtain , the sum yields @xmath744&=2\log(\lambda)(\chi^+_n+\chi^-_n)\\ & + \max_{(\upsilon^+_n,\upsilon^-_n,\mathbf{y}_n)\in \mathcal{o}_n}\ { ( \delta-1){f}_2(\chi^+_n,\chi^-_n , \upsilon^+_n,\upsilon^-_n)+\delta { g}_{2}(\mathbf{y}_n)\}+o(1 ) , \end{aligned}\ ] ] where @xmath745 yet another application of lemma [ lem : maximalimit ] yields @xmath746 where @xmath747 . combining and yields , as desired .
andrei a. bulatov , martin dyer , leslie ann goldberg , mark jerrum , and colin mcquillan .
the expressibility of functions on the boolean domain , with applications to counting csps . , 60(5):32:132:36 , october 2013 .
xi chen , martin e. dyer , leslie ann goldberg , mark jerrum , pinyan lu , colin mcquillan , and david richerby .
the complexity of approximating conservative counting csps .
preprint , 2012 .
available from the arxiv at : http://arxiv.org/abs/1208.1783 andreas galanis , qi ge , daniel tefankovi , eric vigoda , and linji yang .
improved inapproximability results for counting independent sets in the hard - core model . in _ approx - random _ ,
pages 567578 , 2011 .
andreas galanis , daniel tefankovi , and eric vigoda .
inapproximability of the partition function for the antiferromagnetic ising and hard - core models .
preprint , 2012 .
available from the arxiv at : http://arxiv.org/abs/1203.2226[http://arxiv.org/abs/1203.2226v3 ] andreas galanis , daniel tefankovi , and eric vigoda .
inapproximability for antiferromagnetic spin systems in the tree non - uniqueness region .
preprint , 2013 . to appear in _ stoc _ , 2014 .
available from the arxiv at : http://arxiv.org/abs/1305.2902[http://arxiv.org/abs/1305.2902v2 ] qi ge and daniel tefankovi . a graph polynomial for independent sets of bipartite graphs . in _
fsttcs _ , pages 240250 , 2010 .
leslie ann goldberg , mark jerrum , and colin mcquillan . approximating the partition function of planar two - state spin systems .
preprint , 2012 .
available from the arxiv at : http://arxiv.org/abs/1208.4987 liang li , pinyan lu , and yitong yin .
correlation decay up to uniqueness in spin systems . in _ soda _ ,
pages 6784 , 2013 .
available from the arxiv at : http://arxiv.org/abs/1111.7064[http://arxiv.org/abs/1111.7064v2 ] | counting independent sets on bipartite graphs ( # bis ) is considered a canonical counting problem of intermediate approximation complexity .
it is conjectured that # bis neither has an fpras nor is as hard as # sat to approximate .
we study # bis in the general framework of two - state spin systems on bipartite graphs .
we define two notions , nearly - independent phase - correlated spins and unary symmetry breaking .
we prove that it is # bis - hard to approximate the partition function of any 2-spin system on bipartite graphs supporting these two notions . as a consequence ,
we classify the complexity of approximating the partition function of antiferromagnetic 2-spin systems on bounded - degree bipartite graphs . |
understanding the initial mass function ( imf ) is one of the most important problems of stellar astrophysics .
the imf , together with the star formation history , dictates origin , evolution , and fate of the stellar populations , from individual clusters up to entire galaxies .
the distribution of stellar masses has been studied in depth for more than half a century , starting with the pioneering work of @xcite ; the major open question is whether the imf is universal or if it depends on the initial conditions of star formation ( see , e.g. , * ? ? ?
* ; * ? ? ?
whereas it is commonly accepted that the imf seems well reproduced by a power law for masses greater than several tenths of a solar mass , the shape and universality of the imf in the substellar mass regime is still under investigation ( e.g. , * ? ? ?
very young clusters ( few myr old ) in star - forming regions provide a unique tool to investigate the imf across the entire mass spectrum , for a number of reasons .
first , they are usually young enough that neither dynamical processes nor stellar evolution have altered the mass distribution ; therefore the measured distribution of stellar masses coincides with the imf .
moreover , young low - mass stars and brown dwarfs ( bds ) are in their brightest evolutionary stage when they contract towards the main sequence , therefore these objects are more easily detected and characterized when they are young . among the nearby galactic star - forming regions ,
the orion nebula cluster ( onc ) is an ideal site for the study of star formation in particular for low - mass stars and bds .
this cluster counts a few thousand members , 1 - 3 myr old @xcite , spanning the entire mass spectrum ( @xmath4 m@xmath5 ) .
several studies have been conducted in the past decade to measure the imf in the onc ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; they generally find ( similar to the field star population ) a salpeter - like slope above 1 m@xmath5 , which flattens to a broad peak at 0.2 - 0.3 m@xmath5 ( though the shape and position of the imf peak is highly model dependent , @xcite ) , and the mass distribution likely decreases in the substellar mass range .
determining masses in a young region such as the onc is not without difficulties : besides the strong nebular emission , another major impediment is caused by differential reddening , which has strongly limited the ability to derive the stellar parameters of individual sources based on photometry alone . to overcome this shortcoming ,
spectroscopic surveys have been carried out @xcite , but they are either limited to a fraction of the members , or to the very central part of the region , the trapezium cluster . using near - infrared ( nir ) photometry it is possible to asses stellar masses down to the planetary masses ( @xmath6 m@xmath7 ) , for example by de - reddening the measured color - magnitude diagrams ( cmds ) on one isochrone or even to deriving directly the imf from the nir luminosity functions ( lfs ) @xcite .
this second approach , however , is not very accurate : the onc shows a significant luminosity spread , sometimes interpreted as a real age spread @xcite , and stars of different age follow different mass - luminosity relation .
moreover the nir excess originating in the inner circumstellar disks @xcite alters the observed fluxes .
clearly , in order to improve our knowledge of the onc imf , a precise and systematic characterization of the stellar parameters of individual members is needed . to this purpose ,
the _ orion hst treasury program _
@xcite has produced a high spatial resolution photometric survey of the onc with three instruments onboard the hubble space telescope , from @xmath8band to the nir , over a large field of view ( fov ) , with a sensitivity well into the bd mass range . to fully exploit this exceptional dataset , an accurate estimate of the effective temperature ( @xmath9 ) is needed to derive @xmath10 from the observed colors and therefore the stellar luminosities . in our previous works ( * ? ? ?
* ; * ? ? ?
* hereafter paper i and paper ii ) we have collected spectral types from the literature and complemented them with new optical spectroscopy .
moreover , we have presented a new observational technique , based on optical medium - band photometry at 6200 a wavelength where the spectra of cool stars show a deep , @xmath9-dependent tio absorption feature to derive the spectral types of m - type sources .
this allowed us to obtain the stellar parameters of @xmath11 members and derive the most complete hrd of the onc , down to the hydrogen burning limit . in the present paper
we extend our investigation to lower masses , well into the bd regime , and with higher completeness . to this purpose , we take advantage of the same observational strategy presented in paper i , using medium - band photometry to derive spectral types of cool sources , from which @xmath9 and @xmath10 are derived , and place individual sources in the hrd . here , instead of the @xmath12 filter used in paper i , we select two bands at longer optical wavelengths , in order to increase our sensitivity for cooler sources such as bds and further reduce the influence of extinction . in section [ section :
the_data ] we present our new observations , the data reduction and calibration . in section [ section : analysis ] we define two spectro - photometric indices based on our medium - band photometry . using available stellar parameters for a small sample of onc very - low mass stars and bds , we define empirical transformations to convert these indices into @xmath9 and @xmath10 .
we also show that this technique is not significantly affected by optical excesses from mass accretion , typical of young stars and bds . in section
[ section : hrd ] we derive the new hrd for the onc , which now includes @xmath13 sources , reaching masses as low as @xmath14 .
we study the completeness as a function of @xmath9 and bolometric luminosity ( @xmath15 ) accounting for photometric detection as well as differential reddening ; we detect a population of candidate background contaminants , which appears well separated from the onc members in the hrd , at lower luminosity . after excluding the contaminants , in section [ section : imf ] we derive and discuss the imf of the onc .
observations were carried out with the wide field imager ( wfi ) , a focal reducer - type camera mounted at the cassegrain focus of the 2.2-m mpg / eso telescope at la silla .
the field of view ( fov ) of the camera is @xmath16 , allowing us to cover the entire onc with one pointing .
this optical imager offers a particularly large set of broad- , medium- , and narrow - band filters along the entire wavelength range of its ccd sensitivity ( @xmath173000 to 10000@xmath18 ) .
we selected two wfi medium band filters which provide the best sampling of a deep absorption band at @xmath19 .
these are the mb@xmath20753/18-eso848 and the mb@xmath20770/19-eso849 filters , which hereafter will be simply referred to as _ 753 _ and _ 770_. figure [ fig : spectra ] shows the transmission profiles of these filters overlayed to synthetic spectra from @xcite of decreasing temperature . the @xmath21 band is centered on the photospheric continuum , while the 770 is completely covering the tio absorption band whose strength increases with decreasing @xmath22 .
we also used a broad @xmath1-band filter , the bb@xmath20i/203-eso879 , to sample the stellar flux at redder wavelengths , in order to evaluate the photospheric reddening from dust extinction .
this i - band filter is also used in our previous photometric studies of the onc ( paper i and ii ) . table [ tab : bands ] summarizes the characteristics of the 3 filters .
rrrrr 753 & 7536.2 & 184.1 & 191.4 & 1.368@xmath23 + 770 & 7707.9 & 195.2 & 201.3 & 1.275@xmath23 + i & 8620.5 & 1354.9 & 1431.2 & 9.500@xmath24 the wfi imaging was carried out during 5 nights of december 2010 . on each night , observations were performed in all the three bands together so as to minimize the impact of any longer timescale stellar variability .
a standard dithering pattern has been applied , in order to cover the gaps between the 8 ccd chips of wfi and allow for cosmic ray and bad pixel removal .
we did not carry out short exposures to avoid saturation for the bright sources , since the bright end of the onc has been studied well in detail by us in our previous works .
dark , bias and flat field exposures were obtained before and after each night .
we did not observe standard fields for photometric calibration , given the particular filters used in this program and the practicality of performing relative photometric calibration exploiting our large pre - existing data set on orion .
rrrr 753 & 360 & 23 & 9,360 + 770 & 500 & 28 & 14,000 + @xmath1 & 280 & 12 & 3,360 images were processed using the eso / mvm ( multi - vision model ) _ vers . _
1.3.5 package @xcite .
this software automatically performs all common image reduction steps ( e.g. , correction for bias and flat fields ) , and the absolute astrometric calibration of the individual exposures to be merged .
this is accomplished by cross - correlating the images with a reference astrometric catalog ; we utilized to this purpose our previous wfi photometric catalog presented in paper i. the final product of our image reduction is 3 images , one for each filter , obtained by merging all the individual exposures .
table [ tab : observations ] summarizes the total exposures times ; these about 1 hour for the broad band , and about 2.5 and 4 hours for the two medium bands , photometry , both aperture and point - spread - function ( psf ) , was performed using the daophot ii package @xcite .
we computed the psf individually for every image , starting from a coadded sample of few hundreds bright and unsaturated sources , and then refining these with an iterative sigma - clipping method to reject psf stars with poor @xmath25 .
photometry was computed on all the sources brighter than 3@xmath26 above the local sky background noise . in order to eliminate the contamination from spurious detections a potentially significant fraction of the total detections , given the strong variability of the nebular background , and the low threshold for source detection
we rejected sources not present in any of our existing deep photometric catalogs of the onc , both ground based ( wfi @xmath27 , paper i ; ctio / ispi @xmath28 , see @xcite ) or from the hst photometry ( acs @xmath29 , wfpc2 @xmath30 , see @xcite ) .
in particular , the significant depth of the infrared and hst data , over a fov equal or larger than that of the observations described here , guarantees that we are not excluding any real sources in this study .
the photometry has been calibrated to the vegamag system using our previous wfi photometric catalog .
for the @xmath1-band , we simply determined the photometric offset with respect to our previous photometry in the same filter .
for the two medium bands , we used synthetic photometry .
we considered the atmosphere models of @xcite as empirically calibrated by us in paper ii , and computed the two medium - band colors @xmath31 as a function of @xmath9 .
then we considered the actual sources detected in all 3 bands and for which we have @xmath9 and @xmath10 from paper ii , and plotted the measured , extinction - corrected ( @xmath32 ) colors against @xmath9 .
the systematic difference between these colors ( considered individually ) and the synthetic ones is the offset needed to calibrate the medium bands .
specifically , we limited to the stars with @xmath33 k , range where the possible inaccuracies of the atmosphere models are very modest .
we derive the completeness functions for our wfi photometry with an artificial star test , performed separately for each wfi filter .
artificial stars of different magnitude , are added on the reduced wfi images ; then , with the same technique used to detect and extract photometry on the real stars , these artificial stars are recovered or missed depending on their flux , local sky noise , crowding , etc .
the fraction of recovered stars against the total , as a function of magnitude , provides the completeness . in our onc imaging ,
the completeness is spatially variable , being strongly affected by the nebular background , which is brighter at the center where the stellar density is higher .
this correlation , if not accounted for , may bias the derived completeness .
therefore , we first derive the spatial , projected , stellar density distribution of the onc .
this is obtained by merging all catalogs at our disposal , in particular the deep ispi @xmath28 photometry and the hst / acs data .
then , for every magnitude bin , the artificial stars are generated with random positions drawn from the 2 dimensional density distribution .
the derived completeness functions are shown in figure [ fig : plot_completeness_mag ] . in all cases ,
these curves show a characteristic trend : besides the typical steep cutoff at faint luminosities which traces the detection limit , the completeness shows a slow decrease below 1 at bright luminosities .
this is due to the luminous nebular emission and significant crowding of the central part of the onc , where the detection limit is significantly poorer . in the outer regions of the onc , we are able to detect sources all the way down to @xmath34 mag in @xmath1 band ( see figure [ fig : cmd ] ) , and to @xmath35 mag in @xmath21 and @xmath3 .
the final catalog includes 2474 sources with photometry in all 3 bands .
the photometric errors as a function of magnitude are shown in figure [ fig : photometric_errors ] .
figure [ fig : cmd ] shows our new cmd , highlighting the significant improvement in depth with respect to our previous work ( paper ii ) . in particular , our photometry covers a very large range of stellar luminosities , spanning @xmath36 mag in @xmath1-band , and the new observations extend @xmath175 mag deeper than in the previous work , reaching @xmath37 , which corresponds to young sources with masses @xmath38 m@xmath5 at 1 myr ( see figure [ fig : cmd ] ) . in figure [ fig : cmd ] we also observe among the faintest sources a concentration of objects with @xmath39 which , as we show below , comprise a distant , background stellar population that is detectable despite the high extinction through the molecular cloud because of the depth of our observations . in figure
[ fig : color_color_diagram ] we present the @xmath40 vs @xmath41 color - color diagram , which forms the basis of our analysis in all that follows .
the most salient feature of this diagram is the inflection point near ( 0.05,0.25 ) , from which the stars extend in two different directions . by comparison with the orientations of the synthetic isochrone and the extinction vector
, we see that this inflection in the diagram essentially separates the stars into a group ( extending to the left ) that stretches mostly parallel to the isochrone and perpendicular to the reddening vector , and a second group ( extending to the upper right ) that stretches parallel to both the isochrone and the reddening vector .
the inflection in the diagram occurs at @xmath42 k. for stars with @xmath43 k ( i.e. spectral types @xmath44m0 ) , the @xmath41 color is both highly sensitive to @xmath9 , becoming rapidly bluer ( more negative ) as @xmath9 decreases ( see also figure [ fig : spectra ] ) , and largely insensitive to @xmath10 because of the small wavelength separation between the 753 and 770 bands . in our analysis below we exploit this property of the diagram to break the degeneracy between @xmath9 and @xmath10 for these very cool stars , which has previously been one of the most difficult challenges in the derivation of stellar parameters from optical broadband photometry for m - stars and bds ( @xcite ; paper ii ) .
we note here that the synthetic isochrone clearly does not correctly reproduce the slight upward curvature of the stars as one moves to the coolest @xmath9 , but because these cool stars nonetheless extend perpendicular to the reddening vector , we demonstrate below that the systematic error in the @xmath40 colors predicted by the synthetic isochrones can be readily calibrated out for @xmath43 k. we note also that the effects of contamination by faint background sources is mitigated in this region of the diagram , particularly for the coolest objects of interest ( @xmath45 k , i.e. , in the bds regime ) , since background bds are too intrinsically faint to penetrate the very high @xmath10 of the cloud behind the onc . as we show in section [ section : hrd ] , the relatively few remaining background contaminants in this region of the diagram ( mostly red giants ) can be filtered out on the basis of their luminosity .
these desirable features of the @xmath40 vs @xmath41 color - color diagram break down for warmer stars with @xmath46 k. for these warmer stars the colors are largely degenerate in @xmath9 and moreover the isochrone and the reddening vector are parallel .
for these stars we must therefore rely on previous observations by us and from the literature to determine @xmath9 and @xmath10 .
in addition , there is significant background contamination in this region of the diagram ; as we show below , while the brightest of these stars are bona fide onc members , the faintest are mostly background stars .
finally , we note that the apparent lack of low-@xmath10 stars for @xmath47 k is the result of these stars being saturated in our deep images .
such stars are indeed present in the onc , and we include them in our analysis from our own previous work ( paper i and ii ) .
in order to derive the stellar parameters of very low - mass stars ( vlmss ) and bds in the onc , we focus our analysis on the left - hand part of the color - color diagram of figure [ fig : color_color_diagram ] , where the @xmath41 color anti - correlates with @xmath9 .
our basic methodology is to project each source from its observed position in the color - color diagram back along the reddening vector to the theoretical isochrone .
the choice of theoretical isochrone is largely arbitrary because none of the available isochrones exactly reproduce the observed distribution of the stars in the diagram , and consequently we must empirically recalibrate the isochrone based on the observed distribution .
our assumption is that , whichever fiducial isochrone is adopted , the non - reddened stars should lie on the isochrone , and the reddened stars displaced away from it along the reddening vector .
for example , the isochrone shown in figure [ fig : color_color_diagram ] computed using the synthetic spectra of @xcite predicts a much bluer @xmath40 index than our data for low @xmath41 .
if the model were correct , this would imply an unrealistic dependence of the average @xmath10 on @xmath9 , in the sense that the minimum measured extinction ( i.e. the distance , along the reddening vector , between the isochrone and the stars ) increases monotonically with decreasing @xmath9 .
different synthetic isochrones can lead to more realistic predictions of colors .
figure [ fig : color_color_models ] shows our color - color diagram with a number of synthetic isochrones , computed using both the ames - mt spectra of @xcite and the most recent grid the bt - settle models from @xcite .
we also show an isochrone ( red line ) derived using the @xcite atmospheres as recalibrated by us in paper ii . that calibration , appropriate for the onc , involved an empirical constraint on the stellar @xmath48 to produce a synthetic isochrone that matched the observed broadband colors of the onc in @xmath49 .
it is clear that all synthetic isochrones , including that which we used in paper ii , are incompatible with the data in this new medium - band color - color plane .
ames - mt consistently predicts a systematically lower @xmath40 at low @xmath9 , and bt - settle predicts either a turn - over in @xmath41 for low @xmath48 ( young isochrones ) or @xmath40 colors that are too red for high @xmath48 ( the zams model ) .
evidently , current synthetic models , even those that have been empirically calibrated by us to match the broadband colors of young low - mass stars , are unable to correctly reproduce the molecular feature at @xmath50and / or the photospheric continuum level at @xmath51 which form the basis of our new medium - band color - color diagram technique . as a consequence , we must empirically define an isochrone representing both the photospheric colors in our bands and the @xmath9 scale along it .
to proceed , we introduce a new set of coordinates in the color - color plane of figure [ fig : color_color_diagram ] .
we define the @xmath52 $ ] as the distance that a given star is displaced , along the reddening vector , from the @xmath53axis of the color - color diagram , measured in units of @xmath10 : @xmath54= & \bigg(m_{753}-i\bigg)\bigg/\bigg(\frac{a_{753}}{a_v}-\frac{a_i}{a_v}\bigg)\end{aligned}\ ] ] where @xmath55 , @xmath56 and @xmath57 ( values computed from the @xcite reddening law for @xmath58 ) .
we choose the reddening parameter @xmath58 instead of a higher value ( e.g , 5.5 , @xcite ) because , as we found in paper ii , this value better explains the observed broadband colors of the onc members .
we similarly define the @xmath59 $ ] as the projection of a given star , along the direction of the reddening vector , onto the x - axis [ i.e. @xmath60 : @xmath61= & m_{753}\cdot\bigg[1-\big(\frac{a_{753}}{a_v}-\frac{a_{770}}{a_v}\big)/\big(\frac{a_{753}}{a_v}-\frac{a_{i}}{a_v}\big)\bigg ] + \nonumber \\ & + i\cdot\bigg[\big(\frac{a_{753}}{a_v}-\frac{a_{770}}{a_v}\big)/\big(\frac{a_{753}}{a_v}-\frac{a_{i}}{a_v}\big)\bigg ] + \nonumber \\ & - m_{770}\end{aligned}\ ] ] by definition , the @xmath59 $ ] is reddening independent , and so it depends only on @xmath9 .
figure [ fig : color_color_indexes ] shows a schematic representation of the two indices in the color - color diagram .
because the currently available synthetic spectra are unable to accurately predict the stellar colors in our medium - band color - color diagram , we must derive an empirical transformation between the @xmath59 $ ] and @xmath9 .
to this end we utilize the sample of onc stars with directly determined ( i.e. spectroscopic ) spectral types . with the appropriate transformation between the @xmath59 $ ] and @xmath9
, we will be able to reliably determine @xmath9 of all the other members from their measured [ @xmath62 .
we have assembled all available optical spectral types from our paper ii ( for stellar - mass objects ) and nir spectral types ( for substellar - mass objects ) from @xcite and @xcite .
the spectral types were converted to @xmath9 using the temperature scale of @xcite , as we did in paper ii .
figure [ fig : index_teff ] shows the resulting relationship between our measured [ @xmath62 and the spectroscopically determined @xmath9 .
as expected , for late spectral subtypes the @xmath59 $ ] becomes bluer ( i.e. becomes more negative ) with @xmath9 .
this approximately linear dependence holds from @xmath63 k ( @xmath17m2.5 ) down to the coldest available spectral types for bds , and down to our photometric lower limit ( @xmath59\simeq0.8 $ ] ) .
perhaps not surprisingly , we found that the scatter in the [ @xmath62@xmath9 relation could be significantly reduced if we consider only sources that do not show strong spectroscopic evidence of accretion .
we selected the
non - accretors " on the basis of low h@xmath64 emission ( e.w.@xmath65 ) from paper i and low optical veiling ( low @xmath66band excess emission ) from paper ii . figure [ fig : index_teff ] shows that these non - accretors form a relatively tight sequence , while the rest of the stars ( the accretors ) are much more scattered in the diagram .
the accretors are also systematically shifted to higher @xmath9 and/or lower [ @xmath62 , indicating that accretion alters the position of the sources in the diagram , and indeed below we exploit this feature to quantify the degree to which accretion may affect our determination of @xmath9 . using only the sample of non - accretors , we derive an empirical linear relation between @xmath9 and the @xmath59 $ ] : @xmath67 \cdot1150~{\rm k}\ ] ] which is valid for @xmath59<-0.1 $ ] .
the standard deviation of the data relative to this relation is @xmath68 k , roughly corresponding to one - half of a spectral subtype .
this is comparable to the overall uncertainty in our spectroscopically determined @xmath9 from paper ii , suggesting that this scatter is principally due to uncertainty in the ( spectroscopic ) spectral types , and therefore any additional uncertainty introduced through the @xmath59$]@xmath9 transformation is negligible .
the analysis methodology utilizing our medium - band color - color diagram is predicated on the assumption that the stellar colors in our photometric bands depend only on @xmath9 and @xmath10 , permitting us to establish an empirical relationship between the our [ @xmath62 and @xmath9 ( equation [ equation : teff_vs_index ] ) .
however , the colors of young stars can be affected by accretion , whereby infalling matter from the circumstellar disk onto the stellar surface leads to a flux excess , strongest in the uv and in specific emission lines , but also to a smaller extent in the continuum at optical wavelengths .
the latter is typically characterized by a heated photosphere covering @xmath171% of the stellar surface and with @xmath69 k , therefore peaking in the b - band @xcite . since our
photometry uses filters at the red end of the optical spectrum , and there are no emission lines within the wavelength range covered by our filters , we expect that accretion should induce only a modest alteration of our measured colors .
we investigate this hypothesis in a quantitative way , calculating the shift in [ @xmath62 , at constant @xmath9 , obtained by contaminating the photospheric fluxes with varying degrees of accretion luminosity . as in paper
ii , where we performed a similar modeling to derive the change in the @xmath49 colors due to mass accretion , we first model a typical accretion spectrum , then we add it with increasing relative flux fraction on synthetic stellar spectra of different @xmath9 .
finally , by means of synthetic photometry as above , we compute the resulting changes in the @xmath59 $ ] and in turn the error induced in the inferred @xmath9 . specifically , we simulated a typical accretion spectrum using the cloudy photoionization code .
this is similar to that we used in paper ii , but has been further refined to provide a more realistic sed in the ultraviolet range ( manara et al .
2011 , _ in prep .
we use the synthetic spectra that we used in paper ii namely the empirically calibrated ames - mt model from @xcite and for every @xmath9 we add a component of our accretion spectrum to the photospheric spectra , varying the ratio @xmath70 . the result is shown in figure [ fig : index_teff_accretion ] , from which two main results are evident : first , contaminating the stellar flux with accretion excess shifts the [ @xmath62 to higher values , and the effect is more prominent for cooler @xmath9 .
this is intuitive considering that this index traces the ratio between the flux outside ( @xmath71 ) and inside ( @xmath50 ) the tio molecular feature .
the accretion - induced continuum excess partially `` fills in '' the tio absorption feature , leading to higher ( less negative ) values of [ @xmath62 .
second , for typical values of accretion luminosities in a pms cluster ( @xmath72 ) the shift in the [ @xmath62 is negligible . from figure [ fig : index_teff_accretion ] it also appears that the synthetic isochrone does not quite follow our data .
this is not surprising , given that , as discussed in section [ section : the_data ] , this model provides a somewhat inaccurate estimate of the stellar fluxes in our medium bands ( see figure [ fig : color_color_models ] ) .
therefore , in order to obtain a correct , quantitative assessment of the accretion - induced shift in the [ @xmath62 , we must recalibrate the results shown in figure [ fig : index_teff_accretion ] .
we proceed as follows .
the systematic error in the depth of the tio feature as a function of @xmath9 can be treated as a simple systematic error in the @xmath9 scale .
for example , for the lowest @xmath9 modeled in figure [ fig : index_teff_accretion ] ( @xmath73 k ) , the models predict a depth of the tio feature corresponding to [ @xmath62@xmath74 mag , and by adding @xmath75 of accretion luminosity one obtains a shift of the index of @xmath76 mag .
in fact , from figure [ fig : index_teff ] and equation [ equation : teff_vs_index ] , [ @xmath62@xmath74 mag corresponds to a lower @xmath9 , about 2700 k. therefore we can associate this @xmath9 with a shift of @xmath76 mag .
then , we readjust the ratio @xmath77 , using the updated @xmath9 , according to @xmath78 . finally , we can recast the result from a more observational perspective , in which we measure [ @xmath62 for a given star and we do not know the true value of @xmath79 for that star .
deriving @xmath9 via equation [ equation : teff_vs_index ] , which is defined for @xmath80 , we will overestimate the true @xmath9 because accretion will cause the tio feature to be filled - in and consequently we will measure a higher [ @xmath81 .
figure [ fig : accretion_bias ] shows the resulting systematic overestimation of the derived @xmath9 for various @xmath79 .
we find again that the lower the stellar @xmath9 , the larger the error in the @xmath9 estimate due to accretion .
however , if one excludes very strong accretors ( @xmath82 ) , the error in @xmath9 is very small .
for instance , for @xmath83 , the @xmath9 error is always smaller than half spectral subtype , and always @xmath84 k above the h - burning limit ( @xmath85 k ) .
the @xmath9 difference associated to an error of 1 spectral type was computed as the difference between the temperature of two contiguous spectral subtypes .
similarly , the @xmath9 differences corresponding to @xmath86 and @xmath87 spectral subtypes ( dotted lines in figure [ fig : accretion_bias ] ) is equal to that for 1 spectral type , scaled down by a factor of 2 and 4 . in conclusion , the presence of optical excess due to accretion has a negligible effect on the stellar parameters derived from our medium - band survey .
thus , we will neglect accretion in the following analysis , and proceed to derive @xmath9 based on the [ @xmath62 from equation [ equation : teff_vs_index ] . despite having shown that accretion excess not
not alter significantly the measured [ @xmath62 , figure [ fig : index_teff ] showed that accreting onc sources are displaced in the @xmath9 vs. [ @xmath62 plane .
the displacement is such that the accretors lie above the non - accretors , meaning that the spectroscopic @xmath9 plotted on the y - axis of figure [ fig : index_teff ] have been on average overestimated for accreting onc sources .
this overestimation of spectroscopically determined @xmath9 for accreting sources is likely the result of incomplete removal of the optical veiling in the spectra of these sources .
this excess , that we demonstrated above to be minimized at 7700(although it may still be present in extreme cases , see , e.g. , @xcite ) , is likely larger at bluer wavelengths , where other molecular features are typically used to measure the spectral types of m - type stars .
this veiling , if not removed or if incompletely removed , will fill in the molecular absorption lines and lead to a systematic overestimation of the spectral types . by utilizing the [ @xmath62 to derive @xmath9 we are instead largely immune to these effects , and we are now able to correct the previous spectroscopically derived @xmath9 via equation equation [ equation : teff_vs_index ] . in conclusion , our technique to measure stellar @xmath9 is able to derive a more accurate estimate of @xmath9 than optical spectroscopy , being less biased by a possible veiling excess . in section [ subsection
: teff_calibration ] we derived an empirical relation between the @xmath59 $ ] and the stellar @xmath9 , and in section [ subsection : accretion ] we explored the effects of accretion luminosity on there derivations . here ,
similarly , we find the relation between the @xmath52 $ ] , obtained in equation [ equation : avindex ] , and the actual extinction @xmath10 of the onc stars , in magnitudes . in the color - color diagram ,
the extinction @xmath10 of a star is the distance , along the reddening direction , between the observed position and the isochrone , whereas the [ @xmath88 is the distance , also along the same direction and scaled by the appropriate @xmath89 , between the observed colors and the abscissa @xmath90 .
therefore the [ @xmath91 and the true @xmath10 differ only by a constant , which depends on the [ @xmath62 ( i.e. , it is a function of @xmath9 ) . to this end , we plot the extinction - corrected onc stars with independent @xmath10 estimates from paper ii and from @xcite in the [ @xmath88 vs. [ @xmath62 plane ( figure [ fig : aindex_calib ] , left - hand panel ) .
we also use the ispi @xmath28 photometry of the onc presented in @xcite , from which we use stars with the lowest @xmath9 ( low @xmath59 $ ] ) and with the smallest photometric errors . for the coolest @xmath9 considered here ,
the @xmath92 vs @xmath93 isochrone from @xcite is vertical , and its color ( @xmath94 ) is independent of age , therefore @xmath10 can be derived by simply dereddening the sources on the isochrone . by not including the @xmath95-band photometry from @xcite , the infrared - derived extinctions should not be strongly affected by the possible presence of disk excess .
the thick line in the figure is our fit to the data , i.e. , the empirical isochrone corresponding to @xmath10=0 .
this also represents the offset to be subtracted from the measured @xmath52 $ ] , as a function of the [ @xmath62 , to compute the true @xmath10 . among the points in figure [
fig : aindex_calib ] from our previous paper ii , we can distinguish accreting sources from non - accretors , just as in figure [ fig : index_teff ] . in section [
subsection : teff_calibration ] we suggested that previous spectroscopically derived @xmath9 estimates are systematically overestimated for accreting members in the m - type range .
this is now confirmed by figure [ fig : aindex_calib ] : whereas the extinction - corrected non - accretors lie on a curve consistent with the lower edge of the non extinction - corrected population ( right panel of the same figure ) , for accretors we find @xmath10 estimates systematically overestimated .
this is because if @xmath9 is overestimated , the intrinsic colors are underestimated , and therefore the color excess attributed to extinction is overestimated leading to higher @xmath10 .
figure [ fig : aindex_calib ] , right - hand panel , shows again the empirically calibrated isochrone ( i.e. , for @xmath96 ) , together with all the stars in our catalog . unlike in the left - hand panel , we have not subtracted the extinction here .
overall the population lies above the line , i.e. , at positive values of @xmath10 .
as would be expected if the isochrone is correct , the distribution of sources above the isochrone also appears uniform for different @xmath59 $ ] .
there is an overabundance of highly reddened stars at the highest useful limit of the index ( @xmath59 = -0.1 $ ] ) , i.e. , at the highest limit of the @xmath9 range accessible with our medium - band photometry .
this , as will be shown below , is a residual late - type background contaminant population .
having derived two empirical relations , one to obtain the @xmath9 from the @xmath59 $ ] for stars in the spectral subtype range m3m8.5 , the other to derive @xmath10 once the @xmath59 $ ] is known , we now derive the stellar parameters for the sources in our catalog .
sources lying below the isochrone of figure [ fig : aindex_calib]b appear to have negative extinctions , which is unphysical ; 93 stars showing an extinction in the range @xmath97 mag@xmath980 mag are therefore assigned @xmath96 , essentially shifting the measured colors to the closest physically meaningful solution .
another few stars with @xmath99 mag were removed from the sample , being likely spurious detections or faint sources with very inaccurate photometry .
the fraction of sources showing negative extinction ( about 5% ) is much smaller than in previous works ( e.g. , paper ii or @xcite ) , validating the improved accuracy of our methods to derive stellar parameters . in total
, we derived @xmath9 and @xmath10 for 1280 sources , whereas only 544 of them had a previously assigned @xmath9 . merging the new @xmath9 and @xmath10 with those of our previous work , our sample counts 1807 sources with both @xmath9 and @xmath10
from which we can derive an updated h - r diagram for the onc extending to well below the h - burning limit . in order to convert the de - reddened magnitudes into @xmath100 , knowledge of the bolometric corrections ( bcs ) as a function of @xmath9
is required .
unfortunately , we can not rely for such relations on previous works ( e.g. , * ? ? ?
* ; * ? ? ?
* ) for several reasons : 1 ) the photometric bands used here , including @xmath101band , are non - standard ; 2 ) optical bcs usually do not extend in the bd temperature range ; and 3 ) if the optical stellar colors are age dependent , as suggested in paper ii , the bcs may be also . thus , bcs valid for main - sequence dwarfs might not be adequate for a young populations such as the onc .
instead , we use synthetic photometry to derive our bcs , according to the formula : @xmath102,\ ] ] where @xmath103 are synthetic spectra , and @xmath104 the filter throughput .
as we have shown above ( see also the discussion in paper ii ) , current atmosphere models do not correctly predict the optical broad band colors for cool stars .
this clearly affects the computation of bcs , in particular bcs in @xmath101band , @xmath105 . on the other hand , the bolometric flux for a given synthetic spectrum will be trivially correct at a given @xmath9 since , by definition , @xmath106 .
the radii @xmath107 , which could also be inaccurate from the evolutionary models , do not affect the bcs since they introduce an identical proportionality factor in both the numerator and denominator of equation [ equation : bc ] . from the comparison of our data with the synthetic isochrones ( see figure [ fig : color_color_models ] )
we can not ascertain the correctness of the synthetic @xmath1-band fluxes .
it is true that the isochrones are unable to reproduce the data in the color - color diagram , but the shortcomings may be principally in our medium bands and not in the @xmath1-band . at the same time , the bt - settle models @xcite have been validated in the nir for dwarf stars , such that the predicted @xmath92 and @xmath95 fluxes match the observations down to bd masses .
thus , if the predicted @xmath101band magnitudes are also correct , we would expect the predicted @xmath108 color to be also consistent with the observations across all @xmath9 .
we perform this test using the @xmath92-band photometry of the onc from @xcite , together with our @xmath101band magnitudes and ( as in section [ subsection : av_calibration ] ) the @xmath10 estimates from previous works .
figure [ fig : ij_vs_teff ] shows the dereddened @xmath108 color as a function of @xmath9 for our onc sample , in comparison with the same quantity computed using the bt - settle atmospheres , assuming a @xcite 1 myr isochrone to constrain the surface gravity as a function of @xmath9 . as in figure [ fig : ij_vs_teff ] , we highlight sources showing no or little accretion excess , except at the coolest @xmath9 ( @xmath109 k ) , for which we do not have estimates of accretion from paper ii . for @xmath110
k , the isochrone fits the data well .
however , the match is worse at lower @xmath9 , the isochrone predicting lower @xmath108 than shown by our data .
we checked that this conclusion is not strongly dependent on @xmath48 .
trusting the predicted @xmath111band fluxes , as explained above , we conclude that the bt - settle atmosphere models tend to systematically overestimate the @xmath101band fluxes , and this effect is larger for cooler @xmath9 . in figure
[ fig : ij_vs_teff ] we also show the empirical fit to the data .
the difference between the two lines represents the empirical correction needed to calibrate the predicted @xmath1-band synthetic photometry using the bt - settle models . whereas this is small for @xmath112 k , in the bd @xmath9 range the correction increases up to 1 magnitude
thus , we correct the computed bcs , derived from equation [ equation : bc ] , applying this offset as a function of @xmath9 . finally , we derive the bolometric luminosities for all our sources from : @xmath113 where @xmath114 is the absolute magnitude , in wfi @xmath1 band , of the sun
( paper ii ) and dm=8.085 is the distance modulus , adopting @xmath115 pc from @xcite . the resulting h - r diagram is shown in figure [ fig : hrd ] , where we show separately the stars whose parameters have been derived from the new data presented in this work , and those , generally with @xmath116 k , whose @xmath9 and @xmath117 are from paper ii .
the apparent discontinuity in the number of sources around @xmath118 is mainly due to the much higher completeness in the number of low - mass stars with @xmath9 estimates provided by the present work ( see section [ subsection : completenesshrd ] ) . despite the large column density of the orion molecular cloud , which produces an extinction up to @xmath119 mag @xcite , contamination from background
stellar populations starts to become non negligible at faint magnitudes in the onc .
in fact , as suggested by previous works , ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ) , the fraction of background contaminants versus members increases towards the substellar mass range in the onc .
this is both due to the increase in the number of faint background sources seen through the orion molecular cloud , and the decrease of actual members of the young cluster in the bd mass range , because of the declining initial mass function ( imf ) at substellar masses ( see below ) .
the contamination from background sources is evident as a large number ( @xmath120200 stars ) of faint sources in our h - r diagram ( figure [ fig : hrd ] ) , which are well separated from the cluster sequence .
given the relatively low @xmath9 for these objects ( @xmath121 k@xmath122 k ) , they could be either cool dwarf stars or red giants , the latter more easily detectable at large distances and through the orion molecular cloud .
the relative number of these sources , relative to young cluster members , becomes dominant at our detection limit
. a small number of these relatively less - luminous sources might be actual onc members , whose flux may be masked ( e.g. , by edge - on circumstellar disks ) , and therefore are visible only in scattered light . this phenomenon has been observed in several young star - forming regions ( see e.g. , * ? ? ? * ; * ? ? ? * and references therein ) ; however the expected fraction of such objects is very small , including no more than a few percent of the population @xcite .
on the other hand , the faint sources we detect below the young sequence are about 16% of the total number of stars with @xmath123 k. therefore , we can assume that most of these sources are actually contaminants .
moreover , removing a small fraction of faint members erroneously considered as background contaminants will not affect significantly the imf derived in this work , given the small expected number of these objects compared to the entire studied population .
we proceed to attempt to remove the background contamination from the h - r diagram . to this purpose
, we define a cut in the h - r diagram such that stars satisfying the following criterion are considered as non - members : @xmath124 this is shown in figure [ fig : contaminants_selection ] .
all sources located below this line are regarded as non - members and excluded from further analysis .
figure [ fig : contaminants_radec ] presents the spatial distribution of contaminants in comparison with that of the candidate members .
whereas the latter are centrally concentrated , with an evident north - south elongation , contaminants appear uniformly distributed .
also , we find a slight overabundance of contaminants on the east side of our field of view .
the reason for this is that the extinction provided by the orion nebula is higher on a central north - south strip , centered on the onc @xcite .
therefore the density of background objects is expected to be higher on the two sides of this band . here
, the nebular emission of the omc is significantly brighter on the west side of our fov , and this decreases the detection limit in this area .
therefore the east edge of our fov is expected to show the most prominent density of background contaminants . in figure
[ fig : hrd_with_models ] we present once again the h - r diagram of the onc , now including only our candidate members , together with pms evolutionary models .
in particular , we use isochrones and tracks from either @xcite or @xcite . these models , unlike e.g.
, the @xcite or the @xcite , also extend into the substellar mass range , and therefore are more suitable for our sample , which reaches masses as low as 0.02 m@xmath5 ( 20m@xmath125 ) . in section [ subsection :
photometric_completeness ] we have derived the photometric completeness of our survey . starting from this result , we now evaluate the completeness function in the h - r diagram .
to do this , we transform the h - r diagram ( i.e. , @xmath9 and @xmath126 ) into our observational space ( i.e. , the 753 , 770 , and @xmath1-band magnitudes ) , and assess the extent to which reddening causes sources of various intrinsic @xmath9 and @xmath126 to be missed in our survey .
we set a uniform , dense grid of @xmath127 and @xmath126 bins in the h - r diagram , limited to the region @xmath128 . for each bin , we apply in reverse the relations derived in section [ subsection : teff_calibration ] and [ subsection : av_calibration ] to derive from @xmath9 and @xmath126 the @xmath59 $ ] , the @xmath52 $ ] , and ( using the bc ) the unreddened @xmath1-band magnitude , @xmath129 .
then , by inverting equation [ equation:770index ] and [ equation : avindex ] , we also derive the magnitudes @xmath130 and @xmath131 .
thus we have associated to each pair of h - r diagram parameters ( @xmath127,@xmath126 ) the triplet of intrinsic magnitudes ( @xmath132@xmath131,@xmath129 ) for @xmath96 .
next we apply reddening to these magnitudes using the empirical reddening distribution of the onc stellar population . as in paper
ii , we limit ourselves to the most luminous half of the onc population , as these stars sample the full range of @xmath10 . with a monte carlo approach ,
we draw 1000 @xmath10 values from the reddening distribution and apply them to the intrinsic magnitudes ( @xmath132@xmath131,@xmath129 ) associated to each point of the h - r diagram . for every @xmath133-th ( @xmath134 ) triplet of reddened magnitudes ,
we compute the photometric completeness @xmath135 , @xmath136 , @xmath137 .
the product of these three , @xmath138 , averaged over the 1000 values of @xmath10 , provides the `` differential reddening normalized '' completeness for that particular bin in the h - r diagram .
the result is shown in figure [ fig : completeness_hrd]a .
we find that the completeness is mainly dependent on @xmath126 , while its dependence on @xmath9 for a given luminosity is small and appears only at low temperatures ( @xmath139 ) .
we note also that the completeness is rather high ( @xmath140 ) at the low - mass end of our catalog , and thus even without correcting for the completeness , the observed number of onc members in the substellar mass range is correct to within a factor of @xmath172 .
our h - r diagram also includes sources whose stellar parameters are taken from paper ii instead of our new medium - band photometry . for all members with @xmath141 , we simply adopt the completeness function in the hrd derived in that paper .
in addition , there is a group of stars with @xmath142 but which are not present in our new medium - band photometry and thus whose stellar parameters are from paper ii .
the presence of these sources increases the overall completeness in this region of the hrd relative to that computed as above . to account for this
, we divide the hrd into a grid with spacing 0.01 dex in @xmath127 and 0.1 dex in @xmath126 , and we count in each bin the relative number of sources from paper ii versus members classified from our new data .
this ratio , smoothed in @xmath127 and @xmath126 , is added to the completeness shown in figure [ fig : completeness_hrd]a . the final result is shown in figure [ fig : completeness_hrd]b .
the evident discontinuity in completeness at @xmath143 is due to the fact that for larger @xmath9 the stellar parameters are obtained from optical spectroscopy on an incomplete fraction of sources .
from the two - dimensional completeness shown in figure [ fig : completeness_hrd]b , we are able to assign a completeness correction to each our sources , allowing us finally to derive a completeness corrected initial mass function for the onc .
@xmath144 & @xmath145 & @xmath146 & @xmath147 + bcah98 all stars & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 + dm98 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 [ table : imf ] to derive the initial mass function ( imf ) of the onc from our observations , we use pre - main - sequence evolutionary models to convert our @xmath9 and @xmath126 from the h - r diagram into masses and ages .
we do this using both the @xcite and @xcite models ( figure [ fig : hrd_with_models ] ) . in the case of baraffe models , for a large fraction ( about 25% ) of sources
we can not assign masses and ages , as these stars are located above the 1 myr isochrone , the minimum age computed for this family of models .
this not only decreases our stellar sample , but also biases our findings , in particular the mass distribution . to overcome this selection effect , we consider two cases for the mass estimates from baraffe models : a ) we reject all source above the 1 myr isochrone ; b ) we include them by assigning a mass based on the @xmath9-mass relation of the 1 myr isochrone . since for very - low mass stars the pms evolutionary tracks are nearly vertical in the h - r diagram , this approximation is fairly good . in table
[ table : parameters ] we present the derived stellar parameters for the onc sources , using both sets of models . from these masses ,
we derive the mass distribution @xmath158 by binning the onc members in equally - spaced mass bins . for each mass bin
, we account for its exact completeness by adding the inverse of the completeness of each source .
we associate an uncertainty distribution to each measured value of @xmath158 equal to a possion distribution of mean @xmath159 , where @xmath160 is the number of sources in the @xmath161th bin , scaled by a factor equal to the overall completeness correction for that bin . we stress that , strictly speaking , our mass function is actually a `` system '' mass function rather than a proper `` initial '' mass function , in the sense that we do not account for unresolved binaries or multiple systems .
this , however , does not influence significantly our results , since the binary fraction ( accounting both bound systems and visual binaries ) in orion is small ( @xmath162 , @xcite ) , and about half of these are separated more than 1 , therefore resolved in our observations .
it it well established ( e.g. , * ? ? ?
* ) that the imf generally follows a power - law in the intermediate- and high - mass range ( @xmath163 ) , whereas for low - mass stars and bds which is the region of the mass spectrum most relevant for our study this function can be approximated with a shallower power law @xmath164 ( e.g. , * ? ? ? * ) or with a log - normal distribution @xmath165 ( e.g. , * ? ? ?
we use both forms to fit our measured imf in the onc .
we use a monte carlo simulation following @xcite to account for the uncertainties in the measured star counts , as follows .
for every mass bin , we consider the error bars with their statistical distribution , and generate a large number ( @xmath166 ) of points drawn from the error distribution .
then an unweighted fit is run on all these ( @xmath167 times the number of bins ) points .
the best - fit parameters are isolated using a levenberg - marquard minimization algorithm .
the uncertainty associated to the parameters has been computed with a sampling technique as follows : for every one of the @xmath168 bins , we randomly consider only one of the @xmath167 values previously simulated , i.e. a random sample from the distribution describing the error bar of the bin , and we fit the imf function on these @xmath168 data points , deriving the best - fit parameters . by iterating this selection and fit process 1000 times , we derive 1000 sets of parameters .
the standard deviation of each parameter for the 1000 tests is the uncertainty in the estimate of the parameter itself .
the resulting fitted functions for both sets of evolutionary models are shown in figure [ fig : imf ] . in table
[ table : imf ] the fitting parameters are reported : the characteristic mass and the width for the log - normal fit , and the two power law slopes as well as the break point for the 2-phase power law . we find that the two families of evolutionary models lead to significant differences in the derived imf . whereas the baraffe tracks produce a smooth distribution , which appears well fitted by a log - normal distribution characterized by a continuous change in the imf slope , the @xcite models lead to a more peaked distribution , with an evident maximum at @xmath169 m@xmath5 .
the reason for this difference can be attributed to the different shapes of the evolutionary tracks in the very low mass range ( @xmath170 m@xmath171 m@xmath5 ) , where the models of @xcite predict a larger area of the hrd covered by adjacent tracks .
the shape of the imf derived from the @xcite models does not depend significantly on which method we adopt for inclusion of very young stars lying above 1 myr isochrone ( see above ) . in figure
[ fig : imf ] we also compare our measured mass distribution with standard reference imfs . in particular we consider the disk imf from @xcite which is expressed as a log - normal function , with a characteristic mass of either 0.079 m@xmath5 or 0.22 m@xmath5 respectively for the `` single objects '' imf or the `` system '' imf , the latter not corrected for stellar multiplicity . also , we consider the universal imf of @xcite , which is a multi - phase power law with shallower slopes in the vlms and bd regime . depending on the assumed evolutionary models and the type of function fitted , our imf peaks in the range 0.1 m@xmath5 @xmath172 0.3 m@xmath5 ,
the higher values being larger than that most often reported , even in the case of unresolved multiplicity .
this finding , as well as the high dependence on the evolutionary model family used to derive stellar masses , is consistent with what we have found in our previous work ( paper i ) limited to the stellar mass range .
a striking feature of our imf is the steep decline in the bd mass range ( see figure [ fig : imf ] ) ; here our mass distribution is much lower than the standard imfs , implying a significant deficiency of substellar objects in the onc .
we stress that this finding is not biased by detection incompleteness , given that the derived mass distribution has been corrected for completeness star by star , and the completeness function we have derived in section [ subsection : completenesshrd ] remains fairly high ( @xmath173 ) even in the bd mass range .
moreover , even hypothesizing an overestimated completeness for the smallest masses in our sample , the relative scarcity of substellar members is evident up to about 0.1 m@xmath5 ; there simply are not many very cool , very low luminosity sources observed .
such a steep decline of the mass function in the bd mass range differs also from imf determinations in other young regions .
in fact the majority of young clusters have been reported to show substellar mass function compatible with the kroupa or the chabrier mass distributions .
some examples are chamaeleon i @xcite , upper sco @xcite , the @xmath26 ori region @xcite , @xmath174 ori @xcite , ngc 6611 @xcite .
similar findings have also been reported for older open clusters ( e.g. , * ? ? ?
* ) ; in fact for the praesepe cluster , @xcite measure even an increasing imf down to @xmath175 m@xmath7 .
previous studies of the imf in the onc also found a decrease in the relative number of members in the substellar mass range , both in absolute number and relative to the standard imfs .
studies based on photometry alone ( e.g. , * ? ? ?
* ; * ? ? ?
* ) using the conversion of near infrared lfs , find a measured imf slope below the h - burning limit not steeper than @xmath176 , which is the shallowest value we derive assuming @xcite models , and up to 2 units flatter than what we derive with @xcite tracks .
also , @xcite finds an imf for the onc rapidly decreasing below the h - burning limit , but then flattening in the bd mass range .
@xcite reports a secondary peak in the mass distribution at about the deuterium burning limit ( @xmath177 m@xmath5 ) ; @xcite tentatively find a similar feature , although with less significance and at a slightly higher mass .
this mass roughly corresponds to our detection limit .
however , from our results , we do not find any evidence of either a flattening of the imf or an increasing number of sources for the lowest mass bins .
we interpret this inconsistency as an inaccurate estimate of the background contamination in previous works , which we believe is much improved through the methodology that we have developed here .
allard , f. , homeier , d. , & freytag , b. 2010 , asp conf .
cool star 16 , _ in press _
( arxiv:1011.5405 ) allard , f. , hauschildt , p. h. , & schwenke , d. 2000 , , 540 , 1005 baraffe , i. , chabrier , g. , allard , f. , et al .
1998 , , 337 , 403 bastian , n. , covey , k. r. , & meyer , m. r. 2010 , , 48 , 339 bayo , a. , barrado , d. , stauffer , j. , et al .
2011 , arxiv:1109.4917 bergin , e. a. , snell , r. l. , & goldsmith , p. f. 1996 , , 460 , 343 bessell , m. s. , castelli , f. , & plez , b. 1998 , , 333 , 231 bouvier , j. , moraux , e. , & stauffer , j. 2005 , the initial mass function 50 years later , 327 , 61 caballero , j. a. 2009 , american institute of physics conference series , 1094 , 912 calvet , n. , & gullbring , e. 1998 , , 509 , 802 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 chabrier , g. 2003 , , 115 , 763 cohen , m. , & kuhi , l. v. 1979 , , 41 , 743 costero , r. , & peimbert , m. 1970 , boletin de los observatorios tonantzintla y tacubaya , 5 , 229 dantona , f. , & mazzitelli , i. 1998 , brown dwarfs and extrasolar planets , 134 , 442 da rio , n. , robberto , m. , soderblom , d. r. , et al .
2010 , , 722 , 1092 da rio , n. , robberto , m. , soderblom , d. r. , et al .
2009 , , 183 , 261 da rio , n. , gouliermis , d. a. , & henning , t. 2009 , , 696 , 528 de marchi , g. , panagia , n. , guarcello , et al . 2010 , _ submitted _ fischer , w. , edwards , s. , hillenbrand , l. , et al .
2011 , , 730 , 73 flower , p. j. 1996 , , 469 , 355 guarcello , m. g. , damiani , f. , micela , g. , et al .
2010 , , 521 , a18 hillenbrand , l. a. 1997 , , 113 , 1733 hillenbrand , l. a. , & carpenter , j. m. 2000 , , 540 , 236 kraus , a. l. , & hillenbrand , l. a. 2009 , , 703 , 1511 kroupa , p. 2002
, science , 295 , 82 kroupa , p. 2001
, , 322 , 231 lodieu , n. , hambly , n. c. , jameson , r. f. , et al .
2007 , , 374 , 372 lucas , p. w. , weights , d. j. , roche , p. f. , et al .
2006 , , 373 , l60 lucas , p. w. , roche , p. f. , & tamura , m. 2005 , , 361 , 211 lucas , p. w. , roche , p. f. , allard , f. , et al .
2001 , , 326 , 695 lucas , p. w. , & roche , p. f. 2000 , , 314 , 858 luhman , k. l. , rieke , g. h. , young , e. t. , et al .
2000 , , 540 , 1016 luhman , k. l. 2007 , , 173 , 104 menten , k. m. , reid , m. j. , forbrich , j. , et al .
2007 , , 474 , 515 meyer , m. r. , calvet , n. , & hillenbrand , l. a. 1997 , , 114 , 288 muench , a. a. , lada , e. a. , lada , c. j. , et al . 2002 , , 573 , 366 muench , a. a. , lada , e. a. , & lada , c. j. 2000 , , 533 , 358 oliveira , j. m. , jeffries , r. d. , & van loon , j. t. 2009 , , 392 , 1034 padgett , d. l. , strom , s. e. , & ghez , a. 1997 , , 477 , 705 palla , f. , & stahler , s. w. 1999 , , 525 , 772 petr , m. g. , coud du foresto , v. , beckwith , s. v. w. , et al .
1998 , , 500 , 825 reipurth , b. , guimares , m. m. , connelley , m. s. , et al .
2007 , , 134 , 2272 reggiani , m. , robberto , m. , da rio , n. , et al .
2011 , , 534 , a83 riddick , f. c. , roche , p. f. , & lucas , p. w. 2007 , , 381 , 1077 robberto , m. , soderblom , d. r. , scandariato g. , et al .
2010 , , 139 , 950 robberto , m. , et al .
2005 , bulletin of the american astronomical society , 37 , # 146.01 salpeter , e. e. 1955 , , 121 , 161 scandariato , g. , robberto , m. , pagano , i. , et al .
2011 , , 533 , a38 siess , l. , dufour , e. , forestini , m. 2000 , , 358 , 593 slesnick , c. l. , hillenbrand , l. a. , & carpenter , j. m. 2004 , , 610 , 1045 stetson , p. b. 1987 , , 99 , 191 vandame , b. , phd thesis , nice university , 2004 wang , w. , boudreault , s. , goldman , b. , et al .
2011 , , 531 , a164 weights , d. j. , lucas , p. w. , roche , p. f. , et al .
2009 , , 392 , 817 | we present a new census of the orion nebula cluster ( onc ) over a large field of view ( @xmath0 ) , significantly increasing the known population of stellar and substellar cluster members with precisely determined properties .
we develop and exploit a technique to determine stellar effective temperatures from optical colors , nearly doubling the previously available number of objects with effective temperature determinations in this benchmark cluster .
our technique utilizes colors from deep photometry in the @xmath1-band and in two medium - band filters at @xmath2 and @xmath3 nm , which accurately measure the depth of a molecular feature present in the spectra of cool stars . from these colors
we can derive effective temperatures with a precision corresponding to better than one - half spectral subtype , and importantly this precision is independent of the extinction to the individual stars .
also , because this technique utilizes only photometry redward of 750 nm , the results are only mildly sensitive to optical veiling produced by accretion . completing our census with previously available data , we place some 1750 sources in the hertzsprung - russel diagram and assign masses and ages down to 0.02 solar masses . at faint luminosities
, we detect a large population of background sources which is easily separated in our photometry from the bona fide cluster members . the resulting initial mass function of the cluster has good completeness well into the substellar mass range , and we find that it declines steeply with decreasing mass .
this suggests a deficiency of newly formed brown dwarfs in the cluster compared to the galactic disk population . |
more often than not the agents of a social system prefer to combine their efforts in order to achieve results that would be otherwise unattainable by a single agent alone .
a relevant role in the organisation of such systems is therefore played by the emerging patterns of collaboration within a group of individuals , which have been widely and thoroughly investigated in the last few decades @xcite . in a collaboration network ,
two individuals are considered to be linked if they are bound by some form of partnership .
for instance , in the case of scientific collaborations , the nodes of the networks correspond to scientists and the relationship between two authors is testified by the fact that they have co - authored one or more papers @xcite .
another well - known example of collaboration network is that of co - starring graphs , where the nodes represent actors and there is a link between two actors if they have appeared in the same movie .
the study of large collaboration systems has revealed the presence of a surprisingly high number of triangles in the corresponding networks @xcite .
this indicates that two nodes with a common neighbour have a higher probability to be linked than two randomly chosen nodes .
this effect , known as _ transitivity _
@xcite , can be easily explained in terms of a basic mechanism commonly referred to as _
triadic closure _
@xcite , according to which two individuals of a collaboration network have a high probability to connect after having been introduced to each other by a mutual acquaintance @xcite .
some other works have pointed out that triadic closure can also explain other empirical features of real - world collaboration networks , including fat - tailed degree distributions and correlations between the degrees of neighbouring nodes @xcite .
another remarkable feature often observed in social and collaboration networks is the presence of meso - scale structures in the form of _ communities _ , i.e. groups of tightly connected nodes which are loosely linked to each other @xcite .
interestingly , structural communities quite often correspond to functional groups @xcite .
an important observation is that not all the links of a collaboration network are equal , since collaborations can often be classified into a number of different categories .
for instance , scientific co - authorship can be classified according to the research field , while actors often appear in movies of different genres . in these cases , a collaboration network is better described in terms of a _ multi - layer _ or _
multiplex network _
@xcite where links representing collaborations of a specific kind are embedded on a separate layer of the network , and each layer can have in general a different topology .
great attention has been recently devoted to the characterisation of the structure @xcite and dynamics @xcite of multi - layer networks .
in particular , various models to grow multiplex networks have appeared in the literature , focusing on linear @xcite or non - linear @xcite preferential attachment , or on weighted networks @xcite .
less attention has been devoted to define and extract communities in multiplex networks @xcite , for instance by mean of stochastic block models @xcite . in this work
we investigate the multiplex nature of communities in collaboration networks and we propose a simple model to explain the appearance , coexistence and co - evolution of communities at the different layers of a multiplex .
our hypothesis is that the formation of communities in collaboration networks is an intrinsically multiplex process , which is the result of the interplay between intra - layer and inter - layer triadic closure .
for instance , in the case of scientific collaborations , multiplex communities naturally arise from the fact that scientists may collaborate with other researchers in their principal field of investigation and with colleagues coming from other scientific disciplines .
analogously , actors can prefer either to specialise in a specific genre or instead to explore different ( sometimes dissonant ) genres , and these two opposite behaviours undoubtedly have an impact on the kind of meso - scale structures observed on each of the layers of of the system .
the generative model we propose here mimics two of the most basic processes that drive the evolution of collaborations in the real world , namely intra- and inter - layer triadic closure , and is able to explain the appearance of overlapping modular organisations in multi - layer systems .
we will show that the model is able to reproduce the salient micro- , meso- and macro - scale structure of different real - world collaboration networks , including the multi - layer network of co - authorship in journals of the american physical society ( aps ) and the multiplex co - starring graph obtained from the internet movie database ( imdb ) .
we start by analysing the structure of two multiplex collaboration networks from the real world .
the first multiplex is constructed from the aps co - authorship data set , and consists of four layers representing four sub - fields of physics ( respectively , nuclear physics , particle physics , condensed matter i , and interdisciplinary physics ) . in particular , we considered only scientists with at least one publication in each of the four sub - fields , and we connected two scientists at a certain layer if they had co - authored at least a paper in the corresponding sub - field .
the second multiplex is constructed from the internet movie database ( imdb ) and consist of four layers respectively representing the co - starring networks of actors with at least one participation in four different genres , namely action , crime , romance , and thriller movies .
the basic structural properties of each layer of the two multiplexes are summarised in table [ tab : exapsimdbsingle ] ( see methods for more information about the data sets ) .
.[tab : exapsimdbsingle]*basic properties of real - world multiplex collaboration networks . *
we report the number of nodes @xmath0 , the average degree @xmath1 and the clustering coefficient @xmath2 for each layer of a subset of the aps and imdb data sets . in particular , we focus on the multiplex collaboration network of all scientists active in nuclear , particle , condensed matter i and interdisciplinary physics , and the multiplex collaboration network of all actors starring in action , crime , romance and thriller movies
. all the layers of aps have a clustering coefficient @xmath2 in the range @xmath3 $ ] .
conversely , the values of @xmath2 of all the imdb layers are in the range @xmath4 $ ] . [ cols="<,<,<,<",options="header " , ]
human collaboration patterns are inherently multifaceted and often consist of different interaction layers .
scientific collaboration is probably the most emblematic example . as a ph.d .
student you usually join the scientific collaboration network by publishing the first paper with your supervisor in a specific field .
afterwards , you start being introduced by your supervisor to other researchers in the same field , e.g. to some of his / her past collaborators , and you might end up working with them , creating new triangles in the collaboration network of your field ( what we called intra - layer triadic closure ) .
but it is also quite probable that some of your past collaborators will in turn introduce you to researchers working in another -possibly related- area ( what we called an inter - layer triadic closure ) , so that you will easily find yourself participating in more than just one field , and the collaboration network around you will become multi - dimensional .
such multi - level collaboration patterns appear not to be specific of scientific production only , but are instead found in many aspects of human activity .
the multi - layer network framework provides a natural way of modelling and characterising multidimensional collaboration patterns in a comprehensive manner .
in particular , we have argued that one of the classical mechanisms responsible for the creation of triangles of acquaintances , i.e. triadic closure , is indeed general enough to give also account for another interesting aspect of multi - level collaboration networks , namely the formation of cohesive communities spanning more than a single layer of interaction .
it is quite intriguing that the simple model we proposed in this work , based just on the interplay between intra- and inter - layer triadic closure , is actually able to explain much of the complexity observed in the micro- meso- and macroscopic structure of multidimensional collaboration networks of different fields ( science and movies ) , including not just transitivity but also intra- and inter - layer degree correlation patterns and the correspondence between the community partitions at difference layers .
we also remark that such levels of accuracy in reproducing the features of real - world systems have been obtained without the introduction of ad - hoc ingredients .
the results reported in this paper suggest that , despite the apparent differences in the overall dynamics driving scientific cooperation and movie co - starring , triadic closure is a quite generic mechanism and might indeed be one of the fundamental processes shaping the structure of multi - layer collaboration systems .
these findings fill a gap in the literature about modelling growing multidimensional networks , and pave the way to the exploration of other simple models which can help underpinning the driving mechanisms responsible for the emergence of complex multi - dimensional structures .
_ data sets . _ we considered data from the aps and the imdb collaboration networks .
the aps collaboration data set is available from the aps website ` http://journals.aps.org/datasets ` in the form of xml files containing detailed information about all the papers published by all aps journals .
the download is free of charge and restricted to research purposes , and aps does not grant to the recipients the permission to redistribute the data to third parties .
we parsed the original xml files to retrieve , for each paper , the list of authors and the list of pacs codes .
the pacs scheme provides a taxonomy of subjects in physics , and is widely used by several journals to identify the sub - field , category and subject of papers .
we used the highest level of pacs codes to identify the ten main sub - fields of physics , and we considered only the papers published in nuclear physics , particle physics , condensed matter i and interdisciplinary physics , respectively associated to high - level pacs codes starting with 1 ( particle physics ) , 2 ( nuclear physics ) , 6 ( condensed matter i ) and 8 ( interdisciplinary physics ) .
we focused only on the authors who had at least one publication in each of the four sub - fields @xcite .
the co - authorship network of each of those four sub - fields constitutes one of the four layers of the aps multiplex .
in particular , two authors are connected on a certain layer only if they have co - authored at least one paper in the corresponding sub - field . in the construction of the collaboration network of each sub - field
we purposely left out papers with more than ten authors , which represent big collaborations whose driving dynamics might be more complex than just triadic closure .
the imdb data set is made available at the website ` ftp://ftp.fu-berlin.de/pub/misc/movies/database/ ` for personal use and research purposes .
the data set comes in the form of several compressed text files , and we used those containing information about actors , actresses , movies and genres .
we focused only on the co - starring networks of four movie genres , namely action , crime , romance , and thriller @xcite , obtained by merging information about participation of actors and actresses to each movie .
in particular , two actors are connected by a link on a given layer ( genre ) only if they have co - starred in at least one movie of that genre .
we considered only the actors who had acted in at least one movie of each of the four genres .
we chose to restrict our analysis to just four layers for both the aps and the imdb data set , which allowed us to consider the simplest formulation of our model , in which all the layers have the same clustering coefficient @xmath2 .
the use of the aps and the imdb data sets does not require any ethical approval .
_ transitivity and community structure .
_ we measured the transitivity of each level by mean of the clustering coefficient @xmath5 @xcite , where @xmath6 : @xmath7 the similarity of two community partitions can be measured through the normalised mutual information ( nmi ) @xcite . in particular , given the two partitions @xmath8 and @xmath9 respectively associated to layer @xmath10 and layer @xmath11 , we denote the normalised mutual information ( nmi ) between them as @xmath12 where @xmath13 is the number of nodes in common between module @xmath14 of partition @xmath8 and module @xmath15 of partition @xmath16 , while @xmath17 and @xmath18 are respectively the number nodes in module @xmath14 and in module @xmath15 .
the partition in communities on each layer has been obtained through the algorithm infomap @xcite .
_ synthetic multiplex networks .
_ we created synthetic networks according to our multi - layer network model by starting , on each layer , from a seed graph consisting of a triangle of nodes and simulating the intra- and inter - layer triadic closure mechanism for @xmath19 nodes , for different values of the parameters @xmath20 and @xmath21 . for each pair of values
@xmath22 we computed the mean clustering coefficient @xmath2 on each single layer and the normalised mutual information nmi of the community partitions of the two layers over 30 different realisations .
as observed from simulations , once the parameters @xmath23 are fixed , the values of nmi and @xmath2 do not vary substantially as the order @xmath0 of the network increases .
notice that since the most simple formulation of the model we have set an identical value of @xmath20 on both layers , the two layers will end up having the same clustering coefficient ( up to small finite - size fluctuation ) .
_ degree correlations .
_ we study the assortativity of real multiplex collaboration networks in terms of intra - layer , inter - layer and mixed degree correlations .
the trend for intra - layer correlations is analysed by mean of the function @xmath24}_{nn}(k^{[\alpha ] } ) \rangle$ ] , that is the average degree of the nearest neighbours on layer @xmath10 of a node with given degree @xmath25}}$ ] on that layer . in particular , @xmath24}_{nn } \rangle$ ] is obtained as an average of @xmath26}_{nn , i}$ ] over all nodes with the same degree @xmath27}$ ] .
the node term can be computed as @xmath26}_{nn , i}=\frac{\sum_{j \neq i } a^{[\alpha]}_{ij}k_j^{[\alpha]}}{k_i^{[\alpha]}}$ ] , where @xmath28}_{ij}$ ] are the entries of the adjacency matrix at layer @xmath10 . since such measure considers only a layer at a time , the layer index here is not strictly necessary but will be kept for symmetry with the other coefficients .
it is interesting to notice that , in absence of intra - layer degree correlations , @xmath24}_{nn}(k^{[\alpha ] } ) \rangle$ ] is a constant , while @xmath24}_{nn}(k^{[\alpha ] } ) \rangle$ ] is an increasing ( resp . , decreasing )
function of @xmath25}}$ ] if assortative ( resp .
, disassortative ) degree correlations are present . to quantify inter - layer degree correlations we considered the quantity @xmath29}(k^{[\alpha ] } ) \rangle$ ] @xcite , that is the average degree on layer @xmath11 of a node with degree @xmath27}$ ] on layer @xmath10 .
again , @xmath29}(k^{[\alpha ] } ) \rangle$ ] will be an increasing function of @xmath25}}$ ] if nodes tend to have similar degrees on both layers ( assortative inter - layer correlations ) , while @xmath30}(k^{[\alpha ] } ) \rangle$ ] will decrease with @xmath25}}$ ] if a hub on one layer will preferentially have small degree on the other layer , and vice - versa . finally , we measured the presence of mixed correlations through the function @xmath31}_{nn}(k^{[\alpha ] } ) \rangle$ ] , that is the average degree on layer @xmath11 of the nearest neighbours on layer @xmath11 of a node with degree @xmath27}$ ] on layer @xmath10 @xcite . in analogy with the case of intra - layer correlations , the node term is @xmath32}_{nn , i}=\frac{\sum_{j \neq i } a^{[\beta]}_{ij}k_j^{[\beta]}}{k_i^{[\alpha]}}$ ] .
we remark here that there exists another possible definition of mixed correlations coefficient , which considers the nearest neighbours of a node on layer @xmath10 rather then @xmath11 ( see ref .
@xcite for details ) .
the results for the alternative definition of mixed correlations are analogous to those observed for @xmath33}_{nn}(k^{[\alpha ] } ) \rangle$ ] and are not shown in the text .
in general , correlation functions might be affected by the degree sequence at each layer of the multiplex . in the simple scenario considered at first , however
, we do not fit the parameter @xmath14 from the data , to reduce as much as possible the complexity of the model . instead , in order to still perform an accurate comparison between the synthetic multiplex networks constructed by our model and the real ones , in a second step we divided all the correlation functions by their ( constant ) value expected in the corresponding configuration model network .
the correct normalisation for the intra - layer correlation function is @xmath34}})^2 \rangle}}{{\langle k{^{[\alpha ] } } \rangle}}$ ] @xcite , while for the inter - layer correlation function we have to divide @xmath29}(k^{[\alpha ] } ) \rangle$ ] by @xmath35 } } \rangle}$ ] . finally , the mixed correlation function is correctly normalised by @xmath36}})^2 \rangle}}{{\langle k{^{[\alpha ] } } \rangle}}$ ]
f.b . , v.n . and v.l .
acknowledge support by the project lasagne , contract no.318132 ( strep ) , funded by the european commission .
de domenico m , lancichinetti a , arenas a , rosvall m. identifying modular flows on multilayer networks reveals highly overlapping organization in interconnected systems .
physical review x. 2015;5(1):011027 .
iacovacci j , wu z , bianconi g. mesoscopic structures reveal the network between the layers of multiplex data sets .
phys rev e. 2015 oct;92:042806 .
available from : http://link.aps.org/doi/10.1103/physreve.92.042806 . | community structures in collaboration networks reflect the natural tendency of individuals to organize their work in groups in order to better achieve common goals . in most of the cases ,
individuals exploit their connections to introduce themselves to new areas of interests , giving rise to multifaceted collaborations which span different fields . in this paper , we analyse collaborations in science and among movie actors as multiplex networks , where the layers represent respectively research topics and movie genres , and we show that communities indeed coexist and overlap at the different layers of such systems .
we then propose a model to grow multiplex networks based on two mechanisms of intra and inter - layer triadic closure which mimic the real processes by which collaborations evolve .
we show that our model is able to explain the multiplex community structure observed empirically , and we infer the strength of the two underlying social mechanisms from real - world systems . being also able to correctly reproduce the values of intra - layer and inter - layer assortativity correlations , the model contributes to a better understanding of the principles driving the evolution of social networks . |
one of the challenges faced by galaxy formation models is to explain the population of today s red massive quiescent elliptical galaxies . in the current hierarchical galaxy formation paradigm
massive galaxies are formed via mergers of less massive galaxies which in turn fuel intense star formation and feed massive black holes .
one of the main difficulties is to find mechanisms to stop the processes of intense star formation , and to allow galaxies to migrate from the so - called blue cloud or late - type star forming galaxies to the so - called red sequence or early - type quiescent galaxies ( see e.g. , bell et al .
2004 , 2007 ) .
feedback from active galactic nuclei ( agn ) has been proposed as an efficient process for suppressing any further star formation in the late stages of galaxy evolution , while still allowing for continuing black hole growth ( see e.g. , springel , di matteo , & hernquist 2005 and croton et al .
see hopkins et al .
( 2007 , and references therein ) for a detailed discussion on this and other related issues .
some tantalizing evidence of the possible role of agn in galaxy evolution is the location of local optically selected agn ( e.g. , salim et al .
2007 ; martin et al . 2007 ) and moderate-@xmath10 x - ray selected agn ( snchez et al . 2004 ; nandra et al . 2007 ) in the transition between the red sequence and the top of the blue cloud , the region also known as the green valley. these intermediate colors may indicate that agn play a role in causing or maintaining the quenching of star formation .
however , in the local universe agn with strongly accreting black holes tend to be hosted in massive galaxies with blue ( i.e. , star - forming ) disks and young bulges ( kauffmann et al .
2003b , 2007 ) implying a close link between the growth of black holes and bulges .
clearly the relationship between agn and star - formation is a matter of strong debate .
about half of the sources with x - ray luminosities @xmath11 ( i.e. , suggestive of the presence of a moderately luminous agn ) detected in deep ( @xmath12ms ) x - ray surveys do not show broad lines or high excitation lines characteristic of agn in their optical spectra ( see e.g. , barger et al . 2001 ; cohen 2003 ; szokoly et al .
2004 and review by brandt & hasinger 2005 ) . since the agn emission of these optically - dull agn does not dominate their rest - frame uv to near - infrared ( nir ) emission ( rigby et al .
2006 ) , they are the ideal targets to study their host galaxies and investigate the role of agn in galaxy evolution .
in this paper we study the host galaxies of x - ray selected agn with stellar dominated spectral energy distributions ( seds ) at intermediate redshifts ( @xmath13 ) in the _ chandra _ deep field south ( cdf - s ) using uv , optical , nir , and _ spitzer
the agn host galaxy properties are then compared with those of irac - selected ( i.e. , stellar mass selected ) galaxies at similar distances studied by prez - gonzlez et al .
( 2008 ) . throughout this work we assumed the following cosmology : @xmath14 , @xmath15 and @xmath16
we first started with all the cdf - s x - ray sources ( giacconi et al .
2002 ; alexander et al . 2003 ) with spectroscopic and photometric redshifts in the range of @xmath13 ( zheng et al .
then we restricted ourselves to those x - ray sources in the great observatories origins deep survey ( goods ) portion of the cdf - s region which has the deepest _
irac ( fazio et al . 2004 ) and mips ( rieke et al . 2004 ) observations ( see prez - gonzlez et al . 2008 ) .
we cross - correlated the positions of the x - ray sources with the irac ( simultaneous detections at @xmath17 and @xmath18m)-selected galaxies of prez - gonzlez et al .
( 2008 ) using a separation of @xmath19arcsec .
the 75% completeness limits for the cdf - s field in this sample are @xmath1jy and @xmath20jy at @xmath21 m and @xmath18 m , respectively .
we also used the cdf - s photometric catalogs of prez - gonzlez et al .
( 2008 ) to construct the seds of the x - ray sources .
these catalogs include the two other irac bands , uv , optical , nir , and _ spitzer_/mips @xmath22 m data ( see prez - gonzlez et al . 2005 , 2008 for a complete description of the dataset and source matching ) .
the mips @xmath22 m catalog is 75% complete down to @xmath23jy . we finally cross - correlated the x - ray sources with the _
spitzer_/mips @xmath24 m catalog for the cdf - s ( papovich et al .
2007 ) which is @xmath25 complete for sources with flux densities down to @xmath26mjy . out of the 112 x - ray sources with irac detections in the field described above , we selected for this study agn with stellar - dominated uv through nir seds , and in particular with a strong @xmath1 m bump .
our selection thus excluded x - ray sources with agn - dominated seds such as ir power - law galaxies ( alonso - herrero et al .
2006 ; donley et al .
2007 ) , irac color - color selected agn ( lacy et al . 2004
; stern et al . 2005 ) and galaxies without a prominent @xmath1 m bump ( e.g. , daddi et al . 2007 ) .
we also removed from our sample x - ray sources with nearby companions of similar brightness which could contaminate the observed mid - ir ( mir ) seds , in particular the irac bands .
the final sample contains 58 agn .
most ( 52 of the 58 ) agn in our sample have spectroscopic redshifts and type classifications ( szokoly et al .
2004 ; vanzella et al .
2006 ) . the majority ( 46 out of the 52 )
are classified as optically - dull , that is , agn without any evidence for accretion from their optical spectra .
these included agn classified as low excitation and absorption line agn ( see szokoly et al .
2004 ) . the remaining 6 agn with spectroscopic information show high excitation lines or broad lines ( szokoly et al .
2004 ) , and will be referred to as optically - active agn . for the 6 agn in our sample without spectroscopic information we estimated photometric redshifts ( see next section ) .
the rest - frame seds of the 58 agn are shown in figure 1 for two different redshift bins : @xmath27 and @xmath28 , which correspond to approximately similar ranges of cosmic times for the assumed cosmology .
the selected agn have rest - frame absorption - corrected hard ( @xmath29kev ) x - ray luminosities above @xmath30 ( from tozzi et al .
figure 2 compares the x - ray column densities with the absorption - corrected hard x - ray luminosities ( from tozzi et al .
2006 ) for all the x - ray sources ( open circles ) detected by irac in the goods field .
we marked the agn with stellar - dominated seds selected for this study as filled circles .
2 clearly shows that a large fraction of agn with hard x - ray luminosities below @xmath31 have seds dominated by stellar emission ( see also donley et al .
2007 ) . at higher hard x - ray luminosities
only those agn with large x - ray column densities have stellar seds ( see also , polletta et al . 2006 , 2007 ) . at low absorptions ( @xmath32 )
the agn emission becomes more apparent in the integrated seds , but there are some x - ray sources with low @xmath33 with stellar - dominated seds .
these sources will be studied in more detail in 4.1 .
the x - ray properties of our sample of agn ( fig . 2 ) are consistent with recent findings on the nature of optically - dull agn .
rigby et al . ( 2006 ) estimated that in @xmath34 of optically - dull agn the agn emission lines could be diluted by the stellar emission from the host galaxy ( see also moran , filippenko , & chornock 2002 ) , whereas in the rest extinction from the host galaxy may be responsible for hiding the agn optical lines .
the latter conclusion was based on the comparison of the inclination angle distributions of the host galaxies of optically - dull and optically - active agn .
recently , caccianiga et al . ( 2007 ) proposed that the most likely explanation for optical dullness at low x - ray luminosities is dilution by a massive host galaxy , while at high x - ray luminosities the dullness is due to absorption .
this section describes briefly ( see prez - gonzlez et al . 2008 for a full description ) the procedure for modelling the seds at wavelengths @xmath35 m .
this modelling is general to the irac - selected sample of prez - gonzlez et al .
( 2008 ) from which we extracted the stellar masses ( this section ) and sfrs ( next section ) for our sample of agn .
the method involves a two step process in which the galaxies with spectroscopic redshifts are fitted first , and then used as templates to fit the seds ( and photometric redshifts ) of those galaxies without spectroscopic redshifts . for galaxies with spectroscopic redshifts
the stellar emission is generated with the pegase code ( fioc & rocca - volmerange 1997 ) assuming a salpeter imf ( salpeter 1955 ) between 0.1 and 100m@xmath36 .
we assumed that the stellar emission of the galaxies can be described with one or two stellar populations . in the case of the one stellar population model
the star formation rate ( sfr ) is modeled with a declining exponential ( @xmath37 ) .
the four free parameters to fit are : extinction ( using the calzetti et al .
2000 law ) , metallicity , the time scale of the exponential law ( @xmath38 ) , and the age ( @xmath39 ) of the stellar population . in the case of two stellar populations ,
the old stellar population is described as for the one stellar population model ( four free parameters ) , and the young stellar population is assumed to have been formed in an instantaneous burst with three free parameters : extinction , metallicity , and age .
for the two stellar population models the burst strength is another free parameter that relates the mass of young stars with the total mass of the galaxy .
in addition to the stellar emission , we include the hydrogen gas emission in the form of nebular continuum and emission lines .
once the galaxies with spectroscopic redshifts are fitted , they are used as templates for fitting the seds of those galaxies without a spectroscopic redshift . with the best model and best photometric redshift established ,
the stellar mass of the galaxy is obtained by scaling the models to the observed sed .
the value of the stellar mass ( @xmath2 ) is computed as the average of the stellar masses computed for each observed photometric band . although most ( 52 out of 58 ) of the agn in our sample have spectroscopic redshifts ( szokoly et al .
2004 and vanzella et al .
2006 ) , we refitted their seds using the library of templates .
the stellar masses derived with the photometric redshift were then rescaled to the spectroscopic redshift . as discussed in great detail by prez - gonzlez et al .
( 2008 ) there are a variety of effects that can introduce systematics in the determination of stellar masses .
in addition to the intrinsic uncertainties associated with fitting the seds ( typically a factor of @xmath40 ) , other effects include the choice of stellar population libraries , extinction law , and imf , as well as the number of stellar populations ( in our case , one population vs. two populations ) .
these effects introduce uncertainties of the order of or smaller than those intrinsic to the sed fitting .
even though we selected our agn such that their seds are clearly dominated by stellar emission , a valid concern is the possible effect of the agn emission in the determination of the stellar masses . in particular ,
such concern was raised by daddi et al .
( 2007 ) when estimating the stellar masses of @xmath41 galaxies with mir excesses thought to host obscured agn .
these authors argued that the possible agn contribution in the form of hot dust at @xmath42 m might cause overestimation of the stellar masses , but only at a modest level and generally within the systematic errors . daddi et al .
( 2007 ) estimated the stellar masses using empirical calibrations based on @xmath43 , @xmath10 , and @xmath44-band photometry .
we believe our method is less susceptible to the effects of agn because we use population synthesis models to fit the entire observed seds , and the stellar masses are computed as the average of the stellar masses in all the photometric points .
prez - gonzlez et al .
( 2008 ) discussed in detail the effects of the presence of a hot dust component ( which could be associated with an agn ) in the determination of the stellar masses of agn .
they concluded that the agn effects on the estimated stellar masses are negligible for x - ray sources with observed ( i.e. , not corrected for absorption ) luminosities @xmath45 ( see also donley et al .
all the agn in our sample are well below this limit .
in this section , we evaluate whether the mir emission of our sample of agn is likely to be due to the agn or to star formation .
we then use the _ spitzer _ data to model the total ir ( @xmath46 m ) luminosity ( @xmath47 ) and to derive sfrs .
the x - ray column density can be used to classify agn into type 1 ( direct view of the agn , @xmath48 ) and type 2 ( obscured view of the agn ) .
rigby et al . ( 2004 ) using local agn templates showed that for low column densities the @xmath22 m to hard x - ray flux ratio should remain approximately constant , whereas for high column densities this ratio varies by factors of a few .
we test whether the @xmath22 m emission can be solely explained as produced by dust heated by the putative agn , or whether an additional mechanism ( i.e. , star formation ) is needed .
this is tested for the eighteen optically - dull agn ( that is , those without evidence for accretion from the optical spectra ) in our sample with @xmath48 .
to do so , we scaled the median quasar template of elvis et al .
( 1994 ) to the @xmath29kev flux ( from tozzi et al . 2006 ) for each of the 18 optically - dull agn with low @xmath33 .
the predicted agn @xmath22 m flux densities were then compared with observed values .
we only find five optically - dull agn with low @xmath33 ( those in the left panel of fig .
3 ) for which between 30 and 100% of their observed @xmath22 m flux densities could be accounted for with the predictions from their agn hard x - ray fluxes .
these five agn show the highest x - ray fluxes among the optically - dull agn with low @xmath33 . for the rest ( including some galaxies with @xmath49 m excesses ) , the predicted agn @xmath22 m flux densities are below approximately 15% of the observed value , and thus most of their mir emission is probably produced by star formation . in our analysis
, we will assume that star formation dominates the @xmath22 m emission of these galaxies and of those that are similar except for larger absorbing columns for their x - ray sources . for comparison the middle and right panels of fig .
3 show a few examples of seds of optically - active agn with @xmath50 , and their corresponding scaled quasar templates .
the middle panel includes three optically - active agn in our sample ( szokoly 52 , 53 , and 78 ) .
only the mir emission of one of them ( szokoly 53 ) is fully accounted for by the predictions from the agn x - ray flux . from the right panel of fig .
3 , it is clear that the agn signatures ( uv bump and hot dust emission around @xmath51 m and beyond ) become dominant for the most x - ray luminous agn ( see e.g. , barmby et al .
2006 ; polletta et al . 2006 , 2007 ; donley et al .
2007 ) .
after the stellar emission was modeled , as described in 3 , the predicted stellar fluxes were subtracted from the observed photometric data points at @xmath52 m . the resulting dust emission out to the mips @xmath22 m photometric point
was then fitted using the chary & elbaz ( 2001 ) models to derive the ir luminosity . for the @xmath22 m non - detections ( 10 out of the 58 agn in our sample ) ,
the @xmath47 was computed assuming an upper limit to the @xmath22 m flux density of @xmath53jy , which is the 50% completeness limit of our catalog .
the unobscured star formation is assumed to be traced by the uv monochromatic 2800 luminosity , as fitted by the stellar model .
the total sfrs were computed as the sum of the ir and uv luminosities converted to sfrs using the prescriptions of kennicutt ( 1998 ) .
although we do not use the mips @xmath24 m photometric points to model the ir luminosities , we can check if the @xmath54 ratios or upper limits are consistent with star formation , as assumed .
only 3 optically - dull agn are detected at @xmath24 m , all of them in the @xmath55 bin ( see papovich et al . 2007 for more details ) .
none of these three sources with @xmath24 m detections have observed @xmath56 ratios consistent with those expected from hot dust arising from a @xmath57 constant distribution ( @xmath58 , similar to optically selected quasars , elvis et al .
1994 see fig .
3 , or agn - dominated sed , see e.g. , alonso - herrero et al .
neither are these colors consistent with those of ir bright agn such as mrk 231 ( @xmath59 , for our redshift range ) .
the behavior of the @xmath56 ratio of optically - dull agn is however similar to other sources ( both detected and undetected in x - rays ) in the same redshift range ( see papovich et al .
2007 ) , as well as to empirical templates of local star - forming galaxies . since most of the optically - dull agn with low absorptions have only modest x - ray luminosities ( @xmath60 , see fig .
2 ) , we consider the possibility that they do not contain an agn , and thus that their x - ray luminosities could be produced by star formation . to do so , we can compare our total uv+ir sfrs with the sfrs predicted from their hard x - ray luminosities using the relation inferred by ranalli , comastri , & setti ( 2003 ) . in all but one galaxy ( szokoly 577 , which is one of the galaxies with high @xmath61 ratios )
the x - ray based sfrs would be between 3 and 10 times higher than our inferred sfr@xmath62 , indicating that in these galaxies most of the hard x - ray emission is not produced by star formation , and that indeed they contain ( a low luminosity ) agn . in fig . 1
the observed seds ( normalized at @xmath63 m ) are shown for the two redshift bins , and are sorted according to their fitted @xmath61 ratios .
if the agn contribution to the mir emission is small , these ratios are a good proxy for the specific sfrs ( i.e. , star formation rate per unit stellar mass ) . from the modelling of the stellar seds ( see 3 ) we find that most galaxies with high @xmath61 ratios are fitted with younger models than those with low @xmath61 ratios
. an independent way to estimate the average ages of the host galaxy stellar populations is to measure the 4000- break ( see kauffmann et al . 2003a , b and references therein ) .
the optical spectra ( from szokoly et al .
2004 ) of some @xmath27 agn with high and low @xmath61 ratios ( as in fig . 1 ) are shown in figure 4 .
the average @xmath64 values for the agn in figure 4 with high and low @xmath61 ratios are @xmath65 and @xmath66 , respectively .
thus , there is some marginal evidence for the agn host galaxies with the lowest @xmath61 ratios to show larger 4000- breaks ( i.e. , older stellar populations ) than galaxies with high @xmath61 ratios .
these @xmath64 values are similar to those of high-@xmath10 agn identified by kriek et al .
( 2007 ) .
the measured @xmath64 values for our sample of agn indicate relatively young ages , of between 0.8 and 2.5gyr for an instantaneous burst ( using figure 2 of kauffmann et al .
these ages , as well as the ages of local universe type-2 agn ( kauffmann et al .
( 2003b ) and intermediate-@xmath10 type 1 agn ( snchez et al . 2004 ) are younger than those of local quiescent massive galaxies , but consistent with the ages of irac - selected galaxies of similar masses ( see prez - gonzlez et al . 2008 ) . those galaxies in our sample ( about one - third ) with high @xmath61 ratios ( left panels of fig .
1 ) tend to show an excess of @xmath67 m emission over the expectations of the stellar emission for an evolved stellar population .
although in most cases these mir excesses can be accounted for by the gas + stellar emission associated with a young stellar population , we can not rule out some contribution from warm dust emission associated with the putative agn . by our selection criteria ( i.e. , prominent @xmath1 m stellar bump )
, the mir excesses in our galaxies occur at @xmath68 m , unlike the mir excess galaxies of daddi et al . ( 2007 ) and ir power - law galaxies ( alonso - herrero et al . 2006 ; donley et al .
2007 ) , where the mir excesses appear at @xmath69 m .
however , a large fraction ( two - thirds ) of the sample show no mir excesses out to @xmath70 m , with their mir emission being entirely consistent with that of an evolved stellar populations .
thus in these galaxies the agn emission appears completely buried out to @xmath71 m and possibly longer wavelengths .
figure 5 shows the stellar masses versus redshift for our sample of agn compared with the distribution of stellar masses for the irac - selected sample of galaxies of prez - gonzlez et al .
( 2008 ) . at the redshifts probed here
the irac - selected comparison sample is essentially a stellar mass selected sample .
clearly ( x - ray identified ) agn reside in galaxies with a range of about an order of magnitude in mass , including some among the most massive at these intermediate redshifts .
the characteristic stellar masses ( measured as the median of the distributions ) are @xmath72 ( 36 galaxies ) at @xmath27 ( median redshift of 0.67 ) and @xmath73 ( 22 galaxies ) at @xmath74 ( median redshift of 1.07 ) .
we also plot in fig . 5 as solid lines the redshift evolution of the quenching mass ( mass
above which , star formation should be mostly suppressed ) inferred by bundy et al .
( 2006 ) using two different methods .
the fraction of agn above the line is small perhaps indicating that star formation has not been fully suppressed yet in these galaxies ( but see also 5.2 ) . and specific sfrs similar or below the median specific sfrs for irac - selected galaxies ( see more details in 5.2 ) .
the open circles are the stellar masses for the six optically - active agn in our sample with stellar seds .
the dotted line indicates the completeness limit of the sample of prez - gonzlez et al .
( 2008 ) for a maximally - old passively evolving galaxy .
the two solid lines are two empirical determinations ( color and morphology ) of the redshift evolution of the quenching mass ( converted to a salpeter imf ) of bundy et al .
the solid lines reflect two different criteria used by bundy et al .
( 2006 ) to evaluate star formation : color and morphological type . _
lower panel : _ redshift evolution of the stellar mass of irac - selected galaxies ( prez - gonzlez et al .
2008 ) in the same field as our sample of cdf - s agn .
the irac includes all the agn plotted in the upper panel .
only galaxies with @xmath75 are plotted in this comparison.,title="fig:",width=321 ] and specific sfrs similar or below the median specific sfrs for irac - selected galaxies ( see more details in 5.2 ) .
the open circles are the stellar masses for the six optically - active agn in our sample with stellar seds .
the dotted line indicates the completeness limit of the sample of prez - gonzlez et al .
( 2008 ) for a maximally - old passively evolving galaxy .
the two solid lines are two empirical determinations ( color and morphology ) of the redshift evolution of the quenching mass ( converted to a salpeter imf ) of bundy et al .
the solid lines reflect two different criteria used by bundy et al .
( 2006 ) to evaluate star formation : color and morphological type . _
lower panel : _ redshift evolution of the stellar mass of irac - selected galaxies ( prez - gonzlez et al .
2008 ) in the same field as our sample of cdf - s agn .
the irac includes all the agn plotted in the upper panel .
only galaxies with @xmath75 are plotted in this comparison.,title="fig:",width=321 ] it is important to stress that the agn studied here comprise @xmath34 of the x - ray selected agn population at @xmath76 . a pressing question is whether the derived stellar masses are representative of the overall population of x - ray selected agn .
one possibility is that agn with stellar - dominated seds ( mostly optically - dull agn ) might be hosted by massive galaxies ( e.g. , moran et al .
2002 ; severgnini et al .
2003 ; caccianiga et al .
2007 ) causing the agn emission lines to be buried by the galaxy emission . to test this possibiliy , in fig . 6 , which is similar to fig . 2 ,
the agn in our sample are sorted according to their stellar masses .
there is no clear tendency for the most x - ray luminous sources to be hosted by the most massive galaxies . as demonstrated by rigby et al .
( 2006 ) extinction on large scales produced by the host galaxy might also be responsible for hiding the agn lines in these galaxies .
ideally we would like to estimate the stellar masses for all optically - active agn at intermediate redshifts , but this becomes increasingly more uncertain ( see prez - gonzlez et al .
2008 ) , as for more luminous x - ray sources the agn emission in the optical - nir becomes more dominant ( e.g. , barmby et al .
2006 ; polletta et al . 2006 , 2007 , donley et al . 2007 ;
see also next section ) .
the masses of the six optically - active agn in our sample do not appear to be fundamentally different from optically - dull agn ( see fig . 5 ) , although the number statistics is very small . another way to determine
if the stellar masses of optically - dull agn are representative is to compare their absolute magnitudes with those of optically - active agn . if both types of agn reside in similar systems , then optically - active agn should be more luminous in the optical and near - ir because they should have the contributions from the host galaxy and the agn .
rigby et al . ( 2006 , their figure 8) already made this comparison , and found that blagn with @xmath77 tend to show brighter optical magnitudes than optically - dull agn of similar luminosities . again
the number statistics are small , but both comparisons seem to suggest that the stellar masses of optically - dull agn are representative of the whole x - ray selected agn population at these redshifts . optically selected type-2 agn at @xmath78 are also found to reside in the most massive galaxies .
kauffmann et al .
( 2003b ) inferred a characteristic stellar mass of @xmath79 ( for a salpeter imf ) for bright type 2 agn ( see also heckman et al .
2004 ) . at high redshift ( @xmath41 )
x - ray selected agn appear to be hosted by even more massive galaxies ( @xmath80 ) , although the mass estimates have only been done for a few galaxies ( see borys et al . 2005 ; daddi et al .
2007 , but also caveats discussed by alexander et al . 2007 ) .
similarly kriek et al .
( 2007 ) find that @xmath81 agn identified in @xmath44-band selected galaxies are hosted by very massive galaxies , but as they point out it is likely that their results are biased towards the most massive objects .
the stellar masses of agn at @xmath82 and @xmath83 are thus intermediate between those of local agn and high-@xmath10 agn . since the stellar masses of our sample of agn appear to be representative of the whole population of x - ray selected agn ( see fig .
5 ) then we infer that approximately 25% of massive galaxies ( @xmath84 ) at the redshifts probed here contain an x - ray identified agn .
we also took into account agn not included in this study , assuming that they have stellar masses similar to the agn with stellar dominated seds .
this fraction is similar for the two redshift bins considered here , and consistent with the agn fraction in massive galaxies at higher redshifts ( @xmath85 , alexander et al .
2005 ; daddi et al . 2005 ; papovich et al . 2006 ; kriek et al .
our estimated agn fraction in intermediate-@xmath10 massive galaxies is only a lower limit because presumably most compton - thick agn are missed by current x - ray surveys . at @xmath41
the fraction of agn in massive galaxies appears to be as high as @xmath86 ( daddi et al .
2007 ) consistent with the view that the massive black hole growth was higher when the universe was younger . and @xmath87 mass ranges , respectively .
we also show the quartiles of the distribution of specific sfrs of irac - selected galaxies for each of the two mass ranges as the shaded regions .
the median specific sfrs for agn are the filled squares and triangles plotted at the median redshift .
the horizontal bars represent the ranges of redshifts .
we excluded agn with mir excesses or classified as optically - active .
, width=298 ] in this section we quantify the star formation activity of agn with stellar - dominated seds in relation to irac - selected galaxies of similar stellar masses and at similar redshifts .
we use the specific sfrs as an indicator of the star formation activity rather than the absolute sfrs , as the specific sfr measures the rate at which new stars add to the assembled mass of the galaxy ( brinchmann & ellis 2000 ) .
moreover , galaxies show distinct specific sfrs depending on their stellar masses and redshifts ( e.g. , brinchmann et al .
2004 ; zheng et al .
2007 ; prez - gonzlez et al . 2008 ) , so the comparison needs to be made taking this into account .
figure 7 compares the median specific sfrs for intermediate-@xmath10 agn with those of the irac - selected sample of prez - gonzlez et al .
in this comparison we excluded mir - excess galaxies and the six optically - active agn , as they may have an important agn contribution to their observed mir emission which would cause us to overestimate their ir - based sfrs . the comparison between agn and irac - selected galaxies
is done for the two relevant stellar mass ranges of the agn hosts ( see figure 5 ) : @xmath88 and @xmath89 .
as can be seen from figure 7 the agn specific sfrs ( at both redshift and mass intervals ) do not appear to be fundamentally different from those of irac - selected galaxies .
moreover , zheng et al .
( 2007 ) argued against agn feedback as the main process for quenching star formation .
their model predicted that agn feedback would be more effective for more massive galaxies ( see also croton et al .
this would imply a more rapid decline of the specific sfr for massive galaxies ( @xmath90 ) than less massive galaxies ( see also springel et al .
2005 ) , which is observed neither for the zheng et al .
( 2007 ) sample nor for the irac - selected sample of prez - gonzlez et al .
( 2008 , see fig .
in contrast , kriek et al . ( 2007 ) found evidence for a relation between the suppression of star formation and the agn phase for @xmath44-band selected galaxies .
kriek et al .
( 2007 ) however pointed out that their agn sample may be biased toward quiescent galaxies where agn are easier to identify .
in fact , if we consider their two samples of galaxies ( uv and @xmath44-band selected ) then the agn have specific sfrs within the range observed for their non - agn . in the local universe agn
tend to be hosted in massive galaxies with younger stellar ages than non - agn of similar morphological types ( early - type ) and stellar masses ( kauffmann et al . 2003a , 2007 ) .
this was interpreted as evidence that enhanced star formation is a requisite for feeding the agn . at @xmath91 galaxies ( presumably both agn and non - agn ) with @xmath92
are still being assembled ( see prez - gonzlez et al .
2008 ) , so perhaps it is not surprising that their star formation rates are not significantly different .
there are a number of caveats when trying assess the star formation activity of our sample of agn .
as discussed in previous sections , there is the exact agn contribution to the mir emission , although it is found generally not to be large ( 4.1 ) .
also the agn in the @xmath28 redshift range tend to be more luminous in x - rays so the agn become more apparent . even though we excluded galaxies with mir excesses ( as we can not rule out that they might be due to dust heated by the agn ) , the specific sfrs are formally upper limits .
another largely unknown effect in the comparison between agn and irac - selected galaxies is the possible contamination by x - ray identified agn as well as obscured agn in the most massive ( @xmath93 ) systems ( see 5.1 and daddi et al .
2007 ; kriek et al . 2007 ) in the comparison sample .
another concern is the possibility that the obscuring material , which may be responsible in part for the optical dullness , be associated with star formation activity within the host galaxy ( e.g. , ballantyne , everett , & murray 2006 ; martnez - sansigre et al .
this would bias our sample towards star - forming galaxies when compared to the most optically - active agn not included in our sample .
8 shows a comparison between the derived uv+ir sfrs and the x - ray absorption for our sample of agn .
this figure suggests a weak , if any , connection between star formation and obscuration .
it is also clear that a number of highly obscured agn ( @xmath94 ) do not present high sfrs , and in these cases the obscuration is probably not associated with extranuclear dust in the host galaxy .
summarizing , at intermediate-@xmath10 we do not find strong evidence for either highly suppressed star formation activity or increased star formation activity in agn when compared to irac - selected galaxies of similar stellar masses .
we only observe that the most massive galaxies with low specific sfrs ( see fig .
5 ) follow the redshift evolution of the quenching mass inferred by bundy et al .
this may only indicate that the most massive systems hosting an agn are close to being fully assembled at these redshifts .
the relation between the black hole mass and the bulge luminosity and in particular the bulge stellar mass is now well estalished in the local universe ( e.g. , merritt & ferrarese 2001 ; marconi & hunt 2003 ) .
this relationship appears to hold out to @xmath91 ( peng et al .
2006 ) , indicating that massive bulges were fully assembled at this redshift ( see e.g. , glazebrook et al . 2004
; cimatti et al . 2004 ; papovich et al . 2006 ; prez - gonzlez et al .
2008 ) . since the agn activity out to @xmath95 seems to be associated with bulge - dominated galaxies ( see snchez et al .
2004 ; grogin et al . 2005
; pierce et al .
2007 ) we can use the local relationship and our stellar masses to estimate the black hole masses of intermediate-@xmath10 agn . using the marconi & hunt ( 2003 ) relation and assuming that @xmath96
, we find black hole masses for typical agn of approximately between @xmath97 and @xmath98 , with a median value of @xmath99 .
these are in good agreement with the estimates of babi et al .
( 2007 ) for cdf - s @xmath100 agn based on stellar masses and stellar velocity dispersions .
the resulting @xmath101 of optically - dull agn are plotted in figure 9 against the absorption - corrected rest - frame hard x - ray luminositiesm luminosities .
if most of the mir emission was produced by dust heated by an agn , one would expect a proportionality between the absorption corrected x - ray luminosities ( a proxy for the agn luminosity ) and the mir luminosities , which is not observed in general for optically - dull agn ( see also rigby et al .
this again suggests that a large fraction of the mir emission relative to the agn luminosity could be due to star formation . ] . in figure 9
we also show the eddington ratios ( @xmath102 ) calculated using the bolometric corrections ( @xmath103 ) derived for pg quasars by elvis et al .
optically - dull agn at @xmath13 with x - ray luminosities above @xmath104 have eddington ratios close to broad - line agn ( @xmath105 , barger et al .
however for x - ray luminosities below @xmath104 ( comparable to local seyfert galaxies ) the eddington ratios are much lower ( @xmath8 , similar to the findings of babi et al . 2007 ) .
one possibility for the low eddington ratios is that we were overestimating significantly the black hole masses , for instance , if these galaxies were not bulge - dominated .
however , the corrections to the bulge masses ( and thus black hole masses ) would be of the order of @xmath106 to @xmath107 for s0/a and sc types , respectively ( see e.g. , marconi et al .
2004 ; dong & de robertis 2006 ) , assuming that the nir luminosity traces the stellar mass .
recently ballo et al . ( 2007 ) derived bulge magnitudes for x - ray sources in the cdf - s and found that in general in the acs @xmath10-band the bulge to total luminosity ratios are of beween 0.4 and 1 . for the two souces in our sample in common theirs these ratios are @xmath108 . from this , the black hole masses quoted here should be taken as upper limits and eddington ratios lower limits , although the corrections are likely to be relatively small .
the agn hosts in our sample follow very well the redshift evolution of the quasar host mass derived by hopkins et al .
( 2007 ) from the quasar optical luminosity function .
this is interesting because our agn are not x - ray quasars ( i.e. , @xmath109 , see fig . 2 ) .
this is also indirect evidence that the accretion rates of optically - dull agn are lower than those of bright quasars . in 5.1 on the other hand we found that the characteristic stellar masses of intermediate-@xmath10 are between those of local agn and high-@xmath10 ( @xmath110 ) agn .
this would appear to support the interpretation of heckman et al .
( 2004 ) for the agn downsizing phenomenon . in this scenario ,
the fact that the space density of low luminosity agn peaks at lower redshifts than more luminous agn ( see ueda et al .
2003 ) is explained by a decrease of the characteristic mass of actively accreting black holes ( and thus , stellar mass of the hots ) rather than a decreasing accretion rate ( see also barger et al .
since it is these low x - ray luminosity agn that dominate the x - ray background at the redshifts considered here , the agn downsizing is not only due to a decrease in the characteristic stellar mass ( see 5.1 ) , but also lower eddington ratios .
we studied a sample of 58 x - ray selected agn at intermediate redshifts ( @xmath111 ) in the goods portion of the cdf - s .
the agn were selected such that their rest - frame uv to nir seds were dominated by stellar emission , and in particular , showed a prominent @xmath1 m bump .
this selection minimized the agn contamination which is essential for studying the properties of their host galaxies . since agn with stellar dominated seds
comprise approximately 50% of the population of x - ray selected agn , they are a cosmologically important class of agn .
we fitted their rest - frame uv through mir seds using stellar and dust models to derive the stellar masses as well as the total ( uv+ir ) sfrs .
as previously discussed by other works ( e.g. , severgnini et al .
2003 ; rigby et al . 2006
; caccianiga et al . 2007 ) , the optical dullness might be due to various causes .
we find that dilution by a massive host galaxy is only in part responsible , as the mass of the host galaxy is independent of the x - ray luminosity and absorption of the agn .
extinction on large scales may also play a role , and presumably is relatively prevalent in the lower - mass optically dull agn hosts . from this and the derived stellar masses of a few optically - active agn with stellar seds
we conclude that the derived stellar masses of our sample of agn are representative of the entire x - ray selected agn population at these redshifts .
about one - third of our agn show @xmath112 m excesses above the expected stellar emission from an old stellar population together with high @xmath61 ratios .
although these mir excesses could be interpreted as evidence for the putative agn ( i.e. , hot dust ) , these galaxies tend to have smaller 4000 - breaks .
that is , galaxies with mir excesses have younger average stellar populations ( an indication of recent or on - going star formation ) than galaxies with low @xmath61 ratios .
this may indicate that agn emission is responsible for most of the mir emission only in some galaxies . for the rest of the sample
there is no evidence for agn emission out to approximately @xmath70 m , and their mir emission is likely to be produced mostly by star formation .
x - ray identified agn are found to reside in galaxies with a range of stellar masses ( @xmath113 , for a salpeter imf ) , including the most massive galaxies at intermediate redshifts .
we infer characteristic ( median ) stellar masses of @xmath3 and @xmath4 at median redshifts of 0.67 and 1.07 , respectively .
these stellar masses are intermediate between those of local type 2 agn ( @xmath114 , kauffmann et al .
2003b ) and high-@xmath10 agn ( @xmath115 , kriek et al .
2007 ; daddi et al .
2007 ; alexander et al . 2007 ) . from the comparison with the irac - selected sample of prez - gonzlez et al .
( 2008 ) , we find that approximately 25% of massive ( @xmath116 ) galaxies at these redshifts contain an x - ray identified agn .
using the local relation between @xmath117 and @xmath101 of marconi & hunt ( 2003 ) the inferred black hole masses ( @xmath6 ) are similar to those of broad - line agn although with lower eddington ratios ( @xmath118 ) than luminous quasars .
both findings suggest that , at least at intermediate redshifts , the cosmic agn downsizing is probably due not only to a decrease in the characteristic stellar mass of the host galaxy ( as proposed by heckman et al .
2004 ) , but also to less efficient accretion , as also found by babi et al .
( 2007 ) .
finally , we do not find strong evidence in agn host galaxies for either highly suppressed star formation ( expected if agn played a role in quenching star formation ) or intense star formation when compared to irac - selected galaxies of similar stellar masses and redshifts .
this can be understood if we take into account that the host galaxies of agn ( and non - agn of similar mass ) are still being assembled at the redshifts probed here .
the main caveats regarding this conclusion are the possible influence on it of the agn contribution to the observed mir emission ( although it is found generally not to be large ) , and the possible bias towards intense star formation which could be in part responsible for obscuring the agn in these objects .
the authors would like to thank a. marconi , p. hopkins , and m. kriek for interesting discussions .
the authors also thank an anonymous referee for useful suggestions that helped improve the paper .
this work was supported by nasa through contract 1255094 issued by jpl / california institute of technology .
aa - h acknowledges support from the spanish plan nacional del espacio under grant esp2005 - 01480 .
pgp - g acknowledges support from the ramn y cajal fellowship program financed by the spanish government , and from the spanish programa nacional de astronoma y astrofsica under grant aya 200602358 . this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
alexander , d. m. et al .
2003 , aj , 126 , 539 alexander , d. m. , smail , i. , bauer , f. e. , chapman , s. c. , blain , a. w. , brandt , w. n. , & ivison , r. j. 2005 , nature , 434 , 738 alexander , d. m. et al .
2007 , apj , submitted alonso - herrero , a. et al .
2006 , apj , 640 , 167 babi , a. , miller , l. , jarvis , m. j. , turner , t. j. , alexander , d. m. , & croom , s. m. et al .
2007 , a&a , submitted ballantyne , d. , everett .
j. e. , & murray , n. 2006 , apj , 639 , 740 ballo , l. et al .
2007 , apj , 667 , 97 balogh , m. l. et al .
1999 , apj , 527 , 54 barger , a. j. et al .
2001 , aj , 126 , 632 barger , a. j. et al .
2005 , aj , 129 , 578 barmby , p. et al .
2006 , apj , 642 , 126 bell , e. f. et al .
2004 , apj , 608 , 752 bell , e. f. et al .
2007 , apj , 663 , 834 borys , c. , smail , i. , chapman , s. c. , blain , a. w. , alexander , d. m. , & ivison , r. j. 2005 , apj , 635 , 853 brandt , w. n. & hasinger , g. 2005 , ara&a , 43 , 827 brinchmann , j. & ellis , r. s. 2000 , apj , 536 , l77 brinchmann , j. et al .
2004 , mnras , 351 , 1151 bundy , k. et al .
2006 , apj , 651 , 120 caccianiga , a. , severgnini , p. , della ceca , r. , maccacaro , t. , carrera , f. j. , & page , m. j. 2007 , a&a , 470 , 557 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , apj , 533 , 682 caputi , k. i. et al .
2006 , apj , 637 , 727 chary , r. , & elbaz , d. 2001 , apj , 556 , 562 cimatti , a. et al .
2004 , nature , 230 , 184 cohen , j. g. 2003 , apj , 598 , 288 croton , d. j. et al .
2006 , mnras , 365 , 111 daddi , e. et al . 2005 , apj , 626 , 680 daddi , e. et al .
2007 , apj , 670 , 173 dong , x. y. , & de robertis , m. m. 2006 , aj , 131 , 1236 donley , j. l. , rieke , g. h. , prez - gonzlez , p. g. , rigby , j. r. , & alonso - herrero , a. 2007 , apj , 660 , 167 elvis , m. et al .
1994 , apjs , 95 , 1 fazio , g. g. et al . 2004 ,
apjs , 154 , 10 fioc , m. , & rocca - volmerange , b. 1997 , a&a , 326 , 950 giacconi , r. et al .
2002 , apjs , 139 , 369 glazebrook , k. et al .
2004 , nature , 430 , 181 grogin , n. a. et al .
2005 , apj , 627 , l97 heckman , t. m. , kauffmann , g. , brinchmann , j. , charlot , s. , tremonti , c. , & white , s. m. d. 2004 , mnras , 613 , 109 hopkins , p. f. , bundy , k. , hernquist , l. , & ellis , r. s. 2007 , apj , 659 , 976 kauffmann , g. et al . 2003a , mnras , 341 , 33 kauffmann , g. et al . 2003b , mnras , 346 , 1055 kauffmann , g. et al .
2007 , apjs , in press ( astro - ph/0609436 ) kennicutt , r.c .
1998 , ara&a , 36 , 189 kriek , m. et al .
2007 , apj , 669 , 776 kuraszkiewicz , j. k. et al .
2003 , apj , 590 , 128 lacy , m. et al .
2004 , apjs , 154 , 166 marconi , a. , & hunt , l. k. 2003 , apj , 589 , l21 marconi , a. , risaliti , g. , gilli , r. , hunt , l. k. , maiolino , r. , & salvati , m. 2004 , mnras , 351 , 169 martin , d. c. et al .
2007 , apjs , in press ( astro - ph/0703281 ) martnez - sansigre , a. et al . 2006 ,
mnras , 370 , 1479 merritt , d. , & ferrarese , l. 2001 , mnras , 320 , l30 moran , e. c. , filippenko , a. v. , & chornock , r. 2002 , apj , 579 , l71 nandra , k. et al .
2007 , apj , 660 , l11 norman , c. et al . 2004 , apj , 607 , 721 papovich , c. et al .
2006 , apj , 640 , 92 papovich , c. et al .
2007 , apj , 668 , 45 peng , c. y. , impey , c. d. , ho , l. c. , barton , e. j. , & rix , h .- w . 2006 , apj , 640 , 114 prez - gonzlez , p. g. et al .
2005 , apj , 630 , 82 prez - gonzlez , p. g. et al .
2008 , apj , 672 , in press ( astro - ph/0709.1354 ) pierce , c. m. et al .
2007 , apj , 660 , l19 polletta , m. c. et al . 2006 , apj , 642 , 673 polletta , m. c. et al .
2007 , apj , 663 , 81 ranalli , p. , comastri , a. , & setti , g. 2003 , a&a , 399 , 39 rieke , g. h. et al .
2004 , apjs , 154 , 25 rigby , j. r. et al .
2004 , apjs , 154 , 160 rigby , j. r. , rieke , g. h. , donley , j. l. , alonso - herrero , a. , & prez - gonzlez , p. g. 2006 , apj , 645 , 115 salim , s. et al .
2007 , apjs , in press ( astro - ph/07043611 ) salpeter , e. e. 1955 , apj , 121 , 161 snchez , s. f. et al .
2004 , apj , 614 , 586 severgnini , p. et al .
2003 , a&a , 406 , 483 springel , v. , di matteo , t. , & hernquist , l. 2005 , mnras , 361 , 776 stern , d. et al .
2005 , apj , 631 , 163 szokoly , g. p. et al .
2004 , apjs , 155 , 271 tozzi , p. et al .
2006 , a&a , 451 , 457 ueda , y. , akiyama , m. , ohta , k. , & miyaji , t. 2003 , apj , 598 , 886 vanzella , e. et al .
2006 , a&a , 434 , 53 zheng , w. et al .
2004 , apjs , 155 , 73 zheng , x. z. et al .
2007 , apj , 661 , l41 | we study the stellar and star formation properties of the host galaxies of 58 x - ray selected agn in the goods portion of the _ chandra _ deep field south ( cdf - s ) region at @xmath0 .
the agn are selected such that their rest - frame uv to near - infrared spectral energy distributions ( seds ) are dominated by stellar emission , i.e. , they show a prominent @xmath1 m bump , thus minimizing the agn emission contamination. this agn population comprises approximately 50% of the x - ray selected agn at these redshifts . using models of stellar and dust emission we model their seds to derive stellar masses ( @xmath2 ) and total ( uv+ir ) star formation rates ( sfr ) .
we find that agn reside in the most massive galaxies at the redshifts probed here .
their characteristic stellar masses ( @xmath3 and @xmath4 at median redshifts of 0.67 and 1.07 , respectively ) appear to be representative of the x - ray selected agn population at these redshifts , and are intermediate between those of local type 2 agn and high redshift ( @xmath5 ) agn .
the inferred black hole masses ( @xmath6 ) of typical agn are similar to those of optically identified quasars at similar redshifts .
since the agn in our sample are much less luminous ( @xmath7 ) than quasars , typical agn have low eddington ratios ( @xmath8 ) .
this suggests that , at least at intermediate redshifts , the cosmic agn downsizing is due to both a decrease in the characteristic stellar mass of typical host galaxies , and less efficient accretion . finally there is no strong evidence in agn host galaxies for either highly suppressed star formation ( expected if agn played a role in quenching star formation ) or elevated star formation when compared to mass selected ( i.e. , irac - selected ) galaxies of similar stellar masses and redshifts .
this may be explained by the fact that galaxies with @xmath9 are still being assembled at the redshifts probed here . |
we introduce @xmath0-viscosity solutions for the nonlinear parabolic problem @xmath2 we prove an estimate between viscosity solutions and @xmath0-viscosity solutions of ( [ eq : main ] ) under the assumption that the nonlinearity @xmath3 is uniformly elliptic ( see ( f[ellipticity ] ) below ) . as a consequence
, we find a rate of convergence for monotone and consistent implicit finite difference approximations to ( [ eq : main ] ) .
both results generalize the work of caffarelli and souganidis in @xcite and @xcite , who consider the time - independent case .
the nonlinearity @xmath3 is a continuous function on @xmath4 , where @xmath5 is the set of @xmath6 real symmetric matrices endowed with the usual order and norm ( for @xmath7 , @xmath8 ) .
we make the following assumptions : 1 .
[ ellipticity ] @xmath3 is uniformly elliptic with constants @xmath9 , which means that for any @xmath10 and for all @xmath11 , @xmath12 and , 2 . [
last prop of f ] @xmath13 .
we also assume : 1 .
[ assum u ] @xmath14 is a bounded subset of @xmath15 .
the main result is an error estimate between viscosity solutions and @xmath0-viscosity solutions of ( [ eq : main ] ) .
the definition of @xmath0-viscosity solutions is in section [ subsection : delta def ] .
next we present a statement of our main result that has been simplified for the introduction ; the full statement is in section [ section : comparison ] .
[ thm : vague comparison ] assume ( f[ellipticity ] ) , ( f[last prop of f ] ) and ( u1 ) .
assume @xmath16 is a viscosity solution of ( [ eq : main ] ) that is lipschitz continuous in @xmath17 and hlder continuous in @xmath18 .
assume @xmath19 is a family of hlder continuous @xmath0-viscosity solutions of ( [ eq : main ] ) that satisfy @xmath20 for all @xmath0 .
there exist positive constants @xmath21 and @xmath22 that do not depend on @xmath0 such that for all @xmath0 small enough , @xmath23 we remark that if @xmath16 is the solution of a boundary value problem with sufficiently regular boundary data , then the regularity conditions on @xmath16 of theorem [ thm : vague comparison ] are satisfied ( see remark [ remaboutlips ] for more details ) .
we make this an assumption in order to avoid discussing boundary value problems and to keep the setting as simple as possible .
the notion of @xmath0-viscosity solutions was introduced by caffarelli and souganidis in @xcite as a tool for obtaining uniform estimates for viscosity solutions . in @xcite ,
@xmath0-viscosity solutions were used to establish an error estimate for finite difference approximations of nonlinear uniformly elliptic equations . an error estimate between viscosity and @xmath0-viscosity solutions of uniformly elliptic equations
was obtained by caffarelli and souganidis in @xcite .
this was an important step in establishing a rate for homogenization in random media @xcite .
the main challenge in obtaining an error estimate between viscosity and @xmath0-viscosity solutions , in both the elliptic and parabolic setting , is overcoming the lack of regularity of the viscosity solution @xmath16 .
indeed , the proof of the error estimate in @xcite is based on a regularity result ( * ? ? ?
* ; * ? ? ?
* theorem a ) , which says that outside of sets of small measure , solutions of uniformly elliptic equations have second - order expansions with controlled error .
we prove a similar result for solutions of equation ( [ eq : main ] ) .
this is theorem [ thm a ] of this paper , and it is an essential part of our proof of theorem [ thm : vague comparison ] . but even once it is known that the solution @xmath16 of ( [ eq : main ] ) has second - order expansions with controlled error , it is necessary to further regularize @xmath16 and the @xmath0-viscosity solution @xmath24 in order establish an estimate for their difference .
for this we use the classical inf- and sup- convolutions , along with another regularization of inf - sup type , which we call @xmath17-sup- and @xmath17-inf- convolutions . both of these regularizations preserve the notions of viscosity and @xmath0-viscosity solution
. moreover , we show that the regularity we establish in theorem [ thm a ] for solutions @xmath16 of ( [ eq : main ] ) is also enjoyed by the @xmath17-inf and @xmath17-sup convolutions of @xmath16 ( proposition [ prop : more regularity for convolutions ] ) . in this paper
we also study implicit finite difference approximations to ( [ eq : main ] ) and prove an error estimate between the solution @xmath16 of ( [ eq : main ] ) and approximate solutions to ( [ eq : main ] ) .
we have simplified the notation in the introduction in order to state this result here ; see section [ section : approx ] for all the details about approximation schemes and for the full statement of the error estimate .
we write the finite difference approximations as @xmath25(x , t ) = \delta^-_{\tau } u ( x , t ) - \mathcal{f}_h(\delta^2 u(x , t))=0 \text { in } ( \omega \times ( 0,t))\cap e.\ ] ] here @xmath26 is the mesh of discretization , @xmath27 $ ] is the implicit finite difference operator , and @xmath28 and @xmath29 are finite difference quotients associated to a function @xmath16 . we assume : 1 .
[ monotone ] if @xmath30 and @xmath31 are solutions of ( [ eq : main h ] ) with @xmath32 on the discrete boundary of @xmath33 , then @xmath32 in @xmath33 ; and 2 .
[ consistent ] there exists a positive constant @xmath34 such that for all @xmath35 and all @xmath36 , @xmath37|\leq k(h+ h{\ensuremath{||d_x^3\phi||_{l^{\infty}(\omega\times ( 0,t ) ) } } } + h^2{\ensuremath{||\phi_{tt}||_{l^{\infty}(\omega\times ( 0,t))}}}).\ ] ] schemes that satisfy ( s[monotone ] ) and ( s[consistent ] ) are said to be , respectively , _ monotone _ and _ consistent with an error estimate for @xmath3_. we prove : [ thm : mesh comparison vague ] assume ( f[ellipticity ] ) , ( f[last prop of f ] ) and ( u1 ) .
assume that @xmath38 is a monotone implicit approximation scheme for ( [ eq : main ] ) that is consistent with an error estimate .
assume @xmath16 is a viscosity solution of ( [ eq : main ] ) that is lipschitz continuous in @xmath17 and hlder continuous in @xmath18 and that @xmath39 is a hlder continuous solution of ( [ eq : main h ] ) .
assume that for each @xmath36 , @xmath40 there exist positive constants @xmath21 and @xmath22 that do not depend on @xmath41 such that for every @xmath41 small enough , @xmath42 the convergence of monotone and consistent approximations of fully nonlinear second order pde was first established by barles and souganidis @xcite .
kuo and trudinger later studied the existence of monotone and consistent approximations for nonlinear elliptic and parabolic equations and the regularity of the approximate solutions @xmath43 ( see @xcite ) .
they showed , both in the elliptic and in the parabolic cases , that if @xmath3 is uniformly elliptic , then there exists a monotone finite difference scheme @xmath38 that is consistent with @xmath3 , and that the approximate solutions @xmath39 are uniformly equicontinuous .
however , obtaining an error estimate remained an open problem .
the first error estimate for approximation schemes was established by krylov in @xcite and @xcite for nonlinearities @xmath3 that are either convex or concave , but possibly degenerate .
krylov used stochastic control methods that apply in the convex or concave case , but not in the general setting .
barles and jakobsen in @xcite and @xcite improved krylov s error estimates for convex or concave equations . in @xcite krylov improved the error estimate to be of order @xmath44 , but still in the convex / concave case .
in addition , jakobsen @xcite and bonnans , maroso , and zidani @xcite established error estimates for special equations or for special dimensions .
the first error estimate without a convexity or concavity assumption was obtained caffarelli and souganidis in @xcite for nonlinear elliptic equations . to our knowledge ,
theorem [ thm : mesh comparison vague ] is the first error estimate for nonlinear parabolic equations that are neither convex nor concave .
our paper is structured as follows . in section [ preliminaries ]
we establish notation , give the definition of @xmath0-viscosity solutions , and state several known results . in section [ section : regularity ]
we prove theorem [ thm a ] , the regularity result that is essential for the rest of our arguments . in section [ section : two different inf- and sup- convolutions ] we discuss regularizations of inf - sup type and establish proposition [ prop : more regularity for convolutions ] , a version of theorem [ thm a ] for @xmath17-sup and @xmath17-inf convolutions . in section [ section : main prop ] we prove a key estimate between viscosity solutions of ( [ maineqn ] ) and sufficiently regular @xmath0-solutions of ( [ maineqn ] ) .
theorems [ thm : vague comparison ] and [ thm : mesh comparison vague ] are straightforward consequences of this estimate . in section [ section : comparison ]
we give the precise statement and proof of theorem [ thm : vague comparison ] .
section [ section : approx ] is devoted to introducing the necessary notation and stating known results about approximation schemes .
the precise statement and proof of the error estimate for approximation schemes is in section [ section : error est approx ] .
we also include an appendix with known results about inf- and sup- convolutions .
in this section we establish notation , recall the definition of viscosity solutions for parabolic equations , give the definition of @xmath0-viscosity solutions , and recall some known regularity results for solutions of ( [ eq : main ] ) .
points in @xmath45 are denoted @xmath46 with @xmath47 .
the parabolic distance between two points is @xmath48 we denote the usual euclidean distance in @xmath45 by @xmath49 . throughout the argument we consider parabolic cubes , denoted by @xmath50^n\times(t - r^2,t],\ ] ] and forward and backward cylinders , @xmath51 \text { and } y^-_r ( x , t ) = b_{r}(x ) \times ( t - r^2 , t],\ ] ] where @xmath52 is the open ball in @xmath15 .
we write @xmath53 , @xmath54 , @xmath55 and @xmath56 to mean @xmath57 , @xmath58 , @xmath59 and @xmath60 , respectively . for @xmath61 ,
the parabolic boundary of @xmath62 is defined as @xmath63 the time derivative of a function @xmath16 is denoted by @xmath64 or @xmath65 .
the gradient of @xmath16 with respect to the space variable @xmath66 is denoted by @xmath67 , and the hessian of @xmath16 with respect to @xmath17 is denoted by @xmath68 .
we say that a constant is _ universal _ if it is positive and depends only on @xmath69 , @xmath70 and @xmath71 . unless otherwise indicated , all constants depend on @xmath69 , @xmath70 and @xmath71 .
we introduce notation for several families of paraboloids .
let @xmath72 .
we define : 1 . the class of _ convex paraboloids of opening @xmath73 _ : @xmath74 2 . the class of _ concave paraboloids of opening @xmath73 _ : @xmath75 3 . the class of paraboloids of arbitrary opening : @xmath76 4 .
the set of polynomials that are quadratic in @xmath17 and linear in @xmath18 with the only mixed term of the form @xmath77 : @xmath78 we will make use of the following seminorms , norms , and function spaces : the class of continuous functions on @xmath79 is denoted @xmath80 .
the class @xmath81 is the set of functions @xmath82 that are differentiable in @xmath18 and twice differentiable in @xmath17 , with @xmath83 and @xmath84 . for @xmath85 $ ] and @xmath86 ,
we define : @xmath87_{c^{0,\eta}(\omega\times ( a , b ) ) } } } & = \sup_{(x , t ) , ( y , s)\in \omega\times ( a , b)}\frac{|u(x , t)-u(y , s)|}{d((x , t),(y , s))^\eta},\\ [ u]_{c^{0,\eta}_x(\omega\times ( a , b))}&= \sup_{x , y\in \omega ; t\in ( a , b ) } \frac{|u(x , t)-u(y , t)|}{|x - y|^\eta},\\ { \ensuremath{[u]_{c_t^{0,\eta}(\omega\times ( a , b ) ) } } } & = \sup _ { t , s\in ( a , b ) ; x\in \omega } \frac{|u(x , t)-u(x , s)|}{|t - s|^\eta } , \text { and,}\\ { \ensuremath{||u||_{c^{0,\eta}(\omega\times ( a , b))}}}&= { \ensuremath{||u||_{l^{\infty}(\omega\times ( a , b))}}}+{\ensuremath{[u]_{c^{0,\eta}(\omega\times ( a , b))}}}.\end{aligned}\ ] ] for @xmath85 $ ] , we define : @xmath88_{c^{0,\eta}_x(\omega\times(a , b))}+ { \ensuremath{||u||_{l^{\infty}(\omega\times ( a , b ) ) } } } < \infty\right\}.\end{aligned}\ ] ] we will also need the following notion of differentiability : for a constant @xmath34 , we say that @xmath16 satisfies @xmath89 ( resp .
@xmath90 ) _ in the sense of distributions _ if there exists a quadratic polynomial @xmath91 such that @xmath92 and @xmath93 ( resp .
@xmath94 ) for all @xmath95 , for some @xmath96 .
we recall the definition of viscosity solutions for parabolic equations .
[ defn visc ] 1 .
a function @xmath97 is a _ viscosity supersolution _ of ( [ eq : main ] ) if for all @xmath98 , any @xmath99 with @xmath100 on @xmath101 for some @xmath102 and @xmath103 satisfies @xmath104 2 .
a function @xmath97 is a _ viscosity subsolution _ of ( [ eq : main ] ) if for all @xmath98 , any @xmath105 with @xmath106 on @xmath101 for some @xmath102 and @xmath103 satisfies @xmath107 3 .
we say that @xmath97 is a _ viscosity solution _ of ( [ eq : main ] ) if @xmath16 is both a sub- and a super- solution of ( [ eq : main ] ) . in the above definitions
, we require the test function @xmath82 to stay above ( or below ) @xmath16 on a backward cylinder @xmath108 . however , this definition is equivalent to the usual one , which requires the test function to stay above ( or below ) @xmath16 on a euclidean open set around @xmath46 .
this equivalence is proven by crandall , kocan , and wich ( * ? ? ?
* lemma 1.4 ) for @xmath109 viscosity solutions .
their proof carries over into our setting with no modifications .
see section 8 of crandall , ishii , and lions @xcite for further discussion of viscosity solutions of parabolic equations .
we now introduce @xmath0-viscosity solutions for ( [ eq : main ] ) , following the definition of @xcite and @xcite .
[ defn delta ] fix @xmath110 . 1 . a function @xmath111 is a _
@xmath0-viscosity supersolution _ of ( [ eq : main ] ) if for all @xmath98 such that @xmath112 , any @xmath113 with @xmath114 on @xmath115 and @xmath116 satisfies @xmath117 2 .
a function @xmath111 is a _
@xmath0-viscosity subsolution _ of ( [ eq : main ] ) if for all @xmath98 such that @xmath112 , any @xmath113 with @xmath118 on @xmath115 and @xmath116 satisfies @xmath119 3 .
we say @xmath120 is a _
@xmath0-viscosity solution _ of ( [ eq : main ] ) if it is both a @xmath0-viscosity sub- and super- solution .
from now on we will say solution " to mean viscosity solution and @xmath0-solution " to mean @xmath0-viscosity solution . from the definitions ,
it is clear that a viscosity solution of ( [ eq : main ] ) is a @xmath0-solution of ( [ eq : main ] ) .
the main difference between the definitions of viscosity and @xmath0-viscosity solution is that for @xmath120 to be a @xmath0-supersolution ( resp .
subsolution ) , any test polynomial must stay below ( resp . above ) @xmath120 on a set of fixed size .
we introduce the pucci extremal operators , which are defined for constants @xmath121 and @xmath7 by @xmath122 where the @xmath123 are the eigenvalues of @xmath124 ( see caffarelli and cabre @xcite and wang @xcite ) .
we also introduce the so - called upper and lower monotone envelopes of a function , which play the role of the convex and concave envelopes in the regularity theory of elliptic equations .
we follow the notes of imbert and silvestre in ( * ? ? ?
* section 4 ) for our presentation of definition [ defn env ] , lemma [ lem : representation ] and proposition [ pre abp ] below .
[ defn env ] let @xmath14 be a convex subset of @xmath15 and let @xmath125 be continuous . 1
. the _ lower monotone envelope _ of @xmath16 is the largest function @xmath126 that lies below @xmath16 and is non - increasing with respect to @xmath18 and convex with respect to @xmath17 .
it is often denoted by @xmath127 .
the _ upper monotone envelope _ of @xmath16 is the smallest function @xmath126 that lies above @xmath16 and is non - decreasing with respect to @xmath18 and concave with respect to @xmath17 .
it is often denoted by @xmath128 .
we have the following representation formulas for the upper and lower envelopes of @xmath16 ( ( * ? ? ?
* section 4 ) ) : [ lem : representation ] assume @xmath129 .
then @xmath130 , y\in y_\rho^-\}\ ] ] and @xmath131 , y\in y_\rho^-\}.\ ] ] the following fact will play a central role in our arguments . in subsection [ subsec :
proof of pre abp ] of the appendix , we explain how proposition [ pre abp ] follows from parts of the proof of the parabolic version of the alexandroff - bakelman - pucci estimate as presented in ( * ? ? ?
* section 4.1.2 ) .
we refer the reader to ( * ? ? ?
* section 4 ) for a history of the parabolic version of the alexandroff - bakelman - pucci estimate and the relevant references .
[ pre abp ] assume @xmath132 is such that @xmath133 on @xmath134 .
assume that there exists a constant @xmath34 so that @xmath135_{c_t^{0,\eta}(y^-_\rho)}}}\leq k$ ] and @xmath90 in the sense of distributions for every @xmath136 .
let @xmath137 be the lower monotone envelope of @xmath138 in @xmath139 .
there exists a universal constant @xmath21 such that @xmath140 next we state several regularity results .
we will use the following interior hlder gradient estimate ( * ? ? ?
* theorems 4.8 and 4.9 ) .
[ thm : wang s c alpha thm ] assume ( f[ellipticity ] ) and ( f[last prop of f ] ) .
there exist universal constants @xmath141 and @xmath21 such that if @xmath142 is a solution of @xmath143 in @xmath144 , then @xmath145 and @xmath146 with @xmath147 the following proposition follows from theorem [ thm : wang s c alpha thm ] by a standard rescaling and covering argument . [
prop : diff of soln ] assume ( f[ellipticity ] ) and ( f[last prop of f ] ) .
assume that @xmath148 is a subset of @xmath149 with @xmath150 .
there exist universal constants @xmath141 and @xmath21 such that if @xmath151 is a solution of @xmath143 in @xmath152 , then @xmath153_{c^{0,\alpha}(v ) } } } + r{\ensuremath{||d u||_{l^{\infty}(v ) } } } & \leq c({\ensuremath{||u||_{l^{\infty}(\omega \times ( 0,t ) ) } } } + 1)\\ r^{2+\alpha}{\ensuremath{[u_t]_{c^{0,\alpha}(v ) } } } + r^2{\ensuremath{||u_t||_{l^{\infty}(v ) } } } & \leq c({\ensuremath{||u||_{l^{\infty}(\omega \times ( 0,t ) ) } } } + 1).\end{aligned}\ ] ]
in this section , we establish the parabolic version of the regularity result ( * ? ? ? * ; * ? ? ? * theorem a ) , which says that outside of sets of small measure , solutions of uniformly elliptic equations have second - order expansions with controlled error .
given @xmath154 , we define the set @xmath155 by @xmath156 in order to state our result , we introduce another family of subsets of @xmath45 .
we define @xmath157^n \times\left(t , t+ \frac{r^2}{81n}\right]\ ] ] we denote @xmath158 by @xmath159 . we prove the following growth estimate for @xmath160 : [ thm a ] assume @xmath3 satisfies ( f[ellipticity ] ) and ( f[last prop of f ] ) .
assume @xmath16 is a solution of @xmath143 in @xmath161 and @xmath162 and @xmath64 exist and are continuous in @xmath161 .
there exist universal constants @xmath163 and @xmath164 such that for every @xmath165 , latexmath:[\[\label{eq : size bad set } if @xmath16 is a solution of @xmath143 in @xmath152 and @xmath161 is compactly contained in @xmath152 , then @xmath16 satisfies the hypotheses of theorem [ thm a ] because , according to proposition [ prop : diff of soln ] , @xmath162 and @xmath64 exist and are continuous in @xmath161 .
suppose @xmath167 and @xmath168 is the paraboloid given by the definition of @xmath169 . if @xmath16 is a viscosity solution of @xmath143 in @xmath161 , then of course @xmath170 .
on the other hand , suppose @xmath16 is only a @xmath0-solution of @xmath143 in @xmath161 , and @xmath167 with @xmath171 .
in this case we _ can not _ conclude that @xmath170 .
this is because in the definition of @xmath169 we consider paraboloids @xmath168 in the class @xmath172 , while in the definition of @xmath0-solution we allow only paraboloids in the smaller class @xmath173 . in corollary [ cor ]
we establish a version of theorem [ thm a ] with @xmath174 instead of @xmath175
. an essential element of our proof is ( * ? ? ? * theorem 4.11 ) , which says that solutions of uniformly parabolic equations have _ first _
order expansions with controlled error on large sets . to state this result , we recall the following definitions : @xmath176 we observe that if @xmath177 , then there exists @xmath178 such that for all @xmath179 , @xmath180 we now present ( * ? ? ? * theorem 4.11 ) , with notation adapted for our setting : [ thm : wang s good set thm ] assume ( f[ellipticity ] ) and ( f[last prop of f ] ) .
assume that @xmath120 is a solution of @xmath181 in @xmath161 .
there exist universal constants @xmath21 , @xmath164 , and @xmath182 such that for any @xmath165 , @xmath183 the main idea of the proof of theorem [ thm a ] is to apply the growth estimate of theorem [ thm : wang s good set thm ] to the derivatives of @xmath16 . formally , upon differentiating @xmath143 , we obtain that the derivatives @xmath184 and @xmath64 solve the linear parabolic equation @xmath185 .
therefore , the estimates of theorem [ thm : wang s good set thm ] apply to @xmath184 and @xmath64 , and from this the estimates on @xmath16 are deduced . to carry out this plan
, we will first use the following proposition ( * ? ? ?
* theorem 4.6 ) .
[ diff eqn ] assume that @xmath16 is a solution of @xmath143 in @xmath161 .
then @xmath64 and @xmath184 , for @xmath186 , satisfy @xmath187 in @xmath161 .
next we proceed with the proof of theorem [ thm a ] .
we remark that a result similar to ( * ? ? ?
* ; * ? ? ?
* theorem a ) was established later , via similar methods , by armstrong , silvestre and smart ( * ? ? ?
* section 5 ) .
our proof is based on the arguments in ( * ? ? ?
* ; * ? ? ?
* theorem a ) and our presentation follows that of ( * ? ? ?
* section 5 ) .
we proceed with : by proposition [ diff eqn ] , the derivatives @xmath184 , @xmath188 and @xmath64 satisfy ( [ class s ] ) .
therefore , by theorem [ thm : wang s good set thm ] , for any @xmath189 the size of the bad sets " of the @xmath184 and of @xmath64 are controlled : we have @xmath190 where @xmath21 is a universal constant .
we define @xmath191 by @xmath192 we will show @xmath193 so that , by ( [ bd size sets ] ) , @xmath194 to this end , fix @xmath195 .
the goal is to show that @xmath196 ; in other words , to produce a polynomial @xmath168 that is the second order expansion of @xmath16 at @xmath46 . because @xmath197 , there exist @xmath198 such that for @xmath188 and for all @xmath179 , @xmath199 we point out that @xmath200 and @xmath201 .
we define the @xmath6 matrix @xmath202 by @xmath203 and define the paraboloid @xmath204 by @xmath205 we will now show that this polynomial @xmath168 is the second order expansion of @xmath16 at @xmath46 .
we have @xmath206 therefore , @xmath207 we take the absolute value both sides of the above and find @xmath208 we use ( [ eq : u_xi -p^i ] ) to bound each term on the right - hand side of the above , and obtain @xmath209 and combining terms yields @xmath210 [ cor ] assume @xmath3 satisfies ( f[ellipticity ] ) and ( f[last prop of f ] ) .
assume @xmath16 is a solution of @xmath143 in @xmath55 and @xmath162 and @xmath64 exist and are continuous in @xmath55 .
there exists a universal constant @xmath21 such that for any @xmath211 , there exists @xmath212 with @xmath213 and such that on @xmath214 , @xmath215 moreover , @xmath216 ; the general case follows by rescaling : given @xmath16 that solves @xmath217 in @xmath55 , the rescaled function @xmath218 solves the same equation in @xmath161 .
we omit the details of the rescaling argument and proceed with the proof in the case @xmath219 . fix @xmath220
let @xmath221 , where @xmath168 and @xmath222 are as in the proof of theorem [ thm a ] .
it holds that for all @xmath179 , @xmath223 from ( [ eq : u_xi -p^i ] ) we see that for any @xmath224 , @xmath225 taking @xmath226 such that @xmath227 and @xmath228 implies @xmath229 the bound ( [ eq : u - p cor ] ) follows .
we recall the definitions of the _ inf - convolution _ @xmath230 and the _ sup - convolution _ @xmath231 for @xmath232 : @xmath233 and @xmath234 these definitions are quite similar to those used in the regularity theory of elliptic equations , and to the regularizations of ( * ? ? ?
* section 4 ) .
we use the notation @xmath230 , instead of the expected @xmath235 , to distinguish these from the @xmath17-inf- and @xmath17-sup- convolutions that we introduce in the next subsection .
we summarize the basic properties of inf- and sup- convolutions in proposition [ properties of inf - convolution ] of the appendix . in this subsection
we define a regularization of inf - sup type that we call @xmath17-inf and @xmath17-sup convolutions . in the next subsection
, we prove that if @xmath16 is a solution of @xmath143 , then the @xmath17-inf and @xmath17-sup convolutions of @xmath16 enjoy regularity properties similar to those established in theorem [ thm a ] for @xmath16 . for @xmath236 and @xmath237
, we define the _
@xmath17-sup convolution _ @xmath238 and _ @xmath17-inf - convolution _ @xmath239 of @xmath16 by @xmath240 for @xmath241 , we define the set @xmath242 by @xmath243_{c_x^{0,1}(\omega \times ( 0,t ) ) } } } \right\}.\ ] ] in the following proposition we state the facts about @xmath17-inf- and @xmath17-sup- convolutions that we will use in this paper .
their proofs are very similar to those in the elliptic case and for the standard parabolic inf- and sup- convolution ( see ( * ? ? ?
* ; * ? ? ?
* chapter 4 ) or ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* proposition 5.3 ) ) and we omit them . [
prop : x - sup - conv ] assume @xmath244 . then : 1 . if @xmath245 is any point at which the infimum ( resp .
supremum ) is achieved in the definition of @xmath246 ( resp .
@xmath247 ) , then @xmath248 2 .
[ item : bound from above on sup conv ] for @xmath249 , @xmath250 and @xmath251 3 .
[ item : sup conv is semiconvex ] in the sense of distributions , @xmath252 and @xmath253 for all @xmath254 .
[ item : utheta is subsolution ] if @xmath16 is a subsolution of @xmath255 in @xmath149 , then @xmath238 is a subsolution of @xmath255 in @xmath256 .
if @xmath16 is a supersolution of @xmath255 in @xmath149 , then @xmath239 is a supersolution of @xmath255 in @xmath256 .
next we prove that the extra regularity established in theorem [ thm a ] for a solution @xmath16 of @xmath143 carries over to @xmath257 and @xmath258 . to state
this result we introduce the following families of sets : @xmath259\\ k^{\theta}_r ( \bar{x},\bar{t } ) & = \left[\bar{x}-\left(\frac{r}{9\sqrt{n}}-2\theta{\ensuremath{||d u||_{l^{\infty}(\omega\times(0,t))}}}\right ) , \bar{x}+\left(\frac{r}{9\sqrt{n}}-2\theta{\ensuremath{||d u||_{l^{\infty}(\omega\times(0,t))}}}\right)\right]^n \times\left(\bar{t } , \bar{t}+\frac{r^2}{81n}\right].\end{aligned}\ ] ] ( note that for these definitions to be meaningful , @xmath260 can not be too small compared to @xmath261 .
whenever we use this notation we make sure this is not the case . )
[ prop : more regularity for convolutions ] assume ( f[ellipticity ] ) and ( f[last prop of f ] ) .
suppose @xmath97 is a solution of @xmath143 in @xmath262 .
assume @xmath263 and @xmath264 there exist universal constants @xmath265 , and @xmath21 such that for every @xmath189 , there exists a set @xmath266 ( resp .
@xmath267 ) such that for any @xmath268 ( resp .
@xmath269 ) , there exists a polynomial @xmath270 such that @xmath271 , @xmath272 and if @xmath273 , then for all @xmath274 , we have @xmath275 @xmath276 moreover , @xmath277 is quite involved , it is based on the following simple observation : if the supremum in the definition of @xmath278 is achieved at @xmath279 and @xmath16 is touched from below at @xmath279 by some function @xmath82 , then @xmath257 is touched by below at @xmath46 by a translate of @xmath82 .
we demonstrate this below .
let @xmath280 be a point at which the supremum is achieved in the definition of @xmath278 , so that @xmath281 suppose that @xmath282 is a function with @xmath283 and @xmath284 on @xmath285 .
this implies @xmath286 fix @xmath287 .
if @xmath288 is sufficiently far from the boundary of @xmath14 , then we may use @xmath289 as a `` test point '' in the definition of @xmath290 and find @xmath291 but , since @xmath284 on @xmath285 , we have @xmath292 subtracting ( [ toy touching ] ) from the above implies @xmath293 so @xmath294 is touched from below at @xmath46 by @xmath295 . for the proof of proposition [ prop : more regularity for convolutions ] , we consider the map that takes @xmath296 to @xmath17 and investigate what happens to `` good set '' @xmath297 under this map . to do this
, we need the following lemma , which says that @xmath16 is twice differentiable in @xmath17 at any point @xmath296 where the supremum or infimum is achieved in the regularizations @xmath257 or @xmath258 , and gives a bound on @xmath68 at such points .
[ lem : dxu lip ] under the assumptions of proposition [ prop : more regularity for convolutions ] , let @xmath298 . if @xmath296 is any point where the supremum ( resp .
infimum ) is achieved in the definition of @xmath247 ( resp .
@xmath246 ) , then @xmath299 is twice differentiable in @xmath226 at @xmath296 with @xmath300 we postpone the proof of lemma [ lem : dxu lip ] and proceed with the proof of proposition [ prop : more regularity for convolutions ] . we will give the proof for @xmath257 ; the arguments for @xmath258 are similar .
fix @xmath73 and , to simplify the notation , denote @xmath301 by just @xmath302 .
we define the set @xmath303 by @xmath304 we remark that @xmath305 : if @xmath306 , then there exists a corresponding @xmath298 , and , by proposition [ prop : x - sup - conv ] , we have @xmath307 .
the definition of @xmath308 thus implies that @xmath309 . by lemma [ lem :
dxu lip ] , @xmath162 is lipschitz on @xmath303 with @xmath310 for any @xmath306 , the map @xmath311 has a local maximum at @xmath296 .
therefore , @xmath312 so @xmath313 . we define the map @xmath314 by @xmath315 since @xmath162 is lipschitz on @xmath316 , @xmath314 is too , and ( [ du on c ] ) implies that for all @xmath317 , @xmath318 we define the set @xmath319 by @xmath320 by the area formula ( ( * ? ? ?
* ; * ? ? ?
* chapter 3 ) ) , we have @xmath321 the above bound on @xmath322 and the bound ( [ better bdn on theta_m ] ) on @xmath323 therefore imply @xmath324 by @xmath325 fix @xmath326
. we will now show that there exists a polynomial @xmath212 satisfying ( [ cond on p ] ) and ( [ eq : touch utheta from below ] ) . to this end , let @xmath245 be any point at which the supremum in the definition of @xmath278 is achieved , so that @xmath327 .
first we note that since , by proposition [ prop : x - sup - conv ] , @xmath307 and @xmath328 , we have @xmath329 . next we observe that @xmath330 . indeed , if @xmath331 , then @xmath332 , so @xmath333 . however , this is impossible , as we chose @xmath334
. therefore , @xmath330 .
since @xmath330 , there exists a polynomial @xmath212 with @xmath335 such that on @xmath336 , @xmath337 it is easy to check that since @xmath296 is a point where the supremum is achieved in the definition of @xmath278 , we have , for all @xmath338 , @xmath339 where @xmath340 is a paraboloid of the same opening as @xmath202 .
finally , if @xmath341 , then for any @xmath274 , @xmath342 since @xmath343 , proposition [ thm : wang s c alpha thm ] implies @xmath344 and @xmath345 we use these bounds in ( [ size a ] ) and ( [ touch below no bd ] ) to complete the proof . for the proof of lemma [ lem : dxu lip ]
, we first need to establish the following basic fact about how parabolic viscosity solutions relate to viscosity solutions in only @xmath17 : [ lem : visc soln in t and x ] if @xmath16 is a lipschitz in @xmath18 viscosity solution of @xmath143 in @xmath55 , then for any @xmath346 , @xmath347 we will check that @xmath348 is a subsolution ; checking that @xmath348 is a supersolution is analogous .
assume that @xmath349 touches @xmath350 from above at @xmath351 .
let @xmath352 $ ] .
since @xmath16 is lipschitz in @xmath18 , we have @xmath353 therefore , @xmath354 touches @xmath16 from above in @xmath355 $ ] . because @xmath16 is a viscosity solution of @xmath143 , we obtain @xmath356 the uniform ellipticity of @xmath3 implies @xmath357 as desired .
we also recall the harnack inequality for _ elliptic _ equations ( see ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* theorem 4.3 ) ) .
assume that @xmath358 satisfies @xmath359 in @xmath360 and , for some positive constant @xmath361 , @xmath362 there exits a universal constant @xmath21 so that @xmath363 we are now ready to proceed with : we will give the proof for @xmath247 ; the argument for @xmath246 is very similar .
let @xmath296 be a point at which the supremum in the definition of @xmath364 is achieved , so that @xmath365 therefore , as a function of @xmath226 , @xmath299 is touched from above at @xmath296 by the convex paraboloid @xmath366 , where @xmath367 thus , we have @xmath368 next we will show , using the harnack inequality , that @xmath348 is also touched from below at @xmath296 by a paraboloid with bounded opening , thus obtaining a bound from below on @xmath369 .
there exists an affine function @xmath370 with @xmath371 and such that @xmath372 there exists @xmath102 such that @xmath373 .
we fix any @xmath374 and define the function @xmath375 by @xmath376 for @xmath377 , @xmath378 and , by lemma [ lem : visc soln in t and x ] , @xmath379 is a viscosity solution of @xmath380 therefore , the harnack inequality implies that there exists a universal constant @xmath21 such that @xmath381 since @xmath382 , we obtain @xmath383 thus , for all @xmath384 , @xmath385 since this holds for all @xmath374 , we have that @xmath386 is touched from below at @xmath296 by a paraboloid of opening @xmath387 .
therefore , @xmath388 together with ( [ d^2 bound below ] ) , this implies @xmath389 finally , we use proposition [ prop : diff of soln ] to bound @xmath390 and obtain the desired bound .
in this section we prove a proposition that is a key part of the proofs of the two main results , theorems [ thm : vague comparison ] and [ thm : mesh comparison vague ] .
once we have proven proposition [ main lemma ] , we will only need to use the basic properties of inf- and sup- convolutions to verify its hypotheses and establish the two main results .
we define the constant @xmath391 by @xmath392 where @xmath164 is the constant from proposition [ prop : more regularity for convolutions ] .
we remark that since @xmath164 is universal , so is @xmath391 .
[ main lemma ] assume ( f1 ) , ( f2 ) and ( u1 ) and that @xmath241 is a solution of @xmath393 in @xmath149 . for @xmath394 , define the quantity @xmath395 by @xmath396 and the set @xmath397 by @xmath398 1 .
assume @xmath399 is a @xmath0-supersolution of ( [ eqn : mainlemma ] ) in @xmath149 that satisfies 1 .
[ assump : d2v ] @xmath400 in the sense of distributions , and 2 .
[ assump : vt ] @xmath401_{c_t^{0,\eta}(\omega\times(0,t))}}}\leq 3t\delta^{-\zeta}$ ] .
+ there exists a universal constant @xmath402 , a positive constant @xmath403 that depends on @xmath71 , @xmath404 , @xmath69 , @xmath405 , @xmath314 , @xmath406 and @xmath407 , and a positive constant @xmath408 that depends only on @xmath71 , such that , for all @xmath409 , @xmath410 2 .
assume @xmath411 is a @xmath0-subsolution of ( [ eqn : mainlemma ] ) in @xmath149 that is twice differentiable almost everywhere in @xmath33 , with @xmath412 in the sense of distributions and @xmath413_{c_t^{0,\eta}(\omega\times(0,t))}}}\leq 3t\delta^{-\zeta}$ ] .
then for all @xmath409 , @xmath414 where @xmath402 , @xmath403 and @xmath408 are the same as in part 1 of this proposition .
[ rem : where main prop holds ] proposition [ main lemma ] holds in any set @xmath415 , where @xmath416 satisfies ( u1 ) . in this case , @xmath397 becomes @xmath417 we seek to control @xmath418 by the size of its contact set with its upper monotone envelope .
we first regularize @xmath16 by @xmath17-sup convolution and obtain @xmath257 .
we then perturb @xmath257 to make it a strict sub - solution .
let us denote the resulting function by @xmath419 .
we choose this perturbation and the parameter @xmath261 of the @xmath17-sup convolution so that the difference between @xmath16 and @xmath419 is less than @xmath420
. thus it will be enough to bound @xmath421 .
since @xmath419 and @xmath120 are sufficiently regular , we are able to apply proposition [ pre abp ] and find @xmath422 where @xmath423 is the upper monotone envelope of @xmath424 and @xmath34 is large . to complete the proof
, we will show that the contact set is very small compared to @xmath34 .
the key part of our argument is proposition [ prop : more regularity for convolutions ] , which states that there exists a subset @xmath425 of @xmath152 on which @xmath257 is very close to being a polynomial .
this allows us to obtain an upper bound on the size of @xmath426 . moreover , theorem [ thm a ] says that @xmath427 is large , so the size of the remainder of the contact set , @xmath428 , is small .
this completes the outline . for the proof of proposition [ main lemma ] we need a lemma .
[ lem : helper2 ] under the assumptions of proposition [ main lemma ] , assume also that @xmath429 with @xmath430 there exists a positive constant @xmath431 that depends only on @xmath71 and a positive constant @xmath432 that depends on @xmath71 , @xmath404 , @xmath69 , and @xmath407 , such that , if @xmath433 , then there exists @xmath434 and a paraboloid @xmath174 with 1 .
@xmath435 , 2 .
@xmath436 , 3 .
@xmath437 , and 4 . for all @xmath438 , @xmath439 we
postpone the proof of lemma [ lem : helper2 ] and proceed with the proof of proposition [ main lemma ] .
we will give the proof for part 1 of the proposition ; the proof of part 2 is very similar .
in addition , we assume , without loss of generality , that @xmath440 for @xmath391 as given in ( [ constants ] ) , we regularize @xmath16 by taking @xmath17-sup - convolution and denote @xmath441 by @xmath442 .
we also introduce a perturbation @xmath419 of @xmath442 , which we define by @xmath443 by proposition [ prop : x - sup - conv ] , @xmath419 is a subsolution of @xmath444 in @xmath397 . by the definition of @xmath419 , proposition [ prop :
x - sup - conv ] and ( [ main prop wlog ] ) we have @xmath445 where @xmath446 depends on @xmath407 .
. there exist @xmath102 and @xmath448 such that @xmath449 . since @xmath450 on @xmath451 , we may extend @xmath452 by @xmath453 to @xmath454 .
let @xmath455 be the lower monotone envelope of @xmath456 .
we point out that @xmath457 , where @xmath458 is the upper monotone envelope of @xmath459 in @xmath460 .
moreover , we have @xmath461 and @xmath462 we seek to apply proposition [ pre abp ] with @xmath379 instead of @xmath16 and @xmath454 instead of @xmath463 . since @xmath464 , proposition [ prop : diff of soln ] implies that @xmath16 is differentiable in @xmath18 in @xmath397 , with @xmath465 where @xmath21 depends on @xmath71 , @xmath404 , @xmath69 , @xmath406 and @xmath407
. therefore , @xmath466 as well . together with assumption ( [ assump : vt ] ) , this implies @xmath413_{c_t^{0,\eta}(\omega\times(0,t))}}}\leq c_0\delta^{-2\zeta}$ ] , where @xmath467 depends on @xmath71 , @xmath404 , @xmath69 , @xmath314 , @xmath406 and @xmath407 .
in addition , assumption ( [ assump : d2v ] ) on @xmath120 and the properties of @xmath17-sup convolutions ( see item ( [ item : sup conv is semiconvex ] ) of proposition [ prop : x - sup - conv ] ) ) imply @xmath468 in the sense of distributions .
therefore , the hypotheses of proposition [ pre abp ] hold with @xmath469 , so we obtain @xmath470 where @xmath423 is the upper monotone envelope of @xmath471 and @xmath21 depends on @xmath71 , @xmath404 , @xmath69 , @xmath314 , @xmath472 , @xmath406 and @xmath407
. therefore , latexmath:[\[\label{bd cont set below } where @xmath432 depends on @xmath71 , @xmath404 , @xmath69 , @xmath314 , @xmath472 , @xmath406 and @xmath407 .
we claim @xmath474 where @xmath475 is given by ( [ choice o c_1 ] ) .
if ( [ main lemma bound sup ] ) holds , then we have @xmath476 where @xmath477 and @xmath478 .
therefore , to complete the proof of the lemma it will suffice to show that ( [ main lemma bound sup ] ) holds . to this end , we proceed by contradiction and assume that ( [ main lemma bound sup ] ) does not hold , in which case we have a lower bound for the right - hand side of ( [ bd cont set below ] ) and thus obtain @xmath479 we now apply lemma [ lem : helper2 ] with @xmath480 and obtain that there exists a point @xmath481 with @xmath482 and a paraboloid @xmath168 with @xmath436 , @xmath483 and @xmath484 where the error @xmath485 is given by @xmath486 and @xmath487 depends on @xmath71 , @xmath404 , @xmath69 , and @xmath407 .
we recall the representation formula for the upper monotone envelope of a function ( lemma [ lem : representation ] ) : @xmath488 , x\in \omega\}.\ ] ] since @xmath489 is in the contact set of @xmath471 with its upper monotone envelope , the representation formula implies that there exist @xmath490 and @xmath491 such that @xmath492 and @xmath493 for all @xmath17 and for all @xmath494 ( so in particular , for all points in @xmath495 ) .
therefore , for all @xmath496 , we have @xmath497 with equality holding at @xmath498 . by the definitions of @xmath419 and @xmath442 , we find @xmath499 holds on @xmath495 with equality holding at @xmath498 .
we use the bound ( [ touch u at x0t0 used ] ) to find that for all @xmath500 , @xmath501 with equality at @xmath498 . by ( [ y^-in u ] ) ,
since @xmath120 is a @xmath0-supersolution of ( [ eqn : mainlemma ] ) on @xmath149 , we find @xmath503 since the nonlinearity @xmath3 is uniformly elliptic , we have @xmath504 but , according to ( [ what p solves ] ) , @xmath505 , so we obtain @xmath506 using the definition of @xmath485 , we find @xmath507 where @xmath508 is a constant that depends on @xmath71 , @xmath404 , @xmath69 , and @xmath407 . by our choice of @xmath391 , we obtain @xmath509 we choose @xmath510 and find @xmath511 which is impossible , so ( [ main lemma bound sup ] ) must hold .
lemma [ lem : helper2 ] follows from proposition [ prop : more regularity for convolutions ] and a covering argument .
the covering argument is slightly more involved than in the elliptic case because of the need to avoid times @xmath18 that are close to the terminal time @xmath314 .
we will denote the @xmath17-sup convolution @xmath512 by simply @xmath442 .
we will need the sets : @xmath513\\ { \ensuremath{\tilde{k}_{{\ensuremath{r_{\delta}}}}}}(\bar{x},\bar{t } ) & : = k^{\delta^{\zeta}}_{{\ensuremath{r_{\delta } } } } ( \bar{x},\bar{t } ) = \left [ \bar{x } - \delta^{\zeta } , \bar{x } + \delta^{\zeta}\right]^n \times \left(\bar{t } , \bar{t}+\frac{r_\delta^2}{81n}\right ] \\
{ \ensuremath{\tilde{k}_{{\ensuremath{r_{\delta}}}}^t}}(\bar{x},\bar{t } ) & : = \left [ \bar{x } - \delta^{\zeta } , \bar{x } + \delta^{\zeta}\right]^n \times \left(\bar{t}+\frac{r_\delta^2}{162n } , \bar{t}+\frac{r_\delta^2}{81n}\right],\end{aligned}\ ] ] where @xmath514 is the top half ( in terms of @xmath18 ) of @xmath515 .
see figure 1 .
[ sets for lemma ] we remark that there exists a constant @xmath431 that depends only on @xmath71 such that if @xmath433 , then @xmath516 ( we point out that in order for ( [ remark about sets ] ) to hold , it is important that @xmath517 , and not only in @xmath518 . )
we take @xmath519 .
we recall that proposition [ prop : more regularity for convolutions ] gives a lower bound on the size of @xmath520 , the set where @xmath442 has second - order expansions from below . for the rest of the argument we will denote @xmath520 simply by @xmath521 .
we will show that there exists a point in the intersection of @xmath521 and @xmath303 by using the lower bounds on the size of @xmath303 and @xmath522 ( the latter is proved in proposition [ prop : more regularity for convolutions ] ) . since proposition [ prop : more regularity for convolutions ] bounds the size of the complement of @xmath522 inside of sets of the form @xmath515 , we first cover @xmath397 by @xmath523 .
it is clear that the finite collection @xmath524 covers @xmath397 , where @xmath525 , for @xmath467 a constant that depends on @xmath526 , @xmath314 , and @xmath71 .
therefore , there exists an @xmath527 such that @xmath528 the lower bound on @xmath529 and the upper bound on @xmath530 gives @xmath531 in view of the definition of @xmath397 , we have @xmath532 and @xmath533 ; therefore , @xmath534 and @xmath535 therefore , @xmath16 satisfies the hypotheses of proposition [ prop : more regularity for convolutions ] in @xmath536 with @xmath537 and @xmath538 .
thus there exists a set @xmath539 ( the good set " of @xmath442 ) with @xmath540 where @xmath432 is a universal constant .
we take @xmath541 so that @xmath542 since @xmath543 , we have @xmath544 therefore , there exists a point @xmath545 . because of ( [ remark about sets ] ) , we have @xmath546 .
thus , by proposition [ prop : more regularity for convolutions ] , there exists a polynomial @xmath174 such that @xmath547 and , on @xmath548 , @xmath549 there exists a constant @xmath487 that depends on @xmath71 and @xmath407 such that @xmath550 therefore @xmath551 where @xmath508 depends on @xmath71 , @xmath404 , @xmath69 , and @xmath407 .
we use this to bound the coefficients of the error terms in ( [ touch u+ below ] ) and find @xmath552
in this section we give the precise statement and the proof of our main result : [ main result ] assume ( f[ellipticity ] ) , ( f[last prop of f ] ) and ( u1 ) .
assume @xmath553 is a solution of @xmath554 and assume @xmath555 is a family of @xmath0-supersolutions ( resp .
@xmath0-subsolutions ) of ( [ maineqn ] ) such that , for some @xmath556 and for all @xmath110 , @xmath557 and @xmath558 there exist a constant @xmath559 that depends only on @xmath71 , a constant @xmath560 that depends on @xmath71 , @xmath404 , @xmath69 and @xmath561 , and a constant @xmath475 that depends on @xmath71 , @xmath404 , @xmath69 , @xmath405 , @xmath314 , @xmath73 , @xmath562 , and @xmath407 , such that for all @xmath563 , @xmath564 [ remaboutlips ] the assumption @xmath553 is satisfied in the following situation .
assume ( f[ellipticity ] ) , ( f[last prop of f ] ) , ( u1 ) , and that @xmath565 is regular .
take @xmath566 .
then , according to the interior regularity estimates of theorem [ thm : wang s c alpha thm ] and the boundary estimates of ( * ? ? ?
* section 2 ) , the solution @xmath16 of the boundary value problem @xmath567 is indeed lipschitz continuous in @xmath17 and hlder continuous in @xmath18 on all of @xmath152 , with @xmath568 where @xmath21 depends on @xmath71 , @xmath404 , @xmath69 , @xmath569 , @xmath526 and the regularity of @xmath570 .
theorem [ main result ] follows easily from proposition [ main lemma ] .
we give the proof of the bound on @xmath571 in the case that @xmath572 is a @xmath0-supersolution ; the proof of the other case is very similar .
first we regularize @xmath572 by taking inf - convolution : we let @xmath573 and define the constant @xmath574 and the set @xmath575 we point out that @xmath576 is of the form @xmath415 , where @xmath416 satisfies ( u1 ) .
therefore , proposition [ main lemma ] holds in @xmath576 instead of @xmath149 ( see remark [ rem : where main prop holds ] ) .
according to item ( [ inf conv is delta soln ] ) of proposition [ properties of inf - convolution ] , @xmath120 is a @xmath0-supersolution of ( [ eqn : mainlemma ] ) in @xmath576 .
moreover , by items ( [ item : inf conv is semiconcave ] ) and ( [ item : lip bd on v ] ) of proposition [ properties of inf - convolution ] , @xmath120 satisfies the regularity assumptions of proposition [ main lemma ] in @xmath576 .
therefore , according to proposition [ main lemma ] , @xmath577 where @xmath578 and @xmath395 is as given in the statement of proposition [ main lemma ] . since @xmath579 , @xmath580 , and @xmath581 on @xmath582 , there exists a constant constant @xmath583 that depends on @xmath71 , @xmath73 , and @xmath584 such that @xmath585 and @xmath586 by item ( [ item : bound from below on inf conv ] ) of proposition [ properties of inf - convolution ] , @xmath587_{c^{0,\eta}(\omega \times ( 0,t))}}}(2\delta^{\zeta } { \ensuremath{[v_\delta]_{c^{0,\eta}(\omega \times ( 0,t))}}})^{\frac{\eta}{2-\eta } } \leq \sup_{{\ensuremath{\partial_p \tilde}}{u } } ( u - v_\delta ) + c_2\delta^{\frac{\zeta\eta}{2-\eta}},\ ] ] where @xmath588 is a constant that depends on @xmath71 and @xmath73 . by the above line and
( [ bound on bdry u ] ) we have @xmath589 where @xmath590 depends on @xmath71 , @xmath73 and @xmath584 .
so , by ( [ bound on whole set ] ) , ( [ consequence of propo ] ) , and the above line , we have @xmath591 we thus take @xmath592 and @xmath593 to complete the proof of the theorem .
we now present the error estimate for finite difference approximation schemes .
first , we introduce the necessary notation and assumptions , closely following @xcite and @xcite . in the next section
we give the full statement of the error estimate and its proof .
the space - time mesh is denoted by @xmath485 : @xmath594 we fix some @xmath595 and define the subset @xmath596 of @xmath485 by @xmath597 next we introduce finite difference operators for a function @xmath16 : @xmath598 an _ implicit finite difference operator _ is an operator of the form @xmath599 ( x , t ) = \delta^-_{\tau } u ( x , t ) - \mathcal{f}_h(\delta^2 u(x , t)),\ ] ] where @xmath600 is locally lipschitz .
we denote points in @xmath601 by @xmath602 .
we say an operator is _ monotone _
if it satisfies : 1 . there exists a constant @xmath603 and @xmath604 such that for all @xmath605 @xmath606 this is equivalent to the definition given in the introduction . a scheme @xmath38 is said to be _ consistent with @xmath3 _ if for all @xmath607 , @xmath608| \rightarrow 0 \text { as } h\rightarrow 0.\ ] ] in ( * ?
* ; * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* section 4 ) , it was shown that if a nonlinearity @xmath3 satisfies ( f[ellipticity ] ) , then there exists a monotone implicit scheme @xmath38 that is consistent with @xmath143 , and the constants @xmath603 and @xmath604 depend only on @xmath71 , @xmath404 and @xmath69 . in @xcite ,
a monotone and consistent approximation scheme for elliptic equations is explicitly constructed , and the construction in the parabolic case is analogous . in order to obtain an error estimate , we need to make an assumption that quantifies the above rate of convergence . as in @xcite , we assume : 1 . there exists a positive constant @xmath34 such that for all @xmath35 , @xmath37|\leq k(h+ h{\ensuremath{||d_x^3\phi||_{l^{\infty}(\omega\times ( 0,t ) ) } } } + h^2{\ensuremath{||\phi_{tt}||_{l^{\infty}(\omega\times ( 0,t))}}}).\ ] ] schemes that satisfy ( s[consistent ] ) are said to be _ consistent with an error estimate for @xmath3 in @xmath152 with constant @xmath34_. given @xmath609 , we denote @xmath610 by @xmath611 .
we divide @xmath611 into interior and boundary points relative to the operator @xmath38 .
mainly , we define @xmath612 to be the interior points and @xmath613 to be the boundary points .
notice that @xmath27(x , t)$ ] depends only on @xmath614 for @xmath615 and for @xmath616 .
the discrete hlder seminorm of a function @xmath16 on @xmath611 is defined to be @xmath617_{c^{0,\eta}({\ensuremath{\mathcal{u}_h } } ) } } } = \sup_{p , q \in { \ensuremath{\mathcal{u}_h}}}\frac{|u(p)-u(q)|}{d(p , q)^{\eta}}\ ] ] and the discrete hlder norm is @xmath618_{c^{0,\eta}({\ensuremath{\mathcal{u}_h } } ) } } } .\ ] ] in ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* section 4 ) it is shown that solutions of the discrete equation @xmath619 = 0 $ ] are uniformly equicontinuous . we summarize this result : [ ktthm ] assume that @xmath38 is an implicit monotone finite difference scheme and that @xmath43 is a solution of @xmath619 = 0 $ ] .
there exists a constant @xmath21 that depends only on @xmath71 , @xmath603 and @xmath604 such that for all @xmath620 , @xmath621 for a mesh function @xmath622 , we define the inf- and sup- convolutions of @xmath120 at @xmath254 by @xmath623 and @xmath624 for @xmath625 , we introduce the quantity @xmath626 and the set @xmath627 in the appendix we summarize the basic properties of inf- and sup- convolutions of mesh functions ( see proposition [ prop : inf - conv for mesh ] ) . it is a classical fact of the theory of viscosity solutions that if @xmath16 is the viscosity solution of @xmath143 in @xmath152 , then the sup - convolution of @xmath16 is a subsolution of the same equation . in the following proposition
, we establish a similar relationship between solutions @xmath43 of discrete equations and @xmath0-solutions of @xmath143 .
this is a key step in establishing an error estimate for discrete approximation schemes .
[ item : mesh solves equation ] assume that @xmath38 is a monotone and implicit scheme consistent with an error estimate for @xmath3 with constant @xmath34 .
assume @xmath625 . 1 .
if @xmath120 satisfies @xmath628 \geq 0 $ ] in @xmath629 then @xmath630 is a @xmath0-supersolution of @xmath631 in @xmath632 , with @xmath633 .
2 . if @xmath120 satisfies @xmath628 \leq 0 $ ] in @xmath629 , then @xmath634 is @xmath0-subsolution of @xmath635 in @xmath632 , with @xmath633 .
we give the proof of item ( 1 ) ; the proof of item ( 2 ) is analogous .
let @xmath636 and let @xmath637 be a point at which the infimum is achieved in the definition of @xmath638 .
assume that @xmath174 is such that @xmath639 on @xmath640 , with equality at @xmath46 . because @xmath641 on _ all _ of @xmath640 , for each @xmath226 with @xmath642 we have @xmath643 the definition of @xmath644 and item ( 1 ) of proposition [ prop : inf - conv for mesh ] yield @xmath645 ; therefore , @xmath646(x^ * , t^ * ) = \delta^-_\tau v(x^ * , t^ * ) - \mathcal{f}_h(\delta^2 v(x^ * , t^ * ) ) .\ ] ] using ( [ eq : implicit ] ) and the monotonicity of @xmath647 to bound the right - hand side of the above yields @xmath648 that @xmath38 is consistent with an error estimate implies that @xmath649 together with ( [ eq : scheme at x * ] ) , this implies @xmath650
we prove [ prop : convergence for approx schemes ] assume ( f[ellipticity ] ) , ( f[last prop of f ] ) and ( u1 ) .
assume @xmath553 is a solution of @xmath651 in @xmath149 .
assume @xmath38 is an implicit monotone scheme consistent with an error estimate for ( [ h thm main eqn ] ) and that @xmath43 satisfies @xmath619 \geq 0 $ ] ( resp .
@xmath619 \leq 0 $ ] ) in @xmath611 .
assume that for some constant @xmath556 and for all @xmath36 , @xmath652_{c^{0,\eta}({\ensuremath{\mathcal{u}_h}}^i)}}}\leq m\ ] ] and @xmath653 there exist a constant @xmath654 that depends only on @xmath71 , a constant @xmath560 that depends on @xmath71 , @xmath404 , @xmath69 and @xmath561 , and a constant @xmath475 that depends on @xmath71 , @xmath655 , @xmath404 , @xmath69 , @xmath405 , @xmath314 , @xmath73 , @xmath584 , and @xmath407 , such that for all @xmath656 , @xmath657 suppose that @xmath16 is the solution of the boundary value problem @xmath567 and @xmath43 satisfies @xmath658=0 & \quad \text { in } { \ensuremath{\mathcal{u}_h}}^i,\\ v_h = g & \quad \text { on } { \ensuremath{\mathcal{u}_h}}^b , \end{array}\right.\ ] ] where @xmath565 is sufficiently regular and @xmath659 .
then , as explained in remark [ remaboutlips ] , @xmath16 satisfies the hypotheses of theorem [ prop : convergence for approx schemes ] , and @xmath43 satisfies the hypotheses of theorem [ prop : convergence for approx schemes ] because of theorem [ ktthm ] .
we do not know whether our result holds for explicit or mixed schemes .
our method can not be used in those cases , since we are able to verify proposition [ item : mesh solves equation ] only for implicit schemes .
theorem [ prop : convergence for approx schemes ] follows easily from proposition [ main lemma ] .
we give the proof of the bound on @xmath660 ; the proof of the other case is very similar .
first we regularize @xmath43 by taking inf - convolution .
we denote @xmath661 and define @xmath662 and @xmath663 then , according to proposition ( [ item : mesh solves equation ] ) , @xmath664 is a @xmath0-supersolution of @xmath665 in @xmath576 , with @xmath666 .
we point out that @xmath576 is of the form @xmath415 , where @xmath416 satisfies ( u1 ) .
therefore , proposition [ main lemma ] holds in @xmath576 instead of @xmath149 ( see remark [ rem : where main prop holds ] ) .
since we proved proposition [ main lemma ] for @xmath0-solutions with right - hand side @xmath453 , we have to perturb @xmath664 .
we introduce @xmath667 so that @xmath120 is a @xmath0-supersolution of ( [ h thm main eqn ] ) in @xmath576 with @xmath666 .
moreover , by items ( [ item : mesh inf conv is semiconcave ] ) and ( [ item : mesh lip bd on v ] ) of proposition [ prop : inf - conv for mesh ] , @xmath120 satisfies the regularity assumptions of proposition [ main lemma ] .
therefore , according to proposition [ main lemma ] , @xmath668 where @xmath669 and @xmath395 is given in the statement of proposition [ main lemma ] . since @xmath579 , @xmath580 , and @xmath581 on @xmath670 , there exists a constant @xmath446 that depends on @xmath71 , @xmath655 , @xmath314 , @xmath671 and @xmath584 , and a constant @xmath672 that depends on @xmath71 , @xmath404 , @xmath69 and @xmath561 , such that @xmath673 and @xmath674 by item ( [ item : mesh : bound from below on inf conv ] ) of proposition [ prop : inf - conv for mesh ] , @xmath675 so by the above line and ( [ h bound on bdry u ] ) we have @xmath676 where @xmath590 depends on @xmath71 , @xmath655 , @xmath314 , @xmath671 and @xmath584 .
so , by ( [ h bound on whole set ] ) , ( [ h consequence of propo ] ) , and the above , we have @xmath677 we thus take @xmath592 and @xmath678 to complete the proof of the theorem .
we outline the proof of proposition [ pre abp ] , which is a slight modification of part of the proof of the parabolic version of the abp estimate in ( * ? ? ?
* section 4.1.2 ) .
we assume @xmath679 ; the general case follows by rescaling . as in (
* section 4.1.2 ) , we define the function @xmath680 by @xmath681 proposition [ pre abp ] follows from the following three lemmas .
[ lempreabpsup ] suppose @xmath682 and let @xmath683 . then @xmath684 [ lempreabpdg ] if @xmath137 is @xmath685 with respect to @xmath17 and lipschitz with respect to @xmath18 , then @xmath686 is lipschitz in @xmath46 and we have @xmath687 for almost every @xmath688 .
[ lempreabpgamma ] suppose @xmath16 is as in the statement of proposition [ pre abp ]
. then @xmath137 is @xmath685 with respect to @xmath17 and lipschitz in @xmath18 .
lemma [ lempreabpsup ] is very similar to ( * ? ? ?
* lemma 4.13 ) the latter is stated for @xmath16 a supersolution of @xmath689 .
however , it is easy to verify that the proof of ( * ? ? ?
* lemma 4.13 ) holds for general functions @xmath690 , not only supersolutions .
lemma [ lempreabpdg ] is exactly ( * ? ? ?
* lemma 4.4 ) .
the proof of lemma [ lempreabpgamma ] is very similar to that of lemma ( * ? ? ?
* lemma 4.11 ) , which states that if @xmath16 is a supersolution of @xmath689 , then @xmath137 is @xmath685 in @xmath17 and lipschitz in @xmath18 .
we explain the modifications needed to obtain lemma [ lempreabpgamma ] : as in the proof of lemma ( * ? ? ?
* lemma 4.11 ) , it will suffice to show that there exists a constant @xmath34 such that for all @xmath688 , the following holds : @xmath691 in the proof of ( * ? ? ?
* lemma 4.11 ) , the assumption that @xmath16 is a supersolution of @xmath689 is used only to show that ( [ jet condition ] ) holds for @xmath692 .
thus , we need to verify ( [ jet condition ] ) for such points . to this end , assume @xmath692 and @xmath693 satisfies @xmath694 and @xmath695 for all @xmath696 in a neighborhood of @xmath46 . since @xmath137 is the lower monotone envelope of @xmath138 and @xmath692 , we have @xmath697 and @xmath698 for all @xmath696 in a neighborhood of @xmath46 .
our assumption that @xmath699 in the sense of distributions means that there exists a paraboloid @xmath700 of opening @xmath34 that touches @xmath701 from above at @xmath17 .
therefore , @xmath202 touches @xmath702 from above at @xmath17 .
therefore , @xmath703 . that @xmath135_{c_t^{0,\eta}(y^-_\rho)}}}\leq k$ ] implies @xmath704 .
lemma [ lempreabpsup ] , lemma [ lempreabpdg ] and the area formula ( * ? ? ?
* ; * ? ? ?
* chapter 3 ) , which applies to @xmath686 because of lemma [ lempreabpgamma ] , imply @xmath705 because @xmath137 is non - increasing in @xmath18 and convex in @xmath17 , the alexandroff theorem for functions depending on @xmath17 and @xmath18 ( see krylov ( * ? ? ?
* ; * ? ? ?
* appendix 2 ) ) implies that there exists @xmath706 with @xmath707 such that for every @xmath708 , @xmath709 ( in other words , @xmath137 is twice differentiable almost everywhere on @xmath710 ) .
we claim @xmath711 let us fix @xmath712 .
we have @xmath713 and @xmath714 for all @xmath715 .
our assumption that @xmath699 in the sense of distributions means that there exists a paraboloid @xmath700 of opening @xmath34 that touches @xmath701 from above at @xmath17
. therefore , @xmath202 touches @xmath716 from above at @xmath17 . therefore , @xmath717 and so @xmath718 that @xmath135_{c_t^{0,\eta}(y^-_\rho)}}}\leq k$ ] implies @xmath719 .
thus , ( [ claim for 2nd pt ] ) holds .
the bound ( [ cons of abpbd ] ) follows from ( [ claim for 2nd pt ] ) and ( [ abp - like bd ] ) . in the following proposition we state the facts about inf- and sup- convolutions that are used in this paper .
their proofs are very similar to those in the elliptic case ( see ( * ? ? ?
* ; * ? ? ?
* propositions 5.3 and 5.5 ) ) and we omit them . given @xmath720 and @xmath237 , we define the set @xmath721 [ properties of inf - convolution ] assume @xmath722 .
then : 1 . if @xmath637 is a point at which the supremum or infimum is achieved in the definition of @xmath723 , then @xmath724 2 .
[ item : bound from below on inf conv ] if @xmath720 , then for any @xmath725 , @xmath726_{c^{0,\eta}(\omega \times ( 0,t))}}}(2\theta { \ensuremath{[v]_{c^{0,\eta}(\omega \times ( 0,t))}}})^{\frac{\eta}{2-\eta}}\ ] ] and @xmath727_{c^{0,\eta}(\omega \times ( 0,t))}}}(2\theta { \ensuremath{[v]_{c^{0,\eta}(\omega \times ( 0,t))}}})^{\frac{\eta}{2-\eta}}.\ ] ] 3 .
[ item : inf conv is semiconcave ] in the sense of distributions , @xmath728 and @xmath729 for every @xmath725 .
[ item : lip bd on v ] we have @xmath730_{c_t^{0,\eta}(\omega\times(0,t))}}}\leq 3t\theta^{-1}$ ] .
[ inf conv is delta soln ] if @xmath120 is a @xmath0-supersolution of @xmath731 in @xmath149 , then @xmath638 is a @xmath0-supersolution of @xmath731 in @xmath732 .
if @xmath120 is a @xmath0-subsolution of @xmath731 in @xmath149 , then @xmath733 is a @xmath0-subsolution of @xmath731 in @xmath732 . in the following proposition
we summarize the facts that we need about inf and sup convolutions of mesh functions .
the proof is very similar to the elliptic case ( see proposition 2.3 of @xcite ) so we omit it . [
prop : inf - conv for mesh ] assume @xmath625 .
then : 1 .
[ eq : d(q , p^ * ) ] if @xmath734 is a point at which the infimum ( resp .
supremum ) is achieved in the definition of @xmath735 ( resp .
@xmath736 ) , then @xmath737 2 .
[ item : mesh : bound from below on inf conv ] assume @xmath738 and @xmath739 is a closest mesh point to @xmath46
. then @xmath740 and @xmath741 3 .
[ item : mesh inf conv is semiconcave ] in the sense of distributions , @xmath742 and @xmath743 for every @xmath636 .
[ item : mesh lip bd on v ] we have @xmath730_{c_t^{0,\eta}(\omega\times(0,t))}}}\leq 3t\theta^{-1}$ ] .
the author thanks her thesis advisor , professor takis souganidis , for suggesting this problem and for his guidance and encouragement . c. imbert , l. silvestre .
introduction to fully nonlinear parabolic equations .
+ http://blog-cyrilimbert.net/2012/06/20/lecture-notes-on-fully-nonlinear-parabolic-equations/ retrieved february 23 , 2013 .
krylov , n. v. on the rate of convergence of finite - difference approximations for bellman s equations .
algebra i analiz 9 ( 1997 ) , no .
3 , 245256 translation in st .
petersburg math .
j. 9 ( 1998 ) , no . 3 , 639 - 650 | we introduce the notion of @xmath0-viscosity solutions for fully nonlinear uniformly parabolic pde on bounded domains .
we prove that @xmath0-viscosity solutions are uniformly close to the actual viscosity solution , with an explicit error of order @xmath1 . as a consequence
we obtain an error estimate for implicit monotone finite difference approximations of uniformly parabolic pde . |
in this contribution , we briefly review the present status of our ongoing systematic study of the discovery potential of qcd - instanton induced events in deep - inelastic scattering ( dis ) at hera .
hard scattering processes in elementary particle physics are successfully described by the standard model of strong ( qcd ) and electro - weak ( qfd ) interactions in its perturbative formulation .
however , perturbation theory does not exhaust all possible hard scattering processes .
instantons @xcite , fluctuations of ( non - abelian ) gauge fields representing topology changing tunnelling transitions , induce interactions which are absent in conventional perturbation theory . in accord @xcite with the general chiral anomaly relation , instantons give rise to ( hard ) processes in which certain fermionic quantum numbers are violated , notably , chirality ( @xmath0 ) in ( massless ) qcd and baryon plus lepton number ( @xmath1 ) in qfd .
implications of qcd - instantons for long - distance phenomena have been intensively studied in the past , mainly in the context of the phenomenological instanton liquid model @xcite and of lattice simulations @xcite . yet , a direct experimental verification of their existence is lacking up to now .
an experimental discovery of such a novel , non - perturbative manifestation of non - abelian gauge theories would clearly be of basic significance .
the deep - inelastic regime is distinguished by the fact that here hard qcd - instanton induced processes may both be _ calculated _
@xcite within instanton - perturbation theory and possibly _ detected experimentally _ .
as a key feature it has recently been shown @xcite , that in dis the generic hard scale @xmath2 cuts off instantons with _ large size _ @xmath3 ,
over which one has no control within perturbation theory in the instanton background .
the leading instanton ( @xmath4)-induced process in the dis regime of @xmath5 scattering is displayed in fig .
[ ev - displ ] .
the dashed box emphasizes the so - called instanton-_subprocess _ with its own bjorken variables , @xmath6 the @xmath4-induced cross - section in unpolarized deep - inelastic @xmath7 scattering , @xmath8 , can be shown to factorize @xcite ( in the bjorken limit , @xmath9 large ) into a sum of differential parton - parton luminosities , @xmath10 , and @xmath4-subprocess cross - sections , @xmath11 , @xmath12 here @xmath13 denotes the virtual quarks entering the @xmath4-subprocess from the photon side and @xmath14 denotes the target partons . the differential luminosity @xmath10 , accounting for the number of @xmath15 collisions per @xmath16 collision , has a convolution - like structure @xcite , involving integrations over the target - parton density , @xmath17 , the @xmath18-flux , @xmath19 , and the _ known _ @xcite flux @xmath20 of the parton @xmath21 in the @xmath4-background . in eq .
( [ epcross ] ) , the @xmath4-subprocess total cross - section @xmath11 contains the essential instanton dynamics .
we have evaluated the latter @xcite by means of the optical theorem and the so - called @xmath22-valley approximation @xcite for the relevant @xmath23 forward elastic scattering amplitude in the @xmath22 background .
this method resums the exponentiating final state gluons in form of the known valley action @xmath24 and reproduces standard @xmath4-perturbation theory at larger @xmath22 separation @xmath25 .
while a semi - quantitative evaluation of @xmath11 may also be performed by means of standard @xmath4-perturbation theory , the use of the @xmath22-valley method allows to extend our calculations to a somewhat larger range in @xmath26 . corresponding to the symmetries of the theory
, the instanton calculus introduces at the classical level certain ( undetermined ) `` collective coordinates '' like the @xmath27-size parameters @xmath28 and the @xmath22 distance @xmath29 ( in units of the size ) .
observables like @xmath11 must be independent thereof and thus involve integrations over all collective coordinates .
hence , we have generically , @xmath30 the first important quantity of interest , entering eq.([sigma ] ) , is the @xmath4-density , @xmath31 ( tunnelling amplitude ) .
it has been worked out a long time ago @xcite in the framework of @xmath4-perturbation theory : ( renormalization scale @xmath32 ) @xmath33 in terms of the qcd @xmath34-function coefficients , @xmath35 . in this form
it satisfies renormalization - group invariance at the two - loop level @xcite .
note that the large , positive power @xmath36 of @xmath37 in the @xmath4-density ( [ density ] ) would make the integrations over the @xmath38-sizes in eq .
( [ sigma ] ) infrared divergent without the crucial exponential cut - off @xcite @xmath39 arising from the virtual quark entering the @xmath4-subprocess from the photon side .
the second important quantity of interest , entering eq.([sigma ] ) , is the @xmath22-interaction , @xmath40 . in the valley approximation ,
the @xmath22-valley action , @xmath41 $ ] , is restricted by conformal invariance to depend only on the `` conformal separation '' , @xmath42 , and its functional form is explicitly known @xcite .
it is important to note that the interaction between @xmath4 and @xmath43 remains _ attractive _ for all separations @xmath44 ; specifically , the @xmath22-valley action decreases monotonically from 1 at infinite conformal separation to 0 at @xmath45 , which corresponds to @xmath46 and @xmath47 . the collective coordinate integration in the cross - section ( [ sigma ] ) can be performed via _ saddle - point techniques_. one finds @xmath48 and @xmath49 , where the saddle - point solutions @xmath50 and @xmath51 behave qualitatively as @xmath52 thus , the virtuality @xmath53 controls the @xmath38-size : as one might have expected intuitively , highly virtual partons probe only small instantons .
the bjorken - variable @xmath26 , on the other hand , controls the conformal separation between @xmath4 and @xmath43 : for decreasing @xmath26 , the conformal separation decreases .
our quantitative results @xcite on the _ dominating _ cross - section for a target gluon , @xmath54 , are shown in detail in fig .
[ isorho ] , both as functions of @xmath55 ( left ) and of @xmath26 ( right ) .
the dotted curves in fig .
[ isorho ] , indicating lines of constant @xmath50 ( left ) and of constant @xmath56 ( right ) , nicely illustrate the qualitative relations ( [ qual - exp ] ) and their consequences : the @xmath53 dependence essentially maps the @xmath4-density , whereas the @xmath26 dependence mainly maps the @xmath22-interaction .
compared to our earlier estimates based on the one - loop renormalization - group invariant form @xcite of the @xmath4-density @xmath31 , the residual dependence on the renormalization scale has now dramatically weakened ( c.f .
[ rgdep ] ( left ) ) .
this important stabilization of our predictions originates from the use of the _ two - loop _ renormalization - group invariant improvement ( [ density ] , [ b - morretal ] ) of @xmath31 from ref .
@xcite .
intuitivelely one may expect @xcite @xmath57 .
indeed , this guess turns out to match quite well our actual choice of the `` best '' scale , @xmath58 , determined by @xmath59 ( c.f . fig .
[ rgdep ] ( left ) ) .
important information about the range of validity of @xmath4-perturbation for the @xmath4-density and the @xmath22-interaction , in terms of the instanton collective coordinates ( @xmath60 ) , can be obtained from recent ( non - perturbative ) lattice simulations of qcd and translated via the saddle - point relations ( [ qual - exp ] ) into a `` fiducial '' kinematical region ( @xmath61 ) @xcite .
in fact , from a comparison of the ( one - loop ) perturbative expression of the @xmath4-density ( [ density ] ) with recent lattice `` data '' @xcite one infers @xcite semi - classical @xmath4-perturbation theory to be valid for @xmath62 fm ( c.f .
[ rgdep ] ( right ) ) .
similarly , it can be argued @xcite that the attractive , semi - classical valley result for the @xmath22-interaction is supported by lattice simulations down to a minimum conformal separation @xmath63 , i.e. @xmath64 . the corresponding `` fiducial '' kinematical region for our cross - section predictions in dis is then obtained as @xcite @xmath65 \frac{r^\ast}{\rho^\ast}&{\mbox{\,\raisebox{.3ex } { $ > $ } $ \!\!\!\!\!$\raisebox{-.9ex}{$\sim$}}\ , } & 1\\ \end{array}\right\}\rightarrow \left\{\begin{array}{lclcl}q^\prime&\geq & q^\prime_{\rm min}&\simeq & 8 { \rm\ gev};\\[1ex ] x^\prime&\geq & x^\prime_{\rm min}&\simeq & 0.35.\\ \end{array } \right .
\label{fiducial}\ ] ] fig .
[ sighera ] displays the finalized @xmath4-induced cross - section at hera , as function of the cuts @xmath66 and @xmath67 , as obtained with the new release `` qcdins 1.6.0 '' @xcite of our @xmath4-event generator . for the following `` standard cuts '' , @xmath68 including the minimal cuts ( [ fiducial ] ) extracted from lattice simulations , we specifically obtain @xmath69 where the uncertainties result mainly from the experimental uncertainty in the qcd scale @xmath70 .
hence , with the total luminosity accumulated by experiments at hera , @xmath71 pb@xmath72 , there should be already @xmath73 @xmath4-induced events from the kinematical region ( [ standard - cuts ] ) on tape . note also that the cross - section quoted in eq .
( [ minimal - cuts ] ) corresponds to a fraction of @xmath4-induced to normal dis ( ndis ) events of @xmath74 thus , it appears to be a question of _ signature _ rather than a question of _ rate _ to discover @xmath4-induced scattering processes at hera .
hence , we turn next to the _ final states _ of @xmath4-induced events in dis .
a monte - carlo generator , qcdins , for instanton - induced events in dis , interfaced to herwig , has been developed .
the monte - carlo simulation essentially proceeds in three steps .
first , quasi - free partons are produced by qcdins with the distributions prescribed by the hard process matrix elements .
next , these primary partons give rise to parton showers , as described by herwig .
finally , the showers are converted into hadrons , again within herwig . in fig .
[ typical ] we display the lego plot of a typical @xmath4-induced event at hera , as generated by qcdins .
its characteristics directly reflect the essential features of the underlying @xmath4-subprocess and thus may be intuitively understood : after hadronization , the current quark in fig .
[ ev - displ ] gives rise to a current - quark jet .
no further distinct qjets are expected , since the partons from the @xmath4-subprocess are emitted spherically symmetric in the @xmath75 c.m .
system ( `` @xmath4-c.m . system '' ) .
the gluon multiplicities are generated according to a poisson distribution with mean multiplicity @xmath76 . in view of the large required chirality violation @xmath77 ,
the total mean parton multiplicity is large , of the order of ten .
after hadronization , we therefore expect from the @xmath4-subprocess a final state structure reminiscent of a decaying fireball : @xmath78 hadrons are produced , always including _ strange _ ones .
they are concentrated in a `` band '' at fixed pseudorapidity @xmath79 in the ( @xmath79 , azimuth angle @xmath80)-plane . due to the boost from the @xmath4-c.m .
system to the hera - lab system , the center of the band is shifted away from @xmath81 .
its width is of order @xmath82 , as typical for a spherically symmetric event . for @xmath83 and @xmath84 gev ,
the total invariant mass of the @xmath4-system , @xmath85 , is expected to be in the 10 gev range .
all these expectations are clearly reproduced by our monte - carlo simulation .
these features have been exploited by experimentalists at hera to place first upper limits on the fraction of @xmath4-induced events to normal dis events , in a similar kinematical region as our standard cuts ( [ standard - cuts ] ) : from the search of a @xmath86 excess in the `` band '' region , the h1 collaboration could establish a limit of @xmath87 % , while the search of an excess in charged multiplicity yields @xmath88 % @xcite .
the limit from the charged multiplicity distribution has been further improved in ref .
@xcite to about 1 % .
since these experimental upper limits already range close to our estimate ( [ fraction - fiducial ] ) , it becomes extremely interesting to investigate further possibilities to discriminate @xmath4-induced from ndis final states .
a dedicated multi - observable analysis is in progress , with the aim of producing an instanton - enriched data sample .
furthermore , strategies to reconstruct @xmath26 and @xmath55 from the final state are being developped .
99 a. ringwald and f. schrempp , desy 98 - 081 , hep - ph/9806528 , _ phys .
_ b , in press .
s. moch , a. ringwald and f. schrempp , .
a. ringwald and f. schrempp , desy 94 - 197 , hep - ph/9411217 , in : proc . _
quarks 94 _ , vladimir , eds . d. gigoriev et al . ,
170 - 193 .
m. gibbs , a. ringwald and f. schrempp , hep - ph/9506392 , in : _ proc .
dis 95 _ , paris , eds .
f . laporte and y. sirois , pp .
341 - 344 .
a. ringwald and f. schrempp , desy 97 - 115 , hep - ph/9706399 , in : _ proc .
dis 97 _ , chicago , eds . j. repond and d. krakauer , pp .
781 - 786 .
t. carli , a. ringwald and f. schrempp , in preparation .
a. belavin , a. polyakov , a. schwarz and yu .
tyupkin , . g.
t hooft , ; ; ( erratum ) . t. schfer and e.v .
shuryak , .
p. van baal , .
i. balitsky and v. braun , .
a. yung , ; v.v .
khoze and a. ringwald , ; j. verbaarschot , .
t. morris , d. ross and c. sachrajda , .
d. smith and m. teper , hep - lat/9801008 , .
et al . _ , h1 collaboration , ; ; t. carli and m. kuhlen , . | the present status of our ongoing systematic study of the discovery potential of qcd - instanton induced events in deep - inelastic scattering at hera is briefly reviewed .
we emphasize our recent progress in predicting the cross - sections of instanton - induced processes .
our finalized predictions include a dramatic improvement of the residual renormalization - scale dependencies and the specification of a `` fiducial '' kinematical region in the relevant bjorken variables extracted from recent lattice simulations .
published upper limits on instanton - induced cross - sections based on single observables in the final state such as the flow of strange particles and the multiplicity distribution of charged particles are already of the order of our estimate .
thus , a decisive search for instanton - induced events in deep - inelastic scattering at hera , based on a multi - observable analysis , seems feasible . |
the classical secretary problem is a well known optimal stopping problem from probability theory .
it is usually described by different real life examples , notably the process of hiring a secretary .
imagine a company manager in need of a secretary .
our manager wants to hire only the best secretary from a given set of @xmath1 candidates , where @xmath1 is known .
no candidate is equally as qualified as another .
the manager decides to interview the candidates one by one in a random fashion .
every time he has interviewed a candidate he has to decide immediately whether to hire her or to reject her and interview the next one . during the interview process he can only judge the qualities of those candidates he has already interviewed .
this means that for every candidate he has observed , there might be an even better qualified one within the set of candidates yet to be observed .
of course the idea is that by the time only a small number of candidates remain unobserved , a recently interviewed candidate that is relatively best will probably also be the overall best candidate .
there is abundant research literature on this classical secretary problem , for which we refer to ferguson @xcite for an historical note and an extensive bibliography .
the exact optimal policy is known , and may be derived by various methods , see for instance dynkin and yushkevich @xcite , and gilbert and mosteller @xcite .
also , many variations and generalizations of the original problem have been introduced and analysed .
one of these generalizations is the focus of our paper , namely the problem to select one of the @xmath0 best , where @xmath4 is some preassigned number ( notice that @xmath5 is the classical secretary problem ) .
originally , this problem was introduced by gusein - zade @xcite , who derived the structure of the optimal policy : there is a sequence @xmath6 of position thresholds such that when candidate @xmath7 is presented , and judged to have relative rank @xmath8 among the first @xmath7 candidates , with rank 1 being the best , rank 2 being second best , etc . ]
, then the optimal decision says @xmath9 furthermore , @xcite gave an algorithm to compute these thresholds , and derived asymptotic expressions ( as @xmath3 ) for the @xmath10 case .
also frank and samuels @xcite proposed an algorithm , and gave the limiting ( as @xmath3 ) probabilities and limiting proportional thresholds @xmath11 .
the algorithms of @xcite are based on dynamic programming , which means that the optimal thresholds @xmath12 , and the optimal winning probability are determined numerically .
the next interest was to find analytic expressions .
to our best knowledge , this has been resolved only for @xmath10 by gilbert and mosteller @xcite , and for @xmath13 by quine and law @xcite .
although the latter claim that their approach is applicable to produce exact results for any @xmath0 , it is clear that the expressions become rather untractable for larger @xmath0 .
this has inspired us to develop approximate results for larger @xmath0 .
we consider two approximate policies for the general @xmath0 case : single - level policies , and double - level policies .
a single - level policy is given by a single position threshold @xmath14 in conjuction with a rank level @xmath15 , such that when candidate @xmath7 is presented , and judged to have relative rank @xmath8 among the first @xmath7 candidates , then the policy says @xmath16 a double - level policy is given by two position thresholds @xmath17 in conjuction with two rank levels @xmath18 , such that when candidate @xmath7 is presented , and judged to have relative rank @xmath8 among the first @xmath7 candidates , then the policy says @xmath19 we shall derive the exact winning probability for these two approximate policies , when the threshold and level parameters are given .
these expressions can then used easily to compute the optimal single - level and the optimal double - level policies , i.e. , we optimize the winning probabilities ( under these level policies ) with respect to their threshold and level parameters .
the most important result is that the winning probabilities of the optimal double - level policies are extremely close to the winning probabilities of the optimal policies ( with the @xmath0 thresholds ) , specifically for larger @xmath0 . in other words
, we have found explicit formulas that approximate closely the winning probabilities for this generalized secretary problem . as an example , we present in table [ t : resdlp ] the relative errors in percentages for a few @xmath20 combinations ( a more extended table can be found in section [ s : num ] ) .
.__relative errors ( % ) of the optimal double - level policies .
_ _ [ cols="<,>,>,>",options="header " , ]
for the considered generalized secretary problem of selecting one of the @xmath21 best out of a group of @xmath22 we have obtained closed expressions for the probability of success for all possible single- and double - level policies . for
any given finite values of @xmath22 and @xmath21 these expressions can be used to obtain the optimal single - level policy respectively optimal double - level policy in a straightforward manner .
moreover , asymptotically for @xmath23 we have also obtained closed expressions for the winning probability for relevant families of single - level and double - level policies .
optimizing this expression for the family of single - level policies an asymptotic optimal rank level @xmath24 and corresponding optimal position threshold fraction @xmath25 and asymptotic winning probability are easily obtained .
similarly we have done such asymptotic analysis and optimization for the relevant family of double - level policies . both for the single - level and double - level policies we confirmed numerically for @xmath26 that the winning probabilities for optimal finite and double level policies for finite values of @xmath22 converge if @xmath22 increases to the ( respectively single - level and double - level ) optimal asymptotic winning probabilities . finally , we computed for varying @xmath21 and @xmath22 the optimal single - level and double - level policies and corresponding winning probabilities and compared the results to the overall optimal policy which is determined by @xmath21 position thresholds .
we found that the single - level policies and especially the double - level policies perform nearly as well as the overall optimal policy .
in particular for a generalized secretary problem with a larger value of @xmath21 applying the optimal single - level or double - level policy could be considered , because implementation of the overall optimal policy using @xmath21 different thresholds is unattractive compared to using only one or two thresholds for implementing the policy .
besides for large @xmath21 the gain in performance of the overall optimal policy over the optimal double - level policy is very small . 999 dynkin , e.b . and yushkevich , a. 1969 .
_ markov processes : theorems and problems _ , plenum press , new york .
ferguson , t.s .
1989 . who solved the secretary problem ?
_ statistical science _ 4 , 282 - 296 .
frank , a.q . and samuels , s.m . 1980 . on an optimal stopping problem of gusein - zade .
_ stochastic processes and their applications _ 10 , 299 - 311 .
gilbert , j.p . and mosteller , f. 1966 .
recognizing the maximum of a sequence .
_ journal of the american statistical association _ 61 , 35 - 73 .
gusein - zade , s.m .
the problem of choice and the optimal stopping rule for a sequence of indpendent trials .
_ theory of probability and its applications _ 11 , 472 - 476 .
quine , m.p . and
law , j.s .
exact results for a secretary problem .
_ journal of applied probability _ 33 , 630 - 639 .
the dynamic programming method might be applied to find numerically the optimal single- , double- and multiple - level ( ` full ' ) policies . here , we summarize the algorithm for the single - level policy ; it is straightforward how to generalize the algorithm to the double - level , and the multiple - level cases .
define the single - level policy with threshold @xmath14 and level @xmath15 , denoted @xmath27 , by its actions @xmath28 where 1 means to continue , and 0 means to stop and select this candidate .
we restrict to @xmath29 . denote by @xmath30 the probability of winning when @xmath27 is applied .
given level @xmath15 we determine the optimal threshold @xmath31 , defined by @xmath32 we use dynamic programming to find it .
define for @xmath33 the value @xmath34 to be the maximal probability of winning when @xmath35 is observed , and @xmath36 to be the maximal probability of winning when @xmath37 .
the optimality equations are : @xmath38 for @xmath39 @xmath40 and for @xmath41 ( since then surely @xmath35 and @xmath42 ) : @xmath43 one can show that the result of this dp recursion is indeed a slp by setting @xmath44 .
moreover , note that probabilities @xmath45 occuring in the optimality equations can easily be obtained , for example by applying lemma [ l : hypprob ] and bayes rule . | a version of the classical secretary problem is studied , in which one is interested in selecting one of the @xmath0 best out of a group of @xmath1 differently ranked persons who are presented one by one in a random order .
it is assumed that @xmath2 is a preassigned number .
it is known , already for a long time , that for the optimal policy one needs to compute @xmath0 position thresholds , for instance via backwards induction . in this paper
we study approximate policies , that use just a single or a double position threshold , albeit in conjunction with a level rank .
we give exact and asymptotic ( as @xmath3 ) results , which show that the double - level policy is an extremely accurate approximation .
* keywords : * secretary problem ; dynamic programming ; approximate policies |
this article is devoted to the study of the nonlinear integro - differential problem @xmath0 for @xmath1 .
here , @xmath2 @xmath3 denotes an open and bounded set , @xmath4 , @xmath5 and @xmath6 .
the _ range _
kernel @xmath7 is given as a rescaling @xmath8 of a kernel @xmath9 satisfying the usual properties of nonnegativity and smoothness .
we shall give the precise assumptions in section [ sec.results ] .
we shall refer to problem ( [ eq.orig])-([eq.orig2 ] ) as to problem p@xmath10 .
the main results contained in this article are : * theorem [ th.existenceu ] .
the well - posedness of problem p@xmath10 , the stability property of its solutions with respect to the initial datum , and the time invariance of the level set structure of its solutions . * theorem [ th.equiv ] .
the equivalence between solutions of problem p@xmath10 and the one - dimensional problem p@xmath11 , where @xmath12 , and @xmath13 is the decreasing rearrangement of @xmath14 , see section [ sec.dec_re ] for definitions .
* theorem [ th.pde ] .
the asymptotic behavior of the solution of problem p@xmath11 with respect to the window size parameter , @xmath15 , as a shock filter . problem is related to some problems arising in image analysis , population dynamics and other disciplines .
the general formulation in ( [ eq.orig ] ) includes , for example , a time - continuous version of the neighborhood filter ( nf ) operator : @xmath16 where @xmath15 is a positive constant , and @xmath17 is a normalization factor . in terms of the notation introduced for problem
the nf is recovered setting @xmath18 and @xmath19 .
this well known denoising filter is usually employed in the image community through an iterative scheme , @xmath20 with @xmath21 .
it is the simplest particular case of other related filters involving nonlocal terms , notably the yaroslavsky filter @xcite , the bilateral filter @xcite , and the nonlocal means filter @xcite .
these methods have been introduced in the last decades as efficient alternatives to local methods such as those expressed in terms of nonlinear diffusion partial differential equations ( pde s ) , among which the pioneering nonlinear anti - diffusive model of perona and malik @xcite , the theoretical approach of lvarez et al . @xcite and the celebrated rof model of rudin et al .
we refer the reader to @xcite for a review comparing these local and non - local methods .
another image processing task encapsulated by problem is the _ histogram prescription _ , used for image contrast enhancement : given an initial image @xmath14 , find a companion image @xmath22 such that @xmath22 and @xmath14 share the same level sets structure , and the histogram distribution of @xmath22 is given by a prescribed function @xmath23 .
a widely used choice is @xmath24 , implying that @xmath22 has a uniform histogram distribution . in this case ,
@xmath25 and @xmath26 is related to the image size and its dynamic range , see sapiro and caselles @xcite for the formulation and analysis of the problem .
nonlinear integro - differential of the form @xmath27 and other nonlinear variations of it have also been recently used ( andreu et al .
@xcite ) to model diffusion processes in population dynamics and other areas .
more precisely , if @xmath28 is thought of as a density at the point @xmath29 at time @xmath30 and @xmath31 is thought of as the probability distribution of jumping from location @xmath32 to location @xmath29 , then @xmath33 is the rate at which individuals are arriving at position @xmath29 from all other places and @xmath34 is the rate at which they are leaving location @xmath29 . in the absence of external or internal sources
this consideration leads immediately to the fact that the density @xmath22 satisfies the equation ( [ mazon ] ) .
these kind of equations are called nonlocal diffusion equations since in them the diffusion of the density @xmath22 at a point @xmath29 and time @xmath30 depends not only on @xmath35 but also on the values of @xmath22 in a set determined ( and weighted ) by the space kernel @xmath36 .
a thoroughfull study of this problem may be found in the monograph by andreu et al .
observe that in problem , the space kernel is taken as @xmath37 , meaning that the influence of nonlocal diffusion is spread to the whole domain . as noticed by sapiro and caselles @xcite for the histogram prescription problem , and later by kindermann et al .
@xcite for the iterative neighborhood filter ( [ def.gfv ] ) , or by andreu et al .
@xcite for continuous time problems like ( [ mazon ] ) , these formulations may be deduced from variational considerations .
for instance , in @xcite , the authors consider , for @xmath38 , the functional @xmath39 with an appropriate spatial kernel @xmath36 , and a differentiable filter function @xmath40 .
then , the authors formally deduce the equation for the critical points of @xmath41 .
these critical points coincide with the fixed points of the nonlocal filters they study .
for instance , if @xmath42 and @xmath37 , the critical points satisfy @xmath43 which can be solved through a fixed point iteration mimicking the iterative neighborhood filter scheme ( [ def.gfv ] ) .
on the other hand , choosing @xmath44 ( or some suitable nonlinear variant ) and considering a gradient descent method to approximate the stationary solution , equation ( [ mazon ] ) is deduced .
similarly , @xmath45 and @xmath37 leads to the histogram prescription problem . although the functional ( [ def.functional ] ) is not convex in general , kindermann et al .
prove that when @xmath9 is the gaussian kernel then the addition to @xmath41 of a convex fidelity term , e.g. @xmath46 gives , for @xmath47 large enough , a convex functional @xmath48 , see ( * ? ? ?
* theorem 3.1 ) .
thus , the functional @xmath48 may be seen as the starting point for the deduction of problem , representing the continuous gradient descent formulation of the minimization problem modeling gaussian image denoising .
notice that although the convexity of @xmath48 is only ensured for @xmath26 large enough , the results obtained in this article are independent of such value , and only the usual non - negativity condition on @xmath26 is assumed . the outline of the article is as follows . in section [ sec.dec_re ] , we introduce some basic notation and the definition of _ decreasing rearrangement _ of a function .
this is later used to show the equivalence between the general problem and the reformulation p@xmath11 in terms of a problem with a identical structure but defined in a one - dimensional space domain .
this technique was already used in @xcite for dealing with the time - discrete version of problem , in the form of the iterative scheme ( [ def.gfv ] ) .
see also @xcite for the problem with non - uniform spatial kernel .
in section [ sec.results ] , we state our main results .
then , in section [ sec.numerics ] , we introduce a discretization scheme for the efficient approximation of solutions of problem , and demonstrate its performance with some examples . in section [ sec.proofs ] , we provide the proofs of our results , and finally , in section [ sec.conclusions ] , we give our conclusions .
given an open and bounded ( measurable ) set @xmath2 , @xmath3 let us denote by @xmath49 its lebesgue measure and set @xmath12 . for a lebesgue measurable function @xmath50 , the function @xmath51 is called the _ distribution function _ corresponding to @xmath22 .
function @xmath52 is , by definition , non - increasing and therefore admits a unique generalized inverse , called its _
decreasing rearrangement_. this inverse takes the usual pointwise meaning when the function @xmath22 has not flat regions , i.e. when @xmath53 for any @xmath54 . in general , the decreasing rearrangement @xmath55 is given by : @xmath56 notice that since @xmath57 is non - increasing in @xmath58 , it is continuous but at most a countable subset of @xmath58 . in particular , it is right - continuous for all @xmath59 $ ] .
the notion of rearrangement of a function is classical and was introduced by hardy , littlewood and polya @xcite .
applications include the study of isoperimetric and variational inequalities @xcite , comparison of solutions of partial differential equations @xcite , and others .
we refer the reader to the monograph @xcite for a extensive research on this topic .
two of the most remarkable properties of the decreasing rearrangement are the equi - measurability property @xmath60 for any borel function @xmath61 , and the contractivity @xmath62 for @xmath63 , @xmath64 $ ] . for the extension of the decreasing rearrangement to families of functions depending on a parameter ,
e.g. @xmath65 $ ] , we first consider , for @xmath30 fixed , the function @xmath66 given by @xmath67 , for any @xmath68 . then we define @xmath69 by @xmath70 .
our first result ensures the well - posedness of problem for @xmath71 initial data with bounded total variation .
in addition , we show that the level sets structure of the solution is time invariant . before stating our results ,
we collect here the main assumptions on the data problem , to which we shall refer to as * ( h ) * : * @xmath2 is an open , bounded , and connected set ( @xmath72 ) .
* the final time , @xmath73 , which simulate the time horizon of the diffusion process is a real , positive fixed number . *
the parameter @xmath26 is a real , nonnegative fixed number .
* @xmath74 is nonnegative .
* @xmath75 is assumed to be , without loss of generality , non - negative .
basic facts but also advanced results about the space of bounded variation @xmath76 can be found in the book by ambrosio et al .
notice that , depending on the space dimension @xmath77 we have the continuous injections @xmath78 . when @xmath79 we have @xmath80 .
[ th.existenceu ] assume ( h ) .
then there exists a unique solution @xmath81;{{\cal x}})$ ] of problem . in addition , if @xmath82 and @xmath83;{{\cal x}})$ ] are the corresponding solutions to problems @xmath84 then @xmath85 for some constant @xmath86 .
finally , suppose that @xmath87 for some @xmath88 .
then @xmath89 for all @xmath90 $ ] .
the existence and stability results of theorem [ th.existenceu ] may be extended to more general zero - order terms in the equation ( [ eq.orig ] ) of problem . for instance , we can consider a function @xmath91\times\o\times{\mathbb{r}}\to{\mathbb{r}}$ ] satisfying @xmath92 , @xmath93 , and @xmath94 .
this regularity coincides with the initially obtained for the integral term of equation ( [ eq.orig ] ) in the approximation procedure to construct the solution .
in addition , if @xmath87 implies @xmath95 , then the time invariance of level sets holds . replacing the set @xmath96 by @xmath97 and the initial data @xmath98 by @xmath99 , theorem [ th.existenceu ] ensures the existence of a solution of problem p@xmath11 .
observe that @xmath100 is bounded because @xmath2 is bounded ( assumption @xmath101 ) and this implies @xmath102 and @xmath103 .
in the following result we obtain some properties of solutions of the one - dimensional problem p@xmath104 .
although the corollary is valid for any interval in @xmath105 , we keep the notation @xmath97 for simplicity .
the corresponding result for the discrete - time version , with @xmath106 , of problem p@xmath11 may be found in @xcite .
[ th.existencev ] assume ( h ) , and let @xmath107;{{\cal x}}_*)$ ] be the solution of problem p@xmath104 , for some nonincreasing @xmath108
. then 1 .
@xmath109 a.e . in @xmath97 , for all @xmath110 .
2 . for @xmath111 , @xmath112 and @xmath113 .
3 . if @xmath9 is odd then @xmath114 , for @xmath111 .
if @xmath115 and @xmath116 , for @xmath117 and @xmath118 , then @xmath119;w^{m , p}(\o_*))$ ] . 5 .
if @xmath106 and @xmath120 is such that @xmath121 then @xmath122 as @xmath123 .
condition in point 3 is a natural symmetry condition for convolution kernels and it is satisfied , for instance , by the gaussian kernel .
condition in point 5 is also satisfied by the gaussian kernel , if @xmath15 is large enough .
the next result establishes the connection between problems and p@xmath11 .
[ th.equiv ] assume ( h ) .
then , @xmath124;{{\cal x}})$ ] is a solution of if and only if @xmath125;{{\cal x}}_*)$ ] is a solution of p@xmath11 .
theorem [ th.equiv ] implies that the solution of the multi - dimensional problem may be constructed by solving the one - dimensional problem p@xmath11 .
indeed , using the level sets invariance asserted in theorem [ th.existenceu ] , we deduce @xmath126 for all @xmath65 $ ] . when image processing applications are considered , by property 1 of corollary [ th.existencev ] , the solution to
may be understood as a _
contrast change _ of the initial image , @xmath14 .
indeed , this property also implies that if , initially , @xmath14 has no flat regions , and therefore @xmath127 is decreasing , then the solution of p@xmath11 verifies this property for all @xmath111 .
then , theorem [ th.existencev ] implies that the solution of has no flat regions for all @xmath111 .
the last theorem is an extension of a result given in @xcite for the discrete - time formulation with @xmath106 . in it , we deduce the asymptotic behavior of the solution @xmath57 of problem p@xmath11 ( and thus of @xmath22 of problem ) in terms of the window size parameter , @xmath15 .
although we state it for the gaussian kernel , more general choices are possible , see ( * ? ? ?
* remark 2 ) .
[ th.pde ] assume ( h ) with @xmath128 and @xmath98 having no flat regions .
suppose , in addition , that @xmath129
. then , for all @xmath130\times \o_*$ ] , there exist positive constants @xmath131 independent of @xmath15 such that the solution @xmath132;c^3(\bar\o_*))$ ] of p@xmath11 satisfies @xmath133 with @xmath134 and with @xmath135 , and @xmath136 .
two interesting effects captured by ( [ app.nfstar ] ) are the following : 1 . the border effect ( range shrinking ) .
function @xmath137 is _ active _ only when @xmath138 is close to the boundaries , @xmath139 and @xmath140 . for @xmath141
, @xmath142 contributes to the decrease of the largest values of @xmath57 while for @xmath143 we have @xmath144 , increasing the smallest values of @xmath57 .
therefore , this term tends to flatten @xmath57 . in image processing terms , a loss of contrast is induced .
the term @xmath145 is anti - diffusive , inducing large gradients on @xmath146 in a neighborhood of inflexion points . in this sense ,
the scheme ( [ app.nfstar ] ) is related to the shock filter introduced by lvarez and mazorra @xcite @xmath147 where @xmath148 is a smoothing kernel and function @xmath149 satisfies @xmath150 for any @xmath151 . indeed , neglecting the fidelity , the border and the lower order terms , and defining @xmath152 , we render ( [ app.nfstar ] ) to the form ( [ alvarez ] ) .
+ this property can be exploited to produce a partition of the image so the model can be interpreted as a tool for fast segmentation and classification .
an example is proposed in the numerical experiments where a time - continuous version of the nf is implemented .
for the discretization of problem , for @xmath153 , we take advantage of the equivalence result stated in theorem [ th.equiv ] .
thus , we first calculate a numerical approximation , @xmath154 , to the decreasing rearrangement @xmath155 and consider the problem p@xmath156 . then , we discretize this one - dimensional problem and compute a numerical approximation , @xmath157\times\o_*\to{\mathbb{r}}$ ] . by theorem [ th.equiv ] , @xmath158 is , in fact , an approximation to @xmath57 , where @xmath159\times\o\to{\mathbb{r}}$ ] is a solution to problem . then , we finally recover an approximation , @xmath160 , to @xmath22 by defining @xmath161 inspired by the image processing application of problem , we consider a piecewise constant approximation to its solutions .
let @xmath162 be , for simplicity , a rectangle domain and consider a uniform mesh on @xmath96 enclosing square elements ( pixels ) , @xmath163 , of unit area , with barycenters denoted by @xmath164 , for @xmath165 and @xmath166 .
given @xmath167 , we consider its piecewise constant interpolator @xmath168 if @xmath169 .
the interpolator @xmath170 has a finite number , @xmath171 , of quantized levels that we denote by @xmath172 , with @xmath173 .
that is @xmath174 where @xmath175 are the level sets of @xmath170 , @xmath176 since @xmath170 is piecewise constant , the decreasing rearrangement of @xmath170 is piecewise constant too , and given by @xmath177 with @xmath178 for @xmath179 , and @xmath180 , @xmath181 , @xmath182,@xmath183,@xmath184 .
let @xmath158 be a candidate to solve problem p@xmath156 .
due to the time - invariance of the level sets structure of the solution to this problem , see theorem [ th.existenceu ] , we may express @xmath158 as @xmath185 with @xmath186 , for @xmath90 $ ] , @xmath187 , for @xmath179 .
substituting @xmath158 in equation ( [ eq.orig ] ) , we get , for @xmath188 and @xmath179 , @xmath189 with @xmath190 . since , by assumptions ( h ) , the right hand side of ( [ eq.dis ] ) is lipschitz continuous , the existence and uniqueness of a smooth @xmath191\to { \mathbb{r}}_+^q$ ] satisfying ( [ eq.dis ] ) and @xmath192 follows . for the time discretization
, we take a uniform mesh of the interval @xmath193 $ ] of size @xmath194 , and use the notation @xmath195 , with @xmath196 , and @xmath197 then , we consider the following implicit time discretization of problem ( [ eq.dis ] ) . for @xmath179 and @xmath198 , solve @xmath199 since problem ( [ eq.dis2 ] ) is a nonlinear algebraic system of equations , we use a fixed point argument to approximate its solution , @xmath200 , at each discrete time @xmath201 , from the previous approximation @xmath202 .
let @xmath203 .
then , for @xmath204 the problem is to find @xmath205 solving the linear system @xmath206 for @xmath179 .
we choose the stopping criterion @xmath207 , for values of @xmath208 chosen empirically , and then set @xmath209 .
finally , using formula ( [ def.u5 ] ) , the expression of the initial datum ( [ def.u02 ] ) , and the definition ( [ sol.v ] ) , we recover a piecewise constant approximation to the original problem , , taking @xmath210 as an application we consider a grand challenge in biomedical image analysis .
this is a computer vision problem in biomedicine which consists of overlapping cells segmentation and subcellular nucleus and cytoplasm detection , see @xcite , @xcite .
the dataset was downloaded from the _ overlapping cervical cytology image segmentation challenge_carneiro
/ isbi14_challenge / index.html ] , isbi 2014 . the data set is composed by @xmath211 real and synthetic images containing two or more cells with different degrees of overlapping , contrast , and texture .
the phantom images allow the quantitative analysis of segmentation procedures through their ground - truth , which is carried out by using the dice similarity coefficient , @xmath212 : for two sets ( images ) @xmath213 and @xmath214 , @xmath215 observe that values of @xmath212 close to one indicate high coincidence of the images , that is , of the ground - truth segmentation and the segmentation obtained with our method . for running our algorithm ,
that is , providing an approximation , @xmath216 , of ( [ eq.dis ] ) , we consider the usual number of image quantization levels , @xmath217 .
the fidelity term is ignored ( @xmath218 ) , and the range window parameter , @xmath15 , is set as @xmath219 for nucleus detection , and as @xmath220 for cytoplasm detection .
the tolerance in the fixed point loop ( [ eq.dis3 ] ) is taken as @xmath221 . as a stopping criterion
, we consider a combination of a maximum number of time iterations ( 1000 ) , and an energy stabilization criterion , @xmath222 where @xmath223 is the discrete version of the functional given by ( [ def.functional ] ) , for @xmath37 and @xmath224 .
finally , we implement a variable time step , @xmath225 , inspired by the proof of existence of solutions and given by , for @xmath226 , @xmath227 with @xmath228 . in the experiments , we observed that @xmath225 ranges from order @xmath229 in the first iterations to order @xmath230 just before convergence .
we summarize our results for the test90 dataset in table [ table1 ] , where we show the @xmath212 for some specific samples , and the mean @xmath212 of the ninety samples contained in the dataset .
we may check that @xmath212 values are very high for the segmentation of both regions of interest ( cytoplasm and nucleus ) , always above the range obtained in @xcite .
the execution times are given for a matlab implementation of the algorithm , running on a standard laptop ( intel core i7 - 2.80 ghz processor , 8 gb ram ) .
in figure [ fig1 ] , we show the segmentation process for the two regions of interest .
the first column corresponds to the initial image .
the second column , to the background extraction , and the third column to the nucleus segmentation .
thus , the cytoplasm is the difference between the images shown in the third and second column .
finally , the fourth column shows the difference between the ground - truth nucleus segmentation and the obtained with our method .
.example : segmentation results for some samples of test90 dataset . [ cols="<,^,^,^,^,^,^,^,^",options="header " , ] [ table1 ]
* proof of theorem [ th.existenceu ] . *
we divide the proof in several steps .
* step 1 . * existence of a local in time solution to an auxiliary problem with smooth data .
we assume @xmath231 , and consider the following auxiliary problem , obtained using the change of unknown @xmath232 in ( [ eq.orig ] ) , for some positive constant @xmath233 to be fixed : @xmath234 for @xmath235 , and for the initial data @xmath236 . here
, @xmath237 will be fixed later .
_ time discretization . _
let @xmath238 , @xmath239 and @xmath240 , for @xmath241 .
assume that @xmath242 is given and consider the functional @xmath243 given by @xmath244 for @xmath245 .
observe that if @xmath213 has a fixed point @xmath246 , then we may define @xmath247 to get the following semi - implicit version of ( [ eq.aux ] ) @xmath248 we have , @xmath249 therefore , for @xmath250 , the mapping @xmath213 is contractive in @xmath251 , and a unique fixed point , @xmath252 verifying ( [ eq.discrete ] ) does exist .
we have the following uniform estimates for @xmath252 .
one one hand , from ( [ eq.discrete ] ) we obtain @xmath253 which gives ( recall @xmath254 ) the uniform estimate @xmath255 with @xmath256 depending only on @xmath257 . on the other hand , since @xmath258 , we deduce from ( [ eq.discrete ] ) @xmath259 .
this regularity allows to differentiate in ( [ eq.discrete ] ) with respect to the @xmath260th component of @xmath29 , denoted by @xmath261 , to obtain for a.e .
@xmath68 , @xmath262 with @xmath263 from where we deduce @xmath264 solving this differences inequality , we find that , by redefining @xmath233 to satisfy @xmath265 , we obtain the uniform estimate @xmath266 , with @xmath267 depending only on @xmath268 .
this election of @xmath233 is possible by restricting @xmath269 to be @xmath270 _ time interpolators and passing to the limit @xmath271 .
_ we define , for @xmath272\times \o$ ] , the piecewise constant and piecewise linear interpolators @xmath273 using the uniform @xmath274 estimates of @xmath252 and @xmath275 , we deduce the corresponding uniform estimates for @xmath276 , @xmath277 and @xmath278 , implying the existence of @xmath279 and @xmath280 such that , at least in a subsequence ( not relabeled ) , as @xmath281 , @xmath282 in particular , by compactness @xmath283\times \bar \o).\ ] ] since , for @xmath284 $ ] , @xmath285 we deduce both @xmath286 and @xmath287\times \bar \o).\ ] ] considering the shift operator @xmath288 , and introducing the approximation @xmath289 , for @xmath290 $ ] , we may rewrite ( [ eq.discrete ] ) as @xmath291 and due to the convergence properties ( [ conv.1 ] ) and ( [ conv.2 ] ) , we may pass to the limit @xmath271 in ( [ eq.final ] ) to deduce that @xmath36 is a solution of ( [ eq.aux ] ) .
_ continuation of the solution to an arbitrary time @xmath73 .
_ given the solution , @xmath36 , of problem ( [ eq.aux ] ) in @xmath292 , we may consider the same problem for the initial datum @xmath293 .
since @xmath294 and the constant @xmath237 only depends on @xmath295 , @xmath296 and @xmath297 , see ( [ est.w ] ) and ( [ def.t0 ] ) , we obtain a new solution @xmath298 .
clearly , this procedure may be extended to an arbitrarily fixed @xmath73 .
once this is done , a boot - strap argument allows us to deduce @xmath299 , implying that @xmath300;w^{1,\infty}(\o))$ ] is a solution of in @xmath301 . * step 2 .
* non smooth initial data . let us consider a sequence @xmath302 such that , as @xmath303 , @xmath304 where tv denotes total variation with respect to the @xmath29 variable .
let us denote by @xmath305 to the corresponding solution of p@xmath306 .
first , notice that @xmath305 is uniformly bounded in @xmath307 with respect to @xmath308 as a consequence of estimate ( [ est.w ] ) and property ( [ reg1.id ] ) .
we then obtain directly from equation ( [ eq.orig ] ) that @xmath309 since @xmath310 is smooth , we may deduce an @xmath274 bound for @xmath311 as in step 1 , not necessarily uniform in @xmath308 , but which allows us to differentiate equation ( [ eq.orig ] ) with respect to @xmath261 .
after integration in @xmath312 , we obtain @xmath313 with @xmath314 , and @xmath315 since @xmath316 , we have @xmath317 uniformly bounded in @xmath307 and so @xmath318 and @xmath319 .
therefore , using ( [ reg2.id ] ) we deduce from ( [ dsv ] ) that @xmath320 bounds ( [ bound.1 ] ) and ( [ bound.2 ] ) allow to deduce , using the compactness result ( * ? ? ?
* cor . 4 , p. 85 ) , the existence of @xmath321;{{\cal x}})$ ] such that @xmath322 strongly in @xmath323 , for all @xmath324 , and a.e . in @xmath325 .
similarly to the smooth case , this convergence allows to pass to the limit @xmath326 in ( [ eq.orig ] ) ( with @xmath22 replaced by @xmath305 ) and identify the limit @xmath22 as a solution of . again , the property @xmath327 and a boot - strap argument leads to @xmath328 .
_ stability and uniqueness .
_ let @xmath329 and @xmath330;{{\cal x}})$ ] be the corresponding solutions to problems @xmath84 .
set @xmath331 and @xmath332 .
then @xmath22 satisfies @xmath333 for @xmath334 , with @xmath335 . multiplying this equation by @xmath22 , integrating in @xmath96 and using the lipschitz continuity of @xmath336 ( with constant @xmath337 ) and young s inequality , we deduce @xmath338 finally , using jensen s and gronwall s inequalities , we deduce @xmath339 for all @xmath340 , and the result follows . _ time invariance of the level sets . _
the proof of this property is similar to the proof of the stability property .
let @xmath341 and @xmath124;{{\cal x}})$ ] be the corresponding solution to problem @xmath342 .
assume @xmath87 , and set @xmath343 , @xmath344 .
then , from equation ( [ eq.orig ] ) we get @xmath345 then , the lipschitz continuity of @xmath336 and gronwall s lemma allow us to deduce the result .
@xmath346 * proof of corollary [ th.existencev ] . * to prove point 1 , notice that from ( [ dsv ] ) ( in dimension @xmath79 ) we deduce @xmath347 a property that also holds in the limit @xmath326 .
point 2 of the theorem follows from evaluating equation ( [ eq.orig ] ) in @xmath348 and @xmath349 , using that @xmath350 is decreasing for all @xmath111 , and gronwall s inequality .
point 3 is a consequence of the assumption on the symmetry of @xmath9 , under which the integral term in ( [ eq.orig ] ) vanishes when it is integrated in @xmath97 .
point 4 is easily deduced by successive derivation of ( [ dsv ] ) ( which also holds for @xmath351 , under regularity assumptions ) . point 5 is again deduced from ( [ dsv ] ) and the decreasing character of @xmath352 and @xmath158 . since , @xmath353 and , using point 2 , @xmath354 for all @xmath355 , we have that the integral term in ( [ eta ] ) is evaluated inside a closed interval . therefore ,
using the assumptions of point 5 , we get @xmath356 uniformly in @xmath357 .
finally , we obtain the result from ( [ dsv ] ) in the limit @xmath123 and @xmath326 .
@xmath346 * proof of theorem [ th.equiv ] .
* we split the proof in two steps .
* step 1 . *
first we treat the case in which @xmath14 has no flat regions , that is when @xmath53 for any @xmath54 . by the invariance of the level sets structure proven in theorem [ th.existenceu ] we deduce that neither the solution @xmath22 of has flat regions .
then @xmath358 and @xmath146 are strictly decreasing , implying @xmath359 for any @xmath54 . according to ( * ? ? ? * theorem 9.2.1 ) , we have @xmath360 where @xmath361 and we used the notation @xmath362 . integrating ( [ eq.orig ] ) in @xmath363
we get @xmath364 due to the @xmath22 and @xmath57 level sets equi - measure , it is immediate that @xmath365 the equi - measurability property ( [ prop.1 ] ) implies @xmath366 from where we deduce @xmath367 to deal with the term @xmath368 we observe that due to the invariance of the level set structure , as stated in theorem [ th.existenceu ] , we have that , for all @xmath65 $ ] and @xmath369 , there exists @xmath370 such that @xmath371 recalling that @xmath22 and @xmath14 have not flat regions and taking the measure of these sets we deduce @xmath372 .
therefore , @xmath373 finally , substituting in identity ( [ cor.2 ] ) the expressions ( [ cor.1 ] ) , ( [ cor.3 ] ) , ( [ cor.4 ] ) and ( [ cor.5 ] ) , and differentiating with respect to @xmath138 , we deduce the result .
conversely , let @xmath158 be a solution of p@xmath11 . since @xmath14 has not flat regions , @xmath374 in @xmath97 , and by point 1 of corollary [ th.existencev ] we have @xmath375 in @xmath193\times \o_*$ ] .
we define @xmath376 and for all @xmath65 $ ] .
observe that since @xmath14 has not flat regions , we have @xmath377 for all @xmath138 .
therefore , since @xmath378 , we also deduce that @xmath22 has not flat regions . by construction , @xmath379 implying @xmath380 . differentiating in ( [ def.u ] ) with respect to @xmath30 and using
that @xmath158 is a solution of p@xmath11 , we get , for @xmath381 , @xmath382 where we have used again the equi - measurability property ( [ prop.1 ] ) .
* step 2 . *
we now treat the general case in which @xmath383 may have flat regions .
we use the following lemma .
[ lemma.approx ] let @xmath98 .
then there exists a sequence @xmath384 such that @xmath385 has no flat regions and @xmath386 in @xmath387 .
we may then apply the step 1 of this proof to each @xmath385 to obtain that @xmath388;{{\cal x}})$ ] is a solution of @xmath389 ( without flat regions ) if and only if @xmath390 is a solution of @xmath391 .
now we perform the limit @xmath392 .
let @xmath124;{{\cal x}})$ ] and @xmath393 be the solutions of problems @xmath342 and @xmath394 ensured by theorem [ th.existenceu ] .
using the strong continuity of the decreasing rearrangement operation in @xmath395 , see ( [ prop.2 ] ) , and the stability property ( [ stability ] ) applied to problem @xmath342 , we obtain @xmath396 the same arguments in reverse order applied to problem @xmath394 leads to @xmath397 therefore , using the triangle inequality we deduce @xmath398 as @xmath392 .
@xmath346 * proof of lemma [ lemma.approx ] . * in this proof , we rename @xmath399 by @xmath22 and @xmath385 by @xmath400 .
let , for @xmath401 , @xmath402 with @xmath403 , be the collection of flat regions of @xmath22 which is , at most , countable . thus , @xmath404 . let @xmath405 and @xmath406 denote the characteristic function of the set @xmath407 and its perimeter , respectively .
we consider the functions @xmath408 where @xmath409 is a non - negative function without flat regions .
observe that since @xmath410 we have @xmath411 for all @xmath401 .
consider , for @xmath412 , the sequence of @xmath71 functions @xmath413 we have : \(1 ) _
@xmath414 has no flat regions in @xmath96_. let @xmath54 .
we use the decomposition @xmath415 if @xmath416 for some @xmath401 then @xmath417 , and , by definition , @xmath418 if @xmath419 which is not possible . therefore , if @xmath416 we have @xmath420 .
if @xmath421 for all @xmath401 then @xmath422 , so @xmath423 since @xmath158 has no flat regions .
\(2 ) _ @xmath424 in @xmath425 for any @xmath426_. this is immediate , since @xmath427 .
\(3 ) _ @xmath428 and @xmath424 in @xmath429 . _
according to ( * ? ? ?
* proposition 3.38 ) , for each @xmath401 , we can find a sequence @xmath430 with @xmath431 such that @xmath432 in @xmath433 as @xmath434 , and @xmath435 since @xmath410 .
we also introduce a regularizing sequence @xmath436 such that @xmath437 and @xmath438 in @xmath429 .
let @xmath439 .
then , @xmath440 implying that @xmath441 is uniformly bounded in @xmath429 with respect to @xmath15
. therefore , there exists @xmath442 and a a subsequence of @xmath441 ( not relabeled ) such that @xmath443 strongly in @xmath433 as @xmath434 .
since , by the dominated convergence theorem we have @xmath444 in @xmath433 as @xmath434 , we deduce @xmath445 .
taking the limit @xmath434 in ( [ est.bv ] ) we get @xmath446 with @xmath447 independent of @xmath448 and @xmath449 .
thus , using the definition of @xmath400 , the triangle inequality , and ( [ lemma.1 ] ) we get @xmath450 therefore , @xmath451 . finally , from the definition of @xmath400 and ( [ lemma.1 ] ) we have @xmath452 @xmath346 * proof of theorem [ th.pde ] . *
define @xmath453 since @xmath14 has not flat regions , we have @xmath454
. then , due to points 1 and 2 of corollary [ th.existencev ] we have @xmath455 in @xmath456 , and @xmath457 for all @xmath65 $ ] , respectively .
let us consider the inverse of @xmath146 , the distribution function of @xmath22 , @xmath458 . using the change of variable @xmath459 and writing @xmath460
, we obtain from ( [ eq.i ] ) @xmath461 using the explicit form of @xmath9 and integrating by parts , we obtain @xmath462 with @xmath137 given by ( [ def.ktilde ] ) . by assumption ,
function @xmath463 is bounded in @xmath464 $ ] and by point 4 of corollary [ th.existencev ] it is continuously differentiable in @xmath465 . consider the interval @xmath466 . by well known properties of the gaussian kernel , we have @xmath467 and @xmath468 in particular , from ( [ gauss.2 ] ) we get @xmath469 taylor s formula implies @xmath470 therefore , from ( [ th2.i1 ] ) , ( [ th2.4 ] ) and ( [ gauss.1 ] ) we deduce , using @xmath455 , @xmath471 then , the result follows from ( [ th2.3 ] ) substituting @xmath472 by @xmath473 . @xmath346
in this paper we studied a general class of nonlinear integro - differential operators with important imaging applications , such as the denoising - segmentation neighborhood filtering .
although the corresponding pde problem is multi - dimensional , we showed that it can be reformulated as a one - dimensional problem by means of the notion and properties of the decreasing rearrangement function .
we proved the well - posedness of the problem and some stability properties of the solution , as well as the equivalence between the multi - dimensional and the one - dimensional solutions to the problem .
some other interesting properties were deduced for the rearranged one - dimensional version of the problem , such as the time invariance of the level sets of the solution ( inherited by the multi - dimensional equivalent solution ) , and the asymptotic behavior of the solution as a shock - type filter .
future work will point to the use of rearranging techniques for the generalization of the model to include nonlocal effects induced by non - homogeneous spatial kernels , like in equation ( [ mazon ] ) . as already showed for the discrete time problem @xcite , this situation is much more involved suggesting the consideration of the _ relative rearrangement _ functional .
g. galiano , j. velasco , some nonlocal filters formulation using functional rearrangements , scale space and variational methods in computer vision , lecture notes in computer science 9087 , ( 2015 ) 166177 .
d. m. ushizima , a. g. bianchi , c. m. carneiro , segmentation of subcellular compartments combining superpixel representation with voronoi diagrams , http://cs.adelaide.edu.au/@xmath475carneiro/isbi14_challenge /results_release.html , 2014 .
l. zhi , g. carneiro , a. p. bradley , automated nucleus and cytoplasm segmentation of overlapping cervical cells http://cs.adelaide.edu.au/@xmath475carneiro/isbi14_challenge /results_release.html , 2014 . | we study the existence and uniqueness of solutions of a nonlinear integro - differential problem which we reformulate introducing the notion of the decreasing rearrangement of the solution
. a dimensional reduction of the problem is obtained and a detailed analysis of the properties of the solutions of the model is provided .
finally , a fast numerical method is devised and implemented to show the performance of the model when typical image processing tasks such as filtering and segmentation are performed . _
keywords : _ integro - differential equation , existence , uniqueness , neighborhood filters , decreasing rearrangement , denoising , segmentation . |
there abides a deep fascination with the subject of prime terms in naturally occurring integer sequences . the fanfare which accompanies the discovery of each new mersenne prime @xcite bears testimony , as does the search for factorizations of the fermat and fibonacci numbers @xcite .
elliptic curves @xcite give rise naturally to rapidly growing sequences of integers , called _
elliptic divisibility sequences _ , generated from some nontorsion point via the geometric addition law .
the first systematic study of elliptic divisibility sequences was given majestically by morgan ward in @xcite .
recent times have witnessed a great deal of activity ( the following represents a sample ) because of their intrinsic interest @xcite as well as in applications to logic @xcite and cryptography @xcite .
the chudnovsky brothers @xcite first proposed studying primality in elliptic divisibility sequences arising from elliptic curves in weierstrass form .
this was continued in @xcite with the following outcome : a heuristic argument , together with lots of computational evidence and proofs in some special cases , suggests that any elliptic divisibility sequence contains only finitely many prime terms .
this has become known as the _ primality conjecture _ for elliptic curves in weierstrass form . in ( * ? ? ?
* theorem 4.1 ) the same conjecture was proved for curves in homogeneous form .
it is with such curves that our business lies in this paper .
let @xmath1 denote an elliptic curve of the form @xmath2 where @xmath3 is a nonzero integer .
it is sometimes said that @xmath1 is a twist of the fermat cubic .
the set of rational ( or real ) points form a group under the chord and tangent method : the ( projective ) point at infinity with homogeneous coordinates @xmath4 is the identity , and inversion is given by reflection in the line @xmath5 . given a nontorsion point @xmath6 ,
write , in lowest terms , @xmath7 the sequence @xmath8 is a ( strong ) divisibility sequence , see proposition [ strongdivseq ] .
our starting point is the result stated earlier , that only finitely many terms @xmath9 are prime ( * ? ? ?
* theorem 4.1 ) . a related result ( * ? ? ? * theorem 1.2 ) states that @xmath9 has a primitive divisor for all @xmath10 : results from @xcite will be employed in the proofs of our theorems . to focus now upon the question we study : a stronger form of the primality conjecture has been proposed , assuming that the weierstrass equation is in minimal form , predicting that the number of prime terms in an elliptic divisibility sequence is uniformly bounded ( independently of both the curve and the point ) .
unsurprisingly , this conjecture is known as the _ uniform primality conjecture _ for curves in weierstrass form . a conditional proof is known , under lang s conjecture , assuming the generating point is the image of a rational point under a nontrivial isogeny ( * ? ? ?
* theorem 1.4 ) .
this raises the natural question as to whether a uniform primality result holds for curves in homogeneous form and it is this question we address .
the first of our two results is an unconditional result to that effect , and will be stated now .
the proof will appear to be quite short although , in fact , it relies upon two substantial results from @xcite and @xcite .
[ upc]assuming @xmath11 is cube - free , the number of prime ( power ) terms @xmath9 is uniformly bounded .
moreover , assuming the @xmath12 conjecture holds in @xmath13 , there is a uniform upper bound on the index @xmath14 such that @xmath9 is a prime power .
the assumption about @xmath3 being cube - free is necessary given the homogeneous nature of the curve @xmath1 , since one could scale any given elliptic divisibility sequence to obtain as many prime terms as desired
. this will be explained in detail in subsection [ cubefree ] .
it is also probable that a uniform bound on the index @xmath14 would follow from the hall - lang conjecture , as in recent work of mah @xcite .
our second theorem relies upon the existence of a rational isogeny in line with work for curves in weierstrass form @xcite .
there is a @xmath0-isogeny , detailed in @xcite , @xmath15 from the elliptic curve @xmath16 [ expupc ] suppose @xmath11 is cube - free and @xmath17 is the image of a point in @xmath18 under @xmath19 .
then @xmath9 is not a prime power , for all @xmath20 .
further , the number of terms @xmath9 with at most two distinct prime divisors is uniformly bounded .
the conclusion is clearly best possible and what is remarkable is the small size of the bound @xmath21 ; this is certainly much stronger than any corresponding result for elliptic curves in weierstrass form .
the proofs of the two theorems rely upon the same principle , which will be articulated in section [ proveit ] below .
this is preceded by a section discussing possible improvements to our results and is followed by proofs of the theorems in order .
we thank tony flatters and valry mah for their excellent comments and suggestions on a preliminary version of this paper . in particular , the latter pointed out that a ( conditional ) bound on the index could be obtained in theorem [ upc ] .
taking our two theorems in reverse over , theorem [ expupc ] can not be improved . but to how large a class of examples does it apply ?
if @xmath22 is @xmath0 times another rational point then it will be such an image @xcite .
a short analysis is included to shed light in greater generality . in truth , throughout the paper , we will not work explicitly with the isogeny mentioned in the statement of theorem [ expupc ] .
instead , our work relies upon a bi - rational map between @xmath1 and the mordell curve @xmath23 where @xmath24 corresponds to @xmath25 under the bi - rational transformation given by @xmath26 u = \dfrac{2^{2}3^{2}m + y}{6x } , & v = \dfrac{2^{2}3^{2}m - y}{6x}. \end{array}\ ] ] there is a 3-isogeny @xmath27 onto the curve @xmath28 from the elliptic curve @xmath29 given by @xmath30 it is with this isogeny that we work .
the isogeny condition stated in theorem [ expupc ] is equivalent to the point @xmath31 being equal to @xmath32 for some point @xmath33 .
table [ m<100 ] shows the @xmath34 values @xmath35 when this occurs for a point @xmath36 which is a generator for the group @xmath37 .
( thus all points in @xmath37 are in the image of the isogeny . ) in total , there are @xmath38 rank one curves with @xmath39 . we conclude that , at least for small values of @xmath3 , the hypothesis of theorem [ expupc ] is not infrequently met .
.rank 1 curves satisfying theorem [ expupc ] with @xmath39 [ cols="^,^,^",options="header " , ] turning now to theorem [ upc ] , we will draw upon a result of ingram @xcite bounding the number of integral multiples of a given integral point . in its most general form ,
this involves a non - explicit constant . in his paper
, he gives a very strong ( actually best possible ) explicit constant in the case of congruent number curves .
it seems a good bet that his analysis could be re - worked in our context to obtain an explicit ( perhaps even best possible ) form of theorem [ upc ] .
this analysis was done in certain cases by the second author in her thesis @xcite : a special case of ( * ? ? ?
* theorem 1.0.4 ) is recorded here .
assume @xmath40 mod @xmath41 is cube - free .
assuming @xmath31 has @xmath42 and @xmath43 and @xmath44 are not integral then at most one value of @xmath10 yields a prime power term @xmath9 .
of course the most desirable goal is to be to be able to compute all of the prime power terms in a given sequence , as permitted by theorem [ expupc ] .
it is not clear whether theorem [ upc ] could be improved to allow that as an outcome .
however , as part of the proof of theorem [ expupc ] , we do prove the following unconditional result : [ lem:2<n<23 ] suppose @xmath11 is cube - free .
then @xmath9 is not a prime power , for @xmath45 .
there will always be daylight between out theorems as our final remark demonstrates .
it seems likely that without the isogeny hypothesis , there are infinitely many values of @xmath3 and generating points @xmath22 yielding a prime term @xmath46 . for ,
suppose @xmath47 \in c(\mathbb{q})$ ] is an integral point of infinite order .
the point @xmath48 can be expressed in the form @xmath49 suppose @xmath50 . then @xmath51 . applying the bateman - horn conjecture @xcite to the polynomial
@xmath52 implies that @xmath53 is prime for infinitely many positive integers @xmath54 . by a result of erds @xcite
, the corresponding value of @xmath55 will be cube - free for infinitely many of these values . to show that the prime so constructed does not cancel with the numerator , compute the resultants of the two polynomials @xmath56 and @xmath57 in each variable :
they are @xmath58 and @xmath59 so the only non - trivial common division is by a divisor of @xmath41 .
but @xmath60 is coprime to @xmath0 so there can be no cancellation .
this style of argument will be used repeatedly in the proof of theorem [ expupc ] .
this section is elementary although quite technical because of the case - by - case nature of the proofs .
it is the statements of the results which provide important input for sections [ proofofupc ] and [ proof1.2 ] , rather than their proofs .
readers whose taste is not excited by the details could take note of lemma [ cancel ] , proposition [ strongdivseq ] and subsection [ cubefree ] then skip directly to sections [ proofofupc ] and [ proof1.2 ] . from the bi - rational transformation
, we have @xmath61 and @xmath62 where @xmath63 are written in lowest terms .
[ cancel]any cancellation in the fractions and comes from the term @xmath64 and divides @xmath65 .
since @xmath66 and @xmath67 are coprime , the following important principle arises as a consequence of lemma [ cancel ] : 0.1 in * we can be certain that @xmath9 possesses at least two coprime factors if two conditions are met .
the first is that @xmath68 and the second is that the term @xmath64 is not completely cancelled . * 0.1 in consider the fraction on the right - hand side of .
let @xmath69 be a common factor of @xmath70 and @xmath71 with @xmath72 a prime number and @xmath73 the highest order of @xmath72 dividing both terms .
if @xmath74 , then @xmath75 implies @xmath76 , which contradicts the fact that @xmath67 and @xmath77 are coprime .
thus @xmath78 , so that @xmath79 comes from the term @xmath64 . notice , moreover , that any cancellation in the right - hand side of and is the same in each term .
this is because both terms have the same denominators as their cubes sum to an integer .
hence @xmath79 has to divide both @xmath70 and @xmath80 , so that @xmath81 .
thus any cancellation of the fractions and divides @xmath65 .
we will need to be even be more precise about the cancellation that occurs . write @xmath82 for the cancellation in the fraction .
[ cor : cancel ] we have @xmath83 note in particular that if @xmath84 and @xmath85 then @xmath72 is completely cancelled .
thus for primes @xmath84 , we have @xmath86 if and only if @xmath87 . for primes greater than @xmath0 ,
the cancellation is exactly as large as allowed by lemma [ cancel ] ; therefore , whenever @xmath88 divides @xmath66 , we have @xmath89 suppose first that @xmath87 .
by lemma [ cancel ] , we have @xmath90 except possibly
if @xmath84 so suppose we are in this situation .
the defining equation yields @xmath91 whence @xmath92 so @xmath72 can not cancel in .
now suppose @xmath85 and @xmath93 .
from we find @xmath94 since @xmath85 , and it follows that @xmath95 .
finally , suppose @xmath85 and @xmath84 , and put @xmath96 .
the arguments are fiddly but elementary so we only sketch them , beginning with the slightly simpler case @xmath97 . if @xmath98 then , comparing @xmath0-adic valuations , we see that equation is impossible to solve .
hence @xmath99 and we also find that @xmath100 .
thus the numerator of has @xmath0-adic order at least @xmath101 whereas the @xmath0-adic order of the denominator is exactly @xmath101 , giving @xmath102 .
now assume that @xmath103 . if @xmath104 then , dividing through by @xmath105 yields a congruence @xmath106 , which is impossible .
moreover , @xmath107 is impossible by comparing @xmath108-adic valuations in . if @xmath109 then @xmath110
so the numerator of is @xmath111 times the sum of two odd numbers .
thus the @xmath108-adic order of the numerator is at least @xmath112 while that of the denominator is exactly @xmath113 so @xmath114 .
this leaves the case @xmath115 with @xmath116 . here
@xmath117 so the @xmath108-adic order of the numerator of is at least @xmath0 , while that of the denominator is exactly @xmath0 , again giving @xmath114 .
the aim of this subsection is a proof of the following : [ strongdivseq ] the sequence @xmath8 is a strong divisibility sequence : in other words , for all @xmath119 , @xmath120 in particular , @xmath121 , for all @xmath122 .
the divisibility property will be used repeatedly in this paper but we can not find it proved explicitly in the literature .
the strong divisibility property will be used only in subsection [ cubefree ] .
however , it is a natural property and is proved with little more effort .
we admit that the proof we give lacks elegance .
this is due to the evil influence of the primes @xmath108 and @xmath0 and , to a lesser extent , those dividing @xmath3 .
however , the proof is completely self - contained and uses only elementary methods .
a more sophisticated proof of proposition [ strongdivseq ] uses formal groups ( * ? ? ?
* chapter iv ) .
writing @xmath123 and @xmath124 in ( [ defc ] ) yields the equation @xmath125 now the binomial theorem yields a power series @xmath126 which converges @xmath72-adically for all primes @xmath72 with @xmath127 ( only the case @xmath97 is at all tricky ) . applying the geometric group law on the points @xmath128 and @xmath129
now yields a power series @xmath130 the _ formal group _ of our elliptic curve .
the proof of proposition [ strongdivseq ] follows from the standard properties of formal groups ( * ? ? ?
* chapter iv , vii ) : in particular , the filtration into subgroups @xmath131 , @xmath132 , defined by @xmath133 . however arguing as above requires quite a lot of explanation and checking and it would not shorten the paper .
for example , the statement of proposition [ strongdivseq ] per se does not appear in @xcite .
instead , the filtration referred to yields directly the relation ( [ w - ordthing ] ) below .
the divisibility property follows from the following relation : for any prime @xmath72 and @xmath134 , if @xmath86 then @xmath135 much of the spade work here is supplied by the following two relations .
firstly , from ( * ? ? ?
* lemma 3.2 ) , for @xmath88 , if @xmath136 and @xmath137 then @xmath138 secondly , for all primes @xmath72 , if @xmath139 and @xmath140 then @xmath141 the second relation arises from a local analysis using the formal group for the weierstrass equation and follows from ( * ? ? ?
* chapter vii ) . a much more general proof , valid over a dedekind domain , is given in ( * ? ? ?
* proposition 1 ) .
note that implies that @xmath142 is a strong divisibility sequence .
further work is needed to take us from ( [ w - ordthing ] ) to the same conclusion for the sequence @xmath143 : this will be supplied at the end .
assume @xmath72 denotes a prime with @xmath86 .
we will show for all @xmath144 .
there are three cases in total : * @xmath145 , * @xmath146 , * @xmath147 .
* case ( a ) * no cancellation occurs in so @xmath148 if @xmath149 or @xmath150 if @xmath151
. then follows from , the corresponding result for the sequence @xmath142 .
* case ( b ) * since @xmath86 , the two conditions given imply @xmath85 .
when @xmath137 , we have @xmath152 by so we can apply with @xmath153 replacing @xmath14 .
then @xmath154 when @xmath155 , we use the triplication law ( * ? ? ? * ( 8) ) , repeated here for convenience : @xmath156 it follows from and that @xmath157 since @xmath158 we deduce from that @xmath159 . therefore @xmath160 * case ( c ) * since @xmath86 and @xmath84
, we have @xmath87 and , as there is no cancellation , @xmath161 . also , since @xmath87 it follows that @xmath162 for all @xmath163 : this is because the point @xmath164 has good reduction at @xmath72 so @xmath165 has good reduction mod @xmath72 as the set of points with good reduction form a group and it follows that @xmath165 can not reduce mod @xmath72 to a point with zero @xmath166-coordinate .
thus there is no cancellation in when @xmath14 is replaced by @xmath153 .
moreover , as @xmath167 also , the reduction of @xmath164 is a non - identity point and , since @xmath28 has additive reduction modulo @xmath72 , the point @xmath165 reduces to the identity if and only if @xmath168 . ( note that we are reducing the curve @xmath28 , not its minimal form . ) there are now two possibilities : if @xmath169 then @xmath170 , in which case @xmath171 . otherwise @xmath168 and @xmath159 ; it is easily checked by an explicit calculation with the doubling and tripling formulae ( see for the latter ) that @xmath172 so we get @xmath173 now we have proved , the strong divisibility property comes from the following claim : * claim * if a prime @xmath72 divides @xmath174 then @xmath175 . to see that this suffices , let @xmath176 and suppose @xmath177 .
for any prime @xmath72 , one of @xmath178 and @xmath179 is @xmath180 because @xmath181 and @xmath182 are coprime . from , the exact power of @xmath72 dividing @xmath183 is @xmath184 as desired .
it remains to prove the claim .
first let @xmath72 be a prime not dividing @xmath185 .
note that @xmath86 if and only if @xmath186 in this case ( since no cancellation can occur ) , which is equivalent to @xmath164 reducing mod @xmath72 to an element of the subgroup @xmath187 of @xmath188 generated by a @xmath0-torsion point @xmath189 with @xmath190 .
if @xmath191 and @xmath164 both reduce to an element of @xmath187 then so does any integer linear combination ; in particular , so does @xmath192 and we deduce that @xmath175 .
now suppose @xmath88 is a prime dividing @xmath3 such that @xmath193 .
there are three possibilities , and in each case we will show that @xmath175 : 1 .
@xmath194 and @xmath145 , in which case @xmath195 , since @xmath196 is a strong divisibility sequence .
thus @xmath175 , since there is no cancellation .
2 . @xmath194 and @xmath85 , in which case @xmath164 is a point of bad reduction , and @xmath36 is also .
thus @xmath197 and , from the analysis in case ( b ) above , we see that @xmath198 , @xmath199
. then , from , we have @xmath200 so , from strong divisibility , @xmath201 .
moreover , since @xmath202 , we have @xmath203 so , from again , @xmath204 3 .
@xmath205 and @xmath85 , in which case @xmath36 is again a point of bad reduction , @xmath206 , @xmath199 , so @xmath203 . from , we have @xmath207 so that @xmath208 .
finally , suppose @xmath84 .
since @xmath86 we have @xmath87 so @xmath164 is a point of good reduction modulo @xmath72 .
the same is true of @xmath191 and so also of @xmath192 .
thus @xmath209 and we deduce from corollary [ cor : cancel ] that @xmath175 .
this completes the proof of the claim , and of proposition [ strongdivseq ] .
we conclude with a remark justifying the assumption in both our theorems that @xmath3 is cube - free .
the key remark needed follows from proposition [ strongdivseq ] : if @xmath211 then @xmath212 . from (
* theorem 1.2 ) , the terms @xmath9 with @xmath10 all see a primitive prime divisor : in other words , a prime @xmath86 which has not appeared in an earlier term . let @xmath213 denote a finite set of primes . for each @xmath214 , write @xmath215 with @xmath216 and @xmath217 equal to a primitive prime divisor . now write @xmath218 and rescale equation by multiplication with @xmath219 .
if @xmath220 denotes the resulting denominator then @xmath221 is a prime power by the starting remark .
therefore , by expanding @xmath213 arbitrarily , we see that , at the cost only of multiplying @xmath3 by a cube , we may produce elliptic divisibility sequences with arbitrarily large numbers of prime power terms .
we will prove firstly that , for @xmath222 , the term @xmath66 in the denominator is not completely cancelled . from (
* theorem 2.3 ) , all terms @xmath66 with @xmath222 possess a primitive prime divisor @xmath223 . then @xmath224 , because @xmath223 is a primitive prime divisor , so it follows that @xmath36 is a point of good reduction for @xmath223 .
thus all multiples of @xmath36 , in particular @xmath164 , are points of good reduction . however ,
for primes dividing @xmath185 , the point on the reduced curve with @xmath166-coordinate zero is a point of bad reduction so it follows that @xmath225 .
in particular , @xmath223 is not cancelled because any cancellation divides @xmath65 .
secondly , from ( * ? ? ?
* theorem 1 ) , with the notation used there , there is a uniform constant @xmath1 such that @xmath226 forces @xmath227 , except for at most one value of @xmath14 .
the quantity @xmath228 is related to the tamagawa number . since the mordell curve @xmath28 has integral @xmath229-invariant , along the same lines as in @xcite , @xmath28 always has @xmath230 .
it follows that the number of prime power terms @xmath9 is bounded by @xmath231 , a uniform constant .
the idea of the proof of theorem [ expupc ] is to use heights on the weierstrass elliptic curve to show that , for @xmath232 , we are guaranteed to get a prime divisor of @xmath9 coming from the term @xmath67 ; that is , we use the isogeny to make explicit the constants in the argument above using the @xmath12 conjecture .
as we have seen , we also get a prime divisor coming from @xmath66 in this case , so that @xmath9 is not a prime power .
the remaining values @xmath233 are treated case - by - case , using the explicit form of the point @xmath234 as a rational function in the coordinates @xmath235 of @xmath17 . [
> m ] let @xmath28 be defined as in and suppose @xmath31 is the image of a point in @xmath236 under the isogeny @xmath237 as in . then , with @xmath67 as in , @xmath238 for all @xmath239 .
let @xmath240 such that @xmath241 , for some @xmath33 .
write @xmath242 with @xmath243 . from
, @xmath244 * claim : * @xmath227 , provided @xmath245 .
before this claim can be settled , we must examine the fraction on the right - hand side of for possible cancellation .
let @xmath69 be a common factor of @xmath246 and @xmath247 , where @xmath72 is a prime and @xmath73 is the highest order of @xmath72 dividing both terms .
since @xmath243 , either @xmath248 or @xmath249 .
if the latter occurs , then @xmath250 implies @xmath251 , which is impossible as @xmath252 and @xmath253 are coprime .
thus @xmath79 can only come from the term @xmath254 , so that @xmath255 .
we have now that @xmath256 so @xmath257 hence the greatest common divisor of numerator and denominator of the fraction on the right - hand side of , say @xmath258 , has to divide @xmath259 as well . to turn to the claim , if @xmath260 , then @xmath261 on the other hand , if @xmath262 , then @xmath263 implies @xmath264 thus we need to ensure that @xmath265 for our purposes and to that end we turn .
note that the logarithm of the expression on the left is the weil height @xmath266 of @xmath267 which we know , from theorem [ weilcan ] , is close to the canonical height @xmath268 .
write @xmath269 and @xmath270 ; then @xmath271 as @xmath237 is a 3-isogeny .
applying the inequality to the curve @xmath272 with @xmath273 , @xmath274 and @xmath275 , we obtain @xmath276 the height bound ( * ? ? ? * ( 14 ) ) gives @xmath277 for all @xmath278 . then becomes @xmath279
then we can ensure @xmath280 provided that @xmath281 with a manipulation , will be guaranteed for @xmath282 but only for all sufficiently large @xmath3 .
however the amount of checking for smaller values of @xmath3 is infeasible .
more realistically , if @xmath283 then is true provided @xmath239 .
thus it can be concluded that for all @xmath284 , we have @xmath227 if @xmath239 .
the proof for values @xmath285 follows in an appendix , see section [ dontforget ] .
we are now in a position to prove theorem [ expupc ] using lemma [ > m ] . when @xmath222 , the term @xmath66 is not completely cancelled from the denominator of , exactly as before .
also , it follows directly from lemma [ > m ] that @xmath68 for all @xmath232 .
therefore , the term @xmath9 possesses at least two coprime factors for all @xmath232 .
we will go on to prove the same for every @xmath286 case by case . before this
, we give a proof of the second claim in theorem [ expupc ] . under the isogeny hypothesis
, it follows from ( * ? ? ?
* theorem 1.4 ) that the number of prime power terms @xmath67 is uniformly bounded . combining this with our knowledge that @xmath66 has a primitive prime divisor for @xmath222 means that the number of terms @xmath9 which have at most two distinct prime divisors is uniformly bounded . with sleeves rolled up
, we will now show that every term @xmath9 fails to be a prime power when @xmath287 .
we will not invoke the isogeny hypothesis so these results apply in complete generality ; that is , we are giving a proof of lemma [ lem:2<n<23 ] .
finally we will deal with the case when @xmath288 assuming the isogeny hypothesis .
since @xmath8 is a divisibility sequence , it suffices to consider @xmath9 when @xmath14 is an odd prime less than @xmath34 or @xmath289 .
we begin with the case when @xmath290 is an integral point ; at the end of the proof we will explain how the general case follows . * case @xmath291 * we deal firstly with the case when @xmath292 . write @xmath293 where @xmath294 and @xmath295
we may consider the second coordinate , and factorize @xmath296 as a product of four terms : @xmath297 we claim that at least two of these factors can avoid being cancelled by the numerator @xmath298 . choosing to consider @xmath299 and @xmath300
, we can see that the resultants between them and @xmath301 with respect to @xmath54 and @xmath302 are @xmath303 respectively , and also @xmath304 as @xmath54 and @xmath302 are coprime , @xmath305 next we will show that both @xmath306 and @xmath307 are not equal , up to a sign , to any power of @xmath0 .
suppose , for a contradiction , that @xmath308 , for some @xmath309 .
then @xmath310 hence @xmath311 , so @xmath312 . replacing this in the expression of @xmath306 , we get
@xmath313 then @xmath314 or @xmath315 .
since @xmath316 , the former implies @xmath317 , and the latter implies @xmath318 and @xmath319 which is a contradiction as @xmath320 .
thus the possibilities for @xmath181 such that @xmath308 are only @xmath180 and @xmath21 . calculating with pari - gp @xcite , without assuming the grh , shows that the only solutions @xmath235 of the equation @xmath321 are @xmath322 , [ \pm 1 , 0]$ ] and there are no solutions to @xmath323 .
a similar argument will be applied for the second factor @xmath324 suppose @xmath325 for some @xmath326 . as @xmath327 we have @xmath311 , so that @xmath328 replacing each of these in the expression of @xmath300 , we find that there are no solutions to @xmath325 when @xmath326 .
thus it remains to solve the equations @xmath325 when @xmath329 .
the only solutions to @xmath330 , [ 0 , \pm 1 ] , [ -1,1 ] , [ 1 , -1].\ ] ] however these correspond to values of @xmath3 ( @xmath331 or @xmath180 ) which yield rank zero ( or singular ) curves
. finally , gp says there are no solutions at all to @xmath332 when @xmath333 .
we will prove moreover that the multiple @xmath334 can not be a prime power .
as above , @xmath299 and @xmath300 are not powers of @xmath0 , so write @xmath335 where the @xmath336 and @xmath337 are primes other than @xmath0 .
considering the resultant between @xmath299 and @xmath300 , we get @xmath338 .
thus there is at least one prime @xmath336 which is not equal to any prime @xmath337 .
this implies @xmath339 is not a prime power .
* case @xmath340 * the expression for @xmath341 can be written as @xmath342 for convenience , let @xmath343 by the theory of resultants , we obtain @xmath344 since at least one of @xmath345 is not a unit , to complete the proof in this case we have to prove that the denominator @xmath346 is not ( up to sign ) a power of @xmath0 .
suppose @xmath347 for some @xmath309
. then @xmath348 . thus @xmath349 , and hence @xmath350 so @xmath351
this implies @xmath319 which is impossible .
for the remaining values , when @xmath352 the solutions are @xmath353 $ ] and @xmath354 $ ] and when @xmath355 , the only solutions are given by @xmath356 , [ 0 , \pm 1 ] , [ 1 , -1 ] , [ -1 , 1].\ ] ] as before , these all correspond to impossible values of @xmath3 ( @xmath357 ) . since @xmath320 and @xmath54 and @xmath302 are coprime to both @xmath358 and @xmath346 but are not both units ( as @xmath359 ) , @xmath360 possesses at least two coprime divisors .
* case @xmath361 * the proof in this case proceeds exactly in the same way as in the case @xmath292 , by the following steps .
= 0.8cm=0.3cm=-0.5 cm write @xmath362 and factor the denominator @xmath363 as @xmath364 , all of which are homogeneous in @xmath54 and @xmath302 . by the theory of resultants ,
we have found fortunately that for each @xmath14 , @xmath365 divides a power of @xmath0 for every @xmath366 .
pick two factors of @xmath367 , say @xmath368 and @xmath369 .
it can be proved that both of them can not be a power of @xmath0 ( up to sign ) by using the following facts : @xmath370 show that the multiple @xmath371 is not a prime power , for which it is sufficient to prove that the resultant of @xmath372 and @xmath373 is a power of @xmath0 .
* case @xmath374 * in this case , the situation is more complicated . for all @xmath14 , @xmath375 and @xmath363 also behave as in the previous case in steps ( i ) and ( iii )
however , matters are slightly different in step ( ii ) .
we need to employ some facts about newton polygons over @xmath0-adic fields to know about the @xmath0-adic valuation of @xmath372 .
we will show how to do this for @xmath376 ( for other @xmath14 , the proofs will proceed in the same way ) .
we have @xmath377 and @xmath378 & & \quad+21u^{11}v^5 -11u^{10}v^6-u^9v^7 + 27u^8v^8- u^7v^9 - 11u^6v^{10}\\[1pt ] & & \quad\ \quad\ + 21u^5v^{11 } + 11u^4v^{12}-2u^3v^{13}+6u^2v^{14}+2uv^{15}+v^{16}.\end{aligned}\ ] ] we will explore their roots of the polynomials @xmath379 , where we find @xmath380 and @xmath381 & & + 20394x^{10 } + 35496x^9 + 49617x^8 + 55971x^7 + 50814x^6\\[1pt ] & & \
\,+ 36774x^5 + 20871x^4 + 9072x^3 + 2916x^2 + 648x + 81.\end{aligned}\ ] ] the newton polygons for @xmath382 and @xmath383 with @xmath384 , as shown in figures [ g_51 ] and [ g_52 ] , reveal that the slope of the only segment of each polygon is @xmath385 . ] by the standard facts about newton polygons , all roots of @xmath386 ( and all roots of @xmath387 and @xmath388 ) have @xmath0-adic absolute values @xmath389 .
if @xmath390 is a root of @xmath386 then , since @xmath391 , @xmath392 thus , since @xmath393 , @xmath394 where @xmath390 ranges over all roots of @xmath386 ; that is , the @xmath0-adic valuation of @xmath395 is at most @xmath396 .
it remains only to solve the equations @xmath397 , with @xmath398 .
we find that the only solutions to @xmath399 are @xmath322 , [ \pm 1 , 0]$ ] and there are no solutions to @xmath400 for other @xmath181 .
similarly , the only solutions for @xmath401 arise for @xmath402 , with @xmath403 , [ 0 , \pm 1 ] , [ -1 , 1 ] , [ 1 , -1]$ ] , and for @xmath404 , with @xmath353 $ ] and @xmath354 $ ] .
as before , these correspond to inadmissible values of @xmath3 .
] this concludes the proof of lemma [ lem:2<n<23 ] when we consider integral points . in the case of rational points , write @xmath405 in lowest terms .
the condition that @xmath3 is cube - free implies @xmath406 and @xmath407 are coprime .
replacing @xmath54 and @xmath302 in the expressions of @xmath234 in previous cases by @xmath408 and @xmath409 respectively , we obtain @xmath410 now proceed with the proof working with @xmath411 and @xmath412 : the conclusion follows as before .
thus we have proved lemma [ lem:2<n<23 ] . * case @xmath413 * to complete the proof of theorem [ expupc ] it remains only to treat the case @xmath288 ; the salient details follow . here
we begin with the curve @xmath414 in @xmath415 . in a change of notation , we write @xmath416 for a point which maps to @xmath22 under the isogeny @xmath19 .
for now we assume it is integral , noting that this implies @xmath417 since @xmath3 is cube - free .
we write @xmath418 where @xmath419 write @xmath420 for the product of the linear factors and @xmath421 for the degree @xmath422 factor .
firstly , we claim that the factor @xmath420 does not completely cancel with the numerator @xmath423 .
this is easy to check : compute the resultant of each of the linear factors in @xmath420 with @xmath423 .
every time you obtain a power of @xmath0 .
if @xmath420 cancels then each of the linear factors is a power of @xmath0 ( possibly @xmath424 ) .
this can not happen since @xmath425 and the equation @xmath426 has no solutions .
similarly , the factor @xmath421 is not cancelled by @xmath423 . checking resultants shows any common division is a power of @xmath0 . now @xmath427 forces @xmath428
so we only need to consider @xmath429 the only solutions occur when @xmath430 and @xmath431 . in the first case
, they are @xmath432 , [ -1 , 0 ] , [ 1 , -1 ] , [ 0 , -1 ] , [ -1 , 1 ] , [ 0 , 1],\ ] ] each giving the inadmissible value @xmath433 . in the second case
, they are @xmath434 , [ 2 , -1 ] , [ -1 , -1 ] , [ -2 , 1 ] , [ 1 , 1 ] , [ -1 , 2],\ ] ] giving the inadmissible values @xmath435 .
finally , the only common divisor of @xmath436 and @xmath421 is a power of @xmath0 so we must obtain one non - trivial factor of @xmath46 from the first term and a coprime factor from the second , which proves that @xmath46 is not a prime power . when @xmath235 is _ not _ integral , things are slightly different from the situation for @xmath287 above , because @xmath54 and @xmath302 might have different denominators . write @xmath437 and @xmath438 with @xmath439
. then also @xmath440 ( as @xmath3 is cube - free ) .
writing @xmath441 and @xmath442 , which are coprime , we obtain @xmath443 so we can proceed as before .
in order to complete the proof of lemma [ > m ] , it remains to check the statement @xmath444 for all cube - free integers @xmath3 up to @xmath445 , as mentioned at the end of the proof . in this part
, we deal with the particular computations to find a uniform bound , @xmath446 , on the indices @xmath14 for such @xmath3 . ranks and generators of @xmath447
were computed using magma @xcite .
note that when @xmath448 , we were unable to find the generator and rank using magma .
this was found instead using sage @xcite . for rank-1 curves , we tested the elliptic divisibility sequence @xmath449 arising from the generator for @xmath450 .
a special argument is required for the curves of rank 2 , with two parts needed to find the bound @xmath446 .
firstly find the finite set of pairs @xmath451 , @xmath452 , such that the canonical height of each point @xmath453 is less than @xmath454 , where @xmath36 and @xmath455 represent the generators . then compute the elliptic divisibility sequence @xmath449 arising from each point @xmath453 , for @xmath450 .
now we get a bound , say @xmath456 , for the indices @xmath14 from the points of canonical height less than @xmath454 . to treat all cases , when @xmath457 , we return to the proof of lemma [ > m ] again , replacing the estimate ( [ heightbound ] ) by | on the twisted fermat cubic , an elliptic divisibility sequence arises as the sequence of denominators of the multiples of a single rational point . we prove that the number of prime terms in the sequence is uniformly bounded . when the rational point is the image of another rational point under a certain @xmath0-isogeny , all terms beyond the first fail to be primes . |
@xcite , @xcite , and @xcite introduced , developed , and applied a new approach to quantitatively modeling the variability of individual quasars , describing the light curves as damped random walks ( drw ) characterized by two parameters , a timescale and an amplitude .
these parameters are then found to be correlated with the physical properties of the quasar
wavelength , luminosity , and black hole mass .
unfortunately , accurately measuring the parameters of individual quasars requires long ( @xmath4decade ) well - sampled light curves that are generally not available ( see @xcite ) . while this may ultimately be solved by projects such as the panoramic survey telescope and rapid response system ( panstarrs ; @xcite ) and the large synoptic survey telescope ( lsst ; @xcite ) , their cadences are not ideal and it still requires a decade of operations to accumulate the light curves .
this makes it challenging to expand on this promising new approach to probing quasar physics .
there are , however , quasars with almost two decades of well - sampled light curves
the quasars lying in the regions surveyed by microlensing projects such as the optical gravitational lensing experiment ( ogle ; e.g. , @xcite ) , macho ( e.g. , @xcite ) , and the microlensing observations in astrophysics ( moa ; e.g. , @xcite ) . for example
, the combined phases of the ogle survey ( ogle - ii to ogle - iv ) span 15 years with up to 1000 epochs in @xmath5-band and a few dozen in @xmath6-band for approximately 40 million sources with @xmath7 mag in the magellanic clouds , and the light curves continue to be extended by the ogle - iv survey .
the challenge is identifying the rare quasars in these dense stellar fields .
a combination of serendipity ( e.g. , @xcite ) , x - ray ( @xcite ) , and crude variability measures ( ) had succeeded in identifying only 55 and 28 quasars in the lmc and smc , respectively , as of 2010 . in @xcite
, we showed that the mid - ir selection techniques developed for extragalactic surveys ( @xcite ) could also be applied to dense stellar regions , albeit with some contamination from the relatively rare dusty stars ( e.g. , young stellar objects ( ysos ) , planetary nebulae ( pne ) , etc . ) . in @xcite
, we showed that the drw variability parameters of quasars differed from those of variable stars , providing a more quantitative approach to variability - selecting quasars that has been expanded on in @xcite and @xcite . in a single test observation under poor conditions
, @xcite doubled the number of known quasars behind the smc from 28 to 57 .
apart from studying variability , quasars in these fields are important tools for deriving more precise proper motions of the lmc , smc and , potentially , the galactic bulge using the _ hubble space telescope ( hst ) _ or , in the future , the _ james webb space telescope ( jwst)_. @xcite and @xcite based their lmc proper motions on 21 quasars .
further improvements require better measuring the internal motions of the clouds , and this requires larger numbers of quasars well - distributed over the lmc disk . based on @xcite ,
the newly confirmed lmc quasars we discuss here could be used to reduce the proper motion uncertainties by a factor of three , since our new quasars are distributed over a large @xmath8 deg@xmath1 area of the lmc disk .
we will also identify any bright quasars suitable for absorption studies of the lmc s interstellar medium ( ism ) , as we expect 1 , 10 , and 74 quasars brighter than @xmath9 , 18 , and 19 mag in the four observed fields .
the goal of our magellanic quasars survey ( mqs ) is to identify and characterize all these quasars , where we estimate a final sample of roughly @xmath10 quasars in a complete survey of the ogle - iii project regions .
spectroscopy is essential not only for confirmation of the candidates but also to determine the luminosity , rest - wavelength and black hole masses ( through the emission line widths ) of the quasars , so that the variability parameters can be related to other physical properties .
relatively large numbers of quasars will be needed to explore this ( minimally ) three dimensional parameter space quantitatively .
while the sample will be smaller than the @xmath11 quasars we considered in @xcite , the longer , more densely sampled light curves should let us examine the physical correlations with less `` blurring '' from the noisy parameter estimates obtained from the sparsely sampled sloan digital sky survey ( sdss ) light curves , making it far easier to determine the intrinsic parameter distributions .
in this paper , we report our preliminary results for the lmc , sadly reduced in scope by hurricane yasi . in section
[ sec : agnsel ] , we describe the selection methods leading to a sample of 2434 quasar candidates behind lmc . section [ sec : data ] describes the observation plan , the observations , and the data analysis .
section [ sec : newquasars ] introduces the new quasars and section [ sec : contamination ] discusses the contaminating sources .
we discuss the results in section [ sec : discussion ] and summarize the paper in section [ sec : summary ] .
we selected targets based on their mid - ir , x - ray and/or photometric variability characteristics and we describe each method in detail below .
the observations were designed for aaomega , a 400-fiber spectrograph mounted on the 3.9 m anglo - australian telescope ( aat ; @xcite ) .
the aaomega instrument has many more fibers than the expected number of quasars in its field of view , so we kept candidates satisfying any of the selection criteria , thereby aiming for higher completeness at the price of higher contamination .
there is no point in having empty fibers even if it means observing more contaminating stars .
@xcite showed that active galactic nuclei ( agns ) have redder mid - ir colors than galaxies and stars , and that they form a distinctive group in the mid - ir color - color plane ( see @xcite and @xcite , and also @xcite , @xcite , and @xcite for the mid - ir colors of x - ray and variability selected agns ) .
the red mid - ir colors of a typical agn are not due to dust emission , so mid - ir - selected agns are generally normal @xmath12 broad - line quasars ( see @xcite , @xcite ) .
there is also no significant tendency for them to be type ii quasars .
this is particularly true with the addition of a magnitude limit for the optical spectroscopy , since at most redshifts the emission from the disk is required to render the source bright enough to be observed . in @xcite
, we used this approach to identify quasars in the lmc and smc from the _ spitzer / irac _ four - band photometry ( i.e. , 3.6 , 4.5 , 5.8 , and 8.0@xmath13 ) of the surveying the agents of a galaxy s evolution project ( sage ; @xcite )
. the 5000 agn candidates were divided into several categories .
first , all the objects must be located inside the `` stern agn wedge '' ( @xcite ) .
the locus of cool blackbodies passes through the wedge so the quasar sample is divided into group `` a '' if they are far from the blackbody locus , and `` b '' if they can be contaminated with cold stars .
second , the mid - ir color - magnitude diagram ( cmd ) was divided into regions heavily contaminated by ysos and one that should principally contain quasars , based on mid - ir properties of ysos from @xcite and the mid - ir data from the 9 deg@xmath1 of the spitzer deep wide - field survey ( sdwfs ; @xcite ) of spectroscopically confirmed quasars from the agn and galaxy evolution survey ( ages ; @xcite ) .
finally , the candidates were classed as `` a '' if they had the optical - to - mid - ir colors of quasars in ages and `` b '' if they did not .
the most interesting candidates are classed as qso - aa ( quasar region , away from blackbody locus with qso optical - to - mid - ir colors ) , but given the large number of aaomega fibers , we targeted all classes of objects , excluding only those marked as `` outside '' ( of the ogle - iii fields ) or `` faint '' ( not detected on the ogle - iii template images ) .
we searched for the ogle - iii counterparts to the 758 x - ray sources distributed over @xmath14 deg@xmath1 of the lmc from .
the typical positional uncertainly for these x - ray sources is several arcseconds , so we can not effectively target them without some additional selection criterion given the stellar densities . as a second criterion we chose the most optically variable source within @xmath15 of the x - ray position . in this case
, variability simply means that the light curve is inconsistent with a constant flux , without any of the more quantitative variability selection criteria we discuss in the next section .
this led to a sample of 205 x - ray candidates , of which 160 were also mid - ir - selected candidates and 99 satisfy the variability cuts presented in section [ sec : varsel ] .
the ogle - iii database ( @xcite ) contains the data for 9 years ( 20012009 ) of continuously monitoring @xmath035 million objects toward the lmc .
we used this database to search for variable objects that are likely quasars based on the drw model of their light curves ( e.g. , @xcite ) .
we prepared and analyzed the light curves as described in @xcite .
we use the timescale @xmath16 , scaled amplitude @xmath17 , and the likelihood ratio @xmath18 between the best fitting model and a white noise model corresponding to simply expanding the photometric errors .
we also fit a power - law structure function ( sf ) and estimated its slope @xmath19 between 30 days and 2 years , and an amplitude @xmath20 defined by the magnitude difference between the first and third quartile of the sorted light curve . to remove unwanted variable sources , we use the following cuts : _ cut 1 .
_ the average light curve magnitude is @xmath21 mag ; _ cut 2 .
_ @xmath22 ; _ cut 3 .
_ the sf slope @xmath23 ( e.g. , see @xcite ) ; _ cut 4 . _ the @xmath5-band amplitude @xmath24 mag ( this removes large amplitude variable stars ) ; @xcite report that some of @xmath09000 sdss quasars occupy areas outside the @xmath16@xmath25 cut of @xcite , therefore we decided to loosen the cut on the @xmath16 parameter and to ignore @xmath25 parameter , so _
cut 5 _ is @xmath26 .
when simply applied to the ogle light curve database , this process yields @xmath27 candidates .
the vast majority are due to two known systematic issues .
first , bright variable stars typically generate several fake , fainter variable stars in their wings , and these `` ghost '' variables show long term irregular variability .
second , there can be small magnitude shifts between seasons , and ( for one field in particular ) the shifts were being misinterpreted as quasar - like variability .
we carried out a quick visual inspection of all candidates , but this was over kill and would certainly be automated in any subsequent analysis .
we were left with 1063 variability - selected candidates .
the total number of candidates is 2434 , distributed over @xmath030 deg@xmath1 in the lmc .
there are 2049 mid - ir - selected candidates and 385 non - mid - ir - selected candidates . of the 2049 mid - ir ( 385 non - mid - ir ) objects , 708 ( 355 ) are also variable and 160 ( 45 ) have associated x - ray emission .
there are 1063 objects selected as variable , of which 708 are mid - ir sources , 99 are x - ray sources , and 84 both mid - ir and x - ray sources . figure [ fig : venncand ] shows how the samples overlap .
note that the limited overlap of the mid - ir and x - ray - selected agns is also seen in the spectroscopically confirmed samples of agns in the ndwfs field ( e.g. , @xcite ) .
the aaomega fiber allocation software allows to prioritize targets from 1 ( the lowest priority ) to 9 .
we gave priorities of 9 , 8 , and 7 to sources selected by all three methods ( 84 sources ) , sources selected by only two methods ( 715 sources ) , and sources selected by only one method ( 1635 ) , respectively .
lccccr lmc 01 & 04:41:43 & @xmath2869:50:29 & 220 & @xmath29 & @xmath29 + lmc 02 & 04:43:19 & @xmath2868:19:18 & 197 & @xmath29 & @xmath29 + lmc 03 & 04:56:52 & @xmath2867:07:48 & 213 & @xmath29 & @xmath29 + lmc 04 & 05:00:51 & @xmath2870:27:49 & 221 & 41 & 1480 + lmc 05 & 05:01:42 & @xmath2868:49:26 & 269 & @xmath29 & @xmath29 + lmc 06 & 05:14:27 & @xmath2867:35:00 & 189 & @xmath29 & @xmath29 + lmc 07 & 05:19:43 & @xmath2869:31:07 & 307 & 36 & 5400 + lmc 08 & 05:21:54 & @xmath2871:02:48 & 247 & 60 & 3600 + lmc 09 & 05:32:54 & @xmath2868:31:01 & 230 & @xmath29 & @xmath29 + lmc 10 & 05:41:31 & @xmath2871:36:00 & 201 & 37 & 2550 + lmc 11 & 05:41:56 & @xmath2870:11:07 & 275 & @xmath29 & @xmath29 + lmc 12 & 05:52:39 & @xmath2868:57:51 & 183 & @xmath29 & @xmath29 + we are mostly interested in quasars located in the overlapping areas of the sage and ogle - iii surveys .
we divided this lmc area into 12 fields ( see figure [ fig : mqs_lmc_fields ] ) , each @xmath03 deg@xmath1 and corresponding to the field of view of aaomega .
the basic information for each field is given in table [ tab : mqsfields ] .
the fields overlap slightly to avoid having gaps between them , so some of the quasar candidates were observed up to three times if they happened to fall into such an overlapping region .
the total number of targeted objects in all 12 fields was 2678 , of which several ( less than 244 ) were targeted at least twice . on average
there were 220 objects in a field , corresponding to @xmath070 sources deg@xmath30 .
we observed the lmc fields on 2011 february 13 .
unfortunately , due to arrival of tropical storm yasi , we were completely clouded out on the night of february 2 and partly on february 1 and 3 . of the 12 planned fields
, we were able to execute only four ( see table [ tab : mqsfields ] and figure [ fig : mqs_lmc_fields ] ) , and only one with the desired 1.5 hr ( 5400 s ) exposure time .
nevertheless , the data are sufficient to confirm a large number of quasars .
we observed 845 candidates of which we confirm 169 ( 144 new ) in this paper .
the yields divided by the agn selection method are presented in table [ tab : selectionresults ] .
we followed the spectral identification procedures from paper i ( @xcite ) . to find as many quasar emission lines in a spectrum as possible
, we decided to use the widest available spectral coverage of @xmath05100 ( 37508850 ) and a resolution of @xmath31 .
each spectrum consists of a blue ( 580v ) and a red ( 385r ) channel that were spliced together at 5700 .
some of the spectra appear to have a `` wide broad line '' around the splicing wavelength , mimicking the mgii ( 2800 ) line at a redshift of @xmath32 .
we carefully checked all our @xmath33 quasars for possible misidentifications .
our initial goal was to obtain three exposures of @xmath34 s per field , resulting in a median signal - to - noise ratio ( s / n ) of @xmath025 at @xmath35 mag ( calculated from obtained spectra ) . to subtract the sky flux contribution from each spectrum
, we initially selected @xmath050 and finally used @xmath025 sky fibers , each positioned to avoid stellar emission based on the ogle - iii lmc catalogs ( @xcite ) .
these catalogs were also used to find @xmath020 single and well separated bright guide stars for each field , of which 48 were used for guiding .
the aaomega fld files were created with the configure software , and the data reduction was performed using the 2dfdr software ( @xcite ) . as in paper
i , we searched each spectrum for the common ( redshifted ) quasar emission lines ( e.g. , @xcite ) ly@xmath36 at 1216 , h@xmath37 at 4101 , h@xmath19 at 4340 , h@xmath38 at 4861 , h@xmath36 at 6563 , magnesium mgii at 2800 , carbon civ at 1549 and ciii ] at 1909 , as well as the narrow forbidden lines of oxygen [ o ii ] at 3727 , [ o iii ] at 4959 or 5007 . in general , we required the identification of two lines , except in the redshift range @xmath39 , where we can only find mgii despite the broad spectral coverage .
we identified 144 new quasars behind the lmc .
their basic parameters ( including the identified lines ) are presented in table 3 .
we also confirmed the 25 known quasars that were in our target list ( table 4 ) .
of these known quasars , we had flagged four only as probable quasars because their spectra were too noisy for reliable classification .
we missed one known quasar , mqs [email protected] , that had extremely low s / n in our data , although we verified its existence .
these five sources are reported at the bottom of table 4 .
the newly discovered quasars cover the range of redshifts @xmath40@xmath41 as shown in figure [ fig : hist ] .
the deficit at @xmath42 is probably due to problems in convincingly identifying quasars based on only mgii in the region of the dichroic split .
we show ten selected spectra of the newly confirmed agns in figure [ fig : spectra1 ] and also show several spectra of common contaminating sources in figure [ fig : contamination ] .
the remaining spectra are discussed in section [ sec : contamination ] , and are either not quasars or quasars with too low s / n for clear detection .
as we discuss in section [ sec : discussion ] , a fair number of the fainter sources ( @xmath43 mag ) should be quasars .
bins.,width=302 ] one target , mqs [email protected] , shows emission lines corresponding to redshifts of @xmath44 and @xmath45 , as shown in figure [ fig : doubleagn ] .
this could be a chance coincidence or a gravitational lens . in the ogle - iii reference image ( figure [ fig : doubleagnchart ] ) , we find that the agn candidate was a faint source ( @xmath46 , @xmath47 ) in the wings of a brighter extended source .
the standard ogle catalogs may not include all extended sources , so it is likely that the faint source was mistargeted and the brighter galaxy is the agn candidate .
if we assume the bright galaxy is acting as a gravitational lens , then we estimate that it has an einstein radius of @xmath48 arcsec , which is very small because of the small redshift difference .
higher resolution images and spatially resolved spectra are required to clarify the nature of this source .
it displays two sets of the same emission lines at two redshifts , @xmath44 and @xmath49 .
we did not mask the lmc s ism lines [ oi ] ( 6300 ) , h@xmath36 ( 6563 ) , and hei ( 6678).,width=302 ] 700210.1 .
the circle has a diameter of 2 arcsec , corresponding to the aaomega fiber size .
the plus symbol inside the fiber marks the targeted source , while the plus symbol outside the fiber marks the galaxy , most likely hosting an agn .
the image is @xmath50 arcsec .
north is up , east is to the left.,width=302 ]
for the sources that were not quasars , we used our spectra of the known sources to build template yso , pn , blue / be , and red star spectra , respectively , and fit them to the individual spectra .
apart from the automatic fits , we also visually inspected each spectrum .
table 5 reports the object type for each of the sources and figure [ fig : contamination ] shows examples of some of these sources .
we briefly describe each type below .
_ planetary nebulae .
_ we classified sources as candidate pne if they had roughly equally strong h@xmath36 , h@xmath38 , and/or [ oiii ] emission lines along with several additional lines such as hei at 5876 , 6680 and 7065 , heii at 4686 , [ ariii ] at 7136 , [ nii ] at 6548 and 6584 , [ sii ] 6716 and 6731 , and/or [ ariv ] at 4711 and 4740 ( see , e.g. , @xcite ) . of the 17 we classified as pne based on our spectra ,
7 were classified as such in simbad and 5 were classified as ysos .
_ young stellar objects .
_ we classified sources as candidate ysos if they had significantly weaker [ oiii ] and h@xmath38 lines relative to h@xmath36 combined with relatively strong [ sii ] emission lines at 6716 and 6731 , [ oi ] at 6300 , and/or [ nii ] at 6548 and 6584 ( see , e.g. , @xcite ) . of the 115 objects we classified as ysos , 18 were classified as such by simbad and 1 was classified as a pn , leaving 96 new candidate ysos
there were also four additional sources we identify as yso / pn where we were uncertain as to the correct classification .
simbad has two of these four classified as ysos . _
be and blue stars_. blue stars had blue spectra with strong hydrogen balmer series absorption .
they were then further classed as be stars if they had strong hydrogen h@xmath36 ( h@xmath38 ) emission lines .
we identified 63 blue stars of which 39 appear to be be stars . while there is little ambiguity about the be stars , simbad labels two of the blue stars without emission lines as pne , one as a mira - type star and one only as a two micron all sky survey ( 2mass ) infrared source ( @xcite ) .
we also found eight objects that could be either blue stars or ysos .
the median ogle - iii color of these blue stars is @xmath51 mag . _
red stars_. these are stars with red spectra sometimes showing molecular absorption bands .
we identified 68 red sources , where simbad classifies 4 as infrared sources ( one yso from @xcite and three 2mass sources ) , 1 as a pn , 1 as a yso , 1 as an asymptotic giant branch star , and 1 as a galaxy .
the median ogle - iii color of these sources is @xmath52 mag .
the yield of our survey is determined by a combination of contamination and depth , where it is difficult to fully characterize the effects of depth because of the large variation in integration time created by the weather . for discussion , we simply combine the four fields ( table [ tab : selectionresults ] ) . we know from the surface density of candidates compared to quasars ( @xcite , @xcite ) , shown in figure [ fig : cumul ] , that the level of contamination is high , but much of this is by design because of the large number of available fibers .
lccc mid - ir qso - aa & 577 & 136 ( 137@xmath53 ) & 24 + mid - ir qso - ab & 10 & 2 & 20 + mid - ir qso - ba & 62 & 12 & 19 + mid - ir qso - bb & 1 & 0 & 0 + mid - ir yso - aa & 32 & 4 & 13 + mid - ir yso - ab & 16 & 1 & 6 + mid - ir yso - ba & 1 & 0 & 0 + mid - ir yso - bb & 3 & 0 & 0 + mid - ir ( any ) & 702 & 155 ( 156@xmath53 ) & 22 + x - ray ( any ) & 102 & 30 & 29 + var .
( any ) & 350 & 107 & 31 + x - ray + mid - ir ( any ) & 77 & 25 & 32 + mid - ir + var .
( any ) & 228 & 95 & 42 + var . + x - ray ( any ) & 39 & 26 & 67 + + + mid - ir ( only ) & 431 & 57 ( 58@xmath53 ) & 13 + x - ray ( only ) & 21 & 1 & 5 + var .
( only ) & 117 & 8 & 7 + + + x - ray + mid - ir ( only ) & 43 & 3 & 7 + mid - ir + var .
( only ) & 194 & 73 & 38 + var .
+ x - ray ( only ) & 5 & 4 & 80 + + + all three & 34 & 22 & 65 in figure [ fig : venncand ] and table [ tab : selectionresults ] , we present our yields from this observing run , divided by the selection method . the mid - ir - selected sources were divided into several classes ( qso - aa , qso - ab , etc . ;
see section [ sec : mirsel ] ) . as expected the purest class , qso - aa ,
has the highest confirmation rate , at 24% .
next , the two classes qso - ab and qso - ba had confirmation rates of 20% and 19% , respectively .
the yso - aa and yso - ab groups have lower rates of 13% and 6% , as expected from the original division into yso and qso classes .
variability selection had a yield of 31% and x - ray selection 30% .
priority 7 objects , where the source was flagged based on only one of the mid - ir , variability or x - ray criteria , had confirmation rates of 13% ( 57 new agns ) , 7% ( 8) , and 5% ( 1 ) , respectively . of the priority
8 objects selected by any two methods but not the third , the combination of x - ray plus variability yielded the highest confirmation rate , at 80% ( 4 new agns ) , most likely due to low number statistics .
the mid - ir and variability criteria had a yield of 38% ( 73 agns ) and x - ray plus mid - ir criteria had a yield of only 7% , again with low number statistics . not surprisingly , combining all three selection methods was the most efficient with a yield of 65% ( 22 new agns ) .
figure [ fig : cumul ] shows the cumulative distribution of our targets and then their division into quasars , stars , and unclassified objects .
we also show the cumulative distribution of @xmath54 quasars in ages ( @xcite ) .
these quasars are primarily mid - ir and x - ray selected , like much of the present sample , but with a much deeper optical spectroscopic limit .
we have adjusted the ages magnitudes for the @xmath55 mag extinction difference between ages and a typical lmc site line ( @xcite ) , although this may underestimate the correction necessary for a true background population .
the completeness of our present sample relative to ages is roughly 50% ( 30% ) at 18.6 mag ( 19.3 mag ) .
figure [ fig : cumul ] also shows the cumulative distribution of the sdss quasars with @xmath56 from @xcite .
given the change in bandpass ( the @xmath57 mag color term varies significantly with redshift , see figure [ fig : agncolor ] ) and the additional line width and luminosity selection criteria in @xcite it is harder to make a direct comparison , but after matching as best possible , we obtain similar completeness estimates . as expected from filling the fibers with targets ,
most targets are peculiar stars of various types , particularly at the brighter magnitudes .
unclassifiable spectra dominate at the fainter magnitudes , and with the weather - limited integration times , there was an effective magnitude limit of roughly @xmath58@xmath59 mag . and
@xmath54 sources ) .
the red solid ( dashed ) curves show the distribution of @xmath54 ages ( sdss @xmath56 ) quasars for comparison .
our selection methods and bands are more comparable to ages than to sdss . , width=302 ] colors of the mqs lmc quasars ( filled squares ) and other confirmed lmc and smc quasars ( open circles ) as a function of redshift .
overplotted is the color for a pure agn spectrum from @xcite convolved with the @xmath6 and @xmath5 filters ( solid line ) .
the same spectrum was also convolved with the @xmath60 and @xmath5 filters in order to convert ogle magnitudes into sdss @xmath60-band magnitudes
the @xmath61 color is shown as dotted line.,width=302 ] figure [ fig : agncolor ] shows the colors of the confirmed quasars as a function of redshift . while we imposed no optical color selection criteria ,
the quasars generally follow the colors of sdss quasars except at very low @xmath62 , where the host dominates the colors . adding an optical color constraint has no leverage on the selection of quasars , as the colors of quasars are indistinguishable from the colors of many lmc stars . in figures [ fig : tau - sigma][fig : kko9 ] , we examine the selection spaces used to select the candidates .
first , we inspect the @xmath16@xmath25 variability plane from @xcite and @xcite .
@xcite proposed a narrow and long trapezoidal region to select high purity sample of agn candidates . indeed ,
if we had restricted our variability selection to this region , we would have a yield of 70% at the price of a significant reduction in completeness .
in fact , the distribution of the variability parameters of all quasars follows the distribution found by @xcite for sdss quasars , extending outside the @xcite selection boundaries , particularly for long @xmath16 where there is also increased contamination from irregularly varying stars .
the left panel of figure [ fig : tau - sigma ] shows the distribution of objects classified as non - quasars ( pne , ysos , etc . ) or where we failed to classify the spectrum in this variability plane .
we note that the vast majority of classified non - quasars lie outside the @xcite region . in figure
[ fig : schmidt ] , we make the same comparison between the mqs quasars and the other candidates in the @xmath20@xmath19 variability plane of @xcite . we used a cut in this plane to remove more variable sources ( shown as contours ) other than quasars that we included as we greatly relaxed the drw selection criteria of @xcite in these dense stellar fields
. a fair number of common variable stars are periodic with the index of the power - law sf of @xmath63 . on the other hand , objects with very high amplitude variations are unlikely to be quasars as quasar variability amplitudes on long time - scales are expected to be @xmath64 mag ( e.g. , @xcite ) .
our simple cut ( solid lines ) allows higher contamination than that of ( * ? ? ?
* dotted lines ) , but also returns the higher number of confirmed quasars .
we also examined the distribution of the confirmed mqs quasars and quasar candidates relative to the mid - ir selection criteria of ( * ? ? ? * figure [ fig : kko9 ] ) .
the quasars lie in the proposed selection regions confirming the basic principles outlined in @xcite .
the remaining quasar candidates ( shown in the left panels ) are either quasars with low s / n or contaminating sources . in the mid - ir color - color plot ( top panels in figure [ fig : kko9 ] ) ,
the confirmed quasars ( shown in the right panel ) and the faint candidates ( gray squares in the left panel ) occupy nearly identical regions , while the blue and red stars tend to clump in the lower - left region of the `` agn wedge '' area .
also , ysos tend to have a somewhat larger color spread than the more centrally clumped agns .
in the mid - ir cmd ( middle panels of figure [ fig : kko9 ] ) , the confirmed quasars seem to be on average brighter than the remaining faint sources .
the red stars tend to have bluer mid - ir @xmath65-[8.0]$ ] colors than the other sources . in the bottom panels of figure [ fig : kko9 ] ,
we show the optical - to - mid - ir colors of our observed targets .
as expected the vast majority of our quasars occupy the `` a '' ( qso - like optical - to - mid - ir colors ) region , but then so do the most of other sources .
the red stars again have bluer optical - to - mid - ir colors than the other sources .
+ + as expected from earlier surveys of mid - ir - selected quasars ( @xcite ) , we find that 88% of the mid - ir - selected quasars show optical broad lines and are not type ii quasars .
similarly , of the 155 confirmed , mid - ir - selected quasars , 95 ( 61% ) are clearly variable sources .
we observed 845 agn candidates in four @xmath03 deg@xmath1 lmc fields with aaomega to confirm 169 quasars , including 144 new ones .
given the 55 previously known quasars behind the lmc , including 25 in our present fields , we have quadrupled ( to 200 ) the number of known quasars behind the lmc . by observing the eight remaining fields , we should find another @xmath0300 new quasars , although the yield would increase if all data were obtained with the full integration time .
we observed the quasar candidates based on the priorities set by the number of the selection methods that picked them .
if they were selected by only one method , either mid - ir colors , x - ray emission , or photometric variability , the average confirmation yield is low , at @xmath09% . if the candidate was selected based on two methods but not the third , the average confirmation rate is at 40% .
the highest yields come with the simultaneous selection by all three methods , with the yield at 65% .
note , however , that we deliberately chose to fill all the aaomega fibers with candidates in order to achieve higher completeness , guaranteeing high contamination rates because the surface density of quasars is well below the surface density of fibers .
our biggest problem given the availability of fibers is not contamination .
we could assign rough classifications to many contaminating sources , including 115 ysos ( 96 new ) , 17 pne ( 5 new ) , 39 new be and 24 blue stars , 68 red stars , and also 12 objects that are either yso / pn or blue star / yso .
the big problem is that we could not classify 402 spectra due to low s / n , what is also a major problem for making black hole mass estimates from the line widths of the quasar emission lines .
nonetheless , tripling the number of quasars in the lmc and smc using only 5 hr on sky is a strong vindication of the general approach and the advantages created by the large numbers of aaomega fibers . with no improvements in yields due to actually obtaining our target integration times
, we estimate that completing the ogle - iii lmc and smc fields would yield @xmath0700 quasars and completing the ogle - iv fields would yield @xmath03600 quasars .
these remain the best fields for extending studies of quasar variability independent of their additional uses for understanding the galaxy through the proper motions of the lmc or probing the local ism .
we thank sarah brough , our support astronomer , for help during the observing run .
we thank micha jaroszyski for the helpful discussion on mqs [email protected] .
we also thank the anonymous referee , whose comments helped us to improve the manuscript .
this research was made based on observations with the anglo - australian telescope , for which the observing time was granted by the optical infrared coordination network for astronomy ( opticon ) .
this research has made use of the simbad database , operated at cds , strasbourg , france .
this research has also made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory ( jpl ) , california institute of technology ( caltech ) , under contract with the national aeronautics and space administration ( nasa ) .
support for this work was provided by the polish ministry of science and higher education ( mnisw ) through the program `` iuventus plus '' , award number ip2010 020470 to s.k .
, a.m.j . , and
this work has been supported by nsf grants ast-0708082 and ast-1009756 to c.s.k .
the ogle is supported by the european research council under the european community s seventh framework programme ( fp7/2007 - 2013 ) , erc grant agreement no .
246678 to a.u .
dobrzycki , a. , groot , p. j. , macri , l. m. , & stanek , k. z. 2002 , , 569 , l15 dobrzycki , a. , macri , l. m. , stanek , k. z. , & groot , p. j. 2003a , , 125 , 1330 dobrzycki , a. , stanek , k. z. , macri , l. m. , & groot , p. j. 2003b , , 126 , 734 dobrzycki , a. , eyer , l. , stanek , k. z. , & macri , l. m. 2005 , , 442 , 495 | we quadruple the number of quasars known behind the large magellanic cloud ( lmc ) from 55 ( 42 in the lmc fields of the third phase of the optical gravitational lensing experiment ( ogle ) ) to 200 by spectroscopically confirming 169 ( 144 new ) quasars from a sample of 845 observed candidates in four @xmath03 deg@xmath1 anglo - australian telescope / aaomega fields south of the lmc center .
the candidates were selected based on their _ spitzer _ mid - infrared colors , x - ray emission , and/or optical variability properties in the database of the ogle microlensing survey .
the contaminating sources can be divided into 115 young stellar objects ( ysos ) , 17 planetary nebulae ( pne ) , 39 be and 24 blue stars , 68 red stars , and 12 objects classed as either yso / pn or blue star / yso .
there are also 402 targets with either featureless spectra or too low signal - to - noise ratio for source classification .
our quasar sample is 50% ( 30% ) complete at @xmath2 mag ( 19.3 mag ) .
the newly discovered active galactic nuclei ( agns ) provide many additional reference points for proper motion studies of the lmc , and the sample includes 10 bright agns ( @xmath3 mag ) potentially suitable for absorption line studies . their primary use , however , is for detailed studies of quasar variability , as they all have long - term , high cadence , continuously growing light curves from the microlensing surveys of the lmc .
completing the existing magellanic quasars survey fields in the lmc and small magellanic cloud should yield a sample of @xmath0700 well - monitored agns , and expanding it to the larger regions covered by the ogle - iv survey should yield a sample of @xmath03600 agns . |
recent years have seen a surge of interest in wireless networks . unlike the traditional point - to - point communication ,
elementary modes of cooperation such as relaying are needed to improve both the throughput and reliability in a wireless network .
although capacity of a relay channel @xcite is still unknown in general , considerable progress has been made on several aspects , including some achievable capacity results @xcite and capacity scaling laws of large networks @xcite . in parallel , research on the cooperative diversity @xcite , where the relays help the source exploit the spatial diversity of a slow fading channel in a distributed fashion , has attracted significant attention @xcite . in small relay networks where the source signal can reach the destination terminal via a direct link ,
many results have been known in both the channel capacity @xcite and the cooperative diversity .
the capacity results are mostly based on the decode - and - forward ( df ) and the compress - and - forward ( cf ) strategies .
the amplify - and - forward ( af ) scheme , however , is rarely considered in this scenario due to the noise accumulation at the relays . on the other hand ,
the af scheme is widely used for cooperative diversity .
it has been shown in @xcite that the af scheme is as good as the df scheme at high snr as far as the diversity is concerned .
furthermore , it is pointed out in @xcite that not needing to decode the source signal makes the relays more capable of protecting the source signal in some cases . the cf scheme , which works with perfect global channel state information ( csi ) , is usually excluded in the cooperative diversity scenario for practical considerations . in larger relay networks , where direct source - destination links are generally absent
, substantial results on the capacity scaling laws have been obtained in the large network size regime @xcite . however ,
much less is known about the cooperative diversity than in the case of small networks .
this paper analyzes the cooperative diversity in relay networks with a single multi - antenna source - destination terminal pair .
the source signal arrives at the destination via a sequence of @xmath0 hops through @xmath1 layers of relays .
similar channel setting with a single layer has been studied in @xcite in different contexts . using large random matrix theory ,
the ergodic capacity results of some particular relaying schemes have been established for large networks @xcite .
recently , the study has been extended to the case with multiple layers of relays @xcite and the case with multiple source - destination pairs @xcite .
cooperative diversity in this setting was first studied in @xcite for the single - antenna case then in @xcite for the multi - antenna case , with distributed space - time coding .
all the mentioned works assume linear processing at the relays and the df scheme is not considered . actually , one can figure out immediately that the df scheme is not suitable for the multi - antenna setting due to the suboptimality in terms of degrees of freedom .
requiring the relays to decode the source signal restricts the achievable degrees of freedom .
this is one of the fundamental differences between the large networks and small networks : the degrees of freedom of the latter are determined by the source - destination link and not by the relaying strategy . in this work ,
we suppose that the network size is arbitrary ( but fixed ) and the signal - to - noise ratio ( snr ) is large . the multihop channel is investigated in terms of the diversity - multiplexing tradeoff ( dmt ) .
the dmt was introduced in @xcite for the point - to - point multi - antenna ( mimo ) channels to capture the fundamental tradeoff between the throughput and reliability in a slow fading channel at high snr .
it was then extensively used in multiuser channels such as the multiple access channels @xcite and the relay channels @xcite as performance measure and design criterion of different schemes .
our main contributions are summarized in the following paragraphs .
first , we use the information theoretic cut - set bound @xcite to derive an upper bound on the dmt of any relaying strategy . in the clustered case where the relays in the same layer can fully cooperate ,
this bound is shown to be tight .
an optimal scheme is the cooperative df scheme , where the clustered relays perform joint decoding and joint re - encoding . while the clustered channel is equivalent to a series - channel and does not feature the distributed nature of wireless networks , the non - clustered case is studied as the main focus of the paper .
since no within - layer cooperation is considered , linear processing at the relays is assumed .
we start by the af strategy , which seems to be the natural first choice as a linear relaying scheme .
we show that the af scheme is , in the dmt sense , equivalent to the rayleigh product ( rp ) channel , a point - to - point channel whose channel matrix is defined by a product of @xmath0 gaussian matrices .
that being said , we examine the rp channel in great detail .
it turns out that the dmt of a rp channel has a nice recursive structure and lends some intuitive insights into the typical outage events in such channels .
the study of the rp channel leads directly to an exact dmt characterization for the af scheme in multihop channels of arbitrary size .
the closed - form dmt provides simple guidelines on how to efficiently use the available relays with the af scheme .
one such example is how to reduce the number of relays while keeping the same diversity .
while the maximum multiplexing gain is achieved , the achievable diversity gain of the af scheme can be far from maximum diversity gain suggested by the cut - set bound .
specifically , the dmt of the af scheme is limited by a virtual `` bottleneck '' channel .
the following question is then raised : is the dmt cut - set bound tight in the non - clustered case ?
the question is partially answered in this work : there exists a scheme that achieves both extremes of the cut - set bound , that is , the maximum diversity extreme and maximum multiplexing extreme . in order to achieve the maximum diversity gain ,
the key is space - time relay processing . noting that the af scheme is space - only
, we incorporate the temporal processing into the af scheme . the first scheme that we propose
parallel af _ scheme . by partitioning the multihop channel into @xmath2 `` af paths '' , we create a set of @xmath2 parallel sub - channels in the time domain .
a packet that goes through the parallel channel attains an improved diversity if the partition is properly designed .
it is shown that there is at least one partition such that the maximum diversity is achieved .
however , the parallel af scheme does not have the maximum multiplexing gain in general , since the achievable degrees of freedom by the scheme are restricted by those of the individual af paths .
in most cases , the af paths are not as `` wide '' as the original channel in terms of the degrees of freedom . in order to overcome the loss of degrees of freedom , we linearly transform the set of parallel af channels into another set in which each sub - channel has the same degrees of freedom as the multihop channel . in the new parallel channel ,
each relay only need to flip the received signal in a pre - assigned mode , hence the name _ flip - and - forward _ ( ff ) .
it is shown that the ff scheme achieves both the maximum diversity and multiplexing gains .
furthermore , the dmt of the ff scheme is lower - bounded by that of the af scheme . using the results obtained in the non - clustered case , we revisit the clustered case by pointing out that the cooperative df operation might not be needed in all clusters to get the maximum diversity .
we also indicate that cross - antenna linear processing in each cluster helps to improve the dmt only when both transmitter csi and receiver csi are known to the relays .
finally , coding schemes are proposed for all the studied relaying strategies . in the clustered case ,
a series of perfect space - time block codes ( stbcs ) @xcite with appropriate rates and dimensions are used at the source and each relay cluster that performs the cooperative df operation . in the non - clustered case ,
construction of perfect stbcs for general parallel mimo channels is first provided .
the constructed codes can be applied directly to the parallel af scheme and the ff scheme . all
suggested coding schemes achieve the dmt despite of the fading statistics and are thus approximately universal @xcite . regarding the notations , we use boldface lower case letters @xmath3 to denote vectors , boldface capital letters @xmath4 to denote matrices .
@xmath5 represents a complex gaussian random variable with mean @xmath6 and variance @xmath7 .
@xmath8 $ ] stands for the expectation operator .
@xmath9}^{{\scriptscriptstyle \mathsf{t } \!}}},{{[\cdot]}^{{\scriptscriptstyle \dag \!}}}$ ] respectively denote the matrix transposition and conjugated transposition operations .
@xmath10 is the vector norm .
@xmath11 is the frobenius matrix norm .
we define @xmath12 for any matrices @xmath13 s .
the square root @xmath14 of a positive semi - definite matrix @xmath15 is defined as a positive semi - definite matrix such that @xmath16 .
@xmath17 and @xmath18 denote respectively the maximum and minimum eigenvalues of a semi - definite matrix @xmath15 .
@xmath19 means @xmath20 . @xmath21
( respectively , @xmath22 ) is the closest integer that is not smaller ( respectively , not larger ) than @xmath23 .
@xmath24 means @xmath25 .
@xmath26 stands for the base-@xmath27 logarithm . for any quantity @xmath28 , @xmath29 and similarly for @xmath30 and @xmath31 .
the tilde notation @xmath32 is used to denote the ( increasingly ) ordered version of @xmath33 .
let @xmath34 and @xmath33 be two vectors of respective length @xmath35 and @xmath36 , then @xmath37 means @xmath38 , @xmath39 .
@xmath40 means that @xmath34 is a sub - vector of some permutated version of @xmath33 .
the rest of this paper is organized as follows .
section [ sec : system - model ] describes the system model and some basic assumptions in our work . the dmt cut - set bound and the clustered case with the df scheme are presented . in section [ sec : af ] , we study the non - clustered case with the af scheme . the parallel af and the ff schemes
are proposed in section [ sec : para - partition ] . in section
[ sec : clustered ] , the clustered case is revisited .
the approximately universal coding schemes are proposed in section [ sec : stc ] .
section [ sec : nr ] provides some selected numerical examples .
finally , a brief conclusion is drawn in section [ sec : conclusion ] .
most detailed proofs are deferred to the appendices .
the considered @xmath0-hop relay channel model is composed of one source ( layer @xmath41 ) , one destination ( layer @xmath0 ) , and @xmath1 layers of relays ( layer @xmath42 to layer @xmath1 ) . each terminal is equipped with multiple antennas . the total number of antennas in layer @xmath43 is denoted by @xmath44 . for convenience , we define @xmath45 , @xmath46 , and @xmath47 .
we assume that the source signal arrives at the destination via a sequence of @xmath0 hops through the @xmath1 layers and that terminals in layer @xmath43 can only receive the signal from layer @xmath48 . the fading sub - channel between layer @xmath48 and layer @xmath43 is denoted by the matrix @xmath49 .
sub - channels are assumed to be mutually independent , flat rayleigh - fading and quasi - static .
that is , the channel coefficients are independent and identically distributed ( i.i.d . )
complex circular symmetric gaussian with unit variance .
and they remain constant during a coherence interval of length @xmath50 and change independently from one coherence interval to another .
furthermore , the transmission is supposed to be perfectly synchronized . under these assumptions , the signal model within a coherence interval
can be written as @xmath51 = { { { \uppercase{{\pmb{h}}}}}}_{i}{\,}{{{\lowercase{{\pmb{x}}}}}}_{i-1}[l ] + { { { \lowercase{{\pmb{z}}}}}}_{i}[l],\quad l=1,\ldots l,\]]where @xmath52,{{{\lowercase{{\pmb{y}}}}}}_i[l]\in{{\mathbb{c}}}^{n_i\times
1}$ ] denote the transmitted and received signal at layer @xmath43 ; @xmath53\in{{\mathbb{c}}}^{n_i\times 1}$ ] is the additive white gaussian noise ( awgn ) at layer @xmath43 with i.i.d . @xmath54 entries .
since we consider the non - ergodic case where the coherence time interval @xmath50 is large enough , we drop the time index @xmath55 hereafter .
it is assumed that all relays work in full - duplex mode and the transmission is subject to the short - term power constraint @xmath56with @xmath57 being the average transmitted snr per layer .
all terminals are supposed to have perfect channel state information at the receiver and no csi at the transmitter . from now on
, we denote the channel as a @xmath58 multihop channel .
slow fading channels are outage - limited , i.e. , there is an _ outage probability _ @xmath59 that the channel can not support a target data rate of @xmath60 bits per channel use at signal - to - noise ratio @xmath57 . in the high snr regime
, this fundamental interplay between throughput and reliability is characterized by the diversity - multiplexing tradeoff @xcite .
the _ multiplexing gain _ @xmath61 and _ diversity gain _ @xmath62 of a fading channel are defined by @xmath63 a more compact form is @xmath64 note that in the definition we use the outage probability instead of the error probability , since it is shown in @xcite that the error probability of any particular coding scheme with maximum likelihood ( ml ) decoding is dominated by the outage probability at high snr and that the thus defined dmt is the best that one can achieve with any coding scheme . in the rayleigh mimo channel , the dmt has the following closed form .
[ lemma : dmt - rayleigh ] the dmt of a @xmath65 rayleigh channel is a piecewise - linear function connecting the points @xmath66 , @xmath67 , where @xmath68 in the following , we will use the dmt as our performance measure . for convenience of presentation , we provide the following definition .
two channels are said to be _ dmt - equivalent _ or _ equivalent _ if they have the same dmt . before studying any specific relaying strategy
, we establish an upper bound on the dmt of the multihop system as a benchmark .
[ prop : dmt - ub ] for any relaying strategy @xmath69 , we have @xmath70 with @xmath71 where @xmath72 is the dmt of the point - to - point channel between layer @xmath48 and layer @xmath43 . in particular , by defining the maximum diversity gain and multiplexing gain as @xmath73 and @xmath74 , respectively , we have @xmath75 from the information theoretic cut - set bound @xcite , the mutual information between the source and the destination satisfies @xmath76 for any relaying strategy @xmath69 .
thus , the outage probability using a relaying scheme @xmath69 is @xmath77 where @xmath78 is the outage probability of the @xmath43th sub - channel . from ( [ eq : dmt - compact ] ) and ( [ eq : tmp999 ] ) , we prove ( [ eq : dmt - ub ] ) . finally , ( [ eq : dmax ] ) and ( [ eq : rmax ] ) are from the direct application of lemma [ lemma : dmt - rayleigh ] .
if we assume that the relays within the same layer are clustered , i.e. , they can perform joint decoding and joint re - coding operations , then each layer can act as a virtual multi - antenna terminal .
this could happen either when the relays are controlled by a central unit via wired links or when they are close enough to each other to exchange information perfectly . in this case
, the relay channel model is equivalent to a serial concatenation of @xmath0 independent mimo channels
. let us consider the following cooperative decode - and - forward scheme .
each layer tries to cooperatively decode the received signal .
when a successful decoding is assumed , the embedded message is re - encoded and then forwarded to the next layer .
we can show that this simple scheme is dmt optimal .
[ prop : df ] when the relays are clustered , the cooperative df scheme achieves the dmt cut - set bound @xmath79 defined in ( [ eq : dmt - ub ] ) . to show the achievability , note that the cooperative df scheme being in outage implies the outage of at least one of the sub - channels . by the union
bound , @xmath80 at high snr , the probability is dominated by the largest term in the sum of the right - hand side ( rhs ) . from ( [ eq : dmt - compact ] ) , we get @xmath81 in the high snr regime , the union bound defined by the sum operation coincides in the snr exponent with the cut - set bound defined by the minimum operation .
hence , the dmt cut - set bound is tight in the clustered case .
however , relays in wireless networks are not clustered in general . in fact , one of the important and interesting attributes of wireless networks is the distributed nature . in the following two sections ,
we will concentrate on the non - clustered case and analyze the achievable dmt .
in this section , we consider the non - clustered case , where the relays work in a distributed manner and no within - layer communication is allowed . in this case , applying the df scheme at each individual relay might incur loss of degrees of freedom . to see this ,
take the single - layer channel as an example . in the best case where all the relays succeed in decoding ,
they transmit the message using a pre - assigned codebook .
this scheme transforms the relays - destination channel into a @xmath82 virtual mimo channel . before this
could possibly happen , however , the success decoding at the relays must be guaranteed with high probability .
this constraint imposes that the degrees of freedom in this scheme must not be larger than @xmath83 with @xmath84 being the number of antennas at the @xmath85th relay .
while this scheme achieves the maximum multiplexing gain in the single - antenna case , it could fail in the multi - antenna case . since we do not know how to cooperate efficiently in this case , we start by the most obvious and naivest relaying scheme : the amplify - and - forward scheme .
this scheme in the considered setting has been studied in @xcite for the capacity scaling laws , and in @xcite for the dmt .
it is worth noting that , in @xcite , a lower bound on the dmt of the af scheme in a symmetric network ( @xmath86 , @xmath87 ) was obtained , while our work derives the exact dmt for a network of arbitrary dimension with a different approach . in the considered af scheme
, each antenna node normalizes the received signal to the same power level and then retransmits it .
this linear operation can be expressed as @xmath88where , by the power constraint ( [ eq : power - constraint ] ) , @xmath89 the scaling matrix @xmath90 is diagonal due to the antenna - wise nature of the relaying scheme , with the normalization factors in ( [ eq : md ] ) by @xmath42 s . ]
@xmath91thus , the signal model of the end - to - end channel is @xmath92where , for the sake of simplicity , we defined @xmath93 and @xmath94 .
the whitened form of this channel is @xmath95where @xmath96 is the whitened noise and @xmath97 is the whitening matrix with @xmath98 being the covariance matrix of the noise in ( [ eq : multihop_af1 ] ) .
since it can be shown that @xmath99 , @xmath100 can be neglected in the dmt analysis and the af channel is equivalent to the mimo channel defined by the following matrix @xmath101the rest of the section is devoted to the dmt analysis of this channel .
[ def : rp ] let @xmath102 , @xmath103 , be @xmath0 independent complex gaussian matrices with i.i.d .
@xmath54 entries . a @xmath58 _ rayleigh product _
( rp ) channel is a @xmath104 mimo channel defined by @xmath105 where @xmath106 is the channel matrix ; @xmath107 is the transmitted signal with normalized power , i.e. , @xmath108 ; and @xmath109 is the received signal ; @xmath110 is the awgn ; @xmath57 is the snr per receive antenna .
@xmath58 is called the _ dimension _ of the channel and @xmath0 is called the _ length _ of the channel .
while this channel model has been studied in terms of the asymptotic eigenvalue distribution in the large dimension regime @xcite , we are particularly interested in the fixed dimension case in the high snr regime . in this regime , we can define a more general rp channel as @xmath111 [ prop : pdf - general - rp ] the general rp channel is equivalent to * a @xmath58 rp channel , if all the matrices @xmath112 s are square and their singular values satisfy @xmath113 , @xmath114 ; * a @xmath115 rp channel , with @xmath116 being the rank of the matrix @xmath112 , if the matrices @xmath112 s are constant .
see appendix [ sec : proof - pdf - general - rp ] .
hence , we can consider the rp channel from definition [ def : rp ] without loss of generality . recall that @xmath32 is the ordered version of @xmath33 with @xmath117 and @xmath118 .
[ thm : dmt - rp ] the dmt of a rp channel @xmath58 is a piecewise - linear function connecting the points @xmath119 , @xmath120 , where @xmath121 with @xmath122 the dmt depends on the `` near zero '' probability of the singular values of channel matrix . while this probability for the given product matrix is intractable , we can characterize it by induction on the length @xmath0 .
the main idea is that , conditioned on a given product matrix @xmath123 , @xmath124 is gaussian whose singular distribution is tractable .
see appendix [ sec : proof - thm - dmt - rp ] for details .
the following corollaries are given without proofs .
[ coro : per - inv ] the dmt of a rp channel depends only on the ordered dimension @xmath32 .
[ coro : increase ] the dmt is monotonic in the following senses : * if @xmath125 , then @xmath126 * if @xmath127 , then @xmath128 [ coro : sym ] when @xmath129 , we have @xmath130 where @xmath131 and @xmath132 .
corollary [ coro : per - inv ] implies that rp channels with the same ordered dimension belong to the same dmt equivalent class . in the following ,
a precise characterization of the dmt class is obtained . before that
, we need the following definitions . a @xmath133 rp channel
is said to be a _ reduction _ of a @xmath58 rp channel if 1 ) they are equivalent , 2 ) @xmath134 , and 3 ) @xmath135 .
in particular , if @xmath136 , then it is called a _ vertical reduction_. similarly , if @xmath137 $ ] , it is a _ horizontal reduction_. @xmath138 is said to be a _
minimal form _ if no reduction other than itself exists .
similarly , it is called a _ minimal vertical form _
( respectively , _ minimal horizontal form _ ) if no vertical ( respectively , horizontal ) reduction other than itself exists .
a rp channel is said to have _ order _
@xmath139 if its minimal form is of length @xmath140 .
[ thm : reduction ] a @xmath58 rp channel can be reduced to a @xmath141 channel if and only if @xmath142 in particular , it can be reduced to a rayleigh channel if and only if @xmath143 see appendix [ sec : proof - thm - reduction ] .
[ coro : reduc ] the channel order @xmath139 is the minimum integer such that ( [ eq : reduction - cond ] ) is satisfied . the minimal horizontal form is the minimal form @xmath138 and the minimal vertical form is @xmath144 with @xmath145 for instance , the minimal form of a @xmath146 rp channel is @xmath147 , i.e. , a @xmath148 or @xmath149 rayleigh channel .
[ thm : minimal ] the dmt equivalent class is _ uniquely _ identified by the minimal form , i.e. , two rp channels are equivalent _ if and only if _ they have the same minimal form .
see appendix [ sec : proof - thm - minimal ] . in order to interpret the closed - form dmt of theorem [ thm : dmt - rp ] , we derive an equivalent recursive form as shown in the following theorem .
[ thm : recursive ] the dmt @xmath150 defined in ( [ eq : dk ] ) can be alternatively characterized by @xmath151 see appendix [ sec : proof - thm - recursive ] .
a new interpretation of the dmt is as follows .
_ let us consider a multi - layer network of dimension @xmath58 . then
, @xmath150 is the minimum `` cost '' to limit the `` network flow '' between the source and the destination to @xmath85 ( the flow-@xmath85 event ) . in particular , the maximum diversity @xmath152 is the `` disconnection cost '' .
_ now , we can apply the new interpretation to the results of theorem [ thm : recursive ] .
first , @xmath153 says that the most efficient way to limit the flow to @xmath85 is to keep a @xmath154 channel fully connected and to disconnect the @xmath155 residual channel , as shown in fig .
[ fig : interp1 ] .
then , @xmath156 suggests that in order to disconnect a @xmath58 channel , if we allow for @xmath157 flows from the source to some layer @xmath43 , then the @xmath158 channel from the @xmath157 `` ends '' of the flows at layer @xmath43 to the destination must be disconnected ( fig .
[ fig : interp2 ] ) . obviously , the most efficient way is such that the total cost is minimized with respect to @xmath157 . finally , the flow-@xmath85 event takes place when both the flow-@xmath157 ( @xmath159 ) event in the @xmath160 channel and the flow-@xmath85 event in the @xmath161 channel happen at the same time .
we can easily verify that @xmath162 is equivalent to @xmath163 . also note that @xmath156 and @xmath164 hold for any layer @xmath43 , which guarantees the coherence of the interpretation .
the recursive characterization sheds lights on the typical outage event of the rp channel . in the trivial case of @xmath165 ( the rayleigh channel ) , the typical and only way for the channel to be in outage at multiplexing gain @xmath61 approaching to zero
is that all the @xmath166 paths are bad , i.e. , all channel gains are close to zero .
and the disconnection cost is @xmath166 . in the non - trivial cases
( @xmath167 ) where channels are concatenated , there are several types of outage event .
each type is numbered by the index @xmath157 in ( [ eq : interp2 ] ) and ( [ eq : interp3 ] ) .
the cost of the type-@xmath157 event is given by @xmath168 for a certain @xmath157 .
hence , the typical outage event is the one with the minimum cost and it does not necessarily happen when one of the sub - channels being totally bad ( @xmath169 or @xmath170 ) .
mismatch _ of two partially bad sub - channels can also cause outage .
this phenomenon will be detailed later on . from the equivalent channel matrix ( [ eq : afeqchannel ] ) and proposition [ prop : pdf - general - rp ] , the af channel is equivalent to a @xmath171 rp channel . , @xmath114 . to this end
, it is enough to modify the matrices as @xmath172 where @xmath173 is a constant independent of @xmath57 .
note that the @xmath174 is only for theoretical proof and is not used in practice , since we can always set @xmath174 a very large constant but independent of @xmath57 .
] therefore , the dmt of the af channel is @xmath175 from the results of section [ sec : rp ] , several interesting implications on the af scheme with respect to the dmt are summarized below . and @xmath176 symmetric rp channels.,scaledwidth=45.0% ]
* interchanging layers does not influence the dmt . *
the maximum diversity of the af scheme is lower- and upper - bounded as @xmath177 which is obtained via the monotonicity from corollary [ coro : increase ] .
we have set @xmath178 for the upper bound and @xmath179 for the lower bound .
the upper bound shows that there exists a virtual @xmath166 `` bottleneck '' channel that limits the af scheme and that it is not necessarily one of the sub - channels . on the other hand ,
the lower bound is always strictly larger than half the upper bound and is independent of the number of hops @xmath0 . in the symmetric case ( corollary [ coro : sym ] )
, we observe that the dmt degrades with @xmath0 only when @xmath180 and that we have @xmath181 for @xmath182 .
the observation can also be deduced from theorem [ thm : reduction ] applying which we infer that the order of any symmetric rp channel with @xmath183 is @xmath184 . this non - trivial lower bound is somewhat anti - intuition , since it means that at this point introducing extra fading hops does not degrade the diversity any more .
an example illustrating the dmt of the @xmath185 and @xmath186 rp channels of different lengths is in fig .
[ fig : rayleighproduct_n5 ] .
* if one could increase the number of antennas at each relay layer without any constraint , then intuition tells us that the af channel could be reduced to a @xmath65 point - to - point rayleigh mimo channel and the diversity order is @xmath187 .
the relay layers `` disappear '' .
the intuition has been confirmed in @xcite in the single - layer case with the capacity results .
here , the result in theorem [ thm : reduction ] indicates that this happens when there are exactly @xmath188 antennas at each relay layers from the diversity point of view . further increasing the number of antennas is not necessary in the dmt sense .
on the other hand , if the number of available antennas is fixed , then corollary [ coro : reduc ] provides , through the minimal vertical form , the minimum numbers of antennas at each layer to achieve the diversity that could be achieved when all antennas were used . in both cases ,
our results yield simple guidelines to minimize the number of relay antennas ( also the number of relays in general ) without loss of optimality of the dmt . in the same way
, the number of transmit antennas at the source terminal can also be reduced to lower the coding complexity .
a numerical example is given in section [ sec : nr ] .
a simple comparison between the dmt of the af scheme and the cut - set bound ( [ eq : dmt - ub ] ) is carried out as follows .
first , the af scheme is multiplexing optimal and achieves the maximum multiplexing gain @xmath189 of the channel .
then , since @xmath190 the diversity upper bound is generally not achievable by the af scheme for integer multiplexing gain @xmath85
. in particular , the best diversity gain of the af scheme is @xmath191 , while the upper bound is @xmath192 .
finally , for any non - integer multiplexing gain , say @xmath193 , @xmath79 is minimum of linear functions and thus concave , while @xmath194 is linear .
the comparison shows that the _ bottleneck _ of the channel is always one of the hops ( inter - layer sub - channels ) , while the bottleneck of the af scheme is the virtual @xmath166 channel that does not correspond to any physical sub - channel in most cases .
the following remark states the necessary and sufficient condition for the af scheme to achieve the maximum diversity .
[ thm : cond_af ] the af scheme achieves the diversity upper bound @xmath195 if and only if it can be reduced to the bottleneck of the channel , i.e. , @xmath196 where @xmath197 is such that @xmath198 is minimized .
this condition is very stringent .
it means that the two layers with minimum numbers of antennas must stand one next to the other and that the other layers must have a large number of antennas .
moreover , note that the af scheme achieving the maximum diversity does not necessarily mean that it achieves @xmath79 for all @xmath61 . in order to achieve the diversity upper bound ,
intuitively , one should assure that the end - to - end channel is good if each sub - channel is good .
however , this property does not hold for the af scheme that suffers from the _ mismatch _ of adjacent sub - channels .
a concrete example is as follows . in the symmetric two - hop channel with @xmath199 ( fig .
[ fig : ex ] ) , the diversity order of the af scheme is @xmath200 while the upper bound is @xmath201 .
note that the af channel is in outage if the product channel @xmath202 is bad , i.e. , all the singular values of @xmath202 are close to zero .
this probability can be decomposed as @xmath203 where we can verify that the first probability is essentially the probability of the sub - channel @xmath204 being bad and that the second one is essentially the probability of @xmath202 being bad conditioned on the event that @xmath204 is not bad .
as we know , the first probability decays with @xmath57 as @xmath205 . to find out the snr exponent of the second probability , we assume without loss of generality that the vector @xmath206 is strong enough ( since @xmath204 is not bad ) , as shown in fig .
[ fig : ex_canon ] .
now , we apply an orthogonal basis change from the canonical basis to @xmath207 and get the equivalent channel in fig . [
fig : ex_ortho ] . the basis change being an unitary transformation that is independent of the remaining parts of the channel
, it does not affect the statistics of the rest of the channel . as shown in fig .
[ fig : ex_ortho ] , the channel is bad if the three independent edges crossed by the `` minimum cut '' are bad .
the probability for the latter to happen decays as @xmath208 , from which we conclude that the outage probability scales as @xmath209 .
therefore , the mismatch between @xmath210 and @xmath204 is the dominating outage event and the end - to - end diversity of the @xmath211 channel with af scheme is @xmath200 , as compared to @xmath201 given by the cut - set bound .
the naive af scheme presented above can be seen as a space - only processing . in the point - to - point mimo channel
, it has been shown that space - only coding schemes ( e.g. , the v - blast scheme @xcite ) are suboptimal in diversity .
similarly , the af scheme , as a space - only relaying scheme , does not achieve the maximum diversity in the multihop channel due to the mismatch between adjacent sub - channels .
the clue is , just like the space - time codes achieve the maximum diversity in the point - to - point channel , space - time relay processing should be utilized in order to exploit the maximum distributed diversity in the multihop channel .
the first attempt was made in @xcite with a distributed space - time coding scheme . in this scheme ,
each relay performs temporal random unitary transformation on the received signal from the source in an independent way .
then , they forward the transformed signal at the same time as if they were jointly sending a space - time codeword .
the spatial correlation of the codewords is due to the fact that the received signal at different relays is from the same source . the temporal correlation , on the other hand
, is brought in by the temporal transformation . in their setting where a single layer of relays and single - antenna terminals are assumed , the maximum diversity of the channel is achieved . this scheme is then generalized to the multi - antenna case @xcite with structured algebraic transformations @xcite instead of random transformations .
however , generalization of such schemes to the multihop case is difficult and the dmt is hard , if not impossible , to calculate . in the following , we present a different approach to introduce the temporal processing .
this approach does not depend on the dimension of the channel and thus suitable for multihop channels of arbitrary number of hops .
the idea is to partition the relays in each layer .
based on the partition , the relays coordinately amplify - and - forward the received signal in a pre - assigned mode that changes periodically , which creates a parallel channel in the time domain .
such partition is thus called _
parallel partition_. we show that the mismatch is removed in this way and the diversity upper bound is achieved . in order to describe a parallel partition , some definitions and notations
are needed .
a _ supernode _
@xmath212 is a set of indices corresponding to a subset of antenna nodes in the same layer .
the cardinality of @xmath212 is called the _ size _ of the supernode . an _
edge _ is defined as the channel between two antenna nodes from adjacent layers .
an _ af path _ is defined as a sequence of consecutive supernodes from the source to the destination , each supernode performing the af operation .
a _ parallel partition _
@xmath213 is defined as a set of af paths .
the number of af paths in a partition is called the _ partition size _ and denoted by @xmath214 .
an _ independent parallel partition _ is defined as a parallel partition where any two different af paths do not share common edges .
an independent partition of maximum size is called a _
maximum partition_. an independent partition that achieves the maximum diversity @xmath195 is called a _ full diversity partition_. [ lemma : div ] for any fading channel defined by @xmath204 , we have @xmath215 where @xmath216 is the dmt of the channel
. see appendix [ sec : proof - lemma - div ] .
[ lemma : div - para ] let us consider a set of @xmath2 independent parallel af channels @xmath217 where @xmath218 s are statistically independent .
then , the diversity order of the parallel channel is the sum of the diversity order of the individual af sub - channels .
furthermore , if all the sub - channels have the same dmt @xmath219 , then the dmt of the parallel channel is @xmath220 .
see appendix [ sec : proof - lemma - div - para ] .
the independent parallel partition is accomplished in two steps : 1 ) partition each layer into supernodes , and 2 ) find @xmath2 independent af paths connecting the supernodes .
each af path defines a relaying mode : only the supernodes in this path are on and perform the af operation .
assume that a data frame of length @xmath221 is transmitted .
then , the relays change the relaying mode every @xmath222 symbol times .
we call it a _ parallel af scheme _ , since the end - to - end channel is equivalent to a parallel af channel in the time domain .
note that the af scheme is the trivial partition of size @xmath42 with a single `` wide '' af path .
as shown in remark [ thm : cond_af ] , the trivial partition achieves the maximum diversity only when the wide af path satisfies the conditions in ( [ eq : cond_af ] ) .
this being impossible in general , the parallel partition aims to find independent `` narrow '' paths each one of which satisfies the conditions in ( [ eq : cond_af ] ) . and if the number of independent paths is large enough , then the maximum diversity order can be achieved according to lemma [ lemma : div - para ]
. intuitively , the narrower the af path is , the easier the conditions ( [ eq : cond_af ] ) are to be satisfied . in the extreme case with the narrowest af path @xmath223 ,
all conditions in ( [ eq : cond_af ] ) are met .
[ lemma : maxi - par ] in a @xmath224 multihop channel , there are exactly @xmath195 independent single - antenna af paths .
first , the converse is true , since otherwise , at least two af paths share the same edge in the bottleneck of the channel .
then , the achievability is shown by construction : we connect the multihop channel in such a way that 1 ) there are @xmath195 incoming and outgoing edges for each intermediate layer , 2 ) the number of the incoming and outgoing edges is the same for each antenna node ( say , in layer @xmath43 ) and can be either @xmath225 or @xmath226 .
this partition contains @xmath195 independent @xmath223 af paths each one of which has diversity @xmath42 .
the lemma implies that the maximum partition is of size @xmath195 . from lemma [ lemma : div - para ] and
[ lemma : maxi - par ] , the following proposition is immediate . with the parallel af scheme ,
the dmt @xmath227 is always achievable in a multihop channel of arbitrary number of hops and antennas .
the dmt ( [ eq : dmt_rateone ] ) is simply achieved by applying the parallel af scheme with the maximum partition . in this case , @xmath195 single - input - single - output ( siso ) parallel sub - channels are generated , from which we have the dmt ( [ eq : dmt_rateone ] ) .
while the maximum diversity gain is achieved , this scheme only exploits one out of @xmath189 degrees of freedom of the channel .
this is due to the siso nature of the af paths in the maximum partition . in order to improve the achievable multiplexing gain ,
we need parallel partitions with wider af paths .
meanwhile , we still want the maximum diversity , which requires that the af paths should not be too wide .
the following theorem states a necessary and sufficient condition for an independent parallel partition to achieve the maximum diversity .
[ thm : cond_nece ] let the @xmath228 channel be any bottleneck of the @xmath58 multihop channel and @xmath213 be an independent partition of size @xmath2 .
then , @xmath213 is a full diversity partition if and only if 1 ) @xmath229 with @xmath230 and @xmath231 , and 2 ) we have @xmath232 where @xmath233 is the vector of numbers of antennas of the @xmath85th af path . to prove the theorem , let us assume there are respectively @xmath234 and @xmath235 supernodes in the layer @xmath197 and layer @xmath236 , and define @xmath237 .
then , we must have exactly @xmath238 connections between the supernodes from these two layers .
the diversity of the partition @xmath239 is upper - bounded @xmath240note that @xmath241 is achieved if and only if both ( [ eq : tmp987 ] ) and ( [ eq : tmp876 ] ) have equality .
thus , we must have both ( [ eq : cond_nece ] ) according to the conditions in ( [ eq : cond_af ] ) and @xmath229 at the same time .
now , finding full diversity partitions with minimum size is an optimization problem that minimizes the partition size @xmath214 subject to the constraint that @xmath213 must be an independent partition and satisfy the conditions given by theorem [ thm : cond_nece ] .
unfortunately , it remains an open problem for a general multihop channel .
the main difficulty lies in the lack of knowledge on the mathematical structure of the independent partitions for a general multihop channel .
nevertheless , the problem is solved in the two - hop case .
[ prop:2hop ] for a @xmath242 channel , the minimum size of a full diversity partition is @xmath243 see appendix [ sec : proof - prop-2hop ] .
it is achieved by partitioning the relay layer into @xmath2 supernodes of size @xmath244 or @xmath245 .
for example , the minimum partition size of the @xmath246 channel is @xmath27 as compared to the maximum partition size @xmath247 ; and each af path is a @xmath248 channel instead of a @xmath249 channel .
another example is the @xmath250 symmetric channel , where the minimum partition size is @xmath251 as compared to the maximum partition size @xmath252 ; each af path is a @xmath253 channel .
some words regarding the related previous works before proceeding further . in the relay channel with direct link and single layer of relays ,
the @xmath0-relay non - orthogonal af ( naf ) scheme @xcite divides the data frame into @xmath0 sub - frames , each one of which is relayed by one and only one relay . by creating a parallel naf channel , this scheme is optimal in diversity .
similar thought was shown in @xcite in the same channel setting with a different protocol called nd - raf scheme . removing the direct link from the channel setting ,
the scheme in @xcite becomes the parallel af scheme with the maximum partition in the single - antenna single - layer case . with the parallel af scheme , the maximum multiplexing gain of the channel
is achieved only when every af path in the partition achieves the maximum multiplexing gain @xmath254 . in the following ,
we propose a scheme that achieves both the maximum diversity gain and the maximum multiplexing gain .
let us consider an example first .
[ ex : ff222 ] the parallel channel @xmath255 in fig .
[ fig : ex_para ] has maximum diversity gain @xmath201 and multiplexing gain @xmath42 , while the parallel channel @xmath256 in fig .
[ fig : ex_ff ] has maximum diversity gain @xmath201 and multiplexing gain @xmath27 . in this example
, @xmath255 corresponds to the parallel af scheme based on the full diversity partition proposed by proposition [ prop:2hop ] .
however , it suffers from rate - deficiency , since both sub - channels are of rank @xmath42 .
an alternative is the channel @xmath256 shown in fig .
[ fig : ex_ff ] .
note that @xmath257 hence , we have @xmath258from which @xmath259 therefore , according to lemma [ lemma : div ] , they both achieve the maximum diversity gain @xmath201 except that @xmath256 has the maximum multiplexing gain @xmath27 as well .
this scheme is called the _
amplify - flip - and - forward _ ( aff ) s have been neglected for simplicity of demonstration . ]
scheme , or simply the _ flip - and - forward _ ( ff ) scheme . the intuition behind the ff scheme is as follows .
it has been shown that the mismatch between the two hops is the dominating outage event . now , suppose that @xmath260 is bad due to the bad `` angle '' between @xmath261 and @xmath262 both of which are not bad individually .
then , in the second sub - channel , an independent `` rotation '' matrix @xmath263 is used to change the angle . with high probability ,
the new angle is not bad and the mismatch is solved . in the light of the example , we generalize the scheme to arbitrary number of antennas and hops .
three steps are needed to describe the construction . 1 .
find a full diversity independent parallel partition @xmath213 of size @xmath2 .
the partition defines the intermediate supernodes in each layer .
we denote the supernodes in layer @xmath43 by @xmath264 with @xmath265 being the number of supernodes in layer @xmath43 . and
we define the flip matrices @xmath266 s as @xmath267 diagonal matrices with @xmath268 3 .
the ff scheme is composed of @xmath269 parallel sub - channels @xmath270 with @xmath271 where @xmath272 and @xmath273 in other words , the set of relays works in @xmath274 different flip modes , each one being identified by a sequence of flip modes of individual relay layers . and the mapping is effectuated by the functions @xmath275 .
the exact dmt of the ff scheme being difficult to obtain , we get a lower bound instead .
channel with different schemes.,scaledwidth=75.0% ] [ thm : dmt - ff ] the ff scheme constructed above achieves the following dmt @xmath276 see appendix [ sec : proof - theorem - dmt - ff ] .
we can verify that @xmath277 , that is , the maximum diversity of the channel is achieved . furthermore , the ff scheme is always better than the af scheme , especially at low multiplexing gain .
this can be explained by the intuition that the ff scheme solves the mismatch of adjacent hops using all possible combinations of flip modes of individual supernodes .
the equivalent end - to - end channel of the ff scheme can be bad only if at least one of the hops are bad .
the maximum diversity is thus achieved .
[ fig : dmt222 ] shows the dmt of different schemes in the channel of example [ ex : ff222 ] .
while the af and the parallel af schemes achieve respectively the extreme of maximum multiplexing gain @xmath278 and the extreme of maximum diversity gain @xmath279 , the ff scheme achieves both extremes .
the proposed ff scheme is constructed based on the flip matrices that are diagonal with @xmath280 entries .
in fact , it can be shown that a looser sufficient condition is for the matrices to be 1 ) diagonal , 2 ) linearly independent , and 3 ) of unit absolute value ( power constraints ) .
therefore , we can find infinitely many sets of `` flip '' matrices that satisfy the above conditions and they are all diversity optimal .
intuitively , if the matrices are too `` close '' , the ff scheme tends to the af scheme and the promised maximum diversity gain can be achieved only when the snr is very large .
this is translated into a poor power gain of the scheme .
hence , we should choose the matrices such that they are `` far '' from each other .
in this way , with high enough probability , any mismatch can be solved by at least one `` rotation '' and the maximum diversity can be obtained in relatively small snr .
however , what remains open is how to choose the distance metric between the rotation matrices . with independent partition ,
the total diversity is the sum of the diversity of each af path .
we also established some conditions that independent partitions must satisfy to achieve the maximum diversity . in the following ,
we investigate a particular case of non - independent partition .
let us consider a parallel channel defined by @xmath281 with @xmath282where the _ selection matrix _
@xmath283 is a @xmath267 diagonal matrix whose entries are zero except that @xmath284 .
the matrices @xmath218 s are not independent , since they share the common sub - channels @xmath285 and @xmath286 .
however , the rp channels @xmath287 s are independent for different @xmath85 s . despite the dependency between the sub - channels , we can obtain the diversity order of the parallel channel .
[ thm : non - ind ] the diversity order of the channel described above is @xmath288 we use the dmt interpretation given in section [ sec : recur ] to sketch the proof .
one possibility for the parallel channel @xmath281 to be in outage is that one of @xmath289 and @xmath290 is bad .
the diversity is either @xmath291 or @xmath292 .
another possibility is that both @xmath289 and @xmath293 are good and that @xmath281 turns out to be bad . without loss of generality ,
we assume that the flow from the source to layer @xmath48 is @xmath294 and that from layer @xmath295 to the destination is @xmath296 . and we call the outage event a _
type-@xmath297 _ event .
then , it can be shown that @xmath281 is equivalent to @xmath298 with @xmath299 and @xmath300 being gaussian matrices with i.i.d .
@xmath54 entries .
now , we must disconnect all the sub - channels in @xmath301 , which costs @xmath302 .
therefore , the total cost for the type-@xmath297 event is @xmath303 the typical outage event is the one that minimizes the above total cost . for @xmath304 , using ( [ eq : interp2 ] ) , we can show that the minimum total cost is @xmath305 .
similarly , @xmath306 is the minimum total cost for @xmath307 .
since both costs are smaller than @xmath291 and @xmath292 for the monotonicity ( corollary [ coro : increase ] ) , we proved the theorem .
note that with this particular partition at layer @xmath43 , we achieve a diversity order as if layer @xmath43 were clustered and the cooperative df scheme were used .
this result implies that one might achieve the maximum diversity with a partition of small size .
for example , the maximum diversity order of the @xmath308 channel is @xmath201 and all the full diversity independent partitions are of size @xmath309 , i.e. , eight @xmath310 sub - channels . with the non - independent partition described above , we get a couple of @xmath311 sub - channels , i.e. , size @xmath27 . since @xmath312 , the maximum diversity @xmath201 is achieved as well according to theorem [ thm : non - ind ] .
we can apply the ff scheme to the case of non - independent partition . then , in this example , the parallel channel is @xmath270 with @xmath313 where the flip matrix @xmath314 is a @xmath267 diagonal matrix whose entries are one except that @xmath315 if @xmath316 . the channels @xmath270 being a linear invertible transformation of @xmath281 , the generalized ff scheme achieves the diversity given by ( [ eq : dmax_nonind ] ) . with the nice parallel - channel structure , the ff scheme can be extended to various cases .
let us first consider the extension to the mimo relay channel with direct link and a single layer of @xmath0 relays . by applying directly the single - antenna naf scheme @xcite to the multi - antenna case , the source cooperates with one relay at a time .
this is equivalent to using the parallel af scheme in the source - relays - destination link .
the dmt lower bound is obtained in @xcite as @xmath317where @xmath318 is the dmt of the @xmath319 source - destination channel @xmath320 and each relay has @xmath251 antennas .
in fact , this lower bound can be improved to @xmath321by replacing the parallel af scheme in the source - relays - destination link with the ff scheme . comparing the second terms from ( [ eq : tmp9888 ] ) and
( [ eq : tmp9889 ] ) , the gain in diversity of the new scheme over the mimo naf is reflected by @xmath322where the inequality ( [ eq : tmp8788 ] ) becomes strict when @xmath251 is large .
the gain in multiplexing of the source - relays - destination link is obvious when @xmath251 is small , i.e. , @xmath323 . in this case , the ff scheme pools the relay antennas together to provide more degrees of freedom .
another extension is to the multiuser case .
let us take the multiple access channel as an example . for simplicity
, we assume that @xmath324 users try to communicate with the common destination through the same layers of relays .
we the ff scheme , we have an equivalent parallel multiple access channel with @xmath325where @xmath326 is similarly defined as in ( [ eq : tmp1111 ] ) with @xmath327note that only the first hop is distinct for different users . using the techniques of @xcite and the our results for the single - user ff scheme , it is possible to analyze the dmt of the ff scheme in the multiple access channel .
it is trivial to show that similar extension also holds for the broadcast channels with minor modifications .
in section [ sec : ccndf ] , it has been shown that the cooperative df scheme achieves the dmt cut - set bound in the clustered case . in this section
, we would like to study some alternative schemes , since it might be impossible or unnecessary for all the clusters to decode the source message in some cases .
the af and the cooperative df schemes can in fact be seen as two extremes of what we call the _ serial partition _ of multihop channels , defined as follows . a _ serial partition _
is defined by a set of layer indices @xmath328 with @xmath329 , each layer performing cooperative decoding - and - forward operation . with a serial partition ,
the multihop channel becomes a serial concatenation of @xmath330 af channels . as in ( [ eq : dmt - ub ] ) , the dmt of the multihop channel with any partition @xmath331 is easily derived as @xmath332where we defined @xmath333 . to get the maximum diversity gain ,
the question of _ when to decode _ has been answered earlier : when the conditions in ( [ eq : cond_af ] ) are not met .
another question is _ where to decode _
, i.e. , how to find the partition of minimum size that achieves a given diversity order .
[ prop : where_to_decode ] let us take @xmath334 and we succesively decide @xmath335 as the maximum integer in @xmath336 $ ] such that @xmath337 then , the decoding set @xmath338 defines the partition of minimum size that achieves a given diversity @xmath62 ( @xmath339 ) . from ( [ eq : dmt - mixed ] ) , it is easy to show that the proposed partition achieves diversity @xmath62 .
now , we would like to show that the size of the proposed partition is minimized .
to this end , it is enough to show that for any set @xmath340 of decoding points that achieves diversity @xmath62 , we have @xmath341 , @xmath87 .
this is obviously true for @xmath342 , since the diversity of the af channel degrades with the number of hops . by induction on @xmath43
, it is shown that @xmath343 because otherwise @xmath344 and the corresponding diversity of the af scheme can not be larger than @xmath62 according to the monotonicity of the dmt ( corollary [ coro : increase ] ) .
the proposition matches the intuition that we should only decode when we have to , in the diversity sense .
in other words , we allow for the degradation of diversity introduced by the af operation , as long as the resulting diversity is larger than the target @xmath62 . another option is to linear process the received signal at each cluster without decoding it . unlike the af scheme in the non - clustered case , where trivial antenna - wise normalization is performed
, we can run inter - antenna processing based on the available csi at the cluster .
with receiver csi at the relays , let us consider the following project - and - forward ( pf ) scheme . at layer @xmath43 ,
the received signal is first projected to the signal subspace spanned by the columns of the channel matrix @xmath345 .
the dimension of the subspace is @xmath346 , the rank of @xmath345 .
after the component - wise normalization , the projected signal is transmitted using @xmath346 ( out of @xmath44 ) antennas .
it is now clear that @xmath347 is actually composed of the @xmath346 columns of the previously defined @xmath348 , with @xmath349 .
more precisely , the @xmath350 is an orthogonal basis with @xmath351 .
we can rewrite @xmath352 with @xmath353 . for simplicity , we let @xmath354 be obtained by the qr decomposition @xcite of @xmath345 if @xmath355 and be identity matrix if @xmath356 .
the spirit of the pf scheme is not to use more antennas than necessary to forward the signal .
since the useful signal lies only in the @xmath346-dimensional signal subspace , the projection of the received signal provides sufficient statistics and reduces the noise power by a factor @xmath357 . in this case , only @xmath346 antennas are needed to forward the projected signal .
let us define @xmath358 . then , as in the af case , the pf multihop channel is equivalent to the channel defined by @xmath359 the following proposition states that receiver csi and inter - antenna processing do not improve the dmt of the af scheme .
[ thm : pf ] the pf scheme is equivalent to the af scheme . see appendix [ sec : proof - thm - pf ] .
while the pf and af have the same dmt , the pf outperforms the af in power gain for two reasons .
one reason is , as stated before , that the projection reduces the average noise power .
the other reason is that the accumulated noise in the af case is more substantial than that in the pf case .
this is because in the pf case , less relay antennas are used than in the af case .
since the power of independent noises from different transmit antennas add up at the receiver side , the accumulated noise in the af case `` enjoys ''
a larger `` transmit diversity order '' than in the pf case . on the other hand , if we could have receiver _ and
_ transmitter csi at the clusters , the dmt could be improved as shown by the following example .
[ ex : mf ] for a @xmath360 clustered multihop channel , the dmt cut - set bound can be achieved by linear processing within clusters if both transmitter and receiver csi are available at each cluster .
the optimum linear relaying scheme is defined by the processing matrices @xmath361 s with @xmath362 where we assume that @xmath363 is the singular value decomposition of @xmath49 .
the diagonal elements in the singular value matrix @xmath364 are in increasing order .
this scheme matches the adjacent hops by aligning the singular values in the same order .
it is then equivalent to the channel defined by @xmath365 , whose dmt can be shown to be as the @xmath366 rayleigh channel .
now , we need codes that actually attain the dmt promised by the studied relaying strategies . to this end , the construction of perfect stbcs @xcite for mimo channels is extended to the multihop relay channels .
the constructed codes are approximately universal @xcite . the relay clusters that perform the cooperative df operation partition the multihop channel into a series of @xmath330 mimo channels , say , @xmath367 with @xmath368 .
an obvious coding scheme that achieves the dmt is described as follows .
let @xmath61 be the target multiplexing gain .
first , the source terminal encodes the message of @xmath369 bits with a @xmath370 perfect stbc @xmath371 .
then , in a successive manner , layer @xmath335 tries to decode the message .
when a success decoding is assumed , the @xmath369 bits are encoded with a @xmath372 perfect stbc @xmath373 and forwarded .
we can show that as long as @xmath374 with @xmath375 the series of perfect stbcs @xmath376 can be found @xcite .
with the union bound , the end - to - end error probability is upper - bounded @xmath377where @xmath378 is the error probability of @xmath373 in the mimo sub - channel @xmath379 . since @xmath373 is dmt - achieving for any fading statistics ,
we have @xmath380from ( [ eq:170 ] ) and ( [ eq:176 ] ) , the dmt ( [ eq : dmt - mixed ] ) is achieved with coding delay @xmath381 . since the perfect stbcs are approximately universal @xcite , so is this coding scheme .
note that this scheme can be used for the af and pf schemes with @xmath382 . in the non - clustered case , the parallel af and the ff schemes are used .
note that both schemes share the common parallel mimo channel structure @xmath383where @xmath384 and @xmath2 is the number of the parallel sub - channels .
let @xmath385 be a code for the parallel channel .
a codeword is defined by a set of matrices @xmath386 with @xmath387 .
we define a parallel stbc with non - vanishing determinant ( nvd ) as follows .
let @xmath388 be an alphabet that is scalably dense , i.e. , for @xmath389 , @xmath390then , a parallel stbc @xmath385 is called a _
parallel nvd code _ if it 1 .
is @xmath388-linear is @xmath388-linear means that each entry of any codeword in @xmath385 is a linear combination of symbols from @xmath388 . ] ; 2 . has full symbol rate , i.e.
, it transmits on average @xmath391 symbols per channel use from the signal constellation @xmath388 ; 3 . has the nvd property , i.e. , for any pair of different codewords @xmath392 , @xmath393 with @xmath174 a constant independent of the snr .
we have the following result .
[ thm : dmt - para ] the parallel nvd codes are approximately universal over the parallel channel defined by ( [ eq : parallel ] ) .
see appendix [ sec : proof - thm - dmt - para ] . thus , to achieve the dmt of the parallel af and the ff schemes ,
it is enough to construct a parallel nvd codes .
several remarks are made before proceeding to the code construction .
the actual data rate of the nvd codes is controlled by the size of the alphabet @xmath388 and the symbol rate .
efficient decoding schemes ( e.g. , sphere decoding ) may not be implementable when the channel is under - determined or , alternatively speaking , rank - deficient in the sense that @xmath394 .
practical schemes include reducing the symbol rate while increasing the size of the alphabet @xmath388 .
this , however , does not guarantee the dmt - achievability . explicit parallel nvd codes for asymmetric parallel channel ( i.e. , @xmath395 for some @xmath396 ) being hard to construct algebraically , we focus on the symmetric case .
note that in the ff scheme , the equivalent parallel channel is always symmetric . in the parallel af scheme
, the numbers of transmit antennas of different sub - channels may be different .
however , the problem can be overcome by using the same number of antennas ( i.e. , @xmath397 ) .
the resulting parallel channel has at least the same dmt as the original channel .
nevertheless , an alternative code construction that is suitable for both symmetric and asymmetric parallel channels is provided in appendix [ sec : altercod ] for completeness . from a given parallel partition with size @xmath398 ,
the number of the parallel sub - channels @xmath2 is @xmath398 in the parallel af scheme , generally larger than @xmath398 in the ff scheme .
since the minimum coding delay is @xmath399 that grows linearly with @xmath2 , it grows at least linearly with @xmath398 .
moreover , the complexity of decoding can grow up to exponentially with @xmath2 if ml decoding is used .
that is why it is important to find partitions of small size @xmath398 .
a systematic way to construct nvd codes is the construction from cyclic division algebra ( cda ) . for more details on the concept
, the readers can refer to @xcite . in the following , we aim to construct the perfect symmetric parallel nvd codes with quadrature amplitude modulation ( qam ) constellations .
the generalization to hexagonal constellations is straightforward .
we start by the construction of nvd codes for mimo channels ( @xmath400 ) .
let @xmath401 be a cyclic extension of degree @xmath402 on the base field @xmath403 .
we denote @xmath404 the generator of the galois group @xmath405 .
let @xmath406 be such that @xmath407 are non - norm elements in @xmath408 .
then , we can construct a cda @xmath409 of degree @xmath402 .
each element in @xmath410 has the following matrix representation @xmath411where @xmath412 , @xmath87 .
since @xmath410 is a cda , we can show that @xmath413 $ ] and that the determinant is zero only when @xmath414 is a zero matrix .
thus , the nvd property is proved by considering that the difference matrix of each pair of codewords is in the form of @xmath414 .
it is usually desirable to get a stbc with good shaping . to this end
, we can impose the additional constraint that the vectorized codeword is a rotated version of a @xmath415 constellation , as known as the cubic constellation .
rotated constellations constructions from algebraic number fields are well - known now ( see , e.g. , @xcite for a comprehensive tutorial on this topic ) .
this can be made possible if 1 ) @xmath416 s in the matrix @xmath414 belong to some properly chosen ideal @xmath417 @xcite , and 2 ) @xmath418 ( see @xcite for a general method ) .
the thus - constructed nvd codes are well - known as the perfect stbcs .
the construction of parallel nvd codes is similar to the construction presented above .
first , we construct a cda in the same manner as the previous case by simply 1 ) replacing the base field @xmath403 by a new field @xmath420 , a galois extension of degree @xmath2 over @xmath403 ; 2 ) replacing the field @xmath408 by @xmath421 , a cyclic extension of degree @xmath402 over @xmath420 ( same @xmath422 as the previous case ) ; and 3 ) choosing @xmath423 such that @xmath407 are non - norm elements in @xmath424 .
we impose that @xmath425 .
note that the extension @xmath426 remains cyclic with the same galois group as @xmath405 ( fig .
[ fig : fields - extension ] ) .
thus , the constructed cda is @xmath427 .
now , let @xmath428 be the galois group of the extension @xmath429 and define @xmath430 where @xmath414 is the matrix representation of some element in @xmath410 and is in the form ( [ eq : xi ] ) .
now , we have @xmath431 that is in @xmath432 $ ] .
finally , we construct codewords @xmath433 in the form of @xmath434 with qam symbols and we can show that the difference matrix of a pair of different codewords is also in the form of @xmath434 with symbols in @xmath432 $ ] . the nvd condition ( [ eq : nvd ] ) is thus met .
similarly , the cubic shaping can be obtained with the same kind of conditions mentioned before .
an explicit code construction is provided in the following example .
[ ex : code_construction ] let us define @xmath435 .
then , we consider the base field @xmath436 , an extension of @xmath403 of degree @xmath437 and take @xmath438 .
we can verify that @xmath439 is a non - norm element in @xmath424 ( see appendix [ zeta2m - not - norm ] ) .
let @xmath440 and @xmath441 .
the ring of integers of @xmath442 is @xmath443\right\ } $ ] .
and the chosen ideal is principle , i.e. , @xmath444 with @xmath445 .
the matrix @xmath414 is given by @xmath446,\label{eq : golden-2}\ ] ] where @xmath447 $ ] .
we can show that the shaping property is satisfied and finally , this code is a perfect stbc for the parallel channel .
in this section , we present the numerical results on the proposed schemes .
the performance measures are either the outage probability or the symbol error rate probability versus the average received snr per bit .
the results are obtained with monte - carlo simulations .
the first example is to illustrate the impact of vertical reduction of multihop channels , as shown in fig .
[ fig : vertical_reduction ] . in a @xmath448 channel , the necessary antenna number @xmath449 from ( [ eq : minimum_number ] ) is @xmath42 and the minimal vertical form is thus @xmath249 .
we observe that , with the same diversity order @xmath42 , an asymptotic power gain of @xmath450 db is obtained by using only one relay antennas out of four , if the af scheme is used .
the gain is due to the fact that using more relaying antennas hardens of relayed noise . in the @xmath451 channel ,
the necessary number of antennas @xmath449 from ( [ eq : minimum_number ] ) is @xmath27 . as shown in fig .
[ fig : vertical_reduction ] , by restricting the number of relay antennas to @xmath27 , we have a @xmath452 channel and an asymptotic power gain of @xmath27 db is observed .
we can further reduce the number of transmit antennas to @xmath27 to get a @xmath453 channel . unlike the reduction of relay antennas , the reduction of transmit antennas does not provide any gain because it does not impact the relayed noise .
in contrast , it degrades the performance since the first hop @xmath454 is faded more seriously than the original first hop @xmath455 .
nevertheless , the @xmath453 channel is still better than the @xmath451 channel and is only @xmath456 db from the @xmath452 channel .
the coded performance of the @xmath451 channel is then studied fig . [
fig : coded_performance ] .
the diagonal algebraic space - time ( dast ) code @xcite can be used . as shown in fig .
[ fig : coded_performance ] , with the dast code , the symbol error rate performances of in the @xmath451 , @xmath452 and @xmath453 channels have exactly the same behavior as the outage performances of the channels do fig .
[ fig : vertical_reduction ] .
moreover , the reduction in the number of transmit antenna allows us to use the alamouti code @xcite ( the @xmath453 channel ) . as we can see in the figure , the alamouti code , besides the advantage of lower decoding complexity , outperforms all the dast codes
the potential benefits from the vertical reduction are thus highlighted .
then , we consider the parallel partition of two multihop channels : the @xmath457 and @xmath246 channels . the resulting aff scheme is compared to the af scheme in terms of both the outage probability and the symbol error rate . with the aff scheme , we create respectively four and two parallel sub - channels with two transmit antennas for the @xmath457 and @xmath246 channels . specifically , the aff scheme for the @xmath457 channel is based on a partition of four @xmath458 sub - channels and for the @xmath246 channel is a partition of two @xmath248 sub - channels . as shown in fig .
[ fig : outage_aff_af ] , the diversity order of the aff scheme for the @xmath457 ( respectively , @xmath246 channel ) is @xmath201 ( respectively , @xmath247 ) , as compared to that of the af scheme ( @xmath200 and @xmath459 , respectively ) .
the coded performance is also studied .
we apply the construction provided by example [ ex : code_construction ] to get perfect parallel stbcs for two and four sub - channels . as we can observe in fig .
[ fig : coded_aff_af ] , with the use of perfect codes , the symbol error rate performance has similar behaviors as the outage performance . the last example is a @xmath451 channel in the clustered case . through this example
, we would like to address the impact of `` where to decode '' on the end - to - end performance .
the all - af and all - df schemes correspond respectively to the case with no decoding relay cluster and that with two decoding relay clusters . with one decoding cluster , the choice is made between decoding at the first cluster and decoding at the second one .
as shown in fig .
[ fig : outage_3142 ] , the all - af scheme has diversity order two and the all - df scheme has diversity order @xmath200 as analytically expected . with only one decoding cluster ,
the diversity order is also predictable : diversity two in the single - antenna cluster and diversity 3 in the four - antenna cluster .
what is impressive in this example is that the two curves with different choices of decoding cluster joins the all - af and all - df curves respectively at high snr .
therefore , only one decoding cluster is enough to achieve good performance in this case . and
the decoding cluster should not be the single - antenna node .
the diversity of mimo multihop relay channels has been investigated in both the clustered and non - clustered cases .
our results showed that , in both cases , the maximum diversity gain and the maximum multiplexing gain of the multihop channel can be achieved . in the clustered case ,
the optimal scheme is cooperative decode - and - forward that achieves the upper bound on the diversity - multiplexing tradeoff of the channel . in the non - clustered case , the key to achieve the maximum diversity is space - time relay processing . our approach is to introduce temporal processing to the amplify - and - forward scheme by creating a parallel channel in the time domain .
we proposed a flip - and - forward that achieves both the maximum diversity and multiplexing gain of an arbitrary multihop channel in a distributed manner .
we also showed that the ff scheme can be easily extended to the multiuser case . with its low relaying and signaling complexity
, the ff scheme is suitable for wireless _ ad hoc _ networks .
approximately universal coding schemes have been proposed for all the relaying strategies studied in this work .
the followings are some preliminary results that are essential to the proofs .
[ lemma : cal - dmt ] consider a linear fading gaussian channel defined by @xmath204 for which @xmath460 is a function of @xmath461 , a vector of positive random variables .
then , the dmt @xmath462 of this channel can be calculated as @xmath463 where @xmath464 is the exponent of @xmath465 , @xmath466 is the outage event set in terms of @xmath467 and @xmath61 in the high snr regime , and @xmath468 is the exponential order of the pdf @xmath469 , i.e. , @xmath470 this lemma can be justified by ( [ eq : dmt - compact ] ) using laplace s method , as shown in @xcite . the @xmath471 random matrix @xmath472 is a ( central ) complex wishart matrix with @xmath251 degrees of freedom and covariance matrix @xmath100 ( denoted as @xmath473 ) , if the columns of the @xmath474 matrix @xmath204 are zero - mean independent complex gaussian vectors with covariance matrix @xmath100 .
[ lemma : eq - wishart ] the joint pdf of the eigenvalues of @xmath475 is identical to that of any @xmath476 if @xmath477 are the eigenvalues of @xmath478 .
apply the eigenvalue decomposition on @xmath100 and the result is immediate using the unitary invariance property @xcite of wishart matrices .
[ lemma : wishart ] let @xmath479 be a central complex wishart matrix @xmath473 , where the eigenvalues of @xmath100 are distinct are identical , we apply the lhospital rule to the pdf obtained , as shown in @xcite . ] and their ordered values are @xmath480 .
let @xmath481 be the ordered positive eigenvalues of @xmath479 with @xmath482 .
the joint pdf of @xmath461 conditioned on @xmath483 is p(|)= k_m , n ( _ 1 ) _ i=1^m _
i^m - n-1 _ i^n - m _
i < j^m , & [ eq : wishart :
n > m ] + g_m , n ( _ 2 ) _
i < j^m _ i < j^n ( _ i-_j ) , & [ eq : wishart : n < m ] where @xmath484 and @xmath485 are normalization factors ; @xmath486 denotes the absolute value of the determinant @xmath487 ; @xmath488_{i , j=1}^m$ ] and @xmath489 in the non - correlated case with @xmath490 , the joint pdf is @xmath491 now , let us define the _ eigen - exponents _ @xmath492 [ lemma : det ] where @xmath493 and @xmath494 please refer to @xcite for details .
[ lemma : detxi ] where @xmath495 and @xmath496 first , we have @xmath497 then , let us denote the determinant in the rhs of ( [ eq : lemmas : tmp1 ] ) as @xmath498 and we rewrite it as @xmath499where @xmath500 and the product term in ( [ eq : lemmas : tmp2 ] ) is obtained since @xmath501 for all @xmath502 .
let us denote the determinant in ( [ eq : lemmas : tmp2 ] ) as @xmath503 .
then , by multiplying the first column in @xmath503 with @xmath504 and noting that @xmath505 , the first column of @xmath503 becomes all @xmath42 .
now , by eliminating the first @xmath506 `` @xmath42 ' 's of the first column by subtracting all rows by the last row as in ( [ eq : lemmas : tmp3 ] ) and ( [ eq : lemmas : tmp2 ] ) , we have @xmath507 . by continuing reducing the dimension
, we get @xmath508_{i , j=1}^n \prod_{i=1}^{n+1}\mu_i^{m - n-1}\prod_{i = n+2}^m \mu_i^{m - i}\\ & \quad \cdot\prod_{i=1}^n\prod_{j = n+1}^m \left(1-e^{-\lambda_i/\mu_j}\right ) \end{split}\ ] ] from which we prove the lemma , by applying ( [ eq : detexp ] ) . with the preceding lemmas , we have the following lemma that provides the asymptotical pdf of @xmath467 conditioned on @xmath509 in the high snr regime .
[ lemma : pdfrayleighcond ] where @xmath510 and @xmath511 let us replace @xmath512 and @xmath513 in ( [ eq : wishart : n > m ] ) and ( [ eq : wishart : n < m ] ) using the results of lemma [ lemma : det ] and lemma [ lemma : detxi ] .
then , by applying variable changes as done in @xcite , ( [ eq : expcond ] ) can be obtained after some elementary manipulations . when @xmath490 , i.e. , @xmath514 , the joint pdf of @xmath467 is found in @xcite as shown in the following lemma .
[ lemma : pdfrayleigh ] with @xmath515 .
this lemma can be justified either by using ( [ eq : wishart : id ] ) or by setting @xmath516 in ( [ eq : expcond ] ) .
[ lemma : invariance - asymp ] let @xmath517 be any @xmath474 random matrix and @xmath518 be any @xmath519 non - singular matrix whose singular values satisfy @xmath520 .
define @xmath482 and @xmath521 .
let @xmath522 and @xmath523 be the ordered singular values of @xmath517 and @xmath524 , then , we have @xmath525
[ prop : asympt - pdf ] let us denote the non - zero ordered eigenvalues of @xmath526 by @xmath527 with @xmath528 .
then , the joint pdf of the eigen - exponents @xmath467 satisfies where @xmath529 with @xmath530 s defined by ( [ eq : ci ] ) . from lemma [ lemma : cal - dmt ]
, we can derive the dmt with the following optimization problem @xmath531 with @xmath532 being the outage region .
note that @xmath530 is decreasing and @xmath533 is increasing with respect to @xmath43 .
then , the proof of theorem [ thm : dmt - rp ] is immediate .
now , what remains is the proof of proposition [ prop : asympt - pdf ] .
the following lemma will be needed in the proof .
[ lemma : ci ] let @xmath534 $ ] , @xmath535 , be @xmath0 consecutively joint intervals with @xmath536 , @xmath537 , and @xmath538 then , we have @xmath539 @xmath530 defined by ( [ eq : ci ] ) is the minimum of @xmath0 sequences corresponding to the @xmath0 values of @xmath85 .
it is enough to show that each of the @xmath0 sequences dominates in a consecutive manner .
we omit the details here
. the proof will be by induction on @xmath0 . from lemma [ lemma : pdfrayleigh ] ,
the proposition is trivial for @xmath165 .
suppose the proposition holds for some @xmath0 and @xmath540 , we would like to show that it is also true for @xmath541 and @xmath542 . for simplicity , the `` primed '' notations ( e.g. , @xmath543 , @xmath544 , @xmath545 , @xmath546 , @xmath547 , etc . ) will be used for the respective parameters of @xmath548 .
note that @xmath549 for a given @xmath550 , since @xmath551 . according to lemma [ lemma : eq - wishart ]
, the pdf of the eigenvalues @xmath552 of @xmath553 is identical to that of @xmath554 .
hence , the pdf of @xmath543 can be obtained as the marginal pdf of @xmath555 @xmath556where ( [ eq : tmp31 ] ) comes from lemma [ lemma : pdfrayleighcond ] and our assumption that ( [ eq : p - malpha ] ) holds for @xmath0 , with @xmath557being the feasible region ; the exponent @xmath558 in ( [ eq : tmp32 ] ) is @xmath559with @xmath560 . from ( [ eq : expcond ] ) and ( [ eq : ea ] ) , @xmath561it remains to show that @xmath562 with @xmath563by solving the optimization problem ( [ eq : minprob ] ) , which is accomplished in the rest of the section .
we need to distinguish three cases , according to how the value of @xmath564 affects the ordered dimension @xmath545 . in this case , we have @xmath566 .
minimization of @xmath567 of ( [ eq : jointexp ] ) with respect to @xmath467 can be decomposed into @xmath568 minimizations with respect to @xmath569 successively , i.e. , @xmath570 .
we start with @xmath571 . from ( [ eq : feasibility ] ) , the feasible region of @xmath571 is @xmath572 .
since the only @xmath571-related term in ( [ eq : jointexp ] ) is @xmath573 and @xmath574 for @xmath575 , we have @xmath576 . now
, suppose that the minimization with respect to @xmath577 is done and that we would like to minimize with respect to @xmath578 .
for @xmath578 , @xmath579 , we set the initial region as @xmath580 in which we have @xmath581 .
the feasibility conditions in ( [ eq : feasibleregion ] ) require that @xmath578 must not go right across @xmath582 .
the only choice is therefore to go to the left .
each time @xmath578 goes across an @xmath583 from the right to the left , @xmath584 increases by @xmath585 , which increases the coefficient of @xmath583 by @xmath42 and decreases the coefficient of @xmath578 by @xmath42 .
it can be shown that , to minimize the value of @xmath567 with respect to @xmath578 , @xmath578 is allowed to cross @xmath583 only when the current coefficient of @xmath578 in ( [ eq : jointexp ] ) is positive . in ( [ eq : jointexp ] ) is positive , decreasing @xmath533 decreases @xmath567 .
] so , @xmath578 stops moving only in the following two cases : 1 ) it hits the left extreme , @xmath41 ; and 2 ) its coefficient achieves @xmath41 when it is in the interval @xmath586 $ ] for some @xmath587 .
either case , @xmath578-related terms are gone and what remain are the @xmath583 s `` freed '' by @xmath578 from @xmath588 .
same reasoning applies to @xmath578 for @xmath589 , except that the initial region is set to @xmath590 .
therefore , the optimization problem can be solved by counting the total number of freed @xmath583 s . as shown in fig .
[ fig : find - ci1 ] , when @xmath157 is small , the initial coefficient of @xmath578 is large and thus @xmath578 can free out @xmath591 .
we have @xmath592 , which corresponds to the first stopping condition . for large @xmath157 ,
the initial coefficient of @xmath578 is not large enough and only @xmath593 is freed , which corresponds to the second stopping condition . with the above reasoning
, we can get @xmath594 from ( [ eq : tmp48 ] ) and ( [ eq : ci ] ) , we get @xmath595 and @xmath596now , @xmath558 can be obtained .
a more careful analysis reveals that it is always satisfied with the described procedure .
] from fig .
[ fig : find - ci1 ] @xmath597where ( [ eq : tmp89 ] ) is from ( [ eq : gi_inv1 ] ) and the fact that @xmath598 , @xmath599 , @xmath600 ; ( [ eq : tmp54 ] ) can be derived from lemma [ lemma : ci ] , since @xmath601 and therefore the term @xmath602 in ( [ eq : tmp54 ] ) is dominated by @xmath603 for @xmath604 and by @xmath605 for @xmath606 , corresponding to the two terms in ( [ eq : tmp89 ] ) , respectively . in this case
, we have @xmath608 and @xmath609 . from ( [ eq : jointexp ] ) ,
@xmath610 since @xmath611 , @xmath612 , the minimization of @xmath613 with respect to @xmath467 is in exactly the same manner as in the previous case .
therefore , @xmath558 can be obtained from fig .
[ fig : find - ci2 ] with @xmath594 in the same form as ( [ eq : gi ] ) @xmath614 as in the previous case , we have @xmath608 and the same @xmath613 as defined in ( [ eq : jointexp2 ] ) . without loss of generality
, we assume that @xmath616 for some @xmath617 $ ] ( we set @xmath618 ) . then , we have @xmath619and @xmath620unlike the previous case , @xmath621 is not always positive .
let @xmath622 be the smallest integer such that the coefficient @xmath621 of @xmath578 in ( [ eq : jointexp2 ] ) is zero .
it is obvious that for @xmath623 , @xmath624 .
hence , we have @xmath625where the second term is from fig .
[ fig : find - ci3 ] .
furthermore , we can show that @xmath626 , since @xmath627 .
therefore , we get @xmath628now , we would like to show that the coefficient of @xmath583 in ( [ eq : tmp56 ] ) coincides with @xmath629 .
first , for @xmath630 , @xmath631 and lemma [ lemma : ci ] implies that @xmath632 then , for @xmath633 , we have @xmath634hence , @xmath635 where ( [ eq : tmp112 ] ) is from ( [ eq : tmp98 ] ) and ( [ eq : tmp99 ] ) . finally , for @xmath636 , let us rewrite @xmath637 . since @xmath638 , @xmath639
, we have @xmath640from which we have @xmath641 $ ] and @xmath642the proof is complete .
let @xmath643 denote the vector of the eigen - exponents of a matrix @xmath517 as previously defined .
to prove the first case , we use induction on @xmath0 .
suppose that it is true for @xmath0 , which means that the joint pdf of @xmath644 is the same as that of @xmath645 .
furthermore , we know by lemma [ lemma : invariance - asymp ] that @xmath646 .
same steps as ( [ eq : tmp31])([eq : tmp32 ] ) complete the proof . to prove the second statement , we perform a singular value decomposition on the matrices @xmath112 s and then apply the first statement .
let @xmath647 what we should prove is that @xmath648
if and only if ( [ eq : reduction - cond ] ) is true . to this end , it is enough to show that @xmath649if and only if @xmath650 , that is , @xmath651 , and then apply the result successively to show the theorem . note that we need lemma [ lemma : ci ] to eliminate the minimization in ( [ eq:010 ] ) .
the detailed proof is omitted here .
the direct part of the theorem is trivial . to show the converse ,
let @xmath652 and @xmath653 be the two concerned minimal forms .
in addition , we assume , without loss of generality , that @xmath654 with @xmath655 and @xmath656 . now , let us define @xmath657 with @xmath530 defined in ( [ eq : ci3 ] ) .
it can be shown that @xmath324 intervals are non - trivial with @xmath658 , @xmath659 .
the values of @xmath660 s are in the following form @xmath661 same arguments also apply to @xmath32 with @xmath662 and @xmath663 , etc .
it is then not difficult to see that to have exactly the same @xmath660 s ( thus , same @xmath664 s ) , we must have @xmath665 and @xmath666that is , the same minimal form .
to prove the theorem , we will first show the following equivalence relations : @xmath667 the direct parts of @xmath668 , @xmath669 , and @xmath670 are immediate since the rhs are particular cases of the left hand side ( lhs ) . to show the reverse part of ( a )
, we rewrite @xmath671 where @xmath672 is used twice in ( [ eq : tmp01 ] ) and ( [ eq : tmp03 ] ) ; @xmath673 is used in ( [ eq : tmp02 ] ) . as for ( b ) , if @xmath674 holds , then @xmath675which proves @xmath676 . by continuing the process , we can show that @xmath677 is true for all @xmath43 , provided @xmath674 holds . through @xmath668 and @xmath669
, one can verify that the lhs of @xmath678 is equivalent to the rhs of @xmath668 of which the rhs of @xmath678 is a particular case .
hence , the direct part of @xmath678 is shown . the reverse part of ( c )
can be proved by induction on @xmath0 . for @xmath679 , @xmath680
can be shown explicitly using the direct characterization ( [ eq : dk ] ) .
now , assuming that @xmath680 for non - ordered @xmath33 , we would like to show that @xmath681 holds .
let us write @xmath682where the permutation invariance property is used in ( [ eq : tmp001 ] ) ; @xmath674 is used in ( [ eq : tmp002 ] ) since we assume that @xmath680 is trues ; @xmath683 and @xmath684 can be permuted according to @xmath685 .
finally , we should prove the reverse part of ( d ) , i.e. , @xmath686provided that @xmath680 holds for minimal @xmath33 .
if @xmath33 is not minimal , then showing ( c ) is equivalent to showing @xmath687where @xmath139 is the order of @xmath33 with @xmath688 .
therefore , we should show that the minimum is achieved with @xmath169 . according the direct characterization ( [ eq : dk ] ) , this is true only when @xmath689 .
let us rewrite @xmath690 as @xmath691since @xmath692 is always true according to the reduction theorem , we have @xmath693 . the rest of this section is devoted to proving that ( [ eq : tmp004 ] ) holds for minimal @xmath33 .
now , we restrict ourselves in the case of minimal and ordered @xmath33 , i.e. , we would like to prove @xmath694since @xmath695 , the optimal @xmath157 is in the interval @xmath696 $ ] .
now , showing ( [ eq : tmp006 ] ) is equivalent to showing @xmath697which , after some simple manipulations , is reduced to @xmath698where we set @xmath699 for simplicity of notation .
obviously , the minimum of the rhs of ( [ eq : tmp007 ] ) is achieved with such @xmath700 that @xmath701let us decompose @xmath700 as @xmath702 with @xmath703 $ ] .
then , ( [ eq : tmp008 - 1 ] ) becomes @xmath704which also implies that @xmath705 from which @xmath706 the form of @xmath23 suggests that @xmath707 can be decomposed as @xmath708from ( [ eq : tmp009 - 1 ] ) and ( [ eq : tmp009 - 2 ] ) , we have @xmath709 and thus @xmath710 . with the form of optimal @xmath85 and some basic manipulations
, we have finally @xmath711which ends the proof .
first , we have @xmath712 from which @xmath713with @xmath714 being some strictly positive constant
. then , we also have @xmath715since @xmath716 . from ( [ eq : tmp666 ] ) and ( [ eq : tmp667 ] ) , we have @xmath717where @xmath718
is another strictly positive constant .
hence , @xmath719 the lemma is proved since @xmath720 .
let us consider a parallel channel @xmath721 , each sub - channel of rank @xmath722 and with eigen - exponents @xmath723 .
since each sub - channel is an af path , the joint pdf of the eigen - exponents in the high snr regime is @xmath724 from lemma [ lemma : cal - dmt ] , the dmt is @xmath725 with @xmath726 being the outage region .
first , we can deduce that @xmath727 then , if all af paths have the same dmt , they have the same set @xmath728 , i.e. , @xmath729
. we can verify that setting @xmath730 , @xmath731 is without loss of optimality , since 1 ) the objective function is linear and symmetric on different @xmath85 , and 2 ) the constraints are convex and symmetric on different @xmath85 . finally , the optimization problem becomes @xmath732 with @xmath532 is the outage region of each single af path .
the lemma can be proved immediately from here .
without loss of generality , we assume that @xmath733 . then , the bottleneck of the channel is the @xmath82 channel .
since the partition achieves the maximum diversity , by theorem [ thm : cond_nece ] , the partition size is @xmath734 with the @xmath735 ( respectively , @xmath736 ) antennas being partitioned into @xmath737 ( respectively , @xmath738 ) supernodes .
moreover , for any af path @xmath85 in the partition , we have @xmath739 adding all the @xmath2 inequalities up gives @xmath740 the sum in the lhs of ( [ eq : tmp432 ] ) can be upper - bounded by @xmath741 , since each supernode in the transmitter can not be connected to more than @xmath737 nodes .
hence , we have the following inequality after some simple manipulations @xmath742 from which we have the lower bound on the partition size @xmath743 which is obviously increasing with @xmath738 .
therefore , the minimum lower bound is obtained by setting @xmath744 and it coincides with ( [ eq:2 ] ) .
it can be shown that this lower bound is achieved by partitioning the intermediate layer into @xmath2 supernodes with @xmath2 defined by ( [ eq:2 ] ) without partitioning either of the source and the destination antennas .
let us define the selection matrices @xmath745 s as @xmath267 diagonal matrices with @xmath746 first , we would like to prove that the maximum diversity gain is achieved .
this can be done in two steps .
the first step is to prove that the parallel channel @xmath747 with @xmath748 achieves the maximum diversity . to this end , note that by partitioning the rows ( respectively , columns ) of @xmath749 ( respectively , @xmath261 ) according to the indices in @xmath750 ( respectively , @xmath751 ) , the matrix @xmath752 can be partitioned into @xmath753 blocks , each one being an af path from the source to the destination .
therefore , @xmath747 comprises @xmath754 af paths , i.e. , all possible paths .
obviously , these paths include the @xmath2 independent paths @xmath281 in the independent partition .
therefore , the maximum diversity is achieved since @xmath755 the key of the second step is to show that the set of matrices @xmath270 defined in ( [ eq : tmp1111 ] ) is actually an invertible constant linear transformation of @xmath747 , i.e. , @xmath756 in this case , we have @xmath757 and the diversity is lower - bounded by the maximum diversity , according to lemma [ lemma : div - para ] .
hence , the ff scheme also achieves the maximum diversity .
the key point is shown in the following .
first , let us divide the set of indices @xmath758 into @xmath759 groups , each one comprising exactly @xmath760 integers @xmath761 such that @xmath762 and @xmath763 varies from @xmath42 to @xmath760 .
then , we partition the set @xmath270 according to the partition of the indices described above .
hence , the matrices in the same group can be rewritten as @xmath764 with @xmath210 being some matrix .
we have @xmath765 where @xmath766 is composed of @xmath767 blocks of matrices with the @xmath768-th block being @xmath769 if @xmath770 and @xmath771 otherwise .
we can verify that @xmath766 is invertible and with the transformation , the matrices @xmath772 s are replaced by @xmath773 s with the same indices . in the same manner , we can successively replace the matrices @xmath774 with @xmath775 by similar invertible transformations @xmath776 as @xmath766 .
finally , we obtain @xmath747 and the total transformation is invertible , constant and linear .
note that the parallel channel of the ff scheme is in outage for a target rate @xmath777 implies that at least one of the sub - channels is in outage for a target rate @xmath778 .
therefore , one can show that @xmath779 , from which @xmath780 .
finally , by showing that @xmath781 is piece - wise linear with @xmath782 sections , we prove the theorem .
let @xmath783 and @xmath643 denote the vector of the ordered eigenvalues and the corresponding eigen - exponents of a matrix @xmath517 .
the theorem can be proved by showing a stronger result : the asymptotical pdf of @xmath784 in the high snr regime is identical to that of @xmath785 .
we show it by induction on @xmath0 .
for @xmath165 , since @xmath786 , the result is direct .
suppose that the theorem holds for @xmath0 .
let us show that it also holds for @xmath541 .
note that @xmath787 from which we have @xmath788 for a given @xmath550 .
similarly , @xmath789 . at high snr , we can show that @xmath790 where the first equality comes from lemma [ lemma : invariance - asymp ] and the second one holds because @xmath791 finally , since we suppose that the joint pdf of @xmath792 is the same as that of @xmath793 , we can draw the same conclusion for @xmath794 and @xmath795 .
let us consider an equivalent block - diagonal channel of the parallel channel ( [ eq : parallel ] ) in the following form @xmath796 where @xmath797}^{{\scriptscriptstyle \mathsf{t } \!}}}$ ] , and @xmath798 are defined in the same manner .
now , from the parallel nvd code @xmath385 , we can build a block - diagonal code @xmath799 with codewords defined by @xmath800 .
we can verify that @xmath799 is actually a rate-@xmath801 nvd code defined in @xcite with @xmath802 . from ( * ? ? ?
* th . 3 ) , we have @xmath803where @xmath216 is the dmt of the parallel channel ( and thus the block - diagonal channel ) . finally , it is obvious that @xmath804 since using @xmath385 will have the same error performance as using @xmath799 except that the transmission rate is @xmath2 times higher .
we have thus @xmath805 .
it is shown in @xcite that the achievability holds for any fading statistics .
thus , the code is approximately universal .
a simple alternative construction that is approximately universal is described as follows .
let @xmath806 be a @xmath807 full rate nvd code with @xmath808 and @xmath809 .
then , @xmath806 achieves the dmt @xmath216 of the channel ( [ eq : bd ] ) .
it means that by partitioning every codeword matrix @xmath810 into @xmath811 blocks in such a way that the @xmath85th block is of size @xmath812 and sending the @xmath85th block in the @xmath85th sub - channel , the dmt of the original parallel channel is achieved .
although this construction is simple and suitable for both symmetric and asymmetric channels , the main drawback is that the coding delay is roughly @xmath2 times larger than the parallel nvd code constructed in section [ sec : paranvd ] .
decoding complexity of such codes is sometimes prohibitive .
assume that @xmath813 is a norm in @xmath424 , which means@xmath814 consider now the extensions described in fig .
[ fig : fields - extension ] with the proper fields .
from ( [ eq : x - existence ] ) and the left extension of fig .
[ fig : fields - extension ] , we deduce that @xmath815 , since the minimal polynomial of @xmath813 is @xmath816 .
meanwhile , from the right extension of fig .
[ fig : fields - extension ] , we have @xmath817 denote @xmath818
. then the number @xmath819 has an algebraic norm equal to @xmath43 , and belongs to @xmath820 which is in contradiction with the result obtained in @xcite .
so , @xmath813 is a non - norm element .
j. n. laneman and g. w. wornell , `` distributed space - time - coded protocols for exploiting cooperative diversity in wireless networks , '' _ ieee trans .
inf . theory _ , vol .
49 , no .
24152425 , oct .
j. n. laneman , d. n. c. tse , and g. w. wornell , `` cooperative diversity in wireless networks : efficient protocols and outage behavior , '' _ ieee trans .
inf . theory _
50 , no . 12 , pp .
30623080 , dec . 2004 .
k. azarian , h. el gamal , and p. schniter , `` on the achievable diversity - multiplexing tradeoff in half - duplex cooperative channels , '' _ ieee trans .
inf . theory _ ,
51 , no . 12 , pp .
41524172 , dec . 2005 .
s. yang and j .- c .
belfiore , `` towards the optimal amplify - and - forward cooperative diversity scheme , '' _ ieee trans .
inf . theory _ ,
53 , no . 9 ,
2007 , to appear .
[ online ] .
available : http://arxiv.org/pdf/cs.it/0603123 m. yuksel and e. erkip , `` multi - antenna cooperative wireless systems : a diversity multiplexing tradeoff perspective , '' _ ieee trans .
inf . theory _
2007 , special issue on models , theory , and codes for relaying and cooperation in communication networks , to appear .
elia , k. r. kumar , s. a. pawar , p. v. kumar , and h. lu , `` explicit , minimum - delay space - time codes achieving the diversity - multiplexing gain tradeoff , '' _ ieee trans .
inf . theory _ ,
52 , no .
9 , pp . 38693884 , sep .
2006 .
s. borade , l. zheng , and r. gallager , `` amplify and forward in wireless relay networks : rate , diversity and network size , '' _ ieee trans .
inf . theory _ ,
2007 , special issue on relaying and cooperation in communication networks , to appear .
p. wolniansky , g. foschini , g. golden , and r. valenzuela , `` v - blast : an architecture for realizing very high data rates over the rich - scattering wireless channel , '' in _ proc .
of the ursi international symposium on signal , systems , and electronics conference _ , new york , 1998 , pp .
295300 .
p. elia and p. v. kumar , `` approximately universal optimality over several dynamic and non - dynamic cooperative diversity schemes for wireless networks . ''
[ online ] .
available : http://fr.arxiv.org/pdf/cs.it/0512028 f. oggier and e. viterbo , `` algebraic number theory and code design for rayleigh fading channels , '' in _ foundations and trends in communications and information theory _ , 2004 , vol . 1 , no . 3 , pp .
333415 . e.
bayer - fluckiger , f. oggier , and e. viterbo , `` new algebraic constructions of rotated @xmath821-lattice constellations for the rayleigh fading channel , '' _ ieee trans .
inf . theory _
50 , no . 4 ,
pp . 702714 , apr .
2004 .
s. h. simon , a. l. moustakas , and l. marinelli , `` capacity and character expansions : moment generating function and other exact results for mimo correlated channels , '' _ ieee trans .
inf . theory _ ,
52 , no . 12 , pp .
53365351 , dec . 2006 .
s. yang and j .- c .
belfiore , `` diversity - multiplexing tradeoff of double scattering mimo channels , '' _ ieee trans .
inf . theory _ , mar .
2006 , submitted for publication .
[ online ] .
available : http://arxiv.org/pdf/cs.it/0603124 j .- c .
belfiore , g. rekaya , and e. viterbo , `` the golden code : a @xmath185 full - rate space - time code with non - vanishing determinants , '' _ ieee trans .
inf . theory _ ,
51 , no . 4 , pp . 14321436 , apr . 2005 . | we consider _ slow _ fading relay channels with a single multi - antenna source - destination terminal pair .
the source signal arrives at the destination via @xmath0 hops through @xmath1 layers of relays .
we analyze the diversity of such channels with _ fixed
_ network size at _
high snr_. in the clustered case where the relays within the same layer can have full cooperation , the cooperative decode - and - forward ( df ) scheme is shown to be optimal in terms of the diversity - multiplexing tradeoff ( dmt ) .
the upper bound on the dmt , the cut - set bound , is attained . in the non - clustered case ,
we show that the naive amplify - and - forward ( af ) scheme has the maximum multiplexing gain of the channel but is suboptimal in diversity , as compared to the cut - set bound .
to improve the diversity , space - time relay processing is introduced through the parallel partition of the multihop channel .
the idea is to let the source signal go through @xmath2 different `` af paths '' in the multihop channel
. this _ parallel af scheme _ creates a parallel channel in the time domain and has the maximum diversity if the partition is properly designed .
since this scheme does not achieve the maximum multiplexing gain in general , we propose a _
flip - and - forward _ ( ff ) scheme that is built from the parallel af scheme .
it is shown that the ff scheme achieves both the maximum diversity and multiplexing gains in a distributed multihop channel of arbitrary size . in order to realize the dmt promised by the relaying strategies ,
approximately universal coding schemes are also proposed .
relay channel , multiple - input multiple - output ( mimo ) , multihop , diversity - multiplexing tradeoff ( dmt ) , amplify - and - forward ( af ) . |
the soft excess is an important component of the x - ray spectra of many agn and is present in every source in this survey .
the soft excess is defined as the enhanced emission below @xmath02 kev compared to an extrapolation of the approximately power law spectrum in the 2 10 kev band .
this extra emission is generally approximately thermal in shape , and well fit with a black body of energy 0.1 0.2 kev .
this result stands over several decades in agn mass , e.g. walter & fink ( 1993 ) , gierliski & done ( 2004 ) , porquet et al .
this poses problems for the thermal interpretation of the data , as the temperature does not scale in the expected way with luminosity ( @xmath1 ) , as well as being too high for the standard model of shakura & sunyaev ( 1973 ) .
several alternative models have been proposed to account for one or both of these results , none of which have yet been broadly accepted by the community .
this paper investigates photoionized emission blurred relativistically by motion in an accretion disc , which has been previously studied by e.g. ballantyne , iwasawa & fabian ( 2001 ) .
we use the latest models from ross & fabian ( 2005 ) , which include more ionization states and more recent atomic data than earlier versions ( ross & fabian 1993 ) .
in the model of ross & fabian ( 2005 ) a semi - infinite slab of cold optically thick gas of constant density is illuminated by a power law , producing a compton component and fluorescence lines from the ionized species in the gas . to produce the relativistically blurred photoionized disc model this reflected emission
is added to the illuminating power law and the summed emission is convolved with a laor ( 1991 ) profile to simulate the blurring from an accretion disc around a maximally rotating ( kerr ) black hole .
this assumes a geometrically thin and flat accretion disc with clearly defined inner and outer radii , and that the emissivity of the disc as a function of radius is described by a power law .
this combined model allows fitting to the inclination of the disc , the inner and outer radii ( the outer radius is generally poorly constrained as emission is centrally concentrated ) , the emissivity index of the disc , the iron abundance of the gas , the spectral index of the illuminating power law and the ionization parameter ( which we define as @xmath2 , where @xmath3 is the illuminating energy flux and @xmath4 is the hydrogen number density in the illuminated layer , a measure of the ratio of the energy density in the illuminating radiation to the atomic number density in the gas ) .
since the model is physically motivated measuring these free parameters can give us information about the sources .
an example of the model is shown in figure [ 4051_figure ] ( bottom right panel ) .
we used publicly available archival _ xmm - newton _ data on 22 type 1 agn from the palomar - green ( pg ) sample and a selection of 12 other seyfert 1 galaxies with high - quality observations available .
the sample includes several well - studied agn , such as ngc 4051 , pg 1211 + 143 and i zw 001 .
we analyse the longest available observation where the epic ` pn ` camera took data .
we reduced the observation data files in the standard way using ` sas 6.0 ` to produce spectra ( see crummy et al .
2005b for details ) , taking the range 0.3 12.0 kev as the region in which the ` pn ` is accurately calibrated .
we do not include ` mos ` data in our fits .
we grouped the spectra so each bin includes at least 20 source counts , so @xmath5 statistics are applicable .
quoted errors are 90 per cent limits on one parameter ( @xmath6 = 2.706 ) . +
we fit the data with two classes of model using ` xspec ` , the standard power law with a black body to model the soft excess , and the relativistically blurred photoionized disc reflection model .
both models are subject to absorption from cold gas in our galaxy , we fix the amount of absorption at the value given by the ` nh ` ftool ( dickey & lockman 1990 ) .
we also allow for absorbing matter at the agn by fitting for extra cold absorption and up to two absorption edges in the 0.45 1.1 kev band .
these components are redshifted such that they are local to the agn .
we finally allow for cold reflection ( e.g. from distant gas such as a torus ) by including a narrow iron line at 6.4 kev ( in the source frame ) .
we are concerned primarily with the underlying continuum rather than these features , so we include in our final model only those components which improve the overall @xmath5 by @xmath7 2.7 per lost degree of freedom .
since we are performing fits over the entire _ xmm - newton _ band , we only report lines for which ` xspec ` reports a non - zero minimum equivalent width , this avoids fitting spurious lines due to a curvature in the continuum not addressed by the model .
this conservative procedure means faint lines are likely to be missed , although any line that significantly affects the overall goodness of fit will be included . in xspec terminology
our models are ` phabs * zphabs * zedge * zedge * ( powerlaw + zbbody + zgauss ` and ` phabs * zphabs * zedge * zedge * ( kdblur(powerlaw + atable{reflion } ) + zgauss ) ` , where ` kdblur ` is a convolution model using a laor line and ` reflion ` is a table model of reflection from cold gas ( including a redshift)jc / kdblur.html and + http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/models/reflion.html ] . +
our results are presented in full in crummy et al . 2005b .
one of the sources in the sample , pg 1404 + 226 , has been investigated with the same model in a previous paper ( crummy et al .
we find that in general the relativistically blurred photoionized disc reflection model is a better fit to the data , with 25 of the 34 sources showing an improvement in @xmath8 2.7 per degree of freedom ( note that this does not correspond to 90 per cent probability , the @xmath5 distribution is not calibrated across models ) .
6 sources show a significant worsening and 3 sources are inconclusive . in some sources
the improvement is very marked , see figure [ 4051_figure ] .
+ the disc reflection model reproduces the shape of the continuum well , and naturally explains the constant temperature of the soft excess ; see figure [ nufnu_figure ] .
the model also reproduces many features that would be otherwise interpreted as absorption edges in medium resolution ` pn ` data .
when the thermal model is used , 17 of the 34 sources appear to have an absorption edge .
this reduces to just 7 sources when the disc reflection model is used .
the disc reflection model is somewhat less smooth than the thermal model ( see figure [ 4051_figure ] ) , and bumps on the scale of absorption edges are possible .
absorption lines can not be reproduced by the model , therefore the possibility of outflows from absorption edges should be checked by investigating reflection models , or by detecting absorption lines from outflowing matter .
this has implications for the energetics and cosmological evolution of these systems , as some measured outflows can involve similar amounts of power as the radiation e.g. pounds et al .
the model also explains the apparent lack of broad iron lines in most agn spectra .
very few broad iron lines have been clearly detected ( the most well studied being mcg 6 - 30 - 15 , tanaka et al .
1995 ) , which is unexpected since it is believed that these systems are powered by accreting matter near the black hole , where broad iron lines are formed .
our results suggest that the broad iron lines do occur , but are broadened to the point of near undetectability .
a previous paper by gallo et al . ( 2004 ) supports this hypothesis , they fit an extremely broad excess of counts in the hard band of mrk 0586 with a laor profile .
mrk 0586 is included in our sample , and we find an inclination consistent with theirs ; we even fit consistent inclinations when we exclude the hard band from the fit and determine the inclination purely from the blurring effects on the soft excess .
this illustrates how the hard and soft bands are connected , they key to the absence of obvious broad lines is in the shape of the soft excess .
conversely , it is clear that the soft excess is nothing more than a blend of many broad lines .
+ ] as noted in the previous paragraph , it is also possible to measure the inclination of the central disc using the disc reflection model .
our measured inclinations are plotted in figure [ inclinations_figure ] .
we find that the inclinations are inconsistent with being random , with a kolmogorov - smirnov ( k - s ) test probability of 0.06 . performing a k - s test against a distribution which is random over @xmath9
@xmath10 ( ignoring any data points above @xmath10 ) gives a probability of 0.34 , so the data is somewhat consistent with being random over this range .
the deficit at low inclinations may be a selection effect , discs at @xmath0@xmath9 doppler beam all of their emission along the plane of the disc and so would appear dim .
the model also takes no account of limb darkening , which is strong at low inclinations . at high inclinations
the deficit could be due to torus obscuration , with the sources that do have high inclination perhaps not having a torus , or with the torus unaligned with the central disc .
+ abundances may also be measured with the model .
the model assumes a fixed , solar abundance for all elements except iron , for which the abundance is a free parameter .
the measured abundance is consistent with being between solar and @xmath11 solar for 28 of the 34 sources .
no sources have measured abundances below @xmath11 solar , and 6 have abundances which are clearly above solar .
these fairly low iron abundances are another factor in the apparent absence of broad iron lines .
the inability to fit more than one element abundance means these results are tentative , but show that in principle the composition of agn accretion discs can be measured from their x - ray spectra .
+ finally , the rotation of the central black hole may be measured using this model .
the model uses convolution with a laor line , which is based on a maximally rotating ( kerr ) black hole .
it is equally possible to use a model based on a non - rotating ( schwarzschild ) black hole ( ` diskline ` in ` xspec ` , fabian et al .
we also performed fits to all our sources using this non - rotating model , and found that in all cases the non - rotating model has a worse goodness of fit , with only two of the 34 sources where the fit is of comparable quality .
this shows that rotating black holes dominate our sample .
the black hole rotation is not a free parameter in our analysis , but the inner radius of the accretion disc is .
the last stable orbit around a non - rotating black hole is at 6 gravitational radii , and the last stable orbit around a maximally rotating black hole is at 1.235 gravitational radii . inside
this radius matter plunges into the black hole , and does not emit strongly ( section 3.4 of fabian & miniutti , 2005 ) .
all 34 sources have inner disc radii consistent with being below 6 gravitational radii , and 29 of the 34 have inner radii consistent with being below 1.3 gravitational radii .
this is a strong indication that most black holes are maximally rotating , in agreement with theoretical predictions , e.g. volonteri et al .
( 2005 ) predicts that 70 per cent of agn black holes are maximally rotating .
+ one further interesting result from the fits is that the illuminating power - law component is often undetected .
this seems to be a problem , as it is hard to imagine how the disc could be illuminated by radiation we can not see .
however , it is quite possible for relativistic effects due to the black hole to `` hide '' the illuminating continuum from us .
if the source of the illuminating continuum is above the disc , the light from it will be bent and impact on the disc , with little escaping to be observed ( miniutti & fabian 2004 ) . emission from the disc is doppler beamed away from the black hole and will still escape .
this primarily applies to sources which do not share the rotation velocity of the disc , e.g. the base of a weak jet or shocks in a failed jet ( ghisellini , haardt & matt 2004 ) .
we fit a large number of type 1 agn with a thermal model and a relativistically blurred photoionized disc reflection model based on ross & fabian ( 2005 ) .
we find that : * the disc reflection model fits the data better .
* the disc reflection model reproduces all the major features of all the sources , including the soft excess and the absence of obvious broad iron lines .
the model explains the constant temperature of the soft excess across all the sources , since it is formed from a large number of highly broadened lines .
the model reproduces many features that might otherwise be interpreted as absorption edges .
* black holes in agn strongly rotate .
* the central discs around type 1 agn have a wide distribution in inclination , with possible deficits at very low and high inclinations .
the high inclination deficit may indicate torus obscuration . *
the elemental abundances in accretion discs can be measured , iron abundances in these sources are solar or mildly sub - solar .
the relativistically blurred photoionized disc reflection model is an important tool in the study of agn .
taking account of their intrinsically relativistic nature answers several questions about their spectra , as well as providing information about the central regions .
jc , acf and rrr thank pparc , the royal society , and the college of the holy cross for support respectively .
the _ xmm - newton _
satellite is an esa science mission ( with instruments and contributions from nasa and esa member states ) .
ballantyne d.r . , iwasawa k. , fabian a.c . , 2001 , mnras , 323 , 506 + crummy j. , fabian a.c . , brandt w.n . ,
boller th . , 2005a , mnras , 361 , 1197 + crummy j. , fabian a.c . , gallo l. , ross r.r . , 2005b , mnras accepted , astro - ph:0511457 + dickey j.m . , lockman f.j . , 1990 , ara&a , 28 , 215 + fabian a.c . , miniutti g. , 2005 , astro - ph:0507409 + fabian a.c . , rees m.j . , stella l. , white n.e . , 1989 , mnras , 238 , 729 + gallo l.c . , boller th . , brandt w.n . , fabian a.c . , vaughan s. , 2004b , mnras , 355 , 330 + ghisellini g. , haardt f. , matt g. , 2004 , a&a , 413 , 535 + gierliski m. , done c. , 2004 , mnras , 349 , l7 + laor a. , 1991 , apj , 376 , 90 + miniutti g. , fabian a.c . , 2004 , mnras , 349 , 1435 + porquet d. , reeves j.n . , obrien p. , brinkmann w. , 2004 , a&a , 422 , 85 + pounds k.a . , reeves j.n . , king a.r . , page k.l . , obrien p.t . , 2003 , mnras , 345 , 705 + ross r.r . , fabian a.c . , 1993 , mnras , 261 , 74 + ross r.r . , fabian a.c . , 2005 , mnras , 358 , 211 + shakura n.i . , sunyaev r.a .
, 1973 , a&a , 24 , 337 + tanaka y. , et al . , 1995 , nature , 375 , 659 + volonteri m. , madau p. , quataert e. , rees m.j . , 2005 , apj , 620 , 69 + walter r. , fink h.h . , 1993 , a&a , 274 , 105 + wang t. , lu y. , 2001 , a&a , 377 , 52 + | we present a large sample of type 1 agn spectra taken with xmm - newton .
we fit them with the relativistically blurred photoionized disc model of ross & fabian ( 2005 ) .
this model is based on an illuminated accretion disc of fluorescing and compton - scattering gas , and includes relativistic doppler effects due to the rapid motion of the disc and general relativistic effects such as gravitational redshift due to presence of the black hole .
the disc model successfully reproduces the x - ray continuum shape , including the soft excess , of all the sources .
it provides a natural explanation for the observation that the soft excess is at a constant temperature over a wide range of agn properties .
the model also reproduces many features that would conventionally be interpreted as absorption edges .
we use the model to measure properties of the agn such as inclination , black hole rotation , and metallicity . |
over the last two decades , all - sky x - ray monitoring observations of x - ray sources in the milky way and the magellanic clouds have revolutionised our understanding of x - ray binaries , leading to many important discoveries .
for instance , detections of x - ray outbursts of transient sources allow us to discover new black hole candidates , and subsequent identification at other wavelengths enable a detailed study of the nature of the system . by following the x - ray evolution of x - ray transients ,
one can study different intensity / spectral states and investigate the nature of the compact object .
in addition , some persistent x - ray binaries show x - ray variability on timescales from milliseconds to years .
while observations with high timing resolution can probe variability from milliseconds to hours , orbital periods and superorbital periods ( periods longer than the orbital periods ) on timescales of days to years can only be studied with regular monitoring observations .
all these observations have been proven to be very successful for our understanding of x - ray binaries , but the sample is mainly limited to the milky way which suffers difficulty in determining the distance to the source ( hence uncertain luminosity measurements ) and high extinction in some regions ( e.g. , @xcite ) . apart from a few cases in the magellanic clouds , monitoring observations of x - ray binaries are very rare in other nearby galaxies . with the improvement of the spatial resolution and sensitivity of x - ray telescopes ,
such observations have become feasible now .
one intriguing discovery from x - ray observations of nearby galaxies is ultraluminous x - ray sources ( ulxs ) , which are luminous ( @xmath0 ) , non - nuclear x - ray point - like sources with apparent x - ray luminosities above the eddington limit for a typical stellar - mass ( @xmath1 ) black hole .
the majority of ulxs are believed to be accreting objects in binary systems with x - ray flux variability on timescales of hours to years ( e.g , @xcite for short - term variability ) .
assuming an isotropic x - ray emission and that the source does not exceed the eddington limit , a ulx is the best candidate for an intermediate - mass black hole ( imbh ) with a mass of @xmath2 ( e.g. , @xcite ) . while ulxs may represent a missing link between stellar - mass black holes and supermassive black holes in galactic centers , their formation and evolution are still not well understood . unlike galactic x - ray binaries , ulxs in nearby galaxies can not be detected with typical all - sky x - ray monitor instruments because of their x - ray faintness .
however , one can observe ulxs with large instruments regularly . in the past 10 years , _ rxte _ , _ chandra _ , _ xmm - newton _ and _ swift _ have been used for this purpose . for some well - known ulxs ,
long - term time series data have been building up allowing us to study their orbital and superorbital periods .
recently , there are some reports on the possible orbital periods of a few ulxs based on _ swift _ monitoring observations ( e.g. , @xcite ; @xcite ) . due to the limited time span
, the stability of the period which is one of the important indicators for an orbital period requires verification from more observations .
moreover , due to a non - regular sampling and instrumental artefacts , the results have to be cross - checked with different timing analysis techniques . in this paper , we report an investigation of long - term x - ray variability of four ulxs ( eso24349 hlx1 , holmberg ix x1 , m81 x6 , and ngc5408 x1 ) that have been observed regularly with _ swift_.
although the long - term light curves for some of the selected sources in this paper were studied before , we present a systematic timing analysis using both static and dynamic fourier power spectra on a much longer timeline of data .
we also considered the contributions from white noise and red noise to derive the significance of a signal .
finally , we applied a wavelet analysis and hilbert - huang transform to compare with results obtained by fourier analysis . for the present study we concentrate on the usage of different timing analysis techniques for the study of long - term x - ray variability of ulxs
in this study , we select ulxs ( @xmath3 erg s@xmath4 ) that have been continuously monitored by _ swift _ for more than 4 years . because _ swift _
is not an all - sky monitoring instrument , the sample is limited by the observations in the archive that were originally proposed for different purposes . in order to perform meaningful timing analyses
, we require that each target must have more than 200 data points . in total , we have four ulxs ( eso24349 hlx1 , holmberg ix x1 , m81 x6 and ngc5408 x1 ) in our sample that satisfy our criteria ( see table [ obs ] for an observing log ) . in the following subsections ,
we briefly introduce our targets .
the ulx eso24349
hlx1 is the most luminous ulx reaching a peak x - ray luminosities of @xmath5 in 0.2 - 10 kev band @xcite .
assuming an isotropic emission in a super - eddington state , a conservative minimum black hole mass is estimated to be @xmath6 , making it a very promising imbh candidate . during the past few years , the source exhibits several x - ray outbursts . the regular separation between each outburst of eso24349 hlx1 is assumed to be caused by the orbital motion @xcite until recent reports for a delay of the latest outburst @xcite .
holmberg ix x1 is a well - known ulx with an x - ray luminosity of @xmath7 in 0.5 - 10 kev band .
it was first discovered by the _ einstein observatory _ @xcite , and has been observed by all major x - ray observatories throughout the last 20 years @xcite .
recently , @xcite also used _
swift _ to monitor the x - ray evolution of holmberg ix x1 and utilised the co - added spectra taken at different luminosity states to study the spectral behaviour of the source .
the best spectral fits are provided by a dual thermal model with a cool blackbody and a warm disc blackbody .
this suggests that holmberg ix x1 may either be a @xmath8 black hole accreting at seven times above the eddington limit or a @xmath9 maximally rotating black hole accreting at the eddington limit , and we are observing both the inner regions of the accretion disc and outflows from the compact object .
.summary of observations used in analysis [ cols="<,<,<",options="header " , ] m81 x6 , which is also known as ngc 3031 x-11 , is the brightest non - nuclear x - ray source in m81 .
the average x - ray luminosity of the source in 0.3 - 10 kev band is @xmath10@xcite .
it was suggested based on optical colours from _ hubble space telescope _ observations that the system contains a @xmath11 o8 v companion @xcite . using x - ray spectroscopy ,
the black hole mass is estimated to be @xmath12 @xcite .
based on the mass of the secondary star , and by assuming that it fills the roche lobe , an orbital period of about 1.8 days is estimated @xcite . although the cadence of the current _ swift _ monitoring programme for ulxs is usually from a few days to a week and such a short orbital period is difficult to be detected , we can test if a longer periodicity exists .
ngc5408 x1 is one of the most luminous ulxs with an x - ray luminosity of @xmath13 in 0.3 - 10 kev band .
it is also one of the first ulxs observed regularly with _
swift_. using the _ swift_/x - ray telescope ( xrt ) data spanning @xmath14 days , @xcite discovered a @xmath15-day periodicity and suggested that it is due to orbital modulation . if this interpretation is correct , it would support that ngc5408 x1 contains a @xmath16 black hole with a @xmath17 companion based on a theoretical simulation @xcite .
@xcite also used the weighted wavelet @xmath18-transform to confirm the periodicity of 115.5 days , but they only considered the light curve within the first 500 days .
however , @xcite argued that such a modulation is due to a superorbital period ( see @xcite ) by comparing its physical properties to the micro - quasar , ss 433 .
they suggested that the 115.5-day modulation is originated from a precessing jet , and this scenario implies that the system has a much shorter orbital period and the accreting object is a stellar - mass black hole . using 4 years of observations from
_ swift _
@xcite , an orbital origin of the @xmath19-day periodicity can be rejected .
@xcite revisited their interpretation as a quasi - sinusoidal x - ray modulation , and re - determined an orbital period of 230 days for ngc5408 x1 based on the average time scale for the recurrence of sharp dips in the x - ray intensity .
all the data reported in this paper were taken with the x - ray telescope ( xrt ) onboard _
swift_. we only used data taken in the photon counting mode so that we can identify the precise locations of our targets without any contamination .
we extracted xrt light curves by using the xrt products generator @xcite .
it first creates an image from the event list and identifies our target for each observation .
only events with energy in the range 0.310 kev with grades 012 are included .
a circular source extraction region is chosen to match the point - spread - function .
for the background , an annulus centered on the source is used .
then for each observation , source and background counts are extracted .
source count rates are corrected for the good time interval , and losses due to bad pixels and bad columns .
scatter data points can be present at the beginning of each _ swift _ monitoring .
these data are not taken into consideration in our analysis , in order to avoid spurious effects due to large observational gaps .
we also applied a criterion to reject data with uncertainties of the background larger than @xmath20 .
in addition , we removed those data points with anomalously high or low count rates to avoid contamination by the high background or spurious flaring events . some data points with count rate errors larger than their count rate due to very short exposures ( @xmath21 s ) were not included in the temporal analysis as well .
[ lc ] shows the original light curves of all four ulxs in our sample .
since the window function may generate artificial signals , we also re - binned the data samples every 5 days in our analysis to compare with those detections obtained from the original light curve .
because the mean cadence of observations is about 210 days ( excluding obvious data gaps ) , we only considered periodicities longer than 10 days ( i.e. , @xmath22 1/day in frequency ) in our timing analyses . in order to look for any modulation
, we first performed a timing analysis by using the lomb - scargle periodogram ( lsp ; @xcite ; @xcite ) , a modification of the discrete fourier transform which is generalized to the case of unevenly spacing .
our static periodogram / power spectra were generated through the redfit software and corrected for the bias arising from unevenly spaced data in the time domain that overestimates the high - frequency power @xcite .
once period candidates are found , we examine their long - term stability by using a dynamic power spectrum ( dps ) .
finally we cross check the results with a weighted wavelet @xmath18-transform ( wwz ) that has been known as an excellent indicator of signals at the expense of being less sensitive to their powers . in the following
, we will explain the detailed setting of our methods . to derive the lsp using redfit ,
the oversampling ( ofac ) and high - frequency ( hifac ) factors , which set the period range and resolution @xcite , were individually set as 64 and 2 for each process to cover the frequency range in our analysis . in order to compare with the power determined by a typical lsp , we translated the obtained amplitude of the spectrum into lomb - scargle power dividing by a scaling factor related to the average sampling rate and the variance of count rate in our data sets .
the false alarm level of the white noise was set equal to @xmath23 , where p is the highest peak in a periodogram and @xmath24 is the number of independent trial frequencies . in order to simplify our computation
, we adopted the empirical function obtained from simulations @xcite to describe the number of independent frequencies ( horne number ; e.g. , @xcite ) .
we also fit a first - order autoregressive ( ar1 ) process to the time series to estimate the red noise spectrum ( solid line in fig . [ lsp ] ; @xcite ) .
the false alarm level of the red noise was assessed through scaling the theoretical red - noise spectrum by an appropriate percentile of the @xmath25-probability distribution .
the analysis with lsp has been widely applied for studying long - term x - ray variability in x - ray binaries ( e.g. , @xcite ) , and redfit was also successfully applied to the power spectrum to investigate the red - noise level of long - term x - ray variability @xcite . for the dps
, we employed a similar method outlined in @xcite .
the obtained archival data of our targets were analysed with a sliding window to generate the power density , and this window was moved to cover the entire time series .
this method accounts for variations in the number of data points per interval , while instability of quasi - periodicities can be resolved with a dps .
the size of the window depends on the length of the periodicity to be examined in the dps .
the minimum size of the window was set such that it covers at least one complete cycle .
since suspected periodicities of our detections range from @xmath26 to 300 days , the window length was set to 500 days .
we set the step size of the sliding window in the dps at @xmath27 days to avoid artificial effects .
the code implementation of wwz that we used is based on a theoretical study by @xcite . for an observed time series with @xmath28 data sets ( e.g. , @xmath29 ) taken at @xmath28 discrete times @xmath30
, we can define @xmath28-dimensional contravariant vector in sampling space with a representation of the canonical basis .
any function of time @xmath31 can also be defined as a contravariant vector in sampling space , and therefore the inner product of two functions @xmath31 and @xmath32 is defined as @xmath33 where @xmath34 is the statistical weight assigned to uneven data points . for a set of trial functions @xmath35 , the @xmath36-matrix for the sampling space is given by @xmath37 and the coefficients of equally weighted data can be described as @xmath38 for the model function .
the wavelet transform ( wt ) onto the complex trial function at the frequency @xmath39 and the time @xmath40 is @xmath41 where @xmath42 is a constant , and can be regarded as a projection onto the complex trial function of @xmath43 with real statistical weights chosen as @xmath44 .
the wwz consists of the effective number and the weighted variations of uneven data and model functions as below .
according to @xcite , the effective number is @xmath45 ^ 2 } { \sum e^{-2c\omega^2 ( t_\alpha-\tau)^2}}\end{aligned}\ ] ] the weighted variation of the uneven data @xmath46 is given as @xmath47 = < x|x>-<\mathbf{1}| x>^2\end{aligned}\ ] ] and the weighted variation of the model function @xmath48 is @xmath49 ^ 2 = < y|y>-<\mathbf{1}| y>^2\end{aligned}\ ] ] then the wwz is read as @xmath50 notably , the coefficients @xmath51 are directly computed based on the uneven - smapling data . in practice , the choice of @xmath52 could be @xmath53 once a period candidate is determined from the wwz , the amplitude of the signal can be determined as wwa = @xmath54 @xcite .
all the obtained signals in this paper are subtracted by their mean , and the significance can also be examined with wwa or the corresponding effective number ( eq . [ neff ] ) .
here we concentrated on the signals resolved from the evenly sampled data by the wwz in a comparison to those obtained by the dps . comparing with other time - frequency analysis methods ,
the wwz is optimised for non - stationary unevenly time - series analysis , so this method is quite adequate to verify any signal that exists in the obtained data sets .
the long - term light curve of the source is shown in fig .
it is clear that the source exhibits substantial variability . in particular , there are nearly five complete outbursts with a fast - rise exponential - decay profile similar to typical galactic x - ray transients .
the first and second luminosity peaks for eso24349 hlx1 were in august of 2009 and late - august of 2010 @xcite . with the regular monitoring by _ swift _ , the next two cycles of outbursts recurred in 2011 and 2012 @xcite .
the separation between outburst peaks changes from @xmath55 days for the first two outbursts to @xmath56 days for the following ones .
the most recent outburst was detected in early - october of 2013 with a delay by almost a month compared to the previous one @xcite . by performing a lsp analysis ,
the strongest peak higher than both 99% red and white noise significance levels for binned data is @xmath57 days ( fig .
[ lsp ] ) , where the error was estimated by propagating two uncertainties .
the first uncertainty in frequency of the trial periodicity was estimated using @xmath58 where @xmath59 is the total time interval and @xmath60 is the peak power @xcite .
the second uncertainty accounts for measurement errors effects , and was determined by carrying out monte carlo simulations .
we simulated @xmath61 light curves by randomizing the count rate according to a gaussian distribution with standard deviation equal to the count rate uncertainties .
then , we repeated the lsp analysis and derived a standard deviation for the peak frequency of the periodicity .
the major signal also has a power higher than 95% confidence level determined from the red noise model for the power spectrum generated from the original data , indicating that this period is statistically significant .
similarly , the dps in fig .
[ dps ] obtained from the re - binned light curve shows a significant periodicity at about 380 days .
there are two other peaks ( @xmath62 days and @xmath63 days ) in fig .
[ lsp ] with powers more significant than the 99% white noise confidence level , and they might correspond to different harmonics of the strongest signal as we will describe in detail in section 5.1 . according to previous reports @xcite , the separation between the peak luminosity of the first two outbursts and the following outbursts has increased . except for an obvious delay of the outburst occurred in late 2013 , because the poor resolution provided by the dps can not discriminate such a change unambiguously , we considered the lsp with the sub - divided data sets of day @xmath64 ( the black lsp ) and day 365 - 1210 ( the red lsp ) such that both include three outbursts ( see fig .
[ ds_lsp ] ) .
the strongest signals shown in fig .
[ ds_lsp ] are @xmath65 days and @xmath66 days . the difference can be associated with the change in the separation between the two outbursts .
in addition , we found that the power of the signal at @xmath67 days significantly increases in recent observations .
this result can directly be linked to the second harmonic / first overtone with the separation between recent outbursts ( @xmath68 days ) and provides evidence for variation of the duty cycle for each outburst .
this can clearly be seen in fig .
[ dps ] as well , and a bifurcation of the main periodicity @xmath68 days can be found after day @xmath69 .
this might also cause the delay on the most recent outburst in 2013 early - october .
because of the lack of observations between day 12101435 , our results were seriously smeared in this time interval . the aforementioned phenomena discovered in the dps can also be seen from the wwz periodogram ( see fig .
[ wwz ] ) .
there are three strong signals , which represent periodicities of @xmath68 days , @xmath67 days and @xmath70 days .
the major signal ( @xmath68 days ) across the whole data is very strong , but the resolution of the wwz periodogram is not good enough to resolve any variations .
however , we can obviously find that two sub - signals generated from the overtones only get significant after day 720 .
this result totally agrees with those from the dps , indicating a variation of the duty cycle of each outburst .
since eso24349 hlx1 has a regular monitoring before day 1210 , it is an ideal object to be further studied with the hilbert - huang transform ( hht ; @xcite ) since hht analysis requires an evenly sampled data set .
though the data were binned with 5 days per bin , there is also an obvious gap after the 4th outburst between day 12101435 ( only two data points were recorded on days @xmath71 and 1409 ) . because this gap is in the quiescent state and the count rate is expected to be less than 0.003 cts / s according to previous observations , we re - interpolated the light curve using a piecewise cubic hermite spline function @xcite such that the data are evenly sampled without any significant artificial effect . following the similar process to study the superorbital period of smc x-1 with hht @xcite
, we applied an ensemble empirical mode decomposition ( eemd ; @xcite ) method to decompose the original light curve into a number of intrinsic mode functions ( imfs ) with the medium noise level at first .
the decomposed component of the largest energy / power contains four cycles , which correspond to the main modulation obtained from the lsp and dps . here
, we employed a normalised hht ( fig .
[ hht_lsp ] ; @xcite ) , which is sensitive to both inter - wave modulation ( i.e. , the variation of a cycle length ) and intra - wave modulation ( i.e. , the instantaneous frequency changes within one oscillation cycle ) , to represent the instantaneous frequency on the hilbert spectrum and to compare with results obtained from the dps and wwz .
a strong variable signal between @xmath72 250 and 500 days ( frequency @xmath73 1/day ) is seen in the hilbert spectrum , and it fits well to the major signal yielded from the dps .
the variation shows roughly five cycles , which is consistent with the quasi - periodicity detected from the light curve . since the @xmath68-day modulation is highly non - linear ( non - sinusoidal ) , the intra - wave modulation would dominate the variation of instantaneous frequency .
[ lc ] and fig .
[ lsp ] show the long - term light curve and the lsp of holmberg ix x1 , respectively . comparing to @xcite
, holmberg ix x1 has gone through at least three other high / low intensity transitions .
the two period candidates ( @xmath74 and 64.5 days ; @xcite ) found in previous analyses below the 85% white / red noise confidence level in the lsp .
instead , we obtained a significant periodic signal from both unbinned and re - binned data sets at @xmath75 days , which reaches the 99% red noise significance and exceeds over the 99% white noise significance levels in fig .
[ lsp ] . according to the light curve , the peak ( @xmath76 cts / s ) recurs at days 440455 , 11201130 and 17301740 , and each separation is about 625 days
. it can also be seen at days @xmath77 and 12701275 with a separation of @xmath72 625 days .
if we consider the dip features determined from the light curve , we can also find them with a separation of about 600 days at days 135140 , 750760 , and 13251330 .
however , the length of the current light curve used in our analysis is @xmath78 days , which only includes two complete cycles for a periodicity of @xmath79 days . given the scattering of the data and the total length of the observations , the current data set is not suitable for the hht .
since there is an obvious gap without any data points at days 8451085 , we divided the data into two groups to inspect whether there is any periodic signal in a specific time interval .
these results are also presented in fig .
[ ds_lsp ] . because the wwz can also be employed to unevenly sampling data , we double checked those signals resolved from the dps and from the lsp with the sub - divided data . in the wwz periodogram of fig .
[ wwz ] , there is a strong signal at low frequencies ( @xmath79 days ) persisting across the whole data set .
the significance of this signal slightly decreases after day 1050 , and a similar phenomenon can also be observed in fig .
we can not confirm such a long - term periodicity with the current data with limited cycles since the resolution of signals obtained from the wwz is not good enough .
furthermore , no adequate conclusion can be made on this feature from dps because the adopted window time ( 500 days ) is shorter than the periodicity we are discussing . we show the _ swift _ long - term light curve of m81 x6 in fig .
the lsps obtained from both the unbinned data and 5-day average data are presented in fig .
the source varies by a factor of @xmath80 on a time scale of months .
the lsp indicates that the strongest candidate periods are at about 110 days and 370 days , but they are not statistically significant ( @xmath81% red noise confidence level ) when we take into account the entire data sets .
only two signals are more significant than the 99% white noise confidence level in the dps ( fig .
[ dps ] ) , and this is consistent with the lsp in fig . [ ds_lsp ] .
the black curve describes the lsp obtained from the re - binned data sets of @xmath82 day 740 and the red one represents the lsp obtained from the re - binned data sets of @xmath72 day 9651645 .
the period at @xmath72 370 days is only prominent before day 740 while the period at @xmath72 110 days is only significant after day 965 .
there are two obvious gaps without any data points in the light curve after day 965 .
the length of the two gaps ( at days 11601315 and 13501490 ) is 155 days and 140 days .
in addition , we also found that the duration with frequent cadence at days 9651160 and 14901645 is 195 days and 155 days , respectively .
the duration of the gaps and the data with frequent cadence also propagate the fake signals in lsp of fig .
when we examined the signals with the wwz shown in fig .
[ wwz ] , the distinction between the two major signals and others are much more evident .
these findings strongly suggest that those sub - signals found in the lsp of fig .
[ lsp ] are artificial effects . with _
swift_/xrt data covering about 1500 days , we plot the long - term light curve in fig [ lc ] .
we performed a power spectral analysis with lsp and the significance levels of white and red noises in the resulting power spectrum are shown in fig .
we do not find any significant periodicity ; the highest peaks obtained from the original data and the re - binned data sets ( 5 days per bin ) are at @xmath83 days and @xmath84 days , respectively .
these results are consistent with the proposed @xmath85-day quasi - periodic modulation obtained by @xcite and the @xmath72 187-day periodicity detected by @xcite . because there is a large deviation between the obtained peak at @xmath67 days and the proposed orbital period ( @xmath86 days ) of this system @xcite , it is difficult to establish a direct connection between our results and the suggested orbital modulation . in the dps ( fig .
[ dps ] ) , it is clear that the signal near 112 days is quite strong during the first @xmath14 days of observations , but it has become weaker after that . the quasi - periodic signal near 112 days has significantly decreased after day 800 and evolved into a quasi - periodicity of @xmath72 190 days . on the other hand ,
the @xmath72 190-day period has been evident after day 800 and was quite weak before day 500 . in order to clarify this
, we used the lsp obtained in different time intervals to determine the strength of signals obtained in different stages and to avoid any contaminations from artificial signals arisen from the length of empty observations .
the black and red curves in fig .
[ ds_lsp ] represent the lsp resulted from the re - binned data of @xmath72 day 0535 and @xmath72 day 01230 since mjd 54565.8 , respectively .
an exchange of the significance between two major signals ( @xmath72 112 days and 190 days ) is clearly shown although they are not statistically significant , and the similar phenomenon can also be verified by the wwz shown in fig .
we generated lsp and dps for four ulxs from the long - term monitoring _ swift _
xrt data , with the aim of revealing periodic signals embedded in the light curves .
the results were also examined with the wwz . in the case of eso24349 hlx1
, we also used hht to further study its periodicity . here
, we discuss the results for each object .
both the lsp and dps shown in fig . [ lsp ] and fig .
[ dps ] present a strong persistent periodic signal at @xmath68 days .
this detection is consistent with the mean separation between the start of different outbursts or between their peak luminosities ( @xmath72 350 , 373 , 375 and 407 days ) .
the gradual delay of each outburst also reflects on the result of dps , and a clear bifurcation from the main signal can also be seen .
because the biforked feature is only significant in the data @xmath87 day 720 , we divided the data sets observed with similar cadence ( @xmath88 days ) in two segments that both include the entire duty cycle of three outbursts to trace the variation of the periodic signals .
the lsp in fig .
[ ds_lsp ] yielded from the data which include outbursts from the second to the fourth reveals some other quasi - periodic signals that correspond to the harmonics of the main signal at @xmath68 days .
if we reduce the length of the light curve to include the third and the fourth outbursts only , we find that the significance of these signals is enhanced .
the hilbert spectrum and the wwz provide us an opportunity to verify the results yielded from lsp and dps .
we compared the dps and the hilbert spectrum in fig .
[ hht_lsp ] .
the contour represents the power of dps generated with a moving window of 500 days through the re - binned light curve .
the colour map represents the hilbert energy defined as the square of the amplitude , and it was smoothed using a gaussian filter .
it is clear that the hilbert spectrum yielded from the normalised hht passes through the peak of dps , indicating that the result of the hilbert spectrum and the dps are consistent .
furthermore , the hilbert spectrum provides a much higher / better resolution to investigate the structures of the variability / instability caused by the intra - wave modulation , i.e. the frequency modulation within one cycle .
it is caused by the non - sinusoidal nature of the modulation because the modulation shape contains a sharp rise and an exponential decay .
the maximum frequency of the the main signal due to the intra - wave modulation appears at @xmath72 390 , 745 and 1125 days ( fig .
[ hht_lsp ] ) , which are similar to the epochs of the peak luminosity of each outburst . comparing to the dps in fig .
[ dps ] , we do not find a biforked feature on the wwz periodogram .
the reason is that the lomb - scargle method tends to describe the data with a smooth sinusoidal wave characterized by a longer period when the moving window approaches the flat low count rate state .
nevertheless , the gradual delay of the major signal can still be seen in fig .
[ wwz ] if we zoom in the frequency domain .
we can also find that a new signal begins to appear after day 510 , and its power is enhanced after day 760 . because day 510 is at the quiescent state after the second outburst and day 760 is close to the peak of the third outburst , we speculate that the emergence and the enhancement of this new signal might be due to a change of the shape of the light curve in outbursts .
the best fit to the cycle length of the first four complete outbursts can be estimated as @xmath72 370 days from the strongest signal in the lsp .
we then folded the light curve with this periodicity to examine the difference of the outburst light curve profiles in each cycle .
as shown in the top panel of fig . [ flc ] , we use different colours to mark each cycle .
the black curve characterizes the first cycle , while the blue one represents the second cycle .
the third and the fourth outbursts are depicted with green and red colours , repectively .
the most recent outburst denoted by the yellow curve recurred with a relatively long separation ( @xmath89 days ) and clearly shows a delay during its rising phase .
the first four outbursts start with a sharp rise near phase 0 , but the profiles of their decays are different . the duty cycle of the outburst decreases from @xmath90% in first cycle to @xmath26% in the last cycle .
the third and the fourth outbursts have similar phase for each peak , and a good fit to their duty cycle can be modelled by a consecutive sinusoidal signal with a periodicity of 190 days as shown in the bottom panel of fig .
this periodicity is consistent with the signal ( @xmath62 days ) that we obtained in fig .
[ lsp ] . a third signal of @xmath91 days in fig .
[ ds_lsp ] is roughly consistent with the length of the duty cycle of recent outbursts .
these results give an interpretation to the bifurcation of the main signal shown in the dps and the emergence of all the sub - signals shown in the dps and the wwz , and these quasi - periodicities are indeed originated from the harmonics of the main signal . because the distribution of the data is highly non - linear and the duty cycle of the third and fourth outburst is much shorter , the eemd tends to decompose the peak / amplitude into another imf and this feature can also be seen in the hilbert spectrum .
therefore the power / energy of the main signal during the first two outbursts shown in the dps is weaker than that during the next two outbursts ( fig . [ dps ] ) while the hilbert spectrum presents an opposite result .
moreover , we see an obvious signal varying between 125 and 250 days at day 5101310 in the hilbert spectrum as well . since the main periodicity embedded in the light curve of eso24349 hlx1 arises from the separation of each outburst , this signal is not stable because of the delay between outbursts .
we can clearly see this feature in fig .
[ dps ] , and the main period increases to @xmath68 days if we take into account the latest outburst . to explain the @xmath68-day modulation
, it has been suggested that the companion star is in a highly eccentric orbit with the central black hole @xcite .
our analyses , however , indicate that the modulation is not stable .
more recently , @xcite proposed that the orbital period could change between different orbits as a consequence of stochastic fluctuations .
as an alternative explanation we speculate that this modulation represents a superorbital periodicity , while the orbital period of the system is likely @xmath92 days @xcite .
@xcite suggest that the system may be similar to the galactic microquasar ss433 for which the x - ray modulation is due to precession of an x - ray beam .
@xcite detected two strong possible signals near 28.8 and 64.5 days with the _ swift _ data taken between 2006 july and 2009 june .
however , both signals are not statistically significant when the 11 observations before mjd 54200 are removed and only 177 observations after mjd 54801 ( day 878 in fig .
[ lc ] ) are taken into the consideration @xcite . here
, we have more observations covering @xmath78 days to investigate the long - term periodic signal of holmberg ix x1 . by taking red noise into account ,
the two periods are not significant .
we also find a prominent signal corresponding to a quasi - periodicity of @xmath79 days ( @xmath93 1/day in fig .
[ lsp ] ) .
we find that this signal roughly corresponds to the separation between each peak or dip event although some of them seem to appear randomly .
according to fig .
[ dps ] and fig .
[ ds_lsp ] , the signal of @xmath79 days is much more prominent in the beginning ( before day @xmath94 ) of the re - binned light curve , but the length of this segment is shorter than two complete cycles .
although this long - term signal has a stronger significance in the wwz ( see fig .
[ wwz ] ) , the length of the entire light curve we used in the analysis is too short ( only about 2.8 cycles ) to confirm this periodicity .
in addition , we must emphasize that the window size adopted in the current dps is only 500 days and it is also the reason why we only show the signals of interest with periodicity @xmath95 days in most lsps .
all the above reasons make the confirmation of such a long periodicity difficult . because the light curve of holmberg ix x1 is quite spiky
, it is not appropriate to interpolate the data sets to generate an evenly sampled light curve for hht . in order to confirm the long - term periodicity of holmberg ix x1 ,
more observations with similar cadence are required in the future . for m81 x6 , except for a major gap with only two observations at day @xmath96 and two other gaps with no observations at @xmath97 and 13501490
, the mean cadence of the data set is 2.8 days . therefore , the _ swift _ light curve is not ideal to examine the proposed 1.8-day orbital period @xcite , but it provides us an opportunity to study the long - term variability of this high - mass x - ray binary system . although there is no significant period in the lsp ( fig .
[ lsp ] ) , there are two significant periods when we consider the light curves in different epochs ( fig .
[ ds_lsp ] ) . from the dps ( fig .
[ dps ] ) and the lsp obtained from different stages shown in fig . [ ds_lsp ] , we find that the @xmath98-day period is only prominent in the first segment ( @xmath82 day 740 ) of the light curve while the @xmath99-day one is only prominent in the last segment ( @xmath72 day 965 - 1645 ) of the light curve .
the wwz periodogram also clearly demonstrates two strong quasi - periodicities existed in a different time interval .
comparing with the dps , wwz is unaffected by the noise generated from the window effect and all spurious sub - signals are less significant in the wwz map . in the first segment of the light curve , the dip events ( @xmath100 cts / s ) take place at days @xmath101 and @xmath102 and the peak events ( @xmath103 cts / s ) appear at days @xmath104 and @xmath68 .
the separation between the dip or the peak luminosity in the light curve provides a possible explanation to the quasi - periodic signal at @xmath98 days . because the length of the light curve in the first segment is only about 740 days , which only covers about two cycles of the @xmath98-day quasi - periodic signal , it is difficult to confirm the stability of this signal with this limited data set .
the @xmath72110-day quasi - periodic signal is not only the most significant one we obtained in the second segment of the light curve , but also the most significant one in the unbinned data .
the light curve folded with @xmath99 days is shown in fig .
[ m81_flc ] .
although there are two obvious gaps in the 3rd cycle and between the 4th and the 5th cycles , observations in the last segment folded with the @xmath99-day quasi - periodic signal show similar structure as demonstrated in the binned folded light curve . according to the dps and the wwz ,
this signal is still significant from more recent observations .
future observations with similar cadence are required to infer the nature of this quasi - periodicity . with a long - term monitoring of ngc5408 x1 , we find that the most obvious signals from lsp and dps are @xmath105 days and @xmath67 days .
however , none of them are statistically significant , and this result is consistent with @xcite . in the analysis with dps ,
the @xmath105-day period is stronger during the first @xmath14 days of observations , which is consistent with the @xmath106-day period detected by @xcite using the data of the first @xmath107 days .
another quasi - periodicity of @xmath67-day starts to appear after day 400 and becomes prominent after day 800 .
we do not find any strong signal with a period longer than 200 days to support a proposed orbital period of @xmath86 days @xcite .
another weak signal of @xmath108 days arisen at day 600 shown in fig .
[ lsp ] and fig .
[ wwz ] is not related to the proposed orbital period as well , even if we take into account its 95% uncertainty .
since the recurrence of the x - ray dips related to the claimed periodicity of @xmath109 days @xcite only occurs before day 1200 and the time duration of each cycle is quite unstable ( 251 days , 211 days , 276 days and 235 days ) , this signal is unlikely real and future monitoring observations may see a suppression of the signal .
@xcite also indicated that the dip events do not repeat at every cycle .
such phenomenon can also be seen in some galactic transient sources ( e.g. , gro j165540 ; @xcite ) . because the transient behaviour can cause a change on the structure of the accretion disc , the efficiency in x - ray irradiation of the disc will change accordingly causing different behaviours of dipping activities @xcite . however , ngc5408
x1 is a persistent source without any obvious change in x - rays on timescale of years .
a lack of dipping activity in some epochs has been seen in galactic x - ray dipping sources ( e.g. , x1916 - 053 and xb 1254 - 690 , see @xcite and @xcite ) and it may be due to a decrease of the vertical structure of the accretion disc . alternatively , a tilted and precessing disc may provide a possible scenario on the variability of the periodicities and the disappearance of dips @xcite .
the presence of two relatively strong quasi - periodic signals ( @xmath105 days and @xmath67 days ) in the light curve may still have certain relation with the orbital period of ngc5408 x1 .
such a feature can be caused by variations of the accretion rate due to a precession of the accretion disc induced by the tidal force @xcite , or instabilities of disc warping driven by irradiation @xcite .
last but not least , the quasi - periodic signals may be associated with a precessing jet @xcite . we therefore suggest the detected periods to be superorbital periods since these signals are only prominent in a specific time interval of the long - term light curve although it is not entirely clear which mechanism induces these signals .
according to previous investigations between superorbital periodicities and orbital periods of galactic x - ray binaries @xcite , we support the conclusion given by @xcite , who report an orbital period of @xmath110 days .
we have used the _ swift _ archive to investigate the long - term x - ray variability of four ulxs . with timing analyses through the lsp , dps and wwz , we detect the following intriguing periodic signals embedded in their light curves : @xmath68 days for eso24349 hlx1 , and @xmath98 days and @xmath99 days for m81 x6 .
a brief summary of all potential periods is shown in table [ period ] . except for the main signal detected from eso24349 hlx1 ,
all other period candidates are only prominent in a specific time interval .
the @xmath68-day periodic signal of eso24349 hlx1 corresponds to the separation between outbursts , and it was claimed to represent an orbital modulation @xcite . among the four ulxs discussed in this paper , only eso24349 hlx1 has a data set characterized by regular monitoring , which is adequate to be examined with the hht .
this is also the first time that hht has been applied to study long - term x - ray variability of a ulx .
as compared with the dps , the wwz and the hht are independent of the window time , and the hilbert spectrum provides a better resolution to discriminate the evolution of each signal .
we find that the separation between each outburst varies over time , and suggests this main signal to be a superorbital period .
this may also support the idea of a precessing x - ray beam @xcite although a changing orbital period could also be possible @xcite .
at the time of writing , the new outburst has started in 2015 early - january , more than 460 days since the last outburst @xcite .
more monitoring observations in the future are required to distinguish among different models .
ccc source & period / mjd range & possible origin + eso 24349 & @xmath68 days & orbital period ?
+ hlx1 & ( @xmath72mjd 55050 - 56561 ) & superorbital period ? +
holmberg ix & @xmath79 days@xmath111 ? & + x1 & ( @xmath72mjd 54803.7 - 57439.4 ) & + & @xmath98 days & + & ( @xmath72mjd 54924.6 - 56554.7 ) & + & @xmath99 days & + & ( @xmath72mjd 55889.6 - 56554.7 ) & + ngc5408 x1 & none & + the period candidates of other three ulxs may range from @xmath112 days to @xmath113 days . apart from noise and artifacts , all the candidate periods are only significant in a specific epoch .
this suggests that they are not associated with any stable mechanism such as orbital motion . instead , such long - term ( @xmath87 100 days ) x - ray quasi - periodic variations are likely related to superorbital periods that are thought to be due to radiation - driven warping of accretion discs @xcite or tidal interaction - induced disc precession @xcite . alternatively , mass transfer rate - related events such as x - ray state changes and disc instability can also cause long - term modulations @xcite . in particular , there are two intermittent quasi - periodicities for both ngc5408 x1 and m81 x6 , suggesting that the quasi - periods are changing or evolving .
they resemble some galactic x - ray binaries that show similar behaviour ( e.g. , cyg x2 and smc x1 ; @xcite ) and it has been suggested that a warped disc could lead to an unstable steadily precessing disc , causing quasi - periodic behaviour @xcite .
we note that there are many uncertainties on the physical parameters of ulxs .
to determine the origin of superorbital periods of ulxs , one has to know at least the mass ratio between the companion and the compact star ( @xmath114 ) and the binary separation .
unfortunately , it is very difficult to get these parameters for ulxs .
for the three ulxs discussed here ( i.e. , excluding eso24349 hlx1 ) , only m81 x6 has better constraints on the black hole mass and the nature of the companion .
the masses of the black hole and companion star are estimated ( @xmath115 , @xmath116 ) such that @xmath117 can be derived . in this case , we can rule out a tidal interaction - induced disc precession scenario that requires @xmath118 @xcite . for a warped disc , the binary separation and
the mass ratio suggest that m81 x6 lies in the intermediate instability zone for radiation - driven warping in x - ray binaries ( see figure 1 of @xcite ) .
the quasi - periodic variability may represent the switching timescale between a warped disc and a flat disc .
if all of these intriguing signals do represent superorbital periods , the real orbital periodicities of these systems are expected to be much shorter @xcite .
this will give us constraints on the binary separation as well as the black hole mass . in order to understand the physical nature of the quasi - periodic modulations of the four ulxs discussed here
, a regular long - term x - ray monitoring is required in the future .
more importantly , we demonstrate that dynamic timing analysis is important to investigate the long - term x - ray behaviour of ulxs especially when the periodic behaviours are intermittent or varying .
we would like to thank the anonymous reviewer for his / her thorough review and constructive comments to improve the paper .
this work made use of data supplied by the uk swift science data centre at the university of leicester .
this project is partially supported by the ministry of science and technology of the republic of china ( taiwan ) through grant nsc 101 - 2112-m-039 - 001-my3 .
c. p. h. and y. c. are supported by the ministry of science and technology of taiwan through the grant nsc 102 - 1221-m-008 - 020-my3 .
a. k. h. k. is supported by the ministry of science and technology of taiwan through grant 103 - 2628-m-007 - 003-my3 .
d. c. c. y. gets the financial support from the ministry of science and technology of taiwan through nsc 101 - 2115-m-030 - 004 .
p. a. , beardmore a. p. , page k. l. , osborne j. p. , obrien p. t. , willingale r. , starling r. l. c. , burrows d. n. , godet o. , vetere l. , racusin j. , goad m. r. , wiersema k. , angelini l. , et al .
2009 , mnras , 397 , 1177 p. a. , beardmore a. p. , page k. l. , tyler l. g. , osborne j. p. , goad m. r. , obrien p. t. , vetere l. , racusin j. , morris d. , burrows d. n. , capalbi m. , perri m. , gehrels n. , romano p. , 2007 ,
a&a , 469 , 379 | long - term x - ray modulations on timescales from tens to hundreds of days have been widely studied for x - ray binaries located in the milky way and the magellanic clouds .
for other nearby galaxies , only the most luminous x - ray sources can be monitored with dedicated observations .
we here present the first systematic study of long - term x - ray variability of four ultraluminous x - ray sources ( eso24349 hlx1 , holmberg ix x1 , m81 x6 , and ngc5408 x1 ) monitored with _
swift_. by using various dynamic techniques to analyse their light curves , we find several interesting low - frequency quasi - periodicities .
although the periodic signals may not represent any stable orbital modulations , these detections reveal that such long - term regular patterns may be related to superorbital periods and structure of the accretion discs .
in particular , we show that the outburst recurrence time of eso24349 hlx1 varies over time and suggest that it may not be the orbital period . instead
, it may be due to some kinds of precession , and the true binary period is expected to be much shorter .
[ firstpage ] methods : data analysis x - rays : binaries x - rays : individuals ( eso24349 hlx1 , holmberg ix x1 , m81 x6 , ngc5408 x1 ) |
the recent observations of the wilkinson microwave anisotropy probes ( wmap , spergel et al .
2003 ) , combined with many other observations on large scale structures , favor a flat universe in which the dark energy , @xmath2 , dominates the expansion and evolution of the universe , and most of its non - relativistic matter is cold dark matter ( cdm ) .
the ordinary matter , i.e. the baryonic matter , accounts for only a small fraction ( @xmath3 of the non - relativistic matter ) . with these favorable cosmological parameters plus some other reasonable parameters for the hubble constant and for the primordial density fluctuations
, the cdm model can match most of the current observations , including the cosmic microwave background , the intergalactic medium at high redshift , the abundance of rich clusters , and the large - scale distributions of galaxies in the two degree field galaxy redshift survey ( 2dfgrs , colless et al .
2001 ) , in the 2-micron all sky survey ( 2mass , skrutskie et al . 1997 ) , and in the sloan digital sky survey ( sdss , york et al . 2000 ) ( e.g. peacock et al . 2001 ; maller et al . 2003
; tegmark et al . 2004 )
. however , when we compare theoretical models ( such as the cdm model ) with the galaxy distribution revealed by large redshift surveys of galaxies , we have to contend with the bias in the relation between the distribution of luminous galaxies and the underlying dark matter . to understand this relation is therefore one of the main challenges in modern cosmology . with accurate multi - band photometries and medium resolution spectra now available from large redshift surveys such as the 2dfgrs , the sdss and the deep2 galaxy redshift survey ( coil et al .
2004 ) , it is now possible to study in detail a wide range of properties of the galaxy population , such as spatial clustering on different scales , the dependence of clustering on luminosity and color , the luminosity function , the magnitude - color relation , and the environmental dependence of the galaxy population .
these properties can provide important clues to how the galaxies have formed , and important constraints on the underlying cosmological model , such as the primordial density fluctuation , cosmological parameters , and the properties of the dark matter . to fully understand the galaxy properties and to fully make use of large surveys of galaxies to constrain cosmological models ,
it is necessary to understand how galaxies form in the cosmic density field .
however , it is still a challenge to model galaxy formation within the framework of cdm models , though significant progress has been achieved in the last two decades . the parameter space for the big - bang cosmology has now been narrowed greatly , and the uncertainties caused by these parameters are relatively small . using high - resolution @xmath0-body simulations and sophisticated analytical models , the properties of the dark matter distribution are well understood in the cdm scenario .
in particular , a great deal have been learned about the properties of the cdm halo population , which are virialized clumps formed through gravitational instability in the cosmic density field and in which galaxies are assumed to form .
the challenge for current galaxy formation is really to understand the physical processes that govern galaxy formation and evolution in dark matter halos , such as shock heating of gas , radiative cooling , star formation , agn activity and their feedback , and galaxy merging .
there are several ways to link galaxies to dark matter halos in a given cosmological model .
the most straightforward way is to simulate galaxy formation in an expanding universe by numerically solving the gravitational and hydrodynamical equations ( e.g. katz & gunn 1991 ; cen & ostriker 1993 ; bryan et al .
1994 ; navarro & white 1994 ; couchman et al . 1995 ; abel et al . 1997 ; weinberg et al . 1998 ; yoshikawa et al . 2000 ; springel et al .
the hydro / n - body simulation has the advantage of treating the gas dynamical processes in a self - consistent manner .
other known important physical processes , such as star formation and its feedback , are usually input into the simulation `` by hand '' . in order to resolve galaxies as well as to include the effect of the large scale structures , these simulations are required to cover a large dynamical range . with the recent rapid developments of the algorithm , of the physical modeling , and of the computer hardwares , the prospects of developing the hydro / n - body simulations are very promising .
but with the current technology , it is still very time - consuming to run a realistic hydro / n - body simulation of galaxy formation(e.g .
springel & hernquist , 2003 ) , which makes it difficult to study how the galaxy properties change with model parameters .
another powerful tool to link galaxies to dark matter halos is the so - called halo occupation model ( yang et al .
2003 ; van den bosch et al . 2003 ; berlind et al . 2003 ; zheng et al . 2004
; vale & ostriker 2004 ; kravtsov et al .
2004 ; for earlier works , see jing , mo , brner 1998 ; peacock & smith 2000 ; seljak 2001 ; berlind & weinberg 2002 ; bullock , wechsler & somerville 2002 ; cooray & sheth 2002 ) .
the halo occupation model assumes a parameterized form for the conditional luminosity function ( clf ) that quantifies the luminosity distribution of galaxies in a halo of mass @xmath4 .
in contrast to the hydro / n - body simulations that aim to model galaxies from basic principles of physics , the parameterized form of the halo model is motivated by observations and the model parameters are determined by best - fitting the observations such as the luminosity function of galaxies , luminosity and color dependences of galaxy clustering , the mass - to - light ratios of galaxy systems ( yang et al .
2003 ; van den bosch et al . 2003 ; yan et al . 2003
; van den bosch et al . 2004 )
. in the future , more observational data can be incorporated to further refine the halo occupation model , which will provide invaluable insight on how galaxies form in cdm halos .
however , the halo model itself does not tell us about how the observed properties of galaxies have developed through the cosmological evolution . a third method , which has been widely used and will be used here in this paper , is to construct semi - analytical models ( sams ) of galaxy formation ( e.g. , white & frenk 1991 ; lacey & silk 1991 ; kauffmann et al . 1993 ; cole et al . 1994
; somerville & primack 1999 ; cole et al .
the sam approach lies in between the two methods described above .
it incorporates physical processes that are well - known on the basis of theoretical and observational studies , and parameterizes less well understood physical processes by simple functional forms . in this way
, a large number of physical processes can be implemented into sams to a certain degree of accuracy , including hierarchical growth of dark halos , shock heating of the intergalactic gas , radiative cooling of the hot gas in halos , disk formation from cold gas , star formation and its feedback , metal enrichment , starbursts , morphology transformations associated with the mergers of galaxies , and dust extinction .
the parameters involved in the model are determined by matching some well - established observations , e.g. , the tully - fisher relation , and the local luminosity functions of galaxies . with star formation explicitly included in the modeling and
the physical properties predicted for individual galaxies , sam is powerful in making model predictions that can be directly compared with observations . in the last two decades , a number of groups have used the semi - analytical approach to interpret observations to constrain theoretical models ( e.g. , white & rees 1978 ; white & frenk 1991 ; cole 1991 ; lacey & silk 1991 ; kauffmann et al .
1993 ; cole et al . 1994 ; kauffmann et al . 1999a , hereafter kcdw ; somerville & primack 1999 ; cole et al .
2000 ; nagashima et al . 2002 ; menci et al . 2002 ; benson et al .
2003 ; hatton et al .
2003 , helly et al .
2003 ) .
one advantage of the semi - analytic approach is that it can be combined with the merger trees of dark matter halos obtained directly from @xmath0-body simulations to produce model galaxy catalogs ( e.g. kcdw ; springel et al .
2001b , hereafter swtk ; hatten et al . 2003
; helly et al .
2003 ; de lucia et al .
2004 ) . since such catalogs contain information not only about the physical properties of individual galaxies , but also about galaxy distributions in the phase space , they are very useful in making comparisons between model predictions and observations , and in generating mock catalogs to quantify observational bias .
an example here is the gif galaxy catalogs produced by kcdw using their sam , which have been used to study the clustering properties of various types of galaxies at @xmath5 ( kcdw ) and at high redshift ( kauffmann et al .
1999b ) , to produce mock catalogs for the cfa2 redshift survey ( diaferio et al .
1999 ) , and to interpret the galaxy - galaxy lensing observation of the sdss ( yang et al .
2003 ) . despite the large effort ,
there are still outstanding problems in the sams .
the first problem is related to the fact that some important physical processes , such as the feedback from star formation , are still poorly understood , and so there is still a large freedom in the implementations of such processes .
secondly , even with the large number of models that have been investigated , so far we still do not have a single model that can match all the important observational constraints .
for example , most of the sams studied so far have difficulties in simultaneously matching the observed tully - fisher relation and the luminosity function of galaxies ; it is also difficult for the current sams to simultaneously account for the observed luminosity density at the present time and the star formation rate at @xmath6 as inferred from the sub - mm sources ( chapman et al .
2002 ; cowie et al .
2002 ; smail et al .
2002 ; borys et al . 2003 ; baugh et al . 2004 ) .
other problems , such as the faint - end slope of the luminosity function of galaxies , and the sharp break of the galaxy luminosity function at the bright end , still lack satisfactory solutions ( e.g. benson et al .
2003 ) . clearly , further investigations are required within the framework of the sam .
one significant recent development in the studies of structure formation is the finding that cdm halos are not smooth objects , but contain many subhalos instead ( e.g. jing et al . 1995
; ghigna et al . 1998 ; moore et al . 1999 ; klypin et al . 1999 ; swtk ) .
since galaxies may have formed in the centers of subhalos at high redshifts , the properties of the subhalo population may provide important clues about the formation and evolution of galaxies in dark matter halos .
furthermore , following the motion of subhalos can provide more precise information about the positions and velocities of the galaxies hosted by subhalos , which is important for the comparison between theoretical predictions and the observational results derived from redshift surveys .
the results about the subhalo population have yet to be fully incorporated into the sam .
one step in this direction has been taken by swtk and de lucia et al .
( 2004 ) who used their high - resolution halo simulations together with sam to study the formation and evolution of galaxies .
the halos in their simulations have masses typically of a cluster of galaxies , and are simulated with a multi - mass particle technique , so that subhalos are well resolved .
they identified subhalos in the simulation and followed explicitly the formation and evolution of galaxies in the subhalos with the sam .
one of the most interesting results they obtained is that the subhalo - based sam can significantly improve the agreement between the model prediction and the observation for the cluster galaxy luminosity function .
the main reason is that the merger time scale adopted in previous sams underestimated the merging time for bright galaxies ( with luminosity @xmath7 , the characteristic luminosity of the schechter form ) .
this underestimation leads to a reduced survival time for bright galaxies ( thus the number of bright galaxies ) , resulting in central galaxies that are too bright , and in a cluster luminosity function that is much less curved around @xmath8 than the observed one . in this paper
, we consider galaxy formation in a typical cosmological volume , combining sam with a set of high - resolution n - body simulations that can resolve subhalos in massive halos .
our aim is two - fold .
firstly , we want to examine whether resolving subhalos in massive halos can also improve model predictions for the luminosity function of the overall galaxy population , or one has to resolve low - mass halos in order to model the overall luminosity function correctly .
secondly , we want to use a variety of current observational results , including the multi - waveband luminosity functions and color distribution , to constrain the current sams and to identify issues for which further improvements of the model has to be made .
the arrangement of our paper is as follows . in section 2
, we present the simulations and briefly introduce our algorithm to find subhalos . in section 3 , we describe the main physical processes and how to implement them in our sam model . in section 4 ,
we test the model results by comparing with a handful of well - established observations of galaxies . in sec 5 , we summarize and discuss our main results .
the main simulation used in this paper is a @xmath9 cosmological simulation of @xmath10 particles in a box of @xmath11 .
the underlying cosmological model is the standard concordance model with the density parameter @xmath12 , the cosmological constant @xmath13 and the hubble constant @xmath14 in units of @xmath15 .
the initial density field is assumed to be gaussian with a harrison - zeldovich primordial power spectrum and with an amplitude specified by @xmath16 , where @xmath17 is the r.m.s .
fluctuation of the linearly evolved density field in a sphere of radius @xmath18 .
this simulation , which started at redshift @xmath19 , is evolved by 5000 time steps to the present @xmath5 with our vectorized parallel @xmath9 code ( jing & suto 2002 ) at the vpp5000 fujitsu supercomputer of the national astronomical observatory of japan .
the force softening length @xmath20 ( s2 type , hockney & eastwood 1981 ) is @xmath21 comoving , and the particle mass @xmath22 . because these simulation parameters are very similar to those adopted in many high - resolution re - simulations of individual cluster halos ( moore et al .
1999 , jing & suto 2000 , fukushige & makino 2001 , 2003 , power et al .
2003 , diemand et al 2004 ) , we have achieved a resolution that can resolve subhalos within massive halos . as we will see below ,
this is crucial for modeling correctly the mergers of galaxies in dark halos .
the over - merging problem , which exists in many previous studies of structures of dark matter halos and semi - analytic models of galaxy formation based on cosmological @xmath0-body simulations , is greatly alleviated in our model . as an illustration
, we show in figure [ fig : halo ] the density distribution of dark matter particles within the largest dark halo in the simulation .
as one can see , many subhalos remain intact during the evolution .
we have estimated the mass function of the subhalo population ( defined in [ sec : subhalo ] ) , and found a good agreement with the subhalo mass functions obtained in previous halo re - simulations ( e.g. swtk ) down to a mass of about @xmath23 ( 100 particles ) .
this simulation was used to derive the triaxial model for dark matter halos in jing & suto ( 2002 ) , to which we refer the reader for some complementary information about the simulation .
we will call this simulation l100 in the following .
although a box of @xmath11 usually can be regarded as a typical volume in the universe , the most massive clusters with masses larger than @xmath24 may still be under - represented , because of the wavelength cutoff beyond the simulation box .
therefore , we have carried out simulations separately for 20 such massive halos with the nested - grid @xmath9 code of jing & suto ( 2000 ) .
these massive halos are picked up from a cosmological simulation of @xmath25 particles in a box of @xmath26 , and they are the most massive halos in the simulation .
we do not impose any other criterion on the selection of these halos , and so the halo sample should be unbiased . in the re - simulations ,
each of the massive halos is represented by about @xmath27 particles within the virial radius , and so the particle mass is @xmath28 , comparable to the particle mass in l100 .
the force softening is also about @xmath21 , but the simulations have been evolved by 20000 time steps ( this choice of time steps was later found to be too conservative ) .
this sample will be used to populate massive halos with galaxies in a set of simulations of @xmath26 box in a subsequent paper .
here we only use the first halo @xmath29 to compare our results for cluster galaxies with observations and with those obtained by springel et al .
in addition to those simulations , we have also a cosmological simulation of @xmath25 particles in a box of @xmath30 , which has @xmath31 and is evolved with 5000 time steps . this simulation ( referred to as l25 in the following ) has a better mass resolution ( 8 times smaller in particle mass ) than the l100 simulation .
we will use this simulation to examine the mass resolution effect on the properties of galaxies at the faint luminosity end .
all the three sets of simulations have the same model parameters .
the l100 and cl simulations have 60 outputs from @xmath32 equally spaced in @xmath33 , and the l25 has 165 outputs . these outputs are used to construct the merging trees of dark halos .
the dark matter halos are identified in the simulations using the friends - of - friends method ( fof ) with a linking length equal to 0.2 of the mean particle separation .
the technique rarely breaks a physically bound halo into pieces ; rather , it may combine two nearby distinct halos into one halo in some cases , if there is a thin bridge of particles between the two halos .
furthermore , in our high resolution simulations , galactic halos falling into a more massive halo ( say , a cluster halo ) can survive the tidal disruption of the massive halo for a considerable period of time . as emphasized in swtk , it is important to follow the trajectories of these subhalos , not only because they can provide us with the position and velocity of the galaxies they host , but also because they allow us to accurately follow the merging histories of galaxies ( as discussed in 1 ) .
we identify the subhalos within the fof halos with the subfind routine of swtk , which kindly has been made available to us by volker springel .
the subfind routine finds self - bound halos and subhalos in a single output of a simulation . starting from a fof halo , the subfind first locates local overdense regions as subhalo candidates . for each particle in the input fof halo ,
the local density is computed using the common sph technique .
the smoothing length is taken to be the distance to the @xmath34-th nearest neighbor particle , and the density is smoothed by a kernel interpolation of the @xmath34 particles .
the particles are sorted in order of decreasing density , and are assigned to subhalo candidates sequentially .
when the @xmath35-th particle with density @xmath36 is considered , a set of particles @xmath37 is defined as the set of its @xmath38 nearest neighbors . in the set of @xmath37 particles , we select a subset of particles with density higher than @xmath39 and among those a set @xmath40
is selected as the set of the two closest particles .
particle @xmath35 is assigned to a subhalo candidate according to the set of particles @xmath40 .
if there is no particle in @xmath40 , particle @xmath35 is at the local maximum of the smoothed density field , and a new subhalo candidate is created around particle @xmath35 .
if there is only one particle @xmath41 in @xmath40 , the particle @xmath35 is likely within the subhalo that contains particle @xmath41 , and thus is assigned to that subhalo candidate .
if there are two particles @xmath41 and @xmath42 in @xmath40 , there are two possibilities .
either the two particles @xmath41 and @xmath42 are in two different subhalos . in this case
, the particle @xmath35 is said to be a _
point connecting these subhalos , and a new subhalo candidate is formed by joining these two subhalos with particle @xmath35 . or , as the second possibility , the two particles @xmath41 and @xmath42 are in the same subhalo . in this case ,
particle @xmath35 is simply added to this subhalo candidate . in this way ,
a set of subhalo candidates are created for the fof group .
then the particles in each subhalo candidate are checked whether they are bound in the center - of - mass reference frame , and the particles that have positive total energy are removed iteratively until all the particles are bound .
the bound subhalo candidates thus constructed establish a hierarchy of small subhalos in larger subhalos .
the subhalos that have only one saddle point are at the highest level of the hierarchy , and those with more saddle points are at a lower level of hierarchy . in this case
, one particle may be assigned more than once to subhalos at different levels of the hierarchy .
subfind keeps the identity of the particles only at the highest possible level of the hierarchy , thus producing a unique subhalo catalog .
the algorithm can easily deal with the subhalo - within - subhalo problem , as well as the problem of the fof algorithm that two halos with a thin bridge of particles are identified as one halo . with these subhalo catalogs at different outputs , we construct the merging trees for subhalos . in figure
[ submf ] , we show the mass functions of the subhalos in the l100 simulation ( upper panel ) and in the cluster halo re - simulations ( low panel ) . for the l100 simulation ,
the results are the average values for the host halos in the mass ranges indicated in the figure .
for the cluster halo re - simulations , the mass functions are for individual halos .
these mass functions are plotted down to a mass corresponding to 16 simulation particles .
the mass functions for the most massive halos in the l100 simulation and for the individual cluster re - simulations agree well with each other over the whole mass range . for comparison
we show the prediction of swtk by the thick solid line .
it was found that the mass function have a little steeper slope than that of swtk and close to the reported relation @xmath43 by diemand et al ( 2004 ) .
the subhalo mass function for halos with masses @xmath44 has a similar slope to that of cluster halos , but the amplitude is a factor of about 2 lower .
this decrease in amplitude with host halo mass is quantitatively in agreement with the recent results obtained by gao et al .
( 2004 ) based on a cosmological simulation that has slightly lower resolution than our l100 .
the figure also shows that the subhalo mass function breaks away from the power law , @xmath45 , for subhalos containing less than about 20 simulation particles in the l100 simulation , indicating that the mass function may be affected by the resolution effect at such a mass scale .
these results indicate that our l100 simulation can be used to probe subhalos that contain more than 50 simulation particles ( i.e. with masses @xmath46 ) .
for a subhalo with less than 50 particles , we use the l100 simulation to follow its merger history until it reaches the resolution limit , i.e. 10 particles .
we do not use the monte carlo method to follow the merger histories of low mass halos , but instead we will use the l25 simulation ( with a higher mass resolution ) to check the resolution effect . as we will show in section 4.1 ,
the resolution limit has little effect on the predicted luminosity function for @xmath47 , but it under - predicts the number count of faint galaxies ( @xmath48 ) by @xmath49 .
we implement the sam in our @xmath0-body simulations following swtk , but our model also includes the chemical evolutions of galaxies and the inter - galactic medium , as well as dust extinction effects on observational quantities , such as luminosity and color .
our model is therefore closer to the model adopted in a more recent publication from the same group ( de lucia et al .
2004 ) , although there are differences in the choice of model parameters .
for example , de lucia et al .
have chosen a high yield of metal production in order to match the observed high metallicity of the intracluster medium ( icm ) .
since the high metallicity leads to rapid gas cooling , they had to use a rather high feedback efficiency to suppress the formation of luminous galaxies . in our model
, we adopt a metal yield and a feedback efficiency that are more moderate . for completeness
, we briefly describe our sam in this section .
since our main treatments about galaxy formation are similar to swtk , we will compare in some detail our prescription and results with those in swtk and related investigations ( e.g. kcdw , de lucia 2004 ) . in the subhalo scheme ,
there are three distinct populations of galaxies that are associated with fof halos .
the largest subhalo in a fof halo is called the _
main _ halo , and the main halo always hosts a galaxy at its center , which we call the _ central _ galaxy .
the central galaxy is assumed to have the same position and velocity as the most bound particle of the main halo .
any other subhalo in the fof halo also hosts a galaxy at its center .
these galaxies , called _ halo _ galaxies , were central galaxies before their host halos fell into a larger halo .
they are assigned the positions and velocities of the most bound particles of their host subhalos .
the halo galaxies have a physical status very similar to that of the central galaxies , except that the _ central _ galaxies are fed by gas cooling flows , while the _ halo _ galaxies are not .
when two or more halos and/or subhalos merge , the smaller halo ( subhalo ) may be destroyed , thus losing its identity as a subhalo in the subfind algorithm .
a subhalo may also be tidally disrupted after passing through dense regions in the main halo .
a galaxy that was a central or a halo galaxy at an earlier epoch , but whose host ( sub)halos disappeared from the subfind catalog at some later time , is attached to the most bound particle of its host ( sub)halo at the time just before the ( sub)halo is disrupted .
such galaxies are called _ satellite _ galaxies .
the transformation of galaxies from one population to another is governed by the subhalo merging trees constructed from the subfind catalogs at different outputs of the simulation .
consider two successive outputs at redshifts @xmath50 and @xmath51 ( @xmath52 ) .
the subhalo @xmath53 at redshift @xmath54 is said to be a progenitor of a subhalo @xmath55 at a later time @xmath50 if more than half of the @xmath56 most - bound particles of @xmath53 are contained in @xmath55 . as in swtk
, we take @xmath56 to be 10 . after excluding those volatile subhalos at @xmath51 that are not a progenitor to any subhalo or fof group at the later time @xmath50 , each subhalo at @xmath51
must be a progenitor to one and only one subhalo at @xmath50 .
a subhalo at @xmath50 may have more than one progenitor at @xmath51 , and the most massive of them is referred to as the main progenitor . in the case that a subhalo at @xmath50 has no progenitor , the subhalo is usually the main halo of a fof group newly formed between @xmath50 and @xmath51 , and a _ central _ galaxy is created for the subhalo . for a subhalo at @xmath50 that has at least one progenitor , the _ central _ ( _ halo _ ) galaxy of its main progenitor at @xmath51
is updated to be its _ central _ ( _ halo _ ) galaxy , and all the other galaxies of the progenitor subhalos ( including the satellites of the main progenitor ) become the _ satellite _ galaxies of the subhalo at @xmath50 .
the satellite galaxies are assumed to merge later with the central ( halo ) galaxy of the subhalo after a dynamical friction timescale ( see [ sec : merger ] for details ) .
the evolution of galaxies is assumed to be driven by the physical processes to be described in detail in section 3.2 .
we solve the differential equations governing galaxy evolution using a time step smaller than the time interval between two successive simulation outputs .
typically , 20 steps are used for each interval between two successive outputs .
in passing , we mention that the virial mass @xmath57 of the main halo is defined as the mass enclosed by the virial radius @xmath58 , within which the mean density is @xmath59 times the critical density of the universe at the redshift @xmath1 in consideration .
@xmath59 can be calculated analytically from the spherical collapse model ( e.g. kitayama & suto 1996 ) .
here we use the fitting formula provided by bryan & norman ( 1998 ) , @xmath60 where @xmath61 , and @xmath62 is the density parameter at redshift @xmath1 . for a subhalo ,
the virial mass is simply the total mass of all the particles in the subhalo , and the dynamical time @xmath63 and virial velocity @xmath64 are kept at the values the subhalo had just before it merged into a bigger halo .
in addition to the formation of the halo / subhalo population , galaxy formation and evolution also involve many other physical processes . in this paper
, we attempt to take into account all processes that are known to be important for the formation and evolution of galaxies .
these include : ( 1 ) radiative cooling of the hot gas within the main halo ; ( 2 ) star formation in the cold gas , and energy feedback into the cold gas by supernova explosions ; ( 3 ) chemical evolution in the cold and the hot gas , and in stars ; ( 4 ) mergers of galaxies , star bursts , and the transformation of galaxy morphologies ; ( 5 ) spectro - photometric evolution of individual galaxies ; and ( 6 ) the effect of dust extinction on observational quantities . as in white & frenk ( 1991 ) , the hot gas within a halo
is assumed to be distributed in an isothermal sphere @xmath65 with a virial temperature @xmath66 .
the local cooling time at a given radius @xmath67 is @xmath68 where @xmath69 is the mean particle mass , @xmath70 the number density of electron at @xmath67 , and @xmath71 the cooling function , which depends both on the temperature @xmath72 and the metallicity @xmath73 of the hot gas .
we use the cooling functions tabulated in sutherland & dopita ( 1993 ) .
the chemical evolution of the hot gas is described in [ sec : chemical ] . as in kcdw
, we define the cooling radius @xmath74 as the radius at which the cooling time @xmath75 is equal to the age of the universe @xmath76 at redshift @xmath1 . if the cooling radius is smaller than the virial radius , the mass of gas cooled per unit time , i.e. the mass cooling rate , in the halo is approximated by @xmath77 where a dot denotes the derivative with respect to time @xmath78 .
for the isothermal sphere distribution considered here , the cooling rate can be written as @xmath79 where @xmath80 is the total hot gas in the halo .
note that this relation is derived under the assumption that the isothermal sphere is static .
it implies that all the gas within the cooling radius can cool in a time interval that is twice the cooling time . in what follows
we will assume this relation to hold during the evolution of a relatively massive halo , with @xmath80 being the total amount of hot gas at the time in consideration .
for small halos or halos at high redshift , the cooling time can be smaller than the age of the universe even at the virial radius .
in such halos , all the gas within the virial radius can cool , and so the mass cooling rate is limited by the rate at which gas is accreted into the system , rather than by radiative cooling . in this case , we write the mass ` cooling ' rate as @xmath81 where @xmath75 is to be specified below . note that for halos where @xmath82 , there is a jump of a factor of 2 between equation ( [ eq : rcoolsh ] ) and equation ( [ eq : rcoolsh1 ] ) , which is entirely due to the simple prescription we adopt . in reality , the transition must be smooth , which can be made by joining equations ( [ eq : rcoolsh ] ) and ( [ eq : rcoolsh1 ] ) with a smooth function . this does not lead to any noticeable effects , and so we will ignore such details .
we assume a universal baryon fraction @xmath83 for every halo , and the hot gas available for cooling is : @xmath84 \label{eq : mhot}\ ] ] where @xmath85 , and @xmath86 , @xmath87 , and @xmath88 are , respectively , the masses in stars , in cold gas , and ejected from stars .
the sum is over all the galaxies in the halo in consideration . because the ejected mass is assumed to be mixed into the hot gas in the current work , the term @xmath89 is set to be 0 in the above equation .
swtk defined the cooling radius by equating @xmath90 to the dynamical time of the halo , i.e. , @xmath91 . on the other hand ,
kcdm used @xmath92 , i.e. equation ( [ eq : rcoolsh ] ) for large halos where @xmath93 , while @xmath94 for smaller halos ( eq.[eq : rcoolsh1 ] ) .
note that the cooling rate given by @xmath95 is a factor of 2 to 4 higher than that given by @xmath96 , and so the two assumptions in general give quite different results for the galaxy population .
unfortunately , there is no good physical reason to prefer either of the two assumptions .
although radiative cooling is a well - understood process , the gas cooling rate in a halo depends sensitively on the assumed density and temperature profiles of the gas .
the assumption usually made in sams is that the gas is heated by gravitational collapse to a temperature close to the virial temperature of the dark halo and that the gas distribution in a dark halo is close to that of an isothermal gas in hydrostatic equilibrium in the halo potential .
however the heating and cooling of gas in a dark halo are more complicated than this assumption implies . on the one hand , detailed analytical models and hydrodynamic simulations
show that the halo gas in low - mass halos may never be heated to the virial temperature of the halo ( e.g. birnboim & dekel 2003 ; keres et al .
2004 ) , and so the amount of cold gas may be underestimated by the above formulae . on the other hand ,
if gas in protogalactic halos is preheated to some finite entropy , the gas distribution around a small dark matter halo may be much more extended , and the rate of gas cooling in such systems can be significantly reduced ( mo & mao 2002 ; oh & benson 2003 ) . it has been shown by some sams that the suppression of gas cooling in small halos by photoionization heating may have significant impact on the faint - end slope of the galaxy luminosity function ( e.g. somerville 2002 ; benson et al . 2002 ; benson et al .
2003 ) , but such models depend on the details of the implementation of the process . in this paper , instead of introducing such complicated physical models , we consider a simple model , in which @xmath75 is assumed to be equal to @xmath76 .
this assumption is consistent with the results of brinchmann et al .
( 2004 ) who have shown that the local star - forming population of galaxies have roughly constant star formation rates over a time scale comparable to the present age of the universe .
as we will see below , this assumption leads to sufficient suppression of star formation in low - mass halos .
it has been pointed out by kauffmann et al .
( 1993 ) that gas cooling ( or star formation ) must be strongly suppressed in the most massive halos so as not to produce too many super - luminous galaxies ( see also benson et al .
as we will see below , the assumption about this suppression has significant impact on the prediction of the number of bright blue galaxies . as in kauffmann
1993 ) , we consider a case in which gas cooling is simply switched off for halos with circular velocity larger than @xmath97 . since there is no solid physical motivation for this assumption , we will discuss another simple model for the shut - off of cooling in subsection 4.2 .
we model the star formation rate @xmath98 as : @xmath99 in the above equation , @xmath100 is the star formation efficiency , and @xmath101 is the dynamical time of the galaxy ( not the halo ) . as in swtk we use @xmath102 .
note that other choices , such as a constant @xmath101 , have also been used in the literature ( somerville & primack 1999 ) . for a satellite galaxy
, we keep the dynamical time @xmath101 to be equal to the value the galaxy had when it was last a central or halo galaxy .
the star formation efficiency @xmath100 is assumed to depend on the circular velocity of the galaxy ( cole et al .
1994 ; 2000 ) . following de lucia et al .
( 2004 ) , we take @xmath103 where @xmath104 and @xmath105 are treated as free parameters .
note that @xmath100 can not be larger than @xmath106 .
we take @xmath107 if the above equation gives a value larger than 1 .
note that @xmath108 gives a constant star formation efficiency , which is in fact consistent with the observations of high mass galaxies ( e.g. kennicutt 1998 ) . on the other hand
, there is evidence that low - mass galaxies convert cold gas into stars less efficiently than high - mass galaxies ( e.g. boissier et al .
2001 , kauffmann et al .
2003 ) . in this case
, @xmath105 should be positive . as shown by de lucia et al .
( 2004 ) , the assumed value of @xmath105 has significant impact on the predicted cold gas fraction and we note that @xmath105 also plays an important role in determining the metallicity and color of low - mass galaxies . in the present paper , we follow de lucia et al . (
2004 ) by setting @xmath109 . for a given stellar initial mass function ( imf ) ,
the energy released by supernovae associated with massive stars is @xmath110 for a unit solar mass of newly formed stars , where @xmath111 is @xmath112 , the typical kinetic energy of the ejecta from each supernova , and @xmath113 is an factor that depends on the form of the imf . for the scalo imf
( scalo 1986 ) , @xmath113 is about @xmath114 , and for the salpeter ( 1955 ) imf it is about @xmath115 ( note that both cases are assumed to apply in the mass range @xmath116 ) . the released energy is supposed to heat the cold gas and expel it from the cold gas disk . with an amount of @xmath117 of newly formed stars ,
the amount of cold gas that can be heated can be expressed as , @xmath118 where @xmath119 describes the feedback efficiency .
we assume that the cold gas is reheated to the virial temperature of the subhalo in which the galaxy resides .
there are at least two possibilities for the fates of the reheated gas .
the first is that the reheated gas always stays in the halo and is added to the hot gas for later cooling . the sam model based on this treatment of the reheated gas
is usually called the _ retention _ model .
another possibility is that the reheated gas is ejected from the host halo .
this ejected gas may be re - accreted by the halo after some time , say , one dynamical time of the halo .
the sam based on the this assumption is called the _ ejection _ model .
it has been shown by swtk that both the _ retention _ and the _ ejection _ models can be made to match most observations by adjusting model parameters , such as the star formation efficiency @xmath100 and the feedback parameter @xmath119 . in this paper
, we only consider the _ retention _ model .
note that in both retention and ejection cases , there is the issue about the state of the heated gas when it is re - accreted .
usually , the re - heated gas is assumed to have completely forgot its thermal history when it is re - accreted .
in reality , the heated gas may retain part of the entropy it gained and behave differently from the un - heated gas .
unfortunately , this process has not yet been implemented reliably in the current sams .
in the previous subsections , we have described the physical processes that govern the mass and metal exchange among the three main components , stars , cold gas , and hot gas , within a dark matter halo .
here we write down the equations describing these processes , so that we can follow the evolution of each individual component .
we assume that a fraction @xmath120 of the mass in newly formed stars is recycled into the cold gas through stellar winds .
for the scalo imf , @xmath121 , and for the salpeter imf , @xmath122 , according to the standard stellar population synthesis model ( cole et al .
thus , the change of the stellar mass in each galaxy is given by @xmath123 the cold gas in a galaxy is increased by radiative cooling of the hot gas , and is reduced by the formation of new stars and by supernova feedback .
thus , the change in the cold gas in each galaxy is given by @xmath124 in our model , @xmath125 for halo and satellite galaxies .
in the retention model adopted in this paper , all the cold gas reheated by the galaxies is retained in the main halo , thus the change in the hot gas in the main halo is @xmath126 where the summation is over all galaxies in the fof halo . now consider the metal contents in the three components .
we assume a yield @xmath127 of metal for the conversion of a unit mass of cold gas into stars .
all the metals produced are assumed to be returned to the cold gas instantaneously .
thus , the mass changes in heavy elements in the three components are respectively given by @xmath128 @xmath129 and @xmath130 where again the summation is over all the galaxies in the main halo , and @xmath131 in eq.([eq : coldmetal ] ) is set to be zero for all but the central galaxies .
we have used the mean metallicities , @xmath132 and @xmath133 as the metallicities for the cold gas in a galaxy , and for the hot gas in a main halo , respectively .
the ejected metals are assumed to have the same fate as the reheated gas . when solving the above set of differential equations , we divide the time between two successive snapshots typically into 20 steps . as mentioned above
, we keep the cooling rate @xmath131 at a constant level between the two snapshots . with our high - resolution simulations ,
we are able to follow the formation , mass accretion , and mergers of ( sub-)halos in detail . in our implementation of the sam , we deal with these processes according to the ( sub-)halo merger trees at each new snapshot of the simulation . for a newly formed halo of mass @xmath57 ,
there is hot gas of mass @xmath134 associated with it , and a new galaxy is being formed as the gas cools . the accreted mass is considered simply by updating the virial mass of the main halo ( as well as other properties , such as the virial radius , circular velocity ) and its hot gas in eq .
( [ eq : mhot ] ) .
when two or more ( sub-)halos merge , the hot gas , if any , in the smaller halos , is added to that of the main halo that contains them .
the galaxies in the smaller halos either become halo galaxies or satellite galaxies , depending on whether they are the _ central _ galaxies of the subhalos . the merger of a halo galaxy with another halo galaxy or with the central galaxy is explicitly traced by our subhalo merger trees . for a _ satellite _ galaxy where a subhalo is absent in the simulation , we assume that it merges with its _ central _ or _ halo _ galaxy after a dynamical friction time scale @xmath135 .
we use the simple formula given in binney & tremaine ( 1987 ) , @xmath136 to calculate @xmath135 . as shown by navarro et al .
( 1995 ) , this formula provides a good fit to their numerical simulation studying the fate of a satellite @xmath137 orbiting in a halo of circular velocity @xmath138 . in the above equation
, @xmath139 describes the dependence of @xmath135 on the orbit eccentricity @xmath119 , and is approximated by @xmath140 ( lacey & cole 1993 ) .
as in previous sams , we adopt the average value @xmath141 for @xmath139 ( e.g. tormen 1997 ) .
@xmath142 in the above equation is a constant approximately equal to 0.43 , and the coulomb logarithm @xmath143 is @xmath144 .
we take the satellite mass @xmath137 to be the virial mass of the galaxy when it was last a central or halo galaxy . at each new snapshot of the simulations ,
the merger clock is set up for the new _ satellite _ galaxies in each subhalo , according to equation ( [ eq : dft ] ) .
for those satellites whose host subhalos do not change between two successive snapshots , the merger clocks are kept unchanged . in each time step @xmath145 in solving the equations ( [ eq : mstars])([eq : hotmetal ] ) , the merger clock time is reduced by @xmath145 .
a _ satellite _
galaxy is merged to its _ central _ or _ halo _ galaxy when its merger clock time becomes negative , and then is deleted from the galaxy list .
the physical properties of the merger remnant are modeled according to the mass ratio of the two galaxies that merge together .
here the mass of a galaxy is referred to its stellar component only . when the mass ratio of the smaller galaxy to the bigger one in a merger is less than 0.3 , the merger is called a minor merger
; otherwise , it is a major merger . in a minor merger
, the cold gas of the smaller galaxy is put to the cold gas disk of the larger one , and its stellar mass is added to the bulge of the larger one . in a major merger , all the cold gas in the two galaxies is converted into stars instantly , and all the stellar mass in the disk and bulge components is transformed into a new bulge .
we model the photometric evolution of galaxies using the stellar population synthesis ( sps ) model of bruzual & charlot ( 1993 ) .
we use their recent sps library ( in 2000 ) that includes the metallicity effect on the photometric properties .
the spectral energy distribution of a galaxy can be computed through @xmath146 where @xmath147 is the spectral energy distribution for a single - age population of stars , which depends on the imf , metallicity and age of the formed stars .
the model of bruzual & charlot ( 1993 ) is used to produce tables for a single instantaneous burst of a unit stellar mass in which the metallicity changes from @xmath148 to @xmath149 .
we interpolate the tables using a linear interpolation in @xmath78 and @xmath150 , and calculate the magnitudes for both the sdss filters and the standard johnson ubv filters . as in cole et al .
( 2000 ) , we introduce a free parameter , @xmath151 , to take into account the fact that part of the newly formed stars may be invisible brown dwarfs ( @xmath152 ) .
the mass fraction of the visible stars @xmath151 is defined through @xmath153 by definition , @xmath151 can not be larger than @xmath106 .
the inclusion of brown dwarfs makes the predicted luminosity smaller by a factor of @xmath151 . in this paper
, we either assume @xmath154 , or when necessary tune the value of @xmath151 to match the observed luminosity function in the sdss @xmath35 band .
the dust extinction model we use is similar to that of kcdw .
we assume that the optical depth in the @xmath155-band scales with luminosity as @xmath156 with @xmath157 , @xmath158 and @xmath159 , as given by wang & heckman ( 1996 ) .
we use the model of cardelli et al .
( 1989 ) to derive the ratio @xmath160 . for a thin disk in which stars and dust grains have the same distribution , the total galactic extinction in magnitude
is given by tully & fouqu@xmath161 ( 1985 ) : @xmath162 where @xmath163 is the inclination angle between the disk and the line of sight . in our model
we assume that @xmath163 has a random distribution between @xmath164 and @xmath165 .
we apply the above equation only for the disk components , and we assume that there is no dust in the bulge component . as we will see , the inclusion of dust extinction improves the luminosity function in the bright end , especially for the blue band .
note , however , the dust extinction adopted here applies only to quiescent disk galaxies , but not to dust - enshrouded star burst galaxies .
thus , in our model , all star burst galaxies have very blue colors , although in reality they may be rather red due to dust extinction . in the present paper we do not include dust extinction for starburst galaxies , and we should keep this in mind when comparing model predictions with observations .
our sam involves the following set of free parameters : * imf : the initial stellar mass function . in this paper
, we adopt the salpeter imf for the major part of our discussion .
we also consider the scalo imf as a comparison ; * @xmath127 : the yield of metals from a unit mass of newly formed stars ; * @xmath120 : the fraction of mass recycled into the cold gas by evolved stars ; * @xmath166 : the amplitude of the power - law star formation efficiency ; * @xmath105 : the slope of the power - law star formation efficiency ; * @xmath119 : the feedback efficiency ; * @xmath151 : the fraction of newly formed stellar mass that contributes to the luminosity of a galaxy .
note that some of the above - listed free parameters are not independent of each other .
for example , for a given imf , the parameter @xmath120 is determined by stellar evolution theory . for the scalo imf , @xmath120 is about @xmath167 , and for the salpeter imf , @xmath120 is about 0.35 .
we will use these values for @xmath120 . the yield @xmath127 given by stellar evolution theory
is between 0.01 and 0.02 for both the scalo and the salpeter imfs , but the model prediction is still quite uncertain . in practice , we constrain @xmath127 using the color - magnitude relation ( cmr ) of cluster ellipticals , since the cmr is sensitive to the changes in the metallicity and the luminosities and colors of elliptical galaxies are less affected by dust . to fix the other four parameters , we use the following two observational inputs : the sdss galaxy luminosity function in the @xmath35-band , and the cold gas content of the milky way .
the luminosity in the @xmath35-band is neither sensitive to the recent star formation rate nor to the dust extinction , and so can be used quite reliably as an indicator for the total stellar mass .
thus , both the star formation efficiency and the feedback efficiency are expected to have a significant impact on the @xmath35-band luminosity function and the cold gas content of the milky way . as in swtk , we require the predicted cold gas in milky way type galaxies to be about 8@xmath168
. we will always assume @xmath154 ( i.e. the mass locked in brown dwarfs is negligible ) , unless a value of @xmath151 less than 1 has to be introduced in order to match the observed @xmath35-band luminosity function ( only in fig.[lf_cooling ] ) .
we have adjusted the model parameters by trials and tests .
we find that although the model predictions can be affected in complicated ways by changing model parameters , the parameter space allowed by the observations is actually quite limited . in table 1
, we list the parameters that best match the observational inputs .
these values of parameters are in good agreement with those adopted in many earlier papers on sams ( e.g. de lucia et al . 2004 ) . in what follows
we compare model predictions and observational results based on various properties of the galaxy population .
in this section , we compare our model predictions with observations about the galaxy population at @xmath169 . these observations include the multi - waveband luminosity functions of galaxies obtained from the sdss and the 2dfgrs ; the @xmath170-band tully - fisher relation ; the cmr of elliptical galaxies in clusters and in the general field ; the metallicity of galaxies as a function of luminosity or rotation velocity ; and the cold gas fraction in galaxies of different luminosities ; note that some of the comparisons are not independent tests , because they are used to calibrate the model parameters . as we will show , there is general agreement between model and observation for various properties of the galaxy population .
this is encouraging , because it indicates that the basic assumptions made in the model may be not far from reality .
however , we also find significant discrepancies between model prediction and observation , suggesting that further improvement has to be made in the sam .
blanton et al . ( 2003a ) estimated the luminosity functions of galaxies for the sdss survey in the five sdss wavebands ( @xmath171 , @xmath172 , @xmath67 , @xmath35 and @xmath1 ) .
since the median redshift of the galaxies in their sample is about 0.1 , they made the @xmath173-correction and evolution correction to the reference frame at @xmath174 , and shifted the bandpasses by a factor of 1.1 .
for comparison with these results , we shift the sdss filters to the blue by a factor 1.1 when calculating the luminosities of individual galaxies . following the sdss convention ,
the corresponding wavebands are denoted by @xmath175 for the @xmath171-band , @xmath176 for the @xmath172-band , and so on .
the upper panels and the lower left one of fig.[lf ] present the predicted luminosity functions in the @xmath177 bands , with the solid histogram showing the result for the salpeter imf , and the dotted histogram showing that for the scalo imf .
the thick solid lines show the schechter function fits to the corresponding sdss luminosity functions ( blanton et al .
note that the agreement at the @xmath178-band is similar to that of @xmath179 band , and we have omitted the plot .
we note that these schechter function fits provide quite accurate description for the observational data , except at the brightest ends ( @xmath180 ) , where they underestimate the observed number of galaxies .
as one can see from the figure , the predicted luminosity functions match well the observed ones in the red bands . in the lower right panel of fig.[lf ] , we show the predicted k - band luminosity function . the smooth solid curve in this panel shows the result of the schechter function fit obtained by kochanek et al .
( 2001 ) from the 2mass .
this result is similar to that obtained by cole et al .
as we can see , the reasonably good agreement between model prediction and observation also extends to the @xmath173-band , in which galaxy luminosity traces the stellar mass .
thus , the sam considered here can successfully reproduce the observed stellar mass distribution in the universe . as a further test
, we compare our model prediction with the observational result of madgwick et al .
( 2002 ) for the 2df galaxy luminosity function at the @xmath181 band at @xmath5 ( see the lower middle panel of fig.[lf ] ) . in our model
, we estimate the @xmath182-band magnitude of a galaxy using the relation @xmath183 ( blair & gilmore 1982 ) , where @xmath155 and @xmath184 are the rest - frame vega magnitudes in the @xmath155- and @xmath184- wavebands .
since the observed 2df luminosity function has a steeper faint - end slope than the sdss luminosity function in the blue - band , it is better matched by the model prediction at the faint end than is the sdss @xmath172-band luminosity function . from now on
, our discussion about the galaxy luminosity function will be based on results for the @xmath175 , @xmath176 , @xmath185 and @xmath179 bands .
compared with previous predictions of the field luminosity functions based on the sams ( e.g. , kcdw ) , our model gives a much better match to the observed luminosity functions .
for example , kcdw significantly overpredicted the number of galaxies brighter than @xmath186 and underestimates the number of fainter galaxies .
consequently , the luminosity functions predicted in kcdw are much flatter around @xmath186 than the schechter function .
a similar discrepancy was also revealed for the galaxy luminosity function in rich clusters by swtk , when a treatment of the galaxy - galaxy mergers similar to kcdw was adopted .
they attributed the discrepancy to the assumption made in kcdw on the dynamical friction time scale for satellite galaxies . with the use of their resolved subhalos in clusters
, swtk found that the satellite galaxies brighter than @xmath186 in a halo merge into the central galaxy at a rate much slower than that given by the dynamical friction time scale of equation ( [ eq : dft ] ) ( used in kcdw ) .
for the same reason , our model can reproduce well the observed field galaxy luminosity functions at the bright end , because subhalos hosting bright galaxies are well resolved in our @xmath0-body simulations . to demonstrate this ,
we have considered a model in which the merger time scales of satellite galaxies are estimated from equation ( [ eq : dft ] ) rather than using the subhalo scheme . as shown by figure [ lf_sdss_nosub ] ,
the luminosity functions predicted by such a model are much flatter than the observational results and significantly over - predict the number of bright galaxies . although the luminosity function at the bright end can be reduced by adjusting model parameters , for example by reducing the star formation efficiency , this will reduce the luminosity function at @xmath187 , making a large discrepancy at the intermediate luminosity .
it is the shape of the luminosity function that can not be easily adjusted to match the observations in this model . in our subhalo scheme ,
we have used equation ( [ eq : dft ] ) to estimate the merger time for the satellites that are currently not associated with subhalos .
it is therefore possible that the merger time here is also under - estimated .
we have tested whether the uncertainties of the merger time in such galaxies can have a significant impact on our model predictions , by increasing or decreasing the merger time by a factor of 2 .
the results are shown in fig.[lf_sdss_merge ] .
little change was found in the model luminosity functions .
the reason is that the merger time of satellites in these galaxies is relatively long , because of the small mass ratio between the satellite galaxy and the central galaxy , and so mergers are unimportant anyway . for the bright end
, the change of the merger timescale has only a minor impact , because the mass of a satellite galaxy is in general quite small compared to the bright galaxies . in fig.[lf_cluster ] , we compare the predicted cluster luminosity function with the composite luminosity function of rich clusters in the 2dfgrs by de propris et al .
( 2003 ) . here
we use one of our high - resolution cluster simulations , c1 ( other clusters give similar results ) . because the composite lf given by de propris et al is not properly normalized for a given cluster mass
, we adjust the amplitude of the observed lf to best match our model cluster lf and compare only the shape of the lf .
the agreement is quite satisfactory .
the absolute magnitude of the central galaxy in halo c1 is @xmath188 in the subhalo scheme , compared with @xmath189 in the case where galaxy mergers are not based on subhalos .
the subhalo scheme also predicts more sub-@xmath186 galaxies , in better agreement with the observation than the merger scheme solely based on the dynamical friction time scale .
this result for the cluster lf is very similar to those obtained by swtk and de lucia et al .
( 2004 ) , both using the subhalo scheme similar to the one used here . as in the previous works , our result demonstrates that resolving subhalos is an important ingredient to reproduce the exponential ( schechter ) form at @xmath186 of the composite cluster luminosity function . as emphasized above
, the main reason for the success of our model in reproducing the luminosity function of field galaxies is that we are able to follow the trajectories of massive galaxies using subhalos in our high - resolution simulations .
this is an important result , because it implies that the shape of the observed galaxy luminosity function can be understood if galaxy mergers are modeled in a realistic manner .
there is another factor , which also contributes to the success of our model , namely the cooling time scale we have adopted , @xmath96 for all halos .
if we followed swtk in using @xmath94 , we would obtain a much higher amplitude for the luminosity function , because the cooling rate for massive halos at redshift @xmath164 is about a factor of @xmath190 higher . in this case
, we have to multiply the luminosity of each galaxy by @xmath191 , i.e. to assume that about @xmath192 of the stellar mass is locked into brown dwarfs to best match the observed lf ( see the dotted lines in fig.[lf_cooling ] ) .
this required @xmath151 is too small to be realistic ( zhao et al .
1996 , somerville & pirmack 1999 ) .
also , with the assumption @xmath94 , the model predicts more galaxies at the fainter end , making faint - end slope of the luminosity function of cluster galaxies too steep .
as one can see , our model still predicts more faint galaxies ( e.g. @xmath193 in the g - band ) than are present in the sdss observations . detailed analysis of these over - predicted faint galaxies show that most of them are halo ( satellite ) galaxies in massive halos . here
we want to check whether the model prediction is significantly affected by numerical resolution .
to do this , we applied the same sam to the simulation l25 , which has a mass resolution about ten times higher than l100 .
fig.[lf_sdss_box ] shows the results for the l25 simulation ( dotted lines ) in comparison to the results of l100 ( solid lines ) .
the resolution improvement does not seem to has any significant effect ( at most @xmath194 ) on the predicted luminosity function at the faint end with @xmath195 . as we have shown above , the predicted faint end of the luminosity function
is also insensitive to the assumed dynamical friction scale for satellite galaxies .
the other factor we have examined is the feedback efficiency .
we have considered a model assuming a feedback efficiency @xmath196 instead of @xmath149 , and found that such a strong feedback can bring the number of faint galaxies into agreement with the observed lf , but the number density of galaxies with @xmath197 is reduced too much to be compatible with the observation .
it is difficult to quantify how serious the faint end problem is . on the observational side ,
the faint - end slope of galaxy luminosity function is hard to determine , and there are considerable discrepancies among different sky surveys ( lcrs , esp , sdss edr , 2df , sdss dr1 , etc .
see driver & de propris 2003 ) .
our model prediction is in good agreement with the steep faint - end slopes observed in the 2df survey and in the esp survey , but it is out of our scope to examine in which survey the faint galaxies are missed or / and misidentified .
yang et al .
( 2003 ) have measured the mass - to - light ratio of dark halos in the @xmath198 band from the 2dfgrs using the halo model approach .
they found that in terms of this ratio , galaxies are formed most efficiently in the halos of mass @xmath199 , and the galaxy formation is strongly suppressed both in lower and in higher mass halos .
such suppressions are expected in our semi - analytical model , because the hot gas cools less efficiently in cluster halos while the feedback effect can effectively heat up the cold gas and hence reduce the star formation in low - mass halos .
it is interesting to see if our model prediction is in quantitative agreement with the measurement of yang et al .
this comparison is displayed in figure [ mass_light ] . except for @xmath200 , where the model prediction is limited by the simulation resolution and the observational measurement
is limited by the small number statistics of galaxies , the agreement between the model and the observation is reasonable .
based on the results obtained above for the @xmath171-band luminosity function , we see that our model predicts too many bright blue galaxies and insufficient number of blue galaxies with luminosities just below the characteristic luminosity .
our detailed analysis of the model prediction ( with dust extinction ) shows that at @xmath174 the brightest population in the @xmath171-band ( with @xmath201 brighter than @xmath202 ) are dominated by starbursts that have just experienced a major merger .
most of the starbursts are in halos with mass around @xmath203 . as we have mentioned before ,
the model prediction of the blue luminosities of these galaxies is quite uncertain , because our model does not include properly dust extinctions in dust - enshrouded starburst galaxies .
the predicted excess of number of galaxies at lower luminosities ( @xmath201 between @xmath202 and @xmath204 ) is mainly due to central galaxies in dark halos with masses around @xmath205 ( see fig.[galaxy_color_mvir ] ) . in our model
, hot gas in such halos is allowed to cool continuously according to equation ( [ eq : rcoollh ] ) , although gas cooling is prohibited in more massive halos ( see [ ssec : cooling ] ) .
the predicted excess of such blue galaxies is therefore due to the inadequate treatment of suppressing gas cooling in massive halos .
it is possible to solve this problem by considering more realistic feedback processes in massive halos .
the cause of the deficit of blue galaxies with luminosities below @xmath206 is unclear ( also present in fig.[cm_field ] ) .
one possibility is that the model assumes that the star formation rate in a quiescent disk is smooth with time , while in reality it may be intermittent , consisting of small bursts of star formation ( kauffmann et al .
such a change does not affect the total amount of stars that can form , and so does not change the luminosity functions in the red bands , but may produce more blue galaxies at the present time , as some of the low - mass systems may make excursions to the bursting phase at the present time .
another possibility is that the imf is top - heavy , so that the fraction of uv - emitting massive stars is increased with respect to that of the salpeter imf .
clearly , further investigations along these lines are required to see if they are viable solutions to the problem . as shown in baldry et al .
( 2004a , 2004b ) , sdss galaxies seem to exhibit a bimodal distribution in the color - magnitude diagram , in the sense that early - type and late - type galaxies show distinct colors for a given magnitude .
to see if such a distribution can be reproduced in our model , we show in the upper left panel of fig.[cm_field ] the @xmath207 color as a function of @xmath208 for our model galaxies at redshift @xmath164 .
_ central , halo _ and _ satellite _ galaxies are represented by black crosses , cyan squares and red triangles , respectively .
the thick solid lines show respectively the fit results to the means of the gaussian color distributions for the red and blue branches obtained by baldry et al .
( 2004a ) . for comparison with the results of blanton
( 2003b ) , we also show in the upper right panel the @xmath209 color distribution at @xmath174 ( the solid histogram ) , which is less dependent on dust extinction than the @xmath210 color . as in the observations , we see clearly two distinct branches in the case where subhalo scheme is used to follow the mergers of galaxies , with the blue branch more prominent than the red one . in the red branch ,
most galaxies are early - type galaxies .- band bulge - to - disk ratio and the hubble - type @xmath72 found by simien & de vaucouleurs ( 1986 ) .
following swtk , we call galaxies with @xmath211 ellipticals , those with @xmath212 s0 s , and those with @xmath213 spirals and irregulars . ] the mean @xmath207 color in this branch is around @xmath214 for faint galaxies ( @xmath215 ) , and becomes progressively redder for brighter galaxies .
this prediction is very close to the observational result , although the predicted red branch has a larger dispersion and contains smaller number of very bright red galaxies . for the blue branch ,
the predicted mean @xmath210 color for galaxies fainter than @xmath216 is about @xmath217 , slightly redder than the observational result ( @xmath218 ) .
however , the model predicts too many bright galaxies in the blue branch .
this problem is the same as one see in the @xmath171-band luminosity function shown in fig.[lf ] .
the lower right panel shows the @xmath210 color versus stellar mass .
the behavior seen here is similar to the color - magnitude relation .
it is instructive to examine what are the main factors that determine the predicted bimodal color distribution .
we have made tests by changing the prescriptions of gas cooling , dust extinction and chemical enrichment .
these changes can have some effects on the shape of the predicted color - magnitude relation , but do not change the bimodal behavior . on the other hand ,
if we use the old scheme instead of the subhalo scheme to describe the galaxy - galaxy mergers , the predicted bimodal feature is much weaker ( see the lower left panel of fig.[cm_field ] and the dashed histogram in the upper right panel of the same figure ) .
note that in the upper left panel many of the bright galaxies in the red branch are halo galaxies , in which most of the cold gas has turned into stars and current star formation is at a low level , while in the old scheme no halo galaxies exist . the difference between the new and the old schemes can be understood as follows . in the subhalo scheme ,
galaxy mergers in a halo are allowed not only with the central galaxy , but also among halo and satellite galaxies .
since gas cooling is prohibited in halo and satellite galaxies , such mergers involve only evolved stellar population , producing new galaxies that are red .
consequently , many galaxies can stay red in the subhalo scheme . on the other hand , in the old prescription of galaxy - galaxy merging , galaxies in a halo merge relatively fast only with the central galaxy that may still have fresh cold gas . as a result
, the number of red galaxies is reduced and the distinction between the red and blue populations is blurred .
thus , our ability to follow the evolution of subhalos seems to be the key in producing the observed bimodal color distribution .
baldry et al . ( 2004a ) show in their figs.3 and 4 that most of the brightest galaxies are red , that the red branch becomes less prominent and the blue branch becomes more prominent as luminosity decreases . to compare our model prediction with their results ,
we show in fig.[color_strips ] the color distributions for galaxies in different luminosity bins .
overall , the predicted color distributions are similar to the observed ones , and in most cases the color distribution resembles the sum of double gaussian . however , there is marked discrepancy between model and observation . at the high - luminosity end ( @xmath219 ) , the predicted colors in the blue branch are too blue , and the predicted red branch is too weak .
we believe that this failure is related to our simplified treatment of the gas cooling in massive halos . as in many previous works ( cf .
kauffmann et al .
1993 ) , we cut off the gas cooling in very massive halos with the circular velocity above a given value ( @xmath220 in our paper ) .
although there is strong observational evidence that the cooling should be suppressed in massive halos , there is no compelling reason why the gas cooling should be shut off according to the circular velocity .
in fact , the number of the predicted most luminous red galaxies is very sensitive to this assumption .
if we adopt a toy model in which we shut off the gas cooling according to the halo mass ( instead of the circular velocity ) , we find that we can produce more brightest red galaxies and reduce the number of the brightest blue galaxies , while still keeping the luminosity function nearly unchanged ( kang et al . 2005 ) .
a more physical model is needed for suppressing the gas cooling in massive halos .
on the other hand , one also sees that for @xmath221 , model galaxies in the blue branch have colors slightly redder than the observed galaxies .
the observed blue colors imply that these galaxies have newly formed stellar population ( baldry et al .
2004b ; kauffmann et al .
this may indicate that the treatment of star formation in our model is inadequate , and the starbursts triggered by minor mergers may also be an important missing ingredient for producing more blue galaxies . in our model that includes dust extinction ,
the predicted colors for galaxies with @xmath222 are still bluer than observed .
the intrinsic colors of these galaxies are very blue , with @xmath223 , as shown in fig.[cm_field_nodust ] . in order to see which type of galaxies contribute to the over - predicted blue galaxies , here we make an census of the galaxy population in the blue branch . in our model without dust , about @xmath224 percent of the very brightest galaxies with @xmath225 are recent star - burst galaxies in dark halos with virial mass around @xmath203 .
( these are ellipticals because they are associated with major mergers ) . among the bright galaxies with @xmath226 , about @xmath227
are e and s0 type galaxies according to the mass ratio of the merging progenitors and the fraction of star - burst galaxies is decreased to about @xmath228 percent . as we have discuss in
[ sec : dust ] that we apply dust extinction correction to the galaxy with a prominent disk , so the ignore of dust reddening of star - burst galaxies is not reasonable for many observations show that dust correction should be important in those systems .
the bright blue galaxies are still over - predicted even though the star - burst galaxies are extracted .
this is for there are still cold gas available for new stars forming in the e , s0 and late type galaxies .
we also found that the late type galaxies with bigger disks were also over - predicted compared with the morphology - classified k - band luminosity function of 2mass ( kochanek et al .
2001 ) . in order to see the origin of the problem
, we examine where these blue bright galaxies are located . in fig.[galaxy_color_mvir ] we therefore plot the host halo mass for the blue ( @xmath229 ) and red ( @xmath230 ) galaxies in our l100 simulation .
the main population of the blue bright galaxies are the central galaxies in halos of the velocity dispersion about @xmath231 . in our model
the gas cooling are still allowed in the halos and therefore the recent newly formed stars make their colors bluer .
this is due to the inadequate treatment of cooling in massive halos , as discussed in the last subsection .
note that there is a feature in fig.[galaxy_color_mvir ] , also in fig.[cm_field_nodust ] .
this is caused by the discontinuity of the gas cooling described by equation ( [ eq : rcoolsh ] ) and equation ( [ eq : rcoollhalo ] ) .
baldry et al .
( 2004b ) and balogh et al .
( 2004 ) have shown that the color bimodality also depends on the environment defined by the local density of galaxies .
the main properties of their observed color distribution can be summarized as follows : firstly , for a given luminosity , the fraction of galaxies in the red sequence increases with local density . secondly ,
faint galaxies are predominantly blue , except in very dense environments where a large fraction is also red . finally , the most luminous galaxies are all red , independent of local environment . in our model , the first two properties are well reproduced .
however , our model predicts a too strong blue branch in low and mediate density environments .
the main reason for this discrepancy is that our model predicts too many luminous blue galaxies , because of the inadequate treatment of gas cooling in massive halos and because of the neglect of dust attenuation in massive starburst galaxies .
since in the subhalo scheme halo- and satellite- galaxies can survive longer before merging with the central galaxies , our model predicts a large number of bright red galaxies at high redshift . in fig.[red_galaxy ] , we show the distribution of galaxies with respect to the observed - frame @xmath232 color at @xmath233 . here
only bright galaxies with m@xmath234 were considered ; the luminosity limit is chosen to match that of the great observations origins deep survey ( goods ) sample used by somerville et al .
( 2004 ) for our adopted cosmology . as shown by the solid histogram in fig.[red_galaxy ] , about @xmath227 of our sample galaxies
have @xmath235 .
if we do not use the subhalo scheme to follow the galaxy - galaxy mergers , but use the conventional scheme based on dynamical friction , only @xmath236 of the bright galaxies have @xmath237 ( see the dotted histogram ) . based on multi - waveband photometries in deep fields and follow - up spectroscopic observations , such as the k20 survey ( cimatti et al .
2002b ) , the goods ( giavalisco et al . 2004 ) , and the gemini deep deep survey ( gdds , abraham et al . 2004 ) , it is now possible to identify a large number of bright galaxies at high redshifts .
such observations indicate that the number of bright red galaxies at @xmath238 may be larger than earlier predictions based on sam ( e.g. cimatti et al .
2002a ; somerville et al .
2004 ; glazebrook et al .
our results show that resolving subhalos may help alleviating this problem .
we will make a detailed comparison of our model predictions with the observations at high redshift in a separate paper . before leaving from this subsection ,
let us consider the color - magnitude relation for the elliptical galaxies in clusters . in fig.[cm_cluster
] , we present the predicted color - magnitude relation for the elliptical galaxies in the cluster simulation c1 , and compare it with observation .
we identify elliptical galaxies as the early - type galaxies with @xmath239 in the b band , and we show the predicted color - magnitude relation for individual model galaxies as solid triangles .
the solid line shows the fit to the observation of bower et al .
( 1992 ) for elliptical galaxies in the coma cluster .
as one can see , the observed trend is well reproduced in our model .
the main reason for the color - magnitude relation in the model is the metallicity effect : more luminous galaxies have redder colors , because they have higher metallicity .
this result is consistent with that obtained by de lucia et al .
( 2004 ) . in the upper panel of fig.[tf ]
, we plot the tully - fisher relation ( hereafter tf relation ) of the model sb / sc field galaxies against the observation of giovanelli et al .
( 1997 ) at the i band : @xmath240 the best fit of the observational result is shown as the solid line , while the dashed lines show the 1-@xmath241 scatter . to select sb / sc galaxies from our sam , we use the correlation between the @xmath155-band bulge - to - disk luminosity ratio and the hubble - type given by simien & de vaucouleurs ( 1986 ) .
we select sb / sc galaxies according to the criterion , @xmath242 .
we consider only the central galaxies with such hubble types in the _ main _ halos .
the velocity width @xmath243 is set to be 2 times the maximum circular velocity of the disk , and we have assume that the disk maximum circular velocity is @xmath244 larger than the circular velocity of the halo at its virial radius .
this boost of disk maximum circular velocity is expected in a galaxy halo with typical concentration @xmath245 assuming disk mass is negligible ( e.g. mo , mao & white 1998 ) .
the triangles in fig.[tf ] are the results for our model sb / sc galaxies defined in this way .
the figure shows that the scatter predicted by the model is significantly smaller than in the observational result .
note that in our modeling we have not taken into account the scatter in the relation between the line width and the circular velocity , neither have we included the observational errors in photometry and errors due to dust correction .
both can produce scatter in the tf relation .
overall , the predicted tf slope agrees quite well with the observed one , but the predicted luminosity for a given disk maximum circular velocity is lower than observed .
note that if dark halo responds to disk growth adiabatically , then the boost in the disk maximum circular velocity is expected to be larger than what is assumed above , making the discrepancy between model prediction and observation even large .
this problem with the tully - fisher relation in the current @xmath2cdm model has been known earlier and is due to the fact that galaxy halos predicted by this model are too concentrated ( e.g. mo & mao 2000 and references therein ) .
one possible solution to this problem is that some dynamical processes during the formation of galaxies in dark halos can flatten dark matter halos ( e.g. mo & mao 2004 ) , so that the boost in the disk maximum circular is reduced . indeed ,
if the boost is about 10% , then the predicted tully - fisher amplitude can match the observation .
garnett ( 2002 ) studied the correlation between the metallicity of the interstellar gas in a galaxy with the luminosity and rotation velocity of the galaxy , for a sample of spiral and irregular galaxies . in fig.[metal ] , we compare our model predictions with his results .
we select a sample of galaxies from our sam with @xmath246 , corresponding to spiral and irregular galaxies according to our definition .
the metallicity of the interstellar gas is obtained using the chemical evolution model described in [ sec : chemical ] .
the figure shows that the observed correlation is well reproduced by our model .
the vertical dashed line in the lower panel shows the rotation velocity at which the slope of the metallicity - rotation velocity relation changes significantly , as pointed out by garnett . in fig.[coldgas ]
, we show the cold gas fraction as a function of galaxy luminosity . here
again we compare our model prediction with the observational results of garnett ( 2002 ) .
we define the cold gas fraction as @xmath247 . as one can see , the predicted cold gas fraction is larger in fainter galaxies , which is consistent with the observed trend , and with the theoretical results obtained by boissier et al .
( 2001 ) and kauffmann et al .
( 2003 ) . in our model ,
the predicted cold gas depends sensitively on the parameter @xmath105 in equation ( [ eq : sfr2 ] ) .
note that @xmath108 gives a constant gas fraction for all the spiral galaxies , a result obviously in conflict with the observation .
a positive @xmath105 means that smaller galaxies have a lower star formation rate , which also leads to the metallicity trends shown in fig.[metal ] .
in this paper , we have modeled galaxy formation in a series of high - resolution @xmath0-body simulations of the standard @xmath2cdm model using the semi - analytical approach .
one unique aspect of our analysis is that our model uses subhalos resolved in the @xmath0-body simulations to follow the mergers of galaxies in dark halos .
the semi - analytical model we adopted for galaxy formation in dark matter halos is similar to those used in recent semi - analytical studies .
we found that , if galaxy - galaxy mergers are followed by the merger histories of subhalos , rather than based on dynamical friction time scales , the semi - analytical model can match well the luminosity functions of galaxies in the bright end in various wavebands redder than the u - band .
our model also reproduces the main properties of the observed bimodal color distribution better than that model that do not resolve subhalos , although it still predicts too many bright blue galaxies . resolving subhalos
also produces many more bright red galaxies at high @xmath1 , in better agreement with the observations .
once the model parameters are calibrated , the model can also match the observed metallicity - luminosity relation and metallicity - rotation velocity relation of spiral galaxies , the gas fraction in present - day spiral galaxies , and the color - magnitude relation for elliptical galaxies in clusters .
overall , the model is very successful in explaining a wide variety of observational facts .
this is encouraging , because it means that our understanding of galaxy formation in the standard @xmath2cdm model is on the right track .
however , there are several important issues that remain unresolved .
the first is the process that can effectively suppress star formation in low - mass halos . in this paper , instead of assuming an unrealistically high feedback efficiency , we have considered a model in which the gas cooling time in a dark halo is comparable to the age of the system .
with such an assumption the faint - end slope of the luminosity function can be moderated .
although this assumption leads to predictions that can match the observational results better , and a prolonged cooling is expected if the intergalactic medium is preheated ( mo & mao 2002 , 2004 ; oh & benson 2003 ) , the details of this process are still unclear .
the second is concerned with the large number of bright blue galaxies predicted by the model . to address this issue in detail
, we need a realistic dust model for starburst galaxies , as well as a realistic model for the gas cooling in massive halos .
the third problem is that the current model predicts insufficient number of blue galaxies with intermediate luminosities , which indicates that either the imf in disk galaxies is top - heavier than the salpether imf , or star formation in disk galaxies is intermittent .
finally , like many earlier analyses based on sam , we found that the predicted tully - fisher zero point is too low , unless there are some processes that can significantly flatten cdm halos .
clearly , further investigations about the possible solutions are required in order to resolve these issues .
we thank volker springel for providing the subfind code , roberto de propris for providing his data in electronic format , stephane charlot and xu kong for advice in spectral synthesis model .
we also thank simon white and shude mao for stimulating discussions , and an anonymous referee for constructive comments .
this work is partly supported in part by nkbrsf(g19990754 ) , by nsfc(nos.10125314 , 10373012 , 10203004 ) , and by the cas - mpg exchange program .
hjm and gb is supported by the cas famous scholar program .
abel , t. , anninos , p , zhang , y. , norman , m. , 1997 , newa , 2 , 181 abraham , r. g. , et al . , 2004 , aj , 127 , 2455 baldry , i. k. , glazebrook , k. , brinkmann , j. , ivezic , z. , lupton , r. h. , nichol , r. , c. , szalay , a. s. , 2004a , apj , 600 , 681 baldry , i. k. , balogh , m. l. , bower , r. , glazebrook , k. , nichol , r. c. , 2004b , aip conf.proc .
743 , 106 balogh , m. l. , baldry , i. k. , nichol , r. c. , miller , c. , bower , r. , glazebrook , k. , 2004 , apj , 615l , 101 baugh , c. m. , lacey , c. g. , frenk , c. s. , granato , g. l. , silva , l. , bressan , a. , benson , a. j. , cole , s. , 2004 , mnras in press benson , a. j. , cole , s. , frenk , c. s. , baugh , c. m. , lacey , c. g. , 2000 , mnras , 316 , 107 benson , a. j. , lacey , c. g. , baugh , c. m. , cole , s. , frenk , c. s. , 2002 , mnras , 333 , 156 benson , a. j. , bower , r. g. , frenk , c. s. , lacey , c. g. , baugh , c. m. , cole , s. , 2003 , apj , 599 , 38 berlind , a. a. , weinberg , d. h. , 2002 , apj , 575 , 587 berlind , a. a. , weinberg , d. h. , benson , a. j. , baugh , c. m. , cole , s. , et al . , 2003 , apj , 593 , 1 binney , j. , tremaine , s. , 1987 , `` galactic dynamics '' , princeton university press , princeton birnboim , y. , dekel , a. , 2003 , mnras , 345 , 349 blair , m. , gilmore , g. , 1982 , pasp , 94 , 742 blanton et al . 2003a , apj , 592 , 819 blanton et al . 2003b , apj , 594 , 186 boissier , s. , boselli , a. , prantzos , n. , gavazzi , g. , 2001 , mnras , 321 , 733 bond , j. r. , kaiser , n. , cole , s. , efstathiou , g. , 1991 , apj , 379 , 440 borys , c. , chapman , s. c. , halpern , m. , scott , d. , 2003 , mnras , 344 , 385 bower , r. g. , lucey , j. r. , ellis , r. s. , 1992 , mnras , 589 , 600 brinchmann , j. , charlot , s. , kauffmann , g. , white , s. , tremonti , c. , heckman , t. brinkmann , j. , 2004 , mnras , 351 , 1151 bruzual , a. g. , charlot , s. , 1993 , apj , 405 , 538 bryan , g. l. , cen , r. y. , norman , m. l. , ostriker , j. p. , stone , j. m. , 1994 , apj , 428 , 405 bryan , g. l. , norman , m. l. , 1998 , apj , 495 , 80 bullock , j. s. , wechsler , r. h. , somerville , r. s. , 2002 , mnras , 329 , 246 calura , f. , matteucci , f. , menci , n. , 2004 , mnras , 353 , 500 cardelli , j. a. , clayton , g. c. , mathis , j. s. , 1989 , apj , 345 , 245 cen , r. , ostriker , j. p. , 1993
, apj , 417 , 415 chapman , s. c. , scott d. , borys , c. , fahlman , g. g. , 2002 , mnras , 330 , 92 cimatti , a. , et al . , 2002a , a&a , 391 , 1 cimatti , a. , et al .
, 2002b , a&a , 392 , 395 coil , a. l. et al . , 2004 , apj , 609 , 525 cole , s. , 1991 , apj , 367 , 45 cole , s. , arag@xmath248n - salamanca , a. , frenk , c. s. , navarro , j. f. , zepf , s. e. , 1994 , mnras , 271 , 781 cole , s. , lacey , c. g. , baugh , c. m. , frenk , c. s. , 2000 , mnras , 319 , 168 cole , s. , et al . , 2001 , mnras , 326 , 255 colless , m. , et al . , 2001 , mnras , 328 , 1039 cooray , a. , sheth , r. , 2002 , phr , 372 , 1c couchman , h. m. p. , thomas , p. a. , pearce , f. r. , 1995 , apj , 452 , 797 cowie , l. l. , barger , a. j. , kneib , j .- p . , 2002 , aj , 123 , 2197 de lucia g. , kauffmann , g. , springel , v. , white , s. d. m. , 2004 , mnras , 349 , 1101 de propris et al .
, 2003 , mnras , 342 , 725 diaferio , a. , kauffmann , g. , colberg , j. m. , white , s. d. m. , 1999 , mnras , 307 , 537 diemand , j. , moore , b. , stadel , j. , 2004 , mnras , 352 , 535 driver , s. p. , de propris , r. , 2003 , apss , 285 , 175 keres , d. , katz , n. , weinberg , d. h. , dave , r. , 2004 , preprint ( astro - ph/0407095 ) efstathiou , g. , 1992 , mnras , 256 , 43 eke , v. r. , navarro , j. f. , frenk , c. s. , 1998 , apj , 503 , 569 fukushige , t , makino , j. , 2001 , apj , 557 , 533 fukushige , t , makino , j. , 2003 , apj , 588 , 674 gao , l. , white , s. d. m. , jenkins , a. , stoehr , f. , springel , v. , 2004 , 355 , 819 garnett , d. r. , 2002 , apj , 581 , 1019 ghigna , s. , moore , b. , governato , f. , lake , g. , quinn , t. , stadel , j. , 1998 , mnras , 300 , 146 giavalisco , m. , et al . , 2004 ,
apj , 600l , 93 giovanelli , r. , haynes , m. p. , da costa l. n. , freudling , w. , salzer , j. j. , wegner , g. , 1997 , apj , 477 , 1 glazebrook , k. , blake , c. , economou , f. , likky , s. , colless , m. , 1999 , mnras , 306 , 843 glazebrook , k. , et al .
, 2004 , nature , 430 , 181 hatton , s. , devriendt , j. , ninin , s. , et al .
2003 , mnras , 343 , 75 helly , j. c. , cole , s. , frenk , c. s. , baugh , c. m. , benson , a. , lacey , c. , 2003 , mnras , 338 , 903 hockney , r. w. , eastwood , j. w. , 1981 , computer simulation using particles ( mcgraw hill new york ) hughes , d. h. , et al . , 1998 , nature , 394 , 241 jing , y. p. , mo , h. j. , brner , g. , fang , l. z. 1995 , mnras , 276 , 417 jing , y. p. , mo , h. j. , brner , g. , 1998 , apj , 494 , 1 jing , y. p. , suto , y. , 2000 , apj , 529 , 69 jing , y. p. , suto , y. , 2002 , apj , 574 , 538 kang , x. , jing , y. p. , mo , h. j. , brner , g. , 2005 , in preparation katz , n. , gunn , j. e. , 1991 , apj , 377 , 365 kauffmann , g. , white , s. d. m. , guiderdoni , b. , 1993 , mnras , 264 , 201 kauffmann , g. , charlot , s. 1998 , mnras , 294 , 705 kauffmann , g. , colberg , j. m. , diaferio , a. , white , s. d. m. , 1999a , mnras , 303 , 188 ( kcdw ) kauffmann , g. , colberg , j. m. , diaferio , a. , white , s. d. m. , 1999b , mnras , 307 , 529 kauffmann , g. , heckman , t. m. , white , s. d. m. , charlot , s. , tremonti , c. , peng , e. w. , seibert , m. , brinkmann , j. , nichol , r. c. , subbarao , m. , york , d. , 2003 , mnras , 341 , 54 kennicutt , r. c. , 1998 , apj , 498 , 541 keres , d. , katz , n. , weinberg , d. h. , dave , r. , 2004 , preprint , astro - ph/0407095 kitayama , t. , suto , y. , 1996 , apj , 469 , 480 klypin , a. a. , gottber s. , kravtsov , a. v. , khokhlov , a. m. , 1999 , apj , 516 , 530 kochanek , c. s. , pahre , m. a. , falco , e. e. , huchra , j. p. , mader , j. , jarrett , t. h. , chester , t. , cutri , r. , schneider , s. e. , 2001 , apj , 560 , 566 kravtsov , a. v. , berlind , a. a. , wechsler , r. h. , klypin , a. a. , gottloeber , s. , allgood , b. , primack , j. r. , 2004 , apj , 609 , 35 lacey , c. , silk , j. , 1991 , apj , 381 , 14 lacey , c. g. , cole , s. , 1993 , mnras , 262 , 627 madgwick et al .
2002 , mnras , 333 , 133 maller , a. h. , mcintosh , d. h. , katz , n. , weinberg , m. d. , 2004 , preprint , astro - ph/0304005 mathis , h. , lemson , g. , springel , v. , kauffmann , g. , white , s. d. m. , eldar , a. , dekel , a. , 2002 , mnras , 333 , 739 menci , n. , cavaliere , a. , fontana , a. , giallongo , e. , poli , f. , 2002 , apj , 575 , 18 mo , h. j. , mao , s. , white , s. d. m. , 1998 , mnras , 295 , 319 mo , h. j. , mao , s. , 2000 , mnras , 318 , 163 mo , h. j. , mao , s. , 2002 , mnras , 333 , 768 mo , h. j. , mao , s. , 2004 , mnras , 353 , 829 moore , b. , ghigna , s. , governato , f. , lake , g. , quinn , t. , stadel , j. , tozzi , p. , 1999 , apj , 524 , 19 nagashima , m. , yoshii , y. , totani , t. , gouda , n. , 2002 , apj , 578 , 675 nakamura , o. , fukugita , m. , brinkmann , j. , schneider , d. p. , 2003 , preprint , astro - ph/0312519 navarro , j. , white , s. d. m. , 1994 , mnras , 267 , 401 navarro , j. f. , frenk , c. s. , white , s. d. m. , 1995 , mnras , 275 , 56 oh , s. p. , benson , a. , 2003 , mnras , 342 , 664 peacock , j. a. , smith , r. e. , 2000 , mnras , 318 , 1144 peacock , j. a. , et al .
, 2001 , nature , 410 , 169 power , c , navarro , j. f. , jenkins , a. , frenk , c. s. , white , s. d. m. , springel , v. , stadel , j. , quinn , t. , 2003 , mnras , 338 , 14 pozzetti , l. , et al .
, 2003 , a&a , 402 , 837 press , w. h. , schechter , p. , 1974 ,
apj , 187 , 425 salpeter , e. e. , 1955 , apj , 121 , 161 scalo , j. m. , 1986 , fundamentals of cosmic physics , 11 , 1 schneider , p. , ehlers , j. , falco , e. e. , 1999 , gravitational lenses , a&a library .
springer verlag : berlin seljak , u. , 2001 , mnras , 325 , 1359 simien , f. , vaucouleurs , g. , 1986 , apj , 302 , 564 skrutskie m.f .
, et al . , 1997 , in the impact of large - scale near - ir surveys , eds .
f. garzon et al .
kluwer academic publishing company dordrecht , 1997 , p25 smail , i. , ivison , r. j. , blain , a. w. , kneib , j .-
, 2002 , mnras , 331 , 495 somerville , r. s. , primack , j. r. , 1999 , mnras , 310 , 1087 somerville , r. s. , 2002 , apj , 572 , 23 somerville , r. s. , et al . , 2004 ,
apj , 600l , 135s spergel , d. n. , et al .
2003 , apjs , 148 , 175 springel , v. , & hernquist , l. 2003 , , 339 , 289 springel , v. , yoshida , n. , white , s. d , m. , 2001a , newa , 6 , 79 springel , v. , white , s. d. m. , tormen , g. , kauffmann , g. , 2001b , mnras , 328 , 726 ( swtk ) springel , v. , di matteo , t. , hernquist , l. , 2004 , preprint , astro - ph/0409436 sutherland , r. s. , dopita , m. a. , 1993 , apjs , 88 , 253 tegmark et al . , 2004 , phrvd , 69 , 103501 thoul , a. a. , weinberg , d. h. , 1995 , apj , 442 , 480 tormen g. , 1997 , mnras , 290 , 411 tully , r. , fouqu@xmath161 , p. , 1985 ,
apjs , 58 , 67 vale , a. , ostriker , j. p. , 2004 ,
mnras , 353 , 189 van den bosch f.c . , yang x. , mo h.j . , 2003 , mnras , 340 , 771 van den bosch f. c. , yang x. , mo h.j . , norberg p. , 2004
, preprint , astro - ph/0406246 wang , b. , heckman , t. m. , 1996 , apj , 457 , 645 wechsler , r. h. , somerville , r. , bullock , j. s. , kolatt , t. s. , primack , j. r. , blumenthal , g. r. , dekel , a. , 2001 , apj , 554 , 85 weinberg , d. h. , katz , n. , & hernquist , l. 1998 , in origins , eds . j. m. shull , c. e. woodward , & h. thronson ( asp conference series ) white , s. d. m. , rees , m. j. , 1978 , mnras , 183 , 341 white , s. d. m. , frenk , c. s. , 1991 , apj , 379 , 52 yan , r. , madgwick , d. s. , white , m. , 2003 , apj , 598 , 848 yang , x. , mo , h. j. , van den bosch , f. c. , 2003 , mnras , 339 , 1057 yang , x. , mo , h. j. , kauffmann , g. , chu , y. q. , 2003 , mnras , 339 , 387 york , d. g. , et al . , 2000 ,
aj , 120 , 1579 yoshikawa , k. , jing , y. p. , suto , y. , 2000 , apj , 535 , 593 zhao , h. s. , rich , r. m. , spergel , d. n. , 1996 , mnras , 282 , 175 zheng , z. , et al . , 2004 , preprint , astro - ph/0408564 | we model the galaxy formation in a series of high - resolution n - body simulations using the semi - analytical approach . unlike many earlier investigations based on semi - analytical models , we make use of the subhalos resolved in the @xmath0-body simulations to follow the mergers of galaxies in dark halos , and
we show that this is pivotal in modeling correctly the galaxy luminosity function at the bright end and the bimodal nature of galaxy color distribution .
mergers of galaxies based on subhalos also result in many more bright red galaxies at high @xmath1 .
the semi - analytical model we adopt is similar to those used in earlier semi - analytical studies , except that we consider the effect of a prolonged cooling in small halos and that we explicitly follow the chemical enrichment in the interstellar medium .
we use our model to make predictions for the properties of the galaxy population at low redshift and compare them with various current observations .
we find that our model predictions can match the luminosity functions of galaxies in various wavebands redder than the u - band .
the shape of the luminosity function at bright end is well reproduced if galaxy mergers are modeled with the merger trees of subhalos and the steep faint - end slope can be moderated if the gas cooling time in low - mass halos is comparable to the age of the universe . the model with subhalos resolved can reproduce the main features in the observed color bimodal distribution , though it still predicts too many bright blue galaxies .
the same model can also match the color - magnitude relation for elliptical galaxies in clusters , the metallicity - luminosity relation and metallicity - rotation velocity relation of spiral galaxies , and the gas fraction in present - day spiral galaxies .
we also identify areas where further improvements of the model are required . |
the theoretical description of weakly bound exotic nuclei close to the drip - line is one of the most exciting challenges today .
what makes this subject both particularly interesting and difficult , is the proximity of the particle continuum implying strong modification of the effective nucleon nucleon interaction and causing unusual spatial properties of the nucleon density distribution ( halo structures , large diffusivity ) .
many of those nuclei are involved in the chain of thermonuclear reactions and , in the absence of data at relevant energies , the models of stars rely to certain extent on calculated astrophysical factors ( see bahcall ( 1989 ) ) . in weakly bound exotic systems ,
the number of excited bound states or narrow resonances is small and , moreover , they couple strongly to the particle continuum .
hence , these systems should be described in the quantum open system formalism which does not artificially separate the subspaces of ( quasi- ) bound ( the @xmath2-subspace ) and scattering ( the @xmath3-subspace ) states . for well bound nuclei close to the @xmath4-stability line
, microscopic description of states in the first subspace is given by nuclear sm with model - space dependent effective two - body interactions , whereas the latter subspace is treated in terms of coupled channels equations . in this work ,
we question the validity of this basic paradigm of nuclear physics , and propose its modification for weakly stable exotic nuclei by taking into account coupling between @xmath2 and @xmath3 subspaces in terms of residual nucleon nucleon interaction .
this coupling modifies the scattering solutions as well as the spectroscopic quantities for interior bound states . as said before , we are interested in describing low lying bound and quasi - bound states in exotic nuclei .
for that reason , we can restrict description of particle continuum to the subset of one nucleon decay channels .
still , in few rare cases of two - nucleon halo nuclei , this limitation may turn out to be restrictive . in any case ,
further improvement of our model to more complicated channels like , e.g. , @xmath5 - channels , can be done as well ( balashov ( 1964 ) ) .
the influence of scattering continuum on the mean - field properties has been addressed before ( see dobaczewski ( 1996 ) for review ) .
no such analysis has been done so far for the realistic sm .
a possible starting point could be the continuum shell model ( csm ) approach ( fano 1961 , mahaux and weidenmller ( 1969 ) , philpott ( 1977 ) ) , which in the restricted space of configurations generated using the finite - depth potential , has been studied for the giant resonances and the radiative capture reactions probing the microscopic structure of these resonances ( barz ( 1977,1978 ) , fladt ( 1988 ) ) .
this is insufficient for nuclei close to drip lines , where it is essential to have a most realistic description of bound state subspace .
for that reason , the corner - stone of our approach , called the shell model embedded in the continuum ( smec ) , is the _ realistic sm _ itself which is used to generate the @xmath6-particle wavefunctions .
this choice implies that the coupling between sm states and the one - particle scattering continuum must be given by the residual interaction . in our case
, we use the residual interaction in the form : @xmath7 with @xmath8 and @xmath9 .
the key element of both smec and csm is the treatment of single - particle resonances , which on one side may have an important amplitude inside a nucleus and , on other side , they exhibit asymptotic behaviour of scattering wavefunctions ( bartz 1977 ) . the part of resonance for @xmath10 , where @xmath11 is the cut - off radius , is included in @xmath2 subspace , whereas the remaining part is left in the @xmath3 subspace .
the wavefunctions of both subspaces are then renormalized in order to ensure the orthogonality of wavefunctions in both subspaces . in the smec calculations , we solve identical equations as in the csm ( bartz ( 1977 ) ) but due to specificity of exotic nuclei , ingredients of these calculations
are modified . for the bound states we solve the sm problem : @xmath12 , using the code antoine ( caurier ( 1989 ) ) .
@xmath13 is _ identified _ with the realistic sm hamiltonian and @xmath14 are the @xmath6-particle ( quasi- ) bound wavefunctions .
the quasi - bound resonances in the continuum are included as well . 0.8 for the continuum part
, we solve the coupled channel equations : @xmath15 where index @xmath16 denotes different channels and @xmath17 .
the sign @xmath18 characterizes the boundary conditions , i.e. , whether we consider incoming ` @xmath19 ' or outgoing ` @xmath20 ' scattering waves . in our case , we have ingoing wave in the input channel and outgoing waves in all channels .
the structure of @xmath21 - nucleus is given by the sm , whereas one nucleon occupies a scattering state .
the channel states are defined by coupling one nucleon in the continuum to a ` hole state ' of @xmath21 - nucleus .
the sm wavefunction has an incorrect asymptotic behaviour for states in the scattering continuum .
on the other hand , the standard approach consisting of adjusting values of the two - body matrix elements to the experimental data and/or modifying the monopole term of the effective interaction ( dufour and zuker ( 1996 ) ) , makes the definition of average single - particle field and the radial dependence of the single - particle wavefunctions somewhat arbitrary .
therefore , to generate both single - particle resonances and the radial formfactors of occupied orbits entering the coupling matrix elements ( [ force ] ) between states in the subspaces @xmath2 and @xmath3 , we use the finite - depth average potential of saxon - woods type with spin - orbit part included .
the parameters of the average potential are fitted to reproduce experimental single - particle states , whenever their identification is possible .
this is the initial choice of potential , because the microscopic coupling of bound and scattering states generates the correction term to the mean - field which modifies the single - particle wavefunctions for each @xmath22 of the quasi - bound state .
this correction term must be carefully taken into account otherwise , the solutions including coupling of @xmath2 and @xmath3 subspaces could be non - orthogonal .
the third system of equations are the coupled channel equations : @xmath23 with the source term @xmath24 , which is given by the sm structure of @xmath6 - particle wavefunction for state @xmath14 .
these equations define functions @xmath25 which describe the decay of quasi - bound state @xmath14 in the continuum .
the source @xmath24 couples the wavefunction of @xmath6 - nucleon localized states with @xmath21 - nucleon localized states + one nucleon in the continuum .
the formfactor of the source term is given by the same average potential as used in the @xmath3-space ( [ esp ] ) .
the residual coupling ( [ force ] ) is also identical .
the complete solution can be expressed by means of three functions : @xmath14 , @xmath26 and @xmath27 : @xmath28 where @xmath29 is the effective sm hamiltonian including the coupling to the continuum and @xmath30 is the green function for the motion of single particle in @xmath3 .
matrix @xmath31 is symmetric and complex .
it can be diagonalized by the orthogonal transformation : @xmath32 with complex eigenvalues @xmath33 .
the eigenvalues of @xmath31 at energies @xmath34 determine the energies and widths of resonance states . with this changes ,
one obtains : @xmath35^{-1 } < { \tilde { \phi}_i } \mid h \mid { \xi}_{e}^{c } > \ , \end{aligned}\ ] ] for the wavefunction of continuum state modified by the discrete states and : @xmath36 for the wavefunction of discrete state modified by the continuum states .
as an example of smec calculations for bound and resonant states , let us consider the proton - rich nucleus @xmath0b .
1 compares sm spectrum for @xmath38 states of @xmath0b calculated in the @xmath39 - shell using the cohen and kurath ( 1965 ) ( ck ) interaction , with those obtained in the smec for two different strength @xmath40 of the residual interaction ( [ force ] ) . in the smec calculations , the saxon - woods average potential with parameters : @xmath41 mev , @xmath42 mev , @xmath43 fm and @xmath44 was used as a starting guess .
the spectrum of @xmath0b does not show large sensitivity to the choice of these parameters .
this initial average potential is then modified for each @xmath22 state by the coupling to the continuum .
this correction make the average potential deeper , it increases slightly the diffusivity and produceses a maximum at the center of the potential .
it is the corrected potential which is then used to calculate the radial formfactors of coupling matrix elements and single - particle wavefunctions .
the coupling to the continuum mixes the unperturbed sm states .
for example for @xmath45 mev@xmath46@xmath47 in ( [ force ] ) , the wavefunction for the first @xmath48 resonance has an overlap of 0.992 , @xmath49 and @xmath50 with unperturbed sm states @xmath48 , @xmath51 and @xmath52 , respectively .
the calculated width of @xmath48 state is 17 kev and 29 kev for @xmath53 and 1200 mev@xmath46@xmath47 respectively .
the experimental width for this state is @xmath54 kev .
one should notice that also the energy difference between @xmath55 and @xmath48 states is improved by inclusion of the coupling to the continuum . the @xmath57 decay of @xmath0b , which is formed by the reaction @xmath58 at the center of mass ( c.m . )
energy of about 20 kev , is the main source of high energy solar neutrinos . in the absence of data in this region ,
the input of standard model for the solar neutrino problem ( bahcall and ulrich ( 1988 ) ) should be compared with those of 0.95 various calculations . in the smec , the initial wavefunction @xmath59_{j_i^{\pi}=1^{+}})$ ] is : @xmath60^{j_a}\times\chi^{i_t}\biggr]^{(j_i ) } _ { m_i } \end{aligned}\ ] ] and the final wavefunction @xmath61_{j_f^{\pi}=2^{+}})$ ] is : @xmath62^{j_b}\times\chi^{i_t}\biggr]^{(j_f ) } _ { m_f } \ .\end{aligned}\ ] ] @xmath63 and @xmath64 denote the spin of target nucleus and incoming proton , respectively .
@xmath65 is the coefficient of fractional parentage and @xmath66 is the s.p .
wavefunction in the many - particle state @xmath67 . with the wavefunctions
@xmath68 and @xmath69 , we calculate the transition amplitudes : @xmath70 for @xmath71 and @xmath72 and : @xmath73 \cr \hfill + \mu_t ( -1)^{j_f - j_i } \hat i_t
\tilde i_t w(i_t j_b 1 j_i j_f i_t ) \delta_{j_a j_b } \biggr\ } \delta_{l_a l_b } i_{l_a j_a , l_b j_b}^{0,j_i } \cr}\end{aligned}\ ] ] for @xmath74 transitions , respectively . in the above formula , @xmath75 and @xmath76 .
the radiative capture cross section can then be expressed in terms of those amplitudes as : @xmath77 @xmath78 @xmath79 stands for the reduced mass of the system .
2 shows the calculated multipole contributions to the total capture cross section and the astrophysical @xmath80-factor as a function of the c.m .
the calculation is done for @xmath81 mev@xmath82@xmath47 .
the proton threshold energy is adjusted to agree energies of calculated and experimental @xmath83 state .
the photon energy is given by the difference of c.m .
energy of @xmath84_{j_{i}=1_{1}^{+}}$ ] system and the experimental energy of the @xmath55 ground state of @xmath0b .
the value of the @xmath80-factor at c.m .
energy of 20 kev : @xmath85 ev@xmath46b , agrees with certain older experiments ( parker ( 1966 ) , kavanagh ( 1969 ) ) but disagrees with the most recent one of hammache ( 1998 ) .
the ratios of smec @xmath80 factors at different c.m .
energies are : @xmath86 and @xmath87 .
the ratio of @xmath74 and @xmath71 contribution is : @xmath88 , @xmath89 and @xmath90 at 20 , 100 and 500 kev , respectively .
the resonant part of @xmath74 transitions , which yields the contribution of @xmath91 ev@xmath46b at the @xmath48 resonance energy , decreases fast and becomes @xmath92 , @xmath93 , @xmath94 ev@xmath46b at c.m .
energies 500 , 100 , 20 kev , respectively .
this value is proportional to the square of spectroscopic amplitude of @xmath39-states , which for the ck interaction is @xmath95 and @xmath96 for @xmath97 and @xmath98 respectively .
similar small values of spectroscopic amplitudes are obtained for kumar ( 1974 ) and ptbme ( 1992 ) interactions ( see brown ( 1996 ) ) . at the position of @xmath83 resonance ,
the predicted by smec calculations value of @xmath80-factor ( @xmath99 ev@xmath46b ) is smaller than seen in some experiments ( kavanagh ( 1969 ) , vaughn ( 1970 ) , filippone ( 1983 ) ) .
part of this discrepancy could be related to the absence of quenching factor in our calculations .
0.95 fig .
3 shows the same as fig . 2 but for @xmath45 mev@xmath82@xmath47 .
now , the calculated value of the @xmath80 factor at c.m .
energy of 20 kev is @xmath100 ev@xmath46b , rather close to the value obtained for @xmath53 mev@xmath46@xmath47 , and disagrees with the recent experimental value given by hammache ( 1998 ) .
one should stress however , that different experiments show significant fluctuations in the extrapolated values of @xmath80 .
the ratios of @xmath80 factors at different c.m .
energies : @xmath86 and @xmath101 , show much stronger dependence on energy than seen for @xmath53 mev@xmath46@xmath47 .
the resonant part of @xmath74 transitions at the @xmath48 resonance energy , yields a somewhat smaller value : @xmath102 ev@xmath46b , than for @xmath53 mev@xmath46@xmath47 . finally ,
at the position of @xmath83 resonance , the predicted by smec calculations value of @xmath80-factor ( @xmath103ev@xmath46b ) is somewhat smaller than seen in the experiments and for @xmath53 mev@xmath46@xmath47 .
in this work we have shown results of first calculations using the smec which couples the realistic sm solutions for ( quasi- ) bound states with the scattering solutions of one - particle continuum .
the application to @xmath104b reaction yields satisfactory description of different components of the radiative capture cross section , including the resonant components . in future
, more unstable nuclei should be studied in the smec approach to systematically address the problem of effective interactions in the extreme conditions of exotic nuclei . at present
, we are applying the smec to many other reactions of astrophysical interest such as @xmath105 , @xmath106 , @xmath107 .
authors would especially like to thank s. drod and i. rotter for many clarifying discussions .
we are grateful to e. caurier and a. lefebvre for encouragement and useful discussions .
the work was partly supported by kbn grant no .
2 p03 b 14010 and the grant no .
6044 of the french - polish cooperation .
bahcall j n , _ neutrino astrophysics _ ( cambridge university press , 1989 ) . | we present here the first application of realistic shell model ( sm ) including coupling between many - particle ( quasi-)bound states and the continuum of one - particle scattering states to the spectroscopy of @xmath0b and to the calculation of astrophysical factors in the reaction @xmath1 . epsf |
we show how the shell structure of open - shell nuclei can be described in a fully self - consistent extension of the covariant energy density functional theory .
the approach implies quasiparticle - vibration coupling ( qvc ) in the relativistic framework being an extension of the ref .
@xcite for superfluid systems @xcite .
medium - mass and heavy nuclei represent fermi - systems where single - particle and vibrational degrees of freedom are strongly coupled .
collective vibrations lead to shape oscillations of the mean nuclear potential and , therefore , modify the single - particle motion . as a result ,
single mean - field states split into levels occupied with fractional probabilities which correspond to spectroscopic factors of these fragments .
the dyson equation is formulated in the doubled quasiparticle space of dirac spinors and solved numerically for nucleonic propagators in tin isotopes which represent the reference case : the obtained energies of the single - quasiparticle levels and their spectroscopic amplitudes are in excellent agreement with data , see fig .
[ sps](a ) .
sn : relativistic mean field ( rmf , left column ) , qvc ( center ) and experimental data ( right ) . in the qvc and exp cases only the dominant levels are shown .
( b ) single - quasiparticle strength distribution for the orbits around the fermi surfaces in the neutron ( left panels ) and proton ( right panels ) subsystems of the z=120 isotopes calculated in the relativistic quasiparticle - vibration coupling model .
the dashed lines indicate the neutron chemical potentials.,width=566 ] because of high universality of the approach it can be applied to nuclei at the limits of their existence with respect to their proton and neutron numbers , for instance , to superheavy nuclei .
selected results on the single - quasiparticle strength distributions in the neutron and the proton subsystems of the z = 120 isotopic chain are displayed in fig .
[ sps](b ) .
one can see the evolution of these distributions with an increase of the neutron number from n = 172 to n = 184 .
the shell gap in the proton subsystems of the considered nuclei diminishes only little when the neutron number increases , so that the proton number z = 120 remains a rather stable shell closure while the detailed structure of the proton levels shows some rearrangements induced by the neutron addition . in the neutron subsystems
both pairing and qvc mechanisms are active and show a very delicate interplay : pairing correlations tend to increase the shell gap while the qvc alone tends to decrease it and at the same time causes the fragmentation of the states in the middle of the shell . as a result ,
in the presence of both mechanisms the gap in the neutron subsystem remains almost steady while the newly occupied levels jump down across the gap when the neutrons are added .
thus , in contrast to the pure mean field studies @xcite , no sharp neutron numbers appear as the candidates for the spherical shell closures in this region : the shell closures are delocalized .
although last decade the three major concepts in low - energy nuclear theory ( i ) ab - initio approaches , ( ii ) configuration interaction models ( known also as shell - models ) and ( iii ) density functional theory ( dft ) have advanced considerably , they still have to be further developed to meet the requirements demanded by contemporary nuclear experiment and astrophysics .
furthermore , each of these concepts has principal limitations of their applicability in the nuclear physics domain .
[ sir](a ) shows an image of the nuclear chart taken from ref .
the areas of the nuclear landscape which can be described by each of the three theoretical concepts are outlined ( here we focus on the spin - isospin nuclear properties , e.g. , gamow - teller response ) .
the sectors of the nuclear landscape where the applicability areas of the different models overlap are of particular interest because within these sectors the models can constrain each other .
ab - initio models can replace the phenomenological input which is traditionally used in the shell - model with the microscopic effective interaction computed from the first principles . in turn , the shell model with its very advanced configuration interaction concept can guide the dft - based developments beyond its standard random phase approximation .
thus , in contrast to considering different models as independently developing alternatives , we rather admit their complementarity which can be used for their further advancements .
sn calculated within the various theories ( see text for details ) : low - energy and total strength distributions ( upper panels ) and their cumulative sums ( lower panels ) , respectively.,width=566 ] in fig .
[ sir](b ) we present studies of the gamow - teller resonance ( gtr ) in @xmath0sn within the three theoretical concepts : ( i ) non - relativistic quasiparticle random phase approximation ( qrpa ) with realistic g - matrix interaction @xcite , ( ii ) covariant dft - based relativistic random phase approximation ( rrpa ) and its extension to particle - hole@xmath1phonon configurations called relativistic time blocking approximation ( rtba ) @xcite , and ( iii ) shell - model ( sm ) with the configuration interaction ci - jj7a truncated by the tamm - dancoff proton - neutron phonon coupled to particle - hole core vibrations @xcite .
the gross and fine features of the gtr obtained within these models are compared , the advantages and drawbacks of the considered models are discussed .
based on such comparative studies , future directions are outlined for each of the above mentioned microscopic models @xcite .
constraints on the many - body coupling schemes and underlying interactions from measurements at the future rare isotope facilities are anticipated .
excitation energy is another characteristic of excited nuclei which imposes limitations of their existence and plays a very important role in astrophysical modeling .
we consider the finite - temperature effect on the low - energy nuclear response known as upbend phenomenon , which was first reported in ref .
@xcite , later observed systematically in the @xmath2-ray strength functions below neutron threshold of various light and medium - mass nuclei and probed by different experimental techniques @xcite .
studies of ref . @xcite have revealed that this phenomenon , occurring in various astrophysical sites , can have a significant impact on their elemental abundances .
we propose a microscopic approach for the radiative strength function which is based on the statistical description of an excited compound nucleus .
the thermal mean field describes the nuclear excited states of the compound type very reasonably and , at the same time , it is simple enough to allow a straightforward generalization of very complicated microscopic approaches to nuclear response in terms of finite temperature corresponding to the nuclear excitation energy . to describe transitions from a thermally excited state , in the first approximation
we employ the finite - temperature continuum qrpa developed in @xcite .
the two - quasiparticle propagator in nuclear medium is calculated in terms of the matsubara temperature green functions in the coordinate space .
the continuum part of this propagator is responsible for transitions from the thermally unblocked discrete spectrum states to the continuum .
the radiative dipole strength function @xmath3 is determined from the propagator in the standard way .
[ 94mo](a ) displays the @xmath2-strength in @xmath4mo at the excitation energy around its neutron separation energy , that represents the case of radiative thermal neutron capture , compared to @xmath2-strength in the ground state .
the origin of the @xmath2-strength upbend due to the transitions to the continuum is illustrated schematically in fig .
[ 94mo](b ) .
the upbend appears as a typical feature of the @xmath2-strength in medium mass nuclei while in heavy nuclei the strength is flat at @xmath5 @xcite .
the obtained results have an important consequence for astrophysics , namely for the approaches to r - process nucleosynthesis : as shown in ref .
@xcite , the low - energy upbend in the @xmath2-strength can give rise to a considerable enhancement of the neutron capture rates in neutron - rich nuclei and , consequently , influences the global abundance distribution .
based on the obtained results , we expect further advancements of the theoretical approaches discussed in this contribution
. the proposed developments on many - body coupling schemes and underlying interactions will need constraints from data on nuclei away from the valley of stability .
such data will be obtained in experiments performed at existing , and vastly enhanced capabilities presented by future rare isotope facilities . 9 e. litvinova and p. ring , phys .
c * 73 * , 044328 ( 2006 ) .
e. litvinova , phys .
c 85 , 021303(r ) ( 2012 ) .
w. zhang et al . , nucl .
a * 753 * , 106 ( 2005 ) .
bertsch , j. phys .
* 78 * , 012005 ( 2007 ) . j. suhonen , t. taigel and a. faessler , nucl .
* a486 * , 91 ( 1988 ) . t. marketin , e. litvinova , d. vretenar , and p. ring , phys .
b * 706 * , 477 ( 2012 ) .
m. horoi and b. a. brown , phys . rev
. lett . * 110 * , 222502 ( 2013 ) .
fang , t. marketin , b.a .
brown , e. litvinova , and r.g.t .
zegers , in preparation .
a. voinov _
et al . _ ,
93 , 142504 .
m. wiedeking _
108 , 162503 .
a.c . larsen and s. gorieli ,
c 82 , 014318 ( 2010 ) .
litvinova , s.p .
kamerdzhiev , and v.i .
tselyaev , phys .
atomic nuclei 66 , 558 ( 2003 ) .
e. litvinova and n. belov , arxiv:1302.4478 . | we present a short overview of our recent theoretical developments aiming at the description of exotic nuclear phenomena to be reached and studied at the next - generation radioactive beam facilities .
applications to nuclear shell structure and response of nuclei at the limits of their existence , with a special focus on the physics cases of astrophysical importance , are discussed .
address = national superconducting cyclotron laboratory , michigan state university , east lansing , mi 48824 - 1321 , usa , address = department of physics and astronomy and national superconducting cyclotron laboratory , michigan state university , east lansing , mi 48824 - 1321 , usa address = national superconducting cyclotron laboratory , michigan state university , east lansing , mi 48824 - 1321 , usa , altaddress = joint institute for nuclear astrophysics , michigan state university , east lansing , mi 48824 - 1321 , usa address = physics department , faculty of science , university of zagreb , 10000 zagreb , croatia address = physik - department der technischen universitt mnchen , d-85748 garching , germany address = nuclear physics department , st .
petersburg state university , 198504 st .
petersburg , russia address = department of physics and astronomy and national superconducting cyclotron laboratory , michigan state university , east lansing , mi 48824 - 1321 , usa , altaddress = joint institute for nuclear astrophysics , michigan state university ,
east lansing , mi 48824 - 1321 , usa last decades , low - energy nuclear physics has expanded considerably its domain due to the opportunities opened by rare isotope beam facilities of the new generation . in particular ,
the techniques of the isotope separation online and in - flight production have been implemented at the major low - energy nuclear physics facilities with great success : numerous experiments on synthesis of exotic nuclei and on studies of their dynamical properties have been performed .
this has produced a strong catalyzing effect on theoretical developments toward finding a high - precision and highly universal solution of the nuclear many - body problem . in this contribution
, we give a brief overview of our recent developments and applications to nuclear systems toward the limits of mass , isospin and temperature . |
the bethe or bethe - like recursive lattices generally refer to the fractal arrangements of repeating units recursively connected to neighbors only on the sharing vertex , with no connection bond lies crossing layers .
it has become a powerful methodology in various fields such as thermodynamics @xcite , graph theory@xcite , optimization problems @xcite and so on . in statistical physics , one important application of the recursive lattice is to approximate the regular lattice with the identical coordination number to solve a thermodynamic system ( e.g. ising model ) on it . as one of few exactly calculable models , it has been proven to be a reliable method @xcite , with the advantage of exact calculation and simple iterative approach @xcite , to be applied in numerous physical systems , e.g. alloy @xcite , spin glass @xcite , polymers @xcite , biomacromolecule @xcite etc . as a versatile extension of the bethe lattice assembled by single dots and bonds , the husimi lattice employing simple geographic shapes , such as square , triangle , tetrahedron , hexagon , or cube
@xcite has also been developed for decades to describe various systems with particular geographic properties @xcite .
similar to the bethe lattice , the independence feature of units enables the exact calculation on husimi lattice regardless of the dimensions of the geographic unit , and mean field approximation is unnecessary since the interactions are confined within a unit and not shared by others .
the calculation method usually relies on the recursive approach , which is featured as simple and less computation effort costing thanks to the homogeneous self - similar structure @xcite .
nevertheless , the recursive feature is accompanied with several disadvantages of this lattice methodology .
firstly , the repeating structure implies a homogeneous system , it is only suitable to describe systems of uniform texture . some particular however important cases , e.g. the confined geometry or structural transformation , are enormously difficult , if not impossible , to be simulated by the recursive lattices .
therefore , besides a few investigations on the thermodynamics on the surface / thin film employed moderately inhomogeneous structure to present the boundary of a bulk system@xcite , the reports on the application of recursive lattice onto inhomogeneous systems were very rare .
secondly , recursive lattice is considered to be a reliable approximation to regular lattice based on the identical coordination number @xmath4 .
therefore , the manipulation of coordination number(s ) is critical in constructing a recursive lattice for particular requirements .
while it is easy to draw a regular lattice with an arbitrary @xmath4 , achieving an odd @xmath4 in recursive lattice usually requires awkward design of unit selection and branch number , and even worse a prime number of @xmath4 is impossible .
furthermore , when the randomness is necessary in a recursive lattice model , the common method is to add random terms in the hamiltonian , e.g. a random external field as noise or random exchange couplings parameter @xmath5 @xcite , while the structural randomness is difficult to be presented due to the homogeneity of recursions .
therefore , it is expected to be a considerable contribution to this field that if new designs of recursive lattice and calculation methods are developed to handle the above concerns , and make the recursive methodology more versatile in describing inhomogeneous systems .
recently we reported an husimi lattice of random square - cube recursion , on which the simple ising model can be solved by conventional exact calculation technique with moderate modification and exhibits well - defined thermal behaviors @xcite .
following the same principle , in this work we have developed a randomly multi - branched husimi square lattice and solved the simple ferromagnetic ising model on it .
the lattice is featured by randomly two or three square units connected on one vertex , then an inhomogeneous system of variable coordination numbers can be achieved with the identical unit cells . the spontaneous magnetization with critical temperature @xmath6
, thermodynamics around the singularity , and the thermal fluctuation caused by stochastic structure , were investigated with the variation of structural ratio .
the original husimi lattice was a tree - like graph assembled by squares with two units connecting on one vertex . since its development , derivative structures of three and more squares connection have been also investigated . in this way , to achieve an inhomogeneous structure with variable @xmath4
, it is a natural choice to have random number of branches connected in the lattice . to keep the investigation simple , in this work we will only study the randomly 2 or 3-branched husimi square lattice with @xmath7 or @xmath8 as demonstrated in fig.[fig1 ] .
a structural ratio @xmath0 or @xmath1 can be defined to indicate the probability to have 2 or 3 branches on one vertex , with obviously @xmath2 . however , as a probable reason why this type of lattice has not been reported before , the random structure destroys the recursive homogeneity and then makes the iterative approach unfeasible , therefore the lattice shown in fig.[fig1 ] is not the actual model studied in this paper , and particular limitations on the structure must be applied to achieve an exact calculation , which will be detailed later .
while the branch number is randomly 2 or 3 in this lattice , with the structural ratio @xmath3 we can define an analog branch number " as @xmath9 to present the average branch number of the lattice , and similarly an analog coordination number @xmath4 can also be defined as @xmath10 . by this means
, the term cross - dimensional " in this paper refers to the gradient of analog @xmath4 : taking a finite regular cubic lattice with a 2d surface as an example , we have @xmath11 on the surface and @xmath12 in the bulk , then in the near - surface region a randomly sampled site will has a probability to be of either @xmath4 depending on the depth .
therefore , a gradually variation of analog @xmath4 from 5 to 6 well represents the cross - dimensional range from the surface to bulk in this case .
similarly , the case of a 2d layer crossing to thin film can be described by the variation of analog @xmath4 from 4 to 6 . for an additional clarification ,
both 3-branched husimi square lattice and husimi cubic lattice have been proved to be a good approximation to the regular cubic lattice @xcite .
the simplest ferromagnetic spin@xmath13 ising model was applied on the lattice in this paper : @xmath14 without external magnetic field @xmath15 .
the weights of one configuration @xmath16 of a square unit is given by @xmath17 where @xmath18 is the reversal temperature as @xmath19 , the boltzmann constant @xmath20 is set as one .
we have the partition function of the entire system as @xmath21 where the @xmath22 denotes the state of the lattice as an ensemble of unit @xmath23 . in this paper
we setup a uniform ferromagnetic coupling @xmath24 , then the state of system only depends on the structural properties . without external magnetic field , we can expect a half - half probability of spin state on each site at high temperature , a uniform orientation pointing to either up or down of all spins at low temperature , and a spontaneous magnetization occurring in between .
the only question being focused on in this paper is that , how this transition behaves in the cross - dimensional situation on an inhomogeneous lattice .
the lattice is designed of infinite size , nevertheless for an iterative approach it is necessary to imagine an original point where the entire lattice contribute to .
furthermore , the structure must be symmetrical to the original point , and subsequently the symmetry of sub - trees contributing onto one unit is required , otherwise the unique structure of an arbitrary sub - tree is impossible to be tracked and accounted in iterative calculation .
therefore , the unlimited random structure shown in fig.[fig1 ] is not the actual lattice we are going to study , and two important principles have to be settled here : 1 ) the branch number on the vertices of one unit must be the same excluding the base vertex ; 2 ) for any arbitrary square the three sub - trees contributing onto it towards to the original point should be identical . and the branch number on the vertices of different levels are random with the structural ratio @xmath3 . a sample structure is presented in the fig.[fig2]a . although these two limitations confine a locally ordered configuration on the same levels and impair the randomness of the lattice , from a general view of the infinitely large structure , we still have two or three branched vertex randomly appears with a fixed probability .
therefore , we may say that this paper discussed a special case of the ideally random multi - branched lattice with identical sub - trees contributions .
the calculation method basically follows the previous works done by gujrati s @xcite , here we will briefly derive the method with specific modifications for the random - branched lattice . by defining the partition function of a sub - tree with the base spin fixed as @xmath25 or @xmath26 to be the partial partition function ( ppf ) , denoted as @xmath27 , where @xmath28 is the level of the base spin , at the original site @xmath29 ( level @xmath30 ) we have the total partition function : @xmath31 the exponent @xmath32 is the number of sub - trees ( branches ) contributing onto @xmath33 . while in the regular lattice the branch number is a fixed variable , in this work the branch number is a series of random variables on each level , therefore the branch number shall be written as @xmath34 associated to the level @xmath28 . as shown in fig.[fig2]a , for each square unit on level @xmath28 it always takes in @xmath35 sub - trees contributions from upper level and contribute itself to the unit on the lower level , then by defining @xmath36 , for the ppf on the level @xmath37 we have the recursive relations @xmath38 @xmath39 with the sites labeling shown in fig.[fig2]a , on the two sites @xmath40 and @xmath41 neighboring to @xmath42 we have @xmath43 sub - trees contributing from level @xmath44 , and similar to the @xmath45 diagonal to @xmath46 .
the @xmath47 is the local weight of the square confined by the four sites @xmath42 , @xmath40 , @xmath41 , and @xmath45 . in principle
this local weight should exclude the weights of base site @xmath42 , @xmath48 , to conduct a cavity field contribution , however this can be ignored since we set the external field @xmath49 in this paper . with the ppfs defined above ,
we introduce a ratio @xmath50 ( simplified as @xmath51 ) on the site @xmath46 @xmath52 as the weights ratio of ppfs with the fixed spin state of @xmath46 , @xmath53 denotes the sub - tree contribution to the magnetization field of the site @xmath46 , i.e. the cavity contribution @xcite .
therefore this ratio can be treated as the solution " of the system to indicate the magnetization of a site .
obviously , the solution of @xmath54 or @xmath55 indicates a uniform spins orientation to either up or down , and conversely the solution @xmath56 implies a half - half probability to have a up or down spin on one site , i.e. the amorphous state without external magnetic field .
since @xmath51 is a function of ppfs on level @xmath37 , and as shown in eq.[ppf+_recursion ] and [ ppf-_recursion ] a ppf is a function of the ppfs on higher level , then @xmath51 can be derived as a function of @xmath53s on higher levels in a recursive fashion : @xmath57 with an initial seeding input of @xmath58 , the solution can be obtained by applying the eq.[x_recursion ] many times until reaching a fixed - point solution . principally , various input seeds should be tried to exhaust all possible fixed points , however this work deals with the simplest ferromagnetic ising model , and a single solution at a fixed temperature is expected . the exact form of eq.[x_recursion ]
is derived below : starting from @xmath59 we define a compact note @xmath60{c}% x_{n}\text { if } s_{n}=+1\\ y_{n}\text { if } s_{n}=-1 \end{array } \right.\end{aligned}\ ] ] in terms of@xmath61 we have @xmath62 and @xmath63 where the sum is over @xmath64 for @xmath65 , and over @xmath66 for @xmath67 , and where @xmath68 it is related to the polynomials @xmath69 according to @xmath70 in terms of the above polynomials , we can express the recursive relation for the ratio @xmath51 in terms of @xmath71 : @xmath72 the above process is the static recursive calculation of solutions .
for the random multi - branched lattice , on each iteration the program will randomly assign the value of @xmath73 to be 2 or 3 according to the structural ratio @xmath0 , and then execute the calculation for that local level .
therefore , the solutions exhibit a fluctuation around an average value , and the fixed - point solution does not exist . according to our experience ,
the calculation reaches a stable " solutions oscillation @xmath74 in no more than 2000 iterations . in the program we did the calculation 16,000 times and average the last 11,000 @xmath53s to reach a reliable @xmath75 . in this way , although the exact calculation is executed in each iteration , instead of exact results what we actually obtained is a numerical solution distribution .
the thermodynamic calculations again follows the same principle detailed in ref .
@xcite , while a slight modification is necessary to fit the random conformation case .
herein we firstly review the general method to calculate the free energy on a homogeneous @xmath32-branched husimi lattice .
imagine we achieved the fixed - point solution in the region around the original point @xmath76 , the @xmath32 squares joint on @xmath76 are indexed as the first level , then there are @xmath77 squares on the level @xmath78 .
now we cut off these sub - trees and hook them up to form @xmath79 smaller but identical lattices , the partition function of these lattices are @xmath80 the free energy of the left out local squares is @xmath81 we have @xmath82 cites in a square and @xmath73 squares in the local origin region .
the free energy per site is:@xmath83 by substituting @xmath84 and @xmath85 we have @xmath86^{3b^{\prime } } } ) .
\label{fe}\ ] ] an important concern in the inhomogeneous lattice discussed in this paper is that , the original point is merely a conceptual point that indicating the calculation direction , while it can not be realized as an either 2 or 3 branched point .
therefore , the free energy of a local region should be calculated along the recursive process and averaged , in the same fashion of solutions , to approach the free energy of the whole system . by this
means , an replica counterpart calculation was designed to obtain the free energy along the recursive process , as shown in fig.[fig2]b .
after the solution converged to the stable oscillation , once an iteration of solution was calculated on one site , the immediate calculation of free energy of that site was carried out by hooking an identical replica half - tree onto that site ( the shadow part in fig.[fig2]b ) .
then as in a complete lattice with the replica counterpart , the central point can be taken as the ` origin ' and the general method of free energy calculation ( eq.[fe ] ) can be applied by cutting - off and rejoining the sub - trees on the corresponding sites @xmath87 and @xmath88 to yield the free energy of the local region .
although this replica part does not exist in the system , due to the perfect symmetry it provides the correct free energy per site of the local region , and the value calculated on the real half is valid .
we have also tested that , the branch number on the symmetric joint point does not affect the free energy calculation , i.e. hooking one or two imaginary counterparts onto the site to make @xmath89 or @xmath90 present the exact same free energy density , i.e. the thermodynamics only depends on the solution @xmath58 once it was determined .
the average solution @xmath75 with @xmath91 is presented in fig.[fig3]a .
the solution of deterministic cases of @xmath92 and 3 with @xmath93 and 0 are also shown for comparison .
as expected , the solution of @xmath91 with its deviation lies inside the area lined out by the two deterministic solutions .
albeit the rigorous fluctuation in the intimidate temperature range , a distinct phase transitions can be figured out with the orientation preference of spins differentiated from @xmath94 , i.e. the spontaneous magnetization of simple ising model .
figure [ fig3]b shows the detailed solution of @xmath91 around the critical temperature @xmath6 . although the error bar is considerable large merely below @xmath6 , the transition is still distinctive . besides the averaged solution over 11,000 results , we have also investigated several single runs and found that the position of @xmath6 is consistent , i.e. the phase transition of ising spins is independent of the structural stochasticity .
this phenomenon evidenced a well - defined thermodynamics of the randomly multi - branched lattice , and we can be confidential to apply the conventional analysis onto this inhomogeneous lattice .
ten sets of free energies with @xmath0 varying from 2 to 3 by the increment of 0.1 are shown in fig.[fig4]a .
similar to the solutions , free energies of cross - dimensional structures lies inside the area outlined by the 2 and 3-branched free energy curves .
the uneven curves of the free energies of models with @xmath95 imply the thermal fluctuations , which agree with the significant error bar of the solutions presented in fig.[fig3 ] . to more clearly investigate the thermal fluctuation ,
the entropies derived from @xmath96 of five structures are presented in fig.[fig4]b .
rigorous fluctuations can be observed on the entropy behaviors of stochastic models , with a superheating extrapolation unphysically surpass the entropy of the amorphous state .
these singularities should not be understood as the simple effect of randomness , it actually reveals the inhomogeneity in the cross - dimensional range : for the near surface region of a bulk materials , the layer beneath the surface is at the halfway up the energy landscape , and its state is fragile and easily to be frustrated by the energy flow from surface to bulk or vice versa . at the critical range ,
the system is in dramatic dynamics , as the surface begins the order - disorder phase transition while the bulk remains ordered , therefore the energy fluctuations is expected to induce locally superheating in between .
one of the most important aim to design and investigate the inhomogeneous husimi lattice , is its application in the cross - dimensional range .
figure [ fig5]a presents the ten sets of solutions with various analog @xmath73 , and the 3d mapping of these solutions is presented in fig.[fig5]b .
this 3d mapping clearly indicates the gradient in the cross dimensional situation , i.e. the depth from surface to the bulk , or the expansion of a thin film to thick bulk .
herein we can observe from the 3d mapping that it is possible to have different phases on the near - surface region at the same temperature . the edge of the 0.5 plateau is the transition line , which is fitted as @xmath97 with @xmath98 ( fig.[fig5]c ) . before analyzing the exponential behavior of @xmath99 vs. @xmath73 presented in eq.[tranline ] , the variation of ground state energy with the analog @xmath73 should be figured out as a reference .
when the systems is at @xmath100 , all the spins point to one direction as a perfect crystal and we have the ground state energy @xmath101 with entropy @xmath102 . for the homogeneous lattices of @xmath103 and @xmath104 ,
the ground energy state is simply the number of coupling interactions @xmath105 , or @xmath106 ( one interaction is shared by two spins ) . for the cross - dimensional lattices in between , the ground state energies @xmath107 with various analog @xmath73 exhibit linear behavior as presented in table[t1 ] ,
this phenomenon evidenced that the system has a well - defined average property according to the probability ratio .
therefore , the exponential correlation of @xmath99 vs. analog @xmath73 suggests that the stochasticity might be the single reason to affect the transition behavior near the critical temperature , and we may hypothesize that structural randomness reduces the stability of the system and leads to a lower transition temperature .
nevertheless , in the quantitative aspect , the exponential fitting and the meaning of the parameters in eq.[tranline ] is still unclear .
further investigations on this phenomenon is in progress .
.variation of @xmath99 and the ground state energy density with analog @xmath73 . [ cols= " < , < , < " , ]
the ferromagnetic ising model was solved on a randomly multi - branched husimi square lattice .
the lattice was constructed with two or three squares connecting on one with the probability @xmath0 or @xmath1 .
while the system is deterministic on each end , the structures with variable @xmath3 in between describes a cross - dimensional situation from 2d to 3d , such as the thin film or the near - surface case .
the solution @xmath53 representing the magnetization of spins was obtained by the cavity partial partition function method , and the thermodynamics such as free energy and entropy were derived from the solutions .
typical spontaneous magnetization were observed on the random inhomogeneous lattices , the transition occurs on an characteristic @xmath6 regardless of the randomness and thermal fluctuation .
the @xmath6 , however , behaves an exponential correlation to the structural ratio or analog branch number . considering the linear relationship of the ground state energy @xmath107 vs. @xmath73 ,
the cause of these lowered @xmath6s is hypothesized to be the stochasticity that reduced the stability of the system , while the quantitative meaning of the fitting equation remains unclear .
rigorous fluctuations were observed on the entropy behaviors with singularities of locally supercooling or superheating extrapolation , implying the severe inhomogeneity in the cross - dimensional range around the critical temperature .
this work is financially supported by the national natural science foundation of china ( grant no .
11505110 ) , the shanghai pujiang talent program ( grant no .
16pj1431900 ) , and the china postdoctoral science foundation ( grant no .
2016m591666 ) .
the author states that there is no competing financial interests .
g. d. ferraro , c. wang , d. mart and m. mzard , arxiv:1409.3048 [ cond-mat.dis-nn ] a. lage - castellanos and r. mulet , eur .
j. b 65 , 117 - 129 ( 2008 ) r. huang and c. chen , j. phys .
83 , 123002 ( 2014 ) | an inhomogeneous random recursive lattice is constructed from the multi - branched husimi square lattice . the number of repeating units connected on one vertex is randomly set to be 2 or 3 with a fixed ratio @xmath0 or @xmath1 with @xmath2 .
the lattice is designed to describe complex thermodynamic systems with variable coordinating neighbors , e.g. the asymmetric range around the surface of a bulk system .
classical ferromagnetic spin-1 ising model is solved on the lattice to achieve an annealed solution via the local exact calculation technique .
the model exhibits distinct spontaneous magnetization similar to the deterministic system , with however rigorous thermal fluctuations and significant singularities on the entropy behavior around the critical temperature , indicating a complex superheating frustration in the cross - dimensional range induced by the stochasticity .
the critical temperature was found to be exponentially correlated to the structural ratio @xmath3 with the coefficient fitted as 0.53187 , while the ground state energy presents linear correlation to @xmath3 , implying a well - defined average property according to the structural ratio .
husimi lattice , ising model , random multi - branched , spontaneous magnetization , superheating , cross - dimensional |
let @xmath0 be an algebraic variety over a field @xmath1 , which we assume to be of characteristic 0 .
one of the few objects that come naturally with @xmath0 is its sheaf of khler differentials @xmath2 .
it is the sheafified version of the module of khler differentials , which is also an important tool in commutative algebra . in this note
we stick to the geometric language . the sheaf of differentials and its higher exterior powers play an important rle in many contexts , most prominently deformation theory , vanishing theorems and ( if @xmath0 is sufficiently nice ) duality theory . for
a more local example one could mention berger s conjecture that a curve is smooth if and only if its sheaf of differentials is torsionfree ( see e.g. @xcite ) or the zariski lipman conjecture : if @xmath3 is locally free then @xmath0 is regular ( see e.g. @xcite ) . in the context of the minimal model program greb , kebekus , kovcs and
peternell proved strong extension theorems for differential forms @xcite , but instead of the sheaf of khler differentials itself they used its reflexive hull @xmath4 also called module of zariski differentials ( see e.g. @xcite ) . for applications
the following obvious question comes to mind : [ quest ] if @xmath0 has _ mild _ singularities , is @xmath5 reflexive or at least torsionfree ?
phrasing this slightly differently we ask if the natural map @xmath6 , whose kernel is the torsion submodule @xmath7 , is bijective or at least injective . in the terminology of (
* ( 1.6 ) ) we say that @xmath5 has _ cotorsion
_ if @xmath8 is not surjective .
were extensively studied for example by greuel and his collaborators in @xcite and @xcite . ]
if one translates _
mild singularity _ as being a local complete intersection , then indeed a complete answer to question [ quest ] is known .
[ lci ] let @xmath0 be a local complete intersection
. then @xmath5 satisfies serre s condition @xmath9 if and only if @xmath0 is regular in codimension @xmath10 . in particular , if @xmath0 is a normal local complete intersection , then @xmath5 is torsionfree , and it is reflexive if and only if @xmath0 is is nonsingular in codimension 2 .
however , in the context of modern birational geometry one usually measures the singularities of a normal variety in terms of discrepancies , which give rise to the definition of terminal , canonical , log terminal and other singularities ( * ? ? ?
* section 2.3 ) .
even terminal singularities , the mildest class considered , are in general not complete intersections , so theorem [ lci ] does not apply . somewhat contrary to our expectations we will show below that the answer to question [ quest ] is essentially negative if one interprets mild in the sense of birational geometry ; as soon as one leaves the world of local complete intersections one should expect the sheaf of khler differentials to have both torsion and cotorsion .
there are two cheap ways to produce non
lci singularities : quotients of finite groups and affine cones over projective varieties . in the second case
the algebraic description is somewhat simpler and we give a criterion for the existence of torsion differential in section [ criterion ] .
recall that if @xmath11 is a projective variety or scheme given by an ideal sheaf @xmath12 then @xmath13 is called _ projectively normal _ if @xmath14 for all @xmath15 or , equivalently , @xmath16 for all @xmath17 .
[ main ] let @xmath18 be a smooth projective variety over a field @xmath1 of characteristic zero and let @xmath12 be the sheaf of ideals defining @xmath13 . let @xmath19 be the affine cone over @xmath13 . if @xmath20 for all @xmath17 then @xmath21 is torsionfree . if in addition @xmath13 is projectively normal , then @xmath21 is torsionfree if and only if also the first infinitesimal neighbourhood of @xmath13 in @xmath22 is projectively normal . in the cone
situation the study of cotorsion is more problematic but in section [ cotorsion ] we recall results by knighten and steenbrink that give an easy sufficient criterion for cotorsion on finite quotient singularities . in the last section we study our main class of examples , namely @xmath23 , the affine cone over the @xmath10th veronese embedding of @xmath24 .
we can also describe @xmath23 as a cyclic quotient singularity .
such cones have torsion differentials if and only if @xmath25 ( proposition [ vrd ] ) and cotorsion as soon as @xmath26 ( proposition [ vrdkt ] ) ; we have collected some significant cases in table [ tab ] and @xmath27 was also computed in @xcite but the examples were not widely known . ] . [ tab ]
.some veronese cones with torsion or cotorsion in @xmath28 .
[ cols="^,^,^,^,^,^,^",options="header " , ] in the surface case our results are optimal in the following sense : let @xmath0 be a surface singularity . if @xmath0 is terminal then it is smooth and thus @xmath5 is locally free .
if @xmath0 is canonical but not terminal then it is one of the well known ade singularities , thus a hypersurface singularity ; by theorem [ lci ] the sheaf of khler differentials @xmath5 is torsionfree but not reflexive in this case .
the easiest log terminal point is the cone over the twisted cubic @xmath27 and in this case @xmath5 has both torsion and cotorsion .
gorenstein terminal 3fold singularities are hypersurface singularities by a result of reid ( * ? ? ?
* ( 3.2 ) theorem ) ; hence theorem [ lci ] applies to show that the sheaf of khler differentials is reflexive in this case .
it is possible that other classes of mild singularities in small dimensions turn out to have torsionfree or reflexive sheaf of khler differentials as well . using structural results like hilbert burch or buchsbaum eisenbud , low
codimensional singularities might also be accessible ( compare ( * ? ? ? * section 4 ) ) .
in this section we give the proof of theorem [ main ] .
let @xmath11 be a smooth , irreducible , non degenerate projective variety and let @xmath29 be the affine cone over @xmath13 .
we denote by @xmath12 the ideal sheaf of @xmath13 .
the question whether @xmath21 has torsion is purely algebraic , because the cone is affine .
denote by @xmath30 $ ] the polynomial ring with homogeneous maximal ideal @xmath31 and by @xmath32 the homogeneous coordinate ring of @xmath13 whose homogeneous maximal ideal we denote by @xmath33 .
note that @xmath29 is smooth outside the vertex so we are only interested in the local behaviour at the vertex . in the following
we will use some facts about local cohomology all of which can be found in ( * ? ? ?
* appendix 1 ) or in more detail in @xcite . in the following diagram
the middle column is the conormal sequence and the maps to the maximal ideal are given by the contraction with the vectorfield @xmath34 . @xmath35
\ar@{=}[r ] & i / i^2\ar[d]\\ 0\ar[r ] & m \ar[r]\ar[d ] & \omega_s\tensor r \ar[d]\ar[r]^{\xi\lrcorner } & \gothn \ar[r]\ar@{=}[d ] & 0\\ 0 \ar[r ] & n\ar[r]\ar[d ] & \omega_r \ar[r]^{\xi\lrcorner}\ar[d ] & \gothn \ar[r ] & 0\\ & 0 & & } \ ] ] by definition , the modules @xmath36 and @xmath37 are the respective kernels of the contraction maps @xmath38 .
the fact that the composition @xmath39 is zero follows from euler s formula : for every homogeneous element @xmath40 we have @xmath41 . since @xmath33 is torsionfree the torsion
submodule @xmath42 is isomorphic to @xmath43 .
since @xmath29 is smooth outside the vertex we can compute this torsion submodule via local cohomology @xmath44 which we now relate to sheaf cohomology on @xmath45 . introducing the appropriate grading we can transform the above diagram into a diagram of coherent sheaves on @xmath45 , where the middle row is the restriction of the euler sequence to @xmath13 and
the first column becomes the conormal sheaf sequence for @xmath13 .
the map @xmath46 in the latter sequence is injective because @xmath13 is smooth .
@xmath47 & 0 \ar[d]\\ & \ki/\ki^2\ar[d ] \ar@{=}[r ] & \ki/\ki^2\ar[d]\\ 0\ar[r ] & \omega_{\ip^n}\restr x \ar[r]\ar[d ] & \ko_x(-1)^{\oplus n+1 } \ar[d]\ar[r ] & \ko_x\ar[r]\ar@{=}[d ] & 0\\ 0 \ar[r ] & \omega_x\ar[r]\ar[d ] & \widetilde{\omega_r } \ar[r]\ar[d ] & \ko_x \ar[r ] & 0\\ & 0 & 0 } \ ] ] we now use the comparison sequence for local and sheaf cohomology ( ( * ? ? ?
a1.12 ) ) for the first column of the diagram to obtain @xmath48\\ & 0\ar[r ] & i / i^2 \ar[r ] \ar[d ] & \gamma_*(\ki/\ki^2 ) \ar[r ] \ar[d ] & h^1_\gothn(i / i^2)\ar[r]\ar[d ] & 0\\ & 0\ar[r ] & m \ar[r]\ar[d ] & \gamma_*(\omega_{\ip^n}\restr x)\ar[d]\ar[r ] & h^1_\gothn(m)\ar[r]\ar[d ] & 0\\ 0\ar[r ] & h^0_\gothn(n)\ar[r ] & n \ar[r ] \ar[d ] & \gamma_*(\omega_x ) \ar[r ] \ar[d ] & h^1_\gothn(n)\ar[r]&0\\ & & 0 & } \ ] ] where we already used the following lemma . in
the above situation we have 1 .
@xmath49 , 2 .
@xmath50 . 1 .
this follows from the short exact sequence @xmath51 and the fact that @xmath52 and @xmath53 are saturated and have depth at least 2 . 2 .
@xmath36 is a submodule of the free module @xmath54 , hence torsionfree . applying the same arguments as in the proof of the snake lemma to we get an exact sequence @xmath55 and
thus have shown that @xmath56 is torsionfree if @xmath57 .
this latter group can be interpreted as follows : the local cohomology sequence for @xmath51 compared with the sheaf cohomology sequence of @xmath58 yields a diagram @xmath59\ar[d]^\isom & i / i^2 \ar[r]\ar@{^(->}[d ] & 0 \ar[d]\\ \gamma_*(\ki ) \ar[r ] & \gamma_*(\ki/\ki^2)\ar[r]\ar@{->>}[d ] & \bigoplus_d h^1(\ki^2(d))\ar[r]&\bigoplus_d h^1(\ki(d))\\ & h^1_\gothn(i / i^2)\ar@{^(->}[ru ] } \ ] ] we have thus proved the first part of theorem [ main ] from the introduction which we repeat here for the convenience of the reader .
let @xmath18 be a smooth projective variety over a field @xmath1 of characteristic 0 and let @xmath12 be the sheaf of ideals defining @xmath13 .
let @xmath19 be the affine cone over @xmath13 . if @xmath60 for all @xmath15 then @xmath21 is torsionfree . if in addition @xmath13 is projectively normal , then @xmath21 is torsionfree if and only if also the first infinitesimal neighbourhood of @xmath13 is projectively normal
. by definition , projective normality of @xmath13 is equivalent to @xmath61 or equivalently @xmath62 .
the exact sequence in local cohomology @xmath63 induced by @xmath64 , shows that @xmath65 .
together with and that the composition @xmath66 is an isomorphism .
this establishes the desired equivalence .
we briefly recall a result of knighten , also discovered by steenbrink ( * ? ? ?
1.8 ) , describing the double dual of @xmath67 on finite quotient singularities @xmath68 , and deduce a sufficient condition for the existence of cotorsion in the sheaf of khler differentials .
let @xmath69 be a regular local @xmath1algebra and let @xmath70 be a finite group acting on @xmath69 .
we denote by @xmath71 the quotient map .
the natural map @xmath72 is an isomorphism . in concrete situations
the module of invariant differentials is not hard to compute and and the following will turn out to be useful [ cotors ] assume that @xmath68 has an isolated singularity . let @xmath73 be the maximal ideal of @xmath74 and @xmath75 be the embedding dimension of @xmath68
. if the minimal number of generators for @xmath76 ( considered as a @xmath74module ) is bigger than @xmath77 then @xmath67 has cotorsion and is not reflexive . by the conormal sequence the sheaf @xmath67 of khler differentials can be generated by @xmath77 elements and thus can never surject onto @xmath78 if this sheaf needs more than @xmath77 generators .
in this section we study cones over veronese embeddings , including the examples mentioned in table [ tab ] .
the first subsection collects some general properties of these cones , torsion differentials are computed in propostion [ vrd ] , and cotorsion will be discussed in proposition [ vrdkt ] . here
we collect basic properties of the cones over the veronese embeddings .
in particular , we discuss their realisation as cyclic quotient singularities , and compute the discrepancies of the canonical resolution in order to determine in which cases these cones are terminal , canonical , etc . for this
we follow @xcite and ( * ? ? ?
7.2 ) , where one can also find the definition of the singularities appearing in the minimal model program .
let @xmath79 be the cyclic group of order @xmath10 and let @xmath80 be the representation given by choosing a primitive @xmath10th root of unity @xmath81 and sending a generator of @xmath79 to @xmath82 let @xmath83 be the resulting quotient singularity ( in reid s ( * ? ? ?
* ( 4.2 ) ) notation , this is a singularity of type @xmath84 ) . since the ring @xmath85^{\mu_d}$ ]
is generated by all monomials of total degree @xmath10 in the coordinates @xmath86 , the quotient @xmath23 is isomorphic to the cone @xmath87 over the image @xmath88 of the @xmath10th veronese embedding @xmath89 ( so @xmath90 ) .
this isomorphism is induced by the map @xmath91
the blow up @xmath92 of the origin in @xmath23 is smooth , isomorphic to the total space of the line bundle @xmath93 . one can check ( see @xcite ) that the index of @xmath23 is the denominator of @xmath94 , i.e. , @xmath95 .
in particular , the canonical divisor @xmath96 is @xmath97cartier , and @xmath23 is gorenstein if and only if @xmath10 divides @xmath98 .
we next compute the discrepancy of the unique exceptional divisor @xmath99 ; for simplicity we suppress the indices in the notation .
since the canoncial divisor is @xmath97cartier we can write @xmath100 for a rational number @xmath101 . the normal bundle of @xmath102 in @xmath103 is @xmath104 . by adjunction ,
the canonical divisor of @xmath102 is @xmath105 .
hence , by restricting to @xmath102 we obtain @xmath106 it follows that all veronese cones @xmath23 are log terminal , and in addition that @xmath23 is terminal ( canonical ) if and only if @xmath107 ( @xmath108 ) .
we now fix @xmath112 and @xmath10 and denote by @xmath12 the ideal sheaf of the image of the veronese embedding @xmath113 . to avoid confusion we denote by @xmath114 a hyperplane on @xmath45 , so that @xmath115 .
since @xmath116 is projectively normal the second part of theorem [ main ] applies and we only have to check if there is an @xmath117 such that @xmath118 . by a result of wahl ( * ? ? ?
* theorem 2.1 ) we have @xmath119 and thus we only need to consider the case @xmath120
. we will start by showing that @xmath110 is torsionfree if @xmath121 . recall that there is a gaussian map @xmath122 symbolically given by @xmath123 , which in our case is a homomorphism of @xmath124representations .
wahl showed that @xmath125 ( * ? ? ? * proposition 1.8 ) .
note that the representation on @xmath126 is irreducible , cf .
* section 2 ) .
it follows by a straightforward computation that the map @xmath127 is non - trivial and hence surjective . a calculation similar to the ones in lemma [ lem : estimates ] below gives the equality @xmath128 .
consequently , @xmath127 is an isomorphism .
thus @xmath129 .
this proves the only if part of the claim .
if we dentote the graded polynomial ring in @xmath98 variables as @xmath137 then the veronese embedding is induced by the homomorphism of graded rings @xmath138 with kernel a graded ideal @xmath52 . in degree
@xmath117 we get @xmath139 where surjectivity of the last map can easily be checked on monomials .
the formula for the dimension follows from the well known formula for the dimension of a symmetric product .
this proves the first item . from the embedding
@xmath140 we get the ( twisted ) normal bundle sequence @xmath141 the global sections of @xmath142 can be either computed via the euler sequence or read of from bott s formula ( see e.g. @xcite ) . pulling back the euler sequence on @xmath45 with @xmath143 we get @xmath144
the map on global sections @xmath145 is surjective and thus @xmath146 which proves .
it remains to prove .
putting together the formulas from the first two items for @xmath120 , we compute @xmath147 to obtain it therefore suffices to show that @xmath148 for @xmath149 this formula reduces to @xmath150 which is certainly true for @xmath111 .
we now proceed by induction on @xmath112 . for convenience ,
note that for @xmath151 and @xmath152 the expression on the left
hand side of gives 20 , so we may assume @xmath153 if @xmath151 . the induction step follows from the computation below , in which we apply the induction hypothesis twice ( in steps 3 and 5 ) and use some standard identities for binomial coefficients .
@xmath154 where in the last step we used @xmath155 , and @xmath153 if @xmath151 .
this concludes the induction and the proof of .
recall that the ring @xmath158^{\mu_d}$ ] is generated by all monomials of total degree @xmath10 in the coordinates @xmath159 and thus the embedding dimension of @xmath23 is @xmath160 . on the other hand ,
the @xmath161$]-module @xmath162 has a minimal system of homogeneous generators given by all products of monomials of degree @xmath163 with @xmath164 . subtracting the embedding dimension from the number of these generators we obtain @xmath165 where in the last step we use @xmath26 .
it follows that @xmath166 has cotorsion .
we thank duco van straten for an interesting discussion and miles reid for many helpful comments on an earlier version of this note . after posting of the first version of this paper on the arxiv jonathan wahl pointed us to @xcite , which led to a strenghening of proposition [ vrd ] .
the first author was partly supported by the dfg
forschergruppe 790 classification of algebraic surfaces and compact complex manifolds ( bayreuth freiburg ) . during the preparation of the second version of the paper , he enjoyed the hospitality of the mathematics department at princeton university .
he gratefully acknowledges the support of the eliteprogramm fr postdoktorandinnen und postdoktoranden of the baden wrttemberg siftung .
the second author gratefully acknowledges support by the dfg via the sfb / tr 45 periods , moduli spaces and arithmetic of algebraic varieties and his emmy noether project , and partial support by the hausdorff centre for mathematics in bonn .
miles reid .
young person s guide to canonical singularities . in _ algebraic geometry , bowdoin , 1985 ( brunswick , maine , 1985 ) _ , volume 46 of _ proc .
pure math .
_ , pages 345414 .
soc . , providence , ri , 1987 .
j. h. m. steenbrink .
mixed hodge structure on the vanishing cohomology . in
_ real and complex singularities ( proc .
ninth nordic summer school / navf sympos .
oslo , 1976 ) _ , pages 525563 .
sijthoff and noordhoff , alphen aan den rijn , 1977 . | we give a criterion for the sheaf of khler differentials on a cone over a smooth projective variety to be torsionfree . applying this to veronese embeddings of projective space and using known results about differentials on quotient singularities
we show that even for mild , e.g. gorenstein terminal , singularities the sheaf of khler differentials will in general have torsion and cotorsion . |
what is all the fuss about noise ? in this review we endeavor to convey the excitement and promise of studies of the cosmic microwave background ( cmb ) radiation to scientists not engaged in these studies , particularly to particle and nuclear physicists .
although the techniques for both detection and data processing are quite far apart from those familiar to our intended audience , the science goals are aligned .
we do not emphasize mathematical rigor , but rather attempt to provide insight into ( a ) the processes that allow extraction of fundamental physics from the observed radiation patterns and ( b ) some of the most fruitful methods of detection . in section 1 ,
we begin with a broad outline of the most relevant physics that can be addressed with the cmb and its polarization .
we then treat the early history of the field , how the cmb and its polarization are described , the physics behind the acoustic peaks , and the cosmological physics that comes from cmb studies .
section 2 presents the important foreground problem : primarily galactic sources of microwave radiation .
the third section treats detection techniques used to study these extremely faint signals .
the promise ( and challenges ) of future studies is presented in the last two sections . in keeping with our purposes
, we do not cite an exhaustive list of the ever expanding literature on the subject , but rather indicate several particularly pedagogical works . here
we briely review the now standard framework in which cosmologists work and for which there is abundant evidence .
we recommend readers to the excellent book modern cosmology by dodelson ( 2 ) . early in its history
( picoseconds after the big bang ) , the energy density of the universe was divided among matter , radiation , and dark energy . the matter sector
consisted of all known elementary particles and included a dominant component of dark matter , stable particles with negligible electromagnetic interactions .
photons and neutrinos ( together with the kinetic energies of particles ) comprised the radiation energy density , and the dark energy component some sort of fluid with a negative pressure appears to have had no importance in the early universe , although it is responsible for its acceleration today .
matter and radiation were in thermal equilibrium , and their combined energy density drove the expansion of space , as described by general relativity . as the universe expanded , wavelengths were stretched so that particle energies ( and hence the temperature of the universe ) decreased : t(@xmath0 ) = t(0)(1 + @xmath0 ) , where @xmath0 is the redshift and t(0 ) is the temperature at @xmath0 = 0 , or today .
there were slight overdensities in the initial conditions that , throughout the expansion , grew through gravitational instability , eventually forming the structure we observe in todays universe : myriad stars , galaxies , and clusters of galaxies .
the universe was initially radiation dominated .
most of its energy density was in photons , neutrinos , and kinetic motion . after the universe cooled to the point at which the energy in rest mass equaled that in kinetic motion ( matter - radiation equality ) , the expansion rate slowed and the universe became matter dominated , with most of its energy tied up in the masses of slowly moving , relatively heavy stable particles : the proton and deuteron from the baryon sector and the dark matter particle(s ) .
the next important era is termed either decoupling , recombination , or last scattering .
when the temperature reached roughly 1 ev , atoms ( mostly h ) formed and the radiation cooled too much to ionize .
the universe became transparent , and it was during this era that the cmb we see today was emitted , when physical separations were 1000 times smaller than today ( @xmath11000 ) .
at this point , less than one million years into the expansion , when electromagnetic radiation ceased playing an important dynamical role , baryonic matter began to collapse and cool , eventually forming the first stars and galaxies .
later , the first generation of stars and possibly supernova explosions seem to have provided enough radiation to completely reionize the universe ( at @xmath2 ) . throughout these stages ,
the expansion was decelerated by the gravitational force on the expanding matter .
now we are in an era of cosmic acceleration ( @xmath32 ) , where we find that approximately 70% of the energy density is in the fluid that causes the acceleration , 25% is in dark matter , and just 5% is in baryons , with a negligible amount in radiation .
the cmb is a record of the state of the universe at a fraction of a million years after the big bang , after a quite turbulent beginning , so it is not immediately obvious that any important information survives . certainly the fundamental information available in the collisions of elementary particles is best unraveled by observations within nanoseconds of the collision . yet even in this remnant radiation lies the imprint of fundamental features of the universe at its earliest moments .
one of the most important features of the cmb is its planck spectrum .
it follows the blackbody curve to extremely high precision , over a factor of approximately 1000 in frequency ( see figure 1 ) .
this implies that the universe was in thermal equilibrium when the radiation was released ( actually long before , as we see below ) , which was at a temperature of approximately 3000 k. today it is near 3 k. an even more important feature is that , to better than a part in 10@xmath4 , this temperature is the same over the entire sky .
this is surprising because it strongly implies that everything in the observable universe was in thermal equilibrium at one time in its evolution . yet at any time and place in the expansion history of the universe , there is a causal horizon defined by the distance light ( or gravity ) has traveled since the big bang ; at the decoupling era , this horizon corresponded to an angular scale of approximately 1@xmath5 , as observed today .
the uniformity of the cmb on scales well above 1@xmath5 is termed the horizon problem .
the most important feature is that there are differences in the cmb temperature from place to place , at the level of 10@xmath6 , and that these fluctuations have coherence beyond the horizon at the time of last scattering .
the most viable notion put forth to address these observations is the inflationary paradigm , which postulates a very early period of extremely rapid expansion of the universe .
its scale factor increased by approximately 21 orders of magnitude in only approximately 10@xmath7 s. before inflation , the small patch that evolves into our observable universe was likely no larger across than the planck length , its contents in causal contact and local thermodynamic equilibrium . the process of superluminal inflation disconnects regions formally in causal contact .
when the expansion slowed , these regions came back into the horizon and their initial coherence became manifest .
the expansion turns quantum fluctuations into ( nearly ) scale - invariant cmb inhomogeneities , meaning that the fluctuation power is nearly the same for all threedimensional fourier modes .
so far , observations agree with the paradigm , and scientists in the field use it to organize all the measurements .
nevertheless , we are far from understanding the microphysics driving inflation .
the number of models and their associated parameter spaces greatly exceed the number of relevant observables .
new observations , particularly of the cmb polarization , promise a more direct look at inflationary physics , moving our understanding from essentially kinematical to dynamical . for particle physicists , probing microphysics at energy scales beyond accelerators using cosmological observations
is attractive .
the physics of inflation may be associated with the grand unifcation scale , and if so , there could be an observable signature in the cmb : gravity waves .
metric perturbations , or gravity waves ( also termed tensor modes ) , would have been created during inflation , in addition to the density perturbations ( scalar modes ) that give rise to the structure in the universe today . in the simplest of inflationary models
, there is a direct relation between the energy scale of inflation and the strength of these gravity waves .
the notion is that the universe initially had all its energy in a scalar field @xmath9 displaced from the minimum of its potential @xmath10 .
@xmath11 is suitably constructed so that slowly rolls down its potential , beginning the inflationary era of the universe , which terminates only when @xmath9 approaches its minimum .
inflation does not predict the level of the tensor ( or even scalar ) modes .
the parameter @xmath12 is the tensor - to - scalar ratio for fluctuation power ; it depends on the energy scale at which inflation began . specifically , the initial height of the potential @xmath13 depends on @xmath14 , as @xmath15 .
a value of @xmath16 , perhaps the smallest detectable level , corresponds to @xmath17 .
the tensor modes leave distinct patterns on the polarization of the cmb , which may be detectable .
this is now the most important target for future experiments .
they also have effects on the temperature anisotropies , which currently limit @xmath14 to less than approximately 0.3 .
it is a remarkable fact that even a slight neutrino mass affects the expansion of the universe .
when the dominant dark matter clusters , it provides the environment for baryonic matter to collapse , cool , and form galaxies . as described above , the growth of these structures becomes more rapid in the matter - dominated era . if a significant fraction of the dark matter were in the form of neutrinos with electron - volt - scale masses ( nonrelativistic today ) , these would have been relativistic late enough in the expansion history that they could have moved away from overdense regions and suppressed structure growth .
such suppression alters the cmb patterns and provides some sensitivity to the sum of the neutrino masses .
note also that gravitational effects on the cmb in its passage from the epoch ( or surface ) of last scattering to the present leave signatures of that structure and give an additional ( and potentially more sensitive ) handle on the neutrino masses ( see section 1.9.2 ) .
we know from the cmb that the geometry of the universe is consistent with being flat .
that is , its density is consistent with the critical density .
however , the overall density of matter and radiation discerned today ( the latter from the cmb directly ) falls short of accounting for the critical density by approximately a factor of three , with little uncertainty .
thus , the cmb provides indirect evidence for dark energy , corroborating supernova studies that indicate a new era of acceleration . because the presence and possible evolution of a dark energy component alter the expansion history of the universe
, there is the promise of learning more about this mysterious component . in 1965 ,
penzias and wilson ( 3 ) , in trying to understand a nasty noise source in their experiment to study galactic radio emission , discovered the cmb arguably the most important discovery in all the physical sciences in the twentieth century .
shortly thereafter , scientists showed that the radiation was not from radio galaxies or reemission of starlight as thermal radiation .
this first measurement was made at a central wavelength of 7.35 cm , far from the blackbody peak .
the reported temperature was @xmath18 .
however , for a blackbody , the absolute flux at any known frequency determines its temperature .
figure 1 shows the spectrum of detected radiation for different temperatures . there is a linear increase in the peak position and in the flux at low frequencies ( the rayleigh - jeans part of the spectrum ) as temperature increases . ]
multiple efforts were soon mounted to confirm the blackbody nature of the cmb and to search for its anisotropies .
partridge ( 4 ) gives a very valuable account of the early history of the field .
however , there were false observations , which was not surprising given the low ratio of signal to noise . measurements of the absolute cmb temperature are at milli - kelvin levels , whereas relative measurements between two places on the sky are at micro - kelvin levels . by 1967 , partridge and wilkinson had shown , over large regions of the sky , that @xmath19 , leading to the conclusion that the universe was in thermal equilibrium at the time of decoupling ( 4 ) .
however , nonthermal injections of energy even at much earlier times , for example , from the decays of long - lived relic particles , would distort the spectrum .
it is remarkable that current precise measurements of the blackbody spectrum can push back the time of significant injections of energy to when the universe was barely a month old ( 5 ) .
thus , recent models that attribute the dark matter to gravitinos as decay products of long - lived supersymmetric weakly interacting massive particles ( susy wimps ) ( 6 ) can only tolerate lifetimes of less than approximately one month .
the solar system moves with velocity @xmath20 , causing a dipole anisotropy of a few milli - kelvins , first detected in the 1980s .
( note that the direction of our motion was not the one initially hypothesized from motions of our local group of galaxies . )
the first detection of primordial anisotropy came from the cobe satellite ( 7 ) in 1992 , at the level of 10@xmath6 ( 30 @xmath21k ) , on scales of approximately 10@xmath5 and larger .
the impact of this detection matched that of the initial discovery .
it supported the idea that structure in the universe came from gravitational instability to overdensities .
the observed anisotropies are a combination of the original ones at the time of decoupling and the subsequent gravitational red- or blueshifting as photons leave over- or underdense regions . here
we describe the usual techniques for characterizing the temperature field .
first , we define the normalized temperature @xmath22 in direction @xmath23 on the celestial sphere by the deviation @xmath24 from the average : @xmath25 .
next , we consider the multipole decomposition of this temperature field in terms of spherical harmonics @xmath26 : @xmath27 where the integral is over the entire sphere . if the sky temperature field arises from gaussian random fluctuations , then the field is fully characterized by its power spectrum @xmath28 .
the order @xmath29 describes the angular orientation of a fluctuation mode , but the degree ( or multipole ) @xmath30 describes its characteristic angular size .
thus , in a universe with no preferred direction , we expect the power spectrum to be independent of @xmath29 .
finally , we define the angular power spectrum @xmath31 by @xmath32 . here
the brackets denote an ensemble average over skies with the same cosmology .
the best estimate of @xmath31 is then from the average over m. because there are only the ( @xmath33 ) modes with which to detect the power at multipole @xmath30 , there is a fundamental limit in determining the power .
this is known as the cosmic variance ( just the variance on the variance from a finite number of samples ) : @xmath34 the full uncertainty in the power in a given multipole degrades from instrumental noise , finite beam resolution , and observing over a finite fraction of the full sky , as shown below in equation 9 . for historical reasons ,
the quantity that is usually plotted , sometimes termed the tt ( temperature - temperature correlation ) spectrum , is @xmath35 where @xmath36 is the blackbody temperature of the cmb .
this is the variance ( or power ) per logarithmic interval in @xmath30 and is expected to be ( nearly ) uniform in inflationary models ( scale invariant ) over much of the spectrum .
this normalization is useful in calculating the contributions to the fluctuations in the temperature in a given pixel from a range of @xmath30 values : @xmath37 data from other experiments are shown , in addition to the best - fit cosmological model to the wmap data alone .
note the multipole scale on the bottom and the angular scale on the top .
figure courtesy of the wmap science team[wmap - ps ] ] figure 2 shows the current understanding of the temperature power spectrum ( from herewith we redefine @xmath31 to have @xmath38 units by replacing @xmath31 with @xmath39 ) .
the region below @xmath40 indicates the initial conditions .
these modes correspond to fourier modes at the time of decoupling , with wavelengths longer than the horizon scale .
note that were the sky describable by random white noise , the @xmath31 spectrum would be flat and the tt power spectrum , defined by equation 3 , would have risen in this region like @xmath41 .
the ( pleasant ) surprise was the observation of finite power at these superhorizon scales . at high @xmath30 values , there are acoustic oscillations , which are damped at even higher @xmath30 values .
the positions and heights of the acoustic - oscillation peaks reveal fundamental properties about the geometry and composition of the universe , as we discuss below .
the cmb data reveal that the initial inhomogeneities in the universe were small , with overdensities and underdensities in the dark matter , protons , electrons , neutrinos , and photons , each having the distribution that would arise from a small adiabatic compression or expansion of their admixture .
an overdense region grows by attracting more mass , but only after the entire region is in causal contact .
we noted that the horizon at decoupling corresponds today to approximately 1@xmath5 on the sky .
only regions smaller than this had time to compress before decoupling . for sufficiently small regions , enough time elapses
that compression continues until the photon pressure is sufficient to halt the the electrons via thomson scattering , and the protons follow the electrons to keep a charge balance .
inflation provides the initial conditions - zero velocity .
decoupling preserves a snapshot of the state of the photon fluid at that time .
excellent pedagogical descriptions of the oscillations can be found at _
http://background.uchicago.edu/@xmath42whu/_. other useful pages are _ http://wmap.gsfc.nasa.gov/,http://space.mit.edu/home/tegmark/index.html _ and _ http://www.astro.ucla.edu/@xmath42wright / intro.html_. perturbations of particular sizes may have undergone ( a ) one compression , ( b ) one compression and one rarefaction , ( c ) one compression , one rarefaction , and one compression , and so on .
extrema in the density field result in maxima in the power spectrum .
consider a standing wave permeating space with frequency @xmath43 and wave number @xmath44 , where these are related by the velocity of displacements ( the sound speed , @xmath45 ) in the plasma : @xmath46 .
the wave displacement @xmath47 for this single mode can then be written as @xmath48 .
the displacement is maximal at time @xmath49 of decoupling for @xmath50 we add the tt subscript to label these wave numbers associated with maximal autocorrelation in the temperature .
note that even in this tightly coupled regime , the universe at decoupling was quite dilute , with a physical density of less than @xmath51 . because the photons diffuse their mean free path is not infinitely short this pattern does not go on without bound .
the overtones are damped , and in practice only five or six such peaks will be observed , as seen in figure 2 . to help explain these ideas , we reproduce a few frames from an animation by w. hu .
figure 3 shows a density fluctuation on the sky from a single @xmath44 mode and how it appears to an observer at different times .
the figure shows the particle horizon just after decoupling .
this represents the farthest distance one could in principle see approximately the speed of light times the age of the universe .
an observer at the center of the gure could not by any means have knowledge of anything outside this region .
of course , just after decoupling , the observer could see a far shorter distance .
only then could light propagate freely .
the subsequent frames show how the particle horizon grows to encompass more corrugations of the original density fluctuation . at first
the observer sees a dipole , later a quadrupole , then an octopole , and so on , until the present time when that single mode in density inhomogeneities creates very high multipoles in the temperature anisotropy .
it is instructive to think of how the temperature observed today at a spot on the sky arises from the local moments in the temperature field at the time of last scattering .
it is only the lowest three moments that contribute to determining the anisotropies .
the monopole terms are the ones transformed into the rich angular spectra .
the dipole terms also have their contribution : the motion in the fluid oscillations results in doppler shifts in the observed temperatures .
polarization , we see below , comes from local quadrupoles .
these frames show one superhorizon temperature mode just after decoupling with representative photons last scattering and heading toward the observer at the center .
left to right : just after decoupling ; the observer s particle horizon when only the temperature monopole can be detected ; som e time later when the quadrupole is detected ; later still when the 12-pole is detected ; and today , a very high , well aligned multipole , from just this single mode in @xmath44-space , is detected .
figure courtesy of w. hu [ hu-1 ] inflation is a mechanism whereby fluctuations are created without violating causality .
there does not seem to be a better explanation for the observed regularities .
nevertheless , wolfgang pauli s famous statement about the neutrino comes to mind : i have done a terrible thing : i have postulated a particle that can not be detected ! sometimes it seems that inflation is an idea that can not be tested , or tested incisively .
of course pauli s neutrino hypothesis did test positive , and similarly there is hope that the idea of inflation can reach the same footing . still , we have not ( yet ) seen any scalar field in nature .
we discuss what has been claimed as the smoking gun test of inflation the eventual detection of gravity waves in the cmb .
however , will we ever know with certainty that the universe grew in volume by a factor of 10@xmath52 in something like @xmath53 ? experiments have now shown that the cmb is polarized , as expected .
researchers now think that the most fruitful avenue to fundamental physics from the cmb will be in precise studies of the patterns of the polarization .
this section treats the mechanisms responsible for the generation of the polarization and how this polarization is described .
if there is a quadrupole anisotropy in the temperature field around a scattering center , even if that radiation is unpolarized , the scattered radiation will be as shown in figure 4 : a linear polarization will be generated . )
, the incoming quadrupole pattern produces linear polarization along the @xmath54-direction . in terms of the stokes parameters ,
this is @xmath55 , the power difference detected along the @xmath56- and @xmath54-directions .
linear polarization needs one other parameter , corresponding to the power difference between 45@xmath5 and 135@xmath5 from the x - axis .
this parameter is easily shown to be stokes @xmath57 .
right : e and b polarization patterns .
the length of the lines represents the degree of polarization , while their orientation gives the direction of maximum electric field .
frames courtesy of w. hu.,title="fig:",width=163 ] ) , the incoming quadrupole pattern produces linear polarization along the @xmath54-direction . in terms of the stokes parameters ,
this is @xmath55 , the power difference detected along the @xmath56- and @xmath54-directions .
linear polarization needs one other parameter , corresponding to the power difference between 45@xmath5 and 135@xmath5 from the x - axis .
this parameter is easily shown to be stokes @xmath57 .
right : e and b polarization patterns .
the length of the lines represents the degree of polarization , while their orientation gives the direction of maximum electric field .
frames courtesy of w. hu.,title="fig : " ] the quadrupole is generated during decoupling , as shown in figure 3 . because the polarization arises from scattering but said scattering dilutes the quadrupole , the polarization anisotropy is much weaker than that in the temperature field .
indeed with each scatter on the way to equilibrium , the polarization is reduced .
any remaining polarization is a direct result of the cessation of scattering . for this reason , the polarization peaks at higher @xmath30 values than does the temperature anisotropy .
the local quadrupole on scales that are large in comparison to the mean free path is diluted from multiple scattering .
the polarization field is both more complicated and richer than the temperature field . at each point in the sky
, one must specify both the degree of polarization and the preferred direction of the electric field .
this is a tensor field that can be decomposed into two types , termed e and b , which are , respectively , scalar and pseudoscalar fields , with associated power spectra .
examples of these polarization fields are depicted schematically in figure 4 .
the e and b fields are more fundamental than the polarization field on the sky , whose description is coordinate - system dependent .
in addition , e modes arise from the density perturbations ( which do not produce b modes ) that we describe , whereas the b modes come from the tensor distortions to the space - time metric ( which do have a handedness ) .
we mention here that the e and b fields are nonlocal .
their extraction from measurements of polarization over a set of pixels , often in a finite patch of sky , is a well - developed but subtle procedure ( see section 3.3 ) .
the peaks in the ee ( e - polarization correlated with itself ) spectrum should be 180 out of phase with those for temperature : polarization results from scattering and thus is maximal when the fluid velocity is maximal . calculating the fluid velocity for the mode in section 1.6
, we find @xmath58 , defining modes with maximal ee power .
the te ( e - polarization correlated with the temperature field ) spectrum how modes in temperature correlate with those with e polarization is also of cosmological interest , with its own peak structure .
here we are looking at modes that have a maximum at decoupling in the product of their temperature and e - mode polarization ( or velocity ) .
similarly , the appropriate maxima ( which in this case can be positive or negative ) are obtained when @xmath59 thus , between every peak in the tt power spectrum there should be one in the ee , and between every tt and ee pair of peaks there should be one in the te .
figure 5 shows the ee results in addition to the expected power spectra in the standard cosmological model .
measurements of the te cross correlation are also shown .
the pattern of peaks in both power spectra is consistent with what was expected .
what was unexpected was the enhancement at the lowest @xmath30 values in the ee power spectrum .
this is discussed in the next section .
the experiments reported in figure 5 , with 20 or fewer detectors , use a variety of techniques and operate in different frequency ranges .
this is important in dealing with astrophysical foregrounds ( see section 2 ) that have a different frequency dependence from that of the cmb . and that to display features in the very low @xmath30 range , we plot @xmath60.,title="fig : " ] and that to display features in the very low @xmath30 range , we plot @xmath60.,title="fig : " ] limits from current experiments on the b - mode power are now at the level of 1 - 10 @xmath61 , far from the expected signal levels shown in figure 6 .
the peak in the power spectrum ( for the gravity waves ) is at @xmath62 , the horizon scale at decoupling .
the reader may wonder why the b modes fall off steeply above this scale and show no acoustic oscillations .
the reason is simple : a tensor mode will give , for example , a compression in the x - direction followed by a rarefaction in the y - direction , but will not produce a net overdensity that would subsequently contract . in the final section
we discuss experiments with far greater numbers of detectors aimed specifically at b - mode science .
note that such gravity waves have frequencies today of order @xmath63 hz .
however , if their spectrum approximates one of scale invariance , they would in principle be detectable at frequencies nearer 1 hz , such as in the lisa experiment .
this is discussed more fully in reference 10 . in this section
we briefly discuss three important processes after decoupling : rescattering of the cmb in the reionized plasma of the universe , lensing of the cmb through gravitational interactions with matter , and scattering of the cmb from hot gas in galaxy clusters .
although these can be considered foregrounds perturbing the primordial information , each can potentially provide fundamental information . and @xmath64 . the predicted b - mode signal power spectrum due to the distortion of e modes by weak gravitational lensing is also shown .
estimated statistical sensitivities for a new space mission ( pink line ) and two sample ground - based experiments , as considered in reference 9 , each with 1000 detectors operating for one year with 100% duty cycle ( dark and light blue lines ) , are shown .
experiment i observes 4% of the sky , with a 6-arcmin resolution ; experiment ii observes 0.4% of the sky , with a 1-arcmin resolution . ] the enhancement in the ee power spectrum at the very lowest @xmath30 values in figure 5 is the signature that the universe was reionized after decoupling .
this is a subject rich in astrophysics , but for our purposes it is important in that it provides another source for scattering and hence detection of polarization . from the wilkinson microwave anisotropy probe ( wmap ) polarization data ( 11 ) , one can infer an optical depth of order 10% , the fraction of photons scattering in the reionized plasma somewhere in the region of @xmath65 .
this new scattering source can be used to detect the primordial gravity waves .
the signature will show up at very low @xmath30 values , corresponding to the horizon scale at reionization .
figure 6 shows that the region @xmath66 should have substantial effects from gravity waves .
most likely , the only means of detecting such a signal is from space , and even from there it will be very difficult .
the polarization anisotropies for this very low @xmath30 region are comparable to what is expected from the surface of last scattering ( @xmath67 ) .
there are disadvantages to each signature . at the lowest @xmath30 values , galactic foregrounds are more severe , there are fewer modes in which to make a detection , and systematic errors are likely greater . at the higher values ,
there is a foreground that arises from e modes turning into b modes through gravitational lensing ( the topic of the next section ) .
clearly , it will be important to detect the two signatures with the right relative strengths at these two very different scales
. both the temperature and polarization fields will be slightly distorted ( lensed ) when passing collapsing structures in the late universe .
the bending of light means that one is not looking ( on the last scattering surface ) where one thinks .
although lensing will affect both the polarization and t fields , its largest effect is on the b field , where it shifts power from e to b. gravitational distortions , although preserving brightness , do not preserve the e and b nature of the polarization patterns .
figure 6 also shows the expected power spectrum of these lensed b modes .
because this power is sourced by the e modes , it roughly follows their shape , but with @xmath24 suppressed by a factor of 20 .
the peak structure in the e modes is smoothed , as the structures doing the lensing are degree scale themselves .
owing to the coherence of the lensing potential for these modes , there is more information than just the power spectrum , and work is ongoing to characterize the expected cross correlation between different multipole bands .
this signal should be detectable in next - generation polarization experiments .
for our purposes , the most interesting aspect of this lensing is the handle it can potentially give on the masses of the neutrinos , as more massive neutrinos limit the collapse of matter along the cmb trajectories .
all other parameters held fixed , there is roughly a factor - of - two change in the magnitude of the b signal for a 1-ev change in the mean neutrino mass . at very small angular scales
@xmath30 values of a few thousand , way beyond where the acoustic oscillations are damped
there are additional effects on the power spectra that result from the scattering of cmb photons from electrons after the epoch of reionization , including scattering from gas heated from falling deep in the potential wells of galaxy clusters ( the sunyaev- zeldovich , or sz , effect ) .
these nonlinear effects are important as they can help in untangling ( a ) when the first structures formed and ( b ) the role of dark energy . in this section ,
we show how the power spectrum information is used to determine important aspects of the universe .
this is normally known as parameter estimation , where the parameters are those that define our cosmology .
the observable power spectrum is a function of at least 11 such basic parameters .
as we discuss below , some are better constrained than others .
first , there are four parameters that characterize the primordial scalar and tensor fluctuation spectra before the acoustic oscillations , each of which is assumed to follow a power law in wave number .
these four are the normalization of the scalar fluctuations ( @xmath68 ) , the ratio of tensor to scalar fluctuations @xmath14 , and the spectral indices for both ( historically denoted with @xmath69 and @xmath70 ) .
second , there is one equation - of - state parameter ( @xmath71 ) that is the ratio of the pressure of the dark energy to its energy density , and one parameter that gives the optical depth ( @xmath72 ) from the epoch of reionization . finally , there are five parameters that characterize the present universe : its rate of expansion ( hubble constant , with @xmath73 km s@xmath74 mpc@xmath74 ) , its curvature ( @xmath75 ) , and its composition ( baryon density , matter density , and dark energy density ) .
the latter three are described in terms of energy densities with respect to the critical density normalized to the present epoch : @xmath76 , and @xmath77 .
just 10 of these are independent as @xmath78 .
even though the cmb data set itself consists of hundreds of measurements , they are not sufficiently orthogonal with respect to the 10 independent parameters for each to be determined independently ; there are significant degeneracies .
hence , it is necessary to make assumptions that constrain the values of those parameters upon which the data have little leverage . in some cases , such prior assumptions ( priors ) can have large effects on the other parameters , and there is as yet no standard means of reporting results .
several teams have done analyses [ wmap ( 11 , 12 ) , cbi ( 13 ) , boomerang ( 14 ) , see also reference 15 ] . here
we first discuss the leverage that the cmb power spectra have on the cosmological parameters .
then we give a flavor for the analyses , together with representative results .
we consider analyses , done by the several teams , with just the six most important parameters : @xmath79 , and @xmath80 , where the other five are held fixed .
for this discussion we are guided by reference 12 . completely within cmb data
, there is a geometrical degeneracy between @xmath75 , a contribution to the energy density from the curvature of space , and @xmath81 .
however , taking a very weak prior of @xmath82 , the wmap team , using just their first - year data , determined that @xmath83 , that is , no evidence for curvature .
we assume @xmath84 unless otherwise noted .
this conclusion has gotten stronger with the three - year wmap data together with other cmb results , and it is a prediction of the inflationary scenario .
nevertheless , we emphasize that it is an open experimental issue .
the position of the first acoustic peak reveals that the universe is flat or nearly so .
as we describe above , the generation of acoustic peaks is governed by the ( comoving ) sound horizon at decoupling , @xmath85 ( i.e. , the greatest distance a density wave in the plasma could traverse , scaled to today s universe ) .
the sound horizon depends on @xmath86 , and the radiation density , but not on @xmath87 , or the spectral tilt @xmath88 .
the peak positions versus angular multipole are then determined by @xmath89 , where the quantity @xmath90 , the angular diameter distance , is the distance that properly takes into account the expansion history of the universe between decoupling and today so that when @xmath90 is multiplied by an observed angle , the result is the feature size at the time of decoupling . in a nonexpanding universe , this would simply be the physical distance .
the expression depends on the ( evolution of the ) content of the universe . for a flat universe
, we have @xmath91 in this expression , @xmath92 indicates the ( well - known ) radiation density , and the dilutions of the different components with redshift @xmath0 , between decoupling and the present , enter explicitly .
it is easy to see how one in principle determines spectral tilt .
if one knew all the other parameters , then the tilt would be found from the slope of the power spectrum after the removal of the other contributions .
however , there is clearly a coupling to other parameters .
experiments with a very fine angular resolution will determine the power spectrum at very high @xmath30 values , thereby improving the measurement of the tilt .
here we discuss the primary dependences of the acoustic peak heights on @xmath93 and @xmath94 . increasing @xmath93 decreases the peak heights . with greater matter density , the era of equality
is pushed to earlier redshifts , allowing the dark matter more time to form deeper potential wells .
when the baryons fall into these wells , their mass has less effect on the development of the potential so that the escaping photons are less redshifted than they would be , yielding a smaller temperature contrast . as to @xmath94 , increasing it decreases the second peak but enhances that of the third because the inertia in the photon - baryon fluid is increased , leading to hotter compressions and cooler rarefactions ( 16 ) .
the peak - height ratios give the three parameters @xmath95 , and @xmath94 , with a precision just short of that from a full analysis of the power spectrum ( discussed in section 3.4.4 ) .
following wmap , we define the ratio of the second to the first peak by @xmath96 , the ratio of the third to the second peak by @xmath97 , and the ratio of the first to the second peak in the polarization - temperature cross - correlation power spectrum by @xmath98 .
table 1 shows how the errors in these ratios propagate into parameter errors .
we see that all the ratios depend strongly on @xmath88 , and that the ratio of the first two peaks depends strongly on @xmath94 but is also influenced by @xmath93 . for @xmath97 ,
the relative dependences on @xmath94 and @xmath93 are reversed .
finally , the baryon density has little influence on the ratio of the te peaks .
however , increasing @xmath93 deepens potential wells , increasing fluid velocities and the heights of all polarization peaks .
.matrix of how errors in the peak ratios ( defined in text ) relate to the parameter errors .
[ cols="^,^,^,^",options="header " , ] here we discuss experiments needed to detect the b - mode signals of lensing at large @xmath30 values and of gravity waves at intermediate and small @xmath30 values .
table 3 shows the sensitivity required to make 3@xmath99 detections of several target signals .
estimates come from equation 9 , either for the full sky or for a smaller patch where the balance between sample variance and detector noise is optimized .
we give expressions for the sensitivity to a feature in the power spectrum centered at @xmath100 and with width @xmath101 and for the fraction of the sky that accomplishes the balance : ( @xmath102 and @xmath103 . for our purposes , both the signal from lensing and from primordial gravity waves have approximately this shape . here
@xmath104 is the ( peak ) cosmological signal of interest and @xmath105 is the total sensitivity of the experiment , summing over all the observing time and all the detectors . the lensing detection can be accomplished by observing for a year ( from the ground at a good site such as the atacama desert in chile or the south pole ) a patch of approximately 1.6 square degrees , with detectors having 1.5 times the wmap sensitivity .
the table also gives the sensitivities required to detect primordial b modes at various levels of @xmath12 .
foregrounds will be a problem for @xmath106 , perhaps more so for the detection of the signal from the reionized plasma than from the surface of last scattering .
a satellite experiment can detect the signal from the reionized plasma , where the lensing contamination is nearly negligible . the planck experiment ( 2007 )
has an order of magnitude in temperature sensitivity over wmap .
polarization sensitivity was not a primary goal .
still , much has gone into making sure the residual systematic uncertainties ( and foregrounds ) can be understood sufficiently well to allow the extraction of polarization signals around 50 nk , corresponding to @xmath107 .
there is a program of experiments over the coming five to eight years .
these will involve , progressively , tens , then hundreds , and nally a thousand or more detectors per experiment and will test polarization modulation schemes , effective scan strategies , foreground - removal methods , and algorithms for separating e and b modes . experiments with tens of detectors are already underway .
the sister experiments quad and bicep observe from the south pole , using polarization - sensitive bolometers at 100 and 150 ghz .
quad , with a 4-arcmin beam , is optimized for gravitational lensing , whereas bicep , at approximately 40 arcmin , is searching for gravitational waves .
mbi and its european analog brain are testing the idea of using bolometers configured as an interferometer , and pappa is a balloon effort using waveguide - coupled bolometers from the goddard space flight center .
these latter experiments have beams in the range of 0.5@xmath5 to 1@xmath5 .
five initiatives at the level of hundreds of detectors have so far been put forth .
four use tes bolometers at 90 , 150 , and 220 ghz : clover , the lone european effort , with an 8-arcmin beam ; polarbear , with a 4-arcmin beam ; and ebex and spider , balloon - borne experiments with 8- and 20 70-arcmin beams , respectively ( spider and polarbear will use antenna - coupled devices ) .
the fifth uses coherent detectors at 44 and 90 ghz : quiet , initially with a 12-arcmin beam , observing from the atacama desert .
all are dedicated ground - up polarization experiments that build their own optical systems .
the act and spt groups , supported by the national science foundation , deploy very large telescopes to study both the cosmology of clusters via the sz effect and fine - scale temperature anisotropies , and will likely propose follow - up polarimeters .
the reach of the next satellite experiment , termed cmbpol as defined by the three - agency task force in the united states ( 36 ) and termed bpol in europe , is to detect the signal from gravity waves limited only by astrophysical foregrounds . examining figure 6
, we see that @xmath64 , and possibly lower values , can be reached.we should know a great deal from the suborbital experiments well before the 2018 target launch date . for studying polarization at large scales , where foregrounds pose their greatest challenge , information from wmap and planck
will be the most valuable .
there is considerable promise for new , important discoveries from the cmb , ones that can take us back to when the universe s temperature was between the grand unified theory and planck energy scales .
this is particle physics , and while we hope accelerators will provide crucial evidence for , for example , the particle nature of dark matter , exploring these scales seems out of their reach . in some ways
, cosmology has followed the path of particle physics : it has its standard model , accounting for all confirmed phenomena . with no compelling theory ,
parameter values are not of crucial interest .
we can not predict the mass of the top quark , nor can we predict the primordial energy densities .
each discipline is checking consistency , as any discrepancies would be a hint of new physics .
the cmb field is not as mature as particle physics .
it needs considerable detector development , even for current experiments .
there is rapid progress , and overall sensitivity continues to increase .
foregrounds are certainly not sufficiently known or characterized .
there is a great deal of competition in the cmb , like the early days of particle physics before the experiments grew so large that more than one or two teams exploring the same topic worldwide was too costly . for the moment
, this is good , as each team brings something unique in terms of control of systematics , frequencies , regions of the sky scanned , and detection technology .
however , there is a difference in the way results are reported in the two fields . in the cmb field , typically almost nothing is said about an experiment between when it is funded and when it publishes . here
publishing means that results are announced , multiple papers are submitted and circulated , and often there is a full data release , including not only of raw data and intermediate data products but sometimes support for others to repeat or extend the analyses .
the positives of this tradition are obvious .
however , one negative is that one does not learn the problems an experiment is facing in a timely manner .
there is a degree of secrecy among cmb scientists .
there are other differences .
cmb teams frequently engage theorists to perform the final analysis that yields the cosmological significance of the data .
sophisticated analysis techniques are being developed by a set of scientists and their students who do not work with detectors but do generate a growing literature .
there are as yet no standardized analysis techniques ; effectively each new experiment invents its own .
the days appear to be over where the group of scientists that design , build , and operate an experiment can , by themselves , do the full scientific analysis .
another distinction is that there is no one body looking over the field or advising the funding agencies , and private funds sometimes have a major impact .
nearly all cmb scientists are working on multiple projects , sometimes as many as four or five , holding that many grants .
more time is spent writing proposals and reports and arranging support for junior scientists , for whom there is little funding outside of project funds .
this is certainly not the optimum way to fund such an exciting and promising field .
the authors have enjoyed many productive conversations with colleagues at their institutions in the course of preparing this review .
the authors are particularly grateful for helpful comments from norman jarosik , bernd klein , laura la porta , and kendrick smith .
we also wish to acknowledge our collaborators in quiet and capmap for frequent discussions of relevant issues .
this work was partially supported by grants from the national science foundation : phy-0355328 , phy- 0551142 , and astr / ast-0506648 .
* kamionkowski m , kosowsky a. annu .
rev . nucl . part .
sci . 49:77
( 1999 ) * dodelson s. modern cosmology .
acad . press ( 2003 ) * penzias aa , wilson rw .
j. 142:419 ( 1965 ) * partridge rb .
3k : the cosmic microwave background radiation .
cambridge , ny : cambridge univ . press ( 1995 ) * fixsen dj , et al . astrophys .
j. 483:586 ( 1996 ) * feng jl , rajaraman a , takayama f. phys .
d 68:063504 ( 2003 ) * smoot
gf , et al .
j. 396:1 ( 1992 ) * hinshaw g , et al .
j. in press ( 2007 ) * bock j , et al .
astro - ph/0604101 ( 2006 ) * smith tl , kamionkowski m , cooray a. phys . rev .
d 73(2):023504 ( 2006 ) * spergel dn , et al .
j. in press ( 2007 ) * page l , et al .
j. suppl . 148:233 ( 2003 ) * readhead acs , et al .
science 306:836 ( 2004 ) * mactavish cj , et al .
j. 647:799 ( 2006 ) * tegmark m , et al .
d 69:103501 ( 2004 ) * hu w , sugiyama n. phys .
d 51:2599 ( 1995 ) * eisenstein dj , et al .
j. 633:560 ( 2005 ) * bennett cl , et al .
j. suppl . 148:97 ( 2003 )
* page l , et al .
j. in press ( 2007 ) * la
porta l , burigana c , reich w , reich p. astron . astrophys .
455(2):l9 ( 2006 ) * carretti e , bernardi g , cortiglioni s. mnras
373:93 ( 2006 ) * hanany s , rosenkranz p. new astron .
47(11 12):1159 ( 2003 ) * pietranera l , et al .
mnras . in press ( 2007 )
* finkbeiner dp , davis m , schlegel dj .
j. 524:867 ( 1999 ) * vaillancourt je . eur .
publ . ser .
23:147 ( 2007 ) * benoit a , et al .
astron . astrophys .
424:571 ( 2004 ) * ponthieu n , et al .
astron . astrophys .
444:327 ( 2005 ) * huffenberger km , eriksen hk , hansen fk .
j. 651:81 ( 2006 ) * stivoli f , baccigalupi c , maino d , stompor r. mnras 372:615 ( 2006 ) * verde l , peiris hv , jimenez r. j. cosmol .
1:19 ( 2006 ) * barkats d , et al . astrophys .
159(1):1 ( 2005 ) * ryle m. proc .
a 211:351 ( 1952 ) * zmuidzinas j. appl .
optics 42:4989 ( 2003 ) * mather jc
optics 21:1125 ( 1982 ) * tegmark m. phys .
d 56:4514 ( 1997 ) * smith k. new astron .
50(11 12):1025 ( 2006 ) * crawford t. astro - ph/0702608 ( 2007 ) * couchout f , et al .
cmb and physics of the early universe .
cgi - bin / reader / conf.cgi?confid=27 ( 2006 ) * bond jr , jaffe ah , knox l. astrophys .
j. 533:19 ( 2000 ) * hivon e , et al . astrophys .
j. 567:2 ( 2002 ) | we intend to show how fundamental science is drawn from the patterns in the temperature and polarization fields of the cosmic microwave background ( cmb ) radiation , and thus to motivate the field of cmb research .
we discuss the field s history , potential science and current status , contaminating foregrounds , detection and analysis techniques and future prospects . throughout the review
we draw comparisons to particle physics , a field that has many of the same goals and that has gone through many of the same stages . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.